A molecular information ratchet using a cone-shaped macrocycle

1 Introduction The widespread use of molecular machines in biology has long suggested that great rewards could come from bridging the gap between synthetic molecular systems and the machines of the macroscopic world. In the last two decades, it has proved possible to design synthetic molecular systems with architectures where triggered large amplitude positional changes of submolecular components occur. Perhaps the best way to appreciate the technological potential of controlled molecular-level motion is to recognize that nanomotors and molecular-level machines lie at the heart of every significant biological process. Over billions of years of evolution, nature has not repeatedly chosen this solution for performing complex tasks without good reason. When mankind learns how to build artificial structures that can control and exploit molecular level motion and interface their effects directly with other molecular-level substructures and the outside world, it will potentially impact on every aspect of functional molecule and materials design. An improved understanding of physics and biology will surely follow. The first steps on the long path to the invention of artificial molecular machines were arguably taken in 1827 when the Scottish botanist Robert Brown observed the haphazard motion of tiny particles under his microscope. 1,2 The explanation for Brownian motion, that it is caused by bombardment of the particles by molecules as a consequence of the kinetic theory of matter, was later provided by Einstein, followed by experimental verification by Perrin. 3,4 The random thermal motion of molecules and its implications for the laws of thermodynamics in turn inspired Gedankenexperiments (“thought experiments”) that explored the interplay (and apparent paradoxes) of Brownian motion and the Second Law of Thermodynamics. Richard Feynman’s famous 1959 lecture “There’s plenty of room at the bottom” outlined some of the promise that manmade molecular machines might hold. 5,6 However, Feynman’s talk came at a time before chemists had the necessary synthetic and analytical tools to make molecular machines. While interest among synthetic chemists began to grow in the 1970s and 1980s, progress accelerated in the 1990s, particularly with the invention of methods to make mechanically interlocked molecular systems (catenanes and rotaxanes) and control and switch the relative positions of their components. 7−24 Here, we review triggered large-amplitude motions in molecular structures and the changes in properties these can produce. We concentrate on conformational and configurational changes in wholly covalently bonded molecules and on catenanes and rotaxanes in which switching is brought about by various stimuli (light, electrochemistry, pH, heat, solvent polarity, cation or anion binding, allosteric effects, temperature, reversible covalent bond formation, etc.). Finally, we discuss the latest generations of sophisticated synthetic molecular machine systems in which the controlled motion of subcomponents is used to perform complex tasks, paving the way to applications and the realization of a new era of “molecular nanotechnology”. 1.1 The Language Used To Describe Molecular Machines Terminology needs to be properly and appropriately defined and these meanings used consistently to effectively convey scientific concepts. Nowhere is the need for accurate scientific language more apparent than in the field of molecular machines. Much of the terminology used to describe molecular-level machines has its origins in observations made by biologists and physicists, and their findings and descriptions have often been misinterpreted and misunderstood by chemists. In 2007 we formalized definitions of some common terms used in the field (e.g., “machine”, “switch”, “motor”, “ratchet”, etc.) so that chemists could use them in a manner consistent with the meanings understood by biologists and physicists who study molecular-level machines. 14 The word “machine” implies a mechanical movement that accomplishes a useful task. This Review concentrates on systems where a stimulus triggers the controlled, relatively large amplitude (or directional) motion of one molecular or submolecular component relative to another that can potentially result in a net task being performed. Molecular machines can be further categorized into various classes such as “motors” and “switches” whose behavior differs significantly. 14 For example, in a rotaxane-based “switch”, the change in position of a macrocycle on the thread of the rotaxane influences the system only as a function of state. Returning the components of a molecular switch to their original position undoes any work done, and so a switch cannot be used repetitively and progressively to do work. A “motor”, on the other hand, influences a system as a function of trajectory, meaning that when the components of a molecular motor return to their original positions, for example, after a 360° directional rotation, any work that has been done is not undone unless the motor is subsequently rotated by 360° in the reverse direction. This difference in behavior is significant; no “switch-based” molecular machine can be used to progressively perform work in the way that biological motors can, such as those from the kinesin, myosin, and dynein superfamilies, unless the switch is part of a larger ratchet mechanism. 14 1.2 The Effects of Scale Machines need to be designed according to the environment in which they are intended to operate. The significance and consequences of random thermal motion, heat dissipation, solvation, momentum, inertia, gravity, etc., differ significantly at the molecular and macroscopic levels, meaning that nanoscale machines cannot simply mimic the mechanisms of their macroscopic brethren. The forces with the greatest influence on dynamics in the nanoworld are not those we commonly rely on in the macroscopic world. For large objects, inertial terms, which depend on the mass of the particle, dominate motion. As particle size decreases to or below the micrometre scale, viscous forces and Brownian motion become dominant while momentum and gravity become increasingly irrelevant. This effect can be quantified by the Reynolds number (R) for a particle of radius a, velocity v in a medium of density ρ and viscosity μ: 25,26 1 Because the Reynolds number decreases with radius, molecular-sized machines typically operate under low Reynolds number conditions. Furthermore, the larger surface:volume ratio of small particles makes them increasingly adhesive, decreasing mobility. These effects must be carefully considered in the design of molecular machines. 14 1.3 Thought Machines Exploring Brownian Motion Molecular motors rely on exploiting random thermal fluctuations for directional motion by employing ratchet mechanisms. Thought experiments such as Maxwell’s demon, 27,28 Smoluchowski’s trapdoor, 29 and Feynman’s ratchet-and-pawl 30 have investigated potential ways to cause the directional motion of Brownian particles. 31 The Second Law of Thermodynamics states that the entropy of an isolated system tends to increase, leading to an equilibrium distribution with maximum entropy. To achieve any distribution other than the thermodynamic equilibrium, work must be done on the system. Various thought experiments have been proposed that attempt to violate this premise and drive a system away from equilibrium without expending work. In the Maxwell’s Demon thought experiment, particles in a container are sorted by an “intelligent gatekeeper”, the “demon” (Figure 1 ). 32 Gas particles are distributed in a container partitioned into two sections in an isolated system. The spontaneous formation of a heat or pressure gradient would lead to a decrease in entropy and thus violate the Second Law. The gatekeeper is able to detect the velocity of each particle and can control the gate accordingly. The demon allows particles with higher than average speeds (shown in red in Figure 1 a) to pass from the right section to the left, but not from the left to the right. The opposite holds for slower particles (shown in blue in Figure 1 a). The gatekeeper thus causes a nonuniform distribution of particles between the two sections, creating a temperature gradient. Similarly, a demon capable of detecting the direction of the particle and opening the gate accordingly can concentrate particles in one section and thus generate a pressure gradient (Figure 1 b). If the gate is frictionless, then in both cases a nonequilibrium distribution has been achieved seemingly without doing work, which would violate The Second Law. The solution to the apparent paradox lies in considering the information required by the demon to know when to open the door. As the demon must measure the velocity or direction of each particle approaching the door 33 and cannot have infinite memory, 34 at some point the demon must “forget” this information. The destruction of information has a minimum enthalpic cost associated with it 35 as has been experimentally verified, 36 which always exceeds the decrease in entropy in the container. Thus, a local decrease in entropy is paid for by a generalized increase upon information deletion. Figure 1 Maxwell’s demons. (a) A “temperature demon” sorts particles according to velocity, generating a temperature gradient. (b) A “pressure demon” sorts particles according to their direction of movement, generating a pressure gradient. 37 Smoluchowski proposed a system that did not rely on an intelligent observer based on a spring-loaded one-way gate, opened by thermal fluctuations (Figure 2 a). 29,38,39 However, directional transport was proven to be impossible in this system as the gate is unable to dissipate energy from collisions with gas particles. This leads to ever longer open periods for the gate and so to equilibration of the particles across the container. Figure 2 (a) In Smoluchowski’s Trapdoor, a spring-loaded gate separates the two sections of a container. The lack of energy dissipation leads to a longer duration of opening and thus a uniform particle distribution between the sections. 29 (b) Feynman’s ratchet-and-pawl device. The asymmetry of the teeth of the ratchet cog was intended to drive directional movement. However, when T 1 = T 2, the rotation is nondirectional. 30 In Feynman’s ratchet-and-pawl, the asymmetric cog teeth could have allowed directional rotation powered by Brownian motion (Figure 2 b). 30,40−44 Although in line with the second law, no work could be extracted when T 1 = T 2 (the pawl is at the same temperature as the rest of the device and thus fluctuates between open and closed states and so fails to act as a ratchet), but work can be done when T 1 > T 2. This is an important point for the design of nanoscale machines. As the system moves from a nonequilibrium state toward equilibrium (T 1 = T 2), directional work can be done. Hence, if a nonequilibrium situation can be created, its relaxation can drive directional motion. In other words, an energy input is required to do work and to break detailed balance (see section 4.4 ). 1.4 Inspiration from Nature Perhaps the most important lesson to learn from biological systems is that molecular machines are viable and can perform extremely complex tasks. In nature, molecular machines play a vital role, being involved in almost every major biological process and allowing a vast array of chemical and mechanical tasks to be accomplished. 45−47 The ways in which nature has overcome the problems of scale, Brownian motion, viscosity, nonequilibrium distribution and environment provide some general directions for the design of molecular machines. Cellular processes are compartmentalized by lipid barriers, which allow nonequilibrium distributions to be built up and maintained. A range of molecular machines control the transport of various charged and polar species across these membranes using relay channels or mobile carrier entities. 48−50 This transfer can be driven by passive diffusion down a concentration gradient or by the generation of an electrochemical gradient (where a species is moved against its concentration gradient but down another gradient, such as electrical potential). If the motion is driven by the coupled transport of another system, the two species can migrate in either the same (symport) or opposite (antiport) directions across the membrane. Highly selective transport across lipid bilayers is made possible by the large number of specific channels and carriers in biological systems driven by the relaxation of transmembrane electrochemical gradients toward thermodynamic equilibrium. These electrochemical gradients are maintained by only a few ion pumps, most commonly ATPases that hydrolyze ATP and generate directional transport using the resulting release of energy. Although the mechanism of directional motion in these motor proteins is not yet fully understood, conformational changes leading to altered binding affinities seem to be particularly important. 51−60 Access to the binding sites of these pumps is often “closed” by conformational changes after ion binding to increase the directionality of the process. 58,61 Proton pumping across membranes is a particularly important process as it creates the proton motive force that drives ATP synthases’ production of ATP, which in turn powers a large variety of active transport processes. This proton generated transmembrane potential is most commonly created by electron transport chains where redox reactions are used to separate charges across a membrane. 60 Proton pumping powered directly by light, as in bacteriorhodopisin (where cis/trans isomerization leads to directionality), is less common. 62−67 Interlocked structures are often exploited in biology to help ensure high sequentiality, for example, in the ribosome, 68−70 and various DNA polymerases. 71 Biological molecular machines are used to transport cargo about a cell (e.g., kinesin), to power the movement of organisms (e.g., bacterial flagellar motors), to synthesize proteins (e.g., the ribosome), and to separate strands of DNA (e.g., helicases). There are many crucial differences between these machines and those familiar to us in the macroscopic world, in terms of both the tasks they accomplish and the manner in which they do so. Several concepts pertinent to the synthesis of artificial molecular machines can be extrapolated from these natural machines. (1) Biological machines cannot use thermal gradients due to the rapid dissipation of heat at the scale on which they operate. (2) Biological motors often use chemical energy from favorable bond forming/breaking events (e.g., ATP hydrolysis), or concentration gradients to drive their operation. (3) Biomachines operate in solution and on surfaces under high viscosity conditions. (4) Biological machines exploit Brownian motion rather than fight against it. They “ratchet” the random thermal motion of their components and substrates. Brownian motion also ensures the rapid mixing of machines, fuel, and substrates. (5) Friction is irrelevant in such a viscous medium, where constant Brownian motion ensures the mobility of individual parts. (6) Mobile biomachines, such as kinesin, often operate along tracks to reduce the available degrees of freedom of the machine. Kinetic association of the machine on track for the duration of operation is vital to achieve directed transport. (7) Noncovalent interactions are often significant features of the structure and operation of biological systems. (8) Biomachines are often made by self-assembly processes from a limited range of basic motifs such as amino acids, lipids, nucleic acids, and saccharides. (9) Compartmentalization is often required to allow systems to be maintained and operate away from equilibrium. (10) Small binding events can often cause large conformational changes. Although biological molecular machines are of a level of complexity unattainable by the current generation of molecular machines, they are constricted by the evolutionary processes that gave rise to them and the environment in which they operate. Natural selection ensures that the first successful solution to a problem tends to be retained, and improved gradually over many iterations. These restrictions, in terms of chemistry, and solutions, do not apply to manmade systems. Biomachines also operate in a highly cluttered environment and thus must exhibit an extremely high tolerance to collisions with unrelated/unreactive species. Additionally, the study of less elaborate artificial systems may well give rise to a greater understanding of the sophisticated biological systems that inspired them. Nonequilibrium statistical mechanics can provide a more precise understanding of the basic mechanisms and processes used by the current crop of molecular machines and will be explored in the next section. 1.5 Directional Transport and Work under Brownian Motion The Principle of the Detailed Balance provides important constraints on the relationship between equilibrium and rate constants. An understanding of this concept is important in the field of synthetic molecular machines and for the design of potential mechanisms for directional motion. 72−74 The Principle of the Detailed Balance states that, at equilibrium, each transition has an equal probability (and thus rate) of occurring in a forward or reverse direction (e.g., k 1 = k –1). This means that at equilibrium no net flux is generated across any barrier and no task can be performed. This rules out the maintenance of an equilibrium by a cyclic process such as A → B → C → A; rather each substrate must be in equilibrium with every other that it is in exchange with, that is, A⇆ B + B ⇆ C + C ⇆ A instead. It follows that Smoluchowski’s trapdoor could not operate as it would break the Principle of Microscopic Reversibility. Breaking detailed balance is a requirement of doing work at the length scales on which Brownian motion occurs. 75 Stochastic pumping breaks detailed balance using an oscillating parameter to generate net flux and maintain a nonequilibrium steady state, and is seen in nature, for example, in the ATPases. 76,77 Several theoretical frameworks have been developed to explain various fluctuation driven transport modes such as stochastic pumping, which govern transport in Brownian motors and ratchets. 78−126 Some particularly relevant examples are discussed below. 1.6 Ratchet Mechanisms Brownian ratchet mechanisms fall into two general classes: energy ratchets (pulsating and tilting ratchets are subcategories of energy ratchets) and information ratchets. 127−129 1.6.1 Pulsating Ratchets In pulsating ratchets, potential-energy minima and maxima are varied in a periodic or stochastic manner with no reference to the position of the particle on the potential-energy surface. 128 The simplest form is an on–off ratchet where the potential is repeatedly turned on and off more rapidly than the diffusion of the Brownian particles over their potential energy surface. This results in the net directional transport of particles across the surface (Figure 3 ). Figure 3 An on–off ratchet: (a) the particles are located in an energy minima, (b) the potential is turned off so that diffusion can occur for a short time, and (c) the potential is turned on again. As the potential is asymmetric, particles have a greater probability of being trapped in an adjacent well to the right of the original one than to the left. (d) Relaxation into the local energy minima leads to the average position of the particles moving to the right. 128 A flashing ratchet is a subtype of the pulsating ratchet, consisting of a repeating series of maxima and minima. The sequential lowering and raising of parts of this potential energy surface leads to directional transport (Figure 4 ). Figure 4 A flashing ratchet. In (a) and (c), the particle starts in a green or orange well, respectively. Raising this energy minima while lowering the adjacent maxima and minima triggers movement by Brownian motion (b) to (c) or (d) to (e). By continuously varying the relative heights of the energy barriers and minima of the energy wells, the particle can be directionally transported. 128 An asymmetric potential energy surface is not strictly necessary to generate a pulsating ratchet. 128 A periodic array of local minima traveling at a constant velocity (i.e., a traveling wave) can be used to form a traveling potential ratchet. In terms of realistic systems, the traveling potential ratchet mechanism has most relevance for the field-driven processes discussed in section 5 and the self-propulsion mechanisms of section 6 . 1.6.2 Tilting Ratchets In a tilting ratchet, the underlying potential energy surface remains unchanged, and detailed balance is broken by the application of an unbiased driving force to the particles, such as heat. 128 When the applied force is heat, the ratchets are also referred to as temperature or diffusion ratchets (Figure 5 ). In its simplest form, the mechanism is very similar to that of the on–off ratchet. Here, particles cannot cross the energy maxima at low temperature, but a brief increase in temperature allows the diffusion of particles across the surface. A lowering of temperature traps this movement in a manner similar to the on–off pulsating ratchet leading to directional transport. The period of raised temperature must be brief to prevent general diffusion, which would destroy any directionality initially gained. Figure 5 A temperature or diffusion ratchet. (a) The particles are located in an energy minima on the potential-energy surface, with energy barriers ≫k B T 1. (b) The temperature is increased so that the height of the barriers is ≪k B T 2, and free diffusion is allowed to occur for a short time. (c) As the temperature is lowered again, the asymmetric potential energy surface means that the particles have a greater probability of being trapped to the right of their initial position. (d) Relaxation to the local energy minima. 128 Directional motion can also be achieved by the application of a periodic directional force to an asymmetric potential energy surface in a rocking ratchet. Although the net driving force averages to zero, directional motion is driven by exploiting the asymmetry of the potential energy surface (Figure 6 ). Figure 6 A rocking ratchet. (a) Particles are located in an energy minima on the potential energy surface. (b) A directional force is applied to the left. (c) An equal and opposite directional force is applied to the right. (d) Removal of the force and relaxation to an energy minima leads to the average position of the particles moving to the right. 128 Similarly, in a fluctuating force ratchet, the applied driving force is randomly generated and varies stochastically. Finally, in an asymmetric tilting ratchet, the particles lie on a symmetric potential energy surface, but the applied potential provides the required asymmetry. For example, in a rocking ratchet, the tilt in one direction would be longer than in the other. 1.6.3 Information Ratchets Energy ratchets operate by the application of an external force or the modulation of a potential energy surface irrespective of the particle’s position. 128 Information ratchets, on the other hand, rely on raising or lowering an energy barrier according to the position of the particle on the potential energy surface (Figure 7 ). This results in the distribution of particles being driven away from equilibrium. This process requires information transfer from the particle to the potential-energy surface, in a manner reminiscent of Maxwell’s demon. Figure 7 An information ratchet. In (a) and (d) the dotted lines represent the transfer of information about the position of the particle. (b) The position of the particle lowers the energy barrier to movement to the right-hand well but not the left. (c) The particle moves by Brownian motion. 128 1.6.4 Language Necessary To Describe the Operation of Molecular Machines Four terms are useful for describing the relationship of the components/substrates of molecular machines in terms of dynamics: balance, linkage, ratcheting, and escapement. 75 Having the components/substrates in a state of “balanced/unbalanced” or “linked/unlinked” plays a crucial role in determining whether or not a machine can perform a task. “Balance” is the thermodynamically preferred distribution of a particle (or submolecular component) in a molecular system. The trigger for net transportation comes from the balance being broken (i.e., the system being in a nonequilibrium state) (section 4.4 ). “Linkage” is the communication necessary for the transport of a particle between parts of the machine. However, the ability to exchange a particle between two parts of a machine does not necessarily allow a task to be performed, as a driving force is also required. Linking and unlinking operations are not limited to the addition or removal of steric bulk. They refer to any action that has an influence on the rate of a reaction. The movement of a particle up an energy gradient is described as ratcheting. It involves capturing the transitory positional displacement of a particle through the imposition of a kinetic energy barrier. Escapement is the counterpart to ratcheting in which the lowering of a kinetic energy barrier allows a ratcheted substrate to relax from a statistically unbalanced system toward a thermodynamic sink. 2 Controlling Motion in Covalently Bonded Molecular Systems 2.1 Controlling Conformational Changes 2.1.1 Correlated Motion through Nonbonded Interactions At room temperature, organic compounds typically fluctuate between rapidly equilibrating stereoisomeric structures through the perturbation (rotation, lengthening, and shortening) of covalent bonds. The relative stability and energy barrier to interconversion of configurational and conformational isomers depend on intramolecular and intermolecular interactions. The rotational and vibrational freedom of a single substituent of a molecule is not only determined by its connectivity, shape, and size, but also its interactions with neighboring substituents. 130−133 For example, Mislow et al. demonstrated the steric influence of neighboring groups on rotation as part of their work on molecular propellers such as compound 1 where two or more aryl rings are attached to a single atomic center (Z) (Figure 8 ). The rotation of one aryl ring about a C–Z axis is sensed by the other rings. Mislow suggested that the motion of the rings is coupled in the sense that none of the rings move independently of the others and described such correlated motion as “sympathic”. 130,134−138 Figure 8 Chemical structure of a triaryl molecule as an example of a molecular propeller. 130,139 Öki’s group conducted pioneering investigations into restricted rotation about sterically hindered single bonds in molecules such as 9-arylfluorenes and triptycenes. 140−149 These molecules showed very high rotational barriers, and, in some cases, it was possible to isolate the rotational isomers (rotamers). Inspired by this work, Mislow and Iwamura simultaneously introduced the concept of dynamic gearing of proximate substituents in crowded molecules in molecular analogues of propellers and bevel gears. 150−167 The first molecular bevel gear consisted of two 9-triptycyl groups joined through their bridge head carbons to a central atom (Figure 9 ). Experimental and theoretical studies confirmed that the triptycyl groups are tightly intermeshed and undergo correlated disrotatory motion. 150,156 Figure 9 (a) Molecular bevel gear 2, consisting of two 9-triptycyl groups joined through a bridge head carbon to a central atom. 150 (b) X-ray structure of the molecular bevel gear (side view). (c) X-ray structure of the molecular bevel gear (top view). Adapted with permission from ref (156). Copyright 1984 American Chemical Society. Many other examples of dynamic gearing have since been published including a range of triptycene derivatives, 168−174 metallocene-based gears, 175,176 conformationally restricted amides, 177−181 and a series of propeller molecules and other molecules with selective rotation, 132,133,139,182−194 such as 1,2-di-o-tolylnaphthalene moieties. 195,196 The sandwich-shaped trinuclear silver(I) complex [Ag3(3)2] (Figure 10 ) developed by Shionoya and co-workers 197−199 is an example of a different type of a rotational device. The complex consists of two disk-shaped ligands, 3, consisting of three thiazolyl or 2-pyridyl ligands and three p-tolyl groups attached to a central benzene ring. The three p-tolyl groups force the neighboring ligands to adopt a nonplanar arrangement with respect to the central aromatic ring, and make the coordination sites more accessible for silver(I) cations. The exterior rings tilt at 30° in the same sense and thus form helical structures with P and M geometries. A change from P to M helicity, and vice versa, was induced by 120° relative rotation of the ligand in these complexes (Figure 10 ). These observations were further studied using a heterotopic system [Ag3(3)(4)] consisting of a hexa-monodentate ligand 4 and a tris-monodentate thiazolyl ligand 3, which complex three silver(I) cations. 200 In the case of the hexa-monodentate thiazolyl ligand 4, only every second ring is coordinated to a silver cation. Correlated flipping motion of the coordinating rings (and ligand exchange) results in the conversion from M to P and a 60° rotation of the two disks (Figure 10 d). Figure 10 (a) Chemical structure of tris-monodentate disk shaped ligand 3 and hexakis-monodentate ligand 4. (b) Schematic representation of complex [Ag3(3)2] shown as its M-helical enantiomer. (c) X-ray structure of complex [Ag3(3)2]. 197 (d) Schematic representation of the flipping motion of the rings attached to the disks and subsequent ligand exchange from 1-A, 3-E, and 5-C in M to 1-B, 3-F, and 5-D in P. The direction of rotation in the M to P transition is opposite to that of a subsequent P to M′ transition. 200 Reprinted with permission from refs (197) and (200). Copyright 2003 and 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Related heterotopic complexes were later used to transform the rotational motion of the sandwich-shaped trinuclear silver(I) complex into translational motion using a molecular crank mechanism. In molecular crank 5, a [2]rotaxane was attached as a translational segment (Figure 11 ). 201 Figure 11 (a) Schematic illustration of a crank mechanism that translates linear motion into rotary motion in cylinder engines. (b) Schematic representation of a molecular crank mechanism. (c) Chemical structure of a synthetic molecular crank 5. 201 Reprinted with permission from ref (201). Copyright 2010 Royal Society of Chemistry. Recently, Anderson et al. reported a zinc-porphyrin macrocycle coordinated by two templates containing multiple pyridines (wheels) forming a caterpillar track complex. 202 NMR exchange spectroscopy (EXSY) experiments showed that the ring underwent correlated motion with correlated motion of the templating “wheels”. The examples shown in this section clearly show the influence that steric interactions can have on submolecular motion. However, at equilibrium, these motions are nondirectional, even during a partial rotation. 2.1.2 Stimuli-Induced Control of Conformation about a Single Bond An early example of controlled random rotary motion about a single bond was provided by Kelly et al., 203−205 who utilized a cation binding event to halt the free rotation of a triptycene moiety in “molecular brake” 6 (Scheme 1 ). The two pyridine groups must be coplanar to maximize binding with Hg2+, thus raising the energy barrier to rotation by “putting a stick in the spokes”. Sulfur oxidation has similarly been used to inhibit free rotation in 7 with mono- or dioxidation significantly slowing the rotation. 206 Rotation about a single bond has also been promoted by substrate protonation by Shimizu and co-workers in 8. 207 Protonation of the quinoline nitrogen led to an increase in the rate of rotation by 7 orders of magnitude, through a proposed hydrogen-bonding stabilization of the rotation transition state (Scheme 1 ). The same group observed that utilizing an acetate guest’s hydrogen-bonding interactions to “turn on” rotation provided a more modest accelerating effect. 208 Again, this was suggested to be via a lowering of the energy of the rotational transition state. Other examples have also been reported. 209,210 Scheme 1 Molecular Brakes Operated by (a) Hg2+ Binding and (b) Sulfur Oxidation; and (c) Proposed Transition State Stabilization in Shimizu’s System 203,206,207 Singlet oxygen has been used to promote rotation in combination with thermal relaxation in a system based on rotation between the cis and trans isomers of 9 (Scheme 2 ). Reaction with singlet oxygen forms the cis-substituted 9,10-endoperoxide selectively. Thermal reversion generates the cis-anthracene, which upon extensive heating furnishes the more thermodynamically stable trans-anthracene 9, completing the cycle. 211 Scheme 2 A Molecular Switch Flipped by Singlet Oxygen 211 Kelly and co-workers reported an attempt to create a Feynman adiabatic ratchet-and-pawl. 212−214 The design utilized a helicene pawl whose inherent helical chirality was intended to bias the rotation of the triptycene cog. Although this design realizes an asymmetric potential energy surface, 1H NMR showed no directional bias with equal rates of rotation in both directions. This is in line with Feynman’s thought experiment 215 and illustrates that the rate of rotation depends solely on transition state energy and temperature, and not on the shape of the barrier to rotation. State functions, for example, free energy, do not depend on the system’s history (Figure 12 ). Figure 12 (a) Kelly’s ratchet-and-pawl 10. (b) Enthalpy change on rotation of helicene. Reprinted with permission from ref (212). Copyright 1997 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. The key factor required to break detailed balance and allow directional motion of the triptycene cog is an energy input to drive the system away from equilibrium. Although a rapid periodic variation in temperature could in theory cause the system to act as a temperature ratchet, it would be very difficult to verify this experimentally. As such, Kelly et al. instead proposed the modified structure 11, 216,217 where a chemical reaction drives the system from equilibrium (Scheme 3 ). The helicene pawl oscillates in the energy minima between blades, and, ignoring the amine group, all three minima are identical. Formation of the isocyanate from reaction with phosgene activates the system. In the course of random fluctuations, sometimes the helicene is close enough to the isocyanate to react. After reaction, the helicene is trapped part way up the energy barrier; that is, the helicene is “ratcheted” part way to rotation. Random thermal fluctuations may then overcome the smaller barrier to rotation, and the urethane can be cleaved to give overall 120° rotation. This machine is a landmark achievement in the realization of chemically fueled directed rotation in molecular machines and demonstrates most of the basic tenants of a directional motor. However, attempts to modify the rotor to allow 360° rotation have thus far failed. 218 Scheme 3 Chemically Driven Directional 120° Rotation of Kelly’s Triptycene Rotor 11(216) Mock and Ochwat have proposed a minimal molecular motor in which epimerization of a stereogenic center, that is, a formal 180° rotation, is driven by the hydration of a ketenimine fuel. 219 A biaryl lactone motif reported by Branchaud has been used to control motion about a single bond in a nondirectional rotor based on the facially selective ring opening of a chiral lactone. 220 They later improved on this system to report 180° directional rotation. 221 Feringa et al. achieved full 360° directional motion about a single bond in a related biaryl system (Scheme 4 ). 222,223 Rotation was driven by four sequential chemical “power strokes” with felicity of rotation for each step being between 90% and 100%. The cycle involved four conformationally restricted intermediates, 12–15, with 12 and 14 locked by covalent bonds, and 13 and 15 locked by their steric bulk. Although 13 and 15 are free to dynamically equilibrate between helical forms, the chiral information in the reducing agent is used to discriminate between them and thus drive directional motion. The first step involved asymmetric ring opening with the (S)-CBS reagent, and regioselective protection of the resulting phenol with an allyl group. After oxidation and PMB deprotection with spontaneous lactonization, 180° rotation was achieved. Another stereoselective ring opening and regioselective phenol protection as the PMB ether followed, before oxidation and ring closure furnished the product of 360° rotation in high fidelity. Additionally, rotation in the opposite sense can be achieved by utilizing the (R)-CBS reagent and reversing the order of PMB/allyl protection. Scheme 4 Directional 360° Rotation about a Single Bond via Four States A–D Rotation is restricted in A and C by covalent bonds, although helical inversion is allowed, and in B and D by nonbonded interactions, with directional control of rotation being provided by stereospecific covalent bond cleavage. 222 2.1.3 Stimuli-Induced Conformational Control in Organometallic Systems Control of the rotary motion of ligands in metal sandwich or double-decker complexes is somewhat conceptually similar to controlling rotation around covalent single bonds. Aida and co-workers demonstrated that porphyrin ligands in double decker complexes such as 16 can rotate with respect to each other depending on the central metal atom and their steric bulk (Figure 13 a). 224,225 Studies on a variety of metal complexes found that the central metal had an influence on the speed of rotation. 226 Complexes synthesized from two D 2h symmetry porphyrin free-bases are chiral, and therefore their rotary motion corresponds to enantiomerization and can thus be studied by CD spectroscopy. Racemization can be induced by protonation or reduction of the metal center. The second reduces π–π interactions between the two ligands as a consequence of the larger ionic radius of the metal center. 227 Figure 13 (a) X-ray crystal structure of porphyrin double decker complex 16 with cerium(IV). Reprinted with permission from ref (228). Copyright 1989 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. (b) Representation of the cooperative binding of guest molecules to porphyrin double decker complex 17. 229 It was shown that the rotation of the cerium(IV) bis[tetrakis(4-pyridyl)porphyrinate] 17 could be suppressed by successive cooperative guest binding of dicarboxylic acids (Figure 13 b). 229,230 The double decker complex shows a large allosteric effect in this molecular recognition event (once rotation has been suppressed by the first guest molecule, guest binding becomes increasingly favorable). Further guest molecules have been tested such as β-dicarboxylate anions, 231 potassium ions, 232 mono- and oligosaccharides, 233−235 and silver cations. 236 In the case of silver ions, it was shown that their cooperative binding leads to a progressive and nonlinear increase in the rate of rotation. 237−244 The rotation of a similar cerium double decker complex has been monitored on a single molecule level, by the attachment of a bead visible under optical microscopy. 245 Ever more complicated systems such as multidecker pophyrin complexes 246 and mechanically interlocked porphyrin systems like 18 have been synthesized (Figure 14 ). 247−249 The latter consist of cerium(IV) or lanthanum(III) bis(porphyrinate) double decker complexes and one or two orthogonal porphyrin molecules. In these “molecular gears”, the rotation of the top and side units could be partially coupled through mechanical interactions between the different units. A folding ruler based on similar principles has been reported. 250 A dodecanuclear four level complex that mimics a double ball bearing has also been synthesized. 251 Figure 14 Cerium(IV) bis(porphyrinate) double decker (top unit) and a rhodium(III) porphyrin-based side cog. The two units are connected through a coordination bond between rhodium(III) and a pyridyl group. 249 Ferrocene is a prototypical sandwich complex with an iron atom symmetrically aligned between two C5H5 (cyclopentadienyl) rings. 252,253 The barrier to rotation of the two rings about the C5 axis is very small in the gas phase. Using electrospray ionization and theoretical studies, it was shown that ferrocene derivative 19 2– can serve as a two-state rotary switch, which works through proton transfer (Scheme 5 ). Differing electrostatic interactions lock 19 2– and [19·H]− in two different conformations. The trans conformation of 19 2– is more stable because of Coulombic repulsion between the negative charges, while the cis conformation of [19·H]− is stabilized through hydrogen bonding (Scheme 5 ). Scheme 5 Control of Rotation in Ferrocene Complex 19 through Protonation 253 This finding was followed by studies of many more ferrocene derivatives, and the control of their rotary motion through different stimuli such as electron transfer and photochemistry. 254,255 Ferrocene-based rotors have recently been used to form self-assembled nanostructures. 256 Metallocarboranes tend to have rather high barriers to rotation around the metal–ligand axis, but some complexes, for example, nickel complex [Ni(20)2], can be used as electrochemically controlled rotary switches (Figure 15 ). 257−261 Figure 15 (a) Dicarbollide ligand 20 2–. (b) Metallocarborane [Ni(20)2], an electrochemically controlled rotary switch. 261 2.1.4 Stimuli-Induced Conformational Control over Several Covalent Bonds In biological systems, host–guest binding is often used to induce conformational and functional changes. These changes can vary from small, localized, bond deformations to long-range rearrangements across multiple covalent bonds. These deformations are frequently used to alter binding at distal sites, that is, allosteric regulation. There are several synthetic examples where host–guest binding causes a sufficiently large amplitude change to merit discussion in this Review. Some of the earliest examples of allosteric receptors came from Rebek and co-workers. 262 For example, the binding of tungsten by negative allosteric receptor 21 reduces the receptor’s affinity for potassium ions, as in this more rigid conformation binding to potassium would induce a degree of strain (Scheme 6 ). 263,264 The positive allosteric receptor 22 developed by the same group depends on the initial mercury binding event preorganizing the linked receptor for a second, more favorable binding event. 265 These early examples have sired an extraordinary and diverse array of synthetic allosteric receptors. 262,264,266−279 Scheme 6 (a) Negative Allosteric Receptor 21, Where Binding of Tungsten Forces Ring Contraction and Steric Clash of the 3- and 3′-Substituents of the Bipyridyl Unit; and (b) Positive Allosteric Receptor 22(263,264) Organic guests can also cause large-scale changes to host molecules, as they maximize favorable interactions such as in molecular tweezer 23 (Figure 16 ) where the distance between sidewalls halves on binding certain guests. 280,281 A large amplitude change was observed in the flexible system 24 with an extended conformation being exchanged for a more rigid “sandwich” form to maximize guest binding. 282 Guest-induced change is also observed in clip 25 where the sa form predominates in solution in the absence of a guest, but the aa form dominates in the presence of a suitable guest (Figure 16 ). 283−285 This system has been further modified in 26 where the binding of a potassium ion to each crown ether preorganizes the system for organic guest binding with increased affinity. 286 Figure 16 (a) Molecular tweezer 23, arrows indicate direction of contraction. (b) Clip 25, where the aa form is favored in the presence of a guest. (c) Clip 26, where ion binding enhances guest affinity. (d) Crystal structure of uncomplexed and complexed 24. 282,283,286,287 X-ray crystal structure reprinted with permission from ref (282). Copyright 2007 American Chemical Society. Metal coordination has been used by Lehn et al. to switch from an arrangement where parallel anthracene units can bind electron-poor 2,4,7-trinitro-9-fluorenone (TNF) in a tweezer-like manner, to an open form 27 (Scheme 7 ). Here, coordination of copper in a bidentate manner by the aromatic nitrogens causes a rotation about the heterocyclic axis and opens the guest binding cavity, preventing TNF coordination. In related 28, zinc coordination is a prerequisite of TNF binding. 288 Scheme 7 (a) Prevention of Organic Guest Binding on Complexation of Copper to 27; and (b) Organic Guest Is Only Bound in the Zinc-Complexed Form of 28(288) A notable example of chelation control of conformation was provided by trisaccharide 29 (Scheme 8 ). Although the tetra-equatorial isomer is the dominant form in solution, 289 the addition of Pt(II) led to coordination, and a preference for the tetra-axial form, the better to bind the metal ion. This process could be reversed by removal of the Pt(II) with excess NaCN. 290,291 This process has been replicated with a variety of cyclic substrates using both metal ion binding 292−294 and pH variation. 295−297 Sauvage reported the collapse of a porphyrin containing [4] rotaxane on copper removal. 298 Haberhauer’s molecular hinge, 30, based on a 2,2′-bipyridine motif, is another example of large amplitude motion controlled by coordination of a metal ion. In this case, the open system was free to rotate over 180° with its rotation being constrained by an inflexible backbone. Upon coordination of Cu(II), the hinge shut. Upon copper removal, the hinge was again able to open, but only in one direction due to the constraints of the backbone. 299 A light-driven switch and a chirality pendulum based on a similar system have been reported by the same group. 300,301 Scheme 8 Chelation Control of Equatorial/Axial Conformers and Haberhauer’s Molecular Hinge 290,299 , Reprinted with permission from ref (299). Copyright 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Cavitands derived from resorcin[4]arenes have been shown to undergo large conformational changes between open “kite” and closed “vase” forms of 31 (Scheme 9 ). 302 Because of solvation effects, the kite form is normally favored at lower temperatures and the vase at higher. 302−304 However, this change has also been achieved using protonation of or metal coordination to the quinoxaline nitrogens, 305−308 or by altering the design to incorporate redox-active centers. 309,310 Substitution of the cavitand with amide groups can lead to stabilization of the vase form by dimerization or guest complexation. 311−320 As an alternative to coordination of a guest causing a change in conformation, complexation can be intramolecular as in macrocycle 32 reported by Stoddart et al. The system rests in the self-complexed form until reduction of the cyclophane decreases subcomponent interactions, followed by release of the system to an unrestrained, unencapsulated form. 321−326 A similar system, 33, was realized by Feringa and Qu 327 with directional rotation inhibited in the complexed state, but allowed when uncomplexed (Scheme 9 ). Scheme 9 Resorcin[4]arene 31, Vase and Kite Conformers, Stoddart’s Self-Complexing Lock 32, and Feringa’s Self-Complexing Rotor 33(98,321,322,327), Reprinted with permission from refs (321) and (327). Copyright 1997 and 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Hamilton and co-workers introduced conformational switch 34 (Scheme 10 ) where in the initial state the equilibrium is mildly biased toward the benzamide station (1.3:1). Upon protonation of the dimethylamine with TFA, a dramatic reversal of this bias is observed (1:99). 328 The oxidation of a copper center has been used to trigger coupled rotation and copper exchange between two independent switches (based on the structure shown in Scheme 45 ). 329 Scheme 10 Hamilton’s Acid Sensitive Switch 328 Scheme 11 pH-Driven Conformational Change in Oligomeric System 35(330) Oligomeric systems have been used to great effect in generating large-scale conformational changes, often exploiting the helical secondary structures that can be formed by many species. 330−367 Polyamide 35 resides in a helical arrangement, but protonation of four diaminopyridine units causes a dramatic extension to the linear form of 35 (Scheme 11 ). Further protonation leads to another helical form, and the system can be returned to its initial form by deprotonation. 330 An extension of this system was shown to expand from 12.5 to 57 Å upon protonation. 368 An alternating pyridine and pyrimidine oligomer formed helices (due to the favored disposition of neighboring rings being transoid), but upon addition of Pb(II) to 36, coordination led to a linear form with a concurrent increase in length from 7.5 to 38 Å (Scheme 12 ). This process could be made switchable by sequestering the Pb(II) until needed in pH responsive cryptand 37. 369 A similar process has been achieved using Ag(I) to modulate expansion. 370 Scheme 12 Large-Scale Extension of a Ligand Strand upon Pb(II) Complexation 369,370 An interesting recent example was provided by Stadler and Lehn 371 where a linear and a helical domain were synthesized in the same molecule. In this two-domain system, coordination of Pb(II) led to the unfurling of the helical domain with concurrent curling of the linear portion of the molecule. Upon removal of Pb(II), the system reverted to its original state, thus representing a reversible device where an extension process is coupled to a furling one, on both complexation and decomplexation of a metal ion. 2.2 Controlling Configurational Changes Changes in configuration, especially cis/trans isomerization, have been used in many molecular systems to control motion. 372 Although the sometimes small amplitude motion is not always useful for the design of molecular machines, it represents an interesting tool for inducing directional motion in synthetic and supramolecular systems. The most prominent examples of this type of system are the photoisomerization of stilbenes and azobenzenes, 373−377 the reversible electrocyclization of diarylethenes, 378−382 the photochromic reactions of fulgides, 383,384 the interconversion of spiropyrans with merocyanin, 385 as well as chiroptical switching of overcrowded alkenes. 386−388 In recent years, this topic has become an extensive area of research, and a wide range of applications exploiting switchable systems have been proposed and realized. 389−395 Most of these applications rely on intrinsic electronic and spectroscopic changes on interconversion between the two species, but some use small configurational changes in a more mechanical fashion, and the resulting devices can be seen as molecular machines. Configurational changes, especially the photoisomerization of azobenzenes, have been used to induce changes in the structures of biologically relevant molecules such as antibiotics, peptides, and DNA, 374,396−406 but also as part of small-molecule systems such as molecular scissors, tweezers, and other molecular machines. 407−412 Some of them will be introduced in the following section. Scheme 13 (a) Molecular Scissors 38; and (b) Structure of Molecular Hinge 39(414) Irradiation at 350 nm initiates photoisomerization from the trans to cis isomer of the azobenzene moiety (closed to open molecular scissor); irradiation at >400 nm induces the reverse. 413 Molecular scissors 38, described by Aida and co-workers, consist of an azobenzene unit and a ferrocene unit (Scheme 13 a). 413,415 The open and closed forms were interconverted by photoisomerization of the double bond of the azobenzene unit, which led to an angular change of 49° around the cyclopentadienyl rings of the ferrocene. Another example was the synthesis of the molecular hinge 39 where two xanthene rings were connected by −N=N– linkers (Scheme 13 b). 414 Photoisomerization at 366 nm from the trans/trans, which is almost planar, to the cis/cis state resulted in a change of approximately 90° in the angle between the two aromatic rings. Reisomerization could be achieved by irradiation at 436 nm. Azobenzenes have been used for the construction of photoswitchable azo-macrocycles. 389,416 Azo-macrocycles have found applications in host–guest chemistry as they can selectively and reversibly bind ions. Cyclic azobenzenes have been used to switch on or off the rotation of subunits. 417,418 A recent example showed that the rotation of a 2,5-dimethylbenzene rotor in the cyclic azobenzene 40 could be switched off in the E-isomer of the azobenzene, while rotation was allowed in the Z-isomer (Figure 17 ). Figure 17 (a) Chemical structure of a molecular brake 40 with a 2,5-dimethylbenzene unit as the rotor. (b) X-ray crystal structure. (c) Schematic representation of this photoinduced molecular brake. X-ray crystal structure reprinted with permission from ref (418). Copyright 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Azobenzenes have also been used to control motion in molecular shuttles (see also section 4.3 ) and to control the threading and dethreading rate of a macrocycle when attached to the ends of a thread. In a recently published example, E → Z photoisomerization of azo-end groups slowed the threading–dethreading of a ring and turned the pseudorotaxane into a kinetically inert rotaxane. 419 The substitution of an alkene can lead to steric overcrowding around a double bond. When the substituents are large enough, the planarity of the double bond can be disrupted and a variety of twisted and folded conformations can be adopted, including helices. The area where the substituents come into close proximity is often referred to as the fjord region. These types of compounds still undergo photoisomerization and can be used as chiroptical molecular switches. 386,420,421 Feringa et al. have developed several molecular motors exploiting overcrowded alkenes as chiroptical molecular switches. 422−428 The first generation of this molecular motor (Scheme 14 , compound 41) featured two identical halves connected by a central double bond. 429−436 Each of the halves had a stereocenter, which was crucial to controlling the rotational process. Rotation of 360° around the double bond was repetitive and directional. Two light induced cis–trans isomerizations led to a 180° rotation and were each followed by thermally controlled helicity inversion, which effectively blocked reverse rotation. The cis–trans photoisomerization was evident from 1H NMR, and the simultaneous exchange of P to M was detected by circular dichroism (CD). A number of different structures have been synthesized to optimize the rotation process and to explore the limits of the system. 437−444 Scheme 14 Minimization of Steric Interactions between Aromatic Groups and Substituent Results in (P,P)-(trans)-41-Stable Ground-State Conformation, with Axial Substituents (i) Photochemical isomerization leaves substituents in the unstable equatorial conformation (M,M)-(cis)-41-unstable. (ii) Steric strain is released by a thermally activated helicity inversion leading to (P,P)-(cis)-41-stable. (iii) Photoisomerization from cis to trans. (iv) Helical inversion completes a full rotation. 429 In the second generation of Feringa’s molecular motor 42, two halves that are not identical were connected by a central double bond. 445−447 One half was replaced by a tricyclic aromatic group and there was only one stereogenic center present, but the rotation operated according to the same principles as the first generation motor 41 (Figure 18 ). Some dramatic enhancements of rotary speed were achieved in the second generation of motors. 428,448−453 Figure 18 Comparison of the general structures for the first (41) and second (42) generations of Feringa’s molecular motors. 425 Feringa and co-workers have shown that transmission of molecular rotation from one part of a molecule to another is possible. In motor 43, a xylyl unit was attached to the lower half of an overcrowded alkene-based switch. By switching between the cis- and trans-forms, the rotation of the xylyl group could be controlled (Figure 19 ). 454,455 Imines have recently been used as directional light-driven rotors by Lehn and Greb. 456 Figure 19 Structure of molecular brake 43. 454 It has long been known that irradiation of hydrazones can induce E → Z isomerization around the imine bond (C=N). However, the Z-isomer is usually thermodynamically unstable and thus short-lived. 457,458 Intramolecular hydrogen bonding in various hydrazone derivatives can kinetically stabilize the Z-isomer, which is formed by irradiation. 459−461 The Aprahamian group has studied the E/Z isomerization of hydrazones using chemical switching inputs such as protons and zinc ions. 462−467 Scheme 15 Hydrazone-Based Switches (a) Protonation of the pyridyl nitrogen of compound (E)-44 induces E → Z isomerization. 468 (b) The quinoline unit in (E)-45 provides a binding site for a Zn2+ ion, which is now the trigger for E → Z isomerization. 469 In their original system, E→ Z isomerization was induced by protonation of a pyridyl ring that initially acts as an intramolecular hydrogen-bond acceptor in the E-isomer (Scheme 15 , compound (E)-44). 468,470 In the Z-isomer, the carbonyl oxygen stabilizes the Z-H+ form of the molecule because the pyridyl unit is no longer available as a hydrogen-bond acceptor. Upon deprotonation, the process is reversed. However, the Z-isomer is metastable, and as a consequence the removal of the proton does not lead to an immediate change. The thermodynamically more stable E-isomer reappears with time. The Aprahamian group has also replaced the naphthalene unit with a quinoline unit (see (E)-45), thus introducing an additional hydrogen-bond acceptor and an additional metal ion binding pocket next to the hydrazone. 469 In this system, a Zn2+ ion now triggers E → Z isomerization. The most recent system showed that on addition of Zn2+ to (E)-46 and (E)-47, a switching cascade could be initiated. 471 Coordination of Zn2+ to (E)-46 lowers the pK a of the imidazole N–H by 3 units and allows the protonation of (E)-47, which in turn allows switching of this moiety. Without metal coordination, the imidazole is not acidic enough to catalyze switching in (E)-47, making this sequence an elegant example of one switching event acting as the input for another, in one pot. Because the stator and rotor are held in fixed relative positions, the switches developed by Aprahamian and co-workers represent very promising tools for the control of directional motions in molecular machines (Scheme 16 ). A hydrazone switch has been used to dope a liquid crystal with the color of the liquid changing upon switching. 472 Scheme 16 Switching Cascade of Hydrazone Rotary Switches 471 3 Controlling Motion in Supramolecular Systems The use of supramolecular systems in molecular machines is challenging as competing processes such as the association and dissociation of subcomponents from and to the bulk must be controlled. 473,474 These processes can interfere with the integrity and/or processivity of the machine unless the binding is part of the motion generation process itself or the effect of unbound elements is restricted. If host–guest interactions are used to assemble a machine, then exchange with free hosts or guests in solution should be avoided to maintain the integrity and kinetic stability of the machine. Disassembly or exchange can, in certain cases, be intentionally exploited for functional purposes and to provide communication with external species. Controlled motion in supramolecular systems must be built on cooperative interactions of individually weak but collectively strong noncovalent interactions. In biology, large-scale molecular motion generated upon guest binding is a common process. Acetylcholine neurotransmitters are among such guests and bind to a transmembrane channel receptor, inducing a large conformational change, resulting in the controlled flux of ions through the membrane. 475 In this case, any guests in the bulk can act as an inducer of molecular motion and at the same time as a messenger of the microenvironment enabling dynamic communication with the surroundings. The stabilities of most biological functional assemblies are well maintained by restriction of bulk exchange. If this was not so, rapid exchange of the ribosome complex or individual subunits of the ribosome during protein synthesis would detach the enzyme from the substrate mRNA and release a partially synthesized peptide, unlikely to be of any use. However, mRNA and the ribosome form a kinetically stable complex, subunit exchange with the bulk is restricted, and the enzyme processively produces large proteins. 476 3.1 Guest Binding Generating, or Assisted by, Large-Scale Intramolecular Motion Considering the above discussion, molecular machines and devices based on noncovalent assemblies require careful design. The relatively large molecular level motions generated upon guest binding are of great importance in the design of practical molecular machines. Guest-induced molecular motion has been widely studied in the literature with organic compound, 477 anion, 478 cation, 479−481 metal, 464,471,482,483 or proton guests. 484,485 Switches based on pseudorotaxanes have also been reported. 486−518 These supramolecular complexes display interesting operational properties that are useful for device development. 495,519−564 These switches can be used to build relatively sophisticated devices such as the molecular tweezers 23–26. 280,281,287,288,565−575 This concept conventionally refers to receptors having a cavity with two binding motifs. The binding pocket can either readily encapsulate the guest (lock and key type) or evolve to a more suitable conformation upon guest binding (induced fit type). Molecular machines can be regulated by responsive tweezers, with motion induced and regulated through an external stimulus such as the binding of guest molecules. Scheme 17 Light-Operated Molecular Tweezers 575 One of the earliest examples of a molecular machine was a “molecular tweezer” based on the photoisomerization-dependent binding of cations in a crown ether-bearing azobenzene, developed by Shinkai et al. (Scheme 17 ). 575,576 The open trans form can bind smaller cations such as Na+ selectively, whereas the cis form prefers to bind larger Rb+ ions by forming a sandwich complex with the cation. Moreover, the cis isomer can be stabilized by the sandwich complex in the presence of Rb+, and the photostationary state biased toward the cis isomer, causing the molecule to relax to the trans isomer much more slowly. Hence, in addition to photoisomerization-dependent changes in the affinity of the receptor, guest binding can alter the kinetics of the photoisomerization process. Scheme 18 Decreased Drug Binding Affinity by a pH-Induced Conformational Change 573 Intercalation of electron-poor aromatic guests by molecular tweezers having two π-systems has also been explored by several groups. 577−581 The binding and release of the guest is modulated by the motion of the receptor components. Intramolecular motion has been induced allosterically by means of various stimuli including metals, 582−584 anions, 585,586 electrochemical reduction, 587 and pH change. 573 2,6-Substituted pyridine 49 has been shown to change conformation upon protonation of the pyridine, such that the methoxy groups form hydrogen bonds with the pyridinium proton (Scheme 18 ). 573 This hydrogen bonding results in a rotation around the single bond moving the naphthyl moieties away from each other. Intramolecular motion results in the release of the hydrophobic drug mitoxanthron, which had been bound by π–π interactions between two naphthyl receptors. Figure 20 Orthogonal exclusion of an azobenzene and a viologen encapsulated by CB[8]. 588 Extracting work from machines depending solely on supramolecular interactions would be difficult due to the rapid exchange of bound and unbound guests. However, some supramolecular host–guest systems are moderately stable and might be suitable for use in the assembly of molecular devices. A methylviologen (compound 50, MV2+) and a trans-azobenzene derivative (compound 51, trans-AB) were encapsulated by a cucurbit[8]uril (CB[8]) host (Figure 20 ). Each guest has distinct properties: MV2+ is redox active, and the azobenzene can be isomerized to its cis-form (cis-AB) on irradiation. 588 One electron reduction of methylviologen (MV+•) results in a binary encapsulation of the reduced species in the host and trans-AB is kicked out. Photoisomerization of azobenzene to its cis-isomer results in its ejection due to steric repulsion. Therefore, the initial heteroternary complex of MV2+, trans-AB, and CB[8] can be orthogonally switched to give two distinct complexes. In one complex, one-electron reduction of MV2+ forces out the azobenzene to form a homoternary complex (MV+•)2CB[8], while in the other photoisomerization of the azobenzene ejects cis-AB, while MV2+ stays encapsulated to form (MV2+)CB[8]. These three distinct, stimuli-dependent, modes of complexation allow greater control over molecular behavior. To transfer this to the macroscopic world, thiol-functionalized trans-azobenzenes were attached to a gold surface. Upon complexation by CB[8], together with the fluorescently labeled MV2+, the surface became fluorescent, and due to the charged nature of viologen, a substantial increase in wettability was observed. Upon irradiation, the azobenzene was kicked out as its cis isomer, returning the surface to a nonfluorescent, hydrophobic state, after the unbound fluorescent viologen–CB[8] complex was washed away from the surface. The orthogonal control of switching processes is a vital tool for the creation of even more complex molecular devices, especially in memory device construction. 588 One of the interesting emerging properties of supramolecular host–guest systems is the stimuli-dependent expansion and contraction of molecular constructs. This behavior will be revisited in rotaxane systems in section 8 . In double-stranded helicate 52 (bridged by spiroborates at each end), sodium ion-induced reversible extension and contraction was observed with an accompanying directional twisting (Figure 21 ). 589 In the presence of Na+, the molecule was forced to twist its central tetraphenol moiety to better coordinate the cation. The handedness of the helicate was preserved during this process. This process could be reversed by the sequestration of Na+ by [2.2.1]cryptand. Figure 21 Na+-induced twisting in helicate 52. Reprinted with permission from ref (589). Copyright 2010 Nature Publishing Group. The tetraphenolic central compartment could be replaced by two porphyrins to create a hydrophobic cavity for aromatic guests to stack inside. 590 Upon intercalation of electron-deficient aromatic guests between the porphyrin rings, the distance between the two porphyrin units increased from 4.1 to 6.8 Å. This expansion in the central region induced a rotation, which reduced the torsion angle, and, due to steric constraints, the capping spiroborates unwound in a directional manner, demonstrating the first guest-induced rotary motion with accompanying corkscrewing motion. Guest control of screw motion, involving simultaneous linear translation and rotation, has been reported. 591 In this case, the guest acted as a template for the formation of helicates based on hydrogen bonding and biased the formation of one of the two equilibrium structures. Stimulus-driven molecular level conformational change and subsequent control over the binding affinity of guest molecules have been used to regulate anion concentration in solution. 592−594 Flood et al. developed a triazole-based foldamer bearing an azobenzene moiety in the structure (Figure 22 ). 595 Hydrogen bonding between the triazole units and a chloride anion together with π-stacking within the foldamer backbone kept the anion buried in the interior. 592,596−602 Photoisomerization of the azobenzene moieties disrupted the stacking, inducing unfolding and releasing the anions. This photoinduced release and capture of anions is completely reversible. Calculated K values in acetonitrile decrease significantly from 3000 to 380 M–1 upon irradiation with UV-light. In addition, the conductivity of the irradiated sample increased due to release of anions into solution. Helical inversion caused by ion binding has also been observed. 603 Stimuli responsive switches have great potential in drug release and ion-cargo transport applications. Nitschke et al. have reported a complex system, where the addition of various signal molecules to a helical complex led to a cascade of changes in the self-assembled system and various product distributions could be formed. 604 Figure 22 Control of anion concentration in solution by a photoresponsive foldamer. Reprinted with permission from ref (595). Copyright 2010 American Chemical Society. In the above examples, binding properties are regulated by an external stimulus or by allosteric control. Large-scale molecular motions regulated by guest binding are of particular interest in molecular machine design, because the presence of the regulator acts as a chemical communication between the environment and the machine. This dynamic interaction would allow the development of smart molecular machinery that performs advanced operations dependent on the chemical environment, or perhaps allows communication between machines. 3.2 Intramolecular Ion Translocation The molecular basis for the conversion (in archea) of photon energy into more useful potential energy relies on proton translocation between a photoisomerized retinal molecule and the amino acid residues on the protein scaffold in which it is buried. This process occurs with a well-defined directionality, leading to light-mediated directional proton pumping across a membrane. 384,605 In supramolecular systems, for accurate translocation to take place, the ion to be translocated should be more kinetically available for intramolecular motion than the bulk ion. An electrochemical gradient cannot be achieved if each ion acceptor takes ions from the bulk. Ion translocation can be controlled by changing the affinity of organic receptors or metal centers for the ion of interest, through external stimuli such as light or reduction/oxidation. 606 In the case of proton transfer, some molecules have distinct photophysical properties such as a change in the pK a of the molecule in the excited state, which enables rapid dissociation of the proton from one part of the molecule and reassociation with another, intramolecular acceptor, in an “excited-state intramolecular proton transfer” (ESIPT). 607 This process has been extensively studied using 2-(2-hydroxyphenyl)-benzothiazole (HBT, Scheme 19 ). 608 Fast proton transfer from the hydroxyl group to the nearby thiazole nitrogen takes place upon excitation of the molecule to generate a cis-keto form with a distinct, solvent-dependent, spectroscopic character. In this molecule, intramolecular hydrogen bonding mediates this fast proton transfer between different regions of the molecule. Scheme 19 Excited-State Intramolecular Proton Transfer in HBT 607,608 Metals having more than one stable oxidation state can be translocated by oxidation or reduction or incorporated into a scaffold along with other ions that can be translocated. For either process, the molecule should bear more than one binding site, with different affinities, to drive the motion of the ion. Flexible receptors are most often used for ion translocation, but rigid aromatic systems can be suitable as well. These processes are usually accompanied by a visible color change due to changes in absorption after a change in the oxidation state of the metal, or due to ligand exchange. The reversible translocation of a chloride anion from a Cu(II) center to a Ni(III) center was an early example of a flexible system (Scheme 20 a). 609 The concentration-independent nature of the process indicated that anions translocate from the Cu(II) center in an intramolecular fashion rather than via exchange with the bulk. Scheme 20 Intramolecular Anion Translocation Mediated by (a) Redox 609 and (b) Ligand Exchange 610 Work by Mirkin et al. showed reversible intramolecular shuttling of a chloride anion from a bisurea binding site to Rh(I) in the presence of carbon monoxide (Scheme 20 b). 610 The hydrogen-bonding ability of the urea was further enhanced by attaching electron-withdrawing 3,5-bis(trifluoromethyl)benzyl groups. Rh(I) adopts a distorted square-planar geometry with these phosphine and thioether ligands, as shown by the crystal structure. Initially, chloride was encapsulated between the urea tweezers. With the aid of polar solvents (dimethyl sulfoxide or dimethylformamide) and in the presence of CO, chloride was translocated to Rh(I) with accompanying CO coordination. The Rh–S bonds were cleaved concurrently. Translocation was driven by both the weakening of hydrogen-bonding interactions in the urea and the more favored and less distorted geometry adopted at the Rh center with the new ligands. This is an important example of ion translocation in view of the fact that it involves two different types of interaction (hydrogen bonding and metal–ligand coordination) to enable shuttling and may open a path to the development of further molecular devices by the orthogonal control of these two interactions. The first intramolecular cation translocation driven by auxiliary redox reactions was reported by Shanzer et al. 611 A triple-stranded helical scaffold was used with ditopic hydroxamate and bpy ligands, which bind preferentially to Fe(III) and Fe(II), respectively (Scheme 21 ). Chemical reduction of Fe(III) with ascorbic acid relocated the cation to the bpy ligand with a visually observable color change from pale brown to violet-red. Reversibility was obtained by oxidation with (NH4)2S2O8. Even though the process takes place slowly (minutes to hours), the overall movement of the cation is intramolecular. Scheme 21 Redox-Mediated Intramolecular Ion Translocation 611 Several other examples of anion and cation translocation between ligands separated by flexible linkers have been investigated. 480,612−614 Metals are also known to translocate over rigid sp2 hybridized carbon surfaces. 615 Sandwich complexes of two dipalladium units (four palladium nucleii arranged in two separate regions) clustered between π-conjugated ligands showed redox-driven reversible translocation, resulting in a tetrapalladium complex (Scheme 22 ). 616 In a complex with further extended conjugation, redox reaction conditions resulted in reversible carbon–carbon formation between the dipalladium clusters. This barrier (the new C–C bond) built by the mobility driving force used in the first complex (redox reaction) prevented translocation in the second complex and hence confined the metals to their original positions. Scheme 22 Redox-Mediated Pd Translocation in an Aromatic Sandwich Complex 616 Supramolecular host–guest systems with distinct stimuli responsive switching behavior, guest sorting character, and emergent photophysical properties have been developed with implications for the design of molecular machines. To enable controlled mechanical motion, these systems should be kinetically stable enough to avoid exchange of components with the bulk on the time scale of the operation. Mechanically bonded molecular assemblies are viable candidates for machine architectures, considering their stable interlocked structure, and the relative ease of controlling their mobility. 4 Shuttling in Rotaxanes: Inherent Dynamics Motion in rotaxanes is constrained by the nature of the mechanical bond. Very limited motion is permitted orthogonal to the axle, while a “shuttling” motion “powered” by random Brownian motion can be observed along the axle (as bounded by large stoppers). As the templating methods typically used to form interlocked structures normally leave recognition motifs, 23,617−676 it is rare, although not unknown, to form a rotaxane without residual interactions between thread and macrocycle. 504,672,677−689 These residual interactions are typically viewed as “stations” on the rotaxane, and the shuttling of the macrocycle between these stations grants a useful handle in these molecules. The rates of shuttling between stations and their occupancy can be controlled by the strength of station–macrocycle interactions. 4.1 Shuttling in Degenerate, Two-Station, Molecular Shuttles The first two-station degenerate [2]rotaxane 61 was reported by Stoddart et al. and exhibited temperature-dependent shuttling between the two equivalent hydroquinol groups, as shown by 1H NMR. 690 Many rotaxanes based on similar macrocycles and stations have since been reported. 691,692 Similar processes are observed in the amide rotaxanes 62–64 where two diglycine units are separated by various linkers. In 62–64, rapid shuttling of the macrocycle between the degenerate stations was observed at 298 K. Only when the linker was replaced with a bulky N-tosyl group was shuttling inhibited (removal of the tosyl group restored rapid shuttling) (Scheme 23 ). 693 Scheme 23 (a) The First “Molecular Shuttle”, 61; 690 and (b) Degenerate Peptide-Based Molecular Shuttles 62–64 of Varying Linker Length 693 As movement between the binding sites must involve at least partial rupture of the hydrogen-bonding network, the migration can be represented by the simplified energy diagram in Figure 23 . Hydrogen-bonding solvents have been shown to disrupt macrocycle–thread interactions in single station rotaxanes, 694 and here addition of 5% [D4]methanol increased the rate of shuttling 100-fold, consistent with lowering the energy barrier to migration by disrupting station–macrocycle interactions and thus raising ground-state energies. 693,695−697 The effect of water on the rate of shuttling has been investigated and was found to be greatly superior to that of other protic solvents. 698 This effect was attributed to the ability of water to form three-dimensional hydrogen-bonding networks. Figure 23 Idealized model of binding in degenerate molecular shuttles. Barrier height is dependent on the energy required to break interactions between macrocycle and station and a distance-dependent diffusional component. 14 The use of a phenol/phenolate central group where the formation of a phenolate/cation pair significantly slowed shuttling 699 has also been explored. Cleavage of a nitrogen protecting group has been used to induce degenerate shuttling. 700 A large barrier to shuttling could be introduced in 65 where the dimerization of the rotaxane, as templated by Cu(I) addition, prevented shuttling (Scheme 24 ). This could be reversed by copper removal with an ion-exchange resin. 701,702 Berna and co-workers have recently reported a system in which the binding of a guest hydrogen-bond acceptor–donor–acceptor molecule to the thread led to the trapping of the macrocycle to the linker region between the two stations. 703 Upon removal, by external complexation, of the guest, interstation shuttling was restored. Control of shuttling rate has also been achieved by the oxidation/reduction of hydrogen-bonding stations 704 and by photochemical ring contraction of a macrocycle. 705 Scheme 24 Complexation of Copper Leads to the Introduction of a Large Kinetic Barrier to Shuttling 701 4.2 Physical Models of Degenerate, Two-Station Rotaxanes A decrease in the rate of shuttling was observed upon lengthening the alkyl chain of 63 as compared to 62, the magnitude of which corresponds to an increase in activation barrier of ca. 5 kJ mol–1. This effect was attributed to the increased distance the macrocycle must travel between binding stations. 693 This process can be modeled by considering the macrocycle as a particle confined to a one-dimensional potential energy surface (the mechanically interlocked nature of the rotaxane preventing significant motion of the macrocycle orthogonal to the thread). Assuming that thread–macrocycle interactions between stations are negligible leads to the energy profile shown in Figure 24 . The rate of escape from the station energy well can then be modeled by an Arrhenius equation with a contribution from a distance-dependent diffusion factor to the overall rate of shuttling. A quantum mechanical treatment of this system has found that, as the lengthening of the spacer has no effect on the activation for breaking the hydrogen bonds, the effect on the rate of shuttling is due to the widening of the overall potential energy well. This leads to a greater density of states per unit energy, thermal population of a greater number of energy levels, and thus a larger partition function and activation energy. 706 An alternate explanation of the dependence of shuttling rate on linker length was proposed by Brouwer and Günbaş who proposed that shuttling in a similar system was facilitated by hydrogen bonding between the macrocycle and the station to which it is not currently bound. 707 As the linker length increases, this bridging conformation becomes increasingly entropically unfavorable. Some support for this proposal comes from recent examples from Hirose et al. and Sissel and co-workers, 708,709 and Stoddart and co-workers, 710 who in different systems with rigid spacing units found no correlation between spacer length and shuttling rate. Figure 24 Idealized potential energy surface for macrocycle shuttling in a degenerate, two-station molecular shuttle. 4.3 Stimuli Responsive Molecular Shuttles Molecular motion in mechanically interlocked and thus kinetically stable rotaxanes can be controlled using multiple binding sites with affinities for the macrocycle that vary under different conditions. The conditions can be modified by electrochemical redox processes, light, pH, and environmental changes. Once detailed balance is broken by the applied stimulus, as with temperature difference in Feynmann’s ratchet-and-pawl, useful mechanical work can be done, provided there is an additional operation to prevent it being undone. By weakening the existing interaction or increasing the affinity of a competing binding site, the macrocycle can be driven to shuttle to a new equilibrium position. For a reverse or further translational motion to take place, a new stimulus is required to perturb the new equilibrium. 4.3.1 Stimuli Responsive Molecular Shuttles with Single Binding Site Shuttling in a single binding site system can be obtained by reducing the affinity of either the macrocycle for the station or the station to the macrocycle, by the application of an external stimuli. Light controlled switching was observed in a rotaxane made from a “bluebox” macrocycle, an electron-rich dioxyerene station, and redox-active ferrocenyl stoppers. 711 Light-induced electron transfer from the station to the macrocycle was accompanied by electron hole transfer to one of the ferrocenyl stoppers, thereby inducing unfolding of the rotaxane and shuttling of the macrocycle away from the station. Photodriven shuttling in single station rotaxanes was also observed with the increase of steric bulk generated upon photoisomerization of an azobenzene or stilbene station. 712,713 Among these examples, directional shuttling of cyclodextrin (CD) on a symmetrical thread is of particular importance due to the fact that it serves as a ratchet for a directional bias maintained purely by the asymmetric nature of the CD rims. 713 Shuttling was biased to the direction that locates the 6-rim of the CD in a position facing the central stilbene unit (Scheme 25 ). Scheme 25 Light-Driven Directional CD Shuttling in a [2]Rotaxane 712 Solvent-dependent shuttling is generally achieved by switching between hydrogen-bonding and non-hydrogen-bonding solvents. 714 Hydrogen bonding between macrocycle and thread is weakened in the presence of a competing solvent. In a peptide-based rotaxane with a benzyl amide macrocycle, hydrogen-bonding interactions were weakened by addition of MeOH (Scheme 26 ). 715 Because of a solvent-induced relocation of the macrocycle, the chiral center located on the thread now had an influence on the macrocycle. The original system could be restored in CHCl3. More recently, a rotaxane based on a pillar[5]arene macrocycle was shown to shuttle upon either solvent exchange or heating. 716 This system was found to form supramolecular gels in pure DMSO, and in a similar system this shuttling was exploited to form a solvent-driven molecular spring. 717 Scheme 26 Solvent-Dependent Shuttling and Induced CD Response in a [2]Rotaxane 715 4.3.2 Stimuli Responsive Molecular Shuttles with Two or More Binding Sites At a given temperature, the macrocycle of a rotaxane distributes itself among the existing stations according to the binding affinity of each. Perturbation of the binding energies of any of the stations results in the redistribution of the macrocycle toward a new equilibrium state, as driven by thermal motion. Changing the position of the macrocycle in a well-determined way is possible, by making the binding affinity of the less occupied station more favorable (Figure 25 , transformation of station B to B′), 14 or by destabilizing the interaction between the most occupied station and the macrocycle (Figure 25 , transformation of station A to A′). Stimuli responsive modification of the macrocycle can also drive such shuttling processes. Redox, pH, light, and microenvironment (temperature, solvent, etc.) switches are commonly used to control translational movement in rotaxane architectures, and will be discussed in the following sections. Figure 25 Potential energy diagram of a rotaxane-based bistable molecular shuttle. 14 4.3.2.1 pH-Driven Molecular Shuttles Variation in pH is one of the most useful stimuli used to drive translational motion in rotaxanes because hydrogen bonding, electrostatic, and ion–dipole interactions can be effectively controlled by protonation or deprotonation. Indeed, the first molecular shuttle developed by Stoddart et al. depended on acid-driven relocation of the electron poor, positively charged, cyclophane macrocycle from a protonated benzidine station to a biphenol station (Scheme 27 ). 718 Initially, under neutral pH and at 229 K, the macrocycle preferentially rested on the benzidine station (with 84:16 distribution), stabilized by donor–acceptor interactions. Upon protonation of this station, a decrease in the strength of the interaction and an increased electrostatic repulsion caused the macrocycle to relocate to the other station with a new equilibrium distribution of more than 98:2 in favor of the biphenol station. Complexation of the cyclophane macrocycle in similar systems has inspired a variety of computational research. 719−728 Scheme 27 A pH-Driven Molecular Shuttle 718 A relatively low temperature was required to obtain an acceptable distribution bias in favor of the benzidine station under neutral conditions. This necessitated the development of a system, which exclusively resided on one station under standard conditions. Positional discrimination with a ratio of more than 98:2 was obtained in a rotaxane composed of a dibenzocrown ether macrocycle and ammonium-bipyridinium and ammonium-triazole stations. This interaction has been exploited in many systems. 729−743 In a structurally similar rotaxane dimer, pH-induced contraction and expansion was made to function as a “molecular muscle”. 744−746 The pH-dependent relocation of an amide-based macrocycle from a succinamide to an hydroxyl-cinnamate station was observed, with enhanced hydrogen bonding between the deprotonated cinnamate derivative and the macrocycle. 747 Rotaxanes in which repositioning to another station was driven by macrocycle protonation have been reported. 748,749 Shuttling of a metal–macrocycle complex from one ligand to another on a rotaxane as a result of protonation of one of the ligands has been investigated. 750 The pH-driven shuttling of cucurbit[7]uril between bipyridinium and carboxylate stations has recently been reported. 751 Attempts were made to couple this motion to an oscillating pH background reaction, but the pH oscillation was rapidly damped in this case. Autonomous translocation is an important concept in molecular machine design, but requires further research. 4.3.2.2 Redox-Driven Molecular Shuttles Rotaxanes with electron donor–acceptor units or transition metal complexes can be controlled by redox chemistry, provided that redox potentials of the interacting units are carefully chosen. 752−756 After the redox process, the products should be stable on the shuttling time scale, and rapid charge recombination must be prevented. Redox control processes can be chemical, electrochemical, or photochemical. An early example of redox-mediated shuttling was reported by Leigh et al. and took advantage of a redox-active naphthalimide, which was introduced at one end of the thread. The amide-based macrocycle migrated from a succinimide station to the napthalamide station upon one-electron reduction of the naphthalamide (Scheme 28 ). 757−759 Increased electron density on the imide carbonyl of the naphthalamide station outcompeted the succinimide station, and the macrocycle rested almost exclusively at the naphthalamide station. The shuttling was reversible and could be performed through photochemical excitation accompanied by reduction with an external electron donor. Scheme 28 A Redox-Driven Molecular Shuttle 757,758 Benzidine, tetrathiafulvalene, viologen, naphthalimide, and dioxynaphthalene derivatives have been widely used as redox-active stations. 760−777 Redox-induced changes in the coordination preference of transition metal complexes bound to the macrocycle have also been used to drive biased Brownian motion toward different stations. 778 A tristable rotaxane has been reported where the position of the macrocycle over each of the three stations could be controlled solely using electrochemistry. 779 4.3.2.3 Ion-Driven Molecular Shuttles The presence of ions usually affects the conformation and stability of the macrocycle on a station by either interfering with the existing hydrogen bonding, ion–dipole, and dipole–dipole interactions or by sterically disfavoring the binding. This effect can be strong enough to drive translocation of the macrocycle to another station on the thread. In crown ether-based macrocycles, metal ions can be chelated by the macrocycle and thus alter the affinity of the macrocycle for their binding sites, and in some cases induce shuttling. Lithium was reported to mediate the shuttling of crown ether-based macrocycles from a naphthalimide station to a pyromellitic diimide, presumably due to stronger ion–dipole interactions with this station in the presence of a cation. 780 Shuttling could be reversed by the addition of an excess of [18]crown-6, which sequestered the lithium cation, returning the macrocycle to its original equilibrium distribution. Rather than interacting directly with the macrocycle or a station, ions can bind to another position on the rotaxane and affect the affinity of the macrocycle with either station. Shuttling under similar allosteric control was observed using a bis(2-picolyl)amine stopper that chelates cadmium ions. 781,782 Binding to the nearby station was disrupted in the presence of cadmium, and the macrocycle repositioned itself to the distal station. Cadmium could be removed by cyanide complexation to reverse the process. Chloride coordination to a palladium complex was reported to drive shuttling from a triazole station to a pyridinium station (Scheme 29 ). 783 In the absence of chloride anions, the palladium metal was chelated by the macrocycle with the fourth coordinate site being occupied by the triazole unit on the thread. Addition of chloride to the solution as its tetrabutylammonium salt resulted in displacement of the triazole ligand and the macrocycle detached from the station. When this occurred, the crown-ether moiety of the macrocycle was free to interact with the pyridinium station, which promoted translocation. The process could be reversed by addition of AgPF6. Metal-free, anion-induced shuttling between naphthalimide and triazolium stations was also reported in which shuttling could be monitored by UV–vis spectroscopy. 784 Iodide addition has been used to induce shuttling in a halogen/hydrogen-bonding rotaxane. 785 Scheme 29 Chloride-Induced Molecular Shuttling in a [2]Rotaxane 783 In separate work, a shift in fluorescence was obtained by chloride-induced displacement of a macrocycle from a central squaraine station to either of two weakly binding degenerate stations. 786 In a tristable rotaxane bearing urea, ammonium, and phosphine oxide (listed in descending binding ability) stations on the thread, macrocycle shuttling could be induced by the binding of an acetate anion to the urea and subsequently to the ammonium site, which sequentially blocked these higher affinity stations and forced the macrocycle to bind to the weakly coordinating phosphine oxide station. 669 Similarly, the binding of a macrocycle to a guanidinium station was found to be inhibited by PO4 3– addition. 787 In the presence of Zn2+, the macrocycle preferentially bound the central 2,2′-bipyridyl station. In the presence of PO4 3–, but not Zn2+, the macrocycle moved to the weakly coordinating carbamate station. Copper removal followed by guest complexation between porphyrins in the two macrocycles has been used to drive shuttling of a [3]rotaxane. 788 4.3.2.4 Shuttling Induced by Reversible Covalent Modification Dynamic covalent bond formation and reversible reactions are important tools in the synthesis of supramolecular devices and machines. This is due to the ability to control their stability and lability by varying external conditions. These chemical tools enable shuttles to do work by “compartmentalizing” the macrocycle on a station and acting as a physical barrier to prevent random Brownian motion reversing the desired translational motion during re-equilibration steps. This phenomenon will be discussed in detail in the following sections. Reaction-based shuttling can be achieved in several ways. A station can be blocked by introducing a new chemical group, or destabilizing the binding of a macrocycle for steric or electronic reasons. The macrocycle can be chemically trapped on one station. Shuttling between hydrogen-bonding stations has been controlled by Diels–Alder and retro-Diels–Alder reactions, which block and unblock the better-binding fumaramide station. 789 In another example, photoinduced heterolytic cleavage of a C–O bond in a diaryl cycloheptatriene generated a positively charged tropylium station, and thus repelled the cationic cyclophane macrocycle, displacing it to a different station. 790,791 Oxidation or reduction of sulfoxide-based stations and subsequent changes in the hydrogen bonding to a benzylic amide macrocycle have also been used to control shuttling. 627 Scheme 30 Hydrolytic or Entropically Driven Restriction of Shuttling in a [2]Rotaxane 792 Kawai et al. reported intrarotaxane imine formation between a diamino macrocycle and two formyl groups at one of the stations. This reaction locked the macrocycle at the station and maintained a positional discrimination (Scheme 30 ). 792 Hydrolysis of the imine in wet CDCl3 in the presence of acid released the macrocycle from the station and also protonated the amino moieties, making the polyether station more favorable due to hydrogen bonding and ion–dipole interactions. Shuttling of the macrocycle could also be promoted by heating and cooling. This entropy-driven shuttling was possible because of the release of two water molecules on formation of the imine bonds. The entropic favorability of the imine state could be more easily overcome at lower temperatures by enthalpic contributions. Scheme 31 Chemically Driven Shuttling in a [2]Rotaxane 793 As with fumaramides, trans-azodicarboxamide modules have chemical structures preorganized for strong hydrogen bonding with benzylic amide-based macrocycles. The interaction of the macrocycle with the trans-azodicarboxamide station is stronger than that with the succinamide station. Induced shuttling of a macrocycle to the less favored succinamide station could be achieved chemically by blocking the favored station by reaction with triphenylphosphine (Scheme 31 ). 793 This bulky substituent drove the macrocycle away from the station. Under subsequent Mitsunobu reaction conditions, in the presence of a carboxylic acid and an alcohol, the triphenylphosphonium was cleaved from the station as triphenylphosphine oxide. After oxidation, the macrocycle returned to its initial position. 4.3.2.5 Photodriven Molecular Shuttles Photoinduced shuttling is an attractive mode of control over translational motion due to the ease of stimuli introduction and removal. Moreover, if the back reaction or operation is spontaneous and does not require an additional input, then the overall process becomes autonomous; shuttling will occur as long as energy is supplied in the form of light. However, instead of a continuous light source, flashes of light must be used to prevent the system reaching and staying at a steady-state distribution. Although photodriven shuttling can be an efficient approach, the kinetics of translation and photochemical processes have to be carefully considered. If the photoinduced process involves a redox reaction, then rapid charge recombination can take place before positional displacement of the macrocycle. To avoid this, either excited states with a long lifetime are required or external reagents must be used to reduce or oxidize the species. In a carefully designed shuttle reported by Stoddart et al., translational switching was obtained by one-electron reduction of a viologen station by a ruthenium trisbipyridine stopper (Figure 26 ). 794 Relatively slow back electron transfer permitted ca. 10% of the macrocycle to reposition to the dimethyl viologen station on each electron transfer. Continuous irradiation maintained a 95:5 distribution of the macrocycle in favor of the dimethyl viologen station. The system equilibrated back to the original viologen position once irradiation ceased with spontaneous back electron transfer allowing reversion. Figure 26 Photoinitiated redox-driven shuttling in a [2]-rotaxane initiated by (i) irradiation and (ii) subsequent reduction of viologen station. (iii) Competing back electron transfer from one-electron reduced viologen to Ru3+. (iv, v) With continuous irradiation the macrocycle shuttles to the dimethylviologen station. (vi) Ceasing illumination restores the macrocycle to its original position. 794 Alongside the photoinduced redox reactions discussed above, improved excited-state hydrogen-bonding interactions 795 and the photoisomerization of stilbene, 796 azobenzene, 797 and other olefinic moieties leading to altered hydrogen bonding or steric bulk creation have also been used to drive shuttling processes. 798 Absorbance and fluorescence has been used to monitor the progress of shuttling along the rotaxane in some of these systems. 16,799−801 Likewise, other photochromic compounds such as spiropyrans have been exploited to control molecular shuttles, by switching between photoisomers that form hydrogen bonds of different strengths. 802 4.3.2.6 Molecular Shuttles Driven by Changes in the Microenvironment The term “microenvironment” usually refers to the temperature and solvent system. The easiest way to control shuttling via solvent choice is through the disruption of hydrogen-bonding interactions between the macrocycle and the station. Several examples of macrocycle relocation induced by a change in solvent polarity or rotaxane solvation have been reported. 803 In some cases, solvent molecules facilitate a shuttling process already driven by other stimuli. 698,797,804 For entropy-driven shuttles, the rearrangement of solvent molecules in or around the macrocycle is an important parameter controlling the shuttling process. Rotaxanes bearing more than two stations and even polymeric rotaxanes driven by these stimuli have been reported. 797,805,806 In these shuttles, work is undone in each reverse shuttling. Directional shuttling via a ratchet mechanism has been achieved and will be discussed in the next section. 4.4 Compartmentalized Molecular Machines Stimuli-responsive molecular shuttles are rotaxanes in which the net position of the macrocycle on the thread is controlled by external triggers (light, heat, electrons, chemical, pH, binding events, etc.). In general, these triggers change the properties of the binding sites and subsequently have an influence on macrocycle/thread interactions. In this section, various stimuli-induced mechanisms to create and control nonequilibrium conformations will be presented. Rotaxane succ1-74 has two identical stations, which are distinguishable due to the differing stoppers. The thread is divided into two compartments by a bulky silyl ether barrier, preventing the macrocycle from shuttling between the two stations. 75 By removing the silyl ether group, dynamic exchange of the macrocycle was enabled and the system moved toward equilibrium. This resulted in a 50:50 distribution of the macrocycle between the two stations. The operation is irreversible, because the attachment of the bulky silyl group again does not change the average position of the macrocycle, and will not restore the initial distribution of the macrocycle; the position of the macrocycle is not determined by the state of the machine. The machine starts in a statistically unbalanced state (the bulky silyl group prevented the macrocycle restoring balance by moving toward an equilibrium distribution). By “linking” the system, removing the bulky group, the machine could reach its equilibrium state, because free movement of the macrocycle was no longer prevented by the barrier. Therefore, the macrocycle had “escaped” to a lower energy distribution (Figure 27 ). Figure 27 [2]Rotaxane 74 acts as an irreversible mechanical switch. The silyl ether is too bulky to allow macrocycle shuttling between the two succinamide stations. 75 Recently, Coutrot and co-workers synthesized a two-compartment molecular machine consisting of a dibenzo-24-crown-8 macrocycle and a thread with anilinium and monosubstituted pyridinium amide stations. 807 The rotaxane was synthesized via CuAAC chemistry. Subsequent N-benzylation of the triazole moiety, which is located in the middle of the thread, was found to serve as a barrier as well as a third station for the macrocycle. Depending on which station the macrocycle was located, the N-benzylation of the triazole allowed trapping of the macrocycle in either the left or the right compartment. The Leigh group has extended their initial system using the same design described above, but with two different stations 75 (a fumaramide station and succinamide station) (Figure 28 ). 75 Figure 28 Operation of a compartmentalized molecular machine, which corresponds to a two-state Brownian flip-flop. Operation steps: (a) Desilylation (80% aqueous acetic acid); (b) E → Z photoisomerization (hν at 312 nm); (c) resilylation (TBDMSCl); and (d) Z → E thermal isomerization (catalytic piperidine). 75 In this system, the exchange of the macrocycle between stations could be controlled by the introduction of a barrier in the middle of the thread (kinetic control) as well as by altering the binding affinity of the macrocycle to the two different stations (thermodynamic control). Initially, 85% of the macrocycle bound the fumaramide station and 15% the succinamide station. The system was still unlinked because of the kinetic barrier and therefore not in equilibrium. A balance-breaking stimulus (photoisomerization at 312 nm) generated a 49:51 E:Z photostationary state. Removal of the barrier, the linking stimulus, allowed the balance to be restored by the macrocycle equilibration. Restoring the barrier unlinked the system. The last step was Z → E olefin isomerization, which made the system statistically unbalanced, unlinked, and not in equilibrium. After one operational cycle of the machine, 56% of the macrocycles were located on the succinamide station. The thread had therefore changed the net position of the macrocycle. Because the succinamide station binds the macrocycle less strongly than the fumaramide station, this process has moved the macrocycle energetically uphill; that is, it has performed a ratcheting operation, transporting a particle away from equilibrium. This system and its function is an experimental realization of the transportation task required in Smoluchowski’s trapdoor and Maxwell’s pressure demon, which are discussed in the introduction (section 1 ) and are powered by chemical and light energy. Figure 29 shows a schematic representation of the four sequential steps performed by rotaxane 75 to transport the system energetically uphill: balance-breaking 1, linking, unlinking, balance breaking 2 (resetting the machine but not the substrate). Figure 29 Initially the system is balanced (in proportion to the sizes of the two compartments), with an equal density of particles in the left and the right compartments. By changing the volume of the left-hand compartment, the system becomes statistically unbalanced. Opening the door allows the particle to redistribute according to the new size of the compartments. Closing the door ratchets the new distribution of the particles. Restoring the left compartment to its original size results in a concentration gradient of the Brownian particles across the two compartments. Here, the size of the compartment represents the energy level of the macrocycle-station system. 75 There are many different types of compartmentalized molecular machines, some of which have been described in this section. Three main machine types exist. First is the two-state (or multistate) Brownian switch, which is a machine that can reversibly change the distribution of a Brownian substrate between two distinguishable sites as a function of state of the machine. It does this by biasing the Brownian motion of the substrate (section 4.3 ). Second is a two-state (or multistate) Brownian flip-flop, a machine that can reversibly change the distribution of a Brownian substrate between two (or more) distinguishable sites and that can be reset without restoring the original distribution of the substrate. The distribution of the substrate cannot be determined from the state of the flip-flop, but rather it is determined by the history of the operation of the machine. Rotaxane 75 is an example of a two-state Brownian flip-flop. Third, a Brownian motor is a machine that can repetitively and progressively change the distribution of a substrate while the machine can be reset without restoring the original distribution of the substrate. The rotary motors designed by Feringa and co-workers are examples of this category of device. Additional examples will follow in section 4.5 . Brownian ratchet mechanisms fall into two general classes: energy ratchets and information ratchets (section 1 ). The above rotaxane is an energy ratchet in which the energy minima and maxima of the potential energy surface are varied without regard to the particle’s location. In the next section, examples of the second class of Brownian ratchets, information ratchets, will be explored where a barrier is raised or lowered according to the position of a Brownian particle on a potential energy surface, resulting in the particle distribution being directionally driven away from equilibrium. 4.4.1 Molecular Information Ratchets An information ratchet, such as Maxwell’s pressure demon, requires knowledge of the position of each particle and is able to open the door if the particle is approaching from a certain direction. This positional sorting allows the accumulation of particles in one side of the container and thus results in a pressure gradient. The first example of a molecular information ratchet was described by Leigh and co-workers (Scheme 32 ). 808 Molecular machine 76 consisted of a dibenzo-24-crown-8-based macrocycle, which was mechanically locked on a linear thread by bulky 3,5-di-t-butylphenyl stoppers. An α-methyl stilbene unit divided the thread into two compartments, both with ammonium binding stations. The rotaxane used photosensitized energy transfer from the macrocycle to the stilbene unit as the key step in changing the macrocycle distribution. When the stilbene unit adopted the E form, the macrocycle could move randomly along the full length of the thread by Brownian motion, while when the Z form is adopted, the macrocycle was trapped in one or the other of the two compartments. To keep the machine as the Z-isomer, it was operated in the presence of the photosensitizer PhCOCOPh. The sensitized photostationary state of α-methyl stilbene is 82:18 Z:E. Selective “gate opening” was achieved by photosensitized energy transfer from the macrocycle (which has a benzophenone (PhCOPh) moeity attached) to the stilbene unit. This led to a 55:45 Z:E ratio of the α-methyl stilbene. As energy transfer was distance dependent, this photosensitized transfer was more likely to happen when the macrocycle was in the blue compartment (Scheme 32 ). When the macrocycle was on the green station, intramolecular energy transfer from the macrocycle was not efficient and the gate stayed closed, biasing macrocyclic distribution. This is an example of using the positional information on a Brownian particle to break detailed balance, driving macrocycle distribution away from equilibrium. Scheme 32 Structure and Mechanism of the Information Ratchet (a) Gate closed, but energy transfer from the macrocycle is efficient. (b and c) Gate is open, allowing free shuttling of the macrocycle. (d) The macrocycle is on the green station, intramolecular energy transfer (ET) from the macrocycle is inefficient; intermolecular energy transfer from PhCOCOPh dominates (closing the gate). 808 An alternate approach to a molecular information ratchet using chemical fuel to directionally transport a macrocycle was later published by Leigh and co-workers (Scheme 33 ). 809 In this example, a phenyl ester attached to a chiral carbon center separated two degenerate fumaramide binding sites. The close proximity of the stations to the hydroxyl bearing carbon enabled the chiral center to influence the position of the macrocycle. At equilibrium, the macrocycle rested equally at either degenerate station. When a barrier was introduced by benzoylation with an achiral dimethylamino pyridine (DMAP) catalyst, an equal distribution of products was obtained. However, when the chiral DMAP-based catalyst (S)-79 was used, a 33:67 distribution in favor of the pro-(S)-fumaramide (S)-80 station was obtained. Using catalyst (R)-79 resulted in a reverse bias in the position of the macrocycle. Because the stations were identical, macrocycle relocation resulted in a decrease in entropy, with identical enthalpic interactions, and thus a ratcheting operation was achieved. Motion away from an enthalpically favored location was achieved in a similar rotaxane, but with fumaramide and succinamide stations of different binding strengths. Using the same chiral catalyst, 15% of the molecules were relocated to the enthalpically unfavored succinamide station. Scheme 33 Chemically Driven Molecular Information Ratchet 809 This concept was further developed in the three-compartment, chemically driven, molecular information ratchet 82. 810 Here, the compartments contained fumaramide groups and were separated by hydroxyl groups (Scheme 34 ). The macrocycle could be efficiently trapped at either end of the track by sequential benzoylation of the hydroxyl groups in the presence of the chiral catalyst 83 resulting in a 1:21:79 distribution of the macrocycle. The steric barriers formed prevented the macrocycle from passing, and trapped the ring in a certain compartment. It was shown that the macrocycle had an influence on the rate of benzoylation of the hydroxyl groups on the thread resulting in acylation taking place preferentially far from the macrocycle. The distributional bias could be reversed with the enantiomeric catalyst 83, and when both chiral catalysts were used the macrocycle preferentially resided in the central station with a 10:77:13 distribution. Scheme 34 Directional Transport of a Macrocycle in a [2]Rotaxane Three-Compartment Chemical Information Ratchet Directional translational motion along the axis of pseudorotaxane has also been reported (Figure 30 ). 553,811,812 Threading and dethreading of the macrocycle could be controlled by photochemical, chemical, or electrochemical redox reactions. The pseudorotaxane axis has three different chemical motifs responsible for directional translocation. The first is a 2-isopropylphenyl group, a neutral energy barrier at one side of the axle. An electron-rich 1,5-dioxynaphthaline occupied the center, and at the other end was a positively charged 3,5-dimethylpyridinium station. A tetracationic electron-poor cyclobis(paraquat-p-phenylene) macrocycle formed a π-complex with the electron-rich 1,5-dioxynaphthaline group. To form the complex, the macrocycle had to pass one of the energy barriers, and the neutral 2-isopropylphenyl moiety presented a lower energy barrier. Hence, threading took place selectively from this side of the axis. Upon one-electron reduction of the macrocycle, interaction with the 1,5-dioxynaphthaline station was weakened. Additionally, as a result of the reduced positive charge on the macrocycle, electrostatic repulsion from the pyridinium site became less potent, allowing the macrocycle to pass over this barrier with greatly reduced activation energy. Hence, once reduced, the macrocycle dethreaded over the pyridinium group. The macrocycle threaded and dethreaded in a directional manner via a ratcheting mechanism (flashing ratchet) acting on the potential energy surface of the pseudorotaxane axle. The same group later reported a flashing ratchet based on similar principles, in a pseudorotaxane with one stopper, where threading was driven by reduction, and dethreading slowly occurred after rapid shuttling away from a reoxidized bipyridinium station. 813 Recently, Credi et al. also reported the photodriven directional threading/dethreading of a crown ether macrocycle on an axle using a flashing ratchet mechanism. 814,815 Figure 30 Photodriven directional translational motion in pseudorotaxanes 86–87. Reprinted with permission from ref (811). Copyright 2013 American Chemical Society. 4.5 Controlling Rotational Motion in Mechanically Bonded Structures An alternative way to extract work from mechanically interlocked structures would be to control the relative rotation of their components. The rate of macrocycle pirouetting can easily be controlled through temperature, electric field strength, light, structural freedom, or by altering its interaction using other chemical moieties or solvents. 564,698,816−819 Nevertheless, controlling the directionality of this movement requires careful design. 4.5.1 Controlling Rotational Motion in Rotaxanes The pirouetting frequency of a macrocycle around a thread depends strongly on the strength of the interactions between the macrocycle and the thread (and environment), which must be broken and reorganized during the movement of the macrocycle. The rate of macrocyclic pirouetting in some hydrogen-bonding rotaxanes can be controlled by changing the strength of an electric field, which alters the strength of the hydrogen bonds. 820−824 In some cases, direct manipulation of macrocycle/axle interactions can either restrict rotation or increase its frequency by weakening intracomplex interactions. The formation of weaker hydrogen bonds with maleamide units as compared to fumaramide units has been discussed in previous sections. Fumaramide preorganizes two hydrogen-bonding motifs to bind the macrocycle, leading to an increase in binding strength. On the other hand, the cis-form of this olefinic structure can only form hydrogen bonds with one site of the macrocycle. As a consequence, the station–macrocycle interactions are weaker, and thus rotation was observed to increase in frequency by 6 orders of magnitude on photoisomerization. 825 Figure 31 Effect of water on the rate of pirouetting of a macrocycle about an axle. Reprinted with permission from ref (698). Copyright 2013 Nature Publishing Group. Like photoisomerization-dependent disruption of hydrogen bonding, 825 solvent choice can also play a role in controlling these noncovalent interactions and hence influence the rate of pirouetting in a rotaxane system. 826 Comparison of the rate of exchange between axial (E 1) and equatorial (E 2) protons, which was determined by the pirouetting rate, showed that, as compared to heating in dry pyridine, the exchange was greatly accelerated when the pyridine contained 5% D2O (Figure 31 ). 698,827 Addition of water sped up redox-driven shuttling in a different rotaxane system by hydrogen-bonding dependent “lubrication” of the mobile elements. Interestingly, other protic solvents (alcohols, nitromethane) had less effect on the pirouetting rate. This was attributed to the ability of water to form 3-D hydrogen-bonding networks. The effect on macrocycle–thread interactions of the sequential addition of single methanol molecules has been probed by IR spectroscopy of rotaxane-solvent clusters. 828,829 Besides hydrogen-bonding, cation-induced restraint of motion in a crown ether-based macrocycle has been reported with a corresponding increase in pirouetting rate upon demetalation. 817,818,830−834 Redox-dependent switching between coordinating ligands on a macrocycle has been used to control rotational orientation of a macrocycle about a rotaxane axle. 818 In a hybrid organic–inorganic rotaxane, hydrogen bonding between fluorines on the macrocycle and ammonium stations resulted in slow shuttling but fast rotation. 835,836 This rapid rotation was attributed to the array of hydrogen-bonding interactions formed in which a new bond had already started to form before the original was completely broken. 4.5.2 Controlling Rotational Motion in Catenanes The points discussed above for rotaxanes also apply to rotational motion in catenanes. In a catenane, the rate of rotation of one macrocycle with respect to the other depends on the strength of interactions between the two. Therefore, any stimuli altering this interaction should result in a change in rotational rate. The solvent dependence of the rate of macrocycle rotation in catenanes containing benzylic amides has been measured by NMR analysis and also by AFM-based single molecule measurement techniques. 837 In the second method, catenanes with different intercomponent hydrogen-bonding ability were attached to poly(ethylene oxide) polymers and analyzed in dimethylformamide (DMF) and tetrachloroethane (C2H2Cl4). As a measure of equilibrium conformational entropy, the persistence length of the polymer was analyzed for two different compounds in two different solvents to estimate the restriction of motion of the macrocycle in the catenane. In agreement with the NMR analysis, in a single molecule experiment using the AFM method, the mobility of the macrocycle was shown to be accelerated by polar solvents or by disrupting the hydrogen-bonding ability by chemical modification. Ion exchange, 838−840 pH, 841−844 redox, 845−856 demetalation, 842,857−869 light or redox-mediated ligand exchange of metals, 853,855,870 photochemical switching, 870−878 photoisomerization-dependent sequestering of macrocycle 879 and solvent, 826,880−882 photoisomerization-dependent change in hydrogen-bonding potency, 883 and the self-assembly of liquid crystal phases 884 have also been reported to control the rotation of the macrocycle in catenanes. However, directional rotation of one macrocycle with respect to another in a catenane structure has rarely been investigated. Directional rotation was achieved with a [3]catenane in which the presence of a third macrocyclic component helps to restrict the rotational freedom of the molecule (Figure 32 ). 883 The large macrocycle on which the two small macrocycles rotate bears four different interaction sites: two fumaramide motifs with different binding affinities (light green and red, the methylated station has a lower affinity for steric reasons), one succinic amide ester (orange), and finally an amide group (dark green). A benzophenone unit is attached close to the first fumaramide station to enable selective photosensitized isomerization of this olefinic group through energy transfer using a higher wavelength than required for nonsensitized isomerization. Figure 32 Directional circumrotation in a [3]catenane. (i) hν (350 nm), (ii) hν (254 nm), (iii) heating; or heating with catalytic ethylenediamine; or catalytic Br2, hν (400–670 nm). Adapted with permission from ref (883). Copyright 2003 Nature Publishing Group. Initially, in the absence of any stimuli, one of the macrocycles (blue) rests on the most favored fumaramide station, while the second macrocycle (purple) resides on the second most favored, the methyl-fumaramide station (Figure 32 ). Upon photosensitized isomerization of station A, hydrogen-bonding interactions between the blue macrocycle and the station weaken, and because the second most favored site is already occupied by the purple macrocycle, the blue macrocycle moves to the third most favored, succinimide amide ester station C. Further photoisomerization of the methyl-fumaramide station triggers the shuttling of the purple macrocycle to the final binding site, station D. The fumaramide stations are converted to the cis-isomers either by heating in the presence of a catalytic amount of ethylenediamine or via irradiation at 400–670 nm in the presence of a catalytic amount of Br2. This time, however, the relative orientations of the macrocycles force repositioning to the newly formed fumaramide stations in the opposite, with respect to the initial, distribution. A second round of stimuli is necessary to fully reset the system and to bring the macrocycles back to their initial positions. The combination of control of station binding affinity and macrocycle position allowed directional rotation in this catenane. Scheme 35 Directional Circumrotation in [2]Catenane 91(885) A directional [2]catenane rotary motor has also been reported with a simpler chemical/photoinduced mode of operation and well-defined ratcheting mechanism. 885 In this example, the balance breaking operation (to create a new equilibrium distribution) was photoisomerization of the fumaramide station. However, the macrocycle’s rotation to the new energy minima was blocked by two orthogonal protecting groups: an acid labile trityl and base labile tert-butyl-dimethylsilyl. Selective deprotection of one of these protecting groups allowed the movement of the macrocycle to the succinamide station in only one direction. For clockwise rotation, trityl deprotection was required. Under equilibrium conditions, the alcohol had to be protected once more to avoid the work being undone. Full rotation was attained by deprotection and reprotection of the tert-butyldimethylsilyl moiety with reisomerization of the fumaramide station. The sequence of deprotection–reprotection reactions could be reversed to achieve counter clockwise circumrotation. The repositioning of the macrocycle to its new equilibrium position was via only one of two degenerate pathways, and thus the system acts as a directional molecular rotor (Scheme 35 ). 5 Molecular Level Motion Driven by External Fields In most synthetic molecular machines described to date, control over motion arises from the selective restriction of Brownian fluctuations through control of steric and noncovalent bonding interactions (via manipulation of chemical structure). The application of external fields can cause the bulk movement or orientation of a molecular species with electric field-directed alignment of liquid crystals being the most important technological application. Several research groups aim to use electromagnetic fields to control submolecular motion, with most examples concentrating on generating submolecular rotary motion. Molecular rotation involves passage over the minima and maxima of a torsional potential energy surface. An external field can either induce an excited state, where the torsional potential energy surface is altered, or interact with a permanent or induced molecular dipole and so orient the molecule in a particular direction. The interaction between the field and the rotor provides the energy necessary to surmount kinetic barriers and overcome energy dissipation and thermal randomization. To date, molecular dynamics simulations of molecular rotatory systems have dominated the field. 886−930 Only a few examples exist where theoretical studies have led to the synthesis of rotors and their examination under the influence of an electric field. A recent example is the simulation of a single-molecule electric revolving door based on a phenyl-acetylene macrocycle published by Hsu, Li, and Rabitz. 931,932 The authors found, based on simulations, that the opened and closed-door states of 92, whose exchange was accompanied by significant conductance variation, could be operated by an external field. Furthermore, they proposed that due to the large on–off conductance ratio, the molecular machine could also serve as an effective switching device (Figure 33 ). Figure 33 Schematic representation of an electric revolving door. (a) Door closed-switch on leading to high conductance. (b) Door open-switch off leading to low conductance. 932 An earlier example of a computational investigation was published by Ratner and Troisi. 933,934 Compound 93 bound between two electrodes exhibited motion under the influence of an electric field (Scheme 36 ). The authors suggested that the system could find applications in molecular electronics and could be used to create switchable molecular junctions. Theoretical studies revealed that the conformers of 93 (which could be interchanged by an external field) also showed differences in conductance across the junction and could thus be used as a conformational molecular rectifier. 935−937 Scheme 36 A Molecular Rectifier (a) High conductance predicted. (b) Low conductance predicted. 934 Fujimura and co-workers undertook a series of theoretical studies into the mechanism of rotation in gas-phase directional chiral motors driven by picosecond pulses of a linearly polarized laser. 938−950 In one study, aldehyde 94 was examined as a chiral molecular motor with the formyl group as the rotor (Figure 34 ). They considered both enantiomers in their studies and came to the conclusion that directional motion originated from the asymmetric potential-energy surface of the chiral molecules and time-correlated forces created by laser-permanent dipole interactions. 940 The process is reminiscent of a rocking Brownian ratchet (section 1.6.2 ), but here thermal energy is not required to surmount kinetic barriers. 951 Figure 34 One enantiomer of chiral molecule 94, in which directional rotational motion was driven by linearly polarized laser pulses and studied by quantum and classical mechanical simulations. 940 Several examples of surface-mounted molecular rotors have been reported where motion could be driven by alternating electric fields or the absorption of pulses of light. 952−962 Michl and co-workers synthesized, on the basis of computational studies, a series of surface-mounted rotors. 963−965 Figure 35 shows the structure of polar and nonpolar versions of their altitudinal rotor. Polar rotor 96 consisted of a 9,10-dihydrophenanthrene substituted by four fluorine atoms on the central ring and had a calculated dipole moment of 3.7 D. The nonpolar rotor was a 4,5,9,10-tetrahydropyrene whose D 2 symmetry precluded a dipole moment. X-ray photoelectron spectroscopy (XPS), scanning tunneling microscopy (STM), and IR spectroscopy showed that for a fraction of the molecules, the static electric field from an STM tip induced a change in the orientation of the polar rotor, but not in the nonpolar analogue. 966 Several molecular dynamics simulations studying the influence of an electric field or a fluid flow on the rotation of a molecule fixed on a surface have been published by the same group. 967,968 One recent example simulated carborane-based molecular propellers and showed that they could be successfully driven at GHz rates by an oscillating electric field or by a flow of gas. 969 Figure 35 Nonpolar 95 and dipolar 96 altitudinal rotors mounted on an Au(111) surface. Note that rotor and flanking aryl rings are arbitrarily shown perpendicular to the surface for clarity. 966 Molecular machines in the solid state and condensed phase will be discussed in more detail in section 8.6.1 . A number of research groups have been working in the field of crystalline molecular machines with the goal of creating new materials with interesting properties and that are responsive to external stimuli such as external electric fields. 970−975 Amphidynamic crystals are a form of condensed-phase matter with anisotropic molecular order and controlled dynamics, and they offer a good platform for the design of these new materials. 976 Molecular rotors are one of the most promising molecular structures for the synthesis of amphidynamic crystals. Structural designs analogous to macroscopic gyroscopes and compasses are one possibility in the design of such molecular rotors. Several examples have been published consisting of a rotating unit linked to a shielding box or stator by an axle. 977−980 The shielding box generates the local free volume required for unhindered rotation in an otherwise densely packed environment. The study of the orientation of and dipole–dipole interactions in dipolar rotor arrays are important in understanding the dielectric properties of these materials and the dynamics of the dipolar rotors. Furthermore, materials containing dipolar rotors can be controlled and reoriented by external electromagnetic fields or optical stimuli. 981,982 Examples of molecular gyroscopes have been reported by Gladysz et al., who developed transition metal-based systems in which trans phosphorus donor atoms are bridged by three methylene chains or related linkers. For example, the system shown in Figure 36 contains either a Fe(CO)3 or a Fe(CO)2(NO)+ rotor. The latter possesses a dipole moment, which could be a possible handle for external control. 983 A series of studies have been carried out by the same group using gyroscope-like platinum and palladium complexes with trans-spanning bis(pyridine) ligands. 984 Figure 36 (a) Molecular structure of transition-metal-based gyroscopes 97 and 98. (b) X-ray structure of compound 97. X-ray crystal structure reprinted with permission from ref (983). Copyright 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Molecular gyroscopes with two bulky stators such as triarylmethyl or triptycyl groups, creating space for a 1,4-phenylene, have been reported. 985−1002 Some of these wheel and axle motifs (see also section 8.6.4 ) showed a dielectric response as a function of external alternating electric fields. 981,1003 6 Self-Propelled Nanostructures The mechanical self-propelled motion of microscopic objects such as bacteria and the motion of camphor or soap “boats” driven by interfacial forces have long fascinated scientists. More recently, focus has shifted to the control of nanoscale motion in synthetic systems. 1004,1005 These systems are driven not by thermal energy, but by other mechanisms more relevant on the nanoscale. Motion may be generated either autonomously when the energy required for movement is continuously supplied or nonautonomously. 6.1 Propulsion by Manipulation of Surface Tension If surface tension is not balanced between two sides of a droplet of liquid or a bubble of gas, directional transport may occur. This is known as the Marangoni effect and is observed in several natural systems exhibiting spontaneous flow. 1006,1007 Temperature gradients have typically been used in artificial systems to generate Marangoni flows. 1008−1014 A droplet containing a species that absorbs to the surface, irreversibly modifying the surface energy thereof, has also been used to illustrate this effect in a “chemical Marangoni effect”. 1015−1018 The autonomous motion of millimeter scale objects on a liquid surface driven by the catalytic decomposition of hydrogen peroxide was reported by Whitesides et al. 1019 A small area of platinum near the edge of each disc catalyzes this decomposition and releases oxygen bubbles, which drive the movement by a recoil force. Others, however, observed the motion of platinum–gold rods toward the oxygen producing platinum end. 1020−1025 The contributions of driven and Brownian motion could be distinguished at aqueous/organic interfaces, 1022 and a rotor based on the same principles has also been synthesized. 1026 A similar system was reported by Manners and co-workers. 1027 Ozin et al. reported platinum–silica sphere dimers, which showed quasi-linear motion in bulk solution and rotational motion at the solution–glass interface. 1028 Rotors of decreased size and increased rotational rate have since been reported, 1029 both once more based on hydrogen peroxide decomposition. 1030 The catalytic domain has been scaled down to a molecular level by Feringa et al., 1031 who utilized a surface bound dinuclear manganese catalase analogue. Recently, Fischer et al. reported a self-propelling nanoparticle of only 30 nm diameter, a size comparable to certain enzymes, based on a Pt–Au system. 1032 Feringa and co-workers reported the use of carbon nanotube bound enzymes in autonomous propulsion. 1033 As an alternative to the ubiquitous use of hydrogen peroxide to propel nanoparticles, hydrogen evolution by a magnesium domain has been exploited to drive autonomous motion. 1034 Finally, Sen et al. reported the use of ring-opening metathesis polymerization using a modified Grubbs’ catalyst to propel a nanoparticle. Motion was generated by the consumption of monomer only on one side of the nanoparticle, that where the catalyst is bound. This creates a concentration gradient across the nanoparticle and thus movement. 1035 6.2 Mechanical Self-Propulsion In contrast to most synthetic, surface tension driven, self-propelled systems, microorganisms almost exclusively utilize mechanical self-propulsion, akin to a swimming motion or the corkscrew motion of a boat’s motor. The low Reynolds number at the nanoscale limits the number of swimming mechanisms that are viable, 1036 as any useful motion must involve a nonreciprocal motion to break the time reversal symmetry. 26,1037−1044 Various solutions have been proposed (Figure 37 ). 1041,1045−1050 The key point in each is that they must possess at least two hinge points; no system with a sole hinge point can be nonreciprocal. The sequential use of each hinge point allows motion at low Reynolds number. Figure 37 (a,b) Proposed designs for swimmers viable at the nanoscale. (c,d) Potential molecular solutions. 14 No molecular solutions have yet been reported. However, a microscopic artificial swimmer has been reported. 1051 A major difficulty in realizing any molecular design would be carrying out the driven motion rapidly and frequently enough to be observable against random Brownian motion. The repetitive, directional, rotation of a chiral molecule is also a nonreciprocal motion. The field-driven directional motion of an unconfined chiral molecular rotor would result in its propulsion via a screw-propeller type mechanism, such as the bacterial flagella system. Such a system has been theoretically studied as a means to spatially resolve a racemic mixture of propeller-like molecules. 1052,1053 7 Molecular-Level Motion Driven by Atomically Sharp Tools The development of single molecule imaging techniques such as atomic force microscopy (AFM), scanning tunneling microscopy (STM), and optical and magnetic tweezers has significantly enhanced our understanding of the mechanical properties and working mechanisms of molecular motor proteins. 1054−1068 Instead of average statistical information obtained from an ensemble of species, these techniques allow direct measurement of molecular level forces, mechanical properties and motions such as rotation, gliding, and translation, pivoting of an individual molecule, supramolecular host–guest exchange, and coconformational changes within interlocked molecular assemblies. 1061,1069−1089 Beside their imaging abilities, the use of such tools to drive molecular level motion has also been explored. Molecular scale motion is driven by different forces than in the macroscopic world. Gravity is essentially irrelevant at low Reynolds number, and frictional forces differ greatly. 606,1090−1101 The driving forces/interactions must be sufficient to overcome significant thermal fluctuations. For the desired motion to be obtained (i.e., translational motion via rotation of the wheels of a “nanocar” instead of gliding), surface–molecule interactions should be neither too strong nor too weak to balance dissociation and immobility. STM-driven positional displacement of xenon atoms at low temperature was reported by Eigler et al. 1102 The rotation of oxygen and acetylene molecules adsorbed on Pt(111) and Cu(111) surfaces, respectively, was shown to be induced by the tunneling of electrons from an STM tip at low temperatures. 1103,1104 STM- or AFM-driven motions of aromatic molecules on surfaces have been investigated extensively. 1105−1116 The STM-driven directional diffusion of 9,10-dithioanthracene over a Cu(111) surface has been reported. 1117 STM-induced translational motion of porphyrins along the voids of a porphyrin monolayer on a Cu(100) surface was attributed to the rotation of a t-butyl substituent. 1113 Displacement to a region with less structural restriction led to more frequent rotation, and thus the STM probe mediated a switch between a rotating and a nonrotating porphyrin. Cyclic molecules and fluorinated C60 derivatives were also shown to roll across surfaces. 1118−1120 Translational motion mediated through rolling structural units (wheels) has been reported in a number of independent publications. Initial attempts to develop a “molecular wheelbarrow” resulted in overly strong surface–-molecule interactions, and the STM tip caused fragmentation. 1121−1124 STM-driven rolling of a single molecule deposited on a Cu(111) surface was first achieved by Grill et al. 1125 A chemically switchable metalloporphyrin pinwheel was reported by Lambert et al. 1126 The reversible attachment of two CO2 molecules to a diffusing anthraquinone on a Cu(111) surface was an important example of the potential of artificial molecular cargo transporters. 1127 Chiaravalloti et al. designed the first molecular system acting as a rack and pinion device, using self-assembled hexa-tert-butyl-pyrimido-pentaphenylbenzene on Cu(111). 1128 Interlocked arrays formed by self-assembly allowed the rotation and translocation of a single molecule along the array by manipulation with an STM tip. Figure 38 (a) Chemical structure and (b–f) STM micrographs of translational motion of a four-wheeled “nanocar” on an Au(111) surface at 200 °C. Wheels are shown by yellow spots, and the path is highlighted with a white arrow. Reprinted with permission from ref (1129). Copyright 2005 American Chemical Society. In a four-wheeled “nanocar” with fullerene wheels, thermally induced translational motion on a Au(111) surface was observed (Figure 38 ). 1129 Directional motion perpendicular to the axles of the molecule suggested that the observed movement was generated by rolling of the fullerene wheels rather than sliding of the entire molecule on the surface. In addition to thermal activation, an STM tip could be used to drive the motion. Pivoting also took place, clearly visible as small-angle perturbations to the path of translation in the STM images (Figure 38 e,f). In a structurally related three-wheeled molecule, pivoting was observed to be the dominant motion on the surface, suggesting that the four-wheeled molecular structure was important for a rolling translational motion. Similar examples of four- and three-wheeled fluorescent molecules with different wheel sizes (adamantane and p-carboranes as wheels) have been reported, and their diffusion constants on glass surfaces were determined by fluorescence measurement. 1130,1131 The photoisomerization kinetics of a similar four-wheeled molecule functionalized with a rotary motor have also been investigated. 1132 When fullerene wheels were used no photoisomerization took place, whereas with carborane wheels, efficient isomerization was observed in solution. The rotation of dialkylthioethers adsorbed on Cu(111) surfaces has been extensively studied and electrons from an STM tip found to drive a 5% bias in rotational direction. 1133 Electrical excitation of C–H stretching modes in the molecule contributed to ratcheting in the rotation. The direction and rate of rotation depended on the chiralities of both the molecule and the STM tip. In a cerium-centered double-decker molecule self-assembled on a Au(111) surface, an interesting phenomenon was detected (Figure 39 ). 1134 The rotational chirality of individual molecules generated two different orientations on the gold surface, which could clearly be distinguished by STM. Upon scanning the surface with the STM tip, an irreversible, abrupt change in the chirality of some of the molecules was observed. The switch resulted from the rotation of the upper porphyrins, and the irreversibility was attributed to damage caused to the molecules by the high applied voltage. Figure 39 (a) Structure of a cerium-centered double-decker molecule. (b) STM micrographs of the molecules assembled on Au(111) surface. The two distinct isomers, due rotational chirality, are shown in white and blue crosses. (c) Space-filling model of the two chiral species. Reprinted with permission from ref (1134). Copyright 2011 American Chemical Society. Figure 40 (a) Chemical structure of a rotary motor with the groups responsible for double-bond isomerization (red) and helix inversion (blue) highlighted. (b) Schematic representation of a full 360° rotation with sequential double-bond isomerization and helix inversion (hexyl substituents are omitted for clarity). (c) Schematic representation of electron tunneling exciting the molecule and inducing translational motion on the surface. (d) Cartoon representation of the motion. Reprinted with permission from ref (1135). Copyright 2011 Nature Publishing Group. Directional rotary motors undergoing well-defined structural and/or chemical changes on the application of an external stimuli were discussed in section 5 . Feringa et al. developed a system utilizing paddle-wheel type directional motion over a Cu(111) surface upon sequential electronic and vibrational excitation (Figure 40 ). 1135 Electron tunneling induced by a STM tip stimulated double bond isomerization, which was followed by helicity inversion. These sequential configurational and conformational changes in the molecule propelled it across the surface. For the molecule to move on the surface linearly, the four rotary wheels had to rotate in the same direction upon application of the voltage. This is only possible when the molecule was attached to the surface as the meso-(R,S–R,S) isomer. In other isomers, a lack of coherent rotation prevented translational motion or induced spinning. Hao et al. monitored the strength of metal–ligand complexation with an AFM tip. 1136 One terpyridine ligand was attached to a gold surface while another was tethered to a gold-coated AFM tip, and the bonding forces between osmium and the ligands were analyzed. The use of atomic tools to induce relative motion in an interlocked molecular system is of particular interest in the development of molecular machines, because each component can be designed to function like the mechanical components of a macroscopic machine. 1137,1138 In a polyrotaxane architecture adsorbed on MoS2, cyclodextrin ring was translocated by a STM tip along a polyethylene glycol thread. 1139 Such shuttling was thought to be responsible for the conductance switch observed upon the application of a voltage via a STM tip in a bistable [2]rotaxane. 1140 The potential use of interlocked molecular constructs in information storage was highlighted by Leigh et al., who showed that regular patterns of deformations could be successfully generated on a rotaxane monolayer with the aid of a STM tip. 1141 These arrays formed due to the relative ease of intercomponent mobility in these molecular constructs. The effect of electrostatic and steric forces on the shuttling of a bistable [2]rotaxane tethered to an AFM tip has been investigated both experimentally and theoretically. 1142 A molecular shuttle made up of fumaramide and succinamide stations has been attached to a gold surface by thiol linkers (Figure 41 ). 1143 The macrocycle was tethered to an AFM probe using poly(ethylene oxide) (PEO). Strong hydrogen-bonding interactions with the fumaramide station resulted in a 95:5 distribution of macrocycle in favor of this station. As the stretching of the PEO tether continued, and the force exerted by the PEO linker exceeded the hydrogen-bonding forces between the macrocycle and the fumaramide station, the ring moved away from the fumaramide station. Tension in the tether decreased as a result of the shuttling. Further movement of the cantilever resulted in the detachment of the PEO linker from the probe. Figure 41 (a) Chemical structure of a rotaxane with fumaramide and succinamide stations depicted in green and orange, respectively. (b) Schematic representation of macrocycle movement on a thread attached to a gold surface as a result of the force exerted by an AFM probe. Application of force by cantilever movement (I,II) was followed by repositioning of the macrocycle (III) or detachment of the PEO tether (IV) depending on the strength of the force. Reprinted with permission from ref (1143). Copyright 2011 Nature Publishing Group. In addition to molecular level motion generated by atomic tools, modulation of conductivity by molecular switches attached to nanojunctions is of great interest. 1144−1152 A STM tip could cause the rotation of selected moieties in the structure of a polyaromatic scaffold, in the hope of developing switchable nanowires. 1110,1153−1160 Although most examples are still far from practical application, significant progress has been made in this area. 1161−1182 8 Toward Applications of Molecular Machines 8.1 Current Challenges: Constraining, Communicating, Correlating Extracting useful work at the molecular scale requires the restriction of the thermal movement of submolecular components or the exploitation of thermal motion with additional ratcheting. Shuttling, switching, and rotation processes in solution can be modulated externally, and the directionality of each motion can be controlled in single molecules. Considering an ensemble of such molecules in solution, however, the biased motions average to give no net directionality. For various applications, the integrity of the molecular system must be conveyed to the macroscopic world. Although there are possible applications of molecular machines/devices in solution, such as their use in molecular logic gate construction or molecular sensing, they are not compatible with typical solid-state technology. For this reason, molecular machines on solid supports are needed. This challenge is being addressed with molecular devices and machines being built on surfaces, interfaces, and polymer matrixes with a variety of electronic, mechanical, or biological applications. 1183−1209 Controlled drug release through nanovalves, 1210−1226 molecular electronics, 1227−1234 artificial molecular muscles, 18,1224,1235−1242 information storage, 1141,1243,1244 and modulation of surface properties 1245 are topics of active research in this area. Figure 42 Communication between molecular devices. Acid generated upon conversion of merocyanine (MEH+) to spiropyran (SP) protonates a pyridine, and leads to subsequent complexation of the pyridinium ion (103 + ) by a calix[6]arene (104). 1246 Communication between molecular machines is another pertinent challenge in this area. In biological systems, the work done by one machine can be harvested by another and the second can then operate. Raymo and Credi have developed a system that addresses this challenge. 1246,1247 They used a merocyanine-type photoacid to release protons as the molecule transforms from its open form to the closed form. The acid protonated a terpyridine osmium complex and decreased the efficiency of singlet oxygen generation. 1247 These photogenerated protons were also used to reversibly complex and decomplex a 1-alkyl-4,4′-bipyridin-1-ium (103 + ) in a calix[6]arene wheel (104) (Figure 42 ). 1246 Thus, the control of one moiety caused a change in another via the transfer of a proton. Recently, Aprahamian and co-workers successfully coupled a hydrazone switch (section 2.2 ) with a merocyanine unit and used the photogeneration of acid to drive the switch reversibly with high efficiency. 1248 8.2 Reporting Controlled Motion in Solution Submolecular movement can be designed to give a detectable output. The output can be used as a measure of stimulant concentration (sensing) or it can be used for information storage. In theory, any detectable nondestructive output that provides a reliable distinction between states of the system is acceptable (NMR, CD response, etc.). However, for most useful applications, more practical and easily readable optical, electronic, or mechanical outputs are preferred. The response rate can be crucial in certain applications such as memory devices and molecular sensors. In these systems, fast-responding reporters are preferred. For a molecular system to be used repeatedly, stability and fatigue resistance are important. 8.2.1 Conformational Switches in Solution The restriction of molecular motion or switching between bistable conformations can be achieved by external stimuli such as ion or organic molecule binding, pH change, or by photoisomerization of the molecule. Conformational perturbations in polymers or carbon chains with interesting properties have been reported. 1249−1261 Fluorescence readouts have been widely used to detect such submolecular motions in part due to their high signal-to-noise ratios. 16,1262−1266 One way to obtain a fluorescence response is by changing the electron transfer efficiency between a fluorescent component and other moieties in the system by inducing an electronic change in the acceptor or donor units. Ligand binding or protonation can effectively change the HOMO and/or LUMO levels of molecules, which in turn affects electron transfer to or from the fluorescent moiety. The change in relative energy levels upon reduction or oxidization determines the ease of detection of the fluorescence response. A greater response can be obtained by decreasing rotational degrees of freedom or by energy transfer. Because energy transfer processes are highly distance dependent, variations in distance between the donor and acceptor units in different conformational isomers can determine the efficiency of energy transfer. Emission from dimers of certain chromophores is also widely used as a reporter of molecular processes. This emission requires the close proximity of similar (excimer) or different (exciplex) chromophores. Scheme 37 Expansion and Contraction Allosterically Controlled by Ligand Coordination The Zn(II) center acts as a catalytic unit, and diethylaminomethylanthracene is used as the reporter. 1267 Mirkin et al. developed an allosterically regulated supramolecular catalyst based on ligand-dependent contraction or expansion to reveal or conceal a catalytic unit (Scheme 37 ). They were able to measure the conformational change indirectly using the reaction byproduct, acetic acid. 1267 The tetrametallic species was made of two catalytic cofacially aligned Zn(II) centers attached via two Rh(I) centers. The rate of catalysis of the acyl transfer reaction by the core Zn(II) ions was enhanced when the cavity between the two units was larger. Exchange of Rh(I) ligands from labile thioethers to chloride and carbon monoxide led to an expansion and thus an enhancement of catalytic activity. As the acetic acid byproduct protonated the amine of diethylaminomethylanthracene, an increase in fluorescence was observed caused by blocking photoinduced electron transfer from the amine moiety. Lee et al. developed a conformational switch based on a cyclic hydrogen-bonding network in a tris(N-salicylideneaniline) derivative (Figure 43 ). 1268−1271 Structural distortion induced by folding and unfolding processes resulted in a significant change in the electronic properties of the system. BODIPY chromophores were tethered to the molecule as fluorescent energy acceptors. In the folded form, the molecule adopted a highly ordered hydrogen-bonding array, and energy transfer took place from the core tris(N-salicylideneaniline)moiety to the BODIPYs. When a hydrogen-bond acceptor such as fluoride was added, the molecule unfolded, and a substantial decrease in the fluorescence of the BODIPY dyes was detected. Because each mobile element cooperated in the folding process via hydrogen bonding, conformational change was highly cooperative. The initial fluorescence intensity could be recovered when the fluoride anions were captured by trimethylsilyl chloride. Figure 43 (a) Chemical structure of the t-butylphenyl and BODIPY-substituted foldamers, and (b) X-ray structure of the t-butylphenyl functionalized foldamer. Carbon atoms of the t-butyl groups and all hydrogen atoms except OH and NH have been omitted for clarity. (c) Schematic representation of conformational switching. D and A represent energy donor and acceptor modules, respectively. 1268−1270 Parts (a) and (c) are adapted with permission from ref (1268). Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Part (b) is reprinted with permission from ref (1269). Copyright 2006 American Chemical Society. A number of different fluorescent sensors reporting conformational changes have been synthesized on the basis of energy transfer, 1272−1274 excimer formation, 1275−1282 or charge transfer mechanisms. 1283−1286 Optical responses have been obtained by simple stimuli-dependent conformational change, 796,1262,1273,1275,1287−1289 folding processes, 1276−1278,1290 shuttling, 799,1272,1283−1285,1291−1294 switching, 1273 and transitions between coiled and helical structures. 1295 Optical outputs have been used to develop molecular logic gates. 1291,1292,1296 Recently, Tian et al. used both phosphorescence and an induced circular dichroism output to develop an INHIBIT molecular logic gate by taking advantage of the photoisomerization-dependent inclusion of an azobenzene and a bromonaphthalene in the β-CD cavity. 1297 Figure 44 (a) Ligand-induced variation of the chemical shifts (11B (160 MHz) and 1H NMR (500 MHz)) of the helix in CD3OD at 298 K. (b) Schematic representation of fast exchange between two degenerate helical conformers with a single NMR signal and (c) induced bias of helicity upon ligand binding (yellow), and the anisochronous NMR signal generated. Adapted with permission from ref (1298). Copyright 2013 Nature Publishing Group. Techniques other than optical responses have been used to measure conformational change. Flood et al. used a change in conductance to directly measure a folding process because unfolding released a bound chloride anion and changed the conductance (Figure 22 in section 3.1 ). 595 A change in CD response can be a useful indicator of molecular motion and helicity inversion. 1298−1308 Many shape changes can be easily analyzed by NMR spectroscopy. Recently, Clayden et al. showed, by 11B, 1H, and 13C NMR, that the equal distribution of left-handed (M) and right-handed (P) conformers of a helical foldamer could be perturbed by the addition of a chiral ligand (Figure 44 ). 1298 Upon ligand complexation to the boronic ester, chiral information was transferred along the helix, and the isochronous NMR signals of the diastereotopic nuclei were transformed into anisochronous signals, indicating a bias in the distribution of helical conformers. When the meso diastereoisomer of the diol was used to complex the boronic ester, no splitting of the 1H NMR signals was observed due to the symmetry of the ligand. 8.2.2 Rotaxane Switches in Solution Fluorescence spectroscopy has been widely used to detect shuttling processes in interlocked systems, particularly in rotaxanes. The strong distance dependence of fluorescence quenching by electron transfer has frequently been exploited to measure the relative position of molecular subcomponents. Photoisomerization-dependent shuttling of a pyridinium bearing macrocycle along a two station thread was monitored by changes in the fluorescence of an anthracene group tethered near one station. 1285 When the distribution of macrocycle was biased to this station, charge transfer from the anthracene to the pyridinium moiety quenched the fluorescence, thus providing a detectable fluorescence response due to macrocyclic position. The location of a macrocycle along an alkyl chain in a rotaxane structure has also been probed using similar techniques. 1272 Greater rotational and vibrational freedom leads to a greater probability of nonradiative processes and hence a lower fluorescence quantum yield. Any interaction affecting rotation or vibration can influence emission intensity. Tian et al. used this effect with a 4-aminonaphthalimide fluorophore in a stilbene bearing rotaxane where the steric hindrance generated by photoisomerization of the stilbene unit resulted in the relocation of the α-cyclodextrin macrocycle along the thread. 796 As a result of this movement, fluorescence increased by 46%, which was attributed to the restriction of molecular motion of the nearby linker moieties. The same group utilized an azobenzene as an additional photoisomerizable moiety to develop a molecular half adder logic gate (Scheme 38 ). 1291 This logic gate performs binary addition using two inputs and two outputs to provide both XOR and AND gates. When either the azobenzene or the stilbene moiety was isomerized, fluorescence emission of the naphthalimide close to photoisomerized species increased ((Z,E)-112 or (E,Z)-112). However, if none or both of them were isomerized ((E,E)-112 or (Z,Z)-112), fluorescence intensity decreased due to rapid shuttling of the ring between the stations in the nonisomerized form or because the ring was trapped in the center of the thread when both functional moieties were isomerized. These fluorescence outputs collectively result in a XOR gate (output 1 in the truth table). The photoisomerization-dependent decrease in absorbance at 350 nm and increase at 270 nm were used as an additional output for the construction of the AND gate of the half adder. The change in absorbance at the isosbestic point (301 nm) was followed and only breached a predetermined threshold when both units were isomerized. As such, an interlocked molecular system was successfully used as an all-photonic logic gate in solution where the inputs induce submolecular motion and outputs are reporters of their motion. Acid–base and redox switchable bistable rotaxanes with crown ether macrocycles have been used to create an INHIBIT logic gate. 1292 Room-temperature phosphorescence and CD responses have been used in a pseudo[1]rotaxane to create a primitive logic gate. 1297 Scheme 38 Photoisomerization-Dependent Shuttling of an α-Cyclodextrin on an Azobenzene and Stilbene Bearing Thread with Two Naphthalimide Derivatives as Fluorescent Stoppers The percentage of the major isomer in the photostationary state is shown over the reaction arrows. The truth table for a half-adder logic gate is shown, with the inputs being 380 nm (I1) and 313 nm (I2) irradiation, and outputs being the change in absorbance (O2) and fluorescence (O1) values. 1291 Indicator displacement assays involving supramolecular complexes and interlocked molecules are widely used for molecular sensing applications. 1309 In these systems, a fluorescent molecule with complexation-dependent fluorescence properties is usually used as a guest. The optical change generated upon exchange of the fluorescent guest with the analyte is used as the reporting function. Smith et al. used a similar idea to develop a chloride anion sensor based on a rotaxane (Figure 45 ). 1310,1311 A [2]rotaxane (113) was synthesized using a tetralactam macrocycle and a squaraine dye as building blocks. Tetralactam macrocycles are known to bind to and quench the emission of red light from squaraines. 1312 The exposure of this dye, upon chloride-induced displacement of the macrocycle, restored fluorescence. The process could be reversed by chloride precipitation as the NaCl salt. When the same rotaxane was adsorbed on a C-18 coated reverse-phase silica gel plate and dipped into different concentrations of aqueous solutions of chloride, the change in fluorescence was large enough to be visually observable. Later, a ratiometric chloride sensor based on the same squaraine rotaxane was reported. 786 Interlocked systems are useful in such dye-displacement assays, as the displaced dye typically diffuses away preventing reversibility, whereas the stable rotaxane structure allows reversibility and reusability. In a similar example, the sodium ion-dependent shuttling of a [2.2.2]cryptand away from a squaraine station enhanced squaraine emission. 1312 Interlocked architectures have been mounted on metal nanoparticles and shown to keep their switching ability. 1313 Figure 45 (a) Chloride-dependent shuttling of tetralactam macrocycle, and (b) subsequent fluorescence enhancement in CHCl3 upon titration with tetrabutylammonium chloride. (c) Rotaxane solution in the absence (left) or presence (right) of chloride. Adapted with permission from ref (1310). Copyright 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Liu et al. have developed a [2]pseudorotaxane with a dual stimulus luminescent lanthanide switch using a diarylperfluorocyclopentene as the photochromic component (Figure 46 ). 1314 The diarylperfluorocyclopentene (114) bears a benzyl ammonium recognition unit and reacts reversibly to generate its closed form upon absorption of UV irradiation (365 nm). The open form can be obtained via irradiation at 614 nm. The Eu3+ complex of terpyridinyldibenzo-24-crown-8 (115) was used to reversibly complex the ammonium moiety of 114. This Eu3+ center fluoresces at 619 nm through intramolecular energy transfer from an excited terpyridine ligand. Upon complexation of the two compounds through crown ether–ammonium interactions, a small amount of quenching of the lanthanide fluorescence at 619 nm was observed (ca. 10%) due to the poor spectral overlap between the donor emission and acceptor absorbance (which is required for resonant energy transfer (RET)). Upon UV irradiation to form the closed form of compound 114, spectral overlap increased and fluorescence was quenched by 80%, with an accompanying 3-fold decrease in excited-state lifetime. The pseudorotaxane could be reversibly disassembled by the addition of potassium cations and regenerated by the addition of 18-crown-6. Figure 46 (a) Chemical structures of 115, and the open and closed forms of 114. (b) Schematic representation of light modulated switch and K+/18-crown-6-mediated complexation. (c) Fluorescence quenching of the Eu3+ complex upon UV irradiation in 1:1 CH3CN/CHCl3. Fluorescence before (I) and after (II) excitation at 390 nm (c inset). Reprinted with permission from ref (1314). Copyright 2013 American Chemical Society. The responsive nature of induced circular dichroism makes it an interesting phenomenon to study with the CD response of one species changing upon its interaction with another. 387,1315−1319 The transformation of an achiral carbon center to a chiral one upon hydrogen-bonding-dependent symmetry breaking provided an early example of the influence of supramolecular interactions on chirality. 1320 A number of chiral supramolecular assemblies, 1315,1316,1321 chiral spaces (dissymmetric cavities), 1322 and helices 1323−1325 have been reported. The solvent-dependent rearrangement of a macrocycle along a thread has been shown to alter the CD response of a rotaxane (section 4.3.1 ). 715 The position of a macrocycle has been used to control the reactivity of a secondary alcohol toward esterification in the presence of a chiral catalyst in a chemical information ratchet (section 4.4.1 ). 809,810 Similarly, a chiraloptic switch has been reported based on a benzylic amide macrocycle and fumaramide/peptide stations (Figure 47 ). 1285,1326 Photoisomerization-dependent relocation of the macrocycle to the peptide station altered the CD response due to proximity-dependent macrocycle–leucine residue interactions. A large reversible difference in elliptical polarization response (>1500 deg cm2 dmol–1) was obtained upon isomerization of the fumaramide station from E to Z. Figure 47 (a) Chiraloptical switching upon photoinduced shuttling of the macrocycle between fumaramide (green) and peptide (orange) stations; the chiral center of the peptide station is highlighted by a black circle. (b) Percentage of E isomer calculated using 1H NMR (left y axis) and induced CD absorption at 246 nm after alternating irradiation between 254 nm (half integer) and 312 nm (integers) (right y axis). Reprinted with permission from ref (1326). Copyright 2003 American Chemical Society. Li et al. synthesized rotaxane 117 bearing a tetracyanobutadiene (TCBD) stopper, which tends to form aggregates via intermolecular dipole–dipole interactions, to attempt to control this clustering behavior (Figure 48 ). 1327 The position of macrocycle relative to the TCBD group determined the strength of intermolecular interactions and hence the structure of the assemblies formed. In a hexane:chloroform mixture (1/1, v/v), the macrocycle rested on the peptidic station, and the TCBD units were free to interact to generate nanofibers (Figure 48 b). In methanol:chloroform (1/1, v/v), the amphiphilic nature of the molecules generated perforated nanocapsules (Figure 48 c). Finally, when DMSO was used as the solvent, the macrocycle relocated closer to the TCBD group, interfering with aggregation and generating a worm-like nanostructure (Figure 48 d). A similar phenomenon was reported in which a fumaramide bearing rotaxane displayed different nanostructures when the macrocycle bound Zn2+ metal in the presence or absence of irradiation. 1328 Anion and acid/base-dependent control over the formation of an organogel has been achieved using a [2]rotaxane architecture. 1329 In this system, stimuli-induced shuttling of the macrocycle between urea and ammonium stations led to sol–gel phase transitions. Polyrotaxane structures have been formed in a concentration-dependent manner in solution using the interaction of an electron-deficient macrocycle with a monoanionic species. 1330 Recently, light and acid/base-dependent threading and dethreading of pseudorotaxane structures embedded in nanofibers was reported to induce a macroscopic change (more than 1.5-fold enhancement of Young’s modulus upon dethreading). 1331 Figure 48 (a) Solvent-induced shuttling on a tetracyanobutadiene bearing rotaxane 117. (b) Proposed assembly of rotaxane and SEM images in CHCl3/n-C6H14 (1/1, v/v), (c) in CHCl3/MeOH (1/1, v/v), and (d) in DMSO. Reprinted with permission from ref (1327). Copyright 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. The concept of a “molecular muscle”, created from doubly threaded rotaxanes or interlocked daisy chain molecular structures, was developed by Sauvage et al. 1332 In these systems, submolecular movement drives a relative motion that either contracts or expands the entire molecule upon metal exchange, 1332 solvent exchange, 1333,1334 redox reaction, 1236,1238,1335 or acid/base exchange. 18,1336−1339 Photoinduced contraction/expansion has also been reported. 1340 The concept has been used to cause the macroscopic bending of a microcantilever by means of redox-driven shuttling of the surface-adsorbed macrocycle along a thread (section 8.6.4 ). 1238,1341 The design of enzyme-responsive molecular machines is an emerging area and could find applications in biotechnology. Biodegradable polyrotaxanes have been studied as selective drug and gene delivery systems. 1342−1349 Leigh and co-workers published two generations of [2]rotaxane-based propeptides (118, 119) where the peptide axle is protected from degradation from general peptidases by the macrocycle. 1350,1351 The introduced glycosidase-cleavable stopper allows the release of the free peptide in a controlled fashion through treatment with a specific glycosidase. The sequence of reactions triggered by the β-galactosidase leading to the decomposition of the stopper is shown in Scheme 39 . This approach offers a promising alternative delivery system for peptide-based therapies, because many bioactive peptides suffer from in vivo instability and poor bioavailability. Similar principles have been used to deliver a deactivated cancer drug (acting as a stoppering moiety in a rotaxane) via the blood before the macrocycle immolates inside a cell. This releases the thread, and enzyme-mediated hydrolysis furnishes the active drug molecule. The rate of hydrolysis in cancerous cells versus benign cells was somewhat enhanced due to their overexpression of the hydrolyzing enzyme. 1352 Scheme 39 β-Galactosidase-Triggered Release of Triglycyl Peptide 120 from the [2]Rotaxane Propeptide 118 and 119(1351) 8.3 Synthetic Molecular Walkers 8.3.1 Introduction Biological motor proteins have evolved a plethora of functionalities. DNA, RNA, and protein synthesis machineries (DNA polymerases, RNA polymerases, the ribosome) work cooperatively with helper proteins to “unzip” molecules (helicases, 1353 poly(ADP-ribose) polymerases), 1354 release strain in a substrate (gyrase, 1355,1356 topoisomerase), 1357,1358 slide over the substrate, and processively synthesize a product (DNA polymerases, RNA polymerases, the ribosome). 68−70 The ATPase rotary motor is the energy factory of the cell producing ATP via complex mechanical processes. 1359 Pumps 1360 and pores 1361 are essential to maintain communication and transportation across the membranes of different compartments (functioning at organelle, cell–environment, or cell–cell boundaries). Among these machines, molecular walkers are attracting increasing attention as their mechanism of action has been revealed. Dynein, myosin, and kinesin are walker proteins of the ATPase family and differ in their structure, function, and use of energy. 1362−1365 Proteins such as collagenase and exonucleases, which are not traditionally viewed as walker proteins, migrate along their substrate tracks by destroying the track via a “burnt-bridges” mechanism. 1366−1368 Biological walkers have some important characteristics. 1369 First, they must be supplied with energy to do work against random Brownian fluctuations. ATP hydrolysis normally provides this energy. Second, attachment to a track results in a restriction of their degrees of freedom facilitating directional 1-D or 2-D walking in solution. Inertia and momentum under conditions of low Reynold’s number are irrelevant, and work is performed under the influence of viscous and thermal motions. Third, to drive the walker away from equilibrium, that is, to generate directional motion, a ratcheting process (either an energy or an information ratchet) must take place. In addition to the requirements of a Brownian motor, certain additional characteristics are necessary for a motor to be defined as a walker. (i) Processivity is a measure of the integrity of walker–track interactions during its operation. A walker should remain attached over multiple operations to extract useful work. (ii) Directionality is the exclusive or preferential movement of the walker in one direction along the track. (iii) Repetitive operation means the motor should be able to carry out similar mechanical cycles repetitively. (iv) Progressive operation is the ability of the motor to culmulatively perform work with each operating cycle. (v) Autonomous operation allows the motor to function continuously without external intervention, as long as an energy source (fuel) is available. Conventional kinesin (kinesin I) meets all of these criteria and was first isolated by Vale et al. in 1985. 1370 It is a homodimeric protein with two identical heads that interact with a microtubule track. Through cyclical binding, hydrolysis, and then release of ATP, it can walk along the track by an asymmetric hand-over-hand mechanism. 1371−1373 Because at any time one of the heads remains attached to the track during the walking process, processivity is high. On average, this walker takes 100 steps before detaching from the track. 1374−1377 Although walkers with two attachment points or more are common, this is not a requirement for processivity. The KIF1A kinesin protein processively walks along microtubules using a single leg. 1378 In contrast, some 2-legged biological walkers such as myosin-II are nonprocessive. 1365 Even without being processive, myosin-II can still move directionally, and in a large ensemble it can generate macroscopic motions such as muscle contraction. DNA has the notable ability to form predictable hydrogen bonds with a complementary strand and can be synthesized using machine-assisted technologies. Strand displacement provides a versatile design for walking processes, and the ability to make complex 3-D structures can be exploited to perform complex physical tasks. The first artificial walkers were made of DNA. Directional walkers, 1379 autonomous walkers, 1380 walkers using the burnt-bridges mechanism, 1380 cargo-carrying walkers, 1381 and a light-driven walker 1382,1383 based on DNA have been reported, and will be examined in section 9.4 . Aside from DNA-based walkers, the preferential diffusion of molecules along the higher symmetry axis of a metal surface has also been observed. 1117,1127 Huskens et al. reported the gradient-driven diffusion of molecules bearing two adamantane legs across α-cyclodextrin-functionalized surfaces. 1384 Processive crown ether migration on an oligoglycine chain 1385 and migration of metal atoms on aromatic scaffolds 615,1386,1387 have also been explored. Rearrangement reactions enable the migration of a fragment along a molecule (such as the Claisen 1388,1389 and Cope 1390−1393 rearrangements). However, progressive and repetitive operation using sigmatropic rearrangement reactions is generally challenging. For the design of processive small molecule synthetic molecular walkers, mechanically interlocked architectures are good candidates, because the walker (macrocycle) is mechanically bonded to the track (thread). However, their interlocked-structure prevents resetting because each resetting process undoes the walking process, by relocating the walker to its original position. 1369 Moreover, the macrocycle cannot choose a new path without bond breaking, which would result in detachment from the thread. A two-leg system operating under orthogonal conditions could provide a viable synthetic molecular walker. Orthogonal dynamic covalent exchange reactions can provide a suitable balance of reversibility and kinetic stability. A rigid track can decrease the possibility of overstepping. 8.3.2 Spontaneous Walking of Small Synthetic Molecular Walkers Reversible and processive migration of a Michael acceptor along a protein was reported in 1979 by Lawton et al. 1394 More recently, Lehn et al. used dynamic imine exchange to transport a salicylidene walker 125 along an amine bearing track (Figure 49 ). 1395 Deprotection of the amine at one side of the track using methoxyamine initiated the dynamic exchange reactions. The speed of exchange was shown to be modulated when substitution, composition of the solvent, or temperature was altered. 1H NMR analysis in CDCl3 proved that the relative intensity of a characteristic signal corresponding to the transported walker (H6) increased gradually and in 2 days the transported walker was the dominant product in solution. Recently, the same group reported a modified version of their walker with thermodynamic sinks at each end of the track. These trap the stochastic walker at one end in acid as an imine, or as a lactone at the other end under basic conditions. 1396 Figure 49 (a) The structure and operation of walker 125, which uses dynamic imine exchange chemistry. Amine footholds are highlighted in blue and red. Each molecule is labeled with one or two numbers in parentheses indicating the amine footholds to which the walker is attached (foothold amines are assigned with numbers starting from the left). (b) 1H NMR spectra of indicated protons in CDCl3. H7 corresponds to a side product in which the walker is detached from the track. Reprinted with permission from ref (1395). Copyright 2012 American Chemical Society. Figure 50 (a) The chemical structure of the model walker 127, and (b) gradual change in 1H NMR of the model walker in d 6-DMSO upon formation of new positional isomer on the track. Ratio of (1):(2) isomers reaches 1:0.9 after 15 h of operation. (c) Chemical structure of the walker on a larger track bearing an anthracene moiety, 128. (d) Fluorescence quenching of anthracene after 6.5 h of walking. Reprinted with permission from ref (1397). Copyright 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Figure 51 (a) Chemical structure of walker 129, with each positional isomer labeled in a different color, and amines numbered. (b) Change in the partial 1H NMR spectra over 48 h of operation. (c) Occupancy of each foothold over time. Reprinted with permission from ref (1398). Copyright 2013 American Chemical Society. Leigh et al. reported a spontaneous walking process using a Michael addition reaction between an α-methylene-4-nitrostyrene walker and a polyamine track (Figure 50 ). 1397 It was shown that the walker translocated preferentially through successive 1,4-N,N-migration rather than by overstepping. After 48 h, no out of sequence N,N-migration was observed with a model walker (Figure 50 a,b). 1H NMR analysis showed that the walking was highly solvent dependent and accelerated in polar solvents such as dimethylformamide and dimethyl sulfoxide. In d 6-DMSO, the walker reached an equilibrium distribution in 15 h. A ratio of 1:0.9 between the occupancy of initial position and the central amine of the track was achieved over this time period. Intermolecular exchange of less than 6% took place when the model walker was mixed with a longer free track for 3 days. This indicates a high processivity (on average 530 steps taken before detachment). Using a longer track modified with an anthracene at the far end, it was possible to monitor walking via an observed decrease in fluorescence as the walker approached the anthracene. As the nitrostyrene walker approached the edge of the track, it quenched the fluorophore (Figure 50 d). Leigh et al. have since reported an extended system with a naphthalene moiety at one end of the track (Figure 51 a). 1398 In the presence of excess base (iPr2NEt), the walker was “trapped” by the naphthylmethylamine, which was the thermodynamic sink. 1H NMR analysis showed that the steady-state distribution of the walker was biased and 46% of walker molecules had reached the final benzylic foothold in 48 h (Figure 51 b,c). When a longer track (with 9 footholds) was used, the percentage at the final station dropped to 19%. Scheme 40 Structure and Operation of a Small Molecule Walker, Walking along a Molecular Track Footholds are shown in blue and green; the walker unit is depicted in red. Numbering shows the footholds to which the walker is attached. 1399 A walker with two chemically different legs that labilize under orthogonal conditions has been reported by Leigh et al. (130, Scheme 40 ). 1399 The walker unit was attached to the thread by hydrazone and disulfide linkages. Hydrazone exchange took place under acidic conditions, while under these conditions the disulfide bond was stable. Disulfide exchange required basic conditions, which did not affect the hydrazone bond. Successive acid–base cycles led to a 39:36:19 (1,2:2,3:3,4) distribution of walker, with only 6% forming the overstepped (1,4) isomer. Changing the length of the walker alkyl chain from 6 carbons to 5 and using oxidation instead of base exchange at the final step, the distribution could be biased in favor of the (3,4) isomer, which was attributed to strain in the (2,3) isomer. 1400 Recently, Bayley et al. monitored an organoarsenic molecule walking along a cysteine bearing track in a protein pore, by measuring changes in ionic current. The walker “stepped” by thiol exchange reactions and displayed a limited degree of processivity and directionality (the latter due to the thermodynamic sink at the track terminus). 1401 An inchworm walker able to walk on a copper surface has been reported. It could be constrained to one dimension of motion by the use of oligomeric “fences”. 1402 8.3.3 Directional Synthetic Small Molecule Walkers The biased migration of molecules along a track requires either a preference for one isomer over others under the operating conditions or that strain should be released upon stepping. An efficient ratcheting step is needed. Recently, the use of metal complexes in a molecular biped was explored by Leigh et al. (Figure 52 ). 1403 The track contained three different pyridine derivatives: 2,6-dialkyl-4-N,N-dimethylaminopyridine (foothold 1, green), 3,5-dialkylpyridine (foothold 2, red), and 2,6-dialkylpyridine (foothold 3, blue). The walker consisted of pincer ligands able to complex the pyridine foothold via Pd(II) or Pt(II) centers. Initially, the walker was attached to the track by complexation of Pd(II) with foothold 1, and Pt(II) with foothold 2. Upon protonation of the free 2,6-dialkylpyridine foothold (foothold 3) and heating, exchange of the Pd(II) complex between foothold 1 and foothold 3 took place. A distribution of 15:85 in favor of foothold 3 was observed. The proton on foothold 3 was captured by the more strongly basic foothold 1 in this process. Migration could be reversed when the first foothold was deprotonated and the solution was heated. A ratio of 95:5 in favor of the initial position was obtained. Even though this example is not truly a walker, the significant directional bias in the stepping process of the metal-based biped could be used to inform more advanced designs. Figure 52 Toward directional molecular walkers. (a) Chemical structure and schematic representation of the walker attached to the track, and (b) operation of walking through successive acid–base addition and heating cycles. Reprinted with permission from ref (1403). Copyright 2014 American Chemical Society. Finally, a small molecule walker displaying all of the desired properties of an artificial walker, save autonomy, has been reported by Leigh et al. (Scheme 41 ). 1404 The walker was based on the previously published 130 (Scheme 40 ), but a stilbene moiety now linked footholds 2 and 3 (Scheme 41 ). The walker operated by the same hydrazone and disulfide exchange reactions, but with photoisomerization of the stilbene moiety driving directionality. Isomerization led to greater proximity between footholds 2 and 3 (over 1 and 2) in the cis form of the stilbene, resulted in a steady-state distribution of 40:60 in favor of the (2,3) positional isomer. During the second hydrazone exchange step, reisomerization to trans stilbene moved footholds 2 and 3 apart, which led to almost exclusive (5:95) formation of the (3,4) isomer. Scheme 41 Light-Driven Transport of a Molecular Walker along a Molecular Track Footholds are shown in blue and green; the walker unit is depicted in red. Each molecule is labeled with two numbers in parentheses indicating the two footholds to which the walker is attached. (a) (i) hν (365 nm), CD2Cl2, (ii) DBU, DTT, CHCl3; (b) (i) I2, hν (500 nm), CD2Cl2, (ii) TFA, CHCl3. 1404 8.3.4 Challenges Yet To Be Overcome Significant progress has been achieved in the synthesis of synthetic molecular walkers with highly processive, progressive, and in some cases directional walkers being reported. The exploration of additional, reversible, walker–track interactions may lead to greater control over the directionality of motion. Autonomous walking remains a challenge to be addressed. Polymeric tracks could be designed for cargo transport over large distances. The ability to immobilize walkers on surfaces could lead to future technological applications. 8.4 Switchable Catalysts In the living cell, many processes and reactions occur in parallel. To make sure that these reactions and their products do not interfere with each other, these operations must be rigorously controlled and switched on or off when necessary. In nature, enzymes control the outcome and rate of the large majority of reactions taking place in a cell. Recently, a number of biologically inspired systems have been published using molecular machines to control the outcome of reactions by switching on and off catalytic units or by controlling the enantioselectivity of the process. 1405−1412 In this section, switchable catalysts controlled by stimuli such as light, pH change, ion influx, and redox chemistry will be explored. 8.4.1 Photoswitchable Catalysts Several examples of photoswitchable catalytic systems have been reported. 1407,1413−1415 The reactivity of the catalytic unit is typically controlled through alteration of the steric shielding of the active site or by bringing the catalytic units into closer proximity. In these examples, switching between the cis and trans forms of a photochromic group such as an azobenzene or a stilbene, or the electrocyclization of a diarylethene, provides the requisite control. An early example of a photoswitchable catalyst was published by Rebek Jr. and co-workers. 1416 Two adenine receptors capable of forming complexes with purine bases were linked through a trans azobenzene moiety. When the system was isomerized to its cis conformation, the rate of reaction between an amine and p-nitrophenyl ester was markedly increased due to greater proximity between the reactive groups, which increased the effective molarity of these groups and thus increased the rate of reaction. A switchable catalyst based on a light-responsive cavitand was published by the same group. The cavitand was functionalized with an azobenzene switch and the cavity of the trans state accessible to a guest molecule, while the cis conformation could not bind guests. Piperidinium acetate was used as the guest, and it could be shown that the host–guest complex significantly accelerated a Knoevenagel condensation between malononitrile and an aromatic aldehyde. 1417 Hecht and co-workers described the photoswitchable tertiary piperidine derivative 133, which could function as a general base catalyst in a Henry reaction between p-nitrobenzaldehyde and nitroethane (Figure 53 ). Catalysis was regulated through the shielding/deshielding of the catalytic site upon irradiation of an azobenzene unit. 1418 The reversible shielding and deshielding of a catalytic site by photoisomerization has also been used to control the organocatalytic activity of a thiourea and a guanidine catalyst. 1419,1420 Figure 53 (a) Molecular structure of the photoswitchable piperidine base 133, and (b) X-ray structure of the photoswitchable piperidine base 133. X-ray crystal structure reprinted with permission from ref (1418). Copyright 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. The photoswitchable organocatalyst 134 consisting of a photochromic diarylethene (DAE) unit and an imidazolium salt was reported by Bielawski and co-workers. 1421,1422 Under ambient light and in the presence of base, the imidazolium species catalyzed transesterification and amidation. These reactions were considerably slowed upon photoinduced formation of the ring-closed isomer (Scheme 42 ). 1421 Scheme 42 Ring-Opened and Ring-Closed Isomers of the DAE-Modulated Photoswitchable Organocatalyst 134(1421) Exploiting the ability of dithienylethene to switch between ring-closed and ring-opened isomers upon irradiation, Branda et al. designed a photo responsive system mimicking enzyme cofactor pyridoxal 5′-phosphate 135 (PLP) (Scheme 43 ). PLP is responsible for amino acid metabolism and participates in a diverse range of enzymatic reactions such as transamination, racemization, decarboxylation, and various elimination processes. The structural features responsible for the action of PLP are the conjugated aldehyde and pyridinium functional groups. Branda and co-workers replaced the core ring of PLP with a dithienylethene where, in the ring-open form, the pyridinium and aldehyde units were electronically isolated from each other preventing catalytic activity (138-inactive). However, photoirradiation to form the ring-closed (138-active) isomer resulted in a fully conjugated structure, which restored the connectivity between the pyridinium and the aldehyde groups and therefore led to catalytic activity. 1423 Scheme 43 (a) Reaction of an Amino Acid with PLP Showing the Aldimine 136 Initially Produced, and the Molecular Structure of the Quinonoid 137 Formed after Removal of the Amino Acid’s α-Hydrogen, Which Leads to Racemization; and (b) Photoresponsive PLP Mimic 138(1423) Feringa et al. used one of their rotary molecular motors based on a chiral overcrowded alkene, which can perform a directional 360° rotary cycle fueled by light and heat to create a photoswitchable catalyst. During this rotation, the helicity of the motor changes, which changes the overall chirality of the system. A DMAP (dimethylaminopyridine) Brønsted base (Figure 54 ) and a thiourea hydrogen-bonding donor group, which together comprise a bifunctional organocatalyst, perform Michael additions when in close proximity. As a consequence, the catalytic activity as well the stereoselectivity can be controlled. Michael addition of 2-methoxy thiophenol to cyclohexanone was used as a test system. The (P,P)-trans-140 isomer resulted in a racemic (R,S)-thiol adduct in low yield (7%) after a lengthy reaction. When the (M,M)-cis-140 isomer was used, the Michael addition proceeded significantly faster furnishing a 50% yield of product, with an er of 75/25 (S/R). Finally, the (P,P)-cis-140 isomer led an 83% yield with an inversion in enantioselectivity providing an er of 23/77 (S/R). 1424 Recently, the same group reported a bisphosphine derivative of the same photoswitchable rotary motor. Both product enantiomers of a palladium-catalyzed desymmetrization reaction could be formed, although in situ switching experiments were complicated by the photosensitivity of the active palladium complex. 1425 Figure 54 Schematic representation and molecular structure of a bifunctional organocatalyst integrated in a directional light-driven molecular motor. A, DMAP; B, thiourea hydrogen-bonding donor group. (a) A and B are remote. (b) A and B are in close proximity with M helicity in the (M,M)-cis-140 isomer, preferentially providing the (S) enantiomer of the reaction product. (c) A and B are in close proximity with P helicity in the (P,P)-cis-140 isomer, generating (R) enriched product. Step 1: irradiation at 312 nm at 20 °C. Step 2: heating at 70 °C. Step 3: irradiation at 312 nm at −60 °C. Step 4: temperature −10 °C. 1424 8.4.2 pH-Dependent Switchable Catalysts Several rotaxanes containing a catalytic unit have been reported. However, these cannot typically be switched on/off. 1318,1426−1430 Leigh et al. described the rotaxane-based switchable organocatalyst 141, in which catalytic activity could be controlled by pH-dependent macrocycle shuttling (Scheme 44 ). The design consisted of a dibenzo[24]crown-8 macrocycle and an axle containing both triazolium rings and a dibenzylamine/ammonium moiety. The secondary amine/ammonium group was able to carry out iminium catalysis. However, when the rotaxane was protonated, the ammonium group was a better binding site for the macrocycle than the triazolium ring, and so the macrocycle blocked the catalytic center, preventing the reaction. When the secondary amine of the rotaxane was not protonated, the triazolium ring provided a better binding site for the macrocycle and the dibenzylamine group was exposed, and could therefore participate in catalysis. As a model reaction, the Michael addition of an aliphatic thiol to trans-cinnamaldehyde was performed. The free thread was found to efficiently catalyze the reaction in both its protonated and its deprotonated states. 1431 Recently, Leigh and co-workers explored the activation modes of rotaxane catalyst 141 and published a chiral version of this design, with a chiral center next to the secondary amine. 1432,1433 This chiral organocatalyst was able to perform an asymmetric Michael addition with good stereoselectivity. 1433 The pH-driven shuttling of a pyridyl-2,6-dicarboxyamide macrocycle between squaramide (hydrogen bond catalyst) and ammonium (iminium catalyst) stations has been reported. 1434 The macrocycle blocks the site it is bound to, allowing the selective exposure of catalytic sites and the generation of different products. Scheme 44 Acid–Base Switching of the Position of the Macrocycle, Which Conceals or Exposes the Catalytic Site on the Rotaxane 1431 8.4.3 Allosteric Regulation of Switchable Catalysts through Ion Addition A different approach to the regulation of catalytic activity is via the addition of small molecules such as ions, which can change the supramolecular structure of the catalyst, as is observed in many enzymatic processes. Mirkin and co-workers reported the synthesis of triple-layer complex 142 composed of two transition metal hinges, two chemically inert blocking exterior layers, and a single catalytically active interior Al (III)-salen complex, which can act as a ring-opening polymerization catalyst for ε-caprolactone (Figure 55 ). Polymer growth and molecular weight could be controlled by the addition of Cl– (catalytic layer exposed) or of the Cl– abstracting agent NaB(ArF)4, which results in the fully closed complex 142. 1435−1437 Recently, the same research group reported an allosterically regulated photoredox catalyst based on a similar switching mechanism. 1438 Figure 55 (a) Allosteric supramolecular triple-layer complex 142, which regulates the catalytic living polymerization of ε-caprolactone. (b) Molecular structures of the components. 1435 Another ion triggered switch was the self-locking system published by Schmittel and co-workers (Scheme 45 ). 1439−1441 Their design consisted of a zinc porphyrin and a 4-aza-2,2′-bipyridine tether, which either coordinated to the zinc porphyrin or formed a complex with copper and a shielded phenanthroline. In the presence of copper, the tether was removed from the zinc porphyrin center, and piperidine was bound at the coordination site. When copper was removed, piperidine was liberated and able to catalyze Knoevenagel reactions. 1439 Scheme 45 Schematic Representation of the Reversible Locking and Unlocking of Switchable Catalyst 143(1439) 8.4.4 Redox-Driven Switchable Catalysts Canary and co-workers have reported a redox-switchable chiral catalyst capable of delivering either enantiomer of a nitro-Michael addition product depending on the oxidation state of a single copper atom. The design, inspired by previous work carried out in their group and by others, 1442−1444 is based on complexes derived from methionine, which were shown to undergo inner sphere ligand rearrangement upon one-electron oxidation or reduction of copper. The rearrangement is coupled to the orientation of quinolone rings to afford enantiomeric configurations. Urea catalysis produced the (S) enantiomer in 72% ee and 55% yield, and the in situ reduced complex produced the (R) enantiomer in 71% ee and 43% yield. 1445 8.5 Synthesis Using Artificial Molecular Machines Nature provides us with many examples of biological molecular machines that engage in sequence-specific chemical synthesis. For example, the ribosome utilizes sequence information stored in mRNA to synthesize proteins with excellent fidelity. 68−70 The various DNA polymerases provide further examples of sequence-specific polymer synthesis by biological molecular machines. 71 The inherently programmable nature of DNA has been used by synthetic chemists to accomplish sequential oligomer synthesis. A common strategy is exemplified by the seminal work of Liu et al., 1446 where complementary sequences of DNA capped with different, mutually reactive end groups are annealed to create an extremely high local concentration of the two reactive species ensuring intracomplex bond formation predominates. After the first reaction is complete, a biotin tag on one strand allows its selective removal, after which a new reactive partner can be annealed to the existing strand and a second reaction conducted. Peptide, 1446−1448 carbon–carbon, 1449−1452 carbon–heteroatom bond forming reactions, 1453 and even cycloaddition reactions have all been reported on the basis of this type of methodology. 1454 More elaborate DNA systems have been used by Seeman and co-workers to create a “molecular assembly line” where three gold nanoparticles could be combined to form eight differently composed products, illustrating the complexity that can be achieved with DNA-based molecular machines. 1455 In a seminal example of a synthetic system mimicking some of the properties of an enzyme, Nolte, Rowan et al. reported a rotaxane-based catalytic system. 1456 A glycoluril clip-based macrocycle containing a manganese porphyrin catalyst catalyzed the epoxidation of a butadiene polymer, which formed the thread of the rotaxane. A greatly enhanced proportion of cis-epoxide was observed in the product, consistent with previous work showing that when the reaction occurred in the macrocyclic cavity steric constraints favored the cis product. 1457 This system has been further optimized by substitution of the porphyrin macrocycle with protective ligand groups, which increase turnover number (TON) and provide even greater cis selectivity by preventing reaction on the exterior face of the macrocycle (Figure 56 ). 1458 The rate of threading of their macrocycle onto polymeric materials was also studied and found to be highly dependent on interactions between the thread and the outside of the macrocycle preassociating the two moieties. 1459 Figure 56 Porphyrin-catalyzed epoxidation of a butadiene polymer by 144, utilizing a rotaxane architecture. Reprinted with permission from ref (1456). Copyright 2003 Nature Publishing Group. A cyclodextrin dimer has been used as both a catalyst (for the polymerization of δ-valerolactone) and a molecular “clamp”, which guides the output of the polymer. 1460 In this system, an α-cyclodextrin acts as the active site while a β-cyclodextrin acts as the clamp guiding the growing chain away from the active site (Scheme 46 ). An additional example was provided by Takata et al., 1430 where the cyclization of propargyl or allyl urethane groups in the rotaxane backbone to oxazolidinones was catalyzed by a palladium center, bound to the macrocycle. Scheme 46 Cyclodextrin Dimer 145 Polymerization Catalyst 1460 , Reprinted with permission from ref (1460). Copyright 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Recently, Leigh et al. reported the first small molecule artificial molecular machine (146) capable of sequence-specific peptide synthesis. 666 The machine is based on a rotaxane architecture, which is used to ensure processivity in a manner reminiscent of both the ribosome-mRNA structure and the DNA clamp of DNA polymerases. 68−71 Sequence information contained in the track is directly converted into the sequence of the peptide synthesized, and thus the track plays some aspects of the role of mRNA in the ribosomal system. Reactive phenolic ester groups take the place of tRNA-bound amino acids and are sequentially “picked up” by the catalytic arm of the macrocycle. The catalytic arm bears a cysteine group and operates by native chemical ligation (NCL) chemistry where the thiolate of the cysteine unit reacts first with the ester group before transferring the activated amino acid to the end of the growing chain. 1461 The steric bulk of the loaded amino acid prevents the macrocycle passing over the barrier unit before this reaction has taken place but allows free passage of the macrocycle after being cleaved. This prevents the peptide sequence from being scrambled and allows sequential reaction. Rigid spacing units between each loaded amino acid minimize the likelihood of the catalytic arm encountering an out of sequence amino acid. The catalytic unit consists of a cysteine-glycine-glycine motif with the second two amino acids present to prevent an unfavorable 1,8-S,N-acyl shift transition state during the second ligation (Figure 57 ). Figure 57 Leigh’s small molecule peptide synthesizer, 146. 666 After the third and final amino acid has been cleaved from the thread, the macrocycle can dethread and be isolated. The fidelity of sequence transfer in this process was demonstrated by tandem mass spectrometry of the macrocyclic product, identical to an authentic sample synthesized by conventional methods. No out of sequence or abbreviated products were observed by HPLC–MS underlining the extremely high level of control afforded by this system. This molecular machine showed processivity, sequence specificity, and autonomy and an exceptional level of control at the molecular level. However, several problems remain with this design. The rate of reaction is vastly slower than that of the ribosome; one peptide bond takes an average of 12 h to form, whereas the ribosome makes approximately 20 amide bonds every second. 68 Additionally, information is read in a destructive manner; once the macrocycle has cleaved the ester bonds, there is no way to reload the machine for a second run. Finally, the NCL reaction used to catalyze peptide formation limits the scope of amino acids that can be incorporated into the growing peptide chain. Recently, Leigh et al. reported an extension to this system, creating a four barrier machine whose final product is a macrocycle-bound heptapeptide. 661 The sequence specificity of the product was again confirmed by tandem mass spectrometry and compared to an authentic sample. Although no problems were encountered in this system, whose final bond formation involves a 20-membered ring S,N-acyl transfer, the ever-expanding size of the required cyclic transition state will at some point limit the length of potential peptide products formed by the current design (Scheme 47 ). Scheme 47 Operation of Leigh’s Peptide Synthesizer 666 , Reprinted with permission from ref (666). Copyright 2013 American Association for the Advancement of Science (AAAS). 8.6 Controlled Motion on Surfaces, in Solids, and Other Condensed Phases 8.6.1 Controlled Motion in Solids and Condensed Phases A major challenge is to use molecular machines for practical applications by utilizing molecular scale changes to create macroscopic effects. Using molecular machines in the solid state or other condensed-phase matter could lead to new materials with a higher level of complexity with controlled cooperative molecular motion leading to changes in the properties and function of the material on a macroscopic scale. In recent years, several examples of molecular motion in condensed-phase matter have been described, some of which show promise for applications in the fields of electronics and optoelectronics. Garcia-Garibay and co-workers have investigated the rotational dynamics and photophysical properties of the crystalline, linearly conjugated, phenyleneethynylene molecular dirotor 147 (Figure 58 ). A pentiptycene unit was incorporated as a central stator about which the two flanking ethynylphenylene rotators could adopt various torsion angles. X-ray diffraction studies showed that in the solid-state structure of molecular dirotor 147 all of the phenyleneethynylene chromophors are arranged parallel to one another and therefore shared the same rotation axis. The chromophore displayed significant fluorescence changes as a function of interphenylene torsion angles. The authors suggest that with this system external control of the rotation could be achieved by the application of an electric field, which would allow a rapid shift of solid-state fluorescence emission and optical properties upon the application of appropriate stimuli. 1462−1470 Figure 58 (a) Molecular structure of linear conjugated phenylethynylene molecular dirotor 147. Rotation of the phenylene rotor (shown with an arrow) creates rotamers with varying degrees of π-conjugation and so wavelengths of emission. (b) X-ray structure of the dirotor 147. X-ray crystal structure reprinted with permission from ref (1462). Copyright 2013 American Chemical Society. Sozzani and co-workers have published several examples of molecular rotors in crystals with open porosity, as well as molecular rotors embedded in porous frameworks such as aromatic or organosilica frameworks. In some of these systems, the rotational motion can be actively regulated in response to guest molecules such as CO2, I2, tetraethylammonium chloride, and water. These responsive materials may find applications spanning from sensors to actuators, which could capture and release chemicals on command. 1471−1474 Recently, Schurko and Loeb published a metal–organic framework (MOF) material containing dynamically interlocked components. 1475 They used [2]rotaxane 148 as the organic linker and binuclear Cu(II) units as the nodes (Figure 59 ). Void spaces inside the rigid framework allowed the macrocyclic ring of the rotaxane to rotate rapidly. Initially the macrocycle is locked in place through hydrogen bonding from an ether oxygen atom of the macrocycle to a copper-bound H2O. Dynamic motion occurs upon removal of the water molecules by heating to 150 °C under vacuum, destroying hydrogen-bond interactions and allowing rapid circumrotation about the rotaxane axle. Rotation can be quenched by the addition of water. This type of material could be useful for the creation of solid-state molecular switches and machines. 1475 A rotaxane-based molecular shuttle incorporated into the structure of a MOF has recently been reported. 1476 Figure 59 (a) Structure of [2]rotaxane 148. (b) Schematic representation of the structural design components used to create the metal–organic framework. (c) X-ray structure of the tetra-ester precursor to [2]rotaxane 148. (d) X-ray structure of a single unit of the mechanically interlocked molecule, coordinated to four Cu(II) paddlewheel clusters. X-ray crystal structure reprinted with permission from ref (1475). Copyright 2012 Nature Publishing Group. 8.6.2 Solid-State Molecular Electronic Devices In a series of ground-breaking but controversial experiments interfacing switchable rotaxanes and catenanes with silicon-based electronics, molecular shuttles have been employed in solid-state molecular electronic devices. 1146,1477−1485 Bistable [2]rotaxanes and [2]catenanes have been the subject of numerous experimental investigations in the course of the development of such molecular electronic devices. 1233,1478,1486−1495 Here, the bistable [2]catenanes and [2]rotaxanes feature a cyclobis(paraquat-p-phenylene) (CBPQT4+) macrocycle and two stations, often a tetrathiafulvalene (TTF) site and a dioxynaphtalene site (DNP). Initially the macrocycle preferentially resides over the TTF site due to strong aromatic charge-transfer interactions between the components; this is referred to as the ground-state coconformation (GSCC). Electrochemical oxidation of the TTF station to form TTF2+ generates Coulombic repulsion between CBPQT4+ and TTF2+, and drives the translation of the macrocycle to the DNP station, to give the metastable state coconformation (MSCC). The process can be reversed on reduction of TTF2+ to TTF followed by either thermal relaxation of the macrocycle to the TTF station, or reduction of the bipyridinium units in the cyclophane ring to the corresponding radical cations, which reduces the activation barrier to shuttling, restoring the system to the GSCC. These two mechanically distinguishable states exhibit different characteristic tunneling currents. On the basis of quantum mechanical computational studies, the MSCC state is predicted to be the more highly conducting state. The switching cycle can be detected by a number of experimental techniques including time- and temperature-dependent electrochemistry and spectroscopy. Studies have shown that current levels on switching are influenced by temperature, the structure of the rotaxane/catenane, and the environment in which the molecular machines are embedded. 1496−1498 Different environments, including Langmuir–Blodgett (LB) films, 1499−1501 self-assembled monolayers (SAMs), 1502−1506 and solid-state molecular-switch tunnel junctions (MSTJs), have been extensively studied. 1150,1151,1488,1489,1507−1512 In one particular MSTJ, a monolayer of switchable rotaxane 149 was embedded between two conducting electrodes (Figure 60 ). This MSTJ acts as a gate, which can be opened or closed in response to an applied voltage by changes in conductivity and resistance and could be used in molecular logic gate designs. The reported design showed stable switching voltages of −2 and +2 V, with reasonable on/off ratios and low switch-closed currents. Nanometer-scale devices have been built using this approach and connected to form 2-D crossbar circuit architectures. 1144 As a next step, the authors published the design of a 160-kilobit molecular electronic memory circuit consisting of 400 silicon-nanowire electrodes (16 nm wide) and crossed by 400 Ti electrodes sandwiching a monolayer of bistable [2]rotaxanes. 1513 Despite the interesting findings, many remain skeptical about the utility of rotaxanes in electronics, and an array of scientific and engineering challenges remain to be addressed such as device robustness, improved etching tools, and improved switching speed. 1514−1516 Figure 60 (a) Rotaxane 149-based molecular switch tunnel junctions and proposed mechanism for the operation. (i) In the ground state, the tetracationic cyclophane (dark blue) mainly encircles the TTF station (green) and the junction exhibits low conductance. (ii) Application of a positive bias results in one- or two-electron oxidation of the TTF units (green → pink), and increases electrostatic repulsion causing (iii) shuttling of the macrocycle to the DNP station (red). (iv) Returning the bias to near −0 V provides a high conductance state, in which the TTF units have been regenerated, but translocation of the cyclophane has not yet occurred due to a significant activation barrier to movement. Thermally activated decay of this metastable state may occur slowly ((iv) → (i), in a temperature-dependent manner) or can be triggered by the application of a negative voltage (v), which temporarily reduces the cyclophane to its diradical dication form (dark blue → orange), allowing facile recovery of the thermodynamically favored coconformation (vi). (b) Example of one design of a molecular switch. The coloring of the functional units corresponds to that used for the structural diagrams. 1479,1482,1483 Reprinted with permission from ref (14). Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. 8.6.3 Using Mechanical Switches To Affect the Optical Properties of Materials Molecular machines can influence the optical properties of materials in the solid state as well as in solution (section 8.2.2 ). Numerous molecular-machine-functionalized materials where changes can be visually monitored have been reported. Some pertinent examples are discussed below. Liquid crystals have become widespread in numerous technological applications. 1517−1520 The photoalignment of liquid crystals has considerable potential for display and other optoelectronic applications. The reorienting effect of light on nematic liquid crystals is well-known. 1521−1525 However, the effect can be greatly enhanced by the presence of a dopant dichroic dye. Studies have suggested that the mechanism of this effect involves the photoexcited dye molecules acting as Brownian rotors in the nematic liquid crystal. 1526,1527 The change in orientation was attributed to a ratchet mechanism operating via the generation of torque. Feringa and co-workers reported a chiral nematic liquid crystal film doped with a previously reported chiral light-driven molecular motor. Irradiation of the film resulted in directional motion of the molecular motor, which induced a rotation of the rod-like liquid crystal molecules. This rotation led to the alteration of the color of the film over a large part of the visible spectrum (Figure 61 ). 1528,1529 Figure 61 Color changes in a liquid-crystal film doped with a light-driven rotor. Reprinted with permission from ref (1528). Copyright 2002 American National Academy of Sciences. Yamaguchi et al. reported crystalline molecular gyrotop, which showed temperature-dependent changes in optical properties as a result of structural expansion upon rotor acceleration (Figure 62 ). 1530−1532 Utilizing a phenylene moiety as the rotor, they observed that the orientation of the rotor in the crystal was ordered below 270 K, but became disordered above this temperature generating a slight deformation of the crystal lattice. Using the temperature-dependent features of the crystal, the birefringence (Δn) of the crystal could be controlled. At 280 K, the phenylene moiety undergoes a 180° flipping between two equilibrium states, which provides an almost constant Δn. Above this temperature, Δn decreases as the phenylene moiety rapidly rotates, and the cage expands. The dynamic and optical properties are reversible. 1530 Figure 62 (a) Molecular structure of gyrotops 150 and 151. The bulkier 151 does not exhibit rotation. (b) X-ray structure of 150. (c) Single crystal of 150 irradiated with polarized white light. (d) X-ray structure of 151. (e) Single crystal irradiated with polarized white light. For 150, a continuous change in color was observed, due to thermal expansion. Reprinted with permission from ref (1530). Copyright 2012 American National Academy of Sciences. A pseudorotaxane that acts as a thermally driven molecular switch in a crystalline state was recently published by Horie and co-workers. 1533−1535 Crystals of the pseudorotaxane underwent phase transitions upon heating with accompanying changes in optical properties. Leigh et al. used [2]rotaxanes to control fluorescence by distance-dependent intercomponent electron transfer both in solution and on a polymer film. The thread of these rotaxanes included an anthracene fluorophore as a stopper attached to a glycylglycine hydrogen-bonding station and a C11 alkyl chain that could act as a second station. The macrocycles in 152 and 153·2H 2+ contained nitrophenyl and pyridinium moieties, respectively, which are known to quench the fluorescence by distance-dependent electron transfer. Macrocyclic shuttling could be induced by changes in solvent with strongly hydrogen-bonding solvents displacing the macrocycle from the glycylglycine station to the alkyl thread. In non-hydrogen-bonding solvents (e.g., benzene, CH2Cl2, CH3CN), the macrocycle was located on the glycylglycine station and fluorescence was completely quenched. In strongly hydrogen-bonding solvents (e.g., DMSO and NH2CHO), the macrocycle encapsulated the alkyl chain and fluorescence was restored. Polymer/rotaxane hybrids were used to prepare transparent films on quartz slides. Initially, no fluorescence was detected when the slides coated with a 154 containing film were illuminated with UV light (254–350 nm). However, after the slides were exposed to DMSO vapor, prior to illumination, a characteristic blue fluorescence was observed showing that similar shuttling processes were occurring on the polymer film as were observed in solution. Masking regions of the film from the DMSO vapor allowed the transitory etching of patterns on the polymeric films (Figure 63 ). 1536 Figure 63 (a) Chemical structures of rotaxane initiators 152 and 153 and the corresponding PMMA-based polymers 154, 155, and 155·2H 2 + . (b) Images obtained by casting films of polymer 154 on quartz slides, then covering the films with aluminum masks and exposing the unmasked area to DMSO vapor for 5 min. The photographs were taken while illuminating the slides with an 8-W UV lamp (254–350 nm). Reprinted with permission from ref (1536). Copyright 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Polymeric 155·2H 2+ films responded to two different stimuli. Protonation of the macrocycle by CF3CO2H vapor caused quenching of fluorescence, while DMSO vapor induced shuttling of the macrocycle and subsequently restored fluorescence emission. The response of 155·2H 2+ to the different combinations of two stimuli (DMSO and protons) corresponds to an INHIBIT Boolean logic gate (Figure 64 ). 1536 Figure 64 (a) Aluminum grid used in the experiment. (b) Pattern generated when films of 155 were exposed to trifluoroacetic acid vapor for 5 min through the aluminum-grid mask. (c) Mesh pattern obtained by rotation of the aluminum grid by 90° and exposure of the film shown in (b) to DMSO vapor for a further 5 min; only regions exposed to trifluoroacetic acid but not to DMSO were quenched as shown in the magnified view. Inset: Truth table for an INHIBIT logic gate. The photographs were taken while illuminating the slides with an 8-W UV lamp (254–350 nm). Reprinted with permission from ref (1536). Copyright 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Hydrosol–gel systems represent another example of a nonsolution media. The effect of doping these systems with rotaxanes has been studied by various groups. 1208,1537−1541 Light-driven rotaxanes based on α-cyclodextrin and cucurbit[7]uril have been dispersed in thermoreversible hydrogel systems by Tian and co-workers. They observed that when the rotaxanes were embedded in the hydrogel, their optical performance (fluorescence and induced circular dichroism) improved. They attributed this observation to the restriction of movement in the hydrogel system. 1539,1542 8.6.4 Using Mechanical Switches To Affect the Mechanical Properties of Materials Piezoelectric materials provide the most common example of an external stimulus being converted into a mechanical change in the material. Materials incorporating molecular machines could provide a far greater level of control over macroscopic properties than is currently possible, and their development has been the focus of great interest. 1543,1544 Conformational changes in polymers leading to macroscopic mechanical changes have been extensively studied. Hydrogels, consisting of water-swollen cross-linked networks of neutral or ionic amphiphilic polymers, have been shown to undergo reversible phase transitions in response to changes in temperature, solvent, pH, electric field, or irradiation with an up to 100-fold expansion on phase change. 1545−1558 Katchalsky et al. first exploited these properties to develop devices that converted chemical potential energy into macroscopic mechanical work. 1559,1560 A large variety of devices based on the characteristics of these materials have been proposed 1561−1581 including guest binding and release, 1582 control of enzyme activity, 1583 and the macroscopic motion of a polymer gel by selective contraction of alternate sides of a gel. 1571 Finally, guest binding by built-in recognition sites can lead to large-scale changes in the volume of the host gel. 1584−1598 Shape memory materials have also been designed that utilize control at a molecular level. 1599,1600 While most of these systems rely on a temperature variation-induced shape change, a system utilizing a photoinitiated radical reaction to rearrange covalent cross-links and recover the “memorized” shape has been reported. 1601 A reversible cross-linking [2 + 2] cycloaddition has also been used to achieve this control. 1602 Conducting polymers have been used to create stimuli responsive mechanical changes. 1603−1605 Electrochemical oxidation can induce counterion expulsion or inclusion and thus volume change, although these changes tend to be slow in aqueous solution. 1606−1609 Solid state, 1610 gel, 1611 and ionic liquid electrolytes have been shown to accelerate this process. 1612 Conformationally flexible calix[4]arenes have been used as “hinges” to form conducting oligothiophenes where the cone (but not the kite) form of the calix[4]arene maximizes the desired π–π interactions and switching could be utilized to provide macroscopic mechanical motion. 1613,1614 In the previous examples, a macroscopic change was induced by the sum of multiple, relatively uncontrolled conformational changes in a polymeric network; that is, the observed change is not an inbuilt feature of the molecular components. The use of particular submolecular conformational/configurational switches to directly trigger macroscopic changes has numerous benefits such as overcoming problems caused by slow diffusion of guests or solvents and ensuring a uniform response in the material. The use of configurational switches to control long-range order in liquid crystalline phases has been discussed elsewhere (section 8.6.3 ), but a similar application in synthetic polymers often furnishes a useful result. Azobenzenes are often used to control reversible contraction in polymers, a process that has been observed at a molecular level using AFM, 1615 as well as properties such as viscosity and solubility. 1616,1617 In the aforementioned AFM experiment, the application of a “load” to the AFM tip and subsequent isomerization toward the cis isomer provided the first direct measurement of the conversion of light to mechanical energy. Chiral azobenzene side chains can allow control over the screw sense of artificial helical polymers with isomerization leading to interconversion of the P and M forms. 1618−1620 Several liquid crystal polymeric systems have been reported. 209,1621−1635 Macroscopic motion was induced in liquid crystal polymer springs devised by Fletcher, Katsonis et al. 1636 Here, a chiral dopant, a polymerizable liquid crystal, and a polymerizable azobenzene switch were used to generate a liquid crystal polymeric spring. Irradiation with UV light caused cis–trans isomerization of the azobenzene switch, which in turn led to coiling/uncoiling or helical inversion of the spring. More complex behavior such as side to side bending could be induced by careful manipulation of the initial helical state and directed illumination. 1636 Light-driven directional rotor 156 has been used to drive the macroscopic motion of a glass rod. 1637,1638 When doped at 1 wt % into a liquid crystal, the helicity of the rotor induced a helical arrangement in the liquid crystal, giving rise to a characteristic “fingerprint” texture in the liquid crystal’s surface. Irradiation altered the distribution of isomers of the rotor and thus caused the liquid crystal’s helical nature to rearrange. This process could be observed by the clockwise rotation it caused of a glass rod laid on the surface and of the “fingerprint” pattern. Eventually, however, rotation stopped as the system reached a photostationary state with the distribution of molecules between isomers reaching equilibrium. When the light source was removed, the isomer distribution decayed back to its initial state, with concomitant reverse rotation. The use of the opposite isomer of the rotor led to inversion of the direction of rotation (Figure 65 ). Figure 65 (a) Rotor 156. (b) Rotation of micrometer scale glass rod on doped liquid crystal film. Photos taken at 15 s intervals. Reprinted with permission from ref (1637). Copyright 2001 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Shuttling in [3]rotaxane 157 has been used to generate a macroscopic mechanical response, deformation of a microcantilever beam, which had been coated with a monolayer of ca. 6 billion rotaxanes. 1238,1341 Oxidation of the tetrathiafulvalene (TTF, green) station decreased the affinity of the macrocycle for this station, and induced shuttling to the naphthalene (red) station. A deformation of 550 nm was recorded, and the process could be reversed and repeated (with diminishing amplitude) over several cycles (Scheme 48 ). 1238,1341 Microcantilever deformation by guest recognition at the surface has been used by several groups as a “read-out” from molecular detectors. 1639−1646 A polymer has been reported that contracts upon illumination based on a directional light-driven rotor (Scheme 48 ). 1647 A photodriven rotary motor has also been used to induce the disassembly of self-assembled nanotubes. 1648 Scheme 48 Molecular Motion Generating a Macroscopic Mechanical Response 1238,1341,1647 , Reprinted with permission from ref (1647). Copyright 2015 Nature Publishing Group. 8.6.5 Using Mechanical Switches To Affect Interfacial Properties The ability to easily modify the properties of surfaces in a reversible manner would be extremely valuable in the design and synthesis of technologically useful devices. Stimuli-responsive polymer modification of surfaces can lead to control over wettability, adhesive ability, porosity, patterning, and interfacial interactions; this topic has been reviewed elsewhere. 1649−1663 Interlocked architectures such as bistable rotaxane 158 have been used to restrict or allow access to the pores in materials such as mesoporous silica particles. 1664 The positively charged macrocyclic ring initially prefers the TTF station, and in this conformation, diffusion into and out of the nanopores of the silica is allowed. Oxidation of the TTF moiety causes shuttling of the ring to the dioxynaphthalene station, and closure of the nanopores, trapping a portion of the solvent/solute mixture. Solvent exchange and reduction of the oxidized TTF station leads to guest release (Figure 66 ). The aggregation and deaggregation of a polymeric material has been caused by the formation of pseudorotaxanes between strands. This has been used to transiently trap nanoparticles in the pores formed upon aggregation. 1665 Figure 66 (a) Structure of nanopore gate, and (b) controlled release of guest from nanopores. 1664 Reprinted with permission from ref (14). Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. The translation of the macrocyclic ring of 159 has been converted into an electric signal. 1666−1668 A gold electrode was used in place of one stopper of the rotaxane, and a cyclodextrin ring with an attached ferrocene was used as the macrocycle. On photoactivated shuttling, the change in rate of distance-dependent oxidation of the ferrocene moiety allowed detection of the shuttling motion. 1669 When a glucose oxidase enzyme was attached via a rotaxane architecture to a gold electrode, the CBPQT4+ macrocycle of the rotaxane acted as a two-electron shuttle and allowed enzyme operation, which was prevented when the enzyme was attached solely by the thread. 1670 A similar effect was seen in a system involving a CdS nanoparticle in place of the enzyme in which the observed photocurrent was amplified 8-fold in the rotaxane over the thread. 1671 In a simpler analogue of these systems involving a CBPQT4+ macrocycle and a diiminobenzene station (159), electrode-induced shuttling could be directly observed. 1672 The reduction in macrocycle charge on shuttling was shown to decrease the hydrophilicity of the surface, and thus increased the contact angle of a drop of water on the surface (Scheme 49 ). Scheme 49 Electrode Controlled Macrocycle Shuttling, Leading to Control of the Hydrophilicity of the Surface 1672 This control of wettability has been exploited by many groups to form a wide range of interesting systems, especially in the development of “self-cleaning” surfaces. 1673−1679 Surface wettability has been controlled by surface mounted photochromic switches. 1680 Self-assembled monolayers of carboxylate-terminated alkanethiols on gold have been switched between hydrophilic (carboxylate exposed, gold electrode at a negative potential) and hydrophobic (carboxylate surface bound, gold electrode at a positive potential) states by varying the electric potential. 1681 Bipyridinium groups have been used in place of carboxylates to similar effect. 1682 Several methods for the photoswitchable change of surface wettability have been developed. Systems based on the reversible exposure of hydrophobic groups upon shuttling of a cyclodextrin ring, 1683 or the release of a hydrophilic guest, 1684 have been successfully exploited. Feringa and co-workers exploited a tripodal stator to fix a molecular directional rotor to a surface, which upon irradiation exposed or hid a hydrophobic fluorinated chain providing control over surface wettability (Figure 67 ). 1685 Figure 67 Feringa’s tripodal wettability switch 160. 1685 Perhaps the most powerful demonstration of the usefulness of control over wettability is in the macroscopic transport of a liquid across a surface. The earliest example came from Whitesides and Chaudhury who created a hydrophilicity gradient on a silica wafer by reaction with a trichloroalkylsilane vapor and observed the motion of a droplet of water up a 15° incline. 1677 Numerous examples of droplet motion driven by surface energy gradients/steps have since been reported. 1686−1690 However, remotely controlling wettability to generate droplet motion has remained challenging. 1691 This remote control would be particularly useful in “lab-on-a-chip” type applications where it could provide a gentle and cheap alternative to expensive microscopic pumps and strong electric fields. 1692 Azobenzene containing calix[4]resorcinarene 161, when deposited as a monolayer, provided the first example of remote control of surface energy and therefore droplet motion by irradiation. 1693 The monolayer initially consisted of all cis isomers. However, a droplet of olive oil has more favorable interactions with the extended trans form, where interactions with the alkyl chain of the calix[4]arene can be maximized. As such, when an asymmetric light source irradiated the droplet, forming more trans-calix[4]arene on one side of the droplet than the other, a surface energy gradient was created and the droplet moved away from the trans enriched area. This movement could be continued if the light source followed the rear edge of the droplet (Figure 68 ). Figure 68 Calix[4]arene 161, used to control surface wettability. 1693 Photoresponsive control of droplet motion has also been achieved using a system based on bistable rotaxane 162, adsorbed onto a gold surface using a thiol linker. 1245 Diiodomethane drops could be transported on a millimeter scale across a surface using the macrocycle of 162 to either hide or expose the fluoroalkane region of the rotaxane and thus modify surface energy. Using this strategy, a microliter droplet could be moved up a 12° incline. Roughly 50% of the absorbed photon energy was converted into gravitational potential energy of the drop (Figure 69 ). Figure 69 Light switchable rotaxane 162, and transport of a microliter drop of CH2I2 across a flat surface (a–d) and up a 12° incline (e–h). Reprinted with permission from ref (1245). Copyright 2005 Nature Publishing Group. The above examples of macroscopic control utilizing molecular motion underscore the potential importance of this concept for technological applications. Similar control of motion has been achieved with surfaces microscopically patterned with thermoresponsive polymers 1694−1696 and by amplifying the effect of the photochemical switching of a spiropyran-functionalized surface. 1697 9 Artificial Biomolecular Machines 9.1 Hybrid Biomolecular Systems Biological systems make extensive use of motor proteins such as ATPase, kinesin, myosin, and dynein. These biological motors provide both examples of machines viable at the nanoscale and useful building blocks for the creation of hybrid systems. The availability of preformed nanoscale machines has greatly expedited the creation of complex nanotechnological systems. Although a thorough exploration of this topic is beyond the purview of this Review and has been covered elsewhere, 15,47,1364,1698−1702 a brief overview is useful. The initial purpose of these biohybrid experiments was to further probe the complex mechanisms by which biological machines operate. An early example was provided by Kinosita et al., who reported a series of experiments exploring the F1-ATPase rotary motor. 1703−1707 They were able to observe the 360° rotation of the motor, powered by the hydrolysis of ATP, directly, via microscopy, for individual motors mounted on glass. Mechanical rotation has also been observed in an F0F1-ATPase, where the rotor domain is coupled to a proton transporting domain. 1708 Even more remarkably, ATP synthesis has been driven by rotating a magnetic bead attached to an F1-ATPase with electromagnets. 1709 This is a stunning example of chemical synthesis driven by an external mechanical force. ATPase systems have been less exploited in recent years due to the difficulties in using them in synthetic systems. 1710 However, a recent example showed the insertion of ATP synthases into an artificial lipid microcapsule and their subsequent use to generate ATP inside the capsule. 1711 Kinesin, myosin, and dynein systems have been much more extensively utilized. 1712−1722 They have been shown to transport cargos in artificial systems, although cargos must typically be large to outcompete diffusive forces. 1375,1723−1727 These biological walkers have been controlled by the design of appropriate tracks, 1728−1730 or by the application of external forces. Kinesin has been used to stretch a length of DNA, 1715 and to enable the detection of nanomolar concentrations of a targeted protein. 1731 Cross-linked actin and myosin gels were found to move over each other in the presence of ATP. 1732 Microtubule containing gels combined with kinesin and ATP have been shown to spontaneously generate autonomous motility. 1733,1734 As a further benefit to using biological motors, ATPase motors have been shown to operate at a near 100% efficiency, 1707,1735−1737 and kinesin at >50%, 1738 whereas in a typically used polymeric system where phase transitions are used to generate molecular level forces, an efficiency of 0.0001% was observed. The field of biomolecular electronics often uses molecular motion as a switching mechanism, particularly in systems utilizing photoactive proteins (be they native or engineered). 1739−1750 9.2 Hybrid Membrane-Bound Machines Ionophores and ion channels, both synthetic 1751−1755 and biological, 1756−1766 have been extensively explored and exploited. Biological ion channels have been inserted into artificial membranes and used to sequence single strands of DNA, 1756 to open the pore of a nonselective efflux channel, 406,1767,1768 and to effect the light-driven production of ATP by F0F1-ATPase. 1769 Feringa et al. reported the switching of the “mechanosensitive channel of large conductance’ (MscL) of E. coli, which can be opened in response to the introduction of charged entities at a certain location in the protein channel. 406 Initially, a light-cleavable group was used to release acetate anions and thus open the channel, with a reversible version, utilizing a spiropyran switch, also reported (Scheme 50 ). Ion channel proteins modified postsynthetically to be light switchable have been inserted into living cells, and successfully opened and closed. 1770−1777 Unnatural, light switchable amino acids have also been incorporated into ion channel proteins. 400,1778−1780 Scheme 50 (a) Irreversible Photocleavage of 163 Leading to Pore Opening; and (b) Reversible Photoswitching of 164, Leading to Pore Opening MscL = mechanosensitive channel. 406 Active transport between aqueous phases across an organic phase, 1781−1783 and lipid bilayers, 1784 has been observed, driven by differing redox potentials in each aqueous phase. An electron donor–acceptor molecule has been directionally inserted into a lipid bilayer, which on irradiation created a redox potential gradient and ferried protons across the bilayer establishing a pH gradient. However, the quantum yield for the process was only 0.004. 1785 Calcium ions have been transported across a membrane in a similar system. 1786 Work to establish longer lived photoinduced charge separated states, which might lead to greater quantum yields and further applications, has continued. 627,1787−1807 Alternating currents have been induced in a protein-based photoelectrochemical cell by irradiation with intermittent light. 1808 An interesting recent example utilized the light-induced ring-opening of spiropyrans under ultraviolet light and their ring closure under visible light illumination. When a membrane doped with spiropyran 165 was illuminated with ultraviolet light on one side and visible light on the other, the differing proton affinities of ring-opened and ring-closed spiropyran led to proton shuttling across the membrane and the generation of an electric potential of ca. 210 mV and pH gradient of ca. 3.6 pH units (Figure 70 ). 1809 Figure 70 Proton gradient established by spiropyran (165) shuttling upon differential illumination of the two sides of a membrane. Reprinted with permission from ref (1809). Copyright 2014 Nature Publishing Group. 9.3 DNA-Based Motors and Switches Biological building blocks have been used to design and create molecular machines by many research groups. 1810−1828 A large number of DNA-based molecular motors, 1829−1832 walkers, 1382,1833−1838 tweezers, 1839 gears, 1840 springs, 1841 robots, 1842,1843 transporters, 1844 and interlocked structures such as DNA rotaxanes and catenanes can be found in the literature. 1845−1850 All are built by self-assembly, exploiting the sequence-specific interactions that bind complementary oligonucleotides together to form double helix or triplex structures. 1851−1870 Beside base-pairing, other structural motifs can be formed and used in the design of molecular machines such as the pH-induced self-assembly of C-rich sequences into i-motif configurations, 1871,1872 the ion-induced self-organization of G-rich sequences into G-quadruplexes, 1873−1876 and the metal–ion bridging of duplex DNA by T–Hg2+–T or C–Ag+–C complexes. 1877,1878 Various fuels such as single-stranded oligonucleotide fragments, pH variation, metal ions, and light have been used to trigger these DNA devices. 1879−1884 These systems have been used in molecular sensing, drug delivery and other medical applications, the construction of logic gates, the control of chemical transformations, and for many other purposes. 1874,1885−1911 DNA tweezers represent a simple class of DNA machines. 1884,1912 They are two-armed constructs bridged by a DNA linker that can undergo transitions between open and closed states in response to external triggers such as the addition of single-stranded DNA or metal ions, or a change in pH. Willner et al. reported a biomolecular logic gate based on three different DNA tweezers A, B, and C. These were activated by different inputs: protons, Hg2+ ions, and nucleic acid strands (Figure 71 ). The output delivered by this machine depends both on the inputs provided and its initial internal state. Depending on the input, there are eight possible configurations of the three tweezers (open or closed for each). The output could be studied by measuring the Förster resonance energy transfer (FRET) between different pairs of fluorophore and quenching molecules attached to the arms of each of the three tweezers. The linker unit is common to all three tweezers, meaning that tweezers A and B can also be opened by the complementary antilinker. Thus, for any pair of tweezers, there are two different inputs that cause a change in the state of the device. In total, the device can adopt 16 different states and can furthermore be used as a memory storage system because each state and output is dependent not only on the most recent input but also on past states and inputs. 1913,1914 Figure 71 (a) Tweezer A: in the closed form the arms are bound to the linker unit (blue) by Hg2+ ions through T–Hg2+–T bonds. To open the molecular tweezer, Hg2+ is sequestered by the addition of cysteine. (b) Tweezer B: in acid the arms form an i-motif, thus releasing the linker unit, whereas at pH = 7.2, the i-motif is destroyed resulting in the stabilization of the closed structure. (c) Tweezer C: the linker unit can be released by a complementary strand, the antilinker that opens the tweezers. Reprinted with permission from ref (1914). Copyright 2010 American National Academy of Sciences. DNA machines have been used to regulate enzyme cascade reactions. 1890,1891 Liu and co-workers reported a machine containing DNA double crossover (DX) motifs, which formed two rigid arms, joined by an immobile four-way junction (Figure 72 ). 1915 A DNA motor that could switch between stem-loop and double-helix structures, driven by a strand displacement reaction, was incorporated at the center of the machine to cycle between open and closed states. This design amplifies the small motion generated by the DNA motor into a much greater change in separation between the ends of the two arms where glucose oxidase (GOx) and horseradish peroxidase (HRP) were attached. In this biochemical cascade system, GOx first catalyzed the oxidation of glucose to generate gluconic acid and hydrogen peroxide. Hydrogen peroxide is catalytically reduced by HRP into H2O. At the same time, HRP turns ABTS2– into ABTS–, which allowed the kinetics of the peroxidase to be monitored. HRP has a much higher turnover rate then GOx, so the distance hydrogen peroxide must diffuse has a crucial influence on the rate of reaction. Therefore, when the two enzymes are attached to the two arms, the diffusion distance of hydrogen peroxide can be changed from 6 to 18 nm by operation of the DNA motor, regulating the rate of enzymatic reaction. Sequential addition of fuel and antifuel strands showed that this regulation was reversible. Figure 72 DNA machine reported by Liu and co-workers, which could be used to regulate an enzyme cascade reaction. Reprinted with permission from ref (1915). Copyright 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Willner and co-workers have used catenated DNA machines as carriers of Au nanoparticles. 1916 Ordered assemblies of nanoparticles with defined geometries have recently attracted a great deal of interest as these engineered nanoparticle systems are anticipated to show unique plasmonic properties. 1917,1918 The synthesis and structural characterization of DNA catenanes has been reported previously, 1847,1849,1919,1920 as has a three-station directional DNA catenane rotary motor. 1921 A three-ring catenated nanostructure has been used to program the assembly of two Au nanoparticles (of 5 and 10 nm diameter, Figure 73 ). One of the nanoparticles was attached to ring α and the other to ring γ (L1). Ring α included two equivalent domains, I and II, which are complementary to domain III on ring γ. While the initial duplex between rings β and γ in region III is energetically favored, treatment of the linear catenated system L1 with a fuel strand (Fu) displaces ring γ, which then translocates through hybridization to either site I or II of ring α (structures M1 or M2). In the presence of the appropriate blocker units (B), the directional translocation of ring γ proceeds below or above the central ring β to yield the configurations M3 or M4, respectively. The changes in proximity of the nanoparticles led to changes in plasmonic coupling, which was theoretically modeled. Such systems could have uses in nanomedicine and intracellular diagnostics. 1916 Figure 73 (a) Programmed migration of two Au nanoparticles. (b) STEM images corresponding to the different structures; the bar corresponds to 20 nm. Reprinted with permission from ref (1916). Copyright 2013 Nature Publishing Group. 9.4 DNA Walkers The controlled and predictable strand displacement of DNA and RNA has been used to construct complex devices. 1369,1869,1922−1932 Among them, walkers represent a major subclass. The basic principle behind 1-D walking is illustrated in Figure 74 . A relatively rigid and stable DNA track functionalized with protruding single-stranded anchorage points (footholds) is typically used. 1369,1933 Unlike the direct interactions of biological walker proteins with polymeric tracks, the footholds are attached to the track and bind the walker unit by base paring. The walker unit consists of regions with different nucleotide sequences (feet), which are attached to the track via another DNA fragment (the anchor strand), that can hybridize with both the walker and one of the footholds leaving a “toehold region” unpaired. Addition of the fuel strand results in base-pairing with an anchor strand, starting from the protruding toehold, and forms a more thermodynamically favored waste duplex. The removal of the anchor strand liberates the foothold and one foot of the walker. Walking is processive because the other foot is still attached to the track by a holding strand, which provides additional stabilization. The walker can be reattached to the track via the addition of another anchor strand. Figure 74 Schematic representation of a DNA-based walker and track system. Footholds protrude from the track as single-stranded DNA fragments. Anchor and holding strands enable the walker unit to bind to the footholds by hybridization. Areas with functional importance are labeled, and complementary strands are depicted in the same color for clarity. Stimuli-dependent positional DNA switches 1934,1935 and DNA-based devices performing multistep organic synthesis while migrating along a track have been reported. 1447 An in silico “tumbleweed” walker design has also been proposed. 1936 The first DNA-based walker was nonautonomous and bipedal and was reported by Sherman and Seeman. 1834 High dilution conditions were used to prevent the walker from scrambling betwen different tracks. Sequential addition of two different anchor and fuel strands in an aqueous buffer at 16 °C led to the desired walking motion being obtained. The products were characterized by polyacrylamide gel electrophoresis (PAGE). The energy required for directional walking was provided by the additional base-pairing in the waste duplex (red toehold region, Figure 74 ). Mechanistically, this walker is an inchworm walker because the leading foot remains the same throughout the operation. A hand-over-hand DNA walker (the mechanism of operation of kinesin) has been published by Shin and Pierce who used a similar design. 1833 Transport of a cargo over a DNA origami tile and the synthesis of nanoparticle sequences have been reported. It used a DNA origami walker unit with four “feet” for controlled movement, and three “arms” for picking up cargo, each consisting of single strands of DNA (Figures 75 and 76). 1455,1937,1938 Fuel strands (Fi) were used to drive the motion and to remove the anchor strands (Ai). Each station was loaded with a distinct gold nanoparticle cargo and could be switched between states where cargo delivery was possible and where it was not. By manipulating the selective release of each foot from complementary strands of DNA on the DNA tile, the movement of the walker could be controlled. When coupled to the ability of the stations to be switched “on” or “off”, this allowed the formation of eight, differently composed, noncovalently bound products from the full operational sequence. This remarkable level of control on the nanoscale shows that the forces of Brownian motion can be exploited to great effect in the synthesis of complex supramolecular products. Transportation of a DNA cargo on a DNA origami tile over 16 consecutive steps has also been reported. 1939 Figure 75 (a) Structure of the DNA walker with four “feet” (F1–4) and three “hands” (H1–3). (b) Movement of walker across the DNA origami tile driven by sequentially added DNA “fuel” strands labeled FA. (c) Loading of cargo onto DNA walker. Reprinted with permission from ref (1455). Copyright 2010 Nature Publishing Group. Figure 76 (a) Operation of DNA-based walker. (b) AFM images of walker. Reprinted with permission from ref (1455). Copyright 2010 Nature Publishing Group. Turberfield, Reif, and Yan have reported autonomous walking in a DNA-based system. 1379 Three enzymes, the PfIM I and BstAP I restriction enzymes, and T4 ligase, were required for autonomous operation. Successive cleavage and ligation of the walker-foothold duplex resulted in directional walking along the track. Two autonomous burnt-bridges walkers have been reported using either a restriction endonuclease as an additive 1940 or a walker unit with intrinsic DNAzyme activity. 1380,1941,1942 An autonomous DNA motor whose propulsion was driven by random polymerization has been synthesized. 1829 DNA strands were propeled at the growing end of the polymer by the energy gained from hybridization. A number of enzyme-free autonomous DNA walkers have since been published. 1836,1943−1945 Autonomous multipedal walkers 1842,1946 and walkers performing autonomous and progressive acylation reactions have been disclosed. 1447 Autonomous DNA walkers that are driven by pyrene-mediated photocleavage of disulfide bonds, 1947 or photoisomerization-dependent hindrance/exposure of a complementary strand, 1383 have been designed. A vital requirement for useful cargo transport is the ability to choose the correct path at a junction point. The walker should be capable of crossing the junction, and the path choice should be programmable. Turberfield et al. were able to autonomously regulate the transport of a DNA cargo on a branched track (Figure 77 ). 1843 Successive Holliday junctions formed between the selected fuel, foothold, and the cargo as the walker migrated. These led to strand exchange (equivalent to junction migration), followed by loop opening, which allowed the controlled migration of the cargo from one foothold to another. The same group have since used a single nucleoside, adenosine, to control the route taken at the track junction. 1381 Figure 77 Walker with a choice of path at a junction. (a) Initially cargo resides on position W, and (b) the use of different set of fuels (F) leads to transport of cargo to either position. A displacing strand (Dz) was used for the fragmentation of the molecular ensemble and subsequent analysis of the cargo position. Reprinted with permission from ref (1843). Copyright 2011 American Chemical Society. 10 Conclusion and Outlook Perhaps the best way to appreciate the technological potential of controlled molecular-level motion is to recognize that molecular machines lie at the heart of every significant biological process. Over billions of years of evolution, nature has not repeatedly chosen this solution for achieving complex task performance without good reason. In stark contrast to biology, none of mankind’s fantastic myriad of present-day technologies exploit controlled molecular-level motion in any way at all: every catalyst, every material, every polymer, every pharmaceutical, and every reagent all function through their static or equilibrium dynamic properties. When we learn how to build artificial structures that can control and exploit molecular-level motion, and interface their effects directly with other molecular-level substructures and the outside world, it will potentially impact on every aspect of functional molecule and materials design. An improved understanding of physics and biology will surely follow. As indicated by the many examples that appear in this Review, the future for the field of artificial molecular machines is very bright. 1948 After a somewhat difficult period in the 1990s, when chemists struggled to understand the basic differences between machines at the macroscopic and nanometer length scales, scientists now have the know-how and synthetic tools available to enable them to make suitable machine architectures (e.g., catenanes, rotaxanes, overcrowded alkenes, molecules that walk upon tracks, etc.). They can switch the position of components (often by clever manipulation of noncovalent interactions between the various parts), they understand how to use ratchet mechanisms to create motor-mechanisms, and they are learning how to introduce them into more complex molecular machine systems. Yet there are still several basic challenges to overcome for molecular machines to be able to fulfill their potential: (i) In contrast to motor proteins, powered by ATP hydrolysis or proton gradients, there are as yet no chemically driven synthetic small-molecule motors that can operate autonomously (i.e., move or rotate directionally as long as a chemical fuel is present), the closest counter-examples being the Feringa overcrowded alkene motors that rotate continuously under irradiation with light. 429−453 (ii) Most of the synthetic molecular machines reported to date are based on only a single functioning part that moves, stops, fluoresces, catalyzes a reaction, etc. To perform tasks that cannot be accomplished by conventional chemical means, it will be necessary to design systems with multiple integrated parts, each component performing a dedicated role within the machine ensemble. This will not be straightforward because unlike a watch where the second hand, say, does not interfere with the components in the escapement mechanism, the components of a chemical machine are not easily isolated from each other (or the environment) and interference from one reactive part of a machine with another will be a significant issue as complexity increases beyond the current rather trivial systems. (iii) The machines we are familiar with in the macroscopic world are generally stable, operating unchanged through many cycles, and by and large they do not make “mistakes”. This contrasts with those in the biological world where the stochastic nature of molecular dynamics mean that motors stall, or step backward, or detach from tracks, or make errors in synthesis that are spotted or corrected by other machines. These are intrinsic differences that require fundamentally different philosophies in terms of the way machines carry out tasks that chemists have not yet started to tackle. (iv) The “reading” of a sequence of functional groups on a polymer stand (mRNA or DNA) is how biological molecular machines are programmed to carry out synthesis operations in the correct sequence. Although it is possible to use biological polymers to do this, molecular machines made from DNA are obviously far more limited in terms of operating conditions, chemical stability, and functionality than wholly synthetic systems. As yet there is no small-molecule “Turing machine”, 1949 that is, a molecular machine that can read information from a symbol-encoded molecular strand. Programming of small-molecule machines with an information-rich non-DNA strand (rather than the sequential addition of chemical stimuli, say) would be a game-changing development. There are also issues on which protagonists in the field have differing points-of-view. The debate continues over whether it will ultimately prove more productive to build molecular machines based on macroscopic objects (e.g., Stoddart’s “molecular elevator” 730 and “molecular pistons” 527 and Tour’s “nanocars”, 1129−1132 etc.) or based on biological machines (e.g., artificial systems that seek to reproduce aspects of the behavior of kinesin 1398−1400 or the ribosome 666 ). Certainly mimicking biology is not the only way to achieve complex functionality: computer chips are manufactured from silicon wafers rather than being wet and carbon-based like our brains. Yet equally machines have to be designed according to the environment they are intended to operate in, and there may be reasons why biology uses motors and not switches, and tracks and not wheels, to transport molecular cargoes. Or it may be that evolution just did not discover these solutions to such problems and that mankind, with the whole of the periodic table and known synthetic chemistry to work with, can. Perhaps the most productive approach will ultimately be found by following neither of these lines of investigation too closely, for example, by using chemical principles for “molecular robotics” in which ratcheted motions of molecular components (i.e., biologically inspired mechanisms) are used to perform tasks that have their origins in innovations introduced to advance developments in macroscopic technology (e.g., factory assembly lines). The environments that molecular machines can most productively operate in also remain an open question. Efforts to incorporate molecular machines into regular arrays within crystalline solids (e.g., MOFs) are progressing apace and should allow tethered molecular machines to interact effectively with solution- or gas-phase substrates. It will be interesting to see what applications such systems are being tailored for. Molecular machines mounted on surfaces should allow for surface properties to be changed in a facile and effective manner using a minimal amount of the high-cost molecular structures. However, whether monolayers of molecular machines can be sufficiently robust will no doubt be an issue for practical applications. Most biological molecular machines operate while dissolved in solution, a situation that has no equivalent for macroscopic mechanical machines, and so chemists will have to consider carefully how to effectively address and utilize such machines in that environment. Molecular machines have the potential to act productively under each of these conditions; what is important is that the application is appropriate for which to use synthetic molecular machines. In most cases, the cost of synthesis will mean that this will only be the case if there is no way to carry out the task without using a molecular machine. So rotaxane-based “molecular muscles” have to compete with shape-memory polymers (or even stilbene-containing polymers) to be effective, and rotaxane-based switches, which work through the physical displacement of submolecular components, will have to compete with electron movements in silicon for applications. Chemical synthesis is one area where artificial molecular machines may one day be able to perform tasks that are not possible to achieve using conventional chemical methods, a potential “killer app”! A quarter-of-a-century after chemists made the first fledging molecular protomachines, the tantalizing prospect of artificial molecular machines that can perform useful tasks is moving ever closer to becoming a reality. Molecular machines with multiple integrated parts are in design terms a fusion of the familiar with the strange. Examples of mechanical engineering from the world around us provide a conceptual framework for what we want tiny machines to achieve. Biology shows us how machines cope with the nature of the environment in performing tasks at nanometer length scales, and physics explains the often counterintuitive ways that such small objects must behave. Yet ultimately, while drawing on all of these disciplines for ideas, guidance, and inspiration, such machines have to be designed, built, and operated through chemistry, the central science.

Attempts to fabricate mechanical devices on the molecular level have yielded analogues of rotors, gears, switches, shuttles, turnstiles and ratchets. Molecular motors, however, have not yet been made, even though they are common in biological systems. Rotary motion as such has been induced in interlocked systems and directly visualized for single molecules, but the controlled conversion of energy into unidirectional rotary motion has remained difficult to achieve. Here we report repetitive, monodirectional rotation around a central carbon-carbon double bond in a chiral, helical alkene, with each 360 degrees rotation involving four discrete isomerization steps activated by ultraviolet light or a change in the temperature of the system. We find that axial chirality and the presence of two chiral centres are essential for the observed monodirectional behaviour of the molecular motor. Two light-induced cis-trans isomerizations are each associated with a 180 degrees rotation around the carbon-carbon double bond and are each followed by thermally controlled helicity inversions, which effectively block reverse rotation and thus ensure that the four individual steps add up to one full rotation in one direction only. As the energy barriers of the helicity inversion steps can be adjusted by structural modifications, chiral alkenes based on our system may find use as basic components for 'molecular machinery' driven by light.

Motor proteins are nature's solution for directing movement at the molecular level. The field of artificial molecular motors takes inspiration from these tiny but powerful machines. Although directional motion on the nanoscale performed by synthetic molecular machines is a relatively new development, significant advances have been made. In this review an overview is given of the principal designs of artificial molecular motors and their modes of operation. Although synthetic molecular motors have also found widespread application as (multistate) switches, we focus on the control of directional movement, both at the molecular scale and at larger magnitudes. We identify some key challenges remaining in the field.