{"id":"99e52671-0d16-449f-9f77-4049e44ae9de","arxiv_id":"2607.20066","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In vapor-deposited TPD/DO37 films, lowering the deposition rate at fixed temperature enlarges the characteristic phase-separated domain spacing, demonstrating rate-based tuning of morphology.","lead":"This paper shows that slowing the deposition of a two-component organic glass film makes its phase-separated domains grow larger, offering a simple knob to tune nanostructure. The result matters because organic electronics like solar cells and transistors depend on precisely sized domains in their active layers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The rate-dependence of the shorter domain spacing (PSD peak 2) derives from two-peak fits to PSDs that are single broad features at 0.04/0.08 nm/s and a weak shoulder at 0.008 nm/s; without independent validation, the central tunability claim is not yet established.","rationale":"The reader's weakest assumption—that the two-peak decomposition of the PSDs is not independently validated—matches the load-bearing concern identified here. The paper has genuine strengths: open code/data, multiple complementary techniques (AFM, RSoXS, GIWAXS), and a plausible physical mechanism grounded in the surface-equilibration literature. However, the central quantitative evidence for rate-tunable domain size is the high-q peak's shift with deposition rate, and that peak is exactly the component that is least secure in the fits for the intermediate and slow deposition rates. The paper's own statements about single broad features and weak shoulders, and its use of local-maxima detection for the overlapping RSoXS peaks, indicate that the decomposition is not robustly established. Because the reader already returned a CONDITIONAL verdict on essentially these grounds, my stress-test does not move the verdict; it reinforces the condition. A concrete reanalysis of the publicly available data with model comparison and independent real-space statistics would settle whether the rate trend in peak 2 is physical or a fitting artifact.","tokens_in":16614,"tokens_out":4930,"duration_ms":46911,"concrete_test":"Re-run the published AFM analysis on the GitHub data: (1) compute 2D autocorrelation and pair-distance / domain-size statistics from thresholded height and phase images for all five rates; (2) compare one-peak, two-peak, and log-normal PSD fits using BIC/AIC with multiple background choices; (3) take AFM images at ≥3 independent locations per sample and propagate the between-image scatter. If the autocorrelation shows a single dominant spacing at 0.04/0.08 nm/s, or two-Gaussian fits are not preferred, or between-image scatter exceeds the observed peak-2 shift across deposition rates, the tunability claim for the shorter length scale fails. As a second check, fit the 0.04 nm/s RSoXS with an explicit two-peak-plus-background model and examine parameter stability; unstable fits would remove the independent bulk confirmation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is the extraction of a rate-dependent small-spacing length scale ('peak 2') from the AFM PSDs. The central claim—that deposition rate directly tunes center-to-center domain spacing—rests on the monotonic shift of this peak with Rdep in Fig. 3C. However, the paper itself states that at Rdep = 0.04 and 0.08 nm/s the measured PSD contains a single broad feature which the authors 'for now' assume is a convolution of two length scales, and that at 0.008 nm/s peak 2 is only a weak shoulder. Decomposing this feature into two Gaussians after a linear-background subtraction is a modelling assumption; the RSoXS check for 0.04 nm/s also needed a different, local-maxima-based peak finding because the peaks overlap, and RSoXS was only collected for two rates. There is no independent determination—e.g. real-space autocorrelation, pair-distance statistics, or Fourier-filtered binarized images with error bars—that the fitted high-q Gaussian corresponds to a real compositional/droplet periodicity rather than a broad shoulder or fit artifact. If peak 2 is not real or its position is highly fit-dependent, the claimed rate-tuning of the smaller length scale (and the annealing claim that peak 2 is stationary while peak 1 moves) is unsupported. Since that shorter spacing is the quantitative core of the tunability claim, the central argument is conditional on this decomposition.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports physical vapor deposition of co-deposited TPD/DO37 glass blends at fixed substrate temperature (310 K) and five deposition rates, and characterizes the resulting phase-separated morphologies by AFM height/phase imaging, PSD analysis, RSoXS, and post-deposition annealing. The central claim is that the domain morphology—specifically a shorter center-to-center spacing (PSD peak 2) and a longer spacing (peak 1)—can be tuned by deposition rate alone, and that the shorter length scale arises largely from compositional phase separation while the larger length scale also carries a topographic contribution. The authors interpret the results through the surface-equilibration mechanism and a percolation-to-cluster transition framework. The paper includes open data and code, multi-energy RSoXS, and annealing experiments.","tokens_in":17047,"tokens_out":4275,"duration_ms":37790,"significance":"If the central claim holds, the paper provides a direct processing-structure relation: deposition rate as a control parameter for domain size in vapor-deposited organic blends, which is relevant for organic electronic devices. The study is commendable for combining AFM PSD analysis with bulk-sensitive RSoXS and for making data/code openly available. However, the quantitative core of the tunability claim—the rate dependence of the shorter length scale—is built on a two-Gaussian decomposition of PSDs that are, by the authors' own description, single broad features for intermediate rates and a weak shoulder for the lowest rate. The current evidence is therefore conditional on a modelling assumption that has not been independently validated. The paper is potentially valuable, but the key quantitative result needs stronger support before publication.","major_comments":[{"comment":"The central claim that the center-to-center domain spacing is tuned by deposition rate rests on the peak positions in Fig. 3C. For Rdep = 0.04 and 0.08 nm/s, the PSD is described as a single broad feature that the authors 'for now' assume is a convolution of two length scales; for Rdep = 0.008 nm/s, peak 2 is only a weak shoulder near q ~ 0.055 nm^-1. After subtracting a linear background and fitting two Gaussians, the fitted high-q positions are used as quantitative evidence. This is a model assumption, and the manuscript does not report model-selection tests, parameter uncertainties, or independent real-space validation. Please provide one or more of the following: one-peak vs two-peak model comparison (e.g., AIC/BIC), bootstrap confidence intervals on peak positions, replicate AFM scans, or real-space domain spacing statistics from binarized/Fourier-filtered images with error bars. Wi","section":"Results, AFM PSD analysis / Fig. 3"},{"comment":"The RSoXS peak 2 positions differ by roughly a factor of two from the AFM peak 2 positions (0.8 nm/s: 0.045 vs 0.089 nm^-1; 0.04 nm/s: 0.025 vs 0.045 nm^-1). The manuscript attributes this to surface vs bulk aging, but this means that the AFM peak 2 does not directly report the bulk domain spacing. Because the tunability claim is demonstrated at the surface and RSoXS was only collected at two deposition rates, the bulk extent of the rate-tunability is not established. Please state this scope explicitly or provide additional bulk or cross-sectional data showing that the rate trend continues beyond the surface.","section":"Results, Bulk structure analysis with RSoXS / Fig. 4C"},{"comment":"The annealing experiment is used to argue that the two length scales are physically distinct and that peak 2 is stationary while peak 1 evolves. For the Rdep = 0.08 nm/s film, the as-deposited PSD is a single broad feature and the peaks only separate after annealing. The fitted peak positions before and after annealing therefore rely on the same unvalidated decomposition. Figure 6J reports no error bars, and the corresponding fits are not shown. Please report uncertainties and the fitting details for the annealed samples so the independence of peak 2 under annealing can be evaluated.","section":"Results, Morphology evolution during annealing / Fig. 6"},{"comment":"The ISI analysis assumes that the films contain no internal voids and attributes all vacuum contrast to surface roughness. This assumption is load-bearing for the conclusion that peak 2 is essentially 100% compositional contrast while peak 1 has a topographic contribution. The manuscript does not justify the no-void assumption for a phase-separating film; voids can form during deposition or phase separation. Please address this assumption experimentally or explicitly state that the compositional/topographic assignment is conditional on it.","section":"Results, Bulk structure analysis with RSoXS / Eq. 7"}],"minor_comments":[{"comment":"The paragraph ends with the incomplete sentence 'Overall, the PV distances' (page 8). The sentence should be completed or removed.","section":"Results, AFM Surface Morphology and Phase Imaging"},{"comment":"There is an apparent mismatch between the text and the Figure 6 caption. The text refers to the fastest Rdep = 0.8 nm/s film as 'shown in Figure 6A', but the caption assigns panel A to Rdep = 0.008 nm/s. Please correct the cross-reference.","section":"Results, Morphology evolution during annealing / Fig. 6"},{"comment":"Typo: 'homogenous' should be 'homogeneous'.","section":"Abstract"},{"comment":"The ISI model in Eq. (7) has three open parameters (B, A, E0), but the contrast functions Δn_x are derived from NEXAFS data that the authors note may contain a mixed population of DO37 conformers. Please clarify how this uncertainty propagates into the fitted composition weights A, especially for the 100% composition conclusion for peak 2.","section":"Materials & Methods, Resonant Soft X-ray Scattering"},{"comment":"The AFM scan size for the annealed samples is smaller than that for Figure 1, which changes the accessible q range and may influence the observed PSD. Please state the scan sizes and ensure that the before/after comparison is not affected by this difference.","section":"Results, Morphology evolution during annealing / Fig. 6"}],"recommendation":"major_revision","confidential_remarks":"The paper's central claim is plausible and interesting, but it is currently contingent on the two-peak decomposition of PSDs that are not clearly bimodal. The authors should be asked to provide independent validation of the high-q peak before the tunability claim can be accepted. This is fixable with additional analysis and is within the scope of a major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine experimental extension of the group's earlier TPD/DO37 work, with deposition rate replacing substrate temperature as the knob. The AFM PSD trend at the two fastest rates (0.8 and 0.4 nm/s) is clean, the RSoXS bulk confirmation at 0.8 nm/s is a nice complement, and the annealing result—peak 2 stays put while peak 1 coarsens—is genuinely interesting. They also ship code and raw data, which is more than most papers in this subfield.\n\nThe soft spot is exactly the one flagged in the stress test. The paper's own text says the 0.04 and 0.08 nm/s PSDs show a single broad feature that they 'for now' claim is the convolution of two length scales, and for 0.008 nm/s peak 2 is only a weak shoulder. The monotonic shift of peak 2 in Fig. 3C therefore leans heavily on fits to features that are not resolved. RSoXS at 0.04 nm/s needed a different peak-finding method because the peaks overlap, and RSoXS was only taken at two rates. So the 'finely tunable' conclusion is not yet nailed down across the full range. The direction of the trend is probably right—the two high-rate points and the shoulder behavior at low rate are consistent—but the quantitative claim and the error bars on peak position for the middle rates are fit-dependent.\n\nAlso worth noting: the ISI attribution uses three open parameters and NEXAFS contrast functions from their own prior paper. That is acceptable as supporting evidence but too flexible to carry the '100% composition contrast' claim. The island-layer-island explanation for the AFM/RSoXS discrepancy is speculative; they present it as one possibility, which is fair.\n\nBottom line: this deserves a serious referee. The experimental design is sound, the data are openly available, and the authors flag their own assumptions. The referee should ask for independent validation of the two-peak decomposition—for example, real-space autocorrelation or Fourier-filtered binarized images with proper error bars—before the tunability claim is accepted as stated. I would read a revised version.","headline":"Deposition-rate tuning of the small domain spacing is plausible but not yet fully established—the decisive peak-2 positions come from forced fits to single broad PSD features.","tokens_in":17479,"tokens_out":2196,"would_cite":true,"duration_ms":21019,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Deposition rate alone determines the phase-separated domain size and roughness in vapor-deposited TPD/DO37 glass films.","keywords":["vapor-deposited glasses","phase separation","thin-film morphology","organic semiconductor blends","deposition rate","power spectral density","resonant soft X-ray scattering","TPD/DO37"],"falsifier":"Re-measure the middle-rate film (0.08 nm/s) with larger scan areas and different AFM tips and compute the power spectral density: if the small-domain peak shifts with scan size or disappears when the large-scale background is removed, it is an artifact. A second check is bulk scattering near an X-ray energy where the two molecules have nearly identical refractive index—if the small-domain peak still appears there, it cannot be compositional.","tokens_in":16531,"feed_emoji":"🎛️","tokens_out":5348,"duration_ms":50136,"temperature":0.7,"pith_summary":"This paper reports that, in a co-deposited 50/50 blend of TPD and DO37, the size and roughness of phase-separated domains can be controlled by changing only the deposition rate, with substrate temperature and composition held fixed. Slower deposition gives molecules more time in the mobile surface layer to phase-separate and coarsen, so films become rougher and develop larger center-to-center domain spacings; faster deposition traps finer, smoother structures. The authors identify two coexisting length scales with different rate dependences: a larger-scale feature that barely moves with rate and a smaller-scale feature that grows strongly as rate decreases. Bulk X-ray scattering shows the separation runs through the whole film, and annealing shows the short length scale is frozen while the long one continues to coarsen. If correct, the result gives device makers a one-knob control over blend morphology and extends the surface-equilibration picture to phase-separating vapor-deposited systems.","feed_headline":"Slower vapor deposition grows larger domains in organic glass films","feed_subtitle":"At a fixed temperature, changing only deposition rate tunes blend morphology—a useful control for organic electronic devices.","key_machinery":"The surface equilibration mechanism: during deposition, molecules at the free surface have enhanced mobility and can diffuse, phase-separate, and coarsen before being buried and kinetically frozen. The paper's quantitative tool is the power spectral density (PSD) of AFM height and phase images, whose peaks give characteristic real-space lengths d=2π/q; two Gaussian peaks are fit after linear background subtraction. The percolation-to-cluster transition supplies the two-length-scale interpretation: an initially bicontinuous spinodal network breaks into droplet clusters, so a large spinodal-like scale and a small droplet scale coexist. Resonant soft X-ray scattering (RSoXS) provides the bulk c","core_discovery":"The central claim is that kinetic arrest during vapor deposition, not only thermodynamics, sets the final phase-separated morphology of a molecular glass blend. By varying deposition rate from 0.008 to 0.8 nm/s at constant substrate temperature in a TPD/DO37 film, the authors find two distinct, rate-dependent structural length scales at the surface: a low-q peak near q≈0.015 nm⁻¹ that is nearly rate-independent and a high-q peak that shifts to larger length scales as rate is lowered. AFM power spectral densities and resonant soft X-ray scattering both show these two scales, with RSoXS energy-dependent contrast indicating the shorter scale is almost entirely compositional and the longer scale","pith_inferences":["If deposition rate can be programmed during a single run, one could build graded morphologies—small domains near the substrate and larger domains near the surface—without any post-processing step.","The rate dependence of the short length scale resembles effective-rate control of surface equilibration depth; a testable extension is that the same short length scale can be produced by trading rate against substrate temperature according to rate-temperature superposition.","The factor-of-two discrepancy between AFM and RSoXS suggests a growth scenario in which surface islands cap and fuse into larger bulk domains; cross-sectional imaging or depth-resolved scattering could test whether domain size increases with depth.","The decoupling of roughness from composition implies two independent design levers for organic devices: deposition rate sets domain spacing, while a mild anneal can smooth the surface without destroying the desired phase-separated bulk structure."],"forward_implications":["Domain spacing in vapor-deposited organic blends can be continuously tuned by deposition rate alone, without changing temperature or composition.","Because the short length scale is set during deposition and does not evolve on annealing, PVD can create stable small domains that later thermal treatment cannot erase.","Surface AFM measurements significantly underestimate the bulk small-domain spacing (by about a factor of two), so bulk characterization is needed to predict device-relevant morphology.","The energy-dependent RSoXS contrast identifies the short length scale as compositional and the long scale as partly topographical, meaning surface roughness and composition can coarsen independently.","The annealing results indicate capillary levelling smooths surface height fluctuations without changing the underlying compositional domains, decoupling topography from phase-separation structure."],"fun_headline_variants":["Deposition rate tunes domain size in vapor-deposited glass films","Slower vapor deposition yields larger phase-separated domains","Kinetic arrest controls morphology in vapor-deposited organic blends","Vapor deposition rate dictates domain scaling in glass films","Rate-dependent phase separation in vapor-deposited glasses"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"That the two peaks seen in the roughness spectrum are truly two separate physical length scales—especially the smaller-domain peak, which shows up as only a broad hump in the middle-rate films and a weak shoulder in the slowest film; if that peak is a mathematical artifact of the fitting procedure, the claim of independently tunable small-scale structure is not supported.","fun_headline_variants_meta":{"raw":{"variants":["Deposition rate tunes domain size in vapor-deposited glass films","Slower vapor deposition yields larger phase-separated domains","Kinetic arrest controls morphology in vapor-deposited organic blends","Vapor deposition rate dictates domain scaling in glass films","Rate-dependent phase separation in vapor-deposited glasses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000672,"raw_usage":{"total_tokens":2932,"prompt_tokens":812,"completion_tokens":2120,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":2038}},"tokens_in":556,"tokens_out":2120,"duration_ms":12955,"temperature":1.0,"reasoning_tokens":2038,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T10:52:24.323726+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-measure the middle-rate film (0.08 nm/s) with larger scan areas and different AFM tips and compute the power spectral density: if the small-domain peak shifts with scan size or disappears when the large-scale background is removed, it is an artifact. A second check is bulk scattering near an X-ray energy where the two molecules have nearly identical refractive index—if the small-domain peak still appears there, it cannot be compositional.","supporting_citations":[],"review_version":1}