Phosphate flakes on a biological specimen

One application potential of s-SNOM is for biological and medical purposes, by imaging microscopic biomineral structures [Reference Amarie3]. Figure 2a shows images of a polished cross section of a mussel shell consisting of calcium carbonate microcrystals bound by organic material. The calcitic crystal structure in the outer part of the shell (top) abruptly transitions to the periodic arrangement of aragonite microcrystals in the inner part (bottom), leading to the well-known mother-of-pearl iridescent appearance of structural color. With monochromatic IR illumination selected at 980 cm^-1 (corresponding to a wavelength of 10.2 µm), which is away from calcium carbonate vibration frequencies, the IR image shows no contrast between these types of microcrystals, but rather, a consistently higher signal than the organic material in between. Notice that the differing topographical height of the calcite surfaces does not change the observed IR signal, as the latter only depends on refractive index; there is no topography artifact or cross-talk into the IR image.

Figure 2 (a) IR nanoscope image (10 × 10 µm²) of a polished section through a mussel shell; topographical image (left) and back-scattering signal at 10.2 µm wavelength (right). Both the IR and topography images show microcrystals at the same high resolution, known to be calcium carbonate of either the calcite or aragonite crystal structure. Both polymorphs exhibit the same IR signal because their vibration frequencies are far from the probing frequency, but the image also exhibits bright patches indicating a contaminating material (see text) [Reference Amarie3]. (b) Nano-FTIR spectra along a line on the topography image depicted as rectangle. The line crosses a 10-nm thin object in the calcite-aragonite transition region of the mussel shell. Fingerprint vibration lines clearly distinguish both carbonate phases and identify the thin object as a 10 nm thin flake of crystalline phosphate [Reference Amarie3].

Surprisingly, numerous nano-scale features of enhanced scattering were discovered (yellow in the chosen color scale) and shown in Figure 2a. Close inspection of the topography verified that these are about 10 nm high objects. In order to learn about the chemical nature of these unexpected objects, the same location was investigated by the nano-FTIR mode of operation. Three-hundred IR spectra were collected along a line crossing one of the objects and extending 2.5 µm over several calcite and arogonite microcrystals. The result (Figure 2b) shows the expected fingerprint resonance due to the well-known crystal vibrations of trigonal calcite and orthorhomic aragonite, at 873 and 855 cm^-1, respectively. The resonance at about 1,018 cm^-1 identifies the thin object as crystalline phosphate. Close inspection (Figure 2a) reveals that all such flakes observed appear to be 10–20 nm high and have a smoother surface than the carbonate crystals. Even though these features are found in the calcite but not the aragonite, they were tentatively assigned to be just preparation artifacts. In any case, the important conclusion remains that s-SNOM and nano-FTIR are capable of not only detecting but also identifying tiny crystals on a sample surface.

Human bone

Bone samples were nanoscopically inspected after standard poly(methyl methacrylate) (PMMA) embedding and polishing. Figure 3 shows an IR s-SNOM image of a human femoral head at 9.47 µm wavelength. This resonance is associated with the PO₃^- vibration of phosphate nanocrystals, the main mineral component of bone [Reference Geith4]. The marked void designates a canaliculus. It is surrounded by a 1 µm wide region of increased IR signal, which indicates a higher mineral density. The observed nanostructures were suspected to be fibrils of collagen with attached phosphate nanocrystals. The fibril orientation appears to be perpendicular to the sample surface in the neighborhood of the canaliculus, but not so further out. Further images were taken of lacunae [Reference Geith4], and also of tubuli in human dentin, and the structures were verified by scanning electron microscopy (SEM) and BE-SEM. Because SEM can damage the sample and requires conductive coatings, SEM images were taken only after completion of s-SNOM imaging because s-SNOM is sensitive to damage as it probes a thin surface layer only (of 20–50 nm depth) [Reference Keilmann and Hillenbrand2].

Figure 3 IR nanoscopy of polished human bone sample. S-SNOM probing is sketched here, with the probing wavelength of 9.47 µm and the tip at approximately the same scale as the IR image of apatite nanocrystals surrounding a canaliculus [Reference Geith4].

Organic materials for solar cells

Solar cells converting sunlight directly into electric power are installed all over the world; these mostly consist of single-crystal Si material to optimize photon-induced charge production and transport. The minimization of structural defects that could scatter charge carriers allows a conversion efficiency that is currently about 25% [7]. The quest for lower-cost alternative materials is proceeding with great international effort. Organic solar cells consist of suitable conductive polymers or small organic molecules that facilitate large-area flexible devices and low-cost production on many types of substrate by wet-chemical processing or thermal evaporation. In the case of well-defined small molecules, the resulting organic thin films are highly ordered, but they are far from being defect-free. They show a crystalline grain structure on a length-scale on the order of 1 µm (Figure 4a). Key challenges for organic solar cell development are these materials' short diffusion length, due to a high density of structural defects and a short carrier scattering time. Certainly the nanoscale morphology is a key issue for charge-transport properties, and accordingly, great effort has been made toward morphology control [Reference Lin8]. It achieved device conversion efficiency up to 11% [7]. However, it is very difficult to map the nanoscale crystallization, and it remains unclear where the electronic defects are localized.

Figure 4 Pentacene film (about 40 nm thick) on Si, after 20 months of storage. (a) Topography showing a grain structure. (b) Sketch of molecular order for two adjoining polymorphs, thin-film phase (TFP) and bulk phase (BP). (c) Result of X-ray scattering (linear scale) to determine average molecular layer period after preparation (black) and stored for 20 months (red) [Reference Westermeier5].

Nanoscale polymorphism in thin-film pentacene. Today’s best-performing organic solar cells consist of thin films with optimally designed, synthesized molecules. Pentacene has long been the prototypical molecule to study the fundamental electronic and structural properties of organic thin films. It is a small molecule made up as a linear chain of five benzene rings. On evaporation, pentacene molecules stand almost perpendicular to the SiO₂ substrate surface. Continuous thin (typically 40 nm) pentacene films of ordered crystalline grains (Figure 4b) can be p-type semiconductors with mobility up to 10 cm²/Vs, comparable to amorphous Si. X-ray diffraction aimed at determining the molecular layer period shows two distinctly differing values (Figure 4c). This means that two distinct polymorphs, the thin-film phase (TFP) and the bulk phase (BP), coexist and even change over time. The footprint of the X-ray beam on the sample is macroscopic, of the order of 1 mm², and until recently no available microscopy could decide the important question of how both polymorphs may be spatially distributed, and over which lateral or vertical dimensions (see Figure 4b for a possible arrangement). Infrared nanoscopy now straightforwardly decides these issues [Reference Westermeier5].

The essential reason for this success is that the two structural phases differ in their vibration frequencies. For Figure 5b the IR frequency was chosen to highlight the BP polymorph. Indeed there are elongated BP ellipsoidal domains of about 400 µm in width everywhere on the sample, obviously embedded in TFP pentacene. They do not appear to be correlated to the topographic morphology because they even extend through grain boundaries. A repeat of the image acquisition with the illumination frequency changed to highlight TFP pentacene produces an IR image with reversed contrast, as would be expected. As a final experimental proof, nano-FTIR spectra were acquired at different positions of the sample (Figure 5c). Evidently the local spectrum taken on a BP ellipsoid shows a higher resonance frequency (906 cm^-1) than that of the surrounding TFP (904 cm^-1). These values are consistent with published far-field FTIR-spectra of the two polymorphs. Note that a basic agreement of far-field and near-field resonance positions of molecules has been recently confirmed in experiment and theory [Reference Huth6]. Lastly, it may be questioned whether the polymorphs extend throughout the film depth, as sketched in Figure 4b, or whether there may be a sign of inhomogeneity also in the vertical direction. The answer is negative from the following observation: the topography (Figure 5a) shows on close inspection a depression of about 2–3 nm at all BP elliptic regions. Given the film thickness of 40 nm, the depression amounts to about 6% shrinkage, and this agrees quantitatively with the lowering of vertical extent of a BP (1.44 nm=14.4 Å) compared to a TFP pentacene molecule (1.54 nm=15.4 Å). Thus an inhomogeneous ordering of TFP and BP phases in the vertical direction can be ruled out. In fact it is highly probable that an initially homogeneous TFP pentacene film over time changes its conformation, by nucleation and lateral growth of vertically homogeneous BP polymorph ellipsoid domains, keeping the number of molecular layers unchanged [Reference Westermeier5].

Figure 5 High-resolution IR nanoscopy of a similar pentacene film as in Figure 4a. (a) Topography, (b) simultaneously recorded IR map at 907.1 cm^-1 (corresponding to 11 µm wavelength, resonant with BP pentacene), and (c) nano-FTIR spectra at two locations, on and off a BP ellipsoid domain [Reference Westermeier5].