Structure and Growth of Thin Films of Aniline on Silver: Nucleation and

23424

J. Phys. Chem. B 2006, 110, 23424-23432

Structure and Growth of Thin Films of Aniline on Silver: Nucleation and Premelting of Nanocrystallites, Porosity, and Crystallization Grazia Gonella, Minchul Yang,† Susan M. Dounce,‡ and Hai-Lung Dai* Department of Chemistry, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6323 ReceiVed: June 19, 2006; In Final Form: August 28, 2006

The structure and growth of thin films of aniline vapor deposited on Ag(111) and Ag(110) surfaces have been examined using optical second harmonic generation (SHG) and linear optical differential reflectivity (DR). Aniline thin films deposited at 90 K give a detectable SH signal that arises from small polycrystallites with orientation anisotropy in the film. Upon annealing, the SH signal decreases, first due to premelting (at ∼145 K) of the polycrystallites and then sublimation (at ∼180 K) of the film. Quantitative analysis of the SH intensity change by a premelting model [J. Phys. Chem. 1988, 92, 7241] allows the determination of the average size of the crystallites as 1.1 nm in diameter and containing ∼45 aniline molecules. The existence of the nanocrystalline structure and its premelting are confirmed by DR experiments. The DR signal around 145 K exhibits change corresponding to an order-disorder transition. Quantitative analysis of the DR data results in the same nanocrystallite size. Experimental observations indicate that films deposited at 90 K contain not only nanocrystallites but also ∼30% porosity, which can be reduced by annealing. At temperatures above 195 K, micron-size crystallites start to form within the amorphous film, causing a large amount of light scattering while the film sublimates. It appears that, for molecules such as aniline with stronger intermolecular interactions, more enthalpy is released, upon adsorption to the local surrounding molecules, causing them to reorient into crystalline form. The low deposition temperature, on the other hand, prevents diffusion for further crystallization beyond nanocrystallites. The refractive index of the amorphous aniline solid can be determined as 1.68 ( 0.03.

1. Introduction The structure of a molecular thin film is determined by properties intrinsic to the constituent molecules and the supporting substrate such as intermolecular and molecule-substrate interactions as well as by deposition and growth conditions such as substrate structure and temperature, vapor pressure, growth rate, annealing, and the presence of impurities. How each of these factors affects the structure of the deposited molecular thin film has been of fundamental interest in the fields of molecular thin films and supercooled fluids. Furthermore, recently, a great deal of attention has been paid to the understanding as well as to the control of the structure of films made of aromatic organic molecules that have unique electrical properties. The specific structure of an organic thin film is crucial to charge transport properties in its application in electronic devices. It has been established that carrier charge mobilities in organic field effect transistors change dramatically with molecular ordering and the existence of grain boundaries in the organic films.1-3 It is therefore important to be able to probe the structure and phase transitions of organic thin films and understand how the film structure is affected by the molecular characteristics and growth conditions. To understand the film structure and growth mechanism, both thermodynamic and kinetic aspects of the molecular adsorption process need to be considered. For instance, it is normally considered that the adsorption of molecules on a cold substrate * To whom correspondence should be addressed. E-mail: dai@ sas.upenn.edu. † Present address: Naval Research Laboratory, Washington, DC 20375. ‡ Present address: W. L. Gore & Associates, Inc., Elkton, MD 219221320.

leads to the formation of a glassy or amorphous film, although a crystalline structure is more favored thermodynamically. This is because molecules at low temperatures are kinetically arrested with frozen translational and/or rotational motion. The situation can be conceptually described by the energy landscape picture.4,5 For an amorphous phase, there are a large number of local minima on the free energy surface, while there is only one global minimum representing a bulk crystalline structure. Transformation into the bulk crystalline phase is an entropically unfavorable process with a potentially substantial barrier. On the other hand, the possibilities of smaller, localized crystalline structures can facilitate many local minima on the free energy surface. Their formation is thermodynamically metastable, but feasible. The size and structure of the crystallites that can be generated depends on the molecular shape, intermolecular interactions, and the deposition process. Once a specific film structure, corresponding to a local minimum on the free energy surface, is formed during the deposition process, it can be transformed into a different one through annealing. The new structure could be the thermodynamically stable crystalline bulk or another metastable structure, depending on the annealing process. How the postdeposition annealing process affects film structure is another important question to be answered. Considering the complexity in the structure in which a molecular film can be grown and the many factors that may influence the structure, it is desirable that films to be studied are prepared under well-defined conditions (temperature, vapor pressure, substrate structure, etc.) such as those attainable in UHV. Films prepared in such an environment can then be characterized by structural probes with real-time resolution for understanding the kinetics of growth mechanism and structural

10.1021/jp0638080 CCC: $33.50 © 2006 American Chemical Society Published on Web 10/25/2006

Structure and Growth of Thin Aniline Films on Ag transition. In this study, we utilize optical second harmonic generation (SHG) and differential reflectivity (DR) to probe in real time the structure and phase transition of organic films during the film growth and annealing in UHV conditions. In previous studies we have shown that the nonlinear optical SHG technique, with unique sensitivity to the symmetry properties of the films, can be used to characterize the growth rate,6 annealing, structure, and phase transition of thin films of pyridine, a molecule with relatively weak intermolecular interactions.7,8 Linear optical reflectivity, on the other hand, allows measurement of changes in refractive index that can be related to phase transition and light scattering due to the existence of heterogeneous structures.9,10 The study on thin pyridine films has revealed that films deposited at low temperatures are amorphous.7 Upon annealing, domains of micron-size polycrystallites are formed through a cylindrical growth mechanism. Further increase of temperature causes premelting of the polycrystallites into amorphous solid at temperatures much lower than the bulk melting point and the film sublimation temperature. In this report, we present the study on films composed of aniline molecules that have tendencies to form stronger intermolecular interactions compared to pyridine. For aniline, in addition to van der Waals and dipole-dipole interactions, hydrogen bonding is possible. The stronger intermolecular interactions supply more enthalpy released upon condensation of the gaseous molecules into the film. Consequently, as we will show, the vapor-deposited aniline films at low temperature (90 K) consist of nanoscale crystalline clusters instead of being amorphous. The nanocrystallites in the aniline film undergo a premelting process at temperatures much lower than the bulk melting point and below the film desorption temperature. Crystallization into larger size crystallites in the film is also observed at a temperature close to 195 K. It is important to note that the thin films under consideration here are made of molecules in monomer form with a uniform structure and are not in the polymeric form. In the case of aniline, it is known that polyaniline, known as PANI, thin films can be present in amorphous and polycrystalline structures depending on the growth condition and parameters.11,12 Dissociation of aniline deposited on more reactive metal surfaces has been observed: Pd(110),13 Pt(111),14 and Cu(110).15 On the (100) and (111) surfaces of Ni with a partially filled d-level, aniline was reported to polymerize.16,17 In the case of the more inert Ag surfaces with a completely filled d-level, we have found, through temperature programmed desorption and electron energy loss spectroscopy studies, no dissociation or polymerization for the first monolayer and subsequent multilayers deposited at low temperatures.18,19 II. Experimental Section All experiments were performed in an UHV chamber with a base pressure lower than 2 × 10-10 Torr. The silver substrate was cleaned routinely by several cycles of sputtering and annealing before each experiment. Aniline (>99.5%, Aldrich) was purified by several freeze-pump-thaw cycles before use. Aniline was deposited on the clean (111) or (110) surface for SHG or DR experiments, respectively, using a leak valve. During the deposition, the pressure of the molecules was kept at 5.00 ( 0.05 × 10-7 Torr (uncorrected for the ion-gauge sensitivity factor), which roughly corresponds to 1 × 1014 molecules cm-2 s-1 in flux. The sample temperature was maintained within (0.1 K during the deposition and optical measurements. The effect of the two different crystallographic orientations of the substrate on the film growth may result in

J. Phys. Chem. B, Vol. 110, No. 46, 2006 23425

Figure 1. SH intensity change during deposition of (a) pyridine at 90 K, and aniline at (b) 155 K and (c) 90 K. Solid lines in (a) and (b) represent the best fit to the three-media model.

different interfacial layer structures, as reported for pyridine,8,20 but we do not anticipate any difference in the bulk structure for the thick (∼100 nm) films we deal with in this report. For the SHG experiments, the 532 nm output from a Nd: YAG laser (Continuum 580A, 8 ns pulse length, 15 mJ/pulse, 4.5 mm diameter in beam size, linearly polarized with 60° incident angle) was used as the fundamental light. The SH light generated at the silver surface, propagating along the fundamental beam direction and exiting the aniline film, was passed through a filter/monochromator assembly and detected by a photomultiplier. Detailed arrangement of the nonlinear optical experimental setup can be found elsewhere. 20 For the DR experiments, the 632.8 nm linearly polarized output of a He-Ne laser, mechanically chopped at 447 Hz, was used as light source. The laser beam, after passing through a polarizer, was directed to the Ag(110) surface in the UHV chamber with an incident angle of 60° relative to the surface normal. The reflected beam is passed through a polarizing beamsplitter for the separation into the p- and s-components. The intensities of the p- and s-components are detected by photodiodes separately and subtracted from each other with a differential amplifier and processed by a lock-in amplifier. The DR experiments began with a clean surface maintained at the deposition temperature and with the nulling of the DR signal done by adjusting the angle of the polarizer placed before the chamber. A detailed description of the experimental setup has been reported previously.21,22 III. Results and Analysis A. SHG during Film Deposition. The SH intensity as a function of exposure time recorded during aniline film deposition on Ag(111) at two temperatures, 90 and 155 K, is shown in Figure 1. For comparison, the SH intensity recorded during deposition of a pyridine film at 90 K is also shown. As previously reported,6 pyridine films grow on Ag(111) as an optically flat, amorphous solid at 90 K. The SH intensity, due to optical interference of fundamental and SH lights within the growing film, displays an oscillatory pattern as a function of film thickness. For the pyridine film in Figure 1a, the second

23426 J. Phys. Chem. B, Vol. 110, No. 46, 2006 SH peak intensity is weaker than the first one. This is primarily because absorption of 266 nm light by the n-π* electronic transition in pyridine23 centered around 4.5 eV causes the overall SH intensity to decrease exponentially with increasing film thickness. Light scattering due to inhomogeneity of the film may also contribute to this decrease. A similar pattern in SH intensity is observed for aniline films grown at 155 K. The aniline molecule has a similar absorption band around 4.5 eV due to a π-π* electronic transition.7 Solid lines in Figure 1a and b are fits to a model describing the optical interference resulted from multiple reflections of the lights at the boundary of the metal-film and film-vacuum interfaces.6 The excellent fit in Figure 1b to the optical interference model is consistent with the conjecture that at 155 K aniline films grow on Ag (111) as an amorphous solid. The SH intensity pattern in Figure 1c for an aniline film deposited at 90 K is quite different from the other two. First of all, the amplitude of the first peak in Figure 1c is much stronger than those for pyridine film at 90 K (Figure 1a) and the aniline film at 155 K (Figure 1b). The amplitude at the first peak maximum in (c) is 30 times higher than that in (a) and 7 times higher than that in (b). In addition, despite the fact that more light is absorbed within the film through the aniline π-π* electronic transition, the second peak intensity in Figure 1c is stronger than the first. One possible reason for these observations is that the aniline film at 90 K has an ordered structure that generates SH light, providing an additional source of SHG than the Ag surface. In this scenario, the detected SHG should increase with the square of the film thickness when the film is thinner than the optical coherence length, which is on the order of the SH wavelength. On the other hand, for incoherent SHG when the film becomes thicker, the intensity will oscillate with the film thickness due to interference. Overall, the competition between absorption and generation of SH light results in a slight increase of height in the subsequent peak in Figure 1c. A quantitative analysis of the SHG intensity pattern of aniline film at 90 K requires treatment of SH generated in both the film and metal substrate. Moreover, multiple reflection of both fundamental and SH lights as well as absorption of SHG in the film should be considered. Even without this complex treatment, previous works in the literature20,24,25 support our interpretation for the SHG intensity from the aniline film at 90 K. Theory and experiments for SHG in reflection from two nonlinear slabs, ZnSe/GaAs(001), showed oscillating patterns and slowly increasing subsequent peak intensity with increasing thickness.24 Similar patterns were also reported for the system of C60 on fused-quartz25 and crystalline pyridine on Ag(111).20 B. Temperature-Dependent SH Intensity. The SH intensity from a 64 nm thick aniline film, initially deposited on Ag(111) at 90 K, during annealing with a temperature increase rate of 0.05 K/s, is shown as the dotted line in Figure 2. The SH intensity starts to decrease at about 130 K. The decrease is accelerated around 145 K and slows down after 160 K. The SH intensity reaches the bottom and remains at the same level after 196 K. The lowest SH intensity is the same as the SH intensity detected from a clean Ag(111) at about 192 K. Temperature programmed desorption studies of aniline films show that aniline multilayers start desorbing at 160 K, and the desorption rate is maximum at 188 K.18,19 Therefore, we assign the slow decay after 160 K and the sharper decrease around 185 K as due to the loss of the film through desorption. The dramatic decrease of SH intensity from 130 to 160 K, on the other hand, occurs when the film thickness remains unchanged. This decrease must arise from structural changes occurring

Gonella et al.

Figure 2. The change of the SH intensity from a 64 nm thick aniline film initially deposited at 90 K upon temperature increase. The solid line represents the fitting results.

Figure 3. DR signal change during deposition of an aniline film at 90 K. The solid line is the best fit using the three-media model.

inside the aniline film. Since SHG occurs only from ordered structures, it is reasonable to suggest at this point that this ordered structure is disappearing in this temperature range. C. Linear Differential Reflectivity Change during Film Deposition. Figure 3 shows the variation of the DR signal during the deposition of an aniline film on Ag(110) at 90 K. The duration of the deposition reported in Figure 3 represents one period of the optical interference pattern that resulted from interference among He-Ne laser beams reflected at the filmvacuum and film-substrate interfaces. At the low dosing vapor pressure level of 5.00 ( 0.05 × 10-7 Torr, the pressure detected by the ion gauge and the pressure at the crystal surface are not in equilibrium due to diffusion. This delay in pressure equilibrium does not allow accurate recording of the exposure rate during the initial moments of the film deposition. The initial exposure plotted in Figure 3 is therefore not accurate. For this reason, the deposition curve has not been subjected to analysis from the beginning. In the analysis, a delay time is artificially added to mimic the realization of the ion gauge recorded exposure at the crystal surface. The DR interference pattern can be analyzed using the threemedia (vacuum-film-Ag) model that has been reported previously.21,22 Briefly, with the optical setup described above, the measured DR signal, the relative change in reflected p- and

Structure and Growth of Thin Aniline Films on Ag

J. Phys. Chem. B, Vol. 110, No. 46, 2006 23427

Figure 5. DR data measured for 95 nm thick aniline films during temperature treatments. The four curves in the middle are from films that are originally deposited at 90 K, followed by heating to 144, 150, 160, and 170 K, respectively, then cooling to 90 K, and finally, heated to 220 K. The beginning of the process and the direction of the temperature change for each curve are indicated by an arrow. Figure 4. DR change measured during the heating of a 95 nm thick aniline film originally deposited at 90 K. The continuous line is the best model fit to the data.

s-polarized light through the film with thickness d, can be expressed as

∆R(d) ∆Rp(d) ∆Rs(d) ) R Rp Rs

(1)

∆Rp(d) Rp(d) - Rp(0) |rp123|2 - |rp13|2 ) ) Rp Rp(0) |r 13|2 p

∆Rs(d) Rs(d) - Rs(0) |rs | - |rs13|2 ) ) Rs Rs(0) |rs13|2 123 2

R(0) represents the reflectivity of the bare substrate and r13 and r123 are the Fresnel reflection coefficients for the vacuumAg and vacuum-film-Ag systems, respectively.26,27 The complex indices of refraction used in the three-media model are n˜ 1 ) n1 ) 1 for vacuum, n˜ 2 ) n2 for a nonabsorbing film, and n˜ 3 ) 0.27 + i4.18 for the Ag substrate at the He-Ne laser wavelength.28 From the fit of this data with eq 1, we obtain for the aniline film a growth rate of 0.47 ( 0.01 Å/s and a refractive index of 1.450 ( 0.005. If in the fitting procedure, a slightly different refractive index value for the Ag substrate, such as the ones reported by Palik,29 Hass,30 or Born and Wolf27 is used, very little change in growth rate and refractive index of the aniline film is observed. D. Temperature-Dependent Differential Reflectivity Change. Figure 4 shows the DR change measured for a 95 nm thick aniline film, initially deposited on Ag(110) at 90 K, during its annealing with a heating rate of 0.05 K/s. This thickness corresponds to roughly half a period of the interference pattern reported in Figure 3 (∼2000 s). This thickness was chosen because this is the thinnest film for which the DR signal has sufficient sensitivity in revealing any change in the film. Overall, the DR signal shows a monotonic increase with temperature up to 180 K. Between 130 and 160 K, the DR signal displays

a shallow trough, and then at 180 K, a faster change due to desorption of the film. The DR curve for temperature above 180 K follows essentially the reverse of the curve observed during film deposition, with the exception of the appearance of a sharply rising feature around 200 K. Eventually, at 205 K, when the desorption is over, the signal becomes constant. The small dip between 130 and 160 K in DR appears in coincidence with the sharp SH intensity decrease in the same temperature region and may arise from the proposed orderdisorder structure change inside the aniline film. The DR signal is further examined with a deposition/annealing sequence to reveal the reversibility of the process. The results with annealing at temperatures between 144 and 170 K are shown in Figure 5.The DR curves presented are obtained for films initially deposited at 90 K, followed by heating to 144, 150, 160, or 170 K, respectively, cooling back down to 90 K, and heating again to 220 K. For comparison, the DR curve of a film deposited at 90 K and then directly heated to 220 K, as well as one from a film initially deposited at 155 K, cooled to 90 K, and then heated to 220 K, are also shown. All films were dosed at 95 nm thickness. The heating and cooling rates used were + and - 0.05 K/s, respectively. For the film deposited at 155 K and immediately cooled to 90 K, the DR signal increases during the cooling and then stays flat during the subsequent temperature increase. Initial deposition at this higher temperature does not seem to create a film structure that would undergo the transition in the 130-160 K range. For films deposited at 90 K and subjected to annealing (to a designated temperature), cooling, and reheating cycle, it seems that the DR curve behavior, during the last part of the temperature increase, depends on whether the initial annealing has been completed. If, during the initial temperature rise, the transition is interrupted before its completion, as happens when the film is cooled after reaching the temperature of 144 or 150 K (Figure 5), during the subsequent heating of the film, the transition appears to continue from the point at which it was interrupted. Another characteristic in the DR curves is that, during the cooling cycle, the signal also increases. This increase may have the same origin as the continuous increase of the

23428 J. Phys. Chem. B, Vol. 110, No. 46, 2006 signal during the heating process. For the film annealed at 170 K, the transition is complete. In this case, the cooling still produces a slight increase in the signal, but during the following heating process, the signal appears unchanged as if the final stable phase has been reached. IV. Discussion A. Premelting of Nanocrystallites in Thin Aniline Films. The logical interpretation of the observations in Figure 1 is that aniline films deposited at 90 K have an ordered structure that facilitates SHG, whereas those deposited at 155 K, like the pyridine films deposited at 90 K, have amorphous structure that does not generate a SH signal. The aniline films with ordered structure undergo an order-disorder phase transition, or melting, at temperatures close to 145 K but below the onset of desorption at 160 K. Once the film structure becomes amorphous, the SH intensity decreases, as indicated by the sharp decrease starting at 130 K. Such a change in the SH intensity should not arise from a thickness change due to thermal expansion when the temperature is raised above 130 K. Ample examples exist for assessing the thermal expansion of the film thickness in this temperature range. X-ray diffraction31 for an aniline solid (monoclinic) showed that lattice parameters change much less than 1% between 140 and 200 K. MD calculations32 for a pyridine crystal (orthorhombic) showed that the solid volume change is of ∼3% between 159 and 245 K. A similar amount of change in lattice dimensions is expected between 80 and 160 K. Neutron diffraction33 for a deuterated benzene solid (orthorhombic) showed that the solid volume changes less than 5% between 100 and 200 K. An estimation of the SH intensity change due to thermal expansion can be made using the relation between the SH intensity and film thickness (or deposition time) in Figure 1b or c. The SH intensity change caused by the change in film thickness during thermal expansion is estimated to be less than 6% if no significant change in refractive index is assumed. Another consideration is the observation that the SH intensity always decreases at higher temperatures regardless of the film thickness. If thermal expansion at higher temperatures is the reason for significant changes in the SH intensity, the SH intensity should increase (or decrease) at higher temperatures for films thinner (or thicker) than 130 nm at which the SH intensity reaches the apex of the first peak in Figure 1. The change of the silver surface temperature makes a minor contribution to the change in the SH intensity from the filmsilver system. In general, SH intensity from metal surfaces decreases with temperature due to increased electron-phonon scattering. We have observed that the SH intensity from the clean Ag(110) decreased by less than 10% from 90 to 140 K.34 By comparing parts b and c of Figure 1, we judge the Ag surface contribution to the overall SH intensity from the film-Ag system is only about 1/7; this effect is negligible in accounting for the sharp SH intensity decrease between 130 and 160 K. The coincidence of the temperature (130-160 K) of the small dip in the DR curve with that of the change in SH intensity suggests the same origin, the order-disorder transition, for the two phenomena. Order-disorder transition is a first-order phase transition and should have a well-defined transition temperature, whereas the “transition” we observe in both the linear and nonlinear optical responses occurs over a broad range of temperature, as seen in Figures 2 and 4. The transition temperature, ∼150 K, is much lower than the aniline bulk melting point of 267 K. These characteristics, a broad transition in a temperature range much lower than the melting point, is indicative of premelting of crystallites. This phenomenon occurs

Gonella et al. at the surface of crystalline particles. Because of the surface tension, the ordered molecules at the surface region would melt at temperatures lower than the bulk melting point. During the melting process, the dimension of the particle is changing, therefore broadening the transition.35,36 As the surface-to-bulk ratio becomes larger for smaller size particles, the crystalline structure at the particle surface becomes less stable due to a larger contribution of surface energy to free energy. Smaller particles may thus undergo melting at lower temperatures than the bulk melting point. The premelting of the crystallites and its effect on SHG can be quantitatively described. We use a two-state model in which the population ratio of the ordered and disordered states is related to their free energy difference. The number density ratio between ordered phase, No, and disordered phase, Nd, is governed by the free energy difference, ∆G, between the two phases:

No/Nd ) exp(-∆G/kBT)

(2)

∆G as a function of temperature of this order-disorder transition with enthalpy ∆H has been related35 to the average diameter R of the crystallites in the ordered phase and the surface tension σ as:

∆G ) -

4πR3∆H(To - T) + 4πR2NAσ 3VTo

(3)

where To is the bulk melting temperature (267 K for aniline), and V is the molecular volume in the crystallite. It is logical to assume that the SH intensity generated from the aniline film I2ω is proportional to the square of the fraction of the ordered phase

I2ω ) [No/(No+Nd)]2

(4)

After substitution of eqs 2 and 3 into eq 4 and adding in the contribution of the SH intensity from the bare silver surface, IAg, the detected SH intensity can be expressed in the form of

I2ω ) Io[1 + e-A+(B/T)]-2 + IAg

(5)

where A ) 4πR3∆H/(3υkBTo), B ) [4πR3∆H/(3υkB) - 4πR2σNA/ kB], and Io is a proportionality constant. The value of V, 1.31 × 10-22 cm3, for aniline bulk crystal at 150 K, is used for the calculation.31 We adapt a relation from literature for surface tension,37-39 σ ) κT∆H/υ2/3 where κT is constant and equal to 0.32 for organic molecules and 0.4 for metallic films.38,39 In this model, there remain a total of two independent variables, R and ∆H (or σ). They can be determined from a fitting of the temperature dependence of the SHG intensity in Figure 2 using eq 5. Figure 2 shows the excellent fitting result (solid line) to the experimental data (dotted line). The fitting results for a series of films with different thickness between 44 and 152 nm and for different heating rates are shown in Table 1. All the fittings give a narrow distribution for the number of molecules in the ordered-phase crystallite, from 40 to 47. Furthermore, it appears that the size of the ordered-phase cluster is independent of the heating rate in the range of 0.05-1 K/s. In Table 1, the calculated value for ∆H, 0.9-1.8 kJ/mol, is smaller than that for aniline solid bulk (10.5 kJ/mol). The reduction of ∆H for small particle was observed in many systems such as ice,40 organic molecules in porous solids,41 and small gold particles.42 The same model can be applied to account for the DR data where the change in reflectivity is related to change in the linear

Structure and Growth of Thin Aniline Films on Ag

J. Phys. Chem. B, Vol. 110, No. 46, 2006 23429

TABLE 1: Film Thickness, Fitting Parameters (A and B), Enthalpy for Order-Disorder Transition (∆H), Surface Tension (σ), and the Number of Molecules Per Crystallite (N) as Determined from Model Fits of the SHG and DR Data; The Heating Rate (β) is also Listed thickness (nm)

β (K/s)

A

B (× 103 K)

∆H (kJ/mol)

σ (mJ/m2)

N

41a 56a 64a 64a,b 64a,b 95c 107a 119a

0.05 0.05 0.05 0.5 1 0.05 0.05 0.05

28.3 ( 2.4 26.8 ( 1.8 29.2 ( 0.4 22.4 ( 3.1 36.0 ( 5.0 27.4 ( 0.3 19.7 ( 0.6 22.8 ( 1.5

4.32 ( 0.36 3.96 ( 0.26 4.39 ( 0.06 3.46 ( 0.46 5.39 ( 0.80 3.91 ( 0.03 3.00 ( 0.09 3.33 ( 0.20

1.3 ( 0.7 1.4 ( 0.6 1.4 ( 0.1 0.8 ( 0.8 1.8 ( 1.5 1.6 ( 0.1 0.90 ( 0.17 1.2 ( 0.5

2.7 ( 1.5 2.8 ( 1.1 2.9 ( 0.2 1.7 ( 1.7 3.6 ( 3.2 3.3 ( 0.2 1.9 ( 0.4 2.5 ( 0.9

46 ( 23 41 ( 14 44 ( 3 56 ( 49 44 ( 34 37 ( 2 47 ( 8 40 ( 13

a SHG data. b The bigger error in these two cases is due to the smaller S/N ratio in the data acquired using higher heating rates. c DR data. Also obtained from the fit are the polarizability of the disordered phase Rd ) 1.070 ( 0.010 and the thermal coefficient of the refractive index ν ) 3.77 ( 0.01 × 10-4 K-1.

optical properties of the film induced by temperature-dependent structural changes. The refractive index n2 of the film is related to the dielectric constant 2, which in turn is related to the first-

order susceptibility χ(1) as: n2(T) ) x2(T) ) x1+χ(1)(T). The macroscopic susceptibility can be phenomenologically described by a linear combination of the microscopic polarizabilities of the ordered (Ro) and disordered (Rd) phases as:

χ(1)(T) ) fo(T)Ro + fd(T)Rd fo(T) )

No(T) Nd(T) and fd(T) ) NTOT NTOT

(6) (7)

Thus, the refractive index of the film partially in ordered and disordered phases can be expressed as a function of temperature as:

n2(T) )

x ( )( 1+

No +1 Nd

-1

)

No R o + Rd ) Nd

x1 + (e-A+(B/T) + 1)-1(Ro e-A+(B/T) + Rd)

(8)

To describe the temperature dependence of the DR data in Figure 4, we add another term that corresponds to the continuous, linear change of the baseline with temperature which can be related to a reduction of the film porosity (see Subsections B and C). The final form of n2(T) is

n2(T) )

x1 + (e-A+(B/T) + 1)-1(Ro e-A+(B/T) + Rd) +

ν × (T - 90 K) (9)

Here, Ro ) {[n2(T ) 90 K)]2 - 1} and can be calculated from the refractive index of the film obtained by the fit of the deposition pattern (n2(T ) 90 K) ) 1.450 ( 0.005). The four parameters A, B, Rd, and ν can be determined from a fit of the DR data to eq 1 with the film refractive index expressed as in eq 8. The inset in Figure 4 shows the excellent fit (solid line) of the experimental data (dotted line). The values obtained from the fit are reported in Table 1 and are in good agreement with those obtained from the SH intensity fit. From the analysis, we propose that, for deposition at 90 K, the aniline thin films are composed of small ordered structures, nanocrystallites, with an average size of about 45 molecules (radius of 1.1 nm). These nanocrystallites undergo melting transition in a broad temperature range between 140 and 160 K. The vacuum-aniline and aniline-substrate interfaces do not have a significant effect on the observed melting transition because the average size of ordered clusters is independent of

the film thickness and the results are the same on the Ag(111) and Ag(110) surfaces. For pyridine films, the premelting transition of crystallites with sizes of about half a micron has been shown to be reversible.7 In the case of aniline, as indicated by the data in Figure 5, the order-disorder transition is not reversible. Furthermore, the ordered structure is definitely absent in film deposited at 155 K, even when they are cooled to 90 K. This irreversibility can be explained by the small size of the nanocrystallites. If the number of molecules in the crystallite is smaller than that of the critical nucleus for crystallization, melting would cause its permanent destruction. Lowering the temperature of the film after the melting of the crystallites smaller than the critical nucleus does not facilitate the formation of the nuclei because of the unfavorable free energy.43 The size of the crystallites is smaller than the coherence length of SHG, which is on the order of the wavelength of the SH light. The fact that we can detect SHG from the small crystallites implies either that the clusters are not randomly oriented to result in a zero ensemble polarization or that the crystallites have a low density with intercrystallite distance larger than the coherence length. To determine which scenario is the case, we estimate the fraction of the molecules in the ordered phase. The value of No/Nd at 90 K can be calculated using the values of the parameters A and B obtained from the fit, and reported in Table 1, to be between 106 and 109, which means that nearly all of the molecules exist in the ordered, crystalline form. Then the SH signal should arise from anisotropy in the crystallites orientation, which suggests that the crystallites do not orient randomly in the thin film grown on the metal substrate. The films grown at 155 K, in contrast to the ones grown at 90 K, are amorphous. Apparently, in the low-temperature range of 90-160 K, the amorphous state is more stable than the structure consisting in crystallites with sizes smaller than the critical nucleus. Cooling down the sample from 155 to 90 K does not result in a transition from the amorphous to the ordered phase, as evidenced in the DR curve shown in Figure 5, as well as a further heating from 90 to 155 K does not produce any transition. Moreover, as indicated by the annealing curves in Figure 5, if the premelting transition is stopped at some point during the process, the remaining nanocrystallites are still frozen in. Once the temperature is raised again into the transition region, premelting continues. If the annealing temperature is high enough, for instance at 170 K, for the premelting transition to be completed, the resultant film and the one obtained by direct deposition at 155 K are both amorphous and show similar behavior in DR response. B. Refractive Index of the Nanocrystalline and Amorphous Aniline Films. The value of the refractive index that

23430 J. Phys. Chem. B, Vol. 110, No. 46, 2006

Gonella et al.

we obtain from the analysis of the light intensity interference pattern during film deposition at 90 K in Figure 3, n2 ) 1.450 ( 0.005, is low in comparison with the refractive index expected for crystalline aniline. Since the refractive index of aniline in its crystalline form has not been experimentally measured, we obtain this value using the formula44 based on the LorenzLorentz equation:45,46

nC ) 1 + 2

3(dC/dL)(nL2 - 1) nL2 + 2 - (nL2 - 1)(dC/dL)

(10)

This formula allows the calculation of an average refractive index of the crystalline solid phase, provided that the crystalline density dC as well as the density dL and the refractive index nL of the liquid are known. nL and dL of aniline are 1.586 and 1.022 g/cm3, respectively.47 dC of monoclinic aniline at 252 K is 1.174 g/cm3.31 On the basis of the change of the lattice parameters between 252 and 130 K, the volume at 90 K can be estimated to be about 97% of that at 252 K. This suggests dS at 90 K would be equal to 1.210 g/cm3. These values, through eq 10, result in nC ) 1.726 at 90 K. The low value of the film deposited at 90 K can be understood by taking into account the porosity of the film (see Subsection C). The value of Rd ) 1.070 ( 0.010 that we obtain from the fit of the data in Figure 4 by means of eq 9, corresponds to a refractive index of 1.439 ( 0.003 for the amorphous solid film by assuming that the film thickness remains unchanged after premelting. As expected, this value is lower than the one obtained for the crystalline phase (nC ) 1.726), but it is also much lower than the value for the liquid aniline at room temperature (nL ) 1.586). This discrepancy results from the assumption that the film thickness remains unchanged before and after the premelting transition, which is invalid knowing that the film deposited at 90 K is porous and that the porosity can be reduced during the heating process. On the basis of the percentage porosity calculated in Subsection C for the unanneled film deposited at 90 K and further porosity reduction due to heating, we estimate only ∼70% of this film is solid aniline. If we now use a thickness of 65 nm for a nonporous film, we obtain nA ) 1.68 ( 0.03. At this point, we compare this value of 1.68 ( 0.03 with the refractive index 1.409 ( 0.005 obtained for the film directly deposited at 155 K in the amorphous phase from the fit of the interference pattern during the deposition. The value for the amorphous film deposited at 155 K is much smaller than the one for the compact amorphous solid. We deduce that the amorphous film deposited at 155 K is also porous. This porosity can be estimated, following the same procedure described in Subsection C, as 34 ( 2%, consistent with the films deposited at lower temperatures. C. Porosity in Unannealed Aniline Films. It appears that raising the temperature in the range 90-180 K produces two distinct effects in DR: In addition to the small trough around 150 K, which is discussed in Section A, there is a general but slow increase of the signal over the entire temperature range. The slow and continuous increase can be attributed to several factors such as thickness change induced by thermal expansion of the film, substrate refractive index change, and increase of the film refractive index due to structure annealing, for instance, the reduction of cavities/voids inside the film. Observations made from the bare substrate shows no such DR change during temperature variation. Thus, we can rule out any substraterelated effects. In the results presented in Figure 5, the DR curve obtained at the highest annealing temperature displays no

increase from 90 to 180 K, while the ones at lower annealing temperatures show more increase. This observation indicates that this overall increase can be mitigated by annealing and is most likely due to reduction of cavities/voids during the annealing process. The reduction of porosity in the film is not a reversible process, and the effectiveness of annealing depends on the temperature. The fraction of porosity p in the solid can be estimated from comparing the crystalline refractive index nC and the measured refractive index 〈nC〉 using the Maxwell-Garnett equation48,49

p)1-

〈d〉 〈n〉2 - 1 n2 + 2 ‚ )1- 2 d 〈n〉 + 2 n2 - 1

(11)

where 〈d〉 ≡ 〈dC〉 and 〈n〉 ≡ 〈nC〉 are the average density and refractive index of the porous nanocrystalline structure. Using the values obtained for the crystalline solid at 90 K, d ≡ dC ) 1.210 g/cm3 and n ≡ nC ) 1.726, we calculate that the refractive index of 1.450 ( 0.005 corresponds to a porous film with a density of 0.83 ( 0.02 g/cm3 and a porosity fraction of (32 ( 2) %. D. Crystallization in Thin Aniline Films. The DR response as a function of temperature in Figure 4 shows a large dip starting at 180 K. This dip appears for films deposited both at 90 and 155 K. There is a corresponding change starting ∼180 K in the SHG curve in Figure 2. Both features are related to desorption of the film. The SH signal starts to decease at 180 K, when the film begin to sublimate and reduce the film thickness, and reaches the lowest level at 190 K, when sublimation is completed. The change in the DR signal can be explained similarly except that, because the thickness of the film deposited is higher, it takes a higher temperature for the sublimation to complete. An interesting feature in the DR data in Figure 4 is the sharp rise at 200 K. This peak corresponds to an increase in scattering of light that is even visible to the naked eye: The laser spot on the sample, the film inside the UHV chamber, appears bigger and brighter due to the increase of the diffuse reflected light when temperature is increased above 195 K. This scattering causes not only a huge decrease in the intensity of reflected light but also an uneven scattering in the p- and s-polarized components, resulting in a large DR signal. The large light scattering is likely to arise, as suggested from similar observations on other systems, such as water, to crystallization within the molecular thin film. Westley et al.10 observed that for amorphous solid water films with thickness ∼150 µm, this phenomenon occurs around 200 K and is related to the transformation of the amorphous to the hexagonal phase. Seiber et al.50 observed an increase in light scattering at 145-150 K, which they correlated to the crystallization of amorphous ice to the cubic phase. Here for aniline, it appears that crystallization of the amorphous film starts to occur at 195 K. Increasing the temperature leads to an increased crystallization rate as well as sublimation of the film, which by 205 K, is completely desorbed. The crystallization phenomenon referred to here is the formation of crystallites of micron size that causes substantial scattering of visible light. The formation of micron-size crystallites in these micron-thick molecular films on flat, inert metal surfaces appears to occur at temperatures much lower than the bulk crystallization temperature. For pyridine with a melting point of 232 K, this phenomenon occurs at 120 K. For aniline with a melting point of 267 K, this temperature is observed at 195 K. E. Thin Film Growth Mechanisms and Intermolecular Interaction. Comparing the behavior of premelting of

Structure and Growth of Thin Aniline Films on Ag polycrystallites observed for aniline with that of thin films of pyridine,7,20 a molecule with similar size to aniline but very different intermolecular interaction energy, is highly revealing. It was found that, upon adsorption at 90 K, the pyridine film does not contain any ordered phase. Upon annealing, however, the amorphous pyridine film forms polycrystallites with a halfmicron diameter. This behavior appears to be in striking contrast to that of aniline films. The following thermodynamic argument may be used to explain why aniline forms small ordered-phase crystallites at 90 K while pyridine does not. The two molecules have a very similar molecular shape, and thus there should be no significant difference in the orientational entropy. On the other hand, aniline has stronger hydrogen bonding than pyridine, leading to a stronger binding between aniline molecules. The lattice energy of aniline solid (52.8 kJ/mol) is higher than that of pyridine solid (40.3 kJ/mol).51,52 Because the enthalpy gain by forming an ordered phase is more effective for aniline than pyridine, aniline has more chances to form ordered structures than pyridine. Upon adsorption of a gaseous aniline molecule, the stronger intermolecular interaction means that more enthalpy is released to neighbor molecules. This enthalpy release provides the energy for the surrounding molecules to reorient into the energetically more favorable crystalline structure. Further growth of ordered aniline structure in the bulk, on the other hand, is suppressed by restricted diffusion at cold temperature. To generalize the comparison, it appears that deposition of films of weakly interacting molecules on metal surfaces at temperatures much lower than the melting point would result in an amorphous structure as the molecules adsorbed on the film surface retain the random orientation in the gas. Deposition of stronger-interacting molecules, on the other hand, may result in metastable, nanometer size crystalline structure because of the effect of the enthalpy released, upon adsorption on the neighboring molecules. V. Conclusions The structure and growth of thin films of aniline vapor deposited on low-index Ag surfaces have been examined using optical second harmonic generation and linear optical differential reflectivity. Aniline thin films deposited at 90 K give a SH signal much larger than that from the Ag surface. This signal arises from small polycrystallites with orientation anisotropy inside the film. Upon annealing, the SH signal displays a sharp decrease between 130 and 160 K, where premelting of the crystallites occurs. A further decrease of the SH intensity starting at 180 K is observed and is attributed to the film sublimation. The decrease between 130 and 160 K can be quantitatively described by a premelting model of the crystallites. The model analysis allows the determination of the average size of the crystallites as 1.1 nm in diameter and containing 45 aniline molecules. The existence of the nanocrystallites and their premelting phenomenon are supported by the linear optical differential reflectivity experiments. A notable change in the differential reflectivity through the film is observed around 145 K. This change can be quantitatively described by a model depicting a transition from an ordered phase to the amorphous phase in this temperature range. In the model, the phase transition results in a change of the refractive index. The model analysis also results in a crystallite size the same as the one determined from the SH data. The refractive index of the amorphous aniline film can be determined as 1.68 ( 0.03. The linear reflectivity measurements also indicate that the aniline film deposited at 90 K contains ∼30% porous structure,

J. Phys. Chem. B, Vol. 110, No. 46, 2006 23431 which can be removed through annealing at a higher temperature. In contrast, films deposited directly at a higher temperature such as 155 K is amorphous but is also porous. Raising the temperature further to above 195 K, the DR data show that micron-size crystallites form within the amorphous film to greatly increase light scattering while the film sublimates. Combining observations made in previous studies on pyridine, which has weaker intermolecular interactions than aniline, we summarize the following general behavior for vapor-deposited molecular films at low temperatures. For molecules with stronger intermolecular interactions such as aniline, upon deposition of a molecule, the stronger intermolecular interaction causes more enthalpy release into the local surrounding molecules, causing them to reorient into a crystalline form. The low temperature, on the other hand, prevents diffusion and, therefore, further crystallization. Upon annealing, the nanocrystallites, with a size smaller than the critical nucleus, premelt into the amorphous phase. Lowering the temperature does not recreate the crystallites. Upon further increasing the temperature, the film crystallizes and/or sublimates at temperatures much lower than the bulk melting point. For molecules with weaker intermolecular interactions such as pyridine, the film deposited at low temperature is amorphous. Upon increasing the temperature, crystallization occurs within the thin film at a temperature much lower than the bulk melting point. Increasing the temperature either causes the crystallites to premelt and/or the film to sublimate. Acknowledgment. This work is supported by a grant from the Air Force Office of Scientific Research. The equipment for this research was supported by a grant from the National Science Foundation, MRSEC program, no. DMR00-79909. References and Notes (1) Dimitrakopoulos, C. D.; Kymissis, I.; Purushothaman, S.; Neumayer, D. A.; Duncombe, P. R.; Laibowitz, R. B. AdV. Mater. 1999, 11, 1372. (2) Northrup, J. E.; Tiago, M. L.; Louie, S. G. Phys. ReV. B 2002, 66. (3) Heringdorf, F.; Reuter, M. C.; Tromp, R. M. Nature (London) 2001, 412, 517. (4) Ediger, M. D. Annu. ReV. Phys. Chem. 2000, 51, 99. (5) Ediger, M. D.; Skinner, J. L. Science 2001, 292, 233. (6) Li, C. M.; Ying, Z. C.; Dai, H. L. J. Chem. Phys. 1994, 101, 7058. (7) Yang, M.; Dai, H. L. J. Phys. Chem. B 2003, 107, 12233. (8) Yang, M.; Dai, H. L. Langmuir 2004, 20, 37. (9) Diakoumakos, C. D.; Raptis, I. Polymer 2003, 44, 251. (10) Westley, M. S.; Baratta, G. A.; Baragiola, R. A. J. Chem. Phys. 1998, 108, 3321. (11) Li, W.; Wang, H. L. AdV. Funct. Mater. 2005, 15, 1793. (12) Misra, S. C. K.; Mathur, P.; Yadav, M.; Tiwari, M. K.; Garg, S. C.; Tripathi, P. Polymer 2004, 45, 8623. (13) Ramsey, M. G.; Rosina, G.; Steinmuller, D.; Graen, H. H.; Netzer, F. P. Surf. Sci. 1990, 232, 266. (14) Huang, S. X.; Fischer, D. A.; Gland, J. L. J. Vac. Sci. Technol., A 1994, 12, 2164. (15) Plank, R. V.; Dinardo, N. J.; Vohs, J. M. Surf. Sci. 1995, 340, L971. (16) Schoofs, G. R.; Benziger, J. B. J. Phys. Chem. 1988, 92, 741. (17) Myers, A. K.; Benziger, J. B. Langmuir 1989, 5, 1270. (18) Yang, M. C.; Rockey, T. J.; Pursell, D.; Dai, H. L. J. Phys. Chem. B 2001, 105, 11945. (19) Rockey, T. J.; Yang, M.; Dai, H.-L. Surf. Sci. 2005, 589, 42. (20) Sjodin, T.; Troxler, T.; Dai, H. L. Langmuir 2000, 16, 2832. (21) Dvorak, J.; Borguet, E.; Dai, H. L. Surf. Sci. 1996, 369, L122. (22) Dvorak, J.; Dai, H. L. J. Chem. Phys. 2000, 112, 923. (23) Avouris, P.; Demuth, J. E. J. Chem. Phys. 1981, 75, 4783. (24) Yeganeh, M. S.; Qi, J.; Culver, J. P.; Yodh, A. G.; Tamargo, M. C. Phys. ReV. B 1992, 46, 1603. (25) Wilk, D.; Johannsmann, D.; Stanners, C.; Shen, Y. R. Phys. ReV. B 1995, 51, 10057. (26) Azzam, R. M. A.; Bashara, N. M. Ellipsometry and Polarized Light, paperback ed.; North-Holland: New York, 1987.

23432 J. Phys. Chem. B, Vol. 110, No. 46, 2006 (27) Born, M.; Wolf, E. Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light, 7th expanded ed.; Cambridge University Press: New York, 1999. (28) Hagemann, H. J.; Gudat, W.; Kunz, C. J. Opt. Soc. Am. 1975, 65, 742. (29) Palik, E. D.; Ghosh, G. Handbook of Optical Constants of Solids; Academic Press: Orlando, FL, 1985. (30) Kingslake, R. Applied Optics and Optical Engineering; Academic Press: New York, 1969. (31) Fukuyo, M.; Hirotsu, K.; Higuchi, T. Acta Crystallogr., Sect. B 1982, 38, 640. (32) Gamba, Z.; Klein, M. L. Chem. Phys. 1989, 130, 15. (33) Craven, C. J.; Hatton, P. D.; Howard, C. J.; Pawley, G. S. J. Chem. Phys. 1993, 98, 8236. (34) Dounce, S. M.; Yang, M.; Dai, H. L. Phys. ReV. B 2003, 67. (35) Reiss, H.; Mirabel, P.; Whetten, R. L. J. Phys. Chem. 1988, 92, 7241. (36) Bartell, L. S.; Jian, C. J. Phys. Chem. 1992, 96, 8801. (37) Turnbull, D. J. Appl. Phys. 1950, 21, 1022. (38) Dibble, T. S.; Bartell, L. S. J. Phys. Chem. 1992, 96, 2317. (39) Xu, S. M.; Bartell, L. S. J. Phys. Chem. 1993, 97, 13550.

Gonella et al. (40) Bogdan, A.; Kulmala, M.; Avramenko, N. Phys. ReV. Lett. 1998, 81, 1042. (41) Jackson, C. L.; McKenna, G. B. J. Chem. Phys. 1990, 93, 9002. (42) Ercolessi, F.; Andreoni, W.; Tosatti, E. Phys. ReV. Lett. 1991, 66, 911. (43) Kashchiev, D. Nucleation: Basic Theory with Applications; Butterworth Heinemann: Oxford, Boston, 2000. (44) Dejong, S.; Groeneweg, F.; Vader, F. V. J. Appl. Crystallogr. 1991, 24, 171. (45) Lorenz, L. Ann. Phys. 1880, 247, 70. (46) Lorentz, H. A. Ann. Phys. 1880, 245, 641. (47) http://www.sigmaaldrich.com/. (48) Maxwell Garnett, J. C. Philos. Trans. R. Soc. London, Ser. A 1904, 203, 385. (49) Maxwell Garnett, J. C. Philos. Trans. R. Soc. London, Ser. A 1906, 205, 237. (50) Seiber, B. A.; Wood, B. E.; Smith, A. M.; Muller, P. R. Science 1970, 170, 652. (51) Rochester, C. H.; Waters, J. A. J. Chem. Soc., Faraday Trans. 1982, 78, 631. (52) Vriens, G. N.; Hill, A. G. Ind. Eng. Chem. 1952, 44, 2732.