betaFeedback

WorksPhysics and Astronomy

Intermolecular and Interfacial Origins of Layering in a Near-Isotropic Organic Glass Former

Abstract

Physical vapor deposition can produce organic glasses with enhanced stability and pronounced structural anisotropy (layering) even when the constituent molecules are nearly spherical. However, the microscopic origin of the layering in spherical molecular glass formers remains unclear. We investigated vapor-deposited films of 9-(3,5-di(naphthalen-1-yl)phenyl)anthracene (α,α-A) by grazing-incidence wide-angle X-ray scattering, atomistic molecular dynamics, and coarse-grained simulations. Experimentally, the layering signal varies nonmonotonically with deposition temperature and is strongest near 0.8T_g. Simulated scattering calculations reproduce the principal experimental features and show that the layering peak does not originate from shifts in single molecule degrees of freedom. In fact, the simulated deposited and bulk glasses show largely the same single-molecule conformations. Interestingly, a one-site coarse-grained liquid with isotropic interactions develops a comparable interfacial layering signature. These results indicate that layering in α,α-A arises primarily from collective density correlations generated at the free surface during deposition rather than from conformational bias or orientational ordering. The findings connect structural anisotropy in vapor-deposited molecular glasses to the generic interfacial organization of simple liquids.

Introduction

Organic light-emitting diodes (OLEDs) are critical components in modern phones and consumer electronics. Small organic molecules, which transfer negative and positive charges to the emissive layer, are used to create electron and hole transport layers in OLEDs.[1] This charge transport is governed not only by the single molecule electronic structure but also by the alignment of molecules relative to each other, where strong alignment can facilitate orders of magnitude increases in charge transport fluxes.[2, 3] Strong molecular alignment often manifests as anisotropy, or the dependence of the structure and performance on the direction of the material. Thus, understanding the fundamental physics driving this anisotropy could be used to more effectively engineer OLEDs.

Physical vapor deposition (PVD) is a popular method to produce anisotropic glass films with enhanced electronic properties.[4, 5] In PVD, evaporated molecules are deposited onto a heated substrate (TdepT_{\mathrm{dep}}). The presence of this substrate has also been connected to orientational order in films made of nonspherical molecules,[3, 6, 7] and even phase separation.[8] The process leaves a heated free surface exposed to vacuum, where the molecules can reorganize differently than in the bulk.[9] Simulation work has shown that layers can form at the free surface[10] and that the orientation of nonspherical molecules is a feature of equilibrium supercooled liquids,[11, 12] suggesting that anisotropic order could arise from equilibrium effects that are buried during deposition. Hence understanding the general trends in how molecular shape and deposition conditions affect the resulting structure of the film has been a central goal of the field.[5, 13, 14]

Still, much of this work still does not explain the formation of layered structures in near-spherical molecules, which do not exhibit the orientational preferences seen in highly-anisotropic glass formers. Here, we combine experiment and simulation to investigate the deposited structure of a model isotropic glass former, 9-(3,5-di(naphthalen-1-yl)phenyl)anthracene (α,α\alpha,\alpha-A), Tg=364T_g = 364 K, shown in Figure 1. This molecule possesses equal length and width, but the steric repulsion of the large aromatic groups produce frustrated intermolecular interactions. Photoluminescence experiments have shown that this molecule behaves isotropically.[15] However, in PVD films of α,α\alpha,\alpha-A spherical ellipsometry[15] and grazing incident X-ray scattering experiments reveal optical anisotropy[14], showing that birefringence can be due to orientation effects or due to packing in the process of PVD.

Figure 1. The chemical structure of α\alpha,α\alpha-A with characteristic dimensions shown.

We seek to determine how the intramolecular degrees of freedom and center of mass ordering interact to create molecularly layered films. To do so, we analyze the structure of glasses deposited at different temperatures using grazing incidence wide angle X-ray scattering (GIWAXS), which provides nanometer level resolution of the experimental structure which we compare with molecular dynamics (MD) simulations at the all-atom and coarse-grained level. From these simulations, we compute how the interactions at different length scales contribute to the structure.

Our results indicate that the layering of spherical molecules does not impose detectable shifts in the single molecule conformations, and that the total structure is well described by intermolecular correlations of the functional groups and liquid-like centers of mass ordering. We test this finding by comparing results against a highly-coarse-grained equilibrium model of the surface, and find surprisingly good agreement. This result extends previous understanding of the enhanced surface mobility of deposited glasses by identifying an underlying thermodynamic minimum based on a highly general liquid model.

Results

Structural features of the experimental scattering signal

The experimental GIWAXS intensity plot shown in Figure 2a, corresponding to Tdep=303T_{\mathrm{dep}} = 303 K =0.83Tg= 0.83T_g, reveals two prominent signals. The first is the halo at q=1.1q = 1.1-1.4 Å−1^{-1}. The lack of direction dependence indicates that the structural features are isotropic. The real length scale of this signal is about 4.5-5.7 Å, which is the characteristic length scale of π−π\pi-\pi interactions. The second is the strong signal at qz=q_z = 0.5-0.6 Å−1^{-1} concentrated on the qzq_z axis. This anisotropic signal represents strong ordering orthogonal to the substrate at a length scale of about 10-12 Å. The similarity of this distance with the molecular size shown in Figure 1 indicates the molecules of α,α-A form sheets across the substrate with the sheets then stacked on top of each other. This phenomenon is also known as layering, with the anisotropic peak in the GIWAXS signal termed the “layering peak.” Similar layering peaks have been observed in vapor deposited films of spherical and nonspherical molecules.[13, 14, 16, 17]

Figure 2. X-ray scattering results of α,α\alpha,\alpha-A. (a) Experimental 2D GIWAXS image showing the qzq_z vs qrq_r of a vapor-deposited 310 nm film of α,α\alpha,\alpha-A at Tdep=303T_{\mathrm{dep}} = 303 K. (b) Normalized 1D intensity averaged over Ψ\Psi as transformed by Equation 2 for a 310 nm film of α,α\alpha,\alpha-A at Tdep=303T_{\mathrm{dep}} = 303 K. (c) Simulated 2D GIWAXS image calculated following Equation 5. (d) Average of the simulated GIWAXS image over Ψ\Psi.

The two dimensional (qr,qz)(q_r,q_z) intensity plot is projected across the scattering vector magnitude and the angle in between the scattering (Ψ\Psi) to extract the intensity averaged over Ψ\Psi plot against qρq_\rho as shown in Figure 2(b). We fit the projected signal over the range qρq_\rho 0.3-1.8 Å−1^{-1} using four Gaussian distributions, following Equation (2). We identify four peaks: Peak 1, 2, 3, and 4 - where Peak 1 is at the smallest scattering vector magnitude corresponding to the layering process up to Tdep≈0.9TgT_{\mathrm{dep}} \approx0.9T_g. Peak 2 has a weak and broad intensity with regards to the magnitude of the wavevector, but has a high angular character (see § 1 B in the supplementary material). Finally, Peaks 3 and 4 comprise the isotropic halo.

Then, we calculated the X-ray scattering intensity plots from the all-atom simulations following Equation (5). The all-atom configurations are generated using umbrella sampling where a variable strength force drives the simulation towards layered structures (see § 2.6 C in the Methods). The simulations were run at 425 K to compensate for the higher glass transition temperature (about 500 K, see Figure S3) associated with the shorter time scale of simulation.[18] The result shown in Figure 2(c) demonstrates a similar layering peak and isotropic halo. The one dimensional plot in Figure 2(d) allows a quantitative comparison. We again fit four Gaussian distributions restricting the position of each peak to be within 0.1 Å−1^{-1} of the experimental peaks. There are three discrepancies visible: Peak 1 shifted to a higher wavevector magnitude in simulation, indicating the simulation layers are about 14% closer than in experiment; Peak 2 is not visible in simulation; Peak 4 is significantly broader in simulation, suggesting the short range ordering of the aromatic groups is somewhat less regular in simulation than in experiment. This is likely due to the higher temperature and limited time for relaxation in simulation, although there can be some uncertainties attributed to the force-field parameterization. Nonetheless, the layering peak is visible and still very close to the characteristic size of α,α-A and Peak 3 agrees quite well between experiment and simulation. We therefore consider the simulation to be an adequate representation of the atomic structure of the film, and accurate enough to capture the general physics of the situation.

Separation of degrees of freedom

Next, we consider the contributions of the various molecular degrees of freedom to the total structure by using the structure factor of Equation (4). This equation is closely related to the X-ray scattering signal of the previous section but without the added complication of qq-dependent atomic form factors. Using simulation data, we calculated the structure factor of various subsets of the atoms in the simulation. The resulting signal is shown in Figure 3, where we separate the simulation data into layered (a,b,c,d) and unlayered (e,f,g,h) sets. Simulations with a layering peak of at least 60% of the isotropic halo intensity are classified as layered. We calculated the total structure factor (a,e), the average single molecule structure factor (b,f), and the structure factor of the molecular centers of mass (c,g). The convolution of the single molecule and center of mass structure factors is an approximation of the total structure factor that neglects the correlations between the conformations of neighboring molecules. In other words, it represents a structure with the same center of mass positions but with noninteracting molecules. The difference between the total structure factor and the noninteracting convolution we term the “molecular interaction” structure factor, shown in Figure 3(d,h). This is the excess correlation arising from the interactions between nearby molecules.

Figure 3. Decomposition of the simulated two-dimensional structure factor calculated following Equation (4) for layered and unlayered structures. The upper and lower rows show averages over simulations classified as layered and unlayered, respectively. The columns show (a,e) the total structure factor, (b,f) the ensemble-averaged single-molecule contribution, (c,g) the center-of-mass structure factor, and (d,h) the structure factor representing molecular interactions. Inset schematics visualize the physical meaning of each component. The zz-axis is defined normal to the layers. Plots b and f are adjusted following Equation (6) to remove the exponential low qq behavior for finite sized systems. Each column is normalized independently to emphasize the position and angular distribution of the structure factor features; intensities must therefore not be compared quantitatively between columns.

The isotropic halo of the total structure factor comes from the correlations between molecules. The observation that the halo in the single molecule structure factor in Figure 3(b,f) is at a larger length scale (0.9 A˚−1\text{\AA}^{-1}) than the molecular interaction halo in Figure 3(d,h) (1.2 A˚−1\text{\AA}^{-1}) indicates that the functional groups of neighboring molecules approach closer to each other than the groups on the same molecule (Figure 3(b,f)). We also observe that the halo in the total structure factor (1.2 A˚−1\text{\AA}^{-1}, Figure 3(a,e)) matches that of the molecular interactions. This suggests that the molecular level structure is based on the interlocking of the molecular functional groups, rather than each α\alpha-α\alpha molecule taking up their own roughly spherical volume and interacting at the edges of their spheres. This implies that a key barrier for relaxation is the steric frustration of the embedded structure. In addition, the isotropic halo in the experimental GI-WAXS signal corresponds to this length scale, especially Peak 3 (1.14 A˚−1\text{\AA}^{-1}, Figure 2(b)), and would therefore appear to be a direct experimental probe into the tangled structure of the film.

Comparing the layered and unlayered structure factors, the single molecule (Figure 3(b,e)) and molecular interaction structure factors (Figure 3(d,h)) are similar between them. The layering signal arises entirely from the ordering of the centers of mass (Figure 3(c,g)). It may be intuitive to suppose that layering would be accompanied by a change in single molecule conformation or orientation. Perhaps layering accesses an entirely different, more stable molecular level organization. Indeed, for nonspherical molecules, layering is strongly associated with the onset of ensemble level orientational order, [7, 11, 19, 20]. For α\alpha-α\alpha however, we observe no such trend, which would appear as an anisotropic signal in the single molecule structure factor. The lack of orientational order also agrees with evidence from spherical ellipsometry measurement of α\alpha-α\alpha, [15], which implies the layering mechanism is at a larger length scale. It also suggests that the molecular degrees of freedom could be accounted for in a simplified isotropic effective interaction. This interpretation is also supported by the use of machine learning molecular descriptors, see Section IV in the supplementary material.

Mapping to equilibrium simple liquids

Given the significance of the equilibrium interfacial liquid identified in the literature[12, 16, 21] and the results of the all-atom structure factor analysis, we consider a simplified coarse-grained model where each molecule is represented by a single isotropic bead. This approach differs from the coarse-graining approach most commonly used in the field, where a small number of beads are used to roughly approximate a molecule’s shape.[7, 11, 14, 16, 19] We justify the complete neglect of molecular shape by appealing to the isotropy of the single molecule and molecular interaction structure factors (Figure 3(b,d,f,h)).

Determining the optimal coarse-grain interactions for a given system is a complex and open problem. However, as we are interested in more general physical trends, we avoid this problem by simply choosing a pairwise Morse potential for our particles due to its softness and use in highly coarse-grained systems.[22, 23, 24] This simplified model potential is meant to probe the applicability of general liquid physics to the deposited α,α\alpha,\alpha-A glass, rather than the role of specific features of the molecule.

For our simplified equilibrium model, we therefore simulate a Morse liquid of 25,000 particles exposed to vacuum at the each interface (see § 2.6 C in Methods). The comparison between the one dimensional structure factors for the centers of mass from the all-atom simulation and the particles in the Morse simulation, shown in Figure 4(b), shows that the Morse liquid structure factor is more strongly peaked than that of the atomistic centers of mass. The Morse potential in this case is more repulsive at short distances than the actual effective interaction. The intensity plots of Figure 4(c,Figure 4d) reveal the presence of the layering peak in the all-atom and Morse liquid simulations. A discrepancy between the two simulations is the prominence of the isotropic halo in the Morse liquid. This halo represents the underlying bulk liquid ordering, which corresponds to the oscillations of the radial distribution function seen in liquids. A similar feature can be observed in the all-atom structure factor although much fainter and broader, indicating liquid-like ordering is weaker in the all-atom system. The short range repulsiveness of the Morse potential leads to the narrowness of the halo in the liquid and the presence of many particles in the bulk below the surface increases the relative signal of the bulk structure. Importantly, that we observe a strong layering peak even with a simple untuned potential is strong evidence that the center of mass ordering of a deposited glass film comes from a more general phenomenon of simple liquids.

Figure 4. Summary of coarse-grained results. (a) Snapshot of Morse simulation, with a transparency gradient to emphasize the surfaces in the visualization. (b) Structure factor intensity as a function of total wavevector magnitude. (c) Structure factor intensity plot of the centers of masses calculated from layered all-atom simulation. (d) Structure factor intensity plot of the Morse liquid.

To understand this phenomenon in the simplified model, we consider the temperature dependence of the structure factor peaks from the coarse-grained equilibrium model in Figure 5. The peak corresponding to ordering in the bulk (green), which arises from the isotropic ordering of the Morse liquid far from the surface, shows no temperature dependence over the examined range. In contrast, the layering peak (blue) is strongly dependent on the temperature. This peak corresponds to the liquid structure at the surface, and moves to lower wavevector magnitudes as the temperature increases. Figure 5b shows the relative intensity of the layering signal with respect to the bulk signal decays exponentially with temperature. The shift of the layering peak to longer length scales and the decrease in intensity indicates the breakdown of the layered structure at high temperatures, with only the bulk liquid structure commonly associated with liquids remaining. This temperature dependence in the Morse model suggests that the high temperature sensitivity of vapor deposited glasses may similarly correspond to the sensitivity of the underlying equilibrium simple liquid surface.

Figure 5. Temperature dependence of peaks for the Morse liquid. (a) The position of the simulation surface and bulk peaks with respect to temperature. Peaks 1 and 2 are marked with filled circles, with colors matching those of Figure 2(b). (b) The ratio of the layering peak intensity over the bulk peak intensity as a function of deposition temperature. The black crosses are results from simulation of the Morse liquid with a free surface. The simulation temperatures are scaled so that the minimum non-crystallizing temperature maps to the lowest experimental temperature with a layering peak, marked with a red cross in (a).

Discussion

Our results suggest that layering in PVD glass films is the result of the molecular center of masses interacting as an equilibrium liquid. We demonstrate that a simple equilibrium coarse-grained model is able to replicate key features of the all-atom simulation (Figure 4) and demonstrate a surface structure highly sensitive to temperature (Figure 5). We propose, based on structure factor analysis of the all-atom simulation data (Figure 3) and experimental GIWAXS data (Figure 2) that the molecular interactions are mediated by the flexible aromatic groups of α,α\alpha,\alpha-A resulting in a simple, isotropic effective potential. Therefore we conclude that the layering in α,α\alpha,\alpha-A deposited films is closely related to a general equilibrium phenomenon of simple liquids and this effect is likely general to other molecular PVD films. Our results build on previous work concerning the origin of the anisotropic structure and increased stability of PVD glasses. Enhanced mobility has long been used to explain PVD glass’s enhanced stability,[25] although it has been observed that greater surface diffusion does not necessarily lead to more stable glasses, such as in α\alpha,α\alpha-A[26] or in polymeric glasses.[27] The equilibrium surface has been empirically connected to molecular layering,[21] although the reason why this kind of center of mass ordering is the relevant thermodynamic minimum across a wide variety of molecular glasses has not been elaborated upon. Our results suggest that the liquid-like ordering at the surface is a prominent thermodynamic driving force for many small organic glass formers.

We expect these results to be applicable to PVD glasses with mostly spherically shaped molecules, with deviations expected when effective interactions become directional, such as with molecular anisotropy or strong intermolecular interactions[28, 29]. Thus, while the layering effects in isotropic molecules parallel those of non-spherical molecules, the causal mechanism differs.

Conclusion

In this study, we combined grazing-incidence X-ray scattering with atomistic and coarse-grained simulations to identify the microscopic origin of the molecular-scale layering in vapor-deposited α\alpha,α\alpha-A. Our analysis separates translational layering from molecular orientation and intramolecular conformation. We find that the experimentally observed anisotropic structure factor feature is primarily associated with intermolecular center-of-mass correlations and does not require orientational or conformational ordering. Our results are consistent with previous work showing that the unique properties of the deposited glass arise from surface equilibration and subsequent burying, with the temperature sensitivity coming from the careful balance of surface mobility and thermodynamics.[30, 31, 32]

The various features of molecular layering have become well understood in recent years, but the molecular organization within the film and its origin has remained unclear. Our work provides evidence that molecular conformations and interactions need not significantly change from bulk behavior to allow for layering. This work suggests that the intramolecular degrees of freedom mediate simple interactions between the molecules and that the layering of α\alpha,α\alpha-A is a result of the molecular centers of mass behaving as an equilibrium simple liquid at the surface. This finding suggests that the physical driving force of glass morphology entails a competition between surface effects (which would favor layering) and the frustration of local molecular interactions.

Supplementary Material

The supplementary material contains further details on GIWAXS measurements, umbrella sampling protocols, and additional characterization of the simulated all-atom results using the AniSOAP[33, 34] methodology and density calculations.

Acknowledgments

This research was primarily supported by NSF through the University of Wisconsin Materials Research Science and Engineering Center (DMR-2309000). This investigation used the 12-ID beamline of the National Synchrotron Light Source II, a US Department of Energy (DOE) Office of Science User Facility, operated for the DOE Office of Science by Brookhaven National Laboratory under contract no. DE-SC0012704. This work used the Anvil computing center at Purdue University through allocation CHM250054 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by U.S. National Science Foundation grants no. 2138259, 2138286, 2138307, 2137603, and 2138296. Thanks to M. D. Ediger for insightful discussions and Dr. Po-Cheng Kung for insights in interpreting simulation structure factor results.

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author Declarations

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

CC: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing - Original Draft, Writing - Review & Editing; JB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing - Original Draft, Writing - Review & Editing; AL: Formal analysis, Software, Visualization; RS: Resources; PL: Resources; SW: Resources; CC: Resources; DK: Resources; YW: Resources; PW: Resources; PW: Resources; WL: Formal analysis, Supervision, Writing - Review & Editing; ZF: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Visualization, Writing - Original Draft, Writing - Review & Editing; RC: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Software, Validation, Visualization, Writing - Original Draft, Writing - Review & Editing.

Data Availability

Data will be made available in a Zenodo repository upon publication.

Methods

Synthesis and sample preparation

α,α\alpha,\alpha-A (structure shown in Figure 1, Tg=364±2T_g = 364 \pm2 K[14]) was synthesized via the Suzuki cross-coupling reactions and purified as detailed in previous publications.[35, 36] α,α\alpha,\alpha-A in powder form was placed in alumina crucibles and pre-melted in a vacuum oven (Fisherbrand Isotemp 281A, P≈1P \approx1 kPa) at roughly 543 K before being placed into a custom-made deposition chamber with a base pressure of 10−710^{-7} torr for thermal deposition (as detailed in earlier publications.[14, 35, 36])

Films of α,α\alpha,\alpha-A with a thickness of 310 nm were vapor-deposited onto silicon substrates (Si(100) with native oxide, Virginia Semiconductor, Inc.), which were mounted on a bridge between two temperature-controlled copper stages at two different temperatures to produce temperature-gradient (T-grad) samples. Details of the calibration and TdepT_{\mathrm{dep}} determination can be found in the supplementary material and in our previous publications.[14, 35, 36] GIWAXS experiments were performed on T-grad samples held in the TdepT_{\mathrm{dep}} range of 240 K ≥Tdep≥\ge T_{\mathrm{dep}} \ge 340 K, at a deposition rate of rdep=0.02±0.02r_{\mathrm{dep}} = 0.02 \pm0.02 nm/s, monitored with a quartz crystal microbalance (QCM, Inficon STM-2).

After deposition, the temperature of the hot side of each T-grad sample was rapidly quenched, and the cold side was heated to room temperature. Samples were removed and stored at roughly 254 K prior to characterization.

Grazing Incident Wide-Angle X-ray Scattering

GIWAXS data were obtained at room temperature using the 12-ID (Soft Matter Interfaces) beamline at the National Synchrotron Light Source II (NSLS-II) of the Brookhaven National Laboratory. A PILATUS 900 kW detector was used with a sample-to-detector distance of 277 mm. The incident angle was set at 0.1∘0.1^\circ with respect to the plane of the film. The energy of the X-ray beam was set to 16.1 keV (λ=0.77009\lambda= 0.77009 Å). The beam scanned 33-grid points along the T-grad sample, with a distance of 0.2 cm between each measured point. At each point, four images were taken, with the detector being moved horizontally on a fixed arc at 0∘0^\circ, 2∘2^\circ, 18∘18^\circ, and 20∘20^\circ relative to the incident beam, to capture a comprehensive span of scattering angles. Each image was collected with an exposure time of 0.3 s. A Python-based (version 3) package developed at NSLS-II was used to perform post-acquisition data processing.

Figure 2a shows an example of the 2D GIWAXS data. We note that in some images, the integration did not fully mask defects; these features (for example, the dark stripe around qz=1.3q_z = 1.3 Å−1^{-1} in Figure 2a), along with regions with intense scattering from the X-ray source, were masked with white stripes.

As schematically shown in Figure S1, the azimuthal angle (Ψ\Psi) was defined as Ψ=0∘\Psi= 0^\circ corresponding to an out-of-plane direction (qzq_z) and Ψ=±90∘\Psi= \pm90^\circ to an in-plane direction (qrq_r). A transformation to cylindrical coordinates was also performed to obtain the angular dependence of the intensity and to generate averaged intensity profiles. The cylindrical coordinate vectors qρq_\rho and angle Ψ\Psi were defined as:

qρ=qz2+qr2,q_\rho= \sqrt{q_z^2 + q_r^2},
Ψ=arctan⁡(qrqz).(1)\Psi= \arctan\left(\frac{q_r}{q_z}\right). \tag*{(1)}

In the cylindrical geometry, the normalized intensity was fitted to the sum of four Gaussian functions:

Ifit(qρ,Ψ)=∑k=14Gk(qρ−qk,Ψ).(2)I_{\mathrm{fit}}(q_\rho,\Psi) = \sum_{k=1}^{4} G_k(q_\rho-q_k,\Psi). \tag*{(2)}

Where IfitI_{\mathrm{fit}} is the total fitted intensity of the measurement, Gk(qρ,Ψ)G_k(q_\rho,\Psi) represents a Gaussian function with angular-dependent intensity. It was found that the two peaks at lower qq values, G1(qρ,Ψ)G_1(q_\rho,\Psi) and G2(qρ,Ψ)G_2(q_\rho,\Psi) had angular dependencies, while G3(qρ,Ψ)G_3(q_\rho,\Psi) and G4(qρ,Ψ)G_4(q_\rho,\Psi) did not show any angular dependence, even if they were considered as anisotropic functions. As such, for simplicity, they were fitted to isotropic Gaussian functions.

Molecular Dynamics

We used the Optimized Potential for Liquid Simulations (OPLS) all-atom (AA) force field,[37] which has been previously used to study glass formers[38, 39] to perform simulations of α,α\alpha,\alpha-A in LAMMPS.[40] Parameters, including many-body interactions, were generated using the online LigParGen tool using the 1.14*CMLA partial charge model.[41, 42, 43]

To validate this potential, we performed umbrella sampling using LAMMPS and PLUMED[44, 45] to compare the stable layering separation to that exhibited in GIWAXS data. Systems of 320 molecules were initially equilibrated in the NPT ensemble at 1000 K using a low target pressure to relax the simulation cell prior to cooling. The molecules were restrained to their layers using a harmonic restraint (a so-called “bias”) to the layer plane. We then ran simulations at 350 K using an anisotropic barostat until they demonstrated 50 uncorrelated measurements of the volume across time, as calculated by pyMBAR,[46] after which we collected biased sampling data for another 0.5 ns and computed the free energy profile of the separation distances, again using pyMBAR, and as shown in Figure S2 in the supplementary material.

Following this validation, we ran simulations with 512 molecules, again using biasing to study multiple degrees of anisotropic layering, from bulk glass (showing no layering) to systems exhibiting strong layering consistent with experiment. Following simulations at 1000 K and 1 bar, we systematically quenched our systems at a rate of 30 K/ns until reaching 425 K, slightly below the experimental glass transition temperature for this system (as calculated following the method of Suter et al. [47] and shown in Figure S3 in the supplementary material). At 425 K, we ran simulations, again, until sampling 30 uncorrelated volumes across time, after which we removed the simulation biases and collected simulated data in the NVT ensemble for 1 ns. The results described in the main text are based on these final production runs. All simulations were run with a simulation timestep of 1 fs.

Coarse grained simulations of the α,α-A surface were performed using a single bead for each molecule interacting via the isotropic Morse potential,

U(r)=ε(1−exp⁡[−a(r−rc)])2,(3)U(r) = \varepsilon\left(1-\exp\left[-a(r-r_c)\right]\right)^2, \tag*{(3)}

with ε=1\varepsilon= 1, a=3a = 3, and rc=1r_c = 1 chosen to approximate the softness of the α,α-A molecule interactions. Simulations of 25,000 particles were run in the NVT ensemble at dimensionless temperatures from T=0.02−2.0T = 0.02 - 2.0 using OpenMM.[48] We initialized the system to resemble a dense film in contact with vacuum, consistent with PVD procedures. We then ran simulations for 50,000τ50{,}000\tau. We use the last 40,000τ40{,}000\tau to calculate the structure factors of the particles along the film/vacuum interface. Equilibrated density profiles are shown in Figure S5 for T=0.45T = 0.45. Systems below T=0.4T = 0.4 were observed to crystallize.

Simulated X-ray Scattering

The structure factor is related to the Fourier components of the density field,

ρ(q)=∫ρi(r)exp⁡(iq⋅r) dr.\rho(\mathbf{q}) = \int\rho_i(\mathbf{r}) \exp(i\mathbf{q}\cdot\mathbf{r})\,\mathrm{d}r.

where the number density of species ii, ρi\rho_i, comes directly from the NN particles of the simulation,

ρi(r)=∑j=1Niδ(r−rj).\rho_i(\mathbf{r}) = \sum_{j=1}^{N_i} \delta(\mathbf{r}-\mathbf{r}_j).

The structure factor, S(q)S(\mathbf{q}), is the spatial correlation of the number density in reciprocal space,

S(q)=1N∑i,jρi(−q)ρj(q)=∑i,jSij(q),S(\mathbf{q}) = \frac{1}{N}\sum_{i,j}\rho_i(-\mathbf{q})\rho_j(\mathbf{q}) = \sum_{i,j}S_{ij}(\mathbf{q}),

where Sij(q,t)S_{ij}(\mathbf{q},t) are the partial structure factors. The structure factor is often spherically averaged over all wave vectors of the same magnitudes and over time, however here we transform the data from cartesian (qx,qy,qz)(q_x,q_y,q_z) coordinates to circularly averaged from cartesian (qr,qz)(q_r,q_z) coordinate. We use the transformed structure factor,

S(qr,qz)=12πqr∑i,jSij(q) δqr2+qy2−qr2.(4)S(q_r,q_z) = \frac{1}{2\pi q_r}\sum_{i,j}S_{ij}(q)\,\delta_{q_r^2+q_y^2-q_r^2}. \tag*{(4)}

We weight the partial structure factors by the atomic form factors calculated by Waasmaier and Kirfel[49] to get the estimated X-ray scattering from simulation,

I(qr,qz)=12πqr∑i,jfi(q)fj(q)Sij(q) δqr2+qy2−qr2.(5)I(q_r,q_z) = \frac{1}{2\pi q_r}\sum_{i,j}f_i(q)f_j(q)S_{ij}(q)\,\delta_{q_r^2+q_y^2-q_r^2}. \tag*{(5)}

The two dimensional scattering and structure factor surfaces were smoothed using kernel regression as implemented in the python package KDEpy.[50] Here, we use a Gaussian kernel of the form K(x1,x2)=exp⁡(−(x1−x2)2/2σ2)K(x_1,x_2)=\exp(-(x_1-x_2)^2/2\sigma^2) with a σ\sigma value of 0.08 Å−1^{-1}. We define our coordinate system such that the z axis is always orthogonal to layering.

The low q\mathbf{q} behavior of the scattering function arises from the finite size of the system. For single molecules, this behavior can dominate the scattering and obscures the more relevant correlations between the atoms behind this finite size effect. At low q\mathbf{q} magnitudes (which correspond to length scales greater than the longest inter-atomic separation in the molecule), the molecule can be approximately treated as a smooth density distribution. The size effect is often accounted for by approximating this contribution as a normal distribution,[51, 52]

Sexcess(q)=S(q)−S(0)exp⁡(−q2Rg23),(6)S_{\mathrm{excess}}(q)=S(q)-S(0)\exp\left(\frac{-q^2R_g^2}{3}\right), \tag*{(6)}

where Rg2R_g^2 is the radius of gyration, in this case the electron weighted variance of particle positions about the center of mass, and is related to the size of the molecule. This form corresponds to the scattering contribution of a Gaussian approximation of the molecule density.

A common method to relate single molecule structure factor to the total is to take the convolution of the single molecule factor with a lower resolution representation of the system. In Fourier space, this is quite convenient because it is simply a multiplication of the two functions. In literature, the notation often used is

I(q)=P(q)S(q),(7)I(q)=P(q)S(q), \tag*{(7)}

where P(q)P(q) is known as the ‘form factor’, i.e. the single particle contribution, and S(q)S(q) is called the ‘structure factor.’[53, 54, 55] Note that we calculate the scattering signal of simulation in a similar way, except the form factor represents the electronic response of each atom and the structure factor represents each atom position. This makes reciprocal space methods well suited for investigating structure over multiple length scales.

This convolution is a physical approximation. The approximation amounts to taking the centers of mass, adding the ensemble averaged density at each center, and calculating the resulting structure factor of the smooth field. This ‘decoupling’ approximation clearly neglects the correlation of molecular shapes between neighbors, where a molecule is not free to randomly sample the conformational ensemble, but is restricted by the conformations of its neighbors. With this in mind, we use the convolution to account for those aspects of the structure that do not depend on intermolecular correlations, and the difference between it and the total structure factor we ascribe to correlations between aromatic groups of the same molecule or adjacent molecules.

References

  1. [1]L. Jikang, “The application of organic materials used in oled display,” Results in Optics 21, 100856 (2025).DOI
  2. [2]D. Yokoyama, “Molecular orientation in small-molecule organic light-emitting diodes,” Journal of Materials Chemistry 21, 19187–19202 (2011), https://pubs.rsc.org/jm/article-pdf/21/48/19187/2412710/c1jm13417e.pdf.DOI
  3. [3]A. Gujral, L. Yu, and M. D. Ediger, “Anisotropic organic glasses,” Current Opinion in Solid State and Materials Science 22, 49–57 (2018).DOI
  4. [4]P. Luo and Z. Fakhraai, “Surface-mediated formation of stable glasses,” Annual Review of Physical Chemistry 74, 361–389 (2023).DOI
  5. [5]M. D. Ediger, J. de Pablo, and L. Yu, “Anisotropic vapor-deposited glasses: Hybrid organic solids,” Accounts of Chemical Research 52, 407–414 (2019).DOI
  6. [6]A. Gujral, K. A. O’Hara, M. F. Toney, M. L. Chabinyc, and M. D. Ediger, “Structural characterization of vapor-deposited glasses of an organic hole transport material with x-ray scattering,” Chemistry of Materials 27, 3341–3348 (2015).DOI
  7. [7]I. Lyubimov, L. Antony, D. M. Walters, D. Rodney, M. D. Ediger, and J. J. de Pablo, “Orientational anisotropy in simulated vapor-deposited molecular glasses,” The Journal of Chemical Physics 143, 094502 (2015).DOI
  8. [8]A. T. M. M. Alahe, T. J. Ferron, and C. E. Bishop, “Kinetically arrested phase separation leads to tunable domain structures in vapor-deposited glasses,” The Journal of Chemical Physics 165, 124511 (2026).
  9. [9]L. Berthier and M. D. Ediger, “Facets of glass physics,” Physics Today 69, 40–46 (2016).
  10. [10]V. V. Hoang and T. Q. Dong, “Free surface effects on thermodynamics and glass formation in simple monatomic supercooled liquids,” Phys. Rev. B 84, 174204 (2011).DOI
  11. [11]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,” Proceedings of the National Academy of Sciences 112, 4227–4232 (2015).
  12. [12]C. Bishop, J. L. Thelen, E. Gann, M. F. Toney, L. Yu, D. M. DeLongchamp, and M. D. Ediger, “Vapor deposition of a non-mesogen prepares highly structured organic glasses,” Proceedings of the National Academy of Sciences 116, 21421–21426 (2019), https://www.pnas.org/doi/pdf/10.1073/pnas.1908445116.
  13. [13]C. Bishop, Y. Li, M. F. Toney, L. Yu, and M. D. Ediger, “Molecular orientation for vapor-deposited organic glasses follows rate-temperature superposition: The case of posaconazole.” J Phys Chem B 124, 2505–2513 (2020).DOI
  14. [14]A. Zhang, A. R. Moore, H. Zhao, S. Govind, S. E. Wolf, Y. Jin, P. J. Walsh, R. A. Riggleman, and Z. Fakhraai, “The role of intramolecular relaxations on the structure and stability of vapor-deposited glasses,” The Journal of Chemical Physics 156, 244703 (2022).DOI
  15. [15]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).DOI
  16. [16]K. Bagchi, C. Deng, C. Bishop, Y. Li, N. E. Jackson, L. Yu, M. F. Toney, J. J. de Pablo, and M. D. Ediger, “Over what length scale does an organic substrate perturb the structure of a glassy organic semiconductor?” ACS Applied Materials & Interfaces 12, 26717–26726 (2020).DOI
  17. [17]K. Bagchi, N. E. Jackson, A. Gujral, C. Huang, M. F. Toney, L. Yu, J. J. de Pablo, and M. D. Ediger, “Origin of anisotropic molecular packing in vapor-deposited aq3 glasses,” The Journal of Physical Chemistry Letters 10, 164–170 (2019).DOI
  18. [18]A. Madanchi, E. Azek, K. Zongo, L. K. Béland, N. Mousseau, and L. Simine, “Is the future of materials amorphous? challenges and opportunities in simulations of amorphous materials,” ACS Physical Chemistry Au 5, 3–16 (2025).
  19. [19]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,” The Journal of Physical Chemistry Letters 8, 3380–3386 (2017).
  20. [20]T. J. Ferron, J. L. Thelen, K. Bagchi, C. Deng, E. Gann, J. J. de Pablo, M. D. Ediger, D. F. Sunday, and D. M. DeLongchamp, “Characterization of the interfacial orientation and molecular conformation in a glass-forming organic semiconductor,” ACS Applied Materials & Interfaces 14, 3455–3466 (2022), https://pubs.acs.org/aaamick/article-pdf/14/2/3455/12806651/am1c1984.pdf.DOI
  21. [21]S. Singh and J. J. de Pablo, “A molecular view of vapor deposited glasses,” The Journal of Chemical Physics 134, 194903 (2011).DOI
  22. [22]N. Zhang, P. Zhang, W. Kang, D. Bluestein, and Y. Deng, “Parameterizing the morse potential for coarse-grained modeling of blood plasma,” Journal of Computational Physics 257, 726–736 (2014).pubmed.ncbi.nlm.nih.gov/24910470
  23. [23]C. C. Liew and M. Mikami, “A coarse-grained model for particle dynamics simulations of complex fluids,” Chemical Physics Letters 368, 346–351 (2003).DOI
  24. [24]S.-W. Chiu, H. L. Scott, and E. Jakobsson, “A coarse-grained model based on morse potential for water and n-alkanes,” Biophysical Journal 98, 573a–574a (2010).
  25. [25]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, 353–356 (2007).pubmed.ncbi.nlm.nih.gov/17158289
  26. [26]S. Samanta, G. Huang, G. Gao, Y. Zhang, A. Zhang, S. Wolf, C. N. Woods, Y. Jin, P. J. Walsh, and Z. Fakhraai, “Exploring the importance of surface diffusion in stability of vapor-deposited organic glasses,” The Journal of Physical Chemistry B 123, 4108–4117 (2019), https://pubs.acs.org/jpcb/ article-pdf/123/18/4108/11081314/jp9b01012.pdf.DOI
  27. [27]Y. Chai, T. Salez, J. D. McGraw, M. Benzaquen, K. Dalnoki-Veress, E. Raphael, and J. A. Forrest, “A direct quantitative measure of surface mobility in a glassy polymer,” Science 343, 994–999 (2014), https://www.science.org/doi/pdf/10.1126/science.1244845.DOI
  28. [28]M. Tylinski, Y. Z. Chua, M. S. Beasley, C. Schick, and M. D. Ediger, “Vapor-deposited alcohol glasses reveal a wide range of kinetic stability,” The Journal of Chemical Physics 145, 174506 (2016).pubmed.ncbi.nlm.nih.gov/27825204
  29. [29]A. R. Moore, G. Huang, S. Wolf, P. J. Walsh, Z. Fakhraai, and R. A. Riggleman, “Effects of microstructure formation on the stability of vapor-deposited glasses,” Proceedings of the National Academy of Sciences 116, 5937–5942 (2019), https://www.pnas.org/doi/pdf/10.1073/pnas.1821761116.
  30. [30]E. Thoms, J. P. Gabriel, A. Guiseppi-Elie, M. D. Ediger, and R. Richert, “In situ observation of fast surface dynamics during the vapor-deposition of a stable organic glass,” Soft Matter 16, 10860–10864 (2020).DOI
  31. [31]L. Berthier, P. Charbonneau, E. Flenner, and F. Zamponi, “Origin of untrustability in vapor-deposited glasses,” Phys. Rev. Lett. 119, 188002 (2017).DOI
  32. [32]C. Rodríguez-Tinoco, M. Gonzalez-Silveira, J. Råfols-Ribé, A. Vila-Costa, J. C. Martinez-Garcia, and J. Rodríguez-Viejo, “Surface-bulk interplay in vapor-deposited glasses: Crossover length and the origin of front transformation,” Phys. Rev. Lett. 123, 155501 (2019).
  33. [33]C. K. Hu, K. Huguenin-Dumittan, Y.-C. Cho, J. Nigam, and R. K. Cersonsky, “Expanding density-correlation machine learning representations for anisotropic coarse-grained particles,” The Journal of Chemical Physics 161, 074112 (2024).
  34. [34]C. Jorgensen, A. Y. Lin, R. Vasavada, and R. K. Cersonsky, “Interpretable visualizations of data spaces for classification problems,” Machine Learning: Science and Technology (2026), 10.1088/2632-2153/ae466e.
  35. [35]T. Liu, K. Cheng, E. Salami-Ranjbaran, F. Gao, E. C. Glor, M. Li, P. J. Walsh, and Z. Fakhraai, “Synthesis and high-throughput characterization of structural analogues of molecular glassformers: 1,3,5-trisarylbenzenes,” Soft Matter 11, 7558–7566 (2015).DOI
  36. [36]E. E. Wolf, T. Liu, S. Govind, H. Zhao, G. Huang, A. Zhang, Y. Wu, J. Chin, K. Cheng, E. Salami-Ranjbaran, F. Gao, G. Gao, Y. Jin, Y. Pu, T. G. Toledo, K. Ablajan, P. J. Walsh, and Z. Fakhraai, “Design of a homologous series of molecular glassformers,” The Journal of Chemical Physics 155, 224503 (2021).DOI
  37. [37]W. L. Jorgensen, D. S. Maxwell, and J. Tirado-Rives, “Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids,” Journal of the American Chemical Society 118, 11225–11236 (1996).DOI
  38. [38]J. Zhang, Y. Liang, J. Yan, and J. Lou, “Study of the molecular weight dependence of glass transition temperature for amorphous poly(1-lactide) by molecular dynamics simulation,” Polymer 48, 4900–4905 (2007).
  39. [39]P. Sichaouki, K. S. Karadima, V. G. Mavrantzas, and S. N. Pandis, “The effect of functional groups on the glass transition temperature of atmospheric organic compounds: a molecular dynamics study,” Soft Matter 20, 4783–4794 (2024).
  40. [40]A. P. Thompson, H. M. Aktulga, R. Berger, D. S. Bolintineanu, W. M. Brown, P. S. Crozier, P. J. in ’t Veld, A. Kohlmeyer, S. G. Moore, T. D. Nguyen, R. Shan, M. J. Stevens, J. Tranchida, C. Trott, and S. J. Plimpton, “LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales,” Computer Physics Communications 271, 108171 (2022).DOI
  41. [41]L. S. Dodda, I. Cabeza de Vaca, J. Tirado-Rives, and W. L. Jorgensen, “Lipargen web server: an automatic opls-aa parameter generator for organic ligands,” Nucleic Acids Research 45, W331–W336 (2017).DOI
  42. [42]W. L. Jorgensen and J. Tirado-Rives, “Potential energy functions for atomic-level simulations of water and organic and biomolecular systems,” Proceedings of the National Academy of Sciences 102, 6665–6670 (2005).pubmed.ncbi.nlm.nih.gov/15870211
  43. [43]L. S. Dodda, J. Z. Vilseck, J. Tirado-Rives, and W. L. Jorgensen, “1.14*cmla-1bcc: Localized bond-charge corrected cmla charges for condensed-phase simulations,” The Journal of Physical Chemistry B 121, 3864–3870 (2017).
  44. [44]M. Bonomi, G. Bussi, C. Camilloni, G. A. Tribello, P. Banás, A. Barducci, M. Bernetti, P. G. Bolhuis, S. Bottaro, D. Branduardi, R. Capelli, P. Carloni, M. Ceriotti, A. Cesari, H. Chen, W. Chen, F. Colizzi, S. De, M. De La Pierre, D. Donadio, V. Drobot, B. Ensing, A. L. Ferguson, M. Filizola, J. S. Fraser, H. Fu, P. Gasparotto, F. L. Gervasio, F. Giberti, A. Gil-Ley, T. Giorgino, G. T. Heller, G. M. Hocky, M. Iannuzzi, M. Invernizzi, K. E. Jelfs, A. Jussupow, E. Kirilin, A. Laio, V. Limoncelli, K. Lindorff-Larsen, T. Löhr, F. Marinelli, L. Martin-Samos, M. Masetti, R. Meyer, A. Michaelides, C. Molteni, T. Morishita, M. Nava, C. Paissoni, E. Papaleo, M. Parrinello, J. Pfaendtner, P. Piaggi, G. M. Piccini, A. Pietropaolo, F. Petrucci, S. Pipolo, D. Provasi, D. Quigley, P. Raiteri, S. Raniolo, J. Rydzewski, M. Salvaglio, G. C. Sesso, V. Spiwók, J. Sponer, D. W. Swenson, P. Tiwary, O. Valsson, M. Vendruscolo, G. A. Voth, and A. White, “Promoting transparency and reproducibility in enhanced molecular simulations,” Nature Methods 16, 670–673 (2019).DOI
  45. [45]G. A. Tribello, M. Bonomi, D. Branduardi, C. Camilloni, and G. Bussi, “PLUMED 2: New feathers for an old bird,” Computer Physics Communications 185, 604–613 (2014).
  46. [46]M. R. Shirts and J. D. Chodera, “Statistically optimal analysis of samples from multiple equilibrium states,” Journal of Chemical Physics 129, 124105–124105 (2008).
  47. [47]J. L. Suter, W. A. Muller, M. Vassaux, A. Anastasiou, M. Simmons, D. Tilbrook, and P. V. Coveney, “Rapid, accurate and reproducible prediction of the glass transition temperature using ensemble-based molecular dynamics simulation,” Journal of Chemical Theory and Computation 21, 1405–1421 (2025).DOI
  48. [48]P. Eastman, R. Galvelis, R. P. Peláez, C. R. A. Abreu, S. E. Farr, E. Gallicchio, A. Gorenko, M. M. Henry, F. Hu, J. Huang, A. Krämer, J. Michel, J. A. Mitchell, V. S. Pande, J. P. Rodrigues, J. Rodriguez-Guerra, A. C. Simmonett, S. Singh, J. Swails, P. Turner, Y. Wang, I. Zhang, J. D. Chodera, G. De Fabritiis, and T. E. Markland, “OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials,” The Journal of Physical Chemistry. B 128, 109–116 (2024).arxiv.org/abs/2310.03121
  49. [49]D. Waasmaier and A. Kirfel, “New analytical scattering-factor functions for free atoms and ions,” Acta Crystallographica Section A Foundations of Crystallography 51, 416–431 (1995).DOI
  50. [50]Tommy, S. White, Alexandra, R. D. Maio, gunvor, and L. Dyer, “tommyod/kdepy: v1.1.8,” (2023).
  51. [51]F. Brunel, P. Galanopoulo, E. Espinosa Rodriguez, M. Lansalot, and F. D’Agosto, “In situ saxs investigation of vinyl acetate polymerization-induced self-assembly,” Polymer Chemistry 15, 979–990 (2024).DOI
  52. [52]C. Kim, C. Jarumaneeroj, W. Rungswang, K. S. Jin, and M. Ree, “A comprehensive small angle x-ray scattering analysis on morphological structure of semicrystalline linear polymer in bulk state,” Polymer 243, 124610 (2022).
  53. [53]J. S. Pedersen, “Analysis of small-angle scattering data from colloids and polymer solutions: modeling and least-squares fitting,” Advances in Colloid and Interface Science 70, 171–210 (1997).DOI
  54. [54]J.-P. Hansen and I. R. McDonald, “Chapter 11 - molecular liquids,” in Theory of Simple Liquids (Fourth Edition) (Academic Press, Oxford, 2013) fourth edition ed., pp. 455–510.
  55. [55]T. Li, A. J. Senesi, and B. Lee, “Small angle x-ray scattering for nanoparticle research,” Chemical Reviews 116, 11128–11180 (2016), pMID: 27054962.DOI

Paper details

vlnc:W_zdnc7zka
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents