In situ differential reflectance spectroscopy of thin crystalline films of PTCDA on different substrates

We report an investigation of the excitonic properties of thin crystalline films of the archetypal organic semiconductor PTCDA (3,4,9,10-perylenetetracarboxylic dianhydride) grown on poly- and single crystalline surfaces. A sensitive setup capable of measuring the optical properties of ultrathin organic molecular crystals via differential reflectance spectroscopy (DRS) is presented. This tool allows to carry out measurements in situ, i.e., during the actual film growth, and over a wide spectral range, even on single crystalline surfaces with high symmetry or metallic surfaces, where widely used techniques like reflection anisotropy spectroscopy (RAS) or fluorescence excitation spectroscopy fail. The spectra obtained by DRS resemble mainly the absorption of the films if transparent substrates are used, which simplifies the analysis. In the case of mono- to multilayer films of PTCDA on single crystalline muscovite mica(0001) and Au(111) substrates, the formation of the solid state absorption from monomer to dimer and further to crystal-like absorption spectra can be monitored.

Figure 1

Reflectance of the adsorbate covered and bare sample in case of (a) an opaque substrate and (b) a transparent substrate.

Figure 2

(Color online) Simulated DRS of a 2 nm polycrystalline PTCDA film on Au (lower part), on a semi-infinite mica substrate, and a mica sample of a typical thickness used $\left(150\mu \mathrm{m}\right)$ (upper part) in comparison with the absorbance spectra (differential Optical Density, $\Delta \mathrm{OD}$) of that film [optical constants of polycrystalline PTCDA from Ref. 27]. Note the similarity between $\Delta R∕R$ and $\Delta \mathrm{OD}$ in case of the substrate mica: only a small shift of the low energy peak at 2.22 eV of $\sim 8\mathrm{meV}$ occurs.

Figure 3

(Color online) The coefficients $C^{\prime }$ and $C^{″}$ for Au as substrate. Two different spectral regions $E<2.1\mathrm{eV}$ with $C^{\prime }\approx 0$ and $E<2.7\mathrm{eV}$ with $C^{\prime }\approx -2C^{″}$ can be distinguished, representing two limiting cases of the $\Delta R∕R$ signal composition. For comparison, in the case of mica the coefficients have the values of $C^{\prime }\simeq 0.005$ and $C^{″}\simeq 40$ at 2.6 eV, hence the influence of $\mathrm{Re}\left({\widehat{ϵ}}_2\right)$ on the DRS signal can be neglected.

Figure 4

(Color online) (a) Scheme of the UHV growth chamber with the in situ DRS setup. The optical system consists mainly of four lens systems LS1–LS4 and two UV-enhanced Al mirrors M1 and M2. To check for anisotropy, two polarizers $(\mathrm{Pol}.+\mathrm{Anal}.)$ can be inserted. (b) The lens systems LS1, LS2, and LS4 with a front focal length (ffl.) of about 163 mm consist of two commercial ${\mathrm{SiO}}_2$ and ${\mathrm{CaF}}_2$ singlets, while LS3 (not shown here), the system of accumulation, has half the focal length and consists of three singlets. (c) The lowest achievable chromatic aberration of the doublets LS1, LS2, and LS4 in the wavelength range of interest for the optimum lens distances.

Figure 5

(Color online) Views of the PTCDA crystal ($\alpha$ modification): (100) plane (top), (001) plane (right), and (010) plane (bottom). A unit cell and the two types of PTCDA dimers, $A$ (in plane) and $B$ (out of plane, i.e., stacked), are indicated. The unit cell dimensions are $\alpha =3.72\mathrm{\AA },b=11.96\mathrm{\AA }$, and $c=17.34\mathrm{\AA }$. Crystal data were taken from Refs. 31, 32.

Figure 6

(Color online) The differential reflectance spectra of PTCDA on mica during film growth, from 0 ML to a final film thickness of 4.1 ML. During one measurement the increase of the layer thickness is about 0.12 ML, therefore 0.06 ML will read 0–0.12 ML. Note the clear change in the characteristic spectral shape and the two intersection points at about 2.30 and 2.35 eV.

Figure 7

(Color online) (a) The normalized differential reflectance spectra of PTCDA on mica in the thickness range of 0.06–0.8 ML, and the normalized absorbance spectra of PTCDA dissolved in ${\mathrm{CHCl}}_3$ for comparison. (b) Simulated normalized DRS spectra in dependence on the monolayer completion in the described effective medium model with an assumed form factor $f$ of 0.05 which corresponds to distributed PTCDA islands of about 5 nm diameter embedded in vacuum down to isolated molecules with $f\sim 0.1$. With increasing island size, $f$ decreases and the influence on the spectra diminishes with respect to the closed monolayer.

Figure 8

(Color online) $\mathrm{Im}\left(\widehat{ϵ}\right)$ spectra estimated from the differential reflectance spectra of PTCDA on mica, which were recorded with a constant molecular flux of about $0.2\mathrm{ML}∕\min$, from 0 to 12 min (see text, corresponding effective film thickness 0–2.3 ML). Intersections of spectra can be found at about 2.27, 2.41, 2.62, and 2.77 eV, as indicated. For comparison, the $\mathrm{Im}\left(\widehat{ϵ}\right)$ of thick polycrystalline films as reported in Ref. 27, is also shown in the bottom (shifted). We believe that the spectral feature at $\sim 1.87\mathrm{eV}$ in the submonolayer spectra stems from charged molecules, since mica(0001) exhibits a slightly polar surface (Ref. 34). This would imply a transition energy of ionized PTCDA lowered by $\sim 50\mathrm{meV}$ compared to the neutral molecule, and fits well to studies of neutral and ionized perylene (Ref. 45). A similar feature was also reported by Bulovic et al. (Ref. 13) in the absorption of PTCDA dissolved in the strongly polar $N$-methylpyrrolidone (NMP).

Figure 9

(Color online) Experimental integral DRS of the sample from Fig. 6, and simulated integral DRS of PTCDA on mica $\left(190\mu \mathrm{m}\right)$, using the optical constants of polycrystalline PTCDA (Ref. 27). The integration interval was 2–3 eV in correspondence to the position of the $S_0\text{‐}{S}_1$ transition. The demand of equal slopes of the simulated and experimental curve in the higher thickness region allows to refine the final film thickness, being estimated to 4.1 ML. In the inset, the derivative of the experimental integral DRS is plotted. A decreased (observable) oscillator strength per molecule (hypochromism) in the thickness interval of $\sim 1\text{–}2\mathrm{ML}$ is clearly visible.

Figure 10

(Color online) The absorption spectra ($\Delta \mathrm{OD}$, left $y$ axis) of investigated PTCDA on mica films with different final thickness from 0.4 to 5.2 ML. In case of the 5.2 ML sample the $\Delta R∕R$ spectrum is shown for comparison (right $y$ axis). The differential optical density of the films was determined by the difference of the optical density of the sample before and after film deposition. The shape of the $\Delta \mathrm{OD}$ spectra remains constant even when the nominal film thickness becomes as thin as 1 ML.

Figure 11

(Color online) The time dependence of the $\Delta \mathrm{OD}$ spectra of $\sim 0.75\mathrm{ML}$ PTCDA in a desiccated spectrophotometer. The time $t=0$ corresponds to the first measurement after the transfer from the vacuum lock into the spectrophotometer in an exsiccator (the duration is $\sim 3\min$). The spectra were noise filtered by the method of Lee (Ref. 55) $(\text{window size}=5)$. Due to the small thickness of the mica substrate, small interference modulations are visible in the spectra. In measurements after 24 h (not shown here), Peak1 was found redshifted by 17 meV. For comparison, the $\Delta R∕R$ signal of the sample is also shown (gray line, right scale).

Figure 12

(Color online) The shift of the low energy feature Peak1 (uncertainty indicated by error bars) in $\Delta R∕R$ and in absorbance $\left(\Delta \mathrm{OD}\right)$ spectra at different final film thicknesses of PTCDA on mica. In the case of the absorbance, a large shift is already observed at lower film thicknesses $(<4\mathrm{ML})$, due to reordering and clustering of the films in ambient conditions. The dashed curve shows the fit with an on-site-effect model (Ref. 10) ($d^{-1}$ dependence), while the dotted line corresponds to a fit with an simple exciton confinement model ($\sim d^{-2}$-dependence). For details see text.

Figure 13

(Color online) (a) $\mathrm{Im}\left(\widehat{ϵ}\right)$ spectra of PTCDA layers with thickness $N\left(\mathrm{ML}\right)$ obtained by Kramers-Kronig analysis from the $\mathrm{Im}\left(\widehat{n}\right)$ spectra given in the work of Vragovic (Ref. 63). In case of the monolayer, we also assumed a reduced dielectric background for PTCDA when calculating the $\mathrm{Im}\left(\widehat{ϵ}\right)$ spectrum, given by the mica substrate with ${ϵ}_b=2.6$. For the dimer additionally the curve $2\times \mathrm{Im}\left(ϵ\right)$, corresponding to $(\Delta R∕R)∕E$, was plotted as a dashed line. Obviously, no isosbestic behavior would be found in DRS. (b) The monomer dimer transition according to Hennessy et al. (Ref. 67). In contrast to the original work, we used here $\mathrm{Im}\left(\widehat{ϵ}\right)$ obtained from the modeled absorbance spectrum via Kramers-Kronig analysis, and we blueshifted the modeled spectrum by 0.11 eV, to fit the experimental position of Peak1 for the dimer. For the dimer, also $2\times \mathrm{Im}\left(ϵ\right)$ was plotted as dashed line, resulting in isosbestic behavior even in DRS.

Figure 14

(Color online) Simulated differential reflectance spectra of 2 nm PTCDA on $150\mu \mathrm{m}$ mica, using the optical constants of polycrystalline PTCDA (Ref. 27). The PTCDA layer is either solid or composed of a 1 nm solid layer and a 2 nm layer of 50% PTCDA concentration with different shape factors $f$ to simulate increasing surface roughness, see text. The position of the low energy peak is unaffected; it shows an inessential shift of only $\sim 8\mathrm{meV}$ to lower energies.

Figure 15

(Color online) The coverage-dependent differential reflectance spectra of polycrystalline PTCDA on polycrystalline Au, up to a film thickness equivalent to 6 ML. The spectra are characterized by two peaks $A$ and $B$, and can be well described by the known optical constants of pc-PTCDA (Ref. 27), as can be seen from the comparison with a calculated spectrum for 2 nm $(\sim 6\mathrm{ML})$ coverage (dashed line).

Figure 16

(Color online) The coverage dependent differential reflectance spectra of highly ordered PTCDA on Au(111). In addition to the features $A$ and $B$ already present in the polycrystalline films, at 2 ML coverage a new peak $C$ occurs which becomes less pronounced with further increasing film thickness.