Molecular chirality at surfaces


  • This article will be included, in edited form, as a chapter in one of the forthcoming volumes of the Encyclopedia “Surface and Interface Science”, edited by K. Wandelt (Wiley-VCH, Weinheim, 2012), ISBN 978-3-527-41156-6.


With the adsorption of larger molecules being increasingly tackled by surface scientists, the aspect of chirality often plays a role. This paper gives a topical review of molecular chirality at surfaces and gives a phenomenological overview of different aspects of adsorption and self-assembly of chiral and prochiral molecules and the principles of mirror-symmetry breaking at a surface. After a brief introduction into the history of molecular chirality and the important role it played for understanding the spatial structure of molecules, definitions of chirality are presented. Topics treated here are principle ways to create single chiral adsorbates, chiral ensembles, and monolayers by achiral molecules, adsorption of intrinsically chiral molecules at achiral and chiral surfaces, long-range symmetry breaking in two-dimensional (2D) crystals due to additional chiral bias, chiral restructuring of solid surfaces under the influence of chiral molecules, switching the handedness of adsorbates, and chirality at the liquid/air interface. An outlook onto further potential research directions and recommendations for further reading, including nonsurface-related sources of chiral topics completes this paper.

1 Introduction

Many objects in our world have the property that they are incongruent with their mirror image. Such objects are called chiral or enantiomorphous. Examples are quartz crystals, shoes, snail shells, screws, etc. The most significant manifestation of chirality is the appearance of left- and right-handed molecules, so-called enantiomers (see glossary). These have identical physical properties (e.g., melting and boiling points, viscosity, solubility, hardness, etc.), and, as long as interactions with nonhanded (achiral) systems are considered, their chemical activities are identical as well. Chirality is ubiquitous in the biological world, but handedness comes unbalanced. That is, molecules of life, like sugars, proteins, and their building blocks the amino acids, appear basically in only one handedness. This has dramatic consequences, because the biological and pharmaceutical activity of enantiomers is directly related to their handedness and causes different physiological effects. This accords molecular chirality a paramount importance in pharmacology and the fragrance industry. The odor perceptions of limonene enantiomers, e.g., are clearly different. One form appears in most citrus oils and smells like orange fruit, while the other is abundant in conifer needles and smells like turpentine (but certainly not like lemon, as often claimed by those not familiar with the matter) 1.

Two-dimensional (2D) molecular surface science aims at understanding molecular recognition, i.e., the nature and consequences of intermolecular interactions. The identical alignment of molecules due to adsorption imposes a directionality among lateral interactions. In addition, single-crystalline surfaces as substrates introduce symmetry within the plane. Analysis of 2D self-assembly with regard to molecular symmetry, functional groups, and substrate symmetry has become widespread with the rise of molecular surface science. Among these studies, the aspect of chirality at surfaces has also received increasingly attention. Chirality plays an important role in crystallization, and pioneering studies at the liquid/air interface were motivated by this fact 2. In particular enantioselective heterogeneous catalysis triggered studies on molecular chirality at solid surfaces. For instance, a chirally modified surface steers an achiral molecule into a specific chiral alignment so that hydrogenation of C[DOUBLE BOND]C or C[DOUBLE BOND]O double bonds, directed from a metal surface underneath, gives only one enantiomer (Fig. 1) 3, 4. Achiral molecules that turn chiral due to an interaction, e.g., with a surface, are called prochiral. Essentially, two enantioselective heterogeneous catalysts have been established so far: tartaric acid-modified nickel for hydrogenation of β-ketoesters 5 and cinchona-modified platinum catalysts for hydrogenation of α-ketoesters 6.

Figure 1.

(online color at: The principle of enantioselective heterogeneous catalysis. Adsorption of prochiral phenyl acetone leads to chiral adsorbates. Hydrogenation of the double bond by atomic hydrogen coming from the surface is unidirectional. The product is chiral after desorption. If a coadsorbed chiral modifier (not shown) suppresses one of the mirror-like alignments at the surface, a single enantiomer of 2-phenyl isopropanol will be created.

Other important examples are biomineralization and chirally modified surfaces for chiral liquid chromatography. For the latter, enantioselective interactions cause different residence times for enantiomers, thus allowing separation of chiral compounds 7. When organisms create biominerals, like teeth, shells, spines, and spicules 8, chiral macromolecules (proteins and polysaccharides) interact with achiral minerals, perfectly controlling their polymorphism and morphology. This is highly sophisticated surface chemistry, not well understood, and difficult to investigate directly at the molecular level. As for heterogeneous catalysis, these complex materials systems call for defined model studies and the tools of the surface scientist.

It is impossible to cover here all published work concerning chirality at surfaces in this single article. Rather I give a phenomenological overview, addressing different aspects of adsorption and self-assembly of chiral and prochiral molecules and principles of mirror-symmetry breaking. Not all examples given are illustrated with figures, some examples are covered in more detail and others only very briefly, in which it might be necessary to consult the original literature to obtain a more detailed picture of the particular case. There are several reviews that have covered various aspects of the field, in part with a different perspective 9–18.

The remainder of this section introduces the concept of chirality in the form of a brief review of its history, including the problem of determining the absolute configuration of chiral molecules and stating different definitions available. Section 2 discusses principle ways to create single chiral adsorbates from achiral molecules, while Section 3 treats chiral ensembles and monolayers formed by achiral molecules. Adsorption of intrinsically chiral molecules at achiral and chiral surfaces is covered in Sections 4 and 6, respectively, intermitted by the topic of long-range symmetry breaking due to chiral bias (Section 5). Section 7 gives an overview of restructuring solid surfaces under the influence of chiral molecules, Section 8 presents chiral switching of adsorbates, and Section 9 briefly introduces chirality at the liquid/air interface. Finally, Section 10 gives an outlook and recommendations for further reading.

1.1 Historical development of stereochemistry

Although chemists usually draw their chemical formulas in 2D on paper, molecules are arrangements of atoms in space. During the last quarter of the 19th century, when many scientists (among them the great physical chemist Wilhelm Ostwald) had serious doubts if atoms exist at all, organic chemists set out to unravel stereochemistry, i.e., the spatial positions of atoms in a molecule 19. Today, it is difficult for us to comprehend the true novelty of van't Hoff's and Le Bel's concept of the carbon tetrahedron, which has laid the foundation of stereochemistry. A proper perspective is given by Victor Meyer's comment that van't Hoff's work “exposed gold mines” 20.

Before Laue's discovery of the diffraction of X-rays in crystals in 1914 and the introduction of UV–Vis spectroscopy at about the same time, the only known physical phenomenon to address molecular structure had been optical activity. That is, the rotation of plane-polarized light when traveling through matter. Plane-polarized light had been discovered by Malus in 1809 21. The optical activity for quartz as well as its optical rotatory dispersion (ORD), i.e., different optical rotation for different wavelengths of light, was then reported by Arago only 2 years later 22. After Haüy called attention to the enantiomorphism (the right- and left-handedness in shape) of quartz (Fig. 2), it was the British astronomer Herschel who connected the sign of optical rotation with this enantiomorphism of quartz crystals. At that time, Biot had already discovered that organic liquids (e.g., turpentine and citrus oil) and solutions of organic solids, e.g., crystals of tartaric acid (HOOC[BOND]HCOH[BOND]HCOH[BOND]COOH) found in wine casks, show optical activity.

Figure 2.

(online color at: Enantiomorphism of crystals. “Left-” and “right-quartz” can be distinguished by their “hemihedral” faces (arrows), bestowing opposite handedness to the crystals. The microscopic origin lies in the opposite helicity of SiO4 tetrahedra arranged along the c-axis.

It was then the genius Louis Pasteur who went beyond all these known features in a seminal experiment that suggested handedness in molecular structure. When he studied crystals of ammonium sodium salts of the so-called paratartaric acid, a compound isolated from the overheated deposit of fermented grape juice, he found enantiomorphous crystals of opposite handedness. Most intriguingly, their solutions displayed opposite optical activity (Fig. 3) 23. This was the first time that enantiomorphism of a macroscopic crystal had been connected to the properties of an isotropic arrangement of molecules, and hence to the geometric properties of the molecules themselves. Today we know that the overheated tartar is a mixture of left- and right-handed molecules – called a racemate – of tartaric acid. Its ammonium sodium salt had crystallized in Pasteur's laboratory into a so-called conglomerate, i.e., a mixture of “homochiral” crystals, each of which contains only one kind of the two enantiomers. Twelve years later in 1860, Pasteur mentioned in his lectures to the French Academy that, among others, an asymmetric tetrahedron should show such left–right “dissymmetry” 24.

Figure 3.

(online color at: Sketch of Pasteur's experiment. Aqueous solutions of enantiomorphous crystals of ammonium sodium tartrate rotate plane-polarized light either CW (top) or CCW (bottom), thus connecting handedness of macroscopic crystals with those of molecules.

This relation between crystal shape and optical activity motivated Le Bel to reflect on molecular structure 25, while van't Hoff was motivated by the problem of “excess isomers” of lactic acid 26. Both suggested practically simultaneously in 1874 that the ligand atoms of carbon form a stable tetrahedron and that four different ligands will cause so-called “optical isomerism”, i.e., isomers that support optical activity. In the course of further development of stereochemistry chemists were soon able to elucidate unsolved problems with this new theory. In particular, Emil Fischer, who became the second Nobel Prize laureate in chemistry after van't Hoff, deducted the whole chemistry of sugars based on van't Hoff's and Le Bel's proposal. Also noteworthy is the fact that modern coordination theory of metals, as introduced by Alfred Werner in 1893, has been proven by predicting that certain octahedral complexes are chiral and by demonstrating their optical activity after separation of the enantiomers 27.

1.2 Absolute configuration and stereodescriptors

Optical rotation is a pseudoscalar and has opposite signs for enantiomers (Fig. 3). Therefore, these were distinguished early on by their sense of rotation of plane-polarized light as dextro- or laevorotatory and marked with the prefixes d- and l- or (+) and (−), respectively. The sense of rotation, however, does not necessarily designate the absolute positions of atoms in space (configuration) of the optically active molecules. Emil Fischer performed chemical transformations in order to correlate the configurations of asymmetric carbon atoms in different compounds. As denomination for the absolute configuration of an asymmetric carbon atom in carbohydrates Fischer introduced D- and L- as prefixes. Although it was eventually replaced by the much more consistent R/S nomenclature, Fischer's system is often used today. Due to the similarity to the d/l prefixes for optical rotation, however, the D/L system may cause confusion. Being aware that chemical methods alone could not elucidate the absolute configuration, Fischer arbitrarily attributed them to (+)-glucose 28. Rosanoff refined this system later by using (+)-glyceraldehyde as standard by assigning a D-configuration to it (Fig. 4) 29. In the following years chemists could relate many chiral compounds to the glyceraldehyde standard and found that eventually all natural sugars have D-configuration, while the chiral proteomic α-amino acids all have L-configuration. Until the first experimental determination of absolute configuration, these assignments had a 50% probability of being wrong.

Figure 4.

The arbitrary assignment of absolute D-configuration to (+)-glyceraldehyde (top). The molecule is positioned so that the H atom and the hydroxyl group at the asymmetric carbon point toward the observer. For the shown D-enantiomer the OH group is at the right side (for the L-enantiomer OH would be on the left side). The Fisher projection draws the tetrahedron as a cross, with the OH group and the H atom being assumed to be above the paper plane, while the CHO and CH2OH groups are behind the paper plane. Bottom: CIP rules for assigning the absolute configuration. Based on atomic number, the ligands get different ranks in order to determine a CW (R) or CCW (S) progression.

According to Friedel's law, conventional X-ray diffraction patterns are centrosymmetric, providing no information on handedness of molecules 30. In other words, conventional X-ray scattering gives bond distances, but cannot deliver absolute positions of atoms because the path (phase) differences are identical for enantiomers. However, in 1951 Bijvoet and coworkers finally confirmed the Fischer–Rosanoff assignment by determining the absolute configuration of sodium rubidium (+)-tartrate via anomalous X-ray scattering 31. Using zirconium Kα wavelength, which is absorbed by rubidium, a phase lag was introduced, which allowed to distinguish “front-and-back” positions of atoms. Years before, his method was therefore established for studying polar crystals 30. At about the same time when Bijvoet in Utrecht successfully introduced his method, Kirkwood at Cal Tech in Pasadena came to similar conclusions by theoretical prediction of the sense of optical activity for small chiral molecules 32.

Today, absolute configuration is denominated according to the Cahn–Ingold–Prelog (CIP) system, where (R)- and (S)- (from lat. rectus and sinister for right and left) are used as prefixes. An example of an asymmetric carbon atom is shown in Fig. 4. The CIP system also covers chiral molecules without any stereogenic center, like allenes, biphenyls, etc. Helical hydrocarbons without any asymmetric carbon atom (so-called helicenes) are marked with M or P prefixes for left-(minus)- and right-(plus)-handedness, respectively. Although the CIP rules passed extensive tests of consistency, they have the disadvantage that the descriptor can change without changing the configuration at the stereogenic center. Among the common natural proteomic L-α-amino acids, e.g., cysteine and its dimer cystine are the only (R)-enantiomers.

Quite remarkably, modern surface science played a vital role in the quest for determining absolute configurations. During the 1970s, Bijvoet's analysis was challenged by Tanaka, who came to the opposite conclusion 21. Because anomalous X-ray scattering had also been used for the structure determination of polar crystals, the validity of the method was confirmed by ion scattering performed on ZnS surfaces 33. Finally, the absolute configuration of tartaric acid was directly confirmed by X-ray photoelectron diffraction (XPD): Adsorbed on Cu(110) as double-deprotonated tartrate species, the absolute handedness of both tartaric acid enantiomers was determined due to the pronounced forward scattering of photoelectrons caused by neighbor atoms (Fig. 5) 34. The same method delivered the absolute helicity of heptahelicene after uniform alignment on the terraces of a Cu(332) surface 35 and was later confirmed by comparing calculated and measured vibrational circular dichroism spectra 36, i.e., the differential absorption of circular polarized IR light.

Figure 5.

(online color at: Determination of the absolute configuration of tartaric acid adsorbed on Cu(110) 34. (a) Sketch of a carboxylate C 1s photoelectron forward scattering mechanism in tartrate. (b) The hemispherical angular distribution of C 1s photoelectrons reveals directly the positions of the scattering atoms by retrotriangulation. (c) Forward scattering directions of C 1s photoelectrons are numbered and indicated by arrows. These originate from nearby atoms of the electron emitter and lead to the corresponding intensity maxima in (b).

1.3 Definitions of chirality

The word “chiral” stems from χειρ, the Greek expression for hand. It was first mentioned in a footnote in the Second Robert Boyle Lecture, delivered by Lord Kelvin in 1893 at Oxford University 37. Kelvin published in 1904 his definition of chirality as footnote in an appendix of his Baltimore Lectures delivered in 1884 at Johns Hopkins University 38, 39: “I call any geometrical figure, or any group of points, chiral, and say it has chirality, if its image in a plane mirror, ideally realized, cannot be brought to coincide with itself.” Much earlier Kelvin had already introduced the terms “homochiral” and “heterochiral” 40: “The similarity of a right-hand and a left-hand is called heterochiral: that of the two right hands, homochiral. Any object and its image in a plane mirror are heterochirally similar.” Pasteur's “dissymmetry” or alternative expressions describing handedness used in chemistry so far, were progressively dropped, while use of chirality prevailed from about the end of the 1950s 41, 42.

Chirality does allow symmetry – a helix, e.g., has C2 symmetry – and today's official definition, as provided by the International Union of Pure and Applied Chemistry (IUPAC), takes this circumstance into account: [Chirality is] “the geometric property of a rigid object (or spatial arrangement of points or atoms) of being nonsuperposable on its mirror image; such an object has no symmetry elements of the second kind (a mirror plane, a center of inversion, a rotation-reflection axis). If the object is superposable on its mirror image the object is described as being achiral.” Consequently, the symmetry group of an achiral object contains symmetry operations of the second kind (Fig. 6).

Figure 6.

Examples of symmetric chiral (top) and symmetric achiral (bottom) objects.

There is ongoing confusion about chirality when movement or electric and magnetic fields are involved. This motivated Barron to suggest the following definition: “True chirality is exhibited by systems that exist in two distinct enantiomorphous states that are interconverted by space inversion but not by time reversal combined with any proper spatial rotation” 43. The importance of this definition lies in the fact that only time-invariant chirality supports physical observables like ORD.

Unfortunately, with chirality becoming part of surface science the term “2D chirality” is often used nowadays. Planar molecules aligned into the plane (i.e., 2D confinement) by a surface or a physical field still suffer a polarization of their electron cloud perpendicular to that plane, giving it an up and a down; so it is a truly three-dimensional (3D) object. It goes without saying that when surface atoms are also taken into account, i.e., when the adsorbate as a whole is considered (as it always should), we deal with a 3D chiral entity. Although scholarly discussions on chirality in different dimensions can be quite useful 44, 45, 2D chirality remains a purely theoretical construct. For real experimental systems, this term is inappropriate and misleading.

2 Symmetry breaking I: Single achiral molecules at achiral surfaces

The surface-induced “up-down” asymmetry allows only symmetry elements perpendicular to the surface plane. Due to this lower dimensionality, symmetry breaking at surfaces is a common phenomenon. Often, the process of adsorption alone is sufficient to create a chiral adsorbate from a prochiral molecule, but sometimes a polar or azimuthal tilt or the rearrangement of the molecular backbone is responsible for mirror-symmetry breaking. In this section we will list the different mechanisms leading to single chiral adsorbates all the while both ingredients, molecule and surface, are achiral. Without further bias, both enantiomers will always be created upon adsorption. Single chiral adsorbates are sometimes addressed as point chirality 46 and may possess C1, C2, C3, C4, or C6 point symmetry 18.

2.1 Adsorption-induced chirality

Most molecules form a chiral adsorbate, because the possibility to engage in an achiral adsorbate complex is very unlikely for energetic reasons. A simple example is propene on Cu(211) with its enantiotopic sides turning toward the surface. Enantiotopic sides differ only by the stereochemical outcome of their interaction with another reactant (surface or chemical attack). The driving force of adsorption of propene is the interaction of the C[DOUBLE BOND]C double bond with a surface copper atom located at a step 47. An alternative achiral possibility would require binding to the surface via hydrogen atoms, but that would be a weak physisorption. As discussed for the prochiral reactant in heterogeneous enantioselective catalysis (Fig. 1), the two resulting static adsorbate configurations are enantiomers (Fig. 7). Rotation and translation within the plane do not allow superposition of both species. A dynamic process, allowing rolling or flipping of the adsorbed species outside the plane, however, will cause inversion of enantiomers. Although propene is not planar, it might be approximated here as the letter L. Whenever molecules with an L- or Z-shape are adsorbed with their plane parallel to the surface, chirality (with C1 and C2-point symmetry) is the consequence (Fig. 7) 48–51.

Figure 7.

(online color at: Examples of molecules that turn chiral upon adsorption 47–51. (a) Propene, (b) nitronaphthalene, (c) 2,6-dibromoanthrachinone, (d) N,N′-dihexadecylquinacridone, (e) naphtho[2,3-a]pyrene. (c) and (d) have C2 point symmetry, all other examples are of C1. Because of the equal probability that either enantiotopic side faces the substrate underneath, both enantiomers are created.

2.2 Oblique alignment

If an achiral molecule is aligned on the surface in a tilted configuration, a chiral adsorbate may be the result. Figure 8a shows water as an example. A water molecule can be considered to be part of the turn of a helix, and when superposition by translation and rotation within the layer is not possible, it becomes chiral. This effect has also been confirmed by calculating its optical rotatory power 52. A similar case is chiral methane, where chirality originates from different distances of the four hydrogen atoms to the surface 10. For such chirality induction just the presence of a surface is required. Taking also surface crystallinity into account, even a single carbon monoxide molecule can create a chiral adsorbate. Thereto must the molecule be tilted away from any symmetry direction of the surface (Fig. 8b). If adsorbed on a bridge site between different atoms, e.g., on an alloy or oxide surface, a polar tilt even parallel to a crystal axis creates a chiral motif. Although these examples are hypothetical, they illustrate the principles how mirror symmetry can become broken when small molecules adsorb at crystal surfaces.

Figure 8.

(online color at: (a) Chiral water. Side view of the tilted configuration leading to enantiomers that are not superimposable by rotation and translation. (b) CO on a quadratic grit. An oblique tilt away from any symmetry direction on the surface induces a chiral motif, if at least two surface atoms are also taken into account. (c) Ball and stick model for copper phthalocyanine (hydrogen atoms are omitted) on Ag(100) 53. A chiral complex results from CW or CCW alignment. Blue balls represent nitrogen atoms.

Symmetric planar molecules adsorbed parallel to the surface often create a chiral motif as well. Figure 8c shows enantiomers of copper phthalocyanine (CuPc) on Ag(100). The four outer azamethine nitrogen atoms interact directly via their electron lone pair with four silver atoms. It is this special nitrogen–silver interaction that causes the symmetry breaking 53. The four inner pyrrole nitrogen atoms are therefore offset with respect to the nearby Ag surfaces atoms either in a clockwise (CW) or counterclockwise (CCW) manner.

2.3 Chiral footprints

In the previous section, we assumed that the structure of the adsorbed molecule remains unchanged with respect to the gas-phase species. However, when a flexible molecule interacts with a surface, it is unlikely that a molecular symmetry plane is always maintained for the whole adsorbate. Adsorption of the achiral amino acid glycine on Cu(111), Cu(100), and Pt(111) has been studied from early on 54, 55, but the first hint of a chiral footprint did not come before 2003. Photoelectron diffraction and theoretical methods showed that glycine interacts with its amino and carboxyl group with the surface such that one group turns to the left or to the right with respect to the other (Fig. 9a) 56–58.

Figure 9.

(online color at: (a) Top views of chiral footprints of prochiral molecules at surfaces. All carboxyl groups are deprotonated. The achiral amino acid glycine can be oriented so that the amino group points to the right or to the left. Succinic acid and (R,S)-tartaric acid undergo a zigzag distortion on Cu(110) 61. (b) With its benzene ring oriented parallel to the surface, 1,4-bisdodecylbenzene may occupy a chiral or achiral conformation on a surface.

Achiral dicarboxylic acids, like (R,S)-tartaric acid or succinic acid can interact with both carboxylate groups with a Cu(110) surface. This leads to chiral footprints and a zigzag distortion of the molecular frame 59, 60. Such footprints were suggested by density functional theory (DFT) 61, and, as already mentioned, were confirmed for the intrinsically chiral (S,S)- and (R,R)-tartaric acid enantiomers with XPD 34.

A chiral adsorbate can also result by adaptation of an according conformation. Long alkyl chain derivatives have been popular for studying self-assembly on highly oriented pyrolytic graphite (hopg). Arachidic anhydride [(C19H39CO)2O], e.g., is a symmetric molecule, and one could expect to observe symmetric structures. However, only the chiral form was observed on hopg 62. This applies to 1,4-bisdodecylbenzene (Fig. 9b) as well 63. As explanation for the missing achiral forms, a denser packing of the 2D crystal or a better alignment of the alkylchains with the underlying substrate must be considered.

2.4 Chemical reactions

We have seen already that hydrogenation of C[DOUBLE BOND]O double bonds can create chiral alcohols (Fig. 1). Cycloaddition reactions of hydrocarbons on silicon surfaces are also a good example of surface stereochemistry. The Si(100) surface consists of Si dimers and each atom of a dimer has a dangling bond. Unsaturated hydrocarbons undergo cycloaddition reactions at Si(100), i.e., they break up the C[DOUBLE BOND]C double bonds and form new bonds to the Si atoms. Figure 10 shows an example for trans-2-butene 64. The Z-shaped physisorbed trans-2-butene would create a chiral adsorbate, similarly to propene discussed above. The chemical reaction, however, creates a covalently bonded cyclic adsorbate with two asymmetric carbon atoms, allowing the denomination via the CIP system. Both Si2C4H8 enantiomers are presented in Fig. 10. This experimental system was actually used to demonstrate that scanning tunneling microscopy (STM) can reveal the absolute configuration and to show that adsorption of cis-2-butene creates achiral (R,S)-2,3-butylene silicon 64. A similar study was performed with styrene on Ge(100) 65.

Figure 10.

(online color at: A [2 + 2] cycloaddition of trans-2-butene and a Si surface dimer creates a chiral adsorbate 64.

As mentioned in the introduction, chirality played an important role in proving the coordination chemistry in 3D metal complexes. That is, metal atoms or ions and achiral ligands form chiral coordination compounds. A 2D metal complex formed after coadsorption of iron atoms and trimesic acid (TMA) also shows handedness in the way the ligands are oriented around the central atom (Fig. 11) 66. As in the case of CuPc (Fig. 8c), this surface-synthesized adsorbate has C4 point symmetry. However, in contrast to CuPc this metal complex adsorbate is chiral without considering the detailed surface structure (position of surface atoms).

Figure 11.

(online color at: A chiral Fe-TMA complex formed on Cu(100) after codeposition of the individual components 66. The ligands can be attached to the central metal atom in a CW or CWW motif. Red balls present oxygen atoms, dark gray balls carbon atoms. Hydrogen atoms are omitted. Cu(100) substrate atoms are shown in light gray.

2.5 Perturbation of parity eigenstates

In 1927 Friedrich Hund (1896–1997) delivered a quantum-mechanical description of molecular vibrational and rotational spectra. Within this framework he encountered a paradox, namely, that chiral molecules should not exist at all. Because the Hamiltonian is invariant with respect to space inversion, left- and right-handed states are not Hamiltonian eigenstates. Due to a finite tunneling probability the molecules will therefore be found in superpositions of the handed wavefunctions, that is, in parity-even and parity-odd states (Fig. 12) 67. A good (theoretical) example is the isotope-enantiomer of ammonia NHDT (Fig. 12a). One can safely assume that NHDT will be – just like NH3 – in a superposition of handed states (Fig. 12d) and will show the absorption of electromagnetic microwaves due to transition from parity-even (ψ+) to parity-odd (ψ−) levels (Fig. 12d). Hence, the free NHDT constitutes a molecule that should never be found in a chiral state! In principle, this applies to carbon tetrahedrons as well, but clearly contradicts experience. Amino acids, e.g., are clearly found in their chiral states. In the first paper ever dealing with quantum-mechanical tunneling, Hund circumvented this paradox by connecting the tunneling probability with the barrier height of inversion, thus increasing the lifetime of chiral carbon compounds to billions of years 67. However, chiral molecules with low tunneling barrier exist in the parity eigenstates and will not show any optical activity.

Figure 12.

(online color at: Just like ammonia chiral NHDT (a) or H2S2 (b) must be described by the double-well model (c). A finite tunneling probability between left- and right-handed states leads to the superposition of parity-even (ψ+) and parity-odd (ψ−) states (d), in which the molecules do support dipole operator observables (e.g., light absorption), but not optical activity. The chiral buckybowl Me5COR (e) converts at temperatures above 206 K rapidly between its enantiomers (e) 70, but adsorption sets a single handedness, as revealed by STM (f) (image size: 6 nm × 6 nm) 72.

This dilemma encouraged many theoreticians to seek for clarification, predicting that perturbation by the environment via external fields or forces could cause decoherence of left–right superposition states 68. Molecules of the type H2X2 (X = O, S, Se, Te, etc.) with low rotation barriers of the X–X axis (Fig. 12b) are also expected to be in a left–right superposition, but recent calculations show that collisions with He atoms knock D2S2 back into handed states quickly 69. The consequence of adsorption is therefore mundane: the molecule will turn chiral at the surface (as long as the U-shaped achiral planar transition state is not enforced by adsorption). NHDT should also be found in either handed state as long as its electron lone pair interacts strongly with a metal surface.

An experimental example for such dynamic symmetry breaking is the chiral bowl-shaped fullerene fragment-molecule pentamethylcorannulene (Me5COR, Fig. 12e). The free molecule converts between both enantiomeric states at room temperature at a rate of 200 kHz 70. The activation barrier of this conversion is 8.7 kcal mol−1 71. When the conversion is blocked upon adsorption, the sense of handedness becomes defined with a 50% probability. The STM image (Fig. 12f) shows the position of the methyl groups at low temperature, and because the concave side of the bowl faces the surface, handedness is set 72.

3 Chiral ensembles and monolayers of achiral molecules

Of the 230 space groups of 3D crystals 65 are enantiomorphous. Enantiopure molecules always crystallize into one of the enantiomorphous space groups, but racemates pack (almost) exclusively into achiral space groups. Equal numbers of the two enantiomers are in the unit cell, where they are correlated by a center of inversion or a glide plane. Achiral molecules, on the other hand, may crystallize into chiral space groups. The above-mentioned β-quartz is a good example of such spontaneous symmetry breaking in 3D. Other examples are NaClO4, NaIO4, and NiSO4 · 6H2O. The ions are achiral, and in the isotropic phase (solution or melt) no handedness is possessed. In a single crystal of these compounds, however, a chiral space group is taken.

3.1 Symmetry breaking II: 2D assembly induced chirality

In two dimensions, there are just 17 plane groups 73, and only five of them support chirality, i.e., have no mirror symmetry perpendicular to the surface (p1, p2, p3, p4, and p6). For physisorption at the solid/liquid interface, however, 80% of achiral molecules pack into the chiral plane groups 74. It is the combination of close packing of molecules and the provision of specific binding sites on a single crystal surface that leads to adsorbate lattices that break mirror symmetry. Chirality arises here in two different ways. Ignoring the substrate lattice, superposition of the two mirror domain unit cells leaves the molecules nonsuperimposed. The oblique tilt of the adlattice with respect to the substrate lattice, on the other hand, creates itself handedness. For chirality expressed by assembles of chiral or achiral molecules the term organizational chirality has been proposed earlier 46. Even adsorption of atoms may cause mirror-symmetry breaking. For example, a equation imageR19.1° structure has been reported for sulfur and iodine on fcc(111) substrates 75, 76. Nevertheless, this situation is much more often observed for molecules. Benzene, sometimes with coadsorbed carbon monoxide, forms the following structures, e.g.: Ni(111)-equation image R19.1° (Fig. 13), Ru(0001)-equation image R13.9°, Rh(111)-equation image R23.4°, and Os(0001)-equation image R10.9° 77. The example shown in Fig. 13a illustrates the interdigitation of the hydrogen atoms while all C6 rings are located over identical threefold hollow sites. On Ni(111) at lower coverage bridge sites are actually favored. The plane group is p2, but superposition of the unit cells of both mirror domains still leaves the molecules nonsuperposable (13a, bottom). For larger molecules this kind of symmetry breaking becomes actually more probable. For example, the star-shaped hexa-tert-butylhexabenzocoronene on Cu(110) follows this principle 78. Similar structures have been observed for coronene on MoS2, Cu(110), and Cu(100) 15. However, “achiral” (4 × 4) lattices on Ag(111) and Au(111) were observed for coronene as well 15.

Figure 13.

(online color at: (a) Benzene equation imageR19.1° structure on Ni(111) (top) occupying threefold hollow sites 77. The van der Waals (vdW) radii of the molecules are indicated. Congruence of the unit cells of both mirror domains keeps the molecules nonsuperimposed (bottom). (b) STM and LEED reveal mirror domains for corannulene on Cu(110). The unit cell dimensions indicate that specific surface sites govern the self-assembly 79.

Close packing on single-crystal surfaces is always a combination of intermolecular forces and occupation of the favored adsites at saturation coverage. The C5v-symmetric buckybowl corannulene forms mirror domain structures on Cu(110) (Fig. 13b). Although rotation of the structure by 180° brings the unit cell into superposition, the plane group is p1 due to the C5 symmetry of the molecules 79. STM results reveal that the bowl opening points away from the surface. In addition, XPD suggests that the molecule is bonded such that one of the C[BOND]C bonds of the central pentagonal ring is located closest to the surface, possibly on top of a copper atom. The intermolecular distances are larger than observed for the case of Cu(111) at low temperatures 80, but a closer packing on these favored adsites is not possible.

Many close-packed chiral molecular lattices are ruled by interdigitation. Typical examples are porphyrins and phthalocyanines 81–84. These C4v symmetric molecules achieve their densest structures when arranged in a CW or CCW fashion (Fig. 14a). Long alkyl chains at the periphery strongly favor interdigitation and therefore assembly induced chirality (Fig. 14b) 85. This principle applies to 2D self-assembly of the bowl-shaped C3v-symmetric subphthalocyanine (subPC) on Au(111) as well (Fig. 14c) 86. Moreover, when subPc is coadsorbed with C60 on Ag(111), chiral pinwheels are formed in a dense packing 87. Three C60 units are surrounded by three subPC molecules, and six of these pinwheels form a hexagon. Pinwheels of a single domain possess all the same handedness.

Figure 14.

(online color at: Chirality in phthalocyanine (Pc) layers and a chiral Kagomé. (a) Left- and right-handed example of a regular metal-Pc 81. (b) Eight oxyoctyl chains attached to the Pc core enhance the probability of interdigitation 85. (c) Sketch of CW and CCW arrangement of subPC as observed on Au(111) 86. (d) Chiral Kagomé motifs formed by NC[BOND]Ph5[BOND]CN on Ag(111). The shown motifs are part of an extended monolayer 88.

Chiral motifs by self-assembly do not exclusively appear due to close packing and interdigitation. Polar groups and lateral hydrogen bonding may cause mirror-symmetry breaking in supramolecular arrays. A prominent example is dicyanoquinquiphenyl on Ag(111), which forms a chiral Kagomé mesh because of a combination of hydrogen bonding and antiparallel dipole alignment of the nitrile groups (Fig. 14d) 88. Mirror-symmetry breaking was also observed in metalorganic networks of iron terephthalate on Cu(110). Fe2(C6H4(COO)2)4 units combine in CW or CCW motifs via hydrogen bonding between carboxylate groups and hydrogen atoms 89. Multiple H····F hydrogen bonding in networks of fluorinated rigid C3v-symmetric hydrocarbons were also identified to create CW and CCW assemblies at the solid/liquid interface on hopg 90. Even quadrupole–quadrupole interactions have been put forward to explain pinwheel motifs formed by carbon monoxide, flat-lying on hopg at 5 K 91, 92.

For the two last examples of self-assembly induced chirality we return to apolar hydrocarbons. In 1991 Rabe and Buchholz 63 reported the first supramolecular chiral motifs ever observed with an STM. Adsorbed from solution onto hopg para-bisdodecylbenzene self-assembles into supramolecular chiral lamellae (Fig. 15a). Mirror domain boundaries show a glide plane as symmetry element, i.e., there is an offset by approximately half the molecular length plus a reflection. The packing density in the boundary region is almost as good as within domains. This situation reflects nicely a principle that is commonly observed in 3D for racemates: a glide plane acts like a center of inversion, but here at lower dimensionality.

Figure 15.

(online color at: Chirality induced in supramolecular structures of apolar hydrocarbons. (a) Para-bisdodecylbenzene is forced into a chiral conformation and forms handed lamella structures 63. (b) Model of pentacene pinwheels composed of 36 monomers, as observed by STM near monoatomic steps on Bi(111) 93.

A recent STM report by Xue and coworkers 93 revealed an intriguing pinwheel assembly of pentacene on Bi(111). The pinwheels consist of 36 molecules oriented with their molecular plane perpendicular to the surface. Six molecules are oriented parallel forming a hexamer, but adjacent molecules have an offset in position (Fig. 15b), with the two distal molecules of a hexamer shifted to the opposite direction to the proximal four. This gives the hexamer alone a handedness. Only equally handed hexamers are then found in a single pinwheel. Although isolated molecules were found to be oriented with their plane parallel to the surface, within this hexatriaconta-pentacene π–π and H–π forces are at work, which are en bloc larger than the sum of all the dispersion interactions to the bismuth substrate.

3.2 Homochiral versus heterochiral

Because both enantiomers are created from prochiral molecules, a competition of homochiral and heterochiral lateral interactions arises in the monolayer. Either a racemic crystal or a 2D conglomerate is the outcome. The example shown in Fig. 15a can be considered as a conglomerate of a monolayered crystal due to coexisting homochiral domains of both enantiomers. A similar observation was made early with the atomic force microscopy (AFM) 94 and many times thereafter.

Hydrogen-bonded linear chiral structures have been reported for adenine on Cu(110) 95, whereby different intermolecular hydrogen bonding scenarios between homo- or heterochiral pairs were discussed and calculated 96. The energy differences for most of these pairs are minute, but the most stable pair (Fig. 16a) agrees best with the homochiral hydrogen-bridged chain observed in STM 95. Besides hydrogen bonding, the linear chain motif matches well with the substrate lattice. Depending on the handedness of its building blocks, the chain is either aligned parallel along the [1,2] or the equation image direction on the surface. When adenine is coadsorbed with perylene-3,4,9,10-tetracarboxylic dianhydride (PTCDA) on Au(111) it forms a mixed structure with 10 PTCDA and four adenine molecules in the unit cell. All adenine adsorbates are homochiral in a given mirror domain 97.

Figure 16.

Chiral chains and clusters. Linear homochiral structures are observed for adenine on Cu(110) (a) 95, and PVBA on Ag(111) (b) 98. Nitronaphthalene on Au(111) forms enantiomorphous R- and L-decamers of unbalanced heterochiral composition (c,d) 99.

Another example of homochiral self-assembly is pyridylvinyl benzoic acid (PVBA) on Ag(111) 98. Due to specific hydrogen bonding linear chains are formed (Fig. 16b). In addition, hydrogen bonds between pyridyl hydrogen and the carbonyl oxygen induce assembly into double chains. Formation of these hydrogen bonds depends on the handedness and leads to homochiral chains. The chains run along the 〈112〉 directions of the substrate, and at higher coverage parallel running double chains form a grating. Quite remarkably, no opposite-handed chains have been observed within these gratings; i.e., chiral correlation is extended over large distances without apparent direct molecular contact.

A special example of heterochiral recognition is shown in Fig. 16c and d. Nitronaphthalene self-assembles on Au(111) into decamers with eight distal enantiomers surrounding a homochiral pair 99. The clusters nucleate first at the fcc elbows of the herringbone structure, with an equal distribution of left- and right-handed clusters over the surface. The relatively strong intermolecular hydrogen bonds in a decamer allowed lateral manipulation of clusters, i.e., pulling an entire cluster with the STM tip from one spot on the surface to another 48.

Depending on the number of carbon atoms in the side chains, thia-alky anthracenes self-assemble on hopg either into conglomerate or racemate 100. Homochiral chains with identical or alternating handedness are observed for dodecyl or undecyl groups at the sulfur, respectively. When the terminal bonds of the alkyl chain are oriented parallel to the aryl–C1 bonds the highest packing density is achieved (Fig. 17). Consequently, interdigitation of alkyl chains with an even number of bonds aligns the anthracenes of adjacent rows with identical enantiotopic faces toward the surface. Adding one methylene group to the alkyl chain, in turn, switches the assembly from heterochiral (Fig. 17a) to homochiral (Fig. 17b).

Figure 17.

Depending on the interdigitated chain lengths, heterochiral (a) and homochiral (b) domains are observed for bis-1,6-ethylenethiaundecylanthracene and bis-1,6-ethylenethiadodecylanthracene, respectively 100.

Only a few cases have been reported in the literature so far, where homochiral and heterochiral structures of the same molecule coexist on the surface. This implies that racemate versus conglomerate formations in 2D crystals, as in the 3D case, have sufficiently different heat of formations. Adenine on hopg, deposited from solution, forms different phases 101. As for Cu(110), the building blocks are homochiral dimers, but these combine in heterochiral and homochiral domain structures. Neglecting the substrate, the difference in energy was calculated via DFT to be 0.4 eV in favor of the heterochiral structure. If indeed the molecule–substrate interactions are overall identical for both scenarios, the homochiral domains must have been trapped in a metastable phase.

In contrast to this adenine example with a completely covered hopg surface stands the adsorption of dicarboxylstilbene on Cu(110) 102. Islands of racemic and homochiral composition are observed, and between these ordered structures the molecules are mobile. This suggests a delicate balance between conglomerate and racemate, but increasing the temperature favors the racemic structure. It is not obvious, what kind of entropic effects would favor the racemate. There is no difference in entropy between heterochiral and homochiral domains, which was made clear for 3D crystals 103. Perfect, i.e., defect-free, homochiral, and racemic crystals have the same entropy, which is, according to the third fundamental law of thermodynamics, zero at 0 K. Moreover, both phases are in contact with the same 2D gas of both enantiomers. Mixing entropy can only favor the racemate a high mobility when the molecules can swap their sites. Besides mixing, a phase could be favored by entropy, if vibrations become more easily excited at higher temperatures than in the other phase 80.

Sometimes self-assemblies of chiral adsorbates choose a homochiral arrangement, but switch with increasing coverage to a racemic state. The first observation of such a phase transition was reported for nitronaphtalene on Au(111). Instead of the decamers at low coverage (Fig. 16c and d), homochiral double chains are observed in an intermediate coverage regime. At high coverage a close-packed racemic 2D crystal is formed. A similar situation was observed for PVBA on Cu(100) 104. At low coverage, homochiral domains with CW and CCW motifs of four molecules were reported. At a coverage exceeding the critical value of one molecule per 20 Cu atoms, a butterfly structure comprising both enantiomers is formed.

An interesting aspect of a chiral phase transition has been identified in the strength of the binding of alkyl chains to the metal surface. On Cu(110) N,N′-dihexadecylquinacridone only crystallizes into homochiral domains 105 in which the alkyl chains are interdigitated at a saturation coverage of 0.03 molecules/surface atom. The same motif is observed on Au(111), there corresponding to a coverage of 0.023 molecules/surface atom 106. However, the coverage can be increased due to a partial lift-off of the alkyl chains from the surface. With the loss of interdigitation, homochiral recognition no longer dominates, and at the saturation coverage of 0.062 molecules/surface atom the layer is racemic with a glide plane symmetry perpendicular to the surface. Consequently, the stronger vdW interaction of the alkyl chains with the Cu(110) surface retains C2 point symmetry and thus homochiral recognition via chain interdigitation. Lift-off of the chains on gold, on the other hand, allows a much denser structure, but as a racemate crystal.

A temperature-dependent study revealed for 2,6-dibromoanthrachinone (Fig. 7c) on Au(111) a heterochiral phase at low temperatures. Annealing at elevated temperatures caused conglomerate formation, identifying the racemic domains as metastable structures 49. And finally, a chiral phase transition with opposite outcome has been reported for Ni-tetramethyl-tetraazaannulene on Au(111) 107. At low coverage a racemic mixture was observed, but steric interactions favored homochiral domains at saturation coverage. We will return to the issue of conglomerate versus racemate 2D crystallization for intrinsically chiral molecules in the next section.

4 Self-assembly of intrinsically chiral molecules on achiral surfaces

4.1 Homochiral adsorbates

Somorjai and coworkers 54 studied in 1978 with low energy electron diffraction (LEED) the adsorption of (S)-alanine on Cu(111). They reported an enantiomorphous equation imageR13.6° structure (which is better specified as (8 2, −2 6)1 structure in matrix notation 108, 109). This structure breaks the mirror symmetry of the underlying substrate, but mirror image related patterns were not observed. Hence, the handedness of the single enantiomer was somehow transferred into the entire layer. The same paper reports opposite mirror domains for (R)- and (S)-tryptophan on Cu(111) and Cu(100). This early work therefore already exemplified two important manifestations of chiral surface science: (i) the adsorbate lattice of chiral molecules often shows an oblique tilt with respect to the substrate lattice, and: (ii) other than for adsorption-induced chirality, chiral molecules, when adsorbed as pure enantiomers, form only one enantiomorphous domain, and formation of the respective mirror domain requires the other enantiomer to be adsorbed.

It has been emphasized above that tartaric acid was tremendously important for the birth and development of stereochemistry. Interestingly, this compound also played a significant role in chiral surface science. Motivated by the fact that tartaric acid is a modifier in enantioselective heterogeneous catalysis, the (R,R) enantiomer has been studied first on Cu(110) 110. As shown in Fig. 5, this molecule can bind with both carboxylate groups to the surface. In this conformation the (R,R) enantiomers form a (1 2, −8 2) structure and the (S,S) enantiomers assemble into the (1–2, 8 2) enantiomorph (Fig. 18) 111. The enantiomorphism of the lattice is directly connected to the molecular handedness, although the microscopic mechanism is subtle and not completely understood. Other than in normal X-ray diffraction of chiral 3D crystals, LEED patterns of surface enantiomorphs reveal directly the handedness of the lattice (Fig. 18a and b). As mentioned above, Friedel's law does not allow us to distinguish which scatterer is located in the front or in the back in a normal 3D diffraction pattern. To a first approximation, however, an LEED pattern is only concerned with diffraction from the X–Y plane without any “front” or “back”. Diffraction patterns from enantiomorphous 2D systems are still centrosymmetric as their counterparts in 3D, but at a lower dimensionality, which allows identification of the sense of tilt of the enantiomorph with respect to the laboratory frame of the observer.

Figure 18.

(online color at: Diffraction patterns and models of enantiopure 2D crystals of tartaric acid on Cu(110) (a–d). In the models, each molecule is represented as a filled red ellipse, with two legs or four legs, corresponding to one or two carboxylate groups interacting with the surface. (a),(b) The LEED patterns of the 2D lattices of tartrate enantiomers reveal the opposite handedness of both enantiomorphs. (c),(d) Real space (1 2, −8 2) models as suggested by STM 111. The semitransparent yellow lines mark empty channels between triplet rows. (e) Sketch of a singly deprotonated tartaric acid on Cu(110). (f) The “achiral” (2 1, −2 1) lattice built up by homochiral molecules 110. (g) Long-range enantiomorphism at a slightly higher coverage 110.

The models for the two enantiomorphous (1 2, −8 2) structures2 (Fig. 18c and d) are based on STM investigations 111. A local (1 2, −3 1) periodicity is maintained only over three molecular rows, interrupted by an empty channel (yellow bands in Fig. 18c and d). This is a strong indication of surface stress induced by the adsorbed molecule 111.

At room temperature or at high coverage after thermal activation, tartaric acid is only singly deprotonated on Cu(110). This is chemically similar to a bitartrate salt crystal like (HOOC[BOND]CHOH[BOND]CHOH[BOND]COO)2Ca. Apparently, the denser packing at high coverage requires tartrate to be in an upstanding, singly deprotonated form (Fig. 18e). At a coverage of θ = 0.25, a c(2 × 4) or (2 1, −2 1) structure is observed, which does not express any lattice handedness (Fig. 18f). Although all building blocks have the same handedness, the extended lattice itself is achiral. Thus, the substrate registry dictates, and no long-range chirality is expressed. At higher coverage, long-range handedness is restored, and a mirror-symmetry-breaking (4–1, 2 4) phase forms (Fig. 18g) 110. That is, the extreme close packing introduces lattice handedness again.

As for the macroscopic shape of crystals, the sense of handedness expressed in the 2D supramolecular ensemble is difficult to predict. The chiral rosettes built by oligo-para-phenylene derivatives are a good example (Fig. 19) 112. Adsorbed from solution onto hopg, six molecules form a handed motif of the plane group p6. Although all stereogenic centers in the side chains have the same handedness, OPV3T arranges exclusively into CW rosettes, while OPV4T forms CCW rosettes. The expressed handedness is the result of intermolecular hydrogen bonding between the diaminotriazine ends, vdW interactions between the alkyl chains and molecule–substrate interactions. The latter gets contributions from the stereogenic centers and the conjugated backbone, and this balance shifts for OPV4T due to the larger number of stereogenic centers with respect to the somewhat longer backbone.

Figure 19.

(online color at: Oligophenylene derivatives with different backbone lengths (a) form opposite-handed rosettes on hopg 112. (b) STM image (14.4 nm × 14.4 nm) of OPV3T. (C) STM image of a single OPV3T CW-rosette. (d) STM image (18.4 nm × 18.4 nm) of OPV4T. (e) STM image of a single OPV4T CCW rosette.

4.2 Racemic mixtures

We have seen that prochiral molecules always create a racemic mixture at the surface. Besides adsorption of single enantiomers, intrinsically chiral molecules are often adsorbed as a racemic mixture. When a racemic mixture condenses in 3D it may form a racemic compound, in which both enantiomers are present in the same single crystal; a conglomerate, in which the single crystals contain only one enantiomer, but the sample as a whole is racemic; and a solid solution, in which the condensate contains the two enantiomers in a nonordered arrangement 113. Racemic crystals are by far more common as conglomerate formers, and only a few compounds tend to form solid solutions 113, 114, also called a pseudoracemate 115. Based on only eight examples, Wallach 116 came to the conclusion that racemate crystals tend to pack denser, which would explain their abundance.

After the first observation of conglomerate formation by intrinsic chiral molecules in a monolayered crystal with the AFM 117, Walba and coworkers 118 reported the same effect for liquid crystal molecules on hopg with the STM. They confirmed their findings by comparison to monolayer structures formed by pure enantiomers. Concerning UHV experiments, an interesting case is again tartaric acid on Cu(110). The adsorption of the racemate reveals in LEED a superposition of (1 2, −8 2) and (1–2, 8 2) patterns (Fig. 18a and b) 119. This is a strong indication of conglomerate formation, whereby the electron beam probes both homochiral domains. This was first confirmed via the autocatalytic thermal decomposition (“surface explosion”) chemistry 120, and later by STM 121. However, conglomerate formation applies only to the (1 2, −8 2) phase with both carboxylate groups bound to the surface. The racemate forms at higher coverage, just like for the pure enantiomers, a c(2 × 4) structure. This adlattice does not break the symmetry of the underlying substrate, so “mirror domains” would coincide in LEED. However, a profound difference in thermal stability with respect to the enantiopure c(2 × 4) structure suggests a racemic structure 120. Further support for this scenario comes from the fact that the saturation coverage for the pure enantiomers is higher. A formation of the enantiomorphous (4–1, 2 4) phase is not observed for the racemic mixture, which saturates in a c(2 × 4) structure 119.

Other than tartaric acid, racemic heptahelicene ([7]H; C30H18) forms completely different structures from the pure enantiomers on Cu(111) 122–125. This aromatic hydrocarbon is basically a construct of seven ortho-annulated C6 rings and therefore an example of a chiral molecule without a stereogenic center (Fig. 20a). Steric overcrowding leads to left-handed (M-) and right-handed (P-) helical molecules. The pure enantiomers form single enantiomorphous domains. Two examples for the P-enantiomer are shown in Fig. 20b and c. The M-enantiomer shows the corresponding mirror structures, i.e., with opposite handedness (not shown) 122. XPD revealed that three rings on the proximal end are aligned parallel to the surface plane 35, and force-field calculations for the saturation structure provided a model in good agreement with the observations (Fig. 20c and d). The racemate forms coexisting mirror domains (denoted as λ and ρ), but these have different structure with respect to the pure enantiomers (Fig. 20e and f). Calculations suggest that these domains are composed of heterochiral pairs, whereby two possible relative alignments of the enantiomers determine the sense of enantiomorphism 124. This alignment turns the heterochiral MP pair into a chiral entity. However, heterochirality is only present if we just consider the covalently bound molecules as entities. Extending this bonding picture to dispersive forces at the supramolecular level, a single racemic [7]H domain must be considered as homochiral system.

Figure 20.

(online color at: Heptahelicene ([7]H) on Cu(111) 122–125. (a) Space-filling models for M- and P-[7]H. (b,c) STM images of structures formed by P-[7]H. (d) Model for the structure imaged in (c), based on forcefield calculations. (e,f) STM images taken from the racemic mixture, superimposed in part with molecular models of the particular molecules. The semitransparent models in the insets show the lowest-energy results of calculations. The sense of helicity is indicated with the letters M and P. Two enantiomorphs (λ and ρ), composed of both enantiomers, are observed.

A similar picture arises for derivatives of hexaphenylbenzene on Au(111) 126. Steric overcrowding gives this molecule a propeller shape and therefore intrinsic chirality. Mirror domains of honeycomb networks are observed, which are heterochiral at the covalently bound molecule level. Intermolecular hydrogen bonding imparts supramolecular handedness to homochiral trimers, hexamers and decamers, i.e., the molecules in them are CW or CCW aligned. All trimers, hexamers, and decamers in their single enantiomorphous domain have identical supramolecular handedness, although the domain is racemic, i.e., every second trimer, hexamer, or decamer is composed of the other enantiomer.

Only if the coverage is sufficiently high, does [7]H crystallize into an ordered arrangement. Under these conditions, i.e., when short-range repulsive forces are significant, discrimination between homo- and heterochiral ordering has been predicted to be more likely 127. The higher the coverage, the more adsorption energy is gained. Therefore, the closest packing should decide over conglomerate versus racemate crystallization at surfaces. The enantiopure phase shown in Fig. 20d has a higher density than the heterochiral phase observed. Molecular modeling calculations favor the enantiopure phase by 1 kJ mol−1 over the racemate λ/ρ phase 125. However, deposition at room temperature leads to a disordered monolayer due to high mobility. This imposes an entropic penalty of RT ln2, i.e., 1.73 kJ mol−1, not allowing the maximum coverage observed for pure enantiomers to be reached. Moreover, crystallization upon cooling at such coverage, puts a limit on lateral mobility, suppressing phase separation into a conglomerate. Directional forces, like hydrogen bonding have been brought forward for chiral discrimination in Langmuir films at the liquid/air interface 128. [7]helicene is rather apolar, which supports this conclusion. Furthermore, the polar cyano-[7]H on Cu(111) was found to form a conglomerate 129. However, heptahelicene-2-carboxylic acid on calcite(104) forms heterochiral double chains, as concluded by comparison of self-assembly of racemate and pure enantiomers 130, 131.

If enantiopure nuclei are formed in conglomerate crystallization, the wrong enantiomer arriving at the nucleus must be expelled in order to grow the homochiral crystal. This puts a kinetic penalty on conglomerate formation 103. For 2D crystals this effect has been observed by STM for a chiral resorcinol diacid derivative 132. In strong contrast of the pure enantiomer layers with good order, the racemate showed homochiral nuclei of tens of molecules, which were decorated by the other enantiomers, inhibiting conglomerate formation.

This brings us to the question if homochiral recognition prevails over heterochiral interactions in dense crystal lattices or at a lower level, i.e., in dimers or small clusters. For [7]H on Cu(111), calculations favored heterochiral dimers and heterochiral small clusters 125. A similar apolar intrinsically chiral molecule, the aromatic 5,6,11,12-tetraphenyltetracene (rubrene, Fig. 21), however, shows a remarkable hierarchical homochiral recognition 133. The molecule is another example of a chiral structure due to steric overcrowding and the absence of a stereogenic center. The tetracene backbone of rubrene is helically twisted, and its handedness can be directly observed in highly resolved STM images. Spontaneously self-assembled on gold(111), rubrene forms homochiral chains and circular pentagonal clusters of homochiral molecules. Ten pentamers, in turn, assemble into a homochiral supramolecular ring unit of 50 monomers. Within the pentamers, the monomers are equally rotated for close interdigitation in a gear-wheel-like fashion. CCW or CW rotation is representative of the handedness of the monomers. The same applies to the ten pentamers that all are either rotated CW or CCW (Fig. 21). The handedness of the monomer is therefore transferred step-by-step into the pentacontamer.

Figure 21.

(online color at: Transfer of handedness from the single molecule into pentamers and pentacontamers 133. The chiral molecule rubrene (a) appears as a handed three-lobe feature in STM (b). Ten homochiral rings consisting of five molecules assemble into a larger ring.

Polar interactions and surface registry govern the intermolecular interaction in cysteine dimers on Au(110). At low coverage, homochiral pairs are identified in STM and show, depending on molecular handedness, opposite tilt angles with respect to the substrate lattice 134. In combination with specific surface sites occupied by the sulfur atom and the amino group, such homochiral recognition is an example of the classical three-point contact model 135, 136, which postulates that molecular chiral recognition requires interaction at three different locations. DFT calculations suggested further that Au surface atoms are removed from the densely packed row below every cysteine pair. Prolonged annealing at 380 K induces unidirectional growth of cysteine double rows 137, a process that goes hand-in-hand with the deconstruction of the Au surface below the cysteine double rows.

We have seen that molecular recognition at surfaces reflects the interplay between the molecule–substrate and intermolecular interactions. Hence, an isolated flexible adsorbate is expected to be in a different conformation than in a molecular ensemble. This has been beautifully demonstrated for the dipeptide (R,R)- and (S,S)-diphenylalanine on Cu(100) 138. In STM two bright protrusions, associated with the phenyl rings, were observed (Fig. 22a). For the single adsorbate enantiomorphous alignment with respect to the [1equation image0] direction was identified, whereby the (R,R)-enantiomer was rotated CCW by 34° and the (S,S)-enantiomer was rotated CW by 34° (Fig. 22a). In addition, self-assembly into homochiral chains occurred, but with the consequence that the molecular axes in these assemblies are rotated by another 40° (Fig. 22b). These results were obtained from the racemic mixture, but have been confirmed with enantiopure samples. The change of alignment with respect to the surface points to a dynamical rearrangement process due to lateral interactions between the homochiral building blocks. Calculations showed that the molecule binds to the surface in a chiral footprint manner via the two C[DOUBLE BOND]O groups of the peptide group and the carboxyl group as well as the amino group (Fig. 22c). For a given enantiomer, the amino group and the peptide C[DOUBLE BOND]O group in the middle point to the opposite sides with respect to the long molecular axis (Fig. 22c). Upon homochiral aggregation, however, these two groups point to identical sides. This is due to a strong intermolecular interaction of the amino group and the carboxyl group of adjacent molecules and includes a proton transfer from one to the other. Such combination of surface alignment of a peptide on the one hand – mimicking a large protein backbone – and the dynamic rearrangement adapting to an external influence on the other hand, can be considered as a proper model for stereochemical recognition in enzymatic interactions.

Figure 22.

(online color at: The enantiomers of the dipeptide diphenylalanine align on Cu(110) at opposite angles with respect to highly symmetric substrate directions (a) 138. Supramolecular self-assembled chains are homochiral in which single molecules are forced into a different alignment to the substrate (b). DFT calculations reveal the details of different interactions. While the bonding configuration of a single molecule has the peptide carbonyl and the amino group pointing into opposite directions with respect to the long molecular axis (c), the supramolecular chains induce a rearrangement due to strong electrostatic interactions between a carboxylate and an ammonium group (d). The orientation of the long molecular axis switched from ±34° for both enantiomers to ±74° for the homochiral chains.

4.3 Diastereomers and diastereomeric recognition

When intrinsically chiral molecules can occupy opposite-handed footprints (“footedness”), as discussed above for achiral adsorbates (Fig. 9), adsorption-induced diastereoisomerism arises. That is, stereoisomers are formed that are not interrelated by reflection like enantiomers. Figure 23 shows this for alanine. After deprotonation of the carboxyl group a tridentate bonding conformation is occupied, i.e., both oxygen atoms and the nitrogen atom interact with the surface. Two ways of chiral footprints of both enantiomers cause formation of two enantiomeric pairs. One of the four diastereomers is the enantiomer of one of the others. Such diastereoisomerism has been discussed controversially for the (2 × 3) structure of alanine on Cu(110) 139–144. As observed for the glycine (2 × 3) structure, calculations for the pure enantiomers favor a structure including opposite footprints, as if the methyl group would have no influence. For the racemate the two enantiomers coexist in a single domain and remain true enantiomers by occupying the opposite footprint 141–143. Basically, the same conclusion was made for racemic alanine on Cu(100) 142.

Figure 23.

(online color at: Top view onto alaninate diastereomers at a surface. Only (R)/(S) isomers with opposite footedness are enantiomers.

Both footprints are also formed when proline is adsorbed on Cu(110) 145. This amino acid has a pentagonal pyrrolidine ring including the α-carbon atom and the amino group. For the pure enantiomer both diastereomers are found in a (4 × 2) unit cell, alternating along the [1equation image0] direction. Adsorption of the racemic mixture creates all four diastereomers in a regular (4 × 2) lattice, but with a random distribution of the (R)- and (S)-isomers 146. However, the footedness follows the same regular arrangement as observed for the enantiopure lattice. This is in contrast to an earlier investigation for a spiro compound on Au(111). Within homochiral chains, a random distribution of two diastereomers was reported 147.

There are not many studies in which the adsorption of intrinsic diastereomers are compared. One example is (R,S)-tartaric acid on Cu(110) 60, 61. This species differs from the (R,R)- and (S,S) enantiomers only by the relative position of one hydroxyl group at carbon atom 2 or 3. In spite of two stereogenic centers the free molecule has a mirror plane and is therefore achiral. The minute difference in the position of one hydroxyl group, however, leads to a different long-range order with respect to the chiral diastereomers. Instead of the (1 2, −8 2) phase, as observed for chiral tartaric acid (Fig. 18), (R,S)-tartaric acid forms a (1 2, −6 2) phase at roughly the same coverage. Presumably, the doubly deprotonated species forms a chiral adsorbate (Fig. 9) that also bestows strain into the surface. Rather than three molecular rows being separated by an empty channel (Fig. 18), only a double-row model fits the observed lattice periodicity of (R,S)-tartrate. In contrast to the c(4 × 2) observed for chiral tartrate at a coverage of 1/4 molecule per Cu surface atom, the singly deprotonated (R,S)-molecule forms an equally dense c(8 × 2) phase.

When different chiral species interact, they also form diastereomeric pairs. This applies in particular when chiral molecules adsorb onto chiral surfaces, which will be treated in Section 6. Intermolecular diastereoisomerism is in particular present in heterogeneous enantioselective catalysis, where the chiral modifier selectively interacts with the prochiral reactant. Tartaric acid, the modifier for enantioselective hydrogenation of β-ketoesters, has been studied in coadsorption with the prochiral methylacetoacetate (MAA) on Ni(111) 148. At intermediate coverage (R,R)-tartrate indeed steers MAA into a homochiral alignment, so that hydrogenation directly from the surface would only create the (R)-hydroxy enantiomer. However, whether long-range structures support enantioselectivity in heterogeneous catalysis is still debated 149.

As mentioned above, adenine forms homochiral chains on Cu(110) aligned parallel along the (1,2) and the (−1,2) directions on the surface (Fig. 16). Subsequent coadsorption of phenylglycine decorates both sides of the adenine rows with double rows of phenylglycine molecules. Coadsorption of (S)-phenylglycine leads to enantiospecific decoration of adenine [1,2] rows, whilst (R)-phenylglycine decorates only [equation image,2] rows 150. This remarkable enantiospecific interaction is based on substrate-mediated charge transfer and Coulomb repulsion between the amino groups of both species 151.

5 Symmetry breaking III: Single handedness by chiral bias

We have learned that achiral molecules can crystallize into chiral space groups, but both enantiomorphs will appear with equal probability. Under certain conditions, however, crystallization experiments may show spontaneous symmetry breaking and only one enantiomorph will be found in the precipitate 152–155. Without chiral bias both enantiomorphs will be observed at equal frequency when the experiment is repeated several times. With chiral bias, on the other hand, some of these processes will have a defined enantiospecific outcome. This has been shown by Green et al. 156, 157 for helical polyisocyanate polymers with short aliphatic side chains. These show an equal probability for left- and right-handed helicity, giving them overall a zero optical activity. A small number of copolymerized monomers with homochiral side chains led to single helicity in all macromolecules, as judged by large circular dichroism. The small chiral bias and the tendency to avoid energetically unfavorable helix reversals were sufficient to cause this homochirality in the entire polymer. This kind of chiral amplification was coined the “sergeants-and-soldiers” principle, because a few units (chiral side chain sergeants) rule the global outcome (single helicity) by cooperatively influencing the sense of helicity of the turns (soldiers) in the entire chain 158. A similar observation was made for small enantiomeric excess in the side chains and named “majority rule”.

The first 2D sergeant-and-soldiers effect was reported after doping succinic acid monolayers with chiral tartaric acid 159. Due to the footprint chirality, succinic acid forms enantiomorphous domains on Cu(110), observed as superposition of both enantiomorphs in LEED (Fig. 24a) 59. Mixing 2% of a tartaric acid enantiomer into the monolayer completely suppresses formation of one enantiomorph. A reasonable scenario for the mechanism here is the chiral recognition through the substrate. It has been shown that the handedness of the tartaric acid is transferred via intermolecular hydrogen bonds into a chiral footprint 61. This gives the bias to footprints of adjacent succinate, which, in turn, force their neighbors into the same footedness via this cooperative mechanism (Fig. 24b). Because hydrogen bonds between the succinate adsorbates with both carboxylate at the surface can be excluded, one must consider a substrate-mediated mechanism. The chiral footprint onto the surface acts as a chiral bias and suppresses opposite footedness in the adjacent adsorbate complex. In order to allow switching of the succinate backbone, temperatures above 450 K are required. Subsequent cooling with the chiral bias of coadsorbed tartrate present then induces the single handedness throughout the entire monolayer. Achiral (R,S)-tartaric acid acts in the same way when doped with either tartaric acid enantiomer on Cu(110) (Fig. 24c) 160.

Figure 24.

(online color at: The sergeant-and-soldiers principle for 2D lattices on Cu(110) 159, 160. (a) LEED patterns for pure succinic acid (SU) and doped with 2% of tartaric acid enantiomers (TA). (b) Sketch for the mechanism: Zigzag-distorted SU molecules leave a chiral footprint onto the surface. Small amounts of a coadsorbed TA (highlighted in yellow) allow only formation of a single enantiomorph with all SU molecules presumably now occupying the same footedness. The handedness of the TA sergeant is transferred into a defined footprint, which determines the footedness of the SU soldiers in the layer. (c) LEED patterns for (R,S)-tartaric acid (m-TA) and doped with chiral TA, demonstrating the same effect.

Although a heterochiral pair constitutes the building block, [7]H forms enantiomorphous domains on Cu(111) (Fig. 20). Due to the noncovalently bound enantiomers in each pair, the handedness of such building block can be easily switched. A small chiral bias in the form of enantiomeric excess (ee) was found to be sufficient to drive the complete layer into single enantiomorphism 124. Excess of the M-enantiomer favored the λ domain, whilst excess of the P-enantiomer favored the ρ enantiomorph (Fig. 25a and b). Molecular modeling calculations showed that any ee cannot exist within a domain, but rather acts from the domain boundary. Moreover, the calculations revealed that mirror domain boundaries create a higher energy situation, explaining why they were not observed on flat terraces, but rather coincide with single atomic steps of the Cu(111) substrate. Finally, the molecular mechanics calculations of pure enantiomers at the domain boundary produced a detailed insight of the mechanism. Figure 25c shows five M-enantiomers located between a monoatomic substrate step (vertical dashed line) and the boundary of a λ domain. Reducing successively the number of available adsorption sites for these five molecules models the interactions at this 1D interface under close-packing conditions (Fig. 25d). In each step, the lowest-energy sites and orientations of the excess molecules within the available area were determined. The average of the domain boundaries to the left and the right was used, because both sides are different. The interface energies at the λ domain are lower for M-enantiomers for all densities (Fig. 25d). The same scenario was found when one, two, three, or four molecules were considered or if the excess coverage was increased in a fixed area at the boundary. That the P-enantiomer becomes dramatically repulsive at the λ domain must be interpreted such that during 2D crystallization the λ domain will not form at excess of P (ee < 8%). Because of the easy realignment of one enantiomorphous pair into the other, this mechanism requires only small mass transport. For the same reason, a wrong nuclei on a terrace will be rearranged easily. Having only the excess enantiomer at the domain boundary this mechanism would be another 2D analog of the sergeant-and-soldiers principle. However, highly resolved STM images show, that the excess region has both enantiomers at the domain boundary. Therefore, the minority enantiomers are in part forced into a less-favored situation by the majority. Hence, this scenario fits rather the “majority rule” scenario reported for helical polymers 157.

Figure 25.

(online color at: Enantiomeric excess at the domain boundary rules the sense of domain handedness. The STM image (200 nm × 200 nm) shows that only single enantiomorphs exist on flat terraces, while mirror domain boundaries coincide with step edges. At small positive ee the number of observed ρ domains is smaller than the number of λ domains (a). The STM image (200 nm × 200 nm) taken at ee of 8% (M:P = 54:46) shows that exclusively λ domains are formed (b). Molecular modeling calculations reveal that M-enantiomers can be packed more densely at the λ domain boundary, explaining the formation of λ domains at positive ee (c,d) 124.

The [7]H amplification mechanism requires a racemate crystal as ordered phase with a low barrier for switching the heterochiral pair alignment. In contrast, racemic tartrate on Cu(110) forms a conglomerate, i.e., coexisting homochiral (1 2, −8 2) and (1–2, 8 2) mirror domains at a coverage of θ = 0.167 (Fig. 18). At an ee of 20% the enantiomorph of the minority can no longer be observed via LEED because only the majority forms its homochiral ordered phase 121. The minority goes into a mix with other parts of the majority with the molecules mainly oriented in rows along the [1equation image0] direction. This semiordered phase can actually be observed at a smaller extent for the racemate (ee = 0%), coexisting with the enantiopure conglomerate. Therefore, a slight change in composition has drastic consequences. This effect might be based on the balance between intramolecular hydrogen bonding (homochiral) versus intramolecular hydrogen bonding (heterochiral).

Similar observations have been made when one enantiomer of malic acid is mixed with racemic tartrate (Fig. 26). That is, instead of excess of one tartrate enantiomer the chiral conflict is induced by an enantiomer of a different species 161. (R)-malic acid forms a quasiracemic mixture with (S,S)-tartrate and leaves (R,R)-tartaric acid unaffected. The molecules in the mixture are – as observed for tartrate – aligned along the [1equation image0] direction, often in the form of molecular triplets. DFT calculations performed for triplets confirm that the lowest energy configuration is a combination of (S,S)-tartrate with two (R)-malates 165.

Figure 26.

(online color at: Malic acid (a) has one hydroxyl group less than tartaric acid (b). (c) STM image of a mixture of 33% (R)-malic acid with 66% racemic tartaric acid. Only the (1 2, −8 2) enantiomorph of (R,R)-tartaric acid (orange area) remains unaffected. The small green area shows a (1–2, 8 2) (S,S)-tartrate island and the blue area are examples where molecules are aligned parallel to the [1equation image0] direction. (d) Lowest-energy configuration obtained from DFT for a triplet of an (S,S)-tartrate and two (R)-malate 165.

In the previous section we have seen that molecular chirality and supramolecular handedness do not always go hand-in-hand. This applies also to the sergeant-and-soldiers effect, and chiral doping to achieve a homochiral assemble from achiral units does not generally require enantiopure dopants, but may also require structural adjustment. This has been demonstrated for achiral alkyl-chain-interdigitated structures that assemble into CW and CCW motifs. Mere doping with a similar chiral derivative was not as efficient as tuning chain length and handedness of side chains 162. This example shows again that the absolute positions of atoms in space rule the effect and not the conventional assignment of handedness via CIP or other rules.

We return to the system of racemic cysteine adsorption on Au(110)-(1 × 2) in order to address a curious effect of chiral amplification. After deposition of the racemic mixture and annealing an excess of homochiral L,L-pairs were observed. A number of carefully designed control experiments, addressing crystal miscut or sublimation artifacts, have been performed without any conclusive results pointing to the origin of the mechanism 163.

Deposition from the liquid phase provides the possibility to include solvent effects into adsorption. Using chiral solvents, excess of CW or CCW pinwheels in monolayers of achiral oligophenylene derivatives, similar to those shown in Fig. 19, was reported 164. Using these achiral oligophenylenes and dibenzoyl tartaric acid as chiral auxiliaries in the achiral solvent, the same effect was observed 165. Another possibility is using magnetic fields for prealigning liquid-crystal molecules prior to adsorption 166, 167. Due to the cooperative interactions in the liquid-crystal phase, the weak magnetic forces induce polar order. By tuning the B-field such that the prochiral molecules are oriented in an oblique angle with respect to a highly symmetric substrate lattice during adsorption, chiral symmetry breaking is induced, because adsorption into a one-mirror domain is preferred over the other. Finally, it is noteworthy that chiral amplification by the formation of homochiral oligopeptides has been observed, starting from racemic 2D crystalline self-assembled monolayer of amino acids on water 168.

6 Chiral molecules at chiral surfaces

In 1866 Pasteur's former student Gernez 169 discovered that seeding a racemic solution of ammonium sodium tartrate with crystals of either pure enantiomer yielded only crystals of the same handedness as the seed. Werner 170 in 1914 then applied this method to chiral metal complexes. In a modification of the method, preferential crystallization due to the presence of enantiomorphous β-glycine or amino acid crystals was reported in 1908 171, and in 1919 enantioselective inclusion of China Blue dye in NaClO3 was observed via anomalous ORD 172. Today, this type of preferential crystallization is one of the most important means of chiral resolution 19. All these experiments strongly suggest an enantioselective interaction of chiral molecules with chiral crystal surfaces 173. For example, β-quartz was used to induce mirror-symmetry breaking in chemistry and adsorption. In 1932, Schwab and Rudolph 174 reported enantioselective dehydrogenation oxidation of the racemic butane-2-ol over quartz powder, modified with catalytically active metals. However, due to the minute effects the enantioselective adsorption of chiral molecules on quartz has not been widely accepted until 1974 175. The symmetry-breaking influence of a chiral inorganic surface has been demonstrated via a chemical amplification mechanism. The sign of ee in an autocatalytic enantioselective reaction was determined by the handedness of quartz or NaClO3 crystals, which were in contact with the solution 176, 177.

These examples show that inorganic surfaces can have a handedness. Single crystals can be cut such that chiral surfaces are exposed 178. A typical example is the fcc(543) surface possessing chiral kinks (Fig. 27). The space-inverted equation image surface will express the opposite handedness. It is not required that the edges near the kink have a different length, because at a kink site, an fcc surface has always (111), (100), and (110) terraces/step edges connected in a CW or CCW sequence. Even flat terraces can possess chiral structures 178. It is obvious that chiral molecules should show enantiospecific recognition at such sites (Fig. 27). This was first tested in UHV by Gellman and coworkers 179, who found later enantiospecific differences in thermal desorption spectra 180. The first distinct enantioselective reaction, however, was found for electro-oxidation of glucose over Pt(543) 181. Also noteworthy is the enantioselective interaction of cysteine with kinks on gold surfaces 182–184.

Figure 27.

(online color at: The fcc(543) (left) and fccequation image surfaces possess chiral kink sites of opposite handedness. Imaginary alignments of alanine at the kinks are examples of diastereomeric and enantiomeric interactions.

Enantioselective adsorption has been also reported for aspartic acid of scalenohedral calcite. L-aspartic acid was observed to adsorb preferentially on the (3equation image1) face, whereas D-aspartic acid adsorbed preferentially on the (21equation image1) face 185. Systematic studies for understanding chiral recognition at crystal surfaces under ambient conditions were in particular performed during the last five decades at the Weizmann Institute. From the change of crystal shape due to selective adsorption, even the absolute configuration of molecules was deducted 186. The same group investigated the selective adsorption of amino acids from solution onto the different crystal faces of α-glycine crystals, which led to the remarkable consequence of resolution of enantiomers 14, 187. α-glycine packs into centrosymmetric crystals in which the achiral molecules are aligned in a chiral configuration, quite similar to the example shown in Fig. 9a. This makes α-glycine a layered crystal with alternating handedness (Fig. 28), and the (010) and (0equation image0) surfaces of α-glycine expose opposite handedness. When in contact with a racemic solution of α-amino acids with hydrophobic groups (lysine, norvaline, norleucine, etc.) D-enantiomers will adsorb at the (010) surface and L-enantiomers at the (0equation image0) surface, because the opposite scenario would lead to sterically unfavorable packing. Therefore, D-enantiomers will be enantioselectively occluded at the b+ end in a growing α-glycine crystal, while L-enantiomers only incorporate at the b− end. Floating glycine crystals are oriented with either side toward the liquid-air interface. Thus, growth is inhibited on one side, which influences the shape of the crystal. The crystal face pointing to the air becomes more pronounced, while the opposite face becomes smaller due to continuing growth at that end (Fig. 28b). In that case, only one enantiomer of the additive amino acid will be then occluded in this particular crystal. In addition, amino acid additives with a hydrophobic group at the α-carbon, will decorate the air interface and depending on the handedness of the amino acid, either the (010) or (0equation image0) will be turned toward this interface (Fig. 28c). Finally, small enantiomeric excess of the additive will preferentially align the nucleating glycine crystals, which, in turn, will increase the excess, because of occlusion of the opposite enantiomer in the growing uniformly aligned glycine crystals. For D-leucine, e.g., an ee of 6% (D:L = 53:47) was sufficient to achieve 100% L-leucine in the occluded state. Other than the examples discussed in the previous section, this kind of chiral amplification works due to feedback through the solution.

Figure 28.

(online color at: Enantioselective adsorption and occlusion of amino acids into oriented α-glycine crystals 187. (a) The centrosymmetric α-glycine preferentially adsorbs α-amino acid enantiomers on the (010) and the (0equation image0) surface. (b) In a growing crystal, enantiomers become preferentially occluded. If growth on one side is blocked, this crystal face will become more pronounced, while at the same time only one enantiomer is incorporated. (c,d) Amino acids with hydrophobic groups at the α-carbon atom at the liquid/air interface align enantioselectively floating glycine crystals that occlude only one amino acid during growth.

7 Chiral crystal modification and surface reconstruction

The striking feature in biomineralization is control of crystal shape and polymorphism 188. In particular, macromolecules with large numbers of carboxylate groups have been identified as effective modifiers 189. Proteins with large aspartic acid and glutamic acid content, e.g., have been extracted from calcite, aragonite, and hydroxyapatite biominerals 190–192. Because these acidic proteins are homochiral, they were even found to induce chiral shapes of unique handedness into otherwise achiral minerals. As an example stands calcium oxalate crystals extracted from tomato and tobacco leaves with pseudo-tetrahedral shape 193.

In defined crystallization experiments, small molecules that are expected to have chiral footprints, similar to those shown in Fig. 9a, were added during calcite crystal growth 194. For achiral ω-dicarboxylic acids HOOC[BOND](CH2)n[BOND]COOH (n = 1, 2, 3 … 6) the largest deviation from natural crystal shape was observed for small n, i.e., for malonic and succinic acid. Aspartic acid and γ-carboxyglutamic acid also had a large influence on morphology. Because longer charge separation diminished the effect and because the molecule–surface interactions are electrostatic, the extent of space charge has been proposed as a mechanism. Further insight into such crystal modification by additives was achieved by AFM studies, again performed with calcite and aspartic acid 195. The natural {104} faces of calcite show growth hillocks with a glide plane separating two types of acute and two types of obtuse steps on the four sides of the hillocks. After growth in the presence of pure enantiomers, AFM showed that one of the acute steps changed its shape. Calculations confirmed that aspartic acid enantiomers adsorb selectively on one of the acute steps and change the step-edge free energy, inducing a macroscopic handedness of the crystal. Acting as surfactant, aspartic acid was not incorporated into the crystal bulk. The macroscopic crystal shape expressed a handedness by the lengths and shapes of edges of the {104} faces. Similar results were reported after calcite electrodeposition in the presence of tartaric, maleic, and aspartic acid enantiomers 196. Electrodeposition of copper oxide under the influence of tartaric acid enantiomers was found to create homochiral CuO films 197–199. A strong indication that a chiral reconstruction of an inorganic surface occurred came from particle research. Gold nanoparticles, decorated with homochiral molecules, showed a strong circular-differential response of the metal's plasmon resonances 200.

It is known that the change in free energy due to adsorption of additives stabilizes new crystal faces during crystal growth and induces changes in shape. What applies to the so far discussed cases for adsorption from the solution, also applies to UHV. Surface faceting is often observed in STM after adsorption of molecules and sufficient annealing 201. Chiral faceting has been first observed by Zhao and coworkers 202, 203 with the STM. L-alanine as well as L-lysine form on Cu(100) four homochiral {3 1 17} facets. The four other {3 1 17} facets of opposite handedness are not observed. It can be assumed that such chiral faceting is the first step to install a macroscopic chiral shape to a crystal. Such faceting includes mass transport of substrate metal atoms and theoretical investigations on the system tartaric acid/Ni(110) suggest as a first step a chiral footprint reconstruction 204. Such a mechanism of chiral reconstructions of the surface underneath the molecular layer has been discussed before 59, but they are usually invisible by STM due to the presence of the molecular layer on top. In one case, however, pushing the molecule away with the STM tip revealed chiral pits 205. The first direct observation in STM of replaced metal adatoms was made for pure malic acid enantiomers on Cu(110) 206, 207. Careful annealing led in part to removal of molecules without relaxation of the metal surface back to the (1 × 1) structure. Rows of adatoms were then observed in STM between the molecular islands 206. Figure 29 shows a similar result for the racemate. Enantiomorphous domains show malate-decorated copper adatoms rows along the [1,1] and [equation image,1] directions. These structures were not observed for the pure enantiomers and suggest therefore a heterochiral composition. Height profiles along the molecular structure and the adatoms rows strongly support the scenario of molecule-decorated Cu adatom rows.

Figure 29.

(online color at: Chiral reconstruction of the Cu(110) surface due to racemic malic acid 207. (a) LEED and STM images show enantiomorphism in the (1–1, 9 5) phase. Between the bright molecular features, STM reveals Cu adatom rows in [1,1] and [equation image,1] directions (insets). (b) Heterochiral models for the observed enantiomorphs. (c) STM line scans over the bright molecular features (green trace) and the 〈1,1〉 rows (red trace) confirm the expected height differences as well as the periodicity of Cu atoms of 4.4 Å along the 〈1,1〉 directions.

8 Switching handedness

There are chiral compounds whose barrier for interconversion is low so that they switch between both states in liquid phase. This may lead to the situation that one pure crystal enantiomer as seed will induce a homochiral solid phase and turns the whole system to homochirality. Such deracemization has been first reported by Havinga 208, 209 in the early 1940s. For surface-confined systems, we have assumed so far that lateral resolution of enantiomers on surfaces requires mass transport. But what happens if a molecule can switch its handedness at a homochiral domain boundary? Such a case has been observed for prochiral bis-ethynyl-p-phenylene compounds on Au(111) (Fig. 30) 210. Besides the two enantiomers, an achiral meso-form exists at the surface. At both ends of the molecule, tert-butyl groups dominate the STM appearance. Upon thermal activation at least a part of the molecule is lifted off the surface and allows rotation around a single C[BOND]C bond. This converts an enantiomer into the meso-form and eventually into the other enantiomer. Therefore, ripening of homochiral domains becomes possible at higher coverage without lateral mass transport, which is usually hampered under such conditions.

Figure 30.

(online color at: Diastereomers of a compound that switches its adsorbate handedness upon thermal activation 210. Except for the oxygen-containing functional groups, hydrogen atoms are omitted. (a,b) Tert-butyl groups (highlighted in yellow), dominate the STM appearance and help identify the enantiomers. Rotation of a C[BOND]C bond of a partially lifted molecule interconverts one enantiomer into the achiral meso form (c) and in a second rotation step into the other enantiomer.

Photoinduced cistrans isomerization of azobenzene (AB) derivatives is well known from the liquid phase, but has also been applied to adsorbates 211–214. The flat-lying trans-isomer switches into a “helical” cis-isomer, that is, from an adsorption-induced chiral entity into an intrinsically chiral molecule (Fig. 31). This applies especially to the tetra-tert-butyl derivative of AB. Again, the tert-butyl groups dominate the STM appearance, and hence, four lobes are observed for the trans-isomer. Interestingly, the switching process has been found to be chiroselective on Au(111) 215. Two different types of cis-adsorbates are created in their two enantiomeric states, giving rise to four surface diastereomers being available. Depending on the type of the chiral domain in which the trans-isomer is located, certain diastereomers are formed. Molecules in the left-handed trans-domain convert either into cis-type 1, right-handed or into cis-type 2, left-handed. Molecules in the right-handed trans-domains only convert into cis-type 1, left-handed or cis-type 2, right-handed. This observation even allowed insight into the switching mechanism, that is still debated for the liquid phase. A “planar” cis-transition configuration imposes a steric strain due to the bulky tert-butyl groups next to each other. This causes either benzene ring of the molecule to twist out of plane, leading to one or the other of the two observed cis-configurations. The same process in the enantiomorphous trans-domain leads to the two cis-configuration enantiomers.

Figure 31.

(online color at: Chiral photochemistry at surfaces. Adsorbed trans-AB is chiral due to surface confinement, but turns into an intrinsically chiral cis-isomer upon switching 211–214. Depending on which side is lifted, a left-handed or right-handed segment of a helix is created.

Instead of photophysical switching, single AB derivatives have also been switched between their trans- and cis-configurations via inelastic electron tunneling or by the electric field under the STM tip 216, 217. In part, the switching was also thermally activated 218. Interestingly, for amino-nitro-AB and anilino-nitro-AB, the cis-configuration was found to be basically planar with all rings forced down to the surface 219, 220. Hence, the cis-configuration must be considered also as an adsorption-induced chiral entity, although it might be an intrinsically chiral molecule in the liquid or gas phase.

For the propene/Cu(211) system (Fig. 7a) switching between enantiomeric states of single adsorbates via inelastically tunneling electrons has been shown recently 47. The inelastically tunneling electrons deposit energy into molecular vibrational modes that couple anharmonically to action modes like turning, hopping, or even dissociation 221. At sufficient bias and current, conversion between two enantiomers was observed. However, it remained unclear if the molecule flipped over, a structural inversion of the highly vibrationally excited molecular frame occurred, or if proton tunneling from the methyl to the methylene end of the molecule took place. For the lateral manipulation, a rolling motion has been concluded, whereby an uneven number of rolling events caused conversion into the other enantiomer 222. Similar experiments were performed for meta-nitro-chloro-benzene on Au(111) 223, whereby the flipping mechanism of the molecule was favored due to fixation to the surface via the nitro group.

Chiral molecules rotating parallel to the surface plane due to excitation with tunneling electrons has been reported recently 224. A unique approach to unidirectional rotors has been taken by Feringa and coworkers. Based on overcrowded helical systems they designed rotors that depend on photophysical cistrans isomerization followed by thermally induced helix inversion 225, 226. The designed handedness of the rotor determined the sense of rotation. Furthermore, this research group functionalized surfaces with these rotors and made them rotate upon light exposure 227–229. Finally, this concept has been used to build a molecule with four unidirectional chiral rotor units. After adsorption onto the Cu(111) surface and because of the identical sense of rotation of all four rotors, excitation via inelastically tunneling electrons cause an unidirectional propulsion of the molecule 230.

9 Chiral expression at the liquid/air and solid/liquid interface

Due to the difficulty to obtain structural information from liquid surfaces, this article focuses rather on the solid/gas interface. However, two examples of chiral expression at the liquid/air interface are briefly mentioned here. More examples can be found in topical reviews 231, 232. Using chiral amphiphilic molecules 2D crystallization has been studied with grazing incidence synchrotron X-ray diffraction (GIXD) 233. It has been shown that long-alkyl chain α amino acids self-organize into 2-D crystals in which the two enantiomers either form heterochiral domains or spontaneously separate into enantiomorphous islands composed of homochiral molecules. Moreover, diastereomeric acid–base interactions between different chiral amphiphiles induced spontaneous chiral separation at the liquid/air interface. The enantioselective interaction of chiral solutes with the molecules assembled at the surface via Cu-complexation revealed that RS complexes could be packed more densely than SS or RR diastereomers 233.

Nandi and Vollhardt 234, 235 reported recently the chiral discrimination of enantiomers in monolayers of methyl esters of palmitoyl-aspartic acid and NR-stearoyl-serine. The visualization of chiral morphologies was accomplished via Brewster-angle microscopy (BAM). Pure enantiomers on water showed single handedness in the form of spiral motifs, while racemates showed the formation of domains expressing opposite handedness.

As reported here for solid surfaces, amphiphilic prochiral molecules may self-assemble at the air/water interface into enantiomorphous crystalline domains as well 236.

10 Outlook and further reading

We have reviewed the principles of chiral surfaces science that have evolved during the past decade, when self-assembly of chiral molecules at crystal surfaces has been expanded enormously. For the next decade we expect that surface science will further contribute to help understanding unsolved problems from the fields of enantioselective heterogeneous catalysis and crystallization. For the latter, stunning chiral symmetry-breaking effects have been described, but await explanation. This includes, e.g., so-called Viedma ripening, in which grinding induces homochirality in solids that are in contact with saturated racemic or achiral solutions. Under abrasive stirring and continuous dissolution and crystallization a conglomerate of NaClO3 crystals turns into a precipitate of homochiral crystals 153. When chiral molecules interconvert easily in solution, Viedma ripening induces a molecular homochiral solid as precipitate 237. Boiling of supersaturated NaClO3 solutions spontaneously crystallize only one enantiomorph, and the question of chiral recognition in solution, possibly originating from an “Adam crystal” surface and propagating throughout the solution, is equally not understood 155.

The question of how homochirality of biomolecules evolved from an achiral prebiotic environment has intrigued scientists for a long time 238–240. This led to numerous suggestions and a quest for finding the appropriate amplification mechanism for symmetry-breaking scenarios at various stages of the emergence of life 241–245. Among the various symmetry-breaking scenarios, chemical reactions at crystal surfaces or asymmetric crystallization processes have been considered 246–248. If surface mirror-symmetry breaking effects, like those discussed in Section 5, can be combined with larger symmetry-breaking effects, like the Soai reaction 249 or even self-replicating systems chemistry, the reward for enantioselective synthesis as well as insight into the complex symmetry-breaking scenarios of homochirality will be tremendous. However, it remains questionable that this insight will lead us to any valuable clue about the origin of life 250.

These problems require that sophisticated tools are developed that work under ambient conditions rather than in ultrahigh vacuum. Hence, chiral surface science will also largely benefit from “closing the pressure gap”.

The reader should be well equipped now in the matter of chirality at surfaces. But chirality is omnipresent in so many fields ranging from nuclear and astrophysics, biology, mineralogy, materials sciences, social sciences, and many more. Wagnière and Gardener 251, 252 wrote excellent books on general aspects on chirality, basically touching all disciplines. The historical development of chirality from the early days of Fresnel and Biot to Pasteur can be found in Lowry's Optical Rotatory Power 253. Chemistry is especially concerned with asymmetric synthesis in order to avoid the task of enantioseparation 254, which touches the industrial scale 255 and heterogeneous catalysis 256. Chirality of liquids crystals has an enormous technological impact 257, but is also fairly common in nanotechnology and materials sciences 258, 259.

Light scattering and circular dichroism effects in chiral media have been of high interest since the days of Arago. Accounts on optical activity of chiral matter treat electronic and vibrational dichroism, Raman optical activity and nonlinear optics 260–262.

Chirality has a different meaning in nuclear and solid-state physics. For weak bosons chirality is defined as eigenvalue of the Dirac matrix γ5. This topic is treated, among other chirality issues, in Ref. 263. Especially for graphene, chirality describes the connection of the direction along which an electron propagates in this 2D system and the amplitude of its wave function. This has been demonstrated, e.g., by the intensity dependence of angular-resolved photoelectron spectra on the usage left or right circularly polarized light 264. Chiral effects in electron scattering by molecules, e.g., scattering of spin polarized electrons in chiral media is another fascinating topic in physics and will be such soon in materials science 265, 266.

Finally, I recommend an excellent account on the handedness of the human brain and chirality in social sciences by – nomen est omen – McManus 267.


circular dichroismdifferent absorption coefficients for left- and right-circular polarized light
conglomerateeutectic mixture of enantiomers, i.e

., a heterochiral mixture of homochiral crystals (A more general definition is 268

: “A conglomerate is a mixture of mirror-image crystallized phases exhibiting symmetrical enantiomeric excesses.”)
diastereomerstereoisomer, that is not an enantiomer
distal (proximal)anatomical terms for the point farthest away from (closest to) a body. Here used as “inner” and “outer”, i.e., part farthest away from (closest to) the center of an object or ensemble or part farthest away from (closest to) the surface
enantiomermirror image stereoisomers; from Greek for “opposite part”; (“opposite enantiomer” is redundant and not used, while“other”, “both” and “one enantiomer” can be found here, because it is commonly used in the literature)
enantiomeric excess (ee)deviation in % from the racemic state equation image
enantiomorphmirror image object; from Greek for “opposite shape”; applies to single objects and ensembles
enantiotoptwo identical parts (chemical groups) of a molecule when alteration of one of them will make the molecule chiral
heterochiralis an ensemble of chiral molecules when containing both enantiomers
homochiralis an ensemble when containing chiral molecules of identical handedness, i.e., when composed of one enantiomer only
mirror-symmetry breakinga process that destroys the mirror plane as symmetry element, thus creating a chiral entity
prochiralan achiral species that turns chiral upon interaction, e.g., due to adsorption
racemate, racemic mixturemixture of equal amounts of both enantiomers (statistical fluctuations at large numbers neglected)
stereogenic centerthe origin of mirror-symmetry breaking in a molecule, often a carbon atom with four different ligands
stereoisomerisomer with identical structure (molecular formula and bond connections), but whose atoms have different orientations in space
  • 1

    We write here the (2 × 2) transformation matrix that links the adsorbate lattice vectors (b1, b2) to the substrate lattice vectors (a1, a2) via b1 = m11a1 + m12a2 and b2 = m21a1 + m22a2, in the form (m11 m12, m21 m22).

  • 2

    Originally, this phase was denoted as (9 0, ±1 2). For general specification of the structure, no matter the handedness, the “master matrix” formalism is applied here 109.

Biographical Information

Karl-Heinz Ernst, originally trained as a Chemical Technical Assistant (CTA), studied Chemical Engineering and Chemistry at the University of Applied Sciences Berlin (TFH) and the Free University Berlin (FUB). He obtained his PhD from FUB and joined – after postdoctoral research at the University of Washington in Seattle – Empa, the Swiss Federal Laboratories for Materials Science and Technology. In 1995 he founded the Molecular Surface Science Group at Empa, specializing in chirality of two-dimensional molecular crystals, functional surfaces and single-molecule surface dynamics. He lectures at the ETH Zurich and the University of Zurich, where he holds an adjunct faculty position in chemistry. He has been a visiting researcher/professor at the University of California, Berkeley; the University of Washington, Seattle; the Institute of Physics, Chinese Academy of Science, Beijing; and the IBM Almaden Research Center, San José, USA.

original image