Surface equilibration mechanism controls the molecular packing of glassy molecular semiconductors at organic interfaces

Significance

Because glasses are nonequilibrium materials, even a single-component system has an enormous number of distinct glass structures. Physical vapor deposition (PVD) is an important tool for preparing a wide range of these glasses. While the substrate can influence structure for hundreds of nanometers for crystalline and liquid crystalline systems, little is known about the structure of PVD glasses near an underlying substrate. This is important for organic electronics, as layers of PVD glasses as thin as 10 nm are used as active elements in organic light-emitting diodes, and molecular packing is vital to performance. We show that the free surface controls molecular packing of PVD glasses near a buried interface rather than the substrate, in sharp contrast with crystalline systems.

Abstract

Glasses prepared by physical vapor deposition (PVD) are anisotropic, and the average molecular orientation can be varied significantly by controlling the deposition conditions. While previous work has characterized the average structure of thick PVD glasses, most experiments are not sensitive to the structure near an underlying substrate or interface. Given the profound influence of the substrate on the growth of crystalline or liquid crystalline materials, an underlying substrate might be expected to substantially alter the structure of a PVD glass, and this near-interface structure is important for the function of organic electronic devices prepared by PVD, such as organic light-emitting diodes. To study molecular packing near buried organic–organic interfaces, we prepare superlattice structures (stacks of 5- or 10-nm layers) of organic semiconductors, Alq3 (Tris-(8-hydroxyquinoline)aluminum) and DSA-Ph (1,4-di-[4-(N,N-diphenyl)amino]styrylbenzene), using PVD. Superlattice structures significantly increase the fraction of the films near buried interfaces, thereby allowing for quantitative characterization of interfacial packing. Remarkably, both X-ray scattering and spectroscopic ellipsometry indicate that the substrate exerts a negligible influence on PVD glass structure. Thus, the surface equilibration mechanism previously advanced for thick films can successfully describe PVD glass structure even within the first monolayer of deposition on an organic substrate.

For many organic thin films, the substrate can profoundly influence molecular packing over long length scales. For crystalline organic semiconductors, the substrate can alter packing to such a large extent that new polymorphs known as “substrate-induced phases” form at buried interfaces (1–3). In other cases, crystal orientation is determined by substrate choice (4–6). Both polymorph selection and crystal orientation have an immense impact on the performance of organic electronic devices. Similarly, substrate effects are a dominant factor for liquid crystals in which alignment layers are used to control molecular orientation over micrometer length scales (7), and this alignment is crucial for the operation of displays and sensors. For organic glasses, much less is known about how an underlying substrate influences the material’s structure, despite the importance of molecular packing for the operation of thin film devices.

Physical vapor deposition (PVD) is a standard method to prepare thin glassy layers for organic electronics, including OLED (organic light-emitting diode) displays (8, 9), which are a 35-billion-dollar industry (10). While most OLED displays are used for smartphones, large and flexible OLED displays are also prepared by PVD (11). PVD is particularly well suited for preparing the multilayer structures found in OLED devices. PVD can create efficient devices in part because deposition conditions can be chosen to optimize device performance (12–18). In comparison to solution processing, OLED devices prepared by PVD have greater stability against ultraviolet irradiation (19) and do not contain residual solvent, which can reduce efficiency and increase degradation (20). PVD is also a common method to prepare organic photovoltaic cells (OPVs) (21). In some cases, PVD has also been used to prepare organic field effect transistors comprising of glassy layers (22).

Thick glassy molecular films prepared by PVD often exhibit anisotropic molecular packing (23–25). The molecular packing in a PVD glass can be controlled with deposition temperature and rate (26–28) and is rationalized within the framework of the surface equilibration mechanism (29–31). In brief, enhanced mobility at the free surface of a molecular glass (32) allows substantial equilibration during deposition, even below the glass transition temperature T_g. This process allows for dense molecular packing (30, 33), and since the equilibration takes place near the anisotropic surface of the deposited film, the molecular packing trapped into the glass is generally anisotropic. Substrate temperature and deposition rate alter the anisotropic packing by influencing the extent to which surface equilibration occurs during deposition. Over a broad range of substrate temperatures and deposition rates, the anisotropy of PVD glasses can be controlled while simultaneously producing materials with dense molecular packing and high kinetic stability (34, 35).

In contrast to thick films, the structure of PVD glasses near interfaces is not well understood and is important in the context of organic electronics. Interfaces between glassy organic semiconductors are ubiquitous in OLED devices; the core of a typical device consists of a sequentially deposited hole-transport layer, an emissive layer, and an electron transport layer, with each layer 10- to 50-nm thick. The structure of organic glasses within a few nanometers of an organic substrate is critical to charge transport through devices (13, 21, 36), but little is known about the molecular packing near the interface of two organic semiconductors (37). There is no consensus in the literature even regarding the length scale over which the substrate influences the structure of a PVD glass. Using ellipsometry, Yokoyama et al. reported that the substrate affected the molecular orientation of PVD glasses as thick as 100 nm (14), while more recent work by Yang et al. concluded that the substrate can perturb molecular orientation for distances up to 30 nm (38). This is a much larger length scale than the roughly 5-nm-length scale found by Bagchi et al. for PVD glasses deposited on silicon at room temperature (39) or that observed in recent simulations by Yoo et al. (40) This controversy, and the importance of even a few nanometers of substrate-influenced material for charge transport, motivates more precise measurements of glassy packing in order to understand how to control the structure of buried interfaces.

In this work, we study the structure of PVD films with buried glassy interfaces in a geometry that closely resembles organic electronic devices. The 5- and 10-nm superlattices of glassy Alq3 (Tris-(8-hydroxyquinoline)aluminum) and DSA-Ph (1,4-di-[4-(N,N-diphenyl)amino]styrylbenzene) prepared by PVD are compared with bilayers (each layer is a 300-nm-thick PVD film) of the materials with the same average composition to investigate structure near the organic–organic interface. The superlattice structures allow us to gain interfacial sensitivity without compromising the signal-to-noise or signal-to-background ratio. Grazing-incidence wide-angle X-ray scattering (GIWAXS) and spectroscopic ellipsometry (VASE) measurements were performed to probe the structure of the superlattice films. Both GIWAXS and VASE measurements show that the superlattice films are very similar to the bilayer films (and very different from co-deposited mixtures of Alq3 and DSA-Ph). We find that the surface equilibration mechanism controls the anisotropic molecular packing near buried interfaces in the same manner as it controls the structure of thick films. Our work establishes a procedure to control interfacial molecular packing in organic electronic devices such as OLEDs.

Results

Glassy Molecular Packing in Thick Single-Component Films and Bilayers.

While the packing of thick films of both Alq3 (27) and DSA-Ph (41) has been previously studied, it is useful to begin by briefly describing these glassy samples to set the stage for the bilayer and superlattice work. The structure of thick films of Alq3 and DSA-Ph prepared by PVD has been measured using GIWAXS; representative scattering patterns of thick films of each material are shown in Fig. 1. As shown in Fig. 1, Left, glassy PVD films of Alq3, a spherical organic semiconductor, display a molecular layering feature as indicated by the anisotropic scattering along Q_z near 0.75 Å⁻¹. The position of this peak depends weakly on temperature, and we utilize this feature below. In Fig. 1, Right, GIWAXS patterns show that the anisotropic molecular packing of thick films of DSA-Ph, a rod-shaped small molecule, is highly dependent on the substrate temperature (T_sub) during PVD. At low temperatures (290 K), the DSA-Ph molecules pack face on as indicated by the excess scattered intensity along Q_z at 1.4 Å⁻¹. In contrast, GIWAXS pattens of thick films of DSA-Ph glasses deposited at high substrate temperatures (343 K) have excess scattered intensity along Q_xy, indicating a tendency for end-on molecular packing. Control of the anisotropic packing of both Alq3 (27) and DSA-Ph (26, 41) in thick glassy films by the substrate temperature during PVD can be explained by the surface equilibration mechanism. The anisotropic scattering peaks of Alq3 and DSA-Ph (Fig. 1) are observed in different Q ranges, and this separation makes this pair suitable for studying molecular packing in the superlattices.

Fig. 1.

GIWAXS patterns of thick single-component films (600 to 700 nm) of Alq3 (Left) and DSA-Ph (Right) prepared at two different substrate temperatures (indicated in the top right of each image). The molecular structure and glass transition temperature, T_g, of each organic semiconductor are shown. [The data shown for the DSA-Ph sample at T_sub = 290 K was used in a previous publication (39).]

We deposited thick bilayers of Alq3 and DSA-Ph to act as reference samples for the superlattice experiments. Fig. 2B shows the scattering of thick PVD bilayers of Alq3 and DSA-Ph at two substrate temperatures. A comparison of Figs. 1 and 2 shows that the scattering of the bilayer appears to be the sum of the scattering of thick, single-component films for both materials. This is as expected, as previous work has shown PVD glasses with thicknesses in the range of 300 nm have structures that are independent of the underlying substrate (although the structure does depend upon substrate temperature and deposition rate) (39). We compare the superlattices to the bilayers (rather than single-layer films) because the average composition is the same, allowing us to focus on the impact of the different number of buried interfaces.

Fig. 2.

GIWAXS patterns of bilayer and superlattice films of Alq3 and DSA-Ph. (A) Schematic of the film geometries used in this study. Each bilayer and superlattice film has a total thickness of 600 nm and is composed of 50% Alq3 and 50% DSA-Ph by mass; only the slab thicknesses vary. (B) GIWAXS patterns of the Alq3/DSA-Ph bilayer and superlattice films. The deposition temperature is indicated in the upper right corner of each image. Qualitatively, the scattering of the bilayers is very similar to the scattering of the superlattices at both substrate temperatures. This is shown as a cartoon in C, which represents the molecular packing of the Alq3 (orange circles) and DSA-Ph (cyan rods) in a single DSA-Ph slab in a superlattice deposited at T_sub = 290 K.

Comparison of the Molecular Packing for Thick Bilayers and Superlattices.

Comparing the GIWAXS patterns of the bilayer and superlattice films allows us to study the structure near Alq3/DSA-Ph interfaces. In Fig. 2, we see that the GIWAXS patterns of the bilayer and superlattice samples are virtually identical at a given substrate temperature. Additional GIWAXS patterns are shown in SI Appendix, Figs. S1 and S2. We see two main features in each pattern: an Alq3 layering peak near Q_z ∼ 0.75 Å⁻¹ and a nearest neighbor peak for DSA-Ph at Q ∼ 1.4 Å⁻¹. Since the bilayer and superlattice films have the same average composition and are deposited at the same substrate temperatures, the similar scattering indicates that molecular packing in 10-nm layers is essentially the same as the packing of much thicker films. As we discuss in this section, quantitative analysis of the scattering features supports this conclusion. Therefore, the molecular packing near organic–organic interfaces can be controlled by the substrate temperature and understood through the surface equilibration mechanism in the same way as for bulk PVD glasses. X-ray reflectivity (XRR) measurements reveal the superlattice interfaces are compositionally sharp and topographically smooth; we find that the interfacial width is about 1 nm (SI Appendix, Figs. S3 and S4). Thus, we expect that the effects of interdiffusion and surface roughness at the buried interfaces have a negligible impact on the interpretation of these experiments.

Quantitatively, the packing of DSA-Ph molecules is nearly identical in the superlattice and bilayer films at all deposition temperatures. In thick films, DSA-Ph packs with preferentially horizontal or vertical molecular orientation depending upon the substrate temperature, and this control has been rationalized within the framework of the surface equilibration mechanism. The packing can be described with a GIWAXS order parameter, S_GIWAXS, calculated at Q = 1.4 Å⁻¹, the position of the main scattering feature for DSA-Ph. Fig. 3A shows the S_GIWAXS value as a function of substrate temperature for the thick bilayers and the superlattices. As the substrate temperature is increased, the value of S_GIWAXS decreases. A positive or negative S_GIWAXS value is consistent with a tendency for DSA-Ph molecules to pack face-on or end-on, respectively, while an S_GIWAXS value of zero is consistent with isotropic molecular packing. Fig. 3A shows that the S_GIWAXS values for the superlattice samples are identical, within experimental error, to those extracted from the bilayer data at all substrate temperatures.

Fig. 3.

DSA-Ph molecular packing and orientation in the bilayer films (pink circles), 10-nm superlattices (blue triangles), and 5-nm superlattices (green squares) plotted as a function of substrate temperature. (A) S_GIWAXS parameter calculated at ∼1.4 Å⁻¹ for the superlattice and bilayer films as a function of substrate temperature. The diagonal dashed line is a linear fit of the bilayer results, and the line at S_GIWAXS = 0 indicates the point where the scattering is isotropic. (B) The DSA-Ph birefringence in the superlattices as a function of substrate temperature. The birefringence of the superlattice follows the same trend as the birefringence of 70 to 900 nm thick films of DSA-Ph (gray) (data from ref. 26). The horizontal line at DSA-Ph birefringence = 0 indicates where the film is optically isotropic. (C) Schematic illustration of the DSA-Ph packing corresponding to different S_GIWAXS and birefringence values. Each cyan rod represents a DSA-Ph molecule. A–C together indicate that DSA-Ph glasses near buried interfaces have the same structure as thick films; in both cases, the structure is determined by the deposition temperature.

Ellipsometric measurements of birefringence complement the GIWAXS data, supporting the view that the packing of DSA-Ph molecules in the superlattice and bilayer samples is essentially identical. Ellipsometry selectively probes the birefringence of DSA-Ph layers, as Alq3 films are nearly isotropic optically (27, 42). Birefringence has been correlated with the molecular orientation of the DSA-Ph in a PVD glass (26). The birefringence of DSA-Ph in the thick bilayers and superlattices is plotted in Fig. 3B as a function of deposition temperature. The birefringence can be compared with single-component thick films of DSA-Ph glasses; these are plotted as gray circles in Fig. 3B. Positive, negative, and zero birefringence indicate vertical, horizontal, and isotropic orientation of the DSA-Ph long axis, respectively. [Note that the birefringence and the S_GIWAXS values are typically anticorrelated for PVD glasses (23).] In all cases, similar values of the DSA-Ph birefringence are obtained for thick films, bilayers, and superlattices. Therefore, control over the molecular packing in both thick and ultrathin films of DSA-Ph can be understood by the surface equilibration mechanism.

The packing of Alq3 molecules is nearly identical in the superlattices, bilayer films, and thick films at all deposition temperatures as determined by analyzing the scattering feature at Q_z ∼ 0.75 Å⁻¹. This feature describes the out-of-plane layering of the Alq3 molecules in the films (27). The location of this peak along Q_z is plotted in Fig. 4 as a function of the substrate temperature at which the sample was prepared. Data for the thick Alq3/DSA-Ph bilayers and superlattices are plotted along with the Alq3 peak position of thick, single-layer Alq3 films. For all the films presented, the peak is observed at lower Q_z values for films deposited at higher substrate temperatures. In all cases, similar values of the Alq3 peak positions are observed for thick films, bilayers, and superlattices. This indicates that the Alq3 packing in 10-nm-thin films, or 5-nm films deposited at T_sub = 290 K, can be controlled by leveraging the surface equilibration mechanism, allowing the structure of ultrathin Alq3 films to be tuned.

Fig. 4.

The peak position of the Alq3 layering feature along Q_z as a function of substrate temperature in the single layer (gray circles), bilayer (pink circles), and superlattice (blue triangles and green squares) films. The dashed line is a linear fit to the thick (500- to 700-nm) Alq3 film data.

The data in Figs. 3 and 4 support the conclusion that there is minimal intermixing of the Alq3 and DSA-Ph at the vapor-deposited interfaces, as the molecular packing of both Alq3 and DSA-Ph in co-deposited films of Alq3 and DSA-Ph is significantly different from in the superlattices (SI Appendix, Fig. S6). This observation, along with the XRR measurements, provides strong evidence that interdiffusion at the Alq3/DSA-Ph interfaces is minimal and has a negligible effect on the data presented in this manuscript.

Structure of Buried Alq3/DSA-Ph Interfaces.

To understand the molecular packing of a PVD glass near an organic–organic interface, we consider two simple models of our data. Specifically, we examine how the thickness of a DSA-Ph film affects its structure at a given substrate temperature (Fig. 5). We choose the S_GIWAXS order parameter because it is a simple assessment of the molecular packing within films, unlike the model-dependent birefringence measurements obtained from the VASE data. We first model the data with the assumption that DSA-Ph glasses in thick films, bilayers, and superlattices have the same structure (and hence S_GIWAXS value) and that the underlying Alq3 substrate has no impact on the molecular packing. This model is presented as the solid lines in Fig. 5. We see that this is a reasonable description of the data, given that there is spread in the experimental results. A cartoon representation of this model is also presented in Fig. 2C, which shows a slab of DSA-Ph with two buried Alq3 interfaces deposited at T_sub = 290 K. The packing of the DSA-Ph is the same through the ultrathin film. We also present a second model of the data (Eq. 1) in which we allow a single molecular layer at the interface to pack differently than the bulk structure:

$$S_{GIWAXS}=S_{bulk}+\frac{1.4 nm}{h}\left(S_p-S_{bulk}\right).$$ [1]

Fig. 5.

The S_GIWAXS parameter (calculated at 1.4 Å⁻¹) plotted as a function of the DSA-Ph slab thickness within the bilayer and superlattice films. We plot data for the samples vapor deposited at 290 (blue circles), 333 (green squares), and 343 K (red diamonds). The solid horizontal lines indicate the average value of S_GIWAXS across the different thicknesses. The dashed lines fit the data to a two-layer model that perturbs the packing of the molecular layer adjacent to the organic substrate. The vertical gray dashed line marks the thickness of this first layer (1.4 nm).

In Eq. 1, S_bulk is the measured S_GIWAXS value for the bilayer sample at a given substrate temperature, and h is the DSA-Ph slab thickness. The lone fitting parameter in Eq. 1 is S_p, the S_GIWAXS value of the perturbed molecular layer. For specificity, the thickness of the perturbed layer is set to be the thickness of a single isotropic layer of DSA-Ph (1.4-nm thick). This second model, shown in Fig. 5 as dashed lines, also provides a reasonable description of the data.

Several considerations lead us to conclude that the model with no interfacial perturbation (solid lines in Fig. 5) is a better description of the experimental data. The fit quality is hardly improved with the second model (dashed lines in Fig. 5), which introduced the potential for one molecular layer with a different structure than the bulk film. (For more about fitting and CIs, reference SI Appendix, Supplementary Information.) Furthermore, even if an interfacial perturbation is allowed, for two of the three substrate temperatures, the interface properties (S_p) are predicted to be essentially bulk-like (S_bulk). This is especially convincing for the 290-K samples since we have data for thinner (5-nm) superlattices. It seems unlikely that a perturbed layer at the interface would exist only for glasses deposited at 333 K, where the modeled value of S_p is quite different from S_bulk, but we cannot exclude this possibility. Data for thinner DSA-Ph slab thicknesses would improve our understanding of molecular packing in ultrathin films. However, we anticipate that thinner layers may fail to form continuous films, and thus a well-defined superlattice structure may not be formed.

Discussion

We have measured the molecular packing at the buried interface of two organic semiconductor glasses (Alq3 and DSA-Ph) by growing superlattice structures using PVD. We characterized the superlattice films with GIWAXS and VASE and made comparisons with thick films and co-deposited films. The structure of 5-nm and 10-nm glassy films created by PVD is essentially identical to thicker films of the same materials (and quite different from co-deposited films). Thus, we can choose the structure in the ultrathin films deposited onto organic substrates by controlling the substrate temperature at which they are prepared, just as is possible with thick films. In this section, we consider the mechanism controlling this behavior, compare our results to the literature, and consider the significance of these results for organic electronics.

Control of the molecular packing in thick films prepared by PVD can be explained by the surface equilibration mechanism, and we briefly comment on why this mechanism appears to also be valid for films as thin as 5 nm. For thick films, it has been established that the highly mobile surfaces of organic glasses play a key role in defining the properties of PVD glasses (29, 30). Even when the substrate temperature is below T_g, and bulk relaxation processes are extremely slow, molecules near the surface of an organic glass move quite rapidly (32). At a typical deposition rate (∼1 monolayer every 10 s), freshly deposited molecules sample many different packing arrangements before they are trapped into place by further deposition. This equilibration process drives the system toward glasses of high density and high stability, properties that might eventually be obtained through the bulk aging process if one could wait for thousands of years (43). Since equilibration during deposition happens near the free surface (an anisotropic environment), often the molecules are trapped in packing arrangements that result in anisotropic glasses. In the surface equilibration mechanism, the structure of a thick glass film is templated by the free surface and not the underlying substrate. The substrate might reasonably be expected to have an influence on the structure near the substrate, particularly in the case of a crystalline substrate or strong interactions between the deposited molecules and the substrate. For the amorphous organic substrates considered here, our results indicate that even the first monolayer appears to be very similar to the structure of a thick film. Even with stronger interactions between the deposited molecules and the underlying substrate, any influence of substrate on the deposited film is likely quickly lost as the film becomes thicker. In contrast to crystals, glasses have many possible local packing arrangements of nearly equal energy and thus cannot propagate order across large distances.

Previous Work on PVD Glass Structure Near Organic Substrates.

Our experimental results are in remarkable agreement with recent computer simulations by Han and coworkers (40). These investigators used molecular dynamics simulations to prepare PVD bilayer glasses of organic semiconductors. Each layer was a different pure component, and a total of four organic semiconductors were studied. The bilayer glasses prepared by this route had narrow regions of intermixing between the two components (about one molecular layer), similar to the glasses investigated here. Consistent with our results, they reported no significant change between the orientation of a given molecule in the intermixed layer and in the film in which it was essentially a pure component. They also reported that a small number of molecules ended up in the incorrect layer, surrounded by molecules of the other component; these “inter-diffused” molecules had a different molecular orientation than the bulk orientation for that component. Inter-diffused molecules were less likely at lower simulation temperatures. Considering that the simulations are performed at higher temperatures and on much shorter time scales than the experiments, some extrapolation is required for a comparison of the two. Given the lower temperatures of the experiments, it seems reasonable that we do not observe evidence for such inter-diffused molecules. Altogether, the simulations and experiments show remarkably consistent behavior indicating that a buried organic interface has a minimal impact on the packing in PVD glasses.

Recent experiments examined the properties of a buried interface in single-component films prepared by PVD, and these results are broadly consistent with those reported here. Orientational bilayer glasses (without any change in composition) were prepared by switching the substrate temperature in the middle of deposition. DeLongchamp and coworkers used polarized resonant soft XRR (p-RSoXR) to study bilayer glasses of posaconazole (44). Resonant soft reflectivity allows nanometer-scale resolution of the molecular orientation in such a film, including in the vicinity of the buried interface. The authors reported that the molecular orientation transitioned smoothly in the middle of the film with an interfacial breadth of 2.8 nm (roughly one molecular layer). This result is consistent with and extends what was known from earlier studies of orientational bilayers; the earlier work utilized ellipsometry and thus had lower spatial resolution (45). While an orientational bilayer is clearly different from the two-component bilayers studied here, we consider results on these systems to be broadly consistent. In each case, when deposition conditions change, a transition from one type of glassy packing to another occurs on extremely small-length scales. Furthermore, the measurement techniques are complementary. p-RSoXR directly profiles a thin film, while in GIWAXS, our spatial resolution results indirectly from the superlattice structure. On the other hand, GIWAXS directly reports on local packing features, which at best are detected indirectly in p-RSoXR. Bilayers of Alq3 have also been investigated to identify possible substrate effects on the surface potential of PVD glasses. Thick PVD films of Alq3 display a positive surface potential because of the orientation of the molecular dipoles, and Okabayashi et al. created an Alq3 film with a negative surface potential by exposing the backside of Alq3 PVD glasses (46). When a new Alq3 film was deposited onto the backside with the same deposition conditions, the surface potential of the overall film became more positive, indicating that the Alq3 dipole orientation is unaffected by the surface potential of the substrate, at least for films that are hundreds of nanometers thick.

Previous Work on PVD Glass Structure Near Inorganic Substrates.

Recent work exploring PVD glass structure near inorganic substrates provides some relevant context for our results. Bagchi et al. performed GIWAXS measurements on glassy films of DSA-Ph on silicon (with native oxide layer) and gold substrates over a range of film thicknesses from 13 to 600 nm (39). Systematic changes in the scattering with thickness were interpreted in terms of a two-state model. For PVD films prepared at room temperature, the model indicates that roughly 8 nm of material has a perturbed (isotropic) structure, while the rest of the film shows the structure of thick films. The authors indicate that this 8 nm includes changes in structure both at the free surface and at the substrate. Thus, in these experiments, the substrate influenced the structure of the deposited glass for at most 8 nm. Bagchi et al. reported results at a lower substrate temperature indicating that the structure of the deposited glass was influenced for no more than 5 nm. While these length scales are small, it does seem clear that the inorganic substrates influence the glass structure over a larger distance than an organic substrate, and we offer three potential reasons for this. First, the interactions between an organic semiconductor and an inorganic substrate are likely stronger than for the organic–organic case. Simulations support the view that the glass structure can be strongly perturbed over about 3 nm by modifying the interactions between the organic semiconductor and the inorganic substrate (39, 47). Second, thin films of organic semiconductors will dewet some inorganic substrates. During the initial stages of film growth, dewetting will lead to a film with a different structure than that predicted by the surface equilibration mechanism. Finally, it is difficult to guarantee that the inorganic substrates are completely clean before the glass is deposited onto it. By depositing a thick layer of Alq3 to act as a substrate and then immediately depositing the superlattice, we minimize the potential that glassy packing will be affected by contaminants. In comparison to the work on superlattice structures reported here, it is likely that these three factors combine to give rise to the larger length scale observed by Bagchi et al.

Developments in vibrational sum frequency generation (VSFG) spectroscopy have allowed for measurements of molecular orientation directly at buried interfaces, including a study of PVD glasses of CBP (4,4'-Bis(N-carbazolyl)-1,1'-biphenyl) films deposited onto CaF₂ substrates (48). VSFG revealed that the long axis of the CBP molecules adjacent to the CaF₂ substrate had an average molecular orientation of ∼48°, suggesting that the CBP packed edge-on at the inorganic substrate. However, this technique cannot be easily applied to study buried interfaces of PVD glasses because many of these materials (49), including Alq3 (50), have dipolar order that can give rise to an intense VSFG signal from the bulk of the PVD glass; thus, VSFG loses its interfacial specificity. By using the superlattice films discussed in this work, we are able to detect the signal from the PVD glass near the organic–organic interface, even for films with dipolar order.

Significance of Results for Organic Electronics.

Because the molecular structure within the first few nanometers of a glass deposited on an organic substrate is controlled by the same mechanism that drives the bulk structure, thick films of PVD glasses are a good model of the thin (10- to 50-nm) films used in OLED devices. The molecular structure at organic–organic interfaces is instrumental for efficient charge transport across the layers of organic electronic devices, and our understanding of structure in ultrathin PVD glasses may lead to better devices. For example, work by Yokoyama et al. found that horizontal molecular orientation increases charge transport in thick films of organic glasses of linear molecules (14). Horizontal orientation is thought to enhance charge transport because it leads to better overlap of the frontier orbitals. Similarly, Adachi and coworkers find that horizontal molecular orientation of disk-shaped organic semiconductors improves their performance as hole injection layers in OLED devices (16). In the present work, we demonstrate that the surface equilibration mechanism dictates molecular orientation in thin organic layers in model OLED geometries. Molecular orientation, among other factors, at buried interfaces influences charge injection, so the ability to choose molecular orientation at these interfaces by the selection of the substrate temperature is helpful in designing organic electronic devices.

Given the large number of organic semiconductors and the importance of organic–organic interfaces in devices, it is useful to consider how general our results may be. We report that the surface equilibration mechanism controls the molecular packing near the buried interface with an organic substrate in glasses prepared by PVD. Within the error of our measurements, packing in the thin films is unperturbed by the organic substrate onto which they are deposited. It would be useful to test this conclusion with other pairs of organic semiconductors, but we expect our conclusion to be broadly applicable. Many organic semiconductors have chemical functionalities similar to either Alq3 or DSA-Ph. In addition, although dipole–dipole interactions can influence the properties of PVD glasses (51, 52), our work shows that the influence of an underlying organic substrate is small both for deposition of a polar molecule (Alq3) and for a nonpolar molecule (DSA-Ph).

In summary, using spectroscopic ellipsometry and GIWAXS, we find that the structure of alternating 10-nm Alq3 and DSA-Ph films is the same as the structure of thick films of each material. We conclude that the surface equilibration mechanism developed for bulk PVD glasses can be used to precisely control the molecular packing in stacks of ultrathin films of organic semiconductors. Using substrate temperature to control the structure of ultrathin glassy organic semiconductors prepared by PVD should allow for the creation of more efficient devices. This work shows that the surface equilibration mechanism is valid in ultrathin PVD films created on organic substrates, like those used in OLED devices.

Materials and Methods

Alq3 (sublimed grade, 99.995% trace metals basis) was purchased from Sigma-Aldrich, and DSA-Ph (>99% grade) was purchased from Luminescence Technology Corp. Both materials were used without further purification. Films were vapor deposited on <100> cut silicon substrates with ∼2-nm native oxide layers (Virginia Semiconductor, Inc.). The base pressure of the chamber is near 10⁻⁷ Torr. Single-component films were prepared by evaporating each material onto a temperature-controlled silicon substrate. The deposition rate was 0.2 nm/s as measured by quartz crystal microbalance (QCM).

To prepare the bilayers and superlattices, a 100-nm film of Alq3 was first deposited onto the silicon before the bilayer or superlattice. The underlying Alq3 layer was deposited to eliminate any influence of the silicon substrate. The deposition rate of each organic semiconductor was independently controlled to be 0.2 nm/s as measured by QCM. A rotating shield was custom built and installed in the chamber to allow for the preparation of multilayer films with sharp interfaces. The shield blocked one deposition source at a time, allowing the deposition of alternating layers of the superlattices with only about a 1-s pause between the growth of different layers; this minimizes potential equilibration at the free surface of a given layer prior to the deposition of the next layer. The 10-nm superlattices were prepared by alternating depositions of 10 nm Alq3 and DSA-Ph; the total thickness of the superlattice structure is 600 nm. These regular superlattice structures were used for XRR (SI Appendix, Supplementary Information). In order to ascertain if GIWAXS or VASE measurements contained any artifacts because of the regular superlattice structure, similar samples were prepared in which slab thicknesses were randomly varied in the range 8 to 12 nm with an average value of 10 nm. GIWAXS and VASE measurements were identical within error for regular and irregular superlattice structures, and results for both structures are discussed together. The 5-nm superlattices were prepared by alternating deposition of Alq3 and DSA-Ph in which the thickness of each slab was varied from 4 to 6 nm to form superlattices with a total thickness of 600 nm. All GIWAXS and VASE measurements were performed at room temperature outside the deposition chamber.

For comparison with the superlattice and bilayer films, co-deposited glasses of Alq3 and DSA-Ph were prepared by simultaneously evaporating the materials from two independently controlled crucibles. The deposition rate of each material was 0.1 nm/s, so the total deposition rate of the co-deposited films was 0.2 nm/s. This resulted in 600-nm 50:50 co-deposited films of Alq3 and DSA-Ph prepared on silicon substrates. Co-deposited films were prepared at substrate temperatures of 290, 333, and 343 K.

The thickness and birefringence of the glassy films were measured using a J. A. Woollam M-2000U VASE. Data were collected at incidence angles of 50, 60, and 70° and across a wavelength range of 500 to 1,000 nm. The data were modeled using CompleteEASE, a proprietary software from J. A. Woollam Co. All models included the silicon substrate and a 2-nm native oxide layer. The bilayer samples were modeled as three distinct layers: the Alq3 and DSA-Ph in the bilayer and the 100-nm Alq3 layer deposited directly on the silicon substrate. The superlattice films were modeled as two layers: an average layer describing the superlattice and the 100-nm Alq3 layer deposited directly on the silicon substrate. The co-deposited Alq3:DSA-Ph films were modeled as a single layer. The DSA-Ph birefringence at 632.8 nm was determined for the superlattice and co-deposited films using a Cauchy model with uniaxial anisotropy following the procedure outlined by Jiang et al. (53) Briefly, the DSA-Ph birefringence is calculated as the birefringence of the superlattice or co-deposited layer multiplied by two, as equal amounts of DSA-Ph and Alq3 are present in the superlattice or co-deposited film, and all birefringence can be attributed to the DSA-Ph since the Alq3 is optically isotropic (27). We also used VASE to verify the kinetic stability of a superlattice film. As expected, DSA-Ph glassy layers in a 10-nm superlattice are kinetically stable upon heating above the DSA-Ph T_g as described in SI Appendix, Supplementary Information.

GIWAXS measurements were performed at the Stanford Synchrotron Radiation Lightsource on Beamline 11–3. The beam energy was 12.7 keV. The measurements were performed at an incidence angle of 0.14° with an exposure time of 180 s. For the 600-nm DSA-Ph thick films (Fig. 1), measurements were performed with a sample to detector distance of 300 mm. All other GIWAXS measurements were performed with a sample to detector distance of 315 mm. The sample to detector distance was calibrated with LaB₆. The raw diffraction patterns were “chi corrected,” resulting in the missing wedge along Q_z (54). The broad GIWAXS scattering features indicate that the films studied here are not crystalline. This is consistent with previous work showing that PVD glasses of Alq3 (55) and DSA-Ph (26) transform into isotropic supercooled liquids upon heating above T_g; crystalline films would not form the isotropic liquid until heated above the melting temperature.

The GIWAXS order parameter S_GIWAXS was used to quantify the anisotropy of the scattering patterns. Data in the Q range of 1.35 to 1.45 Å⁻¹ were summed, and data in the missing wedge were obtained by extrapolation. To evaluate the background contribution to the scattering, data in the Q range of 2.17 to 2.27 Å⁻¹ were summed. The background subtracted data were used to calculate the order parameter using the following equation:

$${\text{S}}_{\text{GIWAXS}}=\frac{1}{2}\left(3<{\text{cos}}^2\chi >-1\right),$$ [2]

where

$$<{\text{cos}}^2\text{χ}>=\frac{{{\displaystyle \int }}_0^{90}\text{I}\left(\text{χ}\right)\left({\text{cos}}^2\chi \right)\left(\text{sin χ}\right)\text{dχ}}{{{\displaystyle \int }}_0^{90}\text{I}\left(\text{χ}\right)\left(\text{sin χ}\right)\text{dχ}}.$$ [3]

The Alq3 peak position from the GIWAXS was obtained by a procedure similar to the one previously described by Bagchi et al. (27); the peak position was determined in this work by fitting a smaller Q range, 0.3 to 1.1 Å⁻¹.

Data Availability

Representative data of all samples employed in this study are included in the article and/or SI Appendix. Additionally, all the data associated with this publication are publicly accessible at http://digital.library.wisc.edu/1793/82341 (56). Previously published data were used for this work [Fig. 1, GIWAXS pattern of DSA-Ph deposited at Tsub = 290 K (39), and Fig. 3B, DSA-Ph birefringence (26)].