Film growth, adsorption and desorption kinetics of indigo on SiO₂

Abstract

Organic dyes have recently been discovered as promising semiconducting materials, attributable to the formation of hydrogen bonds. In this work, the adsorption and desorption behavior, as well as thin film growth was studied in detail for indigo molecules on silicon dioxide with different substrate treatments. The material was evaporated onto the substrate by means of physical vapor deposition under ultra-high vacuum conditions and was subsequently studied by Thermal Desorption Spectroscopy (TDS), Auger Electron Spectroscopy, X-Ray Diffraction, and Atomic Force Microscopy. TDS revealed initially adsorbed molecules to be strongly bonded on a sputter cleaned surface. After further deposition a formation of dimers is suggested, which de-stabilizes the bonding mechanism to the substrate and leads to a weakly bonded adsorbate. The dimers are highly mobile on the surface until they get incorporated into energetically favourable three-dimensional islands in a dewetting process. The stronger bonding of molecules within those islands could be shown by a higher desorption temperature. On a carbon contaminated surface no strongly bonded molecules appeared initially, weakly bonded monomers rather rearrange into islands at a surface coverage that is equivalent to one third of a monolayer of flat-lying molecules. The sticking coefficient was found to be unity on both substrates. The desorption energies from carbon covered silicon dioxide calculated to 1.67 ± 0.05 eV for multilayer desorption from the islands and 0.84 ± 0.05 eV for monolayer des orption. Corresponding values for desorption from a sputter cleaned surface are 1.53 ± 0.05 eV for multilayer and 0.83 ± 0.05 eV for monolayer desorption.

I. INTRODUCTION

The ongoing search for higher performance electronics and possible future implementations of organic devices into everyday life led to an uprising of a variety of new semiconducting materials.^1,2 Natural dyes such as indigo and Tyrian purple have been used for thousands of years by ancient cultures in India, China, and Egypt to color textiles and food. Indigo is still the most commonly used dye today, thanks to jeans and their typical blue color. Originally, it was obtained from the plants Indigofera tinctoria and Isatis tentoria, although with recent advancements in synthesis and fabrication it is almost exclusively produced artificially today.^3,4 Despite the high level of awareness, the formation of highly-crystalline indigo films upon evaporation and the good charge transport properties have only been discovered recently.^5,6 The combination of a reversible oxidation and reduction electrochemistry plus a low bandgap makes such materials behave as ambipolar semiconductors with fairly high mobilities of 0.01 cm²/Vs. The low cost and low toxicity evoke high levels of interest for a new era of sustainable and bio-degradable materials for organic electronics.^7,8

The indigo molecule (C₁₆H₁₀N₂O_2, see inset in Figure 1) shows a high planarity and forms a double cross-conjugated system with two electron donor groups and two acceptor groups with a central C=C double bond.⁹ Its deep blue color can be explained by the conformation of the two carbonyl groups with respect to the central C=C bond, as well as by the electronic arrangement of the heteroatom and its high polarisability.⁹ One thing that distinguishes indigoids and other natural dyes from more well-known organic semiconductors is the formation of intermolecular hydrogen bonds during stacking. Each molecule is hydrogen bonded to four of its neighbors, while adjacent molecules are also π-stacked to each other, leading to π-π interactions that are reinforced by hydrogen bonds.¹⁰ Air-stable OFETs of indigoids and other amine/carbonyl dye molecules exhibit improved field-effect mobilities and other functionalities inaccessible to conventional organic semiconductors and outperform many established materials in terms of operational stability.^11–14

FIG. 1.

Thermal desorption spectra of indigo from carbon covered silicon dioxide for different exposures. Adsorption temperature T_ad=220 K, heating rate ß=1 K/s. The exposure is given in Hz, as determined by a quartz microbalance: (a) 1–5 Hz, (b) 12–65 Hz. The small peak in (b) at 390 K (Ta) stems from desorption from the Ta mounting clamps. The temperature correction is described in the supplementary material.¹⁷ The chemical structure of indigo is shown in the inset of Figure 1(a).

In this work, we investigated the initial growth behavior of thin indigo films on silicon dioxide for possible applications in transistor fabrication. We prepared high purity thin films by means of physical vapor deposition in ultra-high vacuum and used these well-controlled conditions to study adsorption and desorption behavior with Thermal Desorption Spectroscopy (TDS) and Auger Electron Spectroscopy (AES). In addition we investigated film morphologies and the air stability of ultra-thin layers with Atomic Force Microscopy (AFM). X-Ray Diffraction (XRD) was used to examine crystallinities in the bulk phase.

II. EXPERIMENTAL SETUP

For our experiments indigo, provided by ALDRICH with a purity of 95%, was deposited, after proper outgassing, via physical vapor deposition from a Knudsen cell heated to ≈ 220 °C. Silicon dioxide (150 nm thickness) that has been thermally grown on 0.67 mm thick Si(100) wafers (Siegert Wafer) was used as substrate material. For a quantitative determination of the evaporation rate from the Knudsen cell and therefore for the deposited amount of indigo onto the sample, the frequency change of a quartz microbalance thin film monitor was observed. Due to the geometries of our experiment, the sample had to be temporarily replaced by the microbalance prior to the actual deposition process. According to the relationship

$$d=-S^{-1}\frac{\Delta f}{f^2\rho }$$ (1)

the mean thickness d of a thin film is related to the quartz frequency change Δf via the crystal operating frequency f (about 6 MHz), the quartz sensitivity S (2.26 × 10⁻⁶ cm² s/g) and the density of indigo ρ (1.50 g/cm³).¹⁵ For the provided system a frequency change of 1 Hz equals a mean thickness change of 0.8 Å. One should note that the relationship shown above is only true, if the sticking coefficient of the indigo molecules on the silicon dioxide substrate as well as on the quartz crystal is unity, which is typically the case for organic molecules (and will later be confirmed by experimental data). Deposition rates varied between 6 ng/(min cm²) and 270 ng/(min cm²), equivalent to 0.04 nm/min and 1.8 nm/min. During deposition the sample was cooled via liquid nitrogen to a temperature of 220 K. The residual pressure in the UHV chamber was in the 10⁻⁹ mbar range.

To analyze and investigate changes in the chemical composition of the film and substrate surface we used AES. Besides the anticipated silicon and oxygen peaks, carbon was observed as contaminant on the silica surface. We were able to easily remove any carbon with 10 min of argon sputtering, however, any subsequent temperature treatment resulted in a segregation of carbon atoms from the bulk and thus lead to additional carbon contamination during the heating process. An Auger analysis of deposited indigo films showed carbon, nitrogen, and oxygen signals with no impurities. Unfortunately, it was not possible to perform continuous Auger measurements during different stages of thin film growth, due to the destructive behavior of the electron beam on soft organic surfaces.

The Si/SiO₂ samples were mounted onto a stainless steel sample holder via four tantalum clamps; the steel plate was heated resistively. Attributable to the poor heat conductivity of SiO₂ and the thermal contact resistance between the silicon wafer and the steel plate, a significant temperature lag between the sample surface and the heating plate was observed. A first order temperature correction process was applied and will be discussed in Sec. III A. Further information about the measurement setup and details about the correction process are given in more detail in Ref. 16 as well as in the supplementary material.¹⁷

After material deposition from the gas phase, the sample with a grown thin film on top was placed in front of a mass spectrometer tuned to mass 76 (most prominent fragment in the indigo cracking pattern) and heated linearly until the material desorbed from the surface. This thermal desorption spectroscopy method makes it possible, if the temperature-time relation during the heating process is sufficiently controlled, to investigate thin film characteristics, as well as growth and desorption behaviors.^18,19 Through the peak shape, one can determine the desorption order and whether the activation energy for desorption is constant or a function of the surface coverage. Furthermore, the peak position is an indicator for the bonding strength and can also be used to observe state transitions with increasing surface coverage and film thickness. Technically, this only holds true if the desorption rate curve of one desorbing state can be clearly distinguished from other states. Fortunately, this was the case for all our experiments. An integral over the resulting spectrum versus time or temperature is directly related to the amount of deposited material onto the sample surface. For the described experiments a linear temperature variation with time (T=T₀ + βt) with heating rates β of typically 1 K/s was used.

Analysis of the surface structure and morphology has been provided ex situ by an atomic force microscope in tapping mode (Nanosurf Easyscan 2). Specular X-ray diffraction measurements were performed with a Siemens D501 diffractometer using the radiation of a sealed copper tube. X-ray diffraction pole figures were collected with a Philips X’Pert system equipped with an ATC3 cradle, using the radiation of a chromium tube. In both cases a Bragg-Brentano focusing geometry was used in combination with a flat graphite monochromator at the secondary site.

III. RESULTS AND DISCUSSION

A. Adsorption and desorption of indigo on carbon covered SiO₂

Small organic molecules on reactive surfaces typically show a growth mechanism that is known as Stranski–Krastanov (SK) growth, where the initial layer (wetting layer) consists of lying molecules that are strongly bonded to the substrate with a multilayer of either standing or lying molecules forming on top. Examples for such film formations include para-hexaphenyl on Au(111)²⁰ and mica(001),^16,21,22 pentacene on Au(111),²³ and Si(111),²⁴ as well as quaterphenyl on Au(111)²⁵ and PTCDA on Ag(111).²⁶ Other systems, such as PTCDA on Si(100)²⁷ show layer-like growth for low substrate temperatures which switches to island growth for higher substrate temperatures. TDS shows two desorption peaks for materials, which exhibit SK film formation. Usually a single peak for monolayer desorption at high temperature is followed by a second peak for multilayer desorption at lower temperatures. However, in case of weak interactions between the organic molecules and the substrate, the first layer may already dewet prior to desorption, thus resulting in a single desorption peak.¹⁸

Figure 1 displays desorption spectra of indigo from silicon dioxide for different exposed amounts at adsorption temperatures of 220 K. At this point we need to emphasize again, that after each deposition and subsequent desorption cycle, a specific amount of carbon will remain on the surface. Once an adequate amount of cycles is reached, the accumulated carbon reaches a saturation thickness above which no further decomposition can be observed (Auger ratio C272/O510 ≈ 0.18). This stable and inert substrate was used for the adsorption and desorption experiments in this section.

Figure 1(a) shows a series of spectra for very low exposures between 0.8 and 4.0 Å mean film thickness (see calibration of the coverage below). Initially, for exposures smaller than 0.8 Å, only a single peak at around 400 K (corrected temperature T_corr=290 K) can be observed, designated as α-peak. For exposures between 0.8 and about 2.7 Å a second peak, designated as β-peak, appears at approximately 150 K (corrected value: 52 K) higher temperature. With further increase in coverage the α-peak starts to decrease up to a point where it is not distinctively visible anymore at a coverage of about 9.6 Å. The β-peak increases meanwhile (1.6–4 Å) and, once the initial α-peak has disappeared, continues to increase with a common leading edge for higher coverages (9.6–52 Å), as can be seen in Figure 1(b). For high coverages, the small peak at 390 K, marked as Ta-peak, does not originate from desorption from the actual sample surface, it rather shows multilayer desorption from the tantalum foils, which are used to attach the silicon wafer onto the steel sample holder. Due to the good heat conductivity between those two materials and the direct connection to the thermocouple we can assume that this early peak shows the true desorption temperature of indigo in the multilayer state. Therefore, this can be used to correct the temperature scale (see the supplementary material¹⁷). Thus, the true desorption temperature of the α-peak is at around room temperature and that for the multilayer at about 350–400 K.

Taking only TDS measurements into consideration, we cannot distinguish between flat-lying and side-tilted molecules during initial film formation (α-peak), but we can safely assume, due to the low amount of adsorbed material, that at this point the interactions between the monomers on the surface are either non-existent or negligible. A densely packed monolayer of flat-lying indigo molecules would correspond to a mean thickness of about 3.5 Å, based on the vander-Waals dimensions of the molecule.²⁸ Thus, the change of the desorption peak α into a more strongly bound state β, that occurs already at a mean thickness of below 1 Å, indicates the re-arrangement of the molecules way below a third of a monolayer. This re-arrangement can most probably be described by a dewetting process leading to the formation of islands, in which the molecules are more strongly bound than on the silicon surface. A behaviour, where the molecules in the multilayer islands are more strongly bonded than in the wetting layer is quite unusual for organic materials and has only been observed in a few cases; for the first time by Jacob and Menzel for benzene on Ru(001)²⁹ and more recently also for bithiophene on Cu(110).³⁰ In our laboratory, we have observed such a behaviour for rubicene on SiO₂³¹ and hexaaza-triphenylene-hexacarbonitrile (HATCN) on Au(111)³² and Ag(111).³³ In the first case, a formation of two wetting layers of flat-lying molecules was observed, with a subsequent dewetting and island formation at and above a certain coverage threshold. HATCN on Au(111) and Ag(111) forms a weakly-bonded second layer on top of a strongly bonded wetting layer. At higher coverages the second layer gets incorporated into the multilayer by dewetting, once more leading to desorption at higher temperatures.

Further evaluation of the desorption spectra for indigo films above a certain thickness (Figure 1(b)) shows a common leading edge and a desorption maximum that shifts to higher temperatures for increasing coverage, a behavior characteristic for zero order desorption kinetics from multilayers. For first and higher order desorption (Langmuir desorption), one can deduce the desorption energy from the following relationship (Polanyi-Wigner equation):¹⁹

$$R\left(T\right)=-\frac{d\Theta }{\mathit{dt}}={\nu }_n{\Theta }^n{N}_{\mathit{ML}}^n{e}^{-\frac{E_{\mathit{des}}}{k_{B}T\left(t\right)}},$$ (2)

where R(T) is the desorption rate from a unit surface area, n is the order of the desorption reaction, Θ is the surface coverage, N_ML is the molecule density in one monolayer, and ν_n is the rate constant. In the case of zero order desorption, which describes desorption from multilayers, the desorption energy is equivalent to the heat of evaporation and the desorption rate is coverage independent,

$$R\left(T\right)={\nu }_0{N}_{\mathit{ML}}{e}^{-\frac{E_{\mathit{des}}}{k_{B}T\left(t\right)}}.$$ (3)

Using Eq. (3), a plot of the logarithm of the desorption rate R(T) versus 1/T should yield a straight line, where the slope is equivalent to −E_des/k_B. As outlined above, the exact temperature of the sample surface in our experiments is not directly known, due to the poor heat conductivity between the silicon substrate and the stainless steel heating plate. In order to obtain a reliable value for the desorption temperature, we have conducted experiments with multilayer adsorption and subsequent desorption of indigo directly onto/from the stainless steel sample holder, where the temperature is measured with a thermocouple. A corresponding spectrum (see Figure SI-2 in the supplementary material¹⁷) shows the maximum desorption peak at around 405 K, supporting the assumption made earlier that the peak denoted Ta is due to desorption from the tantalum clamps. Calculations with the corrected temperature values yield a desorption energy of 1.67 ± 0.05 eV, using the 50 Hz curve in Figure 1(b) and the molecule density of indigo in the (100) plane with a value of 3.8 × 10¹⁴ molecules/cm². From the intercept of the linear fit with the y-axis we obtain a frequency factor of 1 × 10²² s⁻¹. By assuming first order desorption we were also able to evaluate the desorption spectra in the monolayer and sub-monolayer regime (α-peak) by using the same temperature correction. In this case ln(N/Θ) was plotted versus 1/T. The evaluation of the 1 Hz spectrum in Figure 1(a) yields a desorption energy E_des of 0.84 ± 0.05 eV and a frequency factor ν of 6 × 10¹² s⁻¹.

At first glance the large difference in the frequency factors for the α and β-peaks seems to be quite astonishing. However, considering the meaning of the pre-exponential factor according to transition state theory (TST), this can easily be interpreted: The pre-exponential factor actually takes the change of all translational and internal degrees of freedom during desorption into account. As a result, this pre-exponential factor is described by³⁴

$$\nu =\frac{k_{B}T}{h}\frac{q^{‡}}{q_{\mathit{ads}}}$$ (4)

with h: Plank’s constant, q_ads: partition function of the adsorbed state, q^‡: partition function of the transition state. In the case of adsorption without an activation barrier for adsorption, the transition state is equal to the final state of the free molecules. Considering a particle, which is already highly mobile in the adsorbed state prior to desorption, i.e., the entropy is nearly the same in the adsorbed state and in the free state, the pre-exponential factor reduces to ν=kT/h ≈ 6 × 10¹² s⁻¹ at room temperature. Thus, one can assume that this is largely fulfilled for desorption from the α-state. On the other hand, desorption from localized adsorption sites, as for example from kink sites of bulk materials, where the partition functions for translation and rotation approach unity, can lead to pre-exponential factors which are orders of magnitudes larger than 10¹³ s⁻¹. This has indeed been frequently observed for desorption from organic multilayer films.^18,35–37

If we want to take a closer look at specific details of the desorption spectra and make arguments about dewetting and the formation of islands as a function of coverage, we need to evaluate the sticking coefficient. For this purpose, one can integrate the area under the individual desorption spectra curves and therefore determine a parameter for the actual amount of adsorbed/desorbed material. This was plotted versus the exposed amount, as measured by the quartz microbalance, in Fig. 2. In both the sub-monolayer and multilayer regime the surface coverage increases linearly with the amount of exposed material, resulting in a constant slope. For organic molecules it is usually assumed that the initial sticking coefficient at and below room temperature is unity.^33,38 Therefore, it stands to reason to assume a sticking coefficient of one for indigo on silicon dioxide. Considering this sticking coefficient, we can convert the exposure values given in Hz into values for the adsorbed amount, described by the mean film thickness: 1 Hz ⩠ 0.8 Å.

FIG. 2.

TDS area for indigo desorbed from carbon covered silicon dioxide as a function of the exposed amount, as measured by the quartz microbalance. Adsorption temperature: 220 K.

B. Adsorption and desorption of indigo on sputter cleaned SiO₂

In order to evaluate the influence of the underlying carbon layer and surface contamination in general, we have performed similar experiments as shown above on a silicon dioxide surface that was cleaned by 10 min of argon sputtering (U=500 V, I_e=30 mA, p_Ar=5 × 10⁻⁵ mbar) prior to each indigo deposition. The cleanliness was always checked by AES. A set of thermal desorption spectra is shown in Figure 3. It is remarkable that for a coverage up to 1.6 Å we observe no desorption peaks at all. Impinging molecules get adsorbed onto the reactive surface and form chemical bonds that are so strong, that no desorption occurs for true surface temperatures of up to 430 K. The existence of some remaining adsorbed material after TDS was checked by AES. After further exposure a single peak (α) at around 450 K (corrected temperature T_corr=307 K) appears, which saturates at a coverage of about 4.8 Å. After that a second peak (β) at higher temperatures starts to grow. Similar to the case of carbon covered SiO₂, the β-peak continues to grow with exposure, whereas the α-peak starts decreasing and disappears completely at, in this case, more than 6.4 Å mean film thickness. Equivalent evaluation of the desorption energies and frequency factors yields E_des=1.53 eV and ν=1.4 × 10²² s⁻¹ for the multilayer, as well as E_des=0.83 eV and ν=1.8 × 10¹² s⁻¹ for the α-peak. The agreement of these values for the multilayer desorption energies and frequency factors between the sputtered and carbon covered substrate is not surprising. However, one would not expect agreement, within the margin of errors, for the α-peaks.

FIG. 3.

Thermal desorption spectra of indigo from sputter cleaned silicon dioxide for different exposures. Adsorption temperature T_ad=220 K, heating rate ß=1 K/s. The exposure is given in Hz, as determined by a quartz microbalance: (a) 2–8 Hz, (b) 6–20 Hz. The temperature correction is described in the supplementary material.¹⁷

The relationship between adsorbed and exposed amount of indigo on the sputter cleaned surface is shown in Figure 4. Above 1.6 Å, the relationship is once again clearly linear and it can therefore be assumed that the sticking coefficient is unity. However, the intersection of the slope with the x-axis does not occur at zero, but rather at a coverage equivalent to 0.8 Å mean film thickness. This explains the small amount of decomposed indigo that stays on the sample after each desorption cycle, as it was described earlier. However, there is another remarkable phenomenon to be observed in Figure 4. Up to about 1.6 Å no desorption takes place at all, but after a minute amount of additionally adsorbed indigo suddenly nearly all the adsorbed material desorbs (besides the remaining 0.8 Å). This suggests that at a particular coverage a rearrangement of some of the already adsorbed molecules has to take place, which leads to less strongly adsorbed particles, which can now desorb more easily.

FIG. 4.

TDS area for indigo desorbed from sputter cleaned silicon dioxide as a function of the exposed amount, as measured by the quartz microbalance. Adsorption temperature: 220 K.

Within our current understanding we can therefore describe the film formation and desorption behavior on the sputter cleaned surface as follows: Initially, single molecules adsorb on the surface and form chemical bonds that proof to be so strong, that no desorption occurs below 800 K (430 K corrected temperature) and no desorption peak is visible in the TD spectra. As soon as a certain coverage threshold of about 2 Å mean thickness is reached, the probability of newly impinging molecules to land on top or in the vicinity of already adsorbed monomers becomes high enough, for a structural re-orientation to take place. We suggest that possibly dimers are formed, where each indigo molecule establishes hydrogen bonds to another indigo molecule. This newly created adsorbate is more weakly bonded to the substrate due to the newly created hydrogen bonds between the active groups in the dimer. Situations where adsorbed molecules rearrange and form dimers with specific H-bonding interactions, thus causing increased diffusion on the surface have been reported by Mitsui et al. and Ranea et al. for water on Pd(111).^39,40 With our experimental capabilities we can of course not present any evidence as to the type of the dimer or multimer formation, but we do suggest a parallel π-stacking configuration where one molecule lies on top of another molecule. Furthermore, one should note that such weakly bonded dimers can only be observed under vacuum conditions with liquid nitrogen cooling. Exposure to room temperature leads to immediate desorption, since the true desorption temperature for the α-adsorption state of indigo lies at around 300 K. At a coverage of 5.6–6.4 Å mean thickness dewetting sets in and island formation takes place, similar to that on the carbon covered silicon oxide surface. The significantly higher saturation coverage of the α-state might be caused by the dimer formation. These dimers, although highly mobile, might not as easily become incorporated in nucleated islands compared to monomers. We will return to this point below, after the description of the island morphologies and island structures.

C. Surface morphology

In order to take a closer look at the morphologies of indigo films and their correlations with the desorption spectra, we used ex-situ atomic force microscopy in tapping mode. In Figure 5(a), an AFM image of an indigo film with a mean film thickness of 40 nm on a sputter cleaned surface is shown. The film consists of large, almost round islands with island heights between 50 and 100 nm, as shown in the cross section (Figure 5(b)). After storage under atmospheric conditions no further dewetting processes or morphology changes (Ostwald ripening) were observable. A comparison of the surface morphology for samples with 0.48 nm, 2.4 nm, and 40 nm mean film thickness, measured with AFM immediately after exposure to air and then again 24 h later, showed no changes in island size, shape, or number. This suggests a stable configuration and complete immobilization of the indigo molecules once they are inherited into the bulk crystal structure, a discovery that is in good agreement with the high heat of evaporation (1.53 eV) found earlier. If we look at an AFM image taken for an indigo film with the same mean film thickness on a carbon covered surface in Figure 5(c), we can see islands with similar shape, but clearly smaller and superior in number. The apparent island heights vary between 10 and 20 nm, as deduced from the cross-section in Figure 5(d). Apparently, in this case a continuous film already exists, so that the actual island height cannot be measured. Similarly, no changes in island size or shape were visible after sample storage under atmospheric pressure for multiple weeks. In consistency with the statements made earlier, we can describe the differences in island size, height, and spacing by an increased indigo dimer diffusion across the reactive silicon oxide surface compared to the monomer diffusion on a carbon covered surface. Spacing between individual islands on a sputter cleaned surface reaches up to a few hundred nanometers, which means that dimers can move very large distances until they finally get incorporated into an island and reach their stable configuration.

FIG. 5.

(a) AFM image (4 μm × 4 μm) of an indigo film on sputter cleaned silicon dioxide, with a mean thickness of about 40 nm. Substrate deposition temperature: 220 K, deposition rate: 0.5 nm/min, (b) Cross section along the black arrow in (a), (c) AFM image (3 μm × 3 μm) of an indigo film on carbon covered silicon dioxide, with a mean thickness of about 40 nm. Substrate deposition temperature: 220 K, deposition rate: 0.5 nm/min, (d) Cross section along the black arrow in (c).

In Figures 6(a) (sputter cleaned) and 6(c) (carbon covered), AFM pictures are shown for indigo films with a mean film thickness of 4.8 Å and 4 Å, respectively. Taking TDS measurement results into account, a mean film thickness of 4.8 Å on a sputter cleaned surface corresponds to the α-state, where the molecules form highly mobile dimers, but desorb already at room temperature. The small amount of material that can be seen is probably the result of some islanding of the α-state during warming to room temperature, prior to venting. However, on a carbon covered SiO₂ surface a mean indigo film thickness of 4 Å already corresponds to the β-state, which consists of stable islands. The cross-section (Figure 6(d)) reveals heights between 5 and 15 nm. Integration over the island volumes provides a quite good agreement with the mean thickness as determined by the quartz microbalance. This corroborates the assumption that in the β-state only negligible desorption takes place during warming to room temperature and venting.

FIG. 6.

(a) AFM image (4 μm × 4 μm) of an indigo film on sputter cleaned silicon dioxide, with a mean thickness of about 4.8 Å. Substrate deposition temperature: 220 K, deposition rate: 0.5 nm/min, (b) Cross section along the black arrow in (a), (c) AFM image (4 μm × 4 μm) of an indigo film on carbon covered silicon dioxide, with a mean thickness of about 4 Å. Substrate deposition temperature: 220 K, deposition rate: 0.5 nm/min. The smeared out islands are an artefact due to the softness of the organic film, (d) Cross section along the black arrow in (c).

D. Structural characterizatione

Two polymorph crystal structures of indigo are known which are denoted as either polymorph A^15,41,42 or polymorph B.^43,44 In both cases the lattice symmetries are the same (monoclinic, space group P2₁/c, two molecules within the unit cell) and even the packing of the molecules is nearly identical. The mass densities are slightly different with 1.50 g/cm³ for polymorph A and 1.46 g/cm³ for polymorph B. The packing of the molecules can be described by piles of stacked molecules where the aromatic planes of the indigo molecules are parallel to each other. In Figure 7, three piles are shown, while only two molecules of a single pile are drawn. It is clearly visible that the molecules of neighboring piles are tilted relative to each other. It is notable, that the distances of intermolecular hydrogen bonds (O… H–N) are significantly shorter (2.17 Å) than the intramolecular ones (2.40 Å). The aromatic planes of stacked molecules are 3.40 Å separated from each other.

FIG. 7.

Package of the molecules in the indigo crystal, together with the crystal unit cell. The Bravais lattice is primitive monoclinic.

In our own experiments the crystallographic properties of two 40 nm thick indigo films on carbon covered and sputter cleaned SiO₂ substrates were determined by X-ray diffraction. A specular scan showed no observable peaks for a film grown on a sputter cleaned surface, hinting at completely random crystal orientation. However, on a carbon contaminated surface we were able to see two reflections that can clearly be distinguished from the noise level at 2Θ angles of about 10.4° and 26.3° (Figure 8). A similar diffraction pattern has also been observed by Irimia-Vladu et al. for 75 nm thick indigo films grown on an aliphatic tetratetracontane surface.⁵ The type of polymorph cannot be unambiguously determined, since calculated peak positions of 100 and 210 of polymorph A as well as −101 and −212 for polymorph with B fit well the observed peak positions (compare Figure 8). Despite different Laue indices for the Bragg peaks observed at 2Θ=10.4° (100 and −101) and for 2Θ=26.3° (210 and −212), the orientation of the molecules relative to the substrate surface would be identical for both phases. In the first case, the piles are aligned parallel to the substrate surface (Figure 9(a)), while in the second case a molecular arrangement with some of the aromatic planes parallel to the substrate surface is observed (Figure 9(b)). Additional pole figure measurements at these angles (see the supplementary material¹⁷) show that the preferred orientations of the crystallites are weakly developed, a variation of ±30° relative to the substrate surface is observed.

FIG. 8.

Differential Θ/2Θ scan for a 40 nm thick indigo film on a carbon covered SiO₂ substrate. The red lines indicate the nearest reflection angles of indigo powder in the α-state (10.74° and 26.58°).⁴²

FIG. 9.

(a) Molecular arrangement of indigo molecules corresponding to the 2Θ peak at 10.4° in Figure 7 viewed from the front (left) and side (right), (b) molecular arrangement corresponding to the 2Θ peak at 26.3° viewed along the front (left) and side (right).

One could speculate as to the reason for the different crystallization behavior of indigo on the sputter cleaned and carbon covered silicon oxide surface, and for the generally weaker tendency of these molecules to crystallize, as compared to rod-like organic molecules, e.g., pentacene^23,24 or hexaphenyl.^16,20–22 We believe that the main reason for this behavior is that the indigo molecules exhibit a chirality in the adsorbed state. Due to the missing mirror symmetry the achiral molecules in the gas phase become two-dimensional chiral after adsorption. As a consequence, when three-dimensional islands have to be formed by diffusion-limited aggregation, only about half of the approaching molecules will have the proper chirality to be incorporated without flipping the molecule. As we can see from the crystal structure (Figure 7) a single indigo crystal is an enantiomerically pure crystal. All molecules within a crystal show the same handedness. Thus, a polycrystalline film will be a racemic conglomerate. It is obvious that in such a case only small crystallites will develop.

The question remains, why no crystallinity can be observed at all for indigo on the sputter cleaned substrate. This might be correlated with the proposed formation of dimers in the adsorbed layer. Of course, we do not know the specific structure of such a dimer; but let us assume that it is a π-stacked dimer with four H-bonds. In this case, the two stacked molecules show different handedness, which makes it impossible to become incorporated into an enantiomeric crystal.

IV. SUMMARY

With the help of thermal desorption spectroscopy, atomic force microscopy, and X-ray diffraction we were able to investigate the initial film formation of indigo on multiple silicon dioxide substrates. A comparison between sputter cleaned and carbon contaminated surfaces led to differences in diffusion behaviour and island growth. In the first case, the substrate is reactive and the indigo molecules are initially very strongly bonded and do not desorb from the surface in a temperature range of up to 430 K. After further adsorption possibly dimers are formed, which de-stabilize the bonding mechanism to the surface and lead to a weakly bonded α-state. Subsequently, at and above a certain coverage, the molecules dewet and form more strongly bonded three-dimensional islands in the so-called β-state. On inert, carbon covered substrates however, initially adsorbed indigo molecules form a similar metastable state, but dewet into islands at much lower coverage. All films with mean film thicknesses corresponding to the β-state were completely stable under atmospheric conditions. Films in the α-state could not be observed with AFM due to immediate desorption at room temperature. The sticking coefficient was found to be coverage independent and unity in all cases. Heat of evaporation calculations yielded desorption energies of 1.67 ± 0.05 eV in the multilayer (β-state) and 0.84 ± 0.05 eV in the monolayer regime (α-state) on a carbon covered and 1.53 ± 0.05 eV (β-state) and 0.83 ± 0.05 eV (α-state) on a sputter cleaned substrate. The frequency factors for desorption were determined to be about 1 × 10²² s⁻¹ (carbon covered) and 1.4 × 10²² s⁻¹ (sputter cleaned) for the ß-state and 6 × 10¹² s⁻¹ (carbon covered) and 1.8 × 10¹² s⁻¹ (sputter cleaned) for the α-state. This hints at a quite localized desorption from the bulk state, but also at a high mobility of the molecules in the adsorbed state prior to desorption. Specular scans of a 40 nm thick film exhibited two weak diffraction peaks for carbon covered samples, which were located at 10.4° and 26.3°. This can be dedicated to a weak crystallographic orientation along the (100) plane and (210) plane, respectively. Films on sputter-cleaned surfaces seem to be completely randomly orientated. The low tendency to crystallize at all can be traced back to the fact that adsorbed indigo molecules show chirality, which might hamper the correct incorporation of the molecules into the 3D crystal. The possible formation of special dimers on the sputter cleaned silicon oxide surface finally can make it even more difficult to form an enantiomeric crystal.