Translated from Chemie in unserer Zeit 1997, 31, 129 -139 and loaded to the Internet with permission of the Editor; please refer to the original publication.

Atomic force microscope and near-field microscope

The construction of an atomic force microscope (AFM) is surprisingly simple. It differs from a tunneling microscope (STM) by scanning mostly van der Waals forces instead of tunneling currents. Both techniques succeed in ambient atmosphere or in ultra high vacuum UHV, at room temperature or at very low temperatures (77 K or 4 K). Figure 1 shows the blueprint of an AFM for ambient conditions in the most frequent version. The sample is fixed to an x/y/z-piezo which moves it in the 3 directions in space by application of suitable voltages. An atomic tip (usually out of Si or Si₃N₄) is approached to the sample under computer control until the van der Waals attractive forces assume values in the range of 10^-7 to 10^-9 N. Only the lateral atomic resolution requires still further approach such that repulsive forces come to work. The tip is part of a miniscule cantilever spring (Si or Si₃N₄) which is coated with gold on its backside. A focussed laser beam hits the backside of the tip and is reflected to a mirror and from there to a split photodiode. If a change in height occurs during the scan in x (usually 512 points) and y (usually 512 lines) the balance of the splid diode is lost and reinstalled by actuation of the z-piezo which receives its voltage from the split photodiode. Thus, the distance (the force) stays constant during the whole scan. Furthermore, the voltage that actuates the z-piezo is transformed poit by point to the AFM image with submicroscopic resolution with the aid of a computer and imaging software.

Fig. 1. Block diagram of an AFM with light deflection as the distance control.

The resolution in z-direction is particularly sensitive. Thus, for instance molecular terrace steps are routinely recognized and resolved even on rough organic surfaces. That is shown in Figure 2. The steps on the (001)-face of 1,4-diphenyltetrazine are 1, 2, 3, 4 nm high. That corresponds precisely to 1 - 4 molecular layers that have a height of 10 Å according to the crystal structure.

Fig. 2 excellent quality (GIF 118 K) or Fig. 2 (JPEG 18 K)

Fig. 2. AFM topography of 1,4-diphenyltetrazine (P21/c) on (001) with molecular terrace steps that were measured at room temperature.

On the other hand organic molecules are very mobile and cannot be laterally resolved as individual objects at room temperature with the AFM. Such resolution is only attainable at 4 K oder 77 K, and an additional requirement are molecularly flat surfaces, which, however, are hard to prepare on crystals and polymers. On the other hand, it will be shown in the next section that molecular resolution succeeds at room temperature with mono- and multilayers (LB-films). Atomic resolution is more easy with inorganic crystals. Organic molecules are better adsorbed to electrically conductive supports and then resolved with the STM. This is shown in Figure 5b. Due to the high sesitivity in z-direction it is also possible to study directly in liquid cells how crystals grow or dissolve (frequently at molecular/atomic steps or screw dislocations, etc.) or how metals are electrochemically deposited, or how living cells devide [5]. It is possible to image the double helix of DNA or to manipulate and cut it into pieces at predetermined sites (a technique called nanodissection). Chemically functionalized tips can be used for direct measurement of chemical forces between single molecules. Atomic patterns have been formed by pushing single atoms over supports at 4 K and spherical fullerenes have been cut open using STM and also AFM tips. A collection of the numerous possibilities is given in a recent review article [5]. In all of these applications AFM provides only topographical information but not chemical information. Some chemical evidence can be obtained by measurement of the lateral displacement of suitably mounted tips which differentiates local friction forces that relate to chemical composition. However, that technique is only useful in flat samples. More distinct chemical information also on rough surfaces requires the application of scanning near-field microscopy SNOM. A light source is approached to a surface at a distance that is much smaller than half of the wavelength of the light that is used (near-field). As light is interacting with the surface, several effects can be used in reflection and transmission mode such as absorption, emission, intensity change, wavelength response, polarization.

An optical near-field microscope SNOM is more complex than an AFM. 8 different basic techniques exist [5]. Only one of these is able to scan rough and transparent or opaque surfaces by chosing light fibers that are pulled to extremely fine tips. Radii down to 15 nm are used for best resolution and the tips are not coated with metals. That option has to be chosen for organic surfaces. Figure 3b,c shows that flat-end metal-coated tips with typical apertures of 100 nm are unsuitable because they cannot follow submicroscopic roughness in constant distance of <=10 nm. The only suitable option are slim uncoated tips that do not heat up [5,6]. Tips that are coated with aluminium (m.p. 660°C) and have very small apertures (e.g. 20 nm) loose the metal at their ends when light is fed in and become far-field apertured tips with much larger apertures (e.g. 400 nm) for special use [6]. Unfortunately metal-coated fiber tips with normal apertures of ·50 nm develop temperatures of 100 - 500°C (up to their destruction) even in ambient atmosphere, while being at 5 - 10 nm distance to the surface [5,6] which is totally unsuitable as well. A setup similar to Figure 3 with uncoated tips was initially developed and used by Courjon [7] and can now be purchased from DME in Copenhagen or from FHR in Dresden. The laserbeam is coupled into the fiber, it hits to the surface that is in the near-field at 5 - 10 nm from the tip end, and is particularly efficiently reflected back into the fiber only in that near-field. Coming back unpolarized it is coupled out at the beam splitter and activates the photodetector point by point. The distance control is acheived by non-contact AFM. The z-piezo uses the damping of the amplitude of the horizontally vibrating tip if it is approached to the surface. Such damping is detected by a red emitting diode laser and transformed in a feedback signal for the z-piezo keeping the distance constant and at the same time creating a topographic non-contact AFM image. While the reasons for the distance depending damping are still not fully understood it is generally, though incorrectly, called "shear-force". The resolution of the reproducible images that are undisturbed from thermal effects (absorbing samples are scanned rapidly) is routinely better than 20 nm [5]. More specialized techniques reach even better resolutions. These are required if, for example, single base pairs of nucleic acids are to be resolved and identified. Clearly, such endeavour will only be successful with cold and slim tips.

Imaging of atoms and molecules with STM and AFM

Atomic and molecular resolution was achieved in STM- and AFM-images [8]. Such lateral resolution is only possible on atomically/molecularly flat surfaces and requires very low temperatures in the case of organic crystals in UHV, in order to freeze out all thermal movements. Graphite, inorganic layer structures or adsorbates and very regular Langmuir-Blodgett (LB) films (including those of polymers [5]) give high resolution at room temperature. However, up till now noboby achieved to image more than every second atom on the very regular surface of graphite (STM, AFM, frequency mixing/rectification techniques)[4]. That failure is a clear indication for the occurance of vibrations that are imposed to the tip by the very regular and defect-free array of atoms on the graphite surface. It was, however, also tried to put forward different theoretical interpretations [8]. Figure 4 shows a remarkably good STM-image of a highly resolved graphite surface. The peak-to-peak distances are 2.5 Å as in the previously reported images. Thus, only any other atom can be seen from the planar 6-membered rings that are connected to further 6-membered rings in the infinite honey-comb sheet. Prove for individually resolved atoms is only given if defect sites are also imaged, as was first demonstrated on a silicon surface [8].

Fig. 4 excellent quality (GIF 88 K) or Fig. 4 (JPEG 17 K)

Fig. 4. Highly resolved STM-surface of graphite at constant tunnel current, measured with a tungsten tip at a RASTERSCOPE 3000. We thank DME, Copenhagen, for the raw data of that image.

Chemical/physical polishing and plasma treatment was applied to metal surfaces in the UHV prior to exposure to oxygen in order to understand their catalytic activities at the level of atomic resolution. Examples are given in Figures 5a,b for oxygen on silver (it catalyzes the reaction of ethylene with oxygen to give ethylene oxide) and on nickel. The distances of the -Ag-O- rows in [001]-direction equalize upon complete coverage [9]. What is seen in the image are the Ag atoms (distance 4.46 Å). Furthermore Stensgaard et al. succeeded in imaging complete benzene molecules that were adsorbed to Ni atoms between -Ni-O- rows on the Ni surface. Four of them are seen in Figure 5b with the correct size.

Ag-O (JPEG 39 K)	Ni-O (JPEG 31 K)

Organic solid-state reactions

Organic crystals may react thermally, photochemically, with gases, or with other crystals without involvement of liquid phases. These chemical transformations occur in a non-topotactic manner in virtually all cases. Only in a few highly engineered cases will a single crystal form a chemically transformed single crystal with, by necessity, same space group and closely identical lattice constants. While solid-state photoreactions do rarely allow for larger scale syntheses the further types are waste-free environmentally benign if they proceed quantitatively giving 100% yield even in larger scale runs without any necessity for workup. Undoubtedly, these will be the synthetic techniques of tomorrow. Right now more than three hundred 100% yield reactions are available in 22 reaction types all across organic chemistry in the author's research group. The recently favored topochemical principle claiming minimal atomic and molecular movements (<2.7 Å [3]) led to unresolvable contradictions in photochemistry [4,5,11] and it could not even think of or deal with gas/solid and solid/solid reactions. The reason for those failures is quite obvious: AFM investigations revealed, that solid-state reactions require long-range molecular movements. If the possibility for such long-range movements is not available due to heavily interlocked crystal packing or due to fixation by very strong ionic bonding, no reaction will occur even if the reaction centers are at close distance and favorably oriented [4,5,12] (the only exceptions are a handful of topotactic reactions [4]). Thus, the opposite principle of maximal molecular movement determines the field. Fortunately, the new experimentally secured principle wipes out all previous inconsistencies of Schmidt's topochemistry. The molecular movements over tens or hundreds of nanometers are easily recognized by submicroscopic features that form at the surface. These features are much larger than the molecular dimensions. They can be classified in eight geometric types and they are strictly related to the crystal packing. They were vividly named according to their appearance as flat covers, craters, volcanoes, egg trays, moving zones, islands, heights and valleys, floes. As it is not only distances between reaction centers that count, X-ray crystal analysis gained much more importance for the understanding and prediction of solid-state reactions than previously. Three stages have to be differentiated which in that sequence enforce 100% yield in most cases particularly in gas-solid reactions just by profiting from the crystal packing:

1) Newly formed molecules have to arrange with the basic lattice around them in the initial stage of the transformation. Molecules move along crystallographically easy paths and thus form characteristic face-specific features by phase rebuilding.

2) If the concentration of product molecules increases, a point will be reached where the distorted initial lattice cannot be kept any more. The product lattice forms suddenly usually with very large changes of the surface topography by (local) phase transformation.

3) The lattices of product and educt are not comensurable in most cases. The newly formed solid phase breaks off, frequently with disintegration of the whole crystal and fresh surface is formed for the next cycle of reaction.

It is that three stage mechanism that leads to 100% yield in many cases by using crystal structure effects very efficiently. It works in all four types of solid-state reactions.

Photoreactions

Crystal photolyses can be investigated at ambient or low temperatures but only until the disintegration of the crystal occurs.

Figures 7c,d show the situation on the (110)-face of 1. Half of the molecules are 23° skew, the other half 76° steep on the long edge [4,5]. Therefore, pressure that has built up inside the crystal by the photodimerisation can only be released here by vertical molecular movements above the crystal surface. During the phase rebuilding stage small volcanoes form everywhere (Fig. 7d). These grow upon further irradiation. Thus, the packing of the molecules determines the shape of the surface features and these are definitely different on different crystallographic faces.

Fig. 7a excellent quality (GIF 111 K) or (JPEG 20 K)
Fig. 7b excellent quality (GIF 92 K) or (JPEG 19 K)
Fig. 7c excellent quality (GIF 94 K) or (JPEG 19 K)
Fig. 7d excellent quality (GIF 88 K) or (JPEG 19 K)

The photodimerization of alpha-trans-cinnamic acid 3 (stable crystal modification) to give overwhelmingly alpha-truxillic acid 4 is a further historically important example.

Fig. 9a excellent quality (GIF 120 K) or (JPEG 23 K)
Fig. 9b excellent quality (GIF 140 K) or (JPEG 17 K)
Fig. 9c excellent quality (GIF 146 K) or (JPEG 19 K)

Thus, chemical uniformity is maintained across all features on the surface. An example is given in Figure 10 showing µm high floes perpendicular to the initial cleavage plane direction (two trenches in that direction) shortly before disintegration. The SNOM image (Fig. 10, right) shows, that despite the very high and steep crystallites that are seen in the 2D-projection (Fig. 10, left) no significant chemical contrast is detected. At the very high and steep edges of the floes only a minor steepness contrast is seen that does not exceed the overall roughness in the image and thus remains insignificant. Such SNOM evidence indicates uniform progress of the photoreaction at all sites of the surface despite the extreme topography.

Fig. 10 (AFM) excellent quality (GIF 127 K) or (JPEG 13 K)
Fig. 10 (SNOM) excellent quality (GIF 182 K) or (JPEG 24 K)

The new theoretical concepts also help in the understanding of rare cis/trans-isomerizations within crystals, that demonstrate space-demanding internal rotations by 180° [4,5,15]. Again, these reactions require a phase rebuilding mechanism which, of course, can be derived from the crystal packing where available [15].

Gas/solid-reactions

If molecules can move in the crystal (for the phase rebuilding) gases may react with organic crystals up to complete formation of product crystals. Local melting may be frequently avoided by cooling. More severe obstacles consist in surface passivation. Fortunately, these are easily recognized with AFM and may be handled by suitable means in many cases. All quantitative syntheses without liquid phases are waste-free and environmentally benign [5]. Several reactions have been performed at the kg-scale with technical bulk chemicals, that are available in sufficient purity. Examples are the reactions of sodium alkoxides with CO₂ to give synthetically versatile (now commercially available) halfester salts of carbonic acid, or of hydroxylamine phosphate with exhaust gases containing acetone to give valuable acetone oxime in pure form, or of sodium (potassium) dithiocarbaminates with highly diluted dichloromethane to give N,N-dialkyldithiocarbaminomethanes in a catalyzed process. Both acetone and dichloromethane are removed below the detection limits. These gases may, of course, also be used at higher gas load if larger scale waste-free production is required. It is seen from these examples that solid-state syntheses may even serve for removal of pollution from the atmosphere [16]. Clearly, the new synthetic techniques are on their way to tomorrows production techniques.

Gaseous acetone and crystalline penicillamine 5 do not react to give the thiazolidine 6. Neither the racemate nor the enantiomer of 5 are successful.

Fig. 11a excellent quality (GIF 87 K) or (JPEG 18 K)
Fig. 11b excellent quality (GIF 84 K) or (JPEG 18 K)
Fig. 11c excellent quality (GIF 85 K) or (JPEG 17 K)
Fig. 11d excellent quality (GIF 100 K) or (JPEG 17 K)

The synthetic chemist has to find ways to overcome that obstacle by avoiding the formation of a closed cover which keeps the gas from approaching the reactive surface or prevents the crystal disintegration. In the present case it proved successful to use the hydrochlorides of racem-5 and of (R)-5, which had been quantitatively prepared by applying HCl gas to the bases. Thus, the reaction succeeded with the less nucleophilic salts that did not undergo passivation. 100% yields of the hydrochlorides of racem-6 and of (R)-6 were easily obtained without waste production [5,16b]. If the same chemical transformations are done in solution wastes are formed because of incomplete conversion and the necessity of product workup as well as disposal of that part of solvents that cannot be recycled.

Solid-state reactions are not always stereospecific. For instance, finely milled crystalline trans-stilbene reacts quantitatively with chlorine and bromine to give the meso- (trans-addition) and the racem-adducts (cis-addition) in the ratios of 18 : 82 (in CH₂Cl₂ 40 : 60) and 25 : 75 (in CS₂ 84 : 16), respectively [17,18]. The cis-addition prevails in the gas/solid reactions. Thus, it is possible to increase the yield in racem-adducts when chosing gas/solid instead of liquid reaction. The stereochemical course for the chlorination of stilbene is given in sawbuck representation. The bromination proceeds presumably via the well-known bromonium ions (similar to 8), that are attacked here preferentially from the frontside by Br- or Br₂.

The addition of chlorine to alpha-trans-cinnamic acid 3 gives the erythro- (trans-addition) and threo-adducts 7 (cis-addition) in the ratio of 12 : 88, and the reaction mixture turns liquid [18]. Conversely the bromination of 3 is governed by the influence of the bromonium ion 8 experiencing backside attack by Br- or Br₂. Only 7 is obtained. The sawbuck representation shows how it works. Both the alpha- and the ß-crystal modification of 3 give the racemate of erythro-7 with 100% yield in the absence of liquefying. Interestingly, both crystal modifications choose exclusively the trans-addition via the known bromonium ion 8, that can here only react from its backside for crystallographic reasons.

The phase rebuildings lead to diverse surface features on (010) of alpha-3 and on (100) and (010) of ß-3. Such behaviour is strictly correlated to the molecular packing data (ß-p-chloro cinnamic acid serves as a model for ß-3). Figure 12a,b show the phase rebuilding upon short application of bromine to the very flat surface of alpha-3. Out of molecular steps 150 nm deep craters and 150 nm high volcanoes with pillars form. The appearance is that of an eggs tray. The (100)-face of ß-3 (Fig. 12c) is predetermined to moving reaction zones as is seen in Figures 12d,e, and the (010)-face of ß-3 produces flat massive floes without pronounced geometric structure (Fig. 12f). Further exposure to bromine leads to crystal disintegration in all three cases.

Fig. 12a excellent quality (GIF 95 K) or (JPEG 17 K)
Fig. 12b excellent quality (GIF 88 K) or (JPEG 15 K)
Fig. 12c excellent quality (GIF 87 K) or (JPEG 17 K)
Fig. 12d excellent quality (GIF 122 K) or (JPEG 16 K)
Fig. 12e excellent quality (GIF 109 K) or (JPEG 15 K)
Fig. 12f excellent quality (GIF 66 K) or (JPEG 14 K)

Packing diagrams serve for an understanding of these striking differences. The stereoscopic representations for alpha-3 in Figure 13 show double bonds that are somewhat shielded by the phenyl groups in the outer half of the double layer. The pairs of molecules lie flat at 30° and have to slide over the surface for reaction with bromine. If the bromonium ion 8 is formed, the molecule is liberated from its H-bond partner and presents its backside for attack by bromide or bromine. The double bond of the left behind partner is now free for reaction with additional bromine (Fig. 13, bottom). That causes an upward transport which becomes faster at the sites where it started. Double layer by double layer has to be worked up and it is to be reasonably assumed that such processes perform best along the "pillars" (Fig. 12b) with formation of the eggs tray structure.

Fig. 13. Space filling stereoscopic packing diagram of alpha-3 on (010); O red, C black, double bond C brown, H white; both components of the double layers that are linked in the crystal by H-bridges are seen. The photodimerization to give 4 that is described in the preceeding section occurs between the molecules of the halves of the double layers (d = 3.590 Å; not imaged); the double bonds of the partners lie between the parallel double bonds of the single layers (d = 5.582 Å) with the result that these cannot react with each other.

The model for the packing of ß-3 (ß-p-chloro cinnamic acid, the crystal packing of which is agreed to be analogous to that of the not directly measurable crystal packing of ß-3) exhibits two halves of the double layer under (100) with the H-bridged molecules 83° steep in two orientations. The double bonds in both half-layers are completely inaccessible (Fig. 14). Thus, the reaction can only start at lattice defects and spread sideways from there after creating access to bromine by upward transport of the more voluminous product molecules 7. By doing so, the reaction eats itself into the surface in various directions.

	Fig. 14. Space filling stereoscopic packing diagrams of the structure model ß-p-chloro cinnamic acid (P21/a) on (100); Cl green, O red, double bond C brown, C black, H white; both components of the double layer are seen that are linked by H-bridges in the crystal; within the double layer the double bonds are also shielded.

On (010) of ß-3 (Fig. 15) the conditions are totally different. Both double bonds of the H-bond pairs are easily accessible. 8 is easily formed, the H-bridge of the 23° flat lying molecules is abandoned, the backside attack by bromide or bromine made possible. The molecules turn away and create access for the reaction in the next layers. All of that explains the formation of massive poorly structured flat floes that settle to the surface (Fig. 12f). The continuation of reaction is apparently secured as the covers separate after phase transformation (into the lattice of 7) from the surface.

	Fig. 15. Space filling stereoscopic packing diagrams of the structure model ß-p-chloro cinnamic acid (P21/a) on (010); the double bonds are freely available on that surface.

Such correlations of the characteristic structures to the crystal packings offer a deep understanding for the molecular movements occuring. And further: the new experimental knowledge allows for reliable predictions and development of new reaction systems by detailed analysis of the crystal packing. If (nano-)liquids occur as in the case shown with alpha-trans-cinnamic acid and chlorine, that fact does not escape the required contact-AFM measurements due to the recrystallization phenomena at the nanometer scale because the tip will always immerse into liquids and transport them over the surface [5,18].

Solid/solid reactions

Solid diazonium nitrates such as 9 are formed quantitatively and without wastes by gas/solid diazotization of anilines that are substituted with electron-withdrawing groups. Conversely, the usual diazotizations in acidic solutions create large quantities of highly corrosive, hard to dispose of, wastes. Such complicated solid-state reactions were equally analyzed with AFM and understood with the crystal packing of the starting solid anilines [19]. Solid diazonium salts are useful reagents for solid/solid-reactions. Thus, the well-known azo coupling with phenols (e.g. 10) to give the salts of azo dyes such as 11 occurs quantitatively just by grinding together the solid components. No intervening liquid phase was found by AFM [5].

It was shown visually and with AFM, that it is the phenol component that migrates through the surface contacts into the diazonium salt crystal. A tiny multicrystal of the diazonium salt 9 was laid down to a single crystal of ß-naphthol 10. The color of the dyestuff (acid form) was only seen in the small multicrystal but not in the naphthol crystal. Figure 16 shows a fresh (001)-surface of 10 (that was crystallized from water), the surface changes after reaction with the small polycrystal of 9 at short distance, and an explanatory sketch. The two crystals stick together immediately and cannot be separated any more. ß-Naphthol molecules 10 migrate across the phase boundary such that further molecules of 10 have to follow from all sides though directed by the crystal lattice, of course. Deep canyons form on (001) that run in direction of the long crystal edge along the cleavage plane and cross connections emerge at right angles, such that solid blocks remain [5]. These features are particularly easily imaged with the 2D map projection. In larger distances the valleys are flatter, the blocks lower. The formation of these features derives from the upward movement of molecules 10 that are under the crystal of 9 and their tendency to filling the voids from the sides while moving along the cleavage planes as is indicated in the sketch. As in the case of geologic erosion of a river valley also the side valleys form. Similar features have been disclosed in further solid/solid reactions with unidirectional motion of only one sort of molecules from one crystal into another crystal [20].

Fig. 16a excellent quality (GIF 97 K) or (JPEG 18 K)
Fig. 16b excellent quality (GIF 73 K) or (JPEG 16 K)
Fig. 16c excellent quality (GIF 128 K) or (JPEG 15 K)

Chemical contrast with SNOM

Surface features that are detected on reacting crystals by AFM do not provide chemical but only topographical information. It is therefore important to check if local differences in chemical composition exist or not. Scanning near-field optical microscopy SNOM is the obvious technique for that task. It has been shown already in the irradiations of 1 and alpha-3 that chemically uniform surfaces do not give SNOM contrast, despite the presence of features. However, if local differences in chemical composition are created on reacting surfaces one detects chemical contrast with the SNOM. Such behaviour is typical for island mechanisms or reactions that are restricted to particular crystal faces as is typical for molecular packings that hide the functional groups under the surface. Such protection of the crystal from chemical reaction can often be removed by fine grinding or milling which creates all crystal faces including those that have the functional groups out [5,21]. Mechanistic evidence for the correct interpretation of the island mechanism has been obtained with one of the first examples of chemical contrast with SNOM: In the gas/solid diazotization of sulfanilic acid 12 to give the inner diazonium salt 13 (the starting material for the pH-indicator methyl orange)

Fig. 17 (AFM) excellent quality (GIF 103 K) or (JPEG 14 K)
Fig. 17 (SNOM) excellent quality (GIF 95 K) or (JPEG 16 K)

It could only succeed with a cold and slim tip according to Figure 3. Chemical SNOM contrast was also obtained recently with reactions of 1, ß-3, 2-mercaptobenzothiazol, benzimidazol and further organic crystals. These results suggest further development of the theory of near-field reflection [6]. That should open a new field of research in physics.

Summary

Supermicroscopic techniques like STM, AFM and SNOM are able to image surfaces down to atomic resolution. Their application to chemically reacting rough organic surfaces leads to unprecedented results which do not accord with the topochemical principle of minimal molecular movement. Rather, feature-forming long-range molecular movements are essential for chemical reactivity. Thus, waste-free solid-state reactions proceed in three steps: phase rebuilding, phase transformation and crystal disintegration. Liquid phases are avoided in solid photochemical, solid thermal, gas/solid and solid/solid syntheses, the upscaling of which will provide the future chemical production techniques. Chemical contrast with SNOM is shown for the first time. It complements the AFM results which are interpreted in terms of crystal structure.