Molecular Reorientation during the Initial Growth of Perfluoropentacene on Ag(110).

Perfluoropentacene (PFP) is an organic material that has been widely studied over the last years and has already found applications in organic electronics. However, fundamental physical questions, such as the structural formation and the preferential orientation of the molecules during deposition on metal surfaces, are still not fully understood. In this work, we report on a unique in-plane molecular reorientation during the completion of the first monolayer of PFP on the Ag(110) surface. To characterize the molecular alignment, we have monitored the deposition process in real time using polarization-dependent differential reflectance spectroscopy and reflectance anisotropy spectroscopy. Abrupt changes in the optical signals reveal an intricate sequence of reorientation transitions of the PFP molecules upon monolayer completion and during the formation of the second monolayer, eventually leading to a full alignment of the long molecular axis along the [001] direction of the substrate and an enhanced structural ordering. Scanning tunneling microscopy and low-energy electron diffraction confirm the observed molecular reorientation upon monolayer compression and provide further details on the structural and orientational ordering of the PFP monolayer before and after compression.

Introduction

One of the many advantages of organic thin layers is that their optical and electronic properties can be systematically modified by molecular design. Several properties like dipole orientation, charge injection, optical absorption, etc. depend strongly on the film structure and on the orientation of the molecules with respect to the underlying substrate. It has been shown that the molecular stacking strongly affects the performance of organic devices, such as field effect transistors[1,2] and organic light-emitting diodes.[3,4] During the deposition of the first monolayer of organic molecules onto a substrate surface, the structural arrangement will be governed by the delicate balance between molecule–molecule and molecule–substrate interactions.[5] Therefore, the study of the structural arrangement of thin molecular films on well-ordered substrates is expected to provide important insight into the structure–property relationship of organic layers.

One approach to probe processes occurring at surfaces or interfaces is optical spectroscopy. Because of its nondestructive nature, optical spectroscopy has been widely applied to study optical, electronic, and even structural properties of molecular thin films and their interfaces with inorganic materials.[6−10] If light is used as a probe, surface sensitivity can be substantially enhanced by eliminating the optical response of the bulk. This is commonly achieved by applying differential techniques, such as differential reflectance spectroscopy (DRS), also known as surface differential reflectance spectroscopy, or by reflectance anisotropy spectroscopy (RAS), also known as reflectance difference spectroscopy. While RAS measures the difference of the complex reflectivity coefficients for light polarized along two orthogonal crystallographic directions of a surface,[11] DRS measures the temporal change of the reflectance of a surface upon physical or chemical modification.[12] DRS can be applied on any surface, whereas RAS is restricted to anisotropic samples. The two techniques also differ by their geometrical configuration. While RAS is typically applied very close to normal incidence (θ ≲ 2°), DRS can be acquired close to normal incidence or at oblique incidence (where the angle of incidence θ is significantly different from zero). With DRS, one can also obtain information on the anisotropy of a surface if linearly polarized light is used. Even in the case of an isotropic sample, the Fresnel equations predict a different reflectance for oblique incidence depending on whether the polarization of light is either perpendicular (s) or parallel (p) to the plane of incidence. As a result, off-normal DRS will produce different signals for s- and p-polarization, which can be measured independently in a polarization-dependent DRS (pol-DRS) setup.[13,14]

Several groups have demonstrated that the growth of organic thin films on various substrates can be monitored successfully by DRS.[6,10,15−18] Similarly, single-wavelength kinetic RAS measurements have been used to monitor the epitaxial growth of inorganic semiconductors[19−21] and organic molecules on anisotropic metal surfaces in real time.[7,22−24]

In this work, we report on the evolution of the molecular alignment of perfluoropentacene (PFP) on Ag(110) surfaces during the formation of the first two monolayers (ML). For this purpose, we employ two complementary optical techniques: DRS and RAS. The data obtained from the optical measurements reveal an intricate sequence of reorientation transitions of the molecules upon compression of the first and condensation of the second monolayer. These observations are further corroborated by scanning tunneling microscopy (STM) imaging and low-energy electron diffraction (LEED).

Experiment

Perfluoropentacene (PFP) is the fully fluorinated derivate of pentacene (PEN). It has similar physical dimensions as PEN but an enhanced electron mobility in the thin-film phase of more than 0.2 cm2/V s.[25] The optical properties of PFP molecules in solution[26] and of PFP thin films on different insulating substrates have been studied extensively during the last years by ellipsometry, photoluminescence, absorption, and DRS.[26−30] It was found that PFP molecules tend to grow in an upright orientation on semiconductor and insulating substrates,[26,29] whereas they are flat-lying on metallic substrates.[31,32] From the literature[24] and previous experiments,[26] it is known that the optical absorption of isolated PFP molecules in the UV–vis spectral range is dominated by three main optical transitions: (i) the S0 → S1 transition (highest occupied molecular orbital–lowest unoccupied molecular orbital transition) located at 1.99 eV with a transition dipole moment oriented along the short axis of the molecule (M) and vibronic replicas at 2.16 and 2.33 eV; (ii) the S0 → S3 transition located at 2.72 eV, which is excited by light polarized along the long molecular axis (L) and has a known vibronic replica at 2.87 eV; and (iii) the S0 → S7 transition at 4.25 eV, which is also polarized along the long molecular axis.

The RAS and pol-DRS experiments reported below were carried out in situ in two different ultrahigh vacuum (UHV) chambers with base pressures below 3 × 10–10 mbar. Prior to the deposition of the organic molecules, the two Ag(110) single-crystal surfaces were cleaned by several cycles of Ar+ sputtering (900 V, ≈3.8 μA/cm2) for 30 min and subsequent annealing at 650 K for 5–10 min. The PFP molecules were evaporated from quartz crucibles, which were kept at a constant temperature of 458–468 K during the deposition. The specific temperature set point depends on the evaporator and the targeted deposition rate. During thin-film deposition, the substrates were kept at room temperature.

In the sample configuration used in the present experiments, the RAS signal is given bywhere r[1̅10] and r[001] denote the complex reflection coefficients for light polarized along the two main crystallographic axes in the fcc(110) surface plane, namely [1̅10] and [001], respectively. The measured Δr/r̅ spectrum is directly related to the optical anisotropy (Δϵ = ϵ[1̅10] – ϵ[001]) of the sample with respect to these two orthogonal crystallographic axes of the substrate.[11,33] The measurements were carried out with a home-built RAS instrument. In this setup, the light beam from a xenon lamp is first linearly polarized such that the polarization axis is oriented at an angle of 45° with respect to the main crystallographic axes of the substrate surface, as marked by an orange arrow in the inset of Figure a. The change in the polarization state of the light after reflection from the sample is determined through a combination of a photoelastic modulator operated at a frequency of 50 kHz and a fixed analyzer. The first and second harmonic of the signal are then spectrally resolved and detected using a grating monochromator and a photomultiplier tube, yielding the imaginary and real part of Δr/r̅, respectively.

Figure 1

(a) Real part of the RA spectrum of the pristine Ag(110) surface. Inset: top view of the sample orientation with respect to the polarization vector of the incident light (orange arrow) in the RAS setup. (b) RA spectra acquired during the deposition of 5.5 ML of PFP. Inset: close-up of the energy region between 1.6 and 3.0 eV. (c, d) pol-DR spectra for p- and s-polarized light, respectively, recorded during the deposition of 5.5 ML of PFP. The insets in (c) show the configuration for pol-DRS with the orientation of the s (blue)- and the surface-projected p-polarization (red) components of the incident light, as well as a schematic of a single PFP molecule with its long (L) and short (M) symmetry axes.

On the other hand, the signal measured with DRS is defined aswhere R(E, t) and R(E, 0) are the reflectances of the sample at time t and at a reference time t = 0, respectively, and E = ℏω denotes the photon energy. The pol-DRS measurements were performed in a UHV chamber that allows the simultaneous characterization by photoelectron emission microscopy (PEEM).[10,13] In this chamber, the incoming light beam has an angle of incidence of θ ≈ 65° with respect to the surface normal and the sample is oriented so that the s-polarized and p-polarized light are parallel to the [001] and the [1̅10] direction of the Ag(110) substrate, respectively. A schematic representation of the reflection geometry for the pol-DRS experiments is shown in the inset of Figure c. In our home-built pol-DRS setup, a superquiet Hamamatsu Xe lamp is used for illumination. The s- and p-polarized components of the reflected light are separated by an angle of 60° via a Glan-Thompson polarizing prism and are collected simultaneously by two Ocean Optics STS-VIS spectrometers covering a spectral range from 1.55 to 3.54 eV (800–350 nm).[13]

Besides the in situ optical characterization during growth, the structure and morphology of the deposited PFP thin films were investigated at a sample temperature of 110 K using a variable temperature scanning tunneling microscope and at room temperature using low-energy electron diffraction (LEED).

Results and Discussion

Prior to the deposition, the surface quality of the Ag(110) substrate was checked by RAS. In Figure a, the real part of the RA spectrum for the pristine Ag(110) surface is shown. We observe the characteristic RAS line shape of a clean and flat Ag(110) surface, i.e., a small peak at 1.7 eV (surface state transition) and a strong negative peak at 3.9 eV.[34,35] These features corroborate the high quality of the silver surface.

The RA spectra acquired during the deposition of up to 5.5 monolayers (ML) of PFP are depicted in Figure b. Besides the strong silver resonance around 3.9 eV, distinct spectral features at 1.7, 2.7, 2.85, and 4.2 eV emerge upon deposition, which can be attributed to the absorption of the PFP molecules.[26,27] The inset in Figure b shows a close-up of the RA spectra in the energy range from 1.5 to 3.0 eV for easier comparison with the pol-DR spectra shown in Figure c,d. The negative peak in the RA spectra at 1.7 eV is also observed in the p-polarized DR spectra (p-DRS), whereas the transitions at ≈2.7 and ≈2.85 eV are prominent only in the s-polarized DR spectra (s-DRS). This is perfectly consistent with the definition of the RAS signal in eq and the known orthogonal orientation of the transition dipole moments of the corresponding optical excitations. The PFP-related peak at 1.7 eV in the RAS and p-DR spectra only appears after the completion of the third monolayer. As can be inferred from PEEM experiments,[36] the filling of the third monolayer corresponds to the completion of the wetting layer, after which three-dimensional (3D) nucleation of PFP crystallites sets in. The appearance of the low-energy transition after completion of the wetting layer points to an enhanced oscillator strength of the M-oriented optical excitation, e.g., due to a different stacking of the molecules within the 3D crystallites. Its origin, however, is still under debate.[26,29] On the other hand, the optical transition at ≈2.7 eV and its vibronic replica at ≈2.85 eV, characteristic of flat-lying molecules, develop in the RA and the s-DR spectra throughout the entire growth, albeit changing in intensity and the exact spectral position, as discussed in detail below.

Considering that the optical transition S0 → S3 around 2.7 eV is excited by light polarized along the long axis (L) of the molecule,[31,37] the evolution of this particular transition during deposition provides information on the in-plane alignment of the molecules on the Ag(110) surface. Consequently, we have plotted in Figure a,b the transients of the pol-DRS signals at 2.7 eV together with the RAS transient recorded at the same photon energy but in a different growth experiment in the other UHV chamber. The RAS signal evolves in four steps, characterized by distinct changes in the slope, in perfect correlation with the pol-DRS transients. These reproducible changes in the slopes also allow us to rescale the deposition rates of the two experiments. During the deposition of the first monolayer, the RAS signal increases almost linearly with PFP coverage, manifesting an increase in the optical anisotropy upon deposition of the molecules. Starting from a coverage of 0.7 ML, the RAS and the pol-DRS transients deviate from this linear behavior and undergo a remarkably abrupt zigzag change, which ends with the completion of the first monolayer. Then, the RAS signal increases almost linearly again, with characteristic but smoother changes of the slope upon completion of the second and third monolayer. According to eq , we can infer from the positive slope of the RAS transient and the orientation of the transition dipole moment of the S0 → S3 transition along the long molecular axis (L) that the deposited PFP molecules must be preferentially aligned with their long axis parallel to the [001] direction of the Ag(110) substrate. The smoothly varying slope of the RAS and pol-DRS transients after the completion of the third ML indicates the transition from a layer-by-layer (two-dimensional (2D)) to a 3D island growth mode, i.e., a Stranski–Krastanov growth with a 3 ML thick wetting layer.[7,10,38] This growth scenario is also in accordance with PEEM results.[36]

Figure 2

Transients of the pol-DRS and RAS signals for a photon energy of 2.7 eV: (a) for the entire growth process and (b) for the initial growth up to a nominal coverage of 2 ML.

The two DRS transients shown in Figure a follow a very similar behavior as the RAS signal but with negative sign. This is expected because upon deposition of PFP, the reflectance of the sample (DRS) decreases as the ultrathin PFP layer absorbs the light, whereas the optical anisotropy (RAS) increases with increasing PFP coverage. The fact that the (negative) amplitudes of the p-DRS and s-DRS signals both increase linearly during the initial growth of PFP suggests that the molecules initially arrange in a flat-lying configuration with either an oblique or more than one azimuthal orientation in the surface plane, while the positive slope of the RAS transient reveals that a molecular alignment toward the [001] direction is preferred.

In fact, for the scattering geometry depicted in the inset of Figure c, the (negative) s-polarized DR signal is proportional to the projection cos2(ϕ), if the long molecular axis of a PFP molecule is rotated by an azimuthal angle ϕ against the [001] direction of the substrate, whereas the (negative) p-polarized DRS signal is proportional to sin2(ϕ). However, at off-normal incidence, the relative weight w of the p-polarized and s-polarized DRS signals also depends on the angle of incidence θ, the anisotropic dielectric function of the adlayer, and the substrate dielectric function ϵAg.

In the energy range around 2.7 eV, we can assume that the PFP molecules only absorb light polarized along the long axis of the molecules (L) such that the orthogonal components of the dielectric function along the in-plane (M) and normal (S) axes of the molecules are real valued.[18] Furthermore, the imaginary part of the Ag substrate in this energy range is also much smaller than its (negative) real part (ϵAg(2.7 eV) = – 7.5 + 0.25i)[39] and can thus be neglected in a first-order approximation. Under these conditions, the thin-film approximation of the pol-DRS signals[12] yields w = (ϵAg – 1)/(αϵAg – 1) with α = cos2(θ)/[1 – sin2(θ)/ϵAg] for the ratio between the p-DRS and s-DRS signals. Using θ = 65° and ϵAg(2.7 eV) ≈ −7.5, the corresponding values are α(2.7 eV) ≈ 0.16 and w(2.7 eV) ≈ 3.85. As a result, the p-DRS signal is geometrically enhanced by a factor w ≈ 3.85 with respect to the s-DRS signal and the latter actually provides the dominant contribution to the optical anisotropy. For instance, during the initial stages of growth the measured ratio of the p-DRS and the s-DRS signals in Figure is c ≈ 1.6, yielding an average projection of the molecular axis ⟨cos2(ϕ)⟩ = w/(c + w) of the order of 0.7. However, from the optical signals alone, we cannot tell whether the molecules are actually rotated with respect to the [001] direction of the substrate or whether a minority of 30% of the molecules is aligned along the [1̅10] direction, whereas the majority of the molecules (70%) is aligned along the [001] direction. In fact, the STM and LEED results reported below suggest that both situations can occur in parallel within two different structural phases observed on the surface.

The prominent zigzag feature between 0.7 and 1.0 ML, which is observed in all three transients, is of particular interest. A close-up to this region is depicted in Figure b. The sudden change in the slope of both pol-DRS transients at ≈0.85 ML and the steep variation in opposite directions up to ≈0.92 ML suggests an azimuthal reorientation of the molecules on the Ag(110) substrate. In fact, the reduction of the p-DRS amplitude in this coverage interval is accompanied by a concomitant increase of the amplitude of the s-DRS signal, suggesting that some of the PFP molecules initially aligned with their long axis toward the [1̅10] direction reorient themselves into the [001] direction of the substrate. According to the approximation described just above, the average projection of the long molecular axis onto the [001] direction changes from about ≈0.65 at 0.85 ML to ≈0.8 at 0.92 ML. The anticorrelation of the zigzag feature in the two pol-DRS signals is paralleled by a similar zigzag in the RAS transient, corroborating the sudden realignment of the flat-lying PFP molecules from the [1̅10] toward the [001] orientation within the narrow coverage interval from 0.85 to 0.92 ML.

Surprisingly, this reorientation of the molecules toward the [001] direction seems to be almost completely reverted upon the final compression of the monolayer between 0.92 and 1 ML as evidenced by the steep slopes in the opposite direction in all three optical transients. In this coverage regime, the second layer may already become occupied with molecules preferentially aligned along the [1̅10] direction and thereby enforce a local realignment of the molecules in the layer below. After a nominal coverage of 1.0 ML, the RA signal and both pol-DRS amplitudes again increase linearly until the slope of the p-DRS transient reverses its sign at a coverage of ≈1.3 ML (with an additional kink at ≈1.6 ML), leading to a significant reduction of the p-DRS amplitude upon completion of the second monolayer. At the same time, the (negative) slope of the s-DRS transient increases as well as the (positive) slope of the RAS transient. This behavior is similar to (but less dramatic than) the one in the coverage range between 0.85 and 0.92 ML and can be interpreted by another reorientation transition, in which the PFP molecules in the second layer change their alignment from [1̅10] to [001] upon condensation from a 2D-gas phase into a 2D-solid phase. In fact, similar characteristics in the RAS transients were found for the deposition of pentacene (PEN) on Cu(110)–(2 × 1)O[23] and during the growth of the second layer of PFP on Cu(110).[24] In both cases, these characteristics could be interpreted in terms of an azimuthal reorientation transition upon 2D condensation. When the second layer is finally completed, the ratio of the p-DRS to s-DRS signal has reached a value of c ≈ 0.39, which corresponds to an average projection of the long molecular axis onto the [001] direction ⟨cos2(ϕ)⟩ ≥ 0.9, i.e., an almost complete alignment of the PFP molecules in both the first and the second layer. Therefore, we may conclude that the condensation and the associated alignment along the [001] of the molecules located in the second layer also enforce a final realignment toward the [001] direction of those molecules, which were still preferentially aligned along the [1̅10] after the full compression of the monolayer.

It must be emphasized that the optical transients recorded at a single photon energy of 2.7 eV may also be influenced by spectral shifts of the peaks rather than only by their amplitudes, especially upon compression of the monolayer and the beginning population of the second layer. Therefore, one should explore the full spectral evolution of the DR spectra between subsequent deposition intervals. To this end, we have calculated the differential DRS (DDRS),[10] defined as the normalized difference between the reflectance measured at deposition times t and tThe DDR spectra thus describe the changes in reflectance originating from those molecules deposited during the time interval Δt = t – t as well as the potential changes (over the same time interval Δt) of molecules already adsorbed on the surface. The DDR spectra for deposition of the first two monolayers of PFP for p- and s-polarized light, p-DDRS and s-DDRS, are shown as 2D false color images in Figure a,c, respectively. In addition, DDR spectra for selected coverages, marked with arrows in the 2D plots, are depicted in Figure b,d. Incremental spectral changes are observed in both, the s-DDR and the p-DDR spectra, with absorption peaks located at ≈2.68 and ≈2.83 eV in the monolayer regime, corresponding to the S0 → S3 transition and its vibrational replica, respectively. A small but clear blue shift of the two peaks by about 0.05 eV is observed right after the completion of the first monolayer in both the s-DDR and p-DDR spectra. However, there are also clear differences for the two polarization states, which can directly be related to the alignment and reorientation of the PFP molecules on the surface. Although the incremental changes are always negative in the s-DDR spectra (bluish colors), the p-DDR spectra reveal both negative and positive changes of the main peak at ≈2.68 eV and its vibronic replica at ≈2.83 eV. The largest (negative) amplitude of these two peaks is found in the p-DDR spectra at a coverage of 0.62 ML (red spectrum in Figure d). On the other hand, the (negative) increment of these two features in the s-DDRS is smallest at this coverage (red spectrum in Figure b), indicating that the relative number of PFP molecules with a preferential [001] orientation decreases at this stage of the growth. More importantly, a complete reversal of the sign of the p-DDR spectra, i.e., positive peak amplitudes (yellow and orange colors in Figure c, green line in Figure d), is observed in the p-DDR spectra in the coverage range between 0.85 and 0.92 ML. In the same coverage interval, the s-DDR spectra exhibit the largest (negative) peak amplitudes (dark blue colors in Figure a, green line in Figure b). These spectral changes nicely correlate with the steep and opposite slopes observed in the pol-DRS transients presented in Figure : the reduction of the amplitude of the transient p-DRS signal at 2.7 eV is, indeed, associated with a full reversal of the sign of the p-DDR spectrum. A positive sign of the p-DDR spectrum, however, corresponds to an overall reduction of the absorption from molecules that are at least partially oriented along the [1̅10] direction, even though additional molecules are deposited during the deposition time interval Δt. This can only be explained by the reorientation of already adsorbed PFP molecules from a more [1̅10] oriented alignment toward the [001] direction of the substrate.

Figure 3

Two-dimensional (2D) false color plots of the DDR spectra for (a) s-polarized and (c) p-polarized light. (b, d) Selected DDR spectra marked with arrows in (a) and (c). The spectra are shown with a vertical offset for clarity. The DDRS data were calculated using eq with time step Δt corresponding to a coverage increment of 1/12 ML.

The DDR spectra in Figure clearly reveal that the optical transitions are changing upon compression of the first layer and are blue-shifted by about 0.05 eV upon adsorption of the molecules in the second layer: The main peaks are now located at ≈2.73 and ≈2.88 eV for the S0 → S3 transitions and its vibrational replica, respectively. As a result, a precise evaluation of the intermittent reversal of the orientation of the molecules in the coverage range between 0.9 and 1 ML, as suggested by the optical transients in Figure , is not straightforward.

However, with increasing second-layer coverage, the large amplitude of the s-DDR peaks compared to those in the p-DDR spectrum indicates a clear preference for the PFP molecules to be aligned along the [001] direction. The situation becomes particularly evident in the spectra for 1.67 ML where the p-DDR peaks even show a positive amplitude. As argued above, this sign reversal of the p-DDR spectrum indicates a reorientation of already adsorbed molecules from a [1̅10] oriented alignment toward the [001] direction and might be associated with the 2D condensation of the molecules in the second layer.[23,24] According to the transients in Figure at 2.7 eV, one would assign the onset of the second-layer condensation to occur between 1.3 and 1.6 ML, whereas the p-DDR and s-DDR spectra with maximum positive and negative amplitudes, respectively, are obtained for a coverage of 1.67 ML (see Figure ). In any case, the reorientation after 2D condensation is rather continuous as it proceeds smoothly until the second layer is completed.

To corroborate the conclusions drawn from the optical data, we performed scanning tunneling microscopy (STM) experiments at 110 K and recorded LEED images at room temperature for selected coverages of 0.5, 0.85, and 1.2 ML. The STM images and the corresponding LEED patterns are depicted in Figure . The STM image in Figure a for a coverage of 0.5 ML shows condensed islands of flat-lying PFP molecules but also the bare Ag(110) substrate. Two different structures are recognized in the STM image: the PFP molecules either condense in a checkerboard-like (CB) pattern, so that their long molecular axis coincides with one of the two main crystallographic axes of the substrate, or they form a structure with a rhombic unit cell (R), in which the long axis of the molecules is slightly inclined with respect to the [001] direction of the substrate. Although the STM images obtained at a sample temperature of about 110 K show well-ordered domains with lateral dimensions up to 50 nm, the LEED pattern acquired on the same sample but at room temperature exhibits a diffuse background around the (0,0) diffraction spot. Four broad intensity maxima are observed along the two main crystallographic axes of the substrate with about equal distance to the (0,0) spot. This distance corresponds to the spacing of the molecules in real space in a side-by-side arrangement with their long axis either parallel to the [001] or the [1̅10] direction. The diffuse LEED pattern suggests a lack of a well-ordered, long-range periodic structure at room temperature. The absence of sharp superstructure spots in the LEED pattern in contrast to the well-resolved structural arrangement in the STM image is certainly related to the sample temperature and can be attributed to a thermally induced positional and/or orientational disorder at room temperature. Upon cooling the sample to 110 K, the 2D disordered phase eventually condenses into an ordered phase with checkerboard or rhombic structure. In fact, high-quality STM images could not be obtained at room temperature.

Figure 4

(a)–(c) STM images recorded at 110 K for coverages of: 0.5 ML, approximately 0.85 ML, and 1.2 ML, respectively. (d) 10 x 10 nm2 image from (c). (e )–(g) The corresponding LEED diffraction patterns acquired with an electron energy of 15 eV at room temperature for the different coverages. The white arrows mark the two main crystallographic axes of the Ag(110) substrate. The diffraction spots for the two rotationally equivalent domains in (g) are highlighted with blue and yellow circles for clarity. (h) Schematic representation of the unit cell of the superstructure M of PFP on the Ag(110) surface derived from the LEED pattern in (g).

Figure b shows an STM image obtained recorded increasing the PFP coverage to 0.85 ML. Although the surface is still partially covered with molecules in a checkerboard (CB) or rhombic phase (R), we now find large areas where the PFP molecules assemble into stacks (S) of molecules in a side-by-side arrangement with their long axes parallel to the [001] direction of the substrate. Figure b thus corroborates the partial reorientation of the molecules as inferred from the optical transients (Figure ) and the DDRS spectra (Figure ). The corresponding LEED pattern shows sharper diffraction spots and two additional elongated streaks at half distance along the high symmetry directions of the substrate, the latter being most pronounced along the [001] direction. This is consistent with an arrangement of stacks of molecules pointing in the [001] direction like in the S-phase observed in the STM image at 110 K.

After the completion of the first monolayer, the STM images displayed in Figure c,d reveal two mirror domains (M, M′) of a long-range ordered structure of close-packed molecules, which are exclusively aligned along the [001] direction. From the STM images, we determine the lattice parameters to b1 = (0.9 ± 0.05) nm and b2 = (1.65 ± 0.05) nm. The unit cell is almost (but not exactly) rectangular, and the long unit cell vector b1 is inclined by about 20° with respect to the [001] direction of the substrate. The corresponding LEED pattern shows sharp superstructure spots, which are well reproduced by an epitaxial matrixThe corresponding spot positions and reciprocal unit cells are overlaid on the LEED image in blue and yellow for the two mirror domains M and M′, respectively. The lattice parameters for this superstructure unit cell are (b1 = 0.906 ± 0.015) nm, b2 = (1.618 ± 0.04) nm, ∠(b⃗1,b⃗2) = (86.1 ± 3)°, and ∠(b⃗2,[001]) = (20.9 ± 0.5)°, all in good agreement with the values obtained from STM. A model of the superstructure unit cell and the molecular arrangement in real space is shown in the STM image and the bottom panel of Figure d.

The structural arrangement does not change upon further deposition of PFP up to the three-layer thick wetting layer. On the other hand, a sharp LEED pattern of the monolayer phase, as shown in Figure c, is only obtained for coverages above 1 ML, suggesting that the final ordering (via molecular reorientation) is enforced by the molecules adsorbed in the second layer.

The superstructure of the ordered monolayer of PFP on Ag(110) (eq ) is similar to the structure observed for PFP on Cu(110).[24] In particular, both structures are commensurate along the [1̅10] direction and the molecules are aligned along the [001] direction. A major difference is the head to tail alignment in the case of PFP on Cu(110) as compared to the skew arrangement on Ag(110), which is probably related to different lattice parameters of the two substrates. The present superstructure is also quite similar to the one reported for pentacene (PEN) on Ag(110),[40] denoted as in ref (40) but equivalent to . Compared to eq , however, this superstructure of PEN on Ag(110) is fully commensurate with a perfectly rectangular unit cell.

The optical data combined with the structural information gained from STM and LEED suggest the following growth scenario: at the beginning of the growth, when the substrate–molecule interaction dominates the epitaxial growth, the PFP molecules adsorb at energetically favorable sites on the Ag(110) surface and in a flat-lying configuration in analogy to its nonfluorinated counterpart PEN.[40] Since there appear to be several possible molecular orientations that are energetically almost degenerate, several structures can be observed in the submonolayer regime, namely a rhombic phase (R) and a checkerboard pattern (CB). Although the phases form more or less ordered, coexisting domains at low temperature (110 K) that can be stably imaged by STM, the structures become dynamically disordered at room temperature, as evidenced by the rather diffuse LEED patterns. The optical data reveal an overall anisotropy of ≈70% in favor of an alignment of the long molecular axis along or close to the [001] direction of the substrate, which is consistent with the STM images in the submonolayer coverage regime. As the molecular density on the surface increases, the PFP monolayer is more and more compressed and molecules in the minority orientations reorient themselves, such that their long axis becomes more aligned along the [001] direction of the Ag(110) surface. In this way, the packing density can be increased and the overall structural ordering is also improved. The optical data suggest this first reorientation transition to occur over a narrow coverage range between 0.85 and 0.92 ML. Then, there seems to be an intermittent reversal of this alignment during the final compression of the monolayer in the coverage range between 0.92 and 1.0 ML, most likely promoted by the beginning population of the second layer with molecules residing in a 2D-gas phase and being oriented along the [1̅10] direction. This conclusion is mainly based on the optical transients recorded at room temperature, and no direct evidence can be found in the STM images (recorded at 110 K). Yet, the LEED patterns (recorded at room temperature) confirm that at this stage the monolayer is considerably disordered. Only after further deposition and likely promoted by the onset of 2D condensation of the molecules in the second layer, there is a final reorientation transition that eventually leads to an almost complete orientation of the molecules along the [001] direction. This final monolayer structure (M) is well ordered and can be described by the epitaxial matrix in eq . Because of the symmetry of the superstructure unit cell, the PFP molecules in the first monolayer form mirror domains (M, M′), in which the lattice vectors are tilted in opposite directions with respect to the main crystallographic axes of the substrate, but all of the molecules in either domain are exclusively aligned along the [001] direction of the substrate.

Conclusions

We have explored the adsorption and growth of PFP on Ag(110) by real-time monitoring using two complementary optical techniques, namely pol-DRS and RAS. We have shown that by analyzing the optical transients and the incremental spectral changes (DDRS), it is possible to obtain detailed information on the growth mode and on the structural and orientational arrangement of the molecules on the surface. Our results indicate that the deposition of PFP on Ag(110) surfaces follows a Stranski–Krastanov growth mode with a 3 ML thick wetting layer formed by flat-lying molecules predominantly aligned with their long axis along the [001] crystallographic axis of the substrate. We find a peculiar sequence of three distinct molecular reorientation transitions during the deposition of the first two monolayers. These three stages have been associated with (i) the 2D compression of the monolayer favoring a uniaxial alignment along the [001] direction, (ii) an intermittent disorder promoted by the beginning population of the second 2D molecular gas phase, in which the molecules are preferentially oriented along the [1̅10] direction, and (iii) the reorientation of the molecules in the second layer along the [001] upon 2D condensation. The PFP adlayer thus undergoes a transition from a dynamically disordered submonolayer phase (at room temperature) to a highly ordered close-packed layer of well-aligned PFP molecules. These results are further corroborated by STM and LEED experiments. Our results demonstrate that polarization-resolved, optical techniques like RAS and pol-DRS are quite sensitive to the orientational order parameter and, hence, to phase transitions, which involve the reorientation of molecules. Being real-time, in situ techniques, they also allow to explore the kinetics of such phase transitions during thin-film growth.