Improvements to the photostability of organic glasses for use in electronic applications have generally relied on the modification of the chemical structure. We show here that the photostability of a guest molecule can also be significantly improved—without chemical modification—by using physical vapor deposition to pack molecules more densely. Photoisomerization of the substituted azobenzene, 4,4′-diphenyl azobenzene, was studied in a vapor-deposited glass matrix of celecoxib. We directly measure photoisomerization of trans- to cis-states via Ultraviolet-visible (UV-Vis) spectroscopy and show that the rate of photoisomerization depends upon the substrate temperature used during co-deposition of the glass. Photostability correlates reasonably with the density of the glass, where the optimum glass is about tenfold more photostable than the liquid-cooled glass. Molecular simulations, which mimic photoisomerization, also demonstrate that photoreaction of a guest molecule can be suppressed in vapor-deposited glasses. From the simulations, we estimate that the region that is disrupted by a single photoisomerization event encompasses approximately 5 molecules.

Organic glasses are amorphous materials that have been widely used in modern technologies, including pharmaceuticals,1 organic light emitting diodes (OLEDs),2,3 polymers,4,5 and organic solar cells.6 For many applications, glassy materials are preferred over their crystalline counterparts. For example, in the pharmaceutical industry, drugs are formulated into amorphous solids to achieve higher solubility and bioavailability than can be attained with crystals.7 In OLEDs, vapor-deposited organic glasses are frequently the active layers that transport charge and emit light. For OLEDs, smooth surfaces and lack of grain boundaries are key advantages of vapor-deposited glasses.2 One potential concern when using organic materials is photochemical stability. Photodegradation can cause the failure of organic electronic devices,8,9 and this is sometimes a more limiting factor than device efficiency.10 

Recently, physical vapor deposition (PVD) has been used to prepare single-component glasses that are highly photostable relative to liquid-cooled (LC) or spin-coated glasses prepared from the same molecules.11 It was reported that photoisomerization of an azobenzene derivative (Disperse Orange 37 or DO37) could be slowed by a factor of 50 by using the optimum substrate temperature during vapor deposition.12 In another example, it was shown that vapor-deposited glasses of indomethacin suppress photodecarboxylation rates by a factor of 2 relative to the liquid-cooled glass.13 The highly stable glasses prepared by PVD also have high density11,14–18 and exceptionally high thermal stability.14,19–27 For example, a glass prepared by vapor deposition onto a substrate near 0.85Tg (where Tg is the glass transition temperature) can have a density as much as 1.3% higher than the liquid-cooled glass. In principle, such high density (and high photostability) could also be achieved if a liquid-cooled glass was annealed for thousands of years,16,21,28 but this is obviously not a practical route for material synthesis. For the two examples above,12,13 enhanced photostability was found to correlate with glass density and explained by the constraint applied by tight molecular packing on the local molecular rearrangements needed for photoreaction. These examples from the literature for single component systems raise the question of whether high density and high thermal stability glasses prepared by PVD can also lead to increased photostability for glass mixtures.

For many applications in organic electronics, glassy mixtures are utilized instead of single-component glasses, and photostability is also a concern for these mixtures. For example, in the typical OLED, the light emitting molecule is diluted in a charge transport layer.3 This arrangement both increases emission efficiency and facilitates tuning of the emission wavelength (by changing emitters) without changing the electrical characteristics of the device.29,30 It is known that switching to a more rigid host matrix can increase the photostability of guest emitters,31,32 but often there are numerous constraints on the choice of the host material—such that this is impractical. Therefore, it would be ideal if photostability could be improved without changing the chemical identity of the guest or the host. As mentioned above, annealing a glass can increase density and rigidity, and some experiments indicate increased photostability as a result. For example, for 4,4′-diphenyl azobenzene (DPA) dilutely dispersed in amorphous polycarbonate, Royal and Torkelson compared the photoreaction in the liquid-cooled glass to the glass after 100 h of annealing (or aging). While little impact was seen on the initial rate of photoisomerization, annealing decreased the steady-state level of photoisomerization by around 10%.33 In another system with a probe that requires a larger isomerization volume, similarly long annealing decreased the steady-state level of photoisomerization by an even larger fraction.34 While these examples support the principle that glass packing can modify the photoreactivity of a guest molecule (without changing the guest or host identity), long-term annealing may not be practical in all situations. In any case, we can ask whether there might be even more effective strategies for inhibiting photoreactions in two-component glasses.

In this study, we test whether PVD can be used to prepare mixed glasses in which the photostability of a guest molecule is substantially increased relative to that of the liquid-cooled glass. As a model system, we investigate the photoisomerization of the guest molecule DPA in vapor-deposited and liquid-cooled glasses of celecoxib. Azobenzene molecules can undergo transcis photoisomerization when irradiated by light, as shown in Scheme 1. We first deposited pure celecoxib onto substrates held at different temperatures and successfully obtained glasses with different initial densities and a wide range of thermal stabilities. This result was expected on the basis of recent calorimetric work showing that PVD glasses of celecoxib have low enthalpy and high thermal stability.22 We then co-deposited 5% DPA with 95% celecoxib and prepared glass mixtures that have the same thermal stability and packing as with the neat films. Ultraviolet-visible (UV-Vis) spectroscopy was used to measure the photoisomerization process directly during light irradiation. The long lifetime of the thermal relaxation from cis to trans (∼10 h) allows unambiguous characterization of the initial rate of photoisomerization.

SCHEME 1.

Photoisomerization of 4,4′-diphenylazobenzene (DPA) under UV irradiation.

SCHEME 1.

Photoisomerization of 4,4′-diphenylazobenzene (DPA) under UV irradiation.

Close modal

We find that, relative to the liquid-cooled glass, the photoisomerization rate of the guest DPA can be decreased tenfold by controlling the packing in co-deposited PVD glasses. We observe that photostability is highly correlated with the density of the glass. This result represents an important demonstration that photostability of a dilute guest can be significantly increased via physical vapor deposition, without changing the guest or host chemical identity. For this system, we show that vapor-deposited glasses inhibit photoisomerization to a much greater extent than can be achieved by annealing a liquid-cooled glass for more than 100 h. Molecular simulations of photoisomerization in vapor-deposited glasses capture the key features of the experiments and provide further molecular-level insight into the mechanism of photostability. In particular, the effect of dilution relative to neat films and the impact of an isomerization event on the surrounding host molecules were investigated. From the simulations, we estimate that the region that is disrupted by a single photoisomerization event contains about 5 molecules.

DPA was synthesized as described elsewhere.34 Celecoxib (99%) was purchased from AvaChem Scientific and used as received. Celecoxib was selected as the host for these experiments because it is a good glass former that facilitates glassy thin film preparation via PVD.22 Celecoxib absorbs light in the deep UV region and has no electronic absorbance above 300 nm. Thus, the absorbance of DPA at 370 nm can be observed by UV-Vis spectroscopy without interference, allowing convenient characterization of the trans-cis isomerization.

PVD was performed in a vacuum chamber with the base pressure of 10−7 Torr. DPA and celecoxib were simultaneously evaporated from two different crucibles and thus co-deposited onto substrates. The total deposition rate was 2 Å/s, as monitored by the quartz crystal microbalance (QCM). The deposition rate for each component was controlled by tuning the heating power for each crucible independently. DPA was deposited at a rate of 0.1 Å/s in order to prepare a 5% mixture with the celecoxib host. A high throughput method was utilized to prepare a library of glasses with different densities and thermal stabilities in a single deposition. The substrate was suspended between two copper fingers; each finger was kept at a different temperature to create a temperature gradient across the sample during deposition. The substrate temperature range was from 240 to 340 K. Fused silica substrates were used for photostability measurements (by UV-Vis), and silicon wafers were used for density and thermal stability measurement (by ellipsometry). Film thicknesses were about 1000 nm for films on silica and 300 nm for films on silicon.

A 365 nm UV lamp (Spectroline ENF-240C, 20 nm bandwidth) was used as the light source to induce photoisomerization of DPA. Samples were intermittently illuminated at normal incidence with an irradiance of 40 µW/cm2 at room temperature. Immediately after each illumination period, a UV-Vis absorbance spectrum was recorded. The rate of photoisomerization was analyzed from the changes in the absorbance with illumination time. For the 1000 nm thick films with 5% DPA that we utilized, 60%-70% of the 365 nm light is absorbed.

The density and thermal stability of vapor-deposited glasses were characterized by spectroscopic ellipsometry (JA Woollam, M-2000). (See Fig. 5 below for sample data.) For all ellipsometry measurements, three incident angles were used (50°, 60°, and 70°), and wavelengths from 500 to 1000 nm were used to fit the measurements to an anisotropic Cauchy model. To measure density and thermal stability, ellipsometry was performed on samples placed on a custom-built hot stage, and the temperature was increased at 1 K/min from near room temperature to 350 K (28 K above Tg of celecoxib). The onset temperature, which characterizes the thermal stability of a glass, was determined from the beginning of the transformation into the supercooled liquid. Immediately after heating, the supercooled liquid was cooled at 1 K/min to room temperature to form the liquid-cooled glass. By comparing the sample thickness before and after heating/cooling runs, the density of the vapor-deposited glass relative to that of the liquid-cooled glass was determined.

We also measured the density changes during light irradiation at an isothermal condition. Spectroscopic ellipsometry was used to characterize the glass thickness at the spot where the 365 nm UV irradiated the sample. Relative density was determined from the inverse of the thickness changes induced by light. Sample data are shown in Fig. 4.

In order to better understand photoisomerization in vapor-deposited glasses, molecular simulations were performed using a coarse-grained model of a glass forming host and a photoactive guest. We used a previous simulation study12 as our starting point and made two important changes in order to make the simulations more realistic for the current experiments. In this paragraph, we briefly recap the simulations of Ref. 12 in order to provide context for the current work. The goal of Ref. 12 was to simulate the photoisomerization of vapor-deposited glasses of neat DO37 (an azobenzene derivative). The simulations began with the vapor deposition of a coarse-grained molecule with four linearly connected beads (as shown in Fig. 8) onto a temperature-controlled substrate. At several different substrate temperatures, films were deposited to a thickness of approximately 40σbb, where σbb is the Lennard-Jones interaction radius of each bead in the molecule. After the film was prepared with all molecules in the trans state, the response of the film to photoexcitation was investigated in the “bulk” region of the film, defined as the section of the film between 15σbb and 28σbb in the z-dimension. During the entire simulation, the dihedral angle potentials had the following functional form:

Udihedral=12k11+cosθ+12k21cos2θ.

During deposition, the parameters used for the dihedral potential were k1 = 20 and k2 = 8; this potential strongly favors the trans state. In Ref. 12, photoexcitation was simulated by randomly selecting a small fraction of the molecules and instantaneously switching the dihedral potential to the one that favored the cis state (k1 = −25 and k2 = 6.25).

In comparison to the procedure used in Ref. 12, the simulation of the photoexcitation process was modified in two respects in order to describe the photoisomerization of the guest DPA in a celecoxib matrix. First, after the deposition stage, we randomly selected a subset of the molecules to be “guests;” only this subset could be photoexcited in subsequent stages of the simulation. Second, to account for the much longer cis-lifetime of DPA as compared to DO37, the dihedral potential describing the resting state for the guest molecules was changed (k1 = 10 and k2 = 25) to have a minimum in the cis configuration (θ = 0) as well as the trans configuration (θ = ±180°). For one set of simulations (described below as the “original potential”), when a photoexcitation event occurred, we switch to the same potential used for photoexcitation in Ref. 12 (k1 = −25 and k2 = 6.25). For a second set of simulations, we increased the driving force for isomerization by a factor of 1.5; for these simulations (described below as the “1.5× original potential”), we utilized k1 = −37.5 and k2 = 9.375 when a photoexcitation event occurred. In this way, the effect of isomerization driving force, which mimics photon energy, on reaction kinetics was investigated.

For the simulations reported here, the photoexcitation procedure was adjusted such that, on average, only 1 out of the approximately 1000 bulk molecules was excited within each iteration. For each photoexcitation iteration, the dihedral potential for a selected guest molecule was switched temporarily from the resting state with minima for cis and trans states to the designated excitation potential with a minimum only in the cis state. A short molecular dynamics simulation of a 100 Lennard-Jones (LJ) time units was then performed, which allowed the excited molecules to isomerize to the cis state if the local packing arrangement allowed it. The dihedral potential was then switched back to the resting potential, and the procedure was repeated with new excited molecules randomly selected from the guest molecules. Due to the cis state minimum in the resting dihedral potential, once a molecule successfully isomerized to the cis configuration, it remained cis for the duration of the simulation. (This was not the case in Ref. 12, where the resting dihedral potential favored the trans state.) For the x-axes in Figs. 8 and 9, the “number of photoexcitations” was calculated by using the number of iterations of the photoexcitation process times the average number of photoexcitations per iteration.

All other aspects of the simulations that are not discussed above were performed as described in Ref. 12.

The photostability of mixed glasses of DPA in celecoxib was monitored by UV-Vis spectroscopy. Figure 1 shows the absorption spectra of a glass vapor-deposited at 273 K and the liquid-cooled glass. Upon irradiation with 365 nm light, the absorbance at the peak near 370 nm for both glasses decreased due to photoisomerization from the trans to cis states. The absorbance of the liquid-cooled glass decayed faster at the initial stages relative to the sample deposited at 273 K, indicating less resistance to light-induced photoisomerization.

FIG. 1.

Photoisomerization of glasses containing 5% DPA guest in a matrix of celecoxib, as monitored by absorbance after light irradiation at 365 nm; irradiation times are specified in the legend. Top panel shows the results for a liquid-cooled glass, and the bottom panel shows the results for a glass vapor-deposited at 273 K. At early times, photoisomerization is much faster in the liquid-cooled glass.

FIG. 1.

Photoisomerization of glasses containing 5% DPA guest in a matrix of celecoxib, as monitored by absorbance after light irradiation at 365 nm; irradiation times are specified in the legend. Top panel shows the results for a liquid-cooled glass, and the bottom panel shows the results for a glass vapor-deposited at 273 K. At early times, photoisomerization is much faster in the liquid-cooled glass.

Close modal

Figure 2 shows the comparison of DPA photoisomerization in vapor-deposited and liquid-cooled glasses and reveals that PVD glasses display significantly lower reaction rates, i.e., enhanced photostability. Immediately after irradiation begins, the normalized absorbance at 370 nm starts to decrease for all glasses, indicating trans to cis isomerization. For the liquid-cooled glass, as represented by the black curve, the reaction half-life is about 10 min and the reaction reaches a stationary state after about 100 min irradiation. By contrast, photoisomerization in vapor-deposited glasses occurs more slowly with the reaction rate depending upon the substrate temperature during deposition. For the optimum PVD glass (Tsub = 273 K = 0.85 Tg), the reaction half-life is about 200 min, which is about 20 times longer than that for the liquid-cooled glass with the same composition.

FIG. 2.

Time dependence of photoisomerization for vapor-deposited and liquid-cooled (LC) glasses of 5% DPA in celecoxib, as determined from normalized absorbance at 370 nm. The glasses were prepared at a range of substrate temperatures as indicated. PVD glasses exhibit much slower photoisomerization than the liquid-cooled glass. Solid lines are exponential fits.

FIG. 2.

Time dependence of photoisomerization for vapor-deposited and liquid-cooled (LC) glasses of 5% DPA in celecoxib, as determined from normalized absorbance at 370 nm. The glasses were prepared at a range of substrate temperatures as indicated. PVD glasses exhibit much slower photoisomerization than the liquid-cooled glass. Solid lines are exponential fits.

Close modal

The kinetics of DPA photoisomerization in mixed glasses with celecoxib are summarized in Fig. 3, showing that the photoisomerization rate varies by a factor of ten depending upon the substrate temperature used during deposition. The rate constant k was obtained from exponential fits of the data shown in Fig. 2; for these fits, we only utilized the first 45 min of the data to ensure that we obtained an accurate description of the initial reaction rate. The liquid-cooled glass, which has the highest reaction rate among the glasses, was prepared by vapor-deposition at a temperature above 322 K (Tg) and cooled back to room temperature at 1 K/min. The most photostable glasses were prepared between 247 and 292 K, which corresponds to glasses with the highest initial densities (see discussion below). As a control experiment, the photoisomerization rate of a liquid-cooled glass was measured before and after aging for 140 h at 273 K. As shown in Fig. 3, 140 h of aging leads to a very small increase in photostability. This result indicates that, as a practical matter, the aging of a liquid-cooled glass cannot achieve the large increases in photostability observed for the PVD glasses. The photoisomerization rate of DPA was also measured in ethanol and is shown in Fig. 3. All of these results are consistent with the idea that a well-packed glassy environment can inhibit the photoreaction of a guest molecule.

FIG. 3.

Photostability of vapor-deposited glasses, with comparison to the liquid-cooled (LC) glass and DPA in solution. The logarithm of the photoisomerization rate constant k is plotted against substrate temperature. Blue and red triangles show the influence of aging on a liquid-cooled glass over 140 h at 273 K. The photoisomerization rate in ethanol solution is also indicated for reference (orange).

FIG. 3.

Photostability of vapor-deposited glasses, with comparison to the liquid-cooled (LC) glass and DPA in solution. The logarithm of the photoisomerization rate constant k is plotted against substrate temperature. Blue and red triangles show the influence of aging on a liquid-cooled glass over 140 h at 273 K. The photoisomerization rate in ethanol solution is also indicated for reference (orange).

Close modal

The impact of DPA photoisomerization on the mixed glass structure was also investigated by using ellipsometry to measure the density changes induced by light irradiation, demonstrating that PVD glasses resist density changes more strongly than the liquid-cooled glass. As shown in Fig. 4, vapor-deposited glasses initially have a higher density than the liquid-cooled glass (see below). For all glasses, upon light irradiation, the glass density decreases immediately after irradiation due to the photo-expansion effect.35 The liquid-cooled glass expands quickly after irradiation and reaches a stationary state at about 100 min, after which no further expansion can be observed. Compared to that, vapor-deposited glasses photo-expand more slowly, which can be found by measuring the time required for the density to change by 0.1%. A density change of this magnitude requires about 10 min for the liquid-cooled glass and about 100 min for the glass deposited at 272 K.

FIG. 4.

Density changes during illumination for vapor-deposited and liquid-cooled glasses of celecoxib with 5% DPA, as a function of irradiation time. Density is normalized to the initial density of the liquid-cooled glass. Relative to the liquid-cooled glass, density changes much more slowly during illumination for the most stable PVD glasses.

FIG. 4.

Density changes during illumination for vapor-deposited and liquid-cooled glasses of celecoxib with 5% DPA, as a function of irradiation time. Density is normalized to the initial density of the liquid-cooled glass. Relative to the liquid-cooled glass, density changes much more slowly during illumination for the most stable PVD glasses.

Close modal

For comparison with the photostability results presented above, the density and thermal stability of PVD glasses of neat celecoxib and DPA/celecoxib mixtures were investigated by spectroscopic ellipsometry. Figure 5 shows an example of a temperature-ramping experiment for celecoxib deposited at Tsub = 276 K (0.86 Tg), as characterized by ellipsometry. Three temperature ramping cycles were performed. In the first cycle, the as-deposited celecoxib was heated from 295 K to 350 K at 1 K/min and then cooled back to 295 K. The subsequent heating/cooling runs were performed at the same rate from 295 K to 340 K. During the first heating, the as-deposited sample transforms into the supercooled liquid as indicated by the large increase in thickness over a small temperature range. The onset temperature, which characterizes the thermal stability of the as-deposited glass, was determined from the beginning of the transformation into the supercooled liquid (334 K). During the subsequent cooling, the liquid-cooled glass was formed at the glass transition temperature (Tg) of 322 K. Thickness data from the second and third cooling runs reproduced closely the data from the first cooling run since the preparation details of the deposition process are erased during transformation to the supercooled liquid.

FIG. 5.

Thickness changes obtained by ellipsometry for a vapor-deposited glass of celecoxib during temperature cycling at 1 K/min. Data are shown for a glass vapor-deposited at Tsub = 276 K (0.86 Tg). This PVD glass is more dense than the liquid-cooled glass (formed by heating/cooling) and has greater thermal stability (since Tonset is greater than Tg). The sample thickness is about 300 nm. The inset shows the molecular structure of celecoxib.

FIG. 5.

Thickness changes obtained by ellipsometry for a vapor-deposited glass of celecoxib during temperature cycling at 1 K/min. Data are shown for a glass vapor-deposited at Tsub = 276 K (0.86 Tg). This PVD glass is more dense than the liquid-cooled glass (formed by heating/cooling) and has greater thermal stability (since Tonset is greater than Tg). The sample thickness is about 300 nm. The inset shows the molecular structure of celecoxib.

Close modal

The PVD glass of celecoxib illustrated in Fig. 5 shows high density and high thermal stability. The change in thickness between the first heating and cooling runs is used to determine the density of the as-deposited glass relative to the liquid-cooled glass. In this case, the as-deposited glass is 1.1% more dense. A high value of Tonset (12 K above Tg) is demonstrated during the first heating run, indicating enhanced thermal stability. PVD glasses of celecoxib have properties similar to PVD glasses of other molecules.15 For comparison, indomethacin has been reported to form PVD glasses with Tonset as much as 18 K above Tg and a density increase of up to 1.3% relative to the liquid-cooled glass.19 Figure 6(a) shows the density for celecoxib glasses vapor-deposited at different substrate temperatures relative to the liquid-cooled glass. All the glasses deposited at Tsub < Tg show higher density than the liquid-cooled glass. We attempted to measure the density of DPA/celecoxib mixtures by the method described in Fig. 5 but could not, as the thickness of the transformed sample was not stable in the second and third heating cycles; this is possibly due to desorption of DPA during the ramping experiment. Given the small amount of DPA present in the mixed glasses, we assume that the relative densities of the mixtures are the same as the neat films shown in Fig. 6(a). This assumption was used in the construction of Fig. 4.

FIG. 6.

Summary of (a) relative density, (b) onset temperature, and (c) birefringence (nz − nxy) of vapor-deposited neat films of celecoxib (blue points) and mixtures with 5% DPA (red points), as characterized by ellipsometry. Liquid-cooled (LC) glasses were deposited at a substrate temperature higher than Tg and then cooled into the glassy state. Horizontal black dashed lines in each panel represent the corresponding measurement for the LC glass.

FIG. 6.

Summary of (a) relative density, (b) onset temperature, and (c) birefringence (nz − nxy) of vapor-deposited neat films of celecoxib (blue points) and mixtures with 5% DPA (red points), as characterized by ellipsometry. Liquid-cooled (LC) glasses were deposited at a substrate temperature higher than Tg and then cooled into the glassy state. Horizontal black dashed lines in each panel represent the corresponding measurement for the LC glass.

Close modal

We find that co-deposition of 5% DPA with celecoxib does not significantly change the thermal stability and structure of PVD glasses in comparison with pure celecoxib glasses. To illustrate this, comparisons of the onset temperature and birefringence for vapor-deposited glasses of celecoxib and DPA/celecoxib mixtures are shown in Figs. 6(b) and 6(c). Glasses deposited above Tg (and subsequently cooled to room temperature) are regarded as liquid-cooled glasses, and this is consistent with the observation that Tonset = Tg = 322 K for these materials. For Tsub lower than Tg, enhanced thermal stability was observed. Glasses with the optimum thermal stability were prepared at Tsub between 0.85 and 0.96 Tg. The increased thermal stability and enhanced density of these vapor-deposited glasses are attributed to enhanced surface mobility during glass formation.36,37 Birefringence of vapor-deposited glasses, indicative of molecular orientation, is shown in Fig. 6(c). Above Tg, the zero birefringence indicates random molecular orientation, which is consistent with the isotropic structure that a liquid-cooled glass inherits from the supercooled liquid. Below Tg, birefringence is not zero and depends upon the value of Tsub. This non-zero birefringence has been reported for many other PVD glasses.17,21,38,39 Both onset temperatures and birefringence data indicate that glass packing of neat celecoxib films and DPA/celecoxib mixtures is similar.

We can now compare the DPA photoisomerization rates (Fig. 3) with the densities for different glasses [Fig. 6(a)]. Figure 7 shows a reasonably strong correlation between the photostability of the DPA guest and the glass density. Increased glass density is associated with a decreased photoisomerization rate. DPA in the liquid-cooled glass photoisomerizes about 10 times faster than it does in the glasses with the highest density. As discussed later, this result is consistent with the view that higher density plays an important role in suppressing photoreaction. The behavior of the aged liquid-cooled glass is qualitatively consistent with this finding. The density change after 140 h of aging was not measured but is expected to be a few tenths of a percent; this small density change would be consistent with the trend shown in Fig. 7, given the slightly lower isomerization rates for the aged sample shown in Fig. 3. It is likely that factors in addition to density play some role in determining photostability as the correlation in Fig. 7 is not perfect. For example, the glasses vapor-deposited at 246 and 260 K have different photostabilities even though they have very similar densities and very similar birefringences, and in addition, the DPA molecules in these films have similar average orientations (based upon initial absorbance). In this regard, photostability is similar to the kinetic stability of vapor-deposited glasses in that the correlation with glass density is strong but not perfect.15,19

FIG. 7.

Correlation of the photoisomerization rate constant with the density of vapor-deposited glasses of the celecoxib host. The orange solid line is the linear fit.

FIG. 7.

Correlation of the photoisomerization rate constant with the density of vapor-deposited glasses of the celecoxib host. The orange solid line is the linear fit.

Close modal

Molecular simulations of vapor-deposition and photoisomerization were performed to provide insight into the enhanced photostability in co-deposited PVD glasses. As explained earlier, the vapor deposition and photoisomerization aspects of the simulation are similar to the method used in a previous study of DO3712 where, to examine the packing effect on the photoisomerization process, DPA was represented by a coarse-grained linearly connected molecule with four beads (inset of Fig. 8). During the vapor deposition process, coarse-grained DPA molecules were held in the trans state and deposited onto substrates at temperatures ranging from 0.76 to 0.97Tg, where Tg was determined to be 0.66 in reduced Lennard-Jones (LJ) units. Consistent with the experimental results, the simulated PVD glasses had a higher density than the corresponding liquid-cooled glass (supplementary material, Fig. S1) when deposited at substrate temperatures below Tg. For the photoisomerization process, we mimic the dilution of DPA in the experiments by randomly selecting 5% of deposited molecules to be guests; only this subset can photoreact. For each photoexcitation cycle, guest molecules are selected with a small probability to be excited; this occurs by switching dihedral potentials such that a force acts to push the molecule from trans to cis. On average, only one guest is selected for excitation during each photoexcitation cycle. Whether or not the cis state is actually attained is determined by the local packing environment. Guest molecules that reach the cis state during photoexcitation remain there for the duration of the simulation, consistent with the long thermal relaxation time for the cis state of DPA.

FIG. 8.

Photoisomerization from trans to cis states as observed in molecular simulations for a mixed glass (5% isomerizable guests), for glasses vapor-deposited at different substrate temperatures. Two different driving forces for isomerization were studied: the solid lines were obtained with the potential utilized in Ref. 12 and the dashed lines represent results for a dihedral potential that has a 50% larger driving force for photoisomerization. Representative structures of the coarse-grained model in trans and cis states are shown in the inset. For both dihedral potentials, vapor-deposited glasses show much slower trans-cis photoisomerization.

FIG. 8.

Photoisomerization from trans to cis states as observed in molecular simulations for a mixed glass (5% isomerizable guests), for glasses vapor-deposited at different substrate temperatures. Two different driving forces for isomerization were studied: the solid lines were obtained with the potential utilized in Ref. 12 and the dashed lines represent results for a dihedral potential that has a 50% larger driving force for photoisomerization. Representative structures of the coarse-grained model in trans and cis states are shown in the inset. For both dihedral potentials, vapor-deposited glasses show much slower trans-cis photoisomerization.

Close modal

Figure 8 shows that simulated glasses prepared by PVD are substantially more photostable than the liquid-cooled glass. For each glass, the fraction of guests isomerized is shown as a function of the number of photoexcitation events. We performed these simulations using two different dihedral potentials for the photoexcitation step, in order to investigate the effect of the driving force on the photoisomerization kinetics. Figure 8 shows the results both for the potential utilized in Ref. 12 (the “original potential”) and the one in which the driving force is 50% larger (“1.5× original potential”).

For the liquid-cooled glass driven by the original potential, the cis fraction of the guest molecules increased from zero to 0.7 after 800 photoexcitations. After the same number of photoexcitations, the fraction of guests that successfully reached cis for the most photostable glass (Tsub = 0.76 Tg) was only 0.04. The simulation result indicates that the initial molecular packing of the vapor-deposited glasses suppresses the ability of excited molecules to change from trans to cis and that the substrate temperature during deposition has a strong impact on photostability. Both of these observations are qualitatively consistent with the experimental results in Fig. 2.

By comparing results from the two different dihedral potentials used in the photoexcitation step, we find that the rate of photoisomerization is significantly increased when a stronger driving force is utilized. As shown in Fig. 8, the optimum glass has a cis fraction of 0.24 after 800 photoexcitation events with the stronger driving force; this is 6 times the cis fraction resulting from the original potential from Ref. 12. We use this result below to compare experiments on azobenzene derivatives that are excited with different photon energies.

The effect of the concentration of guest molecules in mixtures was also investigated in the simulations revealing that, for a given number of photoexcitations, a higher guest concentration leads to greater disruption of glass packing. As shown in Fig. 9, glass density decreases as the number of photoexcitations increases due to the photo-expansion effect, which is consistent with the experimental results in Fig. 4. The density decrease is faster when the concentration of isomerizable guests is larger. For example, after 800 photoexcitations, the sample with the lowest azobenzene concentration (0.03) only photo-expands by 0.35%, while the density of the film in which every molecule is photoisomerizable decreases by 3.4%. Note that in this comparison, the total number of photoexcitation events is the same for glasses with different guest concentrations. Figure 9 shows that when these photoexcitation events are concentrated in a small number of guest molecules, the overall glass density is minimally affected.

FIG. 9.

Simulation results showing photo-induced density changes for glasses with different fractions of isomerizable guests. All glasses were deposited at Tsub = 0.76 Tg. For the same number of excitation events, the density of the glass changes more quickly when the guest fraction is larger.

FIG. 9.

Simulation results showing photo-induced density changes for glasses with different fractions of isomerizable guests. All glasses were deposited at Tsub = 0.76 Tg. For the same number of excitation events, the density of the glass changes more quickly when the guest fraction is larger.

Close modal

By comparing the light-induced density decrease for films in which a different fraction of the molecules is photoisomerizable, we can estimate the extent to which a photoisomerization event disrupts the packing of its surroundings. Figure 10 shows how the density change after 800 photoexcitation events varies as a function of the fraction of photoisomerizable guests. Two regimes of linear dependence can be observed, with the rate of increase becoming slower when the guest fraction surpasses 0.2. We interpret the transition of density disruption to indicate the size of the region that can be influenced by each photoisomerizing guest. At the guest/host ratio of 1:4, the local regions that have been influenced by photoisomerization start to overlap and each isomerizing guest molecule starts to feel the existence of its neighbors.

FIG. 10.

Density change in simulations after 800 photoexcitation events for glasses with different guest fractions. Solid lines are guide to the eye. Density disruption becomes larger as DPA fraction increases, and the transition point indicates the size of the region that can be influenced by each photoisomerizing guest. All the glasses shown were prepared at Tsub = 0.76 Tg.

FIG. 10.

Density change in simulations after 800 photoexcitation events for glasses with different guest fractions. Solid lines are guide to the eye. Density disruption becomes larger as DPA fraction increases, and the transition point indicates the size of the region that can be influenced by each photoisomerizing guest. All the glasses shown were prepared at Tsub = 0.76 Tg.

Close modal

We have shown that photoreaction of a guest molecule can be suppressed by one order of magnitude in a molecular glass mixture by more densely packing the glass, without altering the chemical structures of the guest or host. The strong correlation between glass density and photostability in a two-component system identified here had not been established before, and we therefore anticipate that the results reported in this work will help guide future efforts to increase the lifetime of organic electronic devices. In these experiments, photostability is highly dependent on the substrate temperature used during vapor deposition. For substrate temperatures lower than Tg, the photostability of the DPA guest in PVD glasses with celecoxib can be increased by up to a factor of 12. We also show that photostability correlates with glass density, with higher density glasses being more resistant to reactions. Molecular simulations reproduce important features observed in our experiments, such as enhanced photostability for guests in high-density PVD glasses. Moreover, the simulations provide the means to investigate the effect of driving force (photon energy) and guest concentration and provide insight into the mechanism by which glass packing can suppress photoisomerization.

Compared to other glass preparation methods, such as cooling from the liquid and aging, PVD provides a route to obtain glasses with substantially increased photostability within reasonably short times. Because we used DPA as the guest, a direct comparison can be made between our results and a previous investigation of DPA photoisomerization in aged polymeric glasses. While we show here that photoisomerization of the DPA guest can be slow down by a factor of 10 in vapor-deposited hosts, no significant changes in the initial photoisomerization rate were observed for DPA dispersed in polycarbonate and poly(methyl methacrylate) (PMMA) glasses that had been physically aged for 100 h.33 The authors did report that the steady-state extent of photoisomerization decreased by about 14% after physically aging polycarbonate at 85 °C33 and by about 12% after physically aging PMMA at 60 °C.34 The steady-state extent of photoisomerization of different DPA/celecoxib glasses could not be fairly compared in the present study because molecular orientation in vapor-deposited glasses is generally anisotropic, varies with substrate temperature, and changes as a result of photoisomerization. Thus absorbance is influenced by the initial orientations of the transition dipoles in addition to the guest concentration. It is also interesting to note that, through physical aging, a very large suppression of the extent of photoisomerization has been reported for at least one system.34 For a different guest molecule, diphenyl stilbene (DPS), that requires a larger volume to isomerize, the extent of photoisomerization was reported to be suppressed by about 80% in PMMA after 100 h aging.

In previous studies of one-component PVD glasses, it has been observed that photostability is correlated with glass density. Photostability of vapor-deposited DO37, an azobenzene derivative, was characterized by measuring density and molecular orientation changes during irradiation.12 It was observed that structural changes associated with photoisomerization were suppressed by a factor of 50 by a density increase of 1.3%. In a second example, that of an azobenzene derivative tethered to a PMMA polymer, it was demonstrated that optically induced molecular rearrangement can be hindered by high-density glasses created by high external pressure.40 The glass density at 150 MPa was increased by 2.4%, and the rate of photo-orientation decreased by a factor of nearly 50. These two examples show a strong correlation between glass density and photostability for single-component glasses. The work presented here extends this conclusion to guest stability in a two-component glass.

A significant improvement in our current experiment relative to previous work on vapor-deposited DO37 glasses is that here we provide direct experimental evidence that photoisomerization is blocked by dense packing. In the earlier work with DO37, photostability was assessed by measuring the density and molecular orientation changes induced by light; the impact of glass packing on the trans to cis reaction rate could not be directly monitored.12 The work presented here allows us to directly observe the trans-cis photoreaction by measuring UV-Vis absorbance. In light of these new results, we think that trans-cis isomerization was also blocked in the previous study of DO37, and this conclusion is consistent with the computer simulations reported for that system.12 We imagine that this result is quite general, and so we expect that photoreactions in both one- and two-component glasses can generally be suppressed by preparing high-density glasses by PVD. An issue not resolved by this work is whether density is the glass property that best correlates with increased photostability. For example, for cases where it has been investigated, high-density PVD glasses also have high moduli.41 It is possible that the moduli might correlate with enhanced photostability even better than the density. This would be reasonable as a high modulus glass would be expected to better resist the driving force for trans to cis isomerization.42 

This study provides an important opportunity to investigate the effect of different excitation wavelengths on photoisomerization in glasses. For the DPA/celecoxib experiments presented here, we observe immediate photoisomerization and photo-expansion in Figs. 2 and 4, respectively. By contrast, in our previous DO37 experiment, for the most stable glass, hardly any change in density was observed even after all the molecules had been excited 300 times. In part, we attribute these observations to the fact that DPA was excited by 365 nm light while DO37 was excited by 532 nm light. Since 365 nm photons are 45% more energetic than 532 nm photons, this provides a larger driving force for photoisomerization. Simulation results in Fig. 8 are consistent with this conclusion. When the driving force used to mimic the photoexcitation increases by a factor of 1.5, much larger rates of photoisomerization are observed for the most stable glass.

When a molecule in a glass isomerizes, we expect that the packing of the surrounding molecules is altered, and these simulation results provide an opportunity to estimate the size of this disrupted region. Figure 10 shows that the impact of guest concentration on the disruption of glass structure undergoes a transition when the guest fraction exceeds 20%. We interpret this result to indicate that, when the guest concentration is less than 20%, each photoisomerization of a guest molecule occurs nearly independently of the other guests. When the guest fraction is larger, the regions that are disrupted by photoisomerization overlap, and therefore, the impact of each photoisomerizing guest on the overall density will be diminished. From this interpretation, we infer that each photoisomerizing guest molecule influences a region of the glass that, on average, also includes four host molecules.

Finally, we note that the work presented here builds upon previous vapor-deposition studies of neat celecoxib by Rodriguez-Viejo and co-workers.22 Using calorimetry, these authors reported that glasses of celecoxib with very high thermal stability and low enthalpy were obtained for substrate temperatures near 275 K. These results are in reasonable accord with the ellipsometry results presented in Fig. 6.

We have shown that photoisomerization of guest molecules can be significantly inhibited by improving the packing in a glass without any changes in the chemical identity of the guest or host. Photoisomerization of the guest DPA in a matrix of celecoxib can be suppressed tenfold relative to the liquid-cooled glass by selecting the optimum conditions for vapor deposition. Photostability is observed to correlate reasonably with density, where the tighter packing exerts stronger constraints on the molecular rearrangements required for trans-cis isomerization.

Our work suggests that substrate temperature is an important control variable for optimizing the performance of the multicomponent glassy layers used in organic electronic devices. A previous study has revealed that, for emitters doped in the host materials used for OLEDs, solution-coated films are more susceptible to degradation relative to vapor-deposited glasses.43 Our work has demonstrated that the photostability of two-component vapor-deposited glasses can be further improved by optimizing the glass density. So far, more than thirty molecules have been made into single-component high-density glasses by PVD, and several of these molecules are used as host materials for OLEDs.15 Thus, we expect that the suppression of photoreaction shown in the current work will also be observed for emitter/host glassy layers used in OLEDs. This method for increasing resistance to photodegradation could be important as operational lifetime is considered to be a bottleneck to the further improvement of OLED display performance, especially for blue emitters.10 There are recent reports that the electronic and optical properties of films of organic semiconductors can be more stable over time if vapor deposition conditions are optimized to prepare the highest density glasses.44,45 Our results clearly show the importance of glass density in suppressing photoreactions and, with these other recent results, suggest that optimizing the substrate temperature utilized in PVD is an effective method of extending device lifetime.

See supplementary material for density of vapor-deposited glasses obtained from computer simulations.

The experimental work was supported by the U.S. Department of Energy (DOE), Office of Basic Energy Sciences (BES), Division of Materials Sciences and Engineering, Award No. DE-SC0002161 (to M.D.E. and Y.Q.) and by funds associated with a Walter P. Murphy Professorship (to J.M.T.). The simulations were also supported by the Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division. Additional support from the U.S. Army Research Office through the MURI Program No. W911NF-15-1-0568 for development of light-responsive materials is gratefully acknowledged (J.J.d.P.).

1.
B. C.
Hancock
and
G.
Zografi
, “
Characteristics and significance of the amorphous state in pharmaceutical systems
,”
J. Pharm. Sci.
86
,
1
(
1997
).
2.
D.
Yokoyama
, “
Molecular orientation in small-molecule organic light-emitting diodes
,”
J. Mater. Chem.
21
,
19187
(
2011
).
3.
K.
Sei-Yong
,
J.
Won-Ik
,
M.
Christian
,
P.
Young-Seo
,
K.
Kwon-Hyeon
,
L.
Jeong-Hwan
,
M.
Chang-Ki
,
B.
Wolfgang
, and
K.
Jang-Joo
, “
Organic light-emitting diodes with 30% external quantum efficiency based on a horizontally oriented emitter
,”
Adv. Funct. Mater.
23
,
3896
(
2013
).
4.
G. B.
McKenna
and
S. L.
Simon
, “
50th anniversary perspective: Challenges in the dynamics and kinetics of glass-forming polymers
,”
Macromolecules
50
,
6333
(
2017
).
5.
Y.
Joo
,
V.
Agarkar
,
S. H.
Sung
,
B. M.
Savoie
, and
B. W.
Boudouris
, “
A nonconjugated radical polymer glass with high electrical conductivity
,”
Science
359
,
1391
(
2018
).
6.
M.
Helgesen
,
R.
Sondergaard
, and
F. C.
Krebs
, “
Advanced materials and processes for polymer solar cell devices
,”
J. Mater. Chem.
20
,
36
(
2010
).
7.
B. C.
Hancock
and
M.
Parks
, “
What is the true solubility advantage for amorphous pharmaceuticals?
,”
Pharm. Res.
17
,
397
(
2000
).
8.
Q.
Wang
,
Y.
Luo
, and
H.
Aziz
, “
Photodegradation of the organic/metal cathode interface in organic light-emitting devices
,”
Appl. Phys. Lett.
97
,
063309
(
2010
).
9.
P.
Cheng
and
X.
Zhan
, “
Stability of organic solar cells: Challenges and strategies
,”
Chem. Soc. Rev.
45
,
2544
(
2016
).
10.
Y.
Zhang
,
J.
Lee
, and
S. R.
Forrest
, “
Tenfold increase in the lifetime of blue phosphorescent organic light-emitting diodes
,”
Nat. Commun.
5
,
5008
(
2014
).
11.
M. D.
Ediger
, “
Perspective: Highly stable vapor-deposited glasses
,”
J. Chem. Phys.
147
,
210901
(
2017
).
12.
Y.
Qiu
,
L. W.
Antony
,
J. J.
de Pablo
, and
M. D.
Ediger
, “
Photostability can be significantly modulated by molecular packing in glasses
,”
J. Am. Chem. Soc.
138
,
11282
(
2016
).
13.
Y.
Qiu
,
S. S.
Dalal
, and
M. D.
Ediger
, “
Vapor-deposited organic glasses exhibit enhanced stability against photodegradation
,”
Soft Matter
14
,
2827
(
2018
).
14.
S. F.
Swallen
,
K. L.
Kearns
,
M. K.
Mapes
,
Y. S.
Kim
,
R. J.
McMahon
,
M. D.
Ediger
,
T.
Wu
,
L.
Yu
, and
S.
Satija
, “
Organic glasses with exceptional thermodynamic and kinetic stability
,”
Science
315
,
353
(
2007
).
15.
S. S.
Dalal
,
D. M.
Walters
,
I.
Lyubimov
,
J. J.
de Pablo
, and
M. D.
Ediger
, “
Tunable molecular orientation and elevated thermal stability of vapor-deposited organic semiconductors
,”
Proc. Natl. Acad. Sci. U. S. A.
112
,
4227
(
2015
).
16.
K. L.
Kearns
,
P.
Krzyskowski
, and
Z.
Devereaux
, “
Using deposition rate to increase the thermal and kinetic stability of vapor-deposited hole transport layer glasses via a simple sublimation apparatus
,”
J. Chem. Phys.
146
,
203328
(
2017
).
17.
T.
Liu
,
A. L.
Exarhos
,
E. C.
Alguire
,
F.
Gao
,
E.
Salami-Ranjbaran
,
K.
Cheng
,
T.
Jia
,
J. E.
Subotnik
,
P. J.
Walsh
,
J. M.
Kikkawa
, and
Z.
Fakhraai
, “
Birefringent stable glass with predominantly isotropic molecular orientation
,”
Phys. Rev. Lett.
119
,
095502
(
2017
).
18.
K.
Ishii
,
H.
Nakayama
,
S.
Hirabayashi
, and
R.
Moriyama
, “
Anomalously high-density glass of ethylbenzene prepared by vapor deposition at temperatures close to the glass-transition temperature
,”
Chem. Phys. Lett.
459
,
109
(
2008
).
19.
S. S.
Dalal
,
Z.
Fakhraai
, and
M. D.
Ediger
, “
High-throughput ellipsometric characterization of vapor-deposited indomethacin glasses
,”
J. Phys. Chem. B
117
,
15415
(
2013
).
20.
M. S.
Beasley
,
M.
Tylinski
,
Y. Z.
Chua
,
C.
Schick
, and
M. D.
Ediger
, “
Glasses of three alkyl phosphates show a range of kinetic stabilities when prepared by physical vapor deposition
,”
J. Chem. Phys.
148
,
174503
(
2018
).
21.
T.
Liu
,
K.
Cheng
,
E.
Salami-Ranjbaran
,
F.
Gao
,
C.
Li
,
X.
Tong
,
Y.-C.
Lin
,
Y.
Zhang
,
W.
Zhang
,
L.
Klinge
,
P. J.
Walsh
, and
Z.
Fakhraai
, “
The effect of chemical structure on the stability of physical vapor deposited glasses of 1,3,5-triarylbenzene
,”
J. Chem. Phys.
143
,
084506
(
2015
).
22.
C.
Rodriguez-Tinoco
,
M.
Gonzalez-Silveira
,
J.
Rafols-Ribe
,
G.
Garcia
, and
J.
Rodriguez-Viejo
, “
Highly stable glasses of celecoxib: Influence on thermo-kinetic properties, microstructure and response towards crystal growth
,”
J. Non-Cryst. Solids
407
,
256
(
2015
).
23.
E.
Leon-Gutierrez
,
G.
Garcia
,
M. T.
Clavaguera-Mora
, and
J.
Rodriguez-Viejo
, “
Glass transition in vapor deposited thin films of toluene
,”
Thermochim. Acta
492
,
51
(
2009
).
24.
E.
Leon-Gutierrez
,
A.
Sepúlveda
,
G.
Garcia
,
M. T.
Clavaguera-Mora
, and
J.
Rodríguez-Viejo
, “
Stability of thin film glasses of toluene and ethylbenzene formed by vapor deposition: An in situ nanocalorimetric study
,”
Phys. Chem. Chem. Phys.
12
,
14693
(
2010
).
25.
U.
Cubeta
,
D.
Bhattacharyaa
, and
V.
Sadtchenko
, “
Communication: Surface-facilitated softening of ordinary and vapor-deposited glasses
,”
J. Chem. Phys.
147
,
071101
(
2017
).
26.
S. L. L. M.
Ramos
,
M.
Oguni
,
K.
Ishii
, and
H.
Nakayama
, “
Character of devitrification, viewed from enthalpic paths, of the vapor-deposited ethylbenzene glasses
,”
J. Phys. Chem. B
115
,
14327
(
2011
).
27.
D.
Bhattacharya
and
V.
Sadtchenko
, “
Vapor-deposited non-crystalline phase vs ordinary glasses and supercooled liquids: Subtle thermodynamic and kinetic differences
,”
J. Chem. Phys.
142
,
164510
(
2015
).
28.
K. L.
Kearns
,
S. F.
Swallen
,
M. D.
Ediger
,
T.
Wu
,
Y.
Sun
, and
L.
Yu
, “
Hiking down the energy landscape: Progress toward the Kauzmann temperature via vapor deposition
,”
J. Phys. Chem. B
112
,
4934
(
2008
).
29.
J.
Vygintas
,
D.
Przemyslaw
,
G.
David
,
M.
Callum
,
S.
Jose
,
M. R.
Bryce
,
F. B.
Dias
, and
A. P.
Monkman
, “
Highly efficient TADF OLEDs: How the emitter–host interaction controls both the excited state species and electrical properties of the devices to achieve near 100% triplet harvesting and high efficiency
,”
Adv. Funct. Mater.
24
,
6178
(
2014
).
30.
Y.
Tao
,
C.
Yang
, and
J.
Qin
, “
Organic host materials for phosphorescent organic light-emitting diodes
,”
Chem. Soc. Rev.
40
,
2943
(
2011
).
31.
G.
Valverde-Aguilar
, “
Photostability of laser dyes incorporated in formamide SiO2 ORMOSILs
,”
Opt. Mater.
28
,
1209
(
2006
).
32.
R.
Sastre
and
A.
Costela
, “
Polymeric solid-state dye lasers
,”
Adv. Mater.
7
,
198
(
1995
).
33.
J. S.
Royal
and
J. M.
Torkelson
, “
Photochromic and fluorescent probe studies in glassy polymer matrices. 5. Effects of physical aging on bisphenol a polycarbonate and poly(vinyl acetate) as sensed by a size distribution of photochromic probes
,”
Macromolecules
25
,
4792
(
1992
).
34.
J. S.
Royal
,
J. G.
Victor
, and
J. M.
Torkelson
, “
Photochromic and fluorescent probe studies in glassy polymer matrices. 4. Effects of physical aging on poly(methyl methacrylate) as sensed by a size distribution of photochromic probes
,”
Macromolecules
25
,
729
(
1992
).
35.
J.
Harada
and
K.
Ogawa
, “
X-ray diffraction analysis of nonequilibrium states in crystals: Observation of an unstable conformer in flash-cooled crystals
,”
J. Am. Chem. Soc.
126
,
3539
(
2004
).
36.
L.
Zhu
,
C. W.
Brian
,
S. F.
Swallen
,
P. T.
Straus
,
M. D.
Ediger
, and
L.
Yu
, “
Surface self-diffusion of an organic glass
,”
Phys. Rev. Lett.
106
,
256103
(
2011
).
37.
S.
Ruan
,
W.
Zhang
,
Y.
Sun
,
M. D.
Ediger
, and
L.
Yu
, “
Surface diffusion and surface crystal growth of tris-naphthyl benzene glasses
,”
J. Chem. Phys.
145
,
064503
(
2016
).
38.
D. M.
Walters
,
L.
Antony
,
J. J.
de Pablo
, and
M. D.
Ediger
, “
Influence of molecular shape on the thermal stability and molecular orientation of vapor-deposited organic semiconductors
,”
J. Phys. Chem. Lett.
8
,
3380
(
2017
).
39.
J.
Gómez
,
A.
Gujral
,
C.
Huang
,
C.
Bishop
,
L.
Yu
, and
M. D.
Ediger
, “
Nematic-like stable glasses without equilibrium liquid crystal phases
,”
J. Chem. Phys.
146
,
054503
(
2017
).
40.
Z.
Sekkat
,
G.
Kleideiter
, and
T.
Knoll
, “
Optical orientation of azo dye in polymer films at high pressure
,”
J. Opt. Soc. Am. B
18
,
1854
(
2001
).
41.
K. L.
Kearns
,
S.
Tim
,
F.
George
, and
M. D.
Ediger
, “
High-modulus organic glasses prepared by physical vapor deposition
,”
Adv. Mater.
22
,
39
(
2010
).
42.
J. C.
Dyre
, “
Colloquium: The glass transition and elastic models of glass-forming liquids
,”
Rev. Mod. Phys.
78
,
953
(
2006
).
43.
Y. J.
Cho
and
H.
Aziz
, “
Root causes of the limited electroluminescence stability of organic light-emitting devices made by solution-coating
,”
ACS Appl. Mater. Interfaces
10
,
18113
(
2018
).
44.
Y.
Esaki
,
T.
Komino
,
T.
Matsushima
, and
C.
Adachi
, “
Enhanced electrical properties and air stability of amorphous organic thin films by engineering film density
,”
J. Phys. Chem. Lett.
8
,
5891
(
2017
).
45.
J.
Ràfols-Ribé
,
P.-A.
Will
,
C.
Hänisch
,
M.
Gonzalez-Silveira
,
S.
Lenk
,
J.
Rodríguez-Viejo
, and
S.
Reineke
, “
High-performance organic light-emitting diodes comprising ultrastable glass layers
,”
Sci. Adv.
4
,
eaar8332
(
2018
).

Supplementary Material