Pith. sign in

REVIEW 3 major objections 5 minor 1 references

High Thermal Conductivity of Back-End-of-Line Compatible Diamond Films

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read Diamond films grown below 400 °C conduct heat at 73–86 W/m·K, making them candidates for heat-spreading layers in chip wiring.

desk verdict Plausible new data on low-temperature diamond, but the isotropic TDTR analysis and missing error bars leave the headline numbers underdetermined. read the letter →

arxiv 2608.08534 v1 pith:SUVUL47B submitted 2026-08-09 cond-mat.mtrl-sci cond-mat.mes-hall

classification cond-mat.mtrl-scicond-mat.mes-hall
keywords diamondgrowththermalmanagementback-end-of-linecompatibletime-domainthermoreflectancepolycrystallinelow-temperatureCVDconductivityheat-spreadingdielectric
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper reports that polycrystalline diamond films grown directly on silicon at a substrate temperature below 400 °C retain enough crystalline quality to conduct heat far better than the dielectrics used in modern chip wiring. From time-domain thermoreflectance measurements on films about 760 and 1000 nm thick, the authors extract room-temperature effective thermal conductivities of $\sim 73$ and $\sim 86~\mathrm{W\,m^{-1}\,K^{-1}}$, using an isotropic thermal model because the measured signal is about equally sensitive to in-plane and out-of-plane transport. These values are roughly two orders of magnitude above conventional interlayer dielectrics, so the paper positions low-temperature diamond as a candidate electrically insulating heat-spreading layer for back-end-of-line integration. A sympathetic reader should see the central claim as conditional on the thermal analysis: if the isotropic treatment is valid, the numbers stand as reported and the BEOL application is plausible.

What carries the argument

The central object is the TDTR measurement combined with a sensitivity analysis, applied to submicrometer, highly conductive diamond films. A 10.08 MHz modulated pump beam heats an aluminum transducer on the nucleation side while a probe beam reads the thermoreflectance decay, and the in-phase/out-of-phase ratio is fit to a thermal model. The load-bearing step is the sensitivity analysis: it shows the TDTR ratio responds comparably to cross-plane and in-plane thermal conductivity, which the authors use to justify fixing $\kappa_r = \kappa_z = \kappa_{\mathrm{eff}}$ and reporting a single effective conductivity. Raman spectroscopy and SEM do supporting work by establishing diamond phase purity and large grains, making the high extracted conductivity physically plausible.

What would settle it

Fit the same TDTR data with an anisotropic model that allows separate in-plane and cross-plane conductivities, using multiple laser spot sizes and modulation frequencies. If the best-fit cross-plane conductivity differs from the reported $73$ or $86~\mathrm{W\,m^{-1}\,K^{-1}}$ by more than the measurement uncertainty, the isotropic assumption fails and the headline numbers are not the through-plane conductivities.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that sub-400 °C microwave-plasma CVD can grow polycrystalline diamond on Si with large grains, a sharp diamond Raman peak near 1332 cm$^{-1}$, and only minor non-diamond carbon, despite the low thermal budget. TDTR performed on the nucleation side after mechanical transfer of the film yields effective thermal conductivities of $\sim 73~\mathrm{W\,m^{-1}\,K^{-1}}$ for the 760 nm film and $\sim 86~\mathrm{W\,m^{-1}\,K^{-1}}$ for the 1000 nm film at room temperature, with a decreasing empirical exponential trend as temperature rises to 100 °C. Because a sensitivity analysis shows comparable response to in-plane and out-of-plane conductivity, and SEM indicates nearly isotropic grain structure, the films are analyzed with an isotropic model in which $\kappa_r = \kappa_z = \kappa_{\mathrm{eff}}$. The paper presents these values as demonstration that low-temperature diamond can serve as a BEOL-compatible dielectric heat-spreading layer, not as a replacement for bulk diamond but as an order-of-magnitude improvement over conventional dielectrics.

Load-bearing premise

The load-bearing premise is that these thin diamond films can be treated as thermally isotropic, so that one effective conductivity describes the heat flow a chip would actually experience; if the films are anisotropic, the reported $73$ and $86~\mathrm{W\,m^{-1}\,K^{-1}}$ values may not be the through-plane number a BEOL heat-spreading layer needs.

Editorial extensions

If this is right

  • If the extracted values hold, sub-400 °C diamond offers a BEOL-compatible dielectric heat spreader with thermal conductivity about two orders of magnitude above conventional interlayer dielectrics.
  • The 1000 nm film conducting better than the 760 nm film is consistent with grain coarsening reducing boundary scattering, so thicker low-temperature diamond films should conduct even better.
  • Because the measurements are taken from the nucleation side, where grains are smaller, growth-side measurements would likely give values at least as high, strengthening the case for heat spreading.
  • The measured values give device thermal simulations a concrete input for estimating temperature reduction in 3D-stacked chips with diamond heat-spreading layers.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: Because the reported numbers come from the nucleation side, the growth-side cross-plane conductivity of the same films is plausibly higher; a growth-side TDTR measurement would test this directly.
  • Inference: If the films are mildly anisotropic, the isotropic effective value could overstate the through-plane conductivity that matters for BEOL heat spreading; a multi-spot-size fit separating in-plane and cross-plane components would settle it.
  • Inference: Applying the same growth recipe to a thickness series from roughly 0.3 to 2 µm would map how conductivity climbs with grain coarsening and reveal whether an optimum thickness exists before the nucleation layer dominates.
  • Inference: The comparison table suggests reported low-temperature diamond conductivities scatter widely; re-measuring nominally identical films with one standardized TDTR configuration, including sensitivity reporting, would likely narrow the spread more than growth changes alone.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript reports polycrystalline diamond films grown directly on Si at a substrate temperature below 400 °C by MPCVD. Two films with average thicknesses of about 760 and 1000 nm were characterized by Raman spectroscopy and SEM, and their thermal conductivities were measured by time-domain thermoreflectance (TDTR) after mechanical release and transfer to Kapton tape, with the Al transducer on the nucleation side. A sensitivity analysis is used to argue that the TDTR signal is comparably sensitive to in-plane and cross-plane thermal conductivity, and the data are therefore fit with an isotropic model. The paper reports room-temperature effective thermal conductivities of approximately 73 and 86 W m-1 K-1 for the two films, with a decaying temperature dependence up to 100 °C, and argues that these values are about two orders of magnitude larger than conventional BEOL dielectrics, supporting the use of low-temperature diamond as a heat-spreading layer.

Significance. If the extracted thermal conductivity values are reliable, the paper provides a useful data point for BEOL-compatible diamond integration: the films are grown below 400 °C, show diamond Raman peaks with FWHM values of 6.7 and 8.7 cm-1, and the reported conductivities are substantially higher than those of conventional interlayer dielectrics. The paper also includes a helpful comparison table of thin diamond film thermal conductivity values from the literature. However, the central quantitative claim rests on an isotropic assumption that is not established and on a TDTR analysis that lacks uncertainty quantification. The authors themselves note that low sensitivity can allow different parameter combinations to fit the data, but they do not provide the residual, correlation, or confidence-interval analysis needed to show that their extracted values are uniquely determined. The decisive missing check is an anisotropic refit of the same data.

major comments (3)
  1. [Results and Discussion, Fig. 3b and following paragraph] The statement that 'the TDTR ratio exhibits comparable sensitivity of both the in-plane and out-of-plane thermal conductivity' is used to justify fitting an isotropic model, but comparable sensitivity is not evidence of isotropy; it is evidence that both components influence the measured signal. With a single modulation frequency (10.08 MHz) and a single pair of spot sizes (4.38 and 2.7 μm), an anisotropic fit with κr and κz as independent parameters is the appropriate test. The prior work cited by the authors (Sood et al., Ref. 16) reports κr/κz ≈ 46/89 for a 1-μm-thick nucleation-side polycrystalline diamond film, so the possibility of significant anisotropy is concrete and must be ruled out before the isotropic κeff values can be assigned to the films.
  2. [Results and Discussion, Fig. 3 and final paragraph before Summary] The paper correctly warns that when sensitivity is low, similar agreement between the experimental data and the thermal model may be obtained with different combinations of thermal properties, but it does not provide the uncertainty analysis needed to show that this does not apply here. For the 760 nm film, L/κeff ≈ 1.0×10^-8 m²K/W, whereas the fitted interface resistances are 1/TBC_Al/diamond ≈ 1.3×10^-8 m²K/W and 1/TBC_diamond/tape ≈ 5.6×10^-8 m²K/W; the film resistance is thus a minority of the total series resistance, and κeff can trade off against the two TBCs. No confidence intervals, residual analysis, or parameter-correlation information is reported for κeff, TBC_Al/diamond, or TBC_diamond/tape, and Fig. 3a shows no error bars. Without this, the claimed values of 73 and 86 W m-1 K-1 are not fully supported.
  3. [Results and Discussion, Fig. 3b caption and sensitivity discussion] The sensitivity analysis is presented only in qualitative terms: the sensitivity coefficients are described as 'comparable', but no absolute sensitivity magnitudes, noise levels, or quantitative comparisons are provided. The Fig. 3b caption is also incomplete, with parameter symbols missing so that the exact parameter set used for the calculation cannot be reconstructed. Because the central claim is that the measurement is sufficiently sensitive to justify the extracted conductivities, the authors should report quantitative sensitivity coefficients and a parameter-uncertainty propagation, or show how the extracted values change under a plausible range of model assumptions.
minor comments (5)
  1. [Introduction, Lundh et al. sentence] The sentence containing '45±25 W m-1 K-1 sdf' has an extraneous 'sdf' that should be removed.
  2. [Results and Discussion, after Fig. 3] The lines 'Alk diamondk Alth diamondth tapek' appear to be floating equation fragments from a corrupted equation; they should be cleaned up or replaced with the intended equation.
  3. [Fig. 3b caption] The caption uses bare '=' signs without variable names (e.g., '= 200 W m-1 K-1', '= 73 W m-1 K-1', '= 78 nm'). Please restore the symbols so the parameter set is fully specified.
  4. [Temperature dependence, Fig. 3a] The 'empirical exponential decay temperature dependence relationship' is mentioned but no functional form or fitting parameters are given; please provide the equation or a reference that specifies it.
  5. [Table 1] The entries for this work list no uncertainty; if uncertainty estimates are available for the reported values, they should be included in the table for comparison with the literature values.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported thermal conductivities are TDTR fits, not predictions pre-wired to the model inputs.

full rationale

The paper's central quantities, the effective thermal conductivities of 73 and 86 W/mK, are extracted by fitting a standard isotropic TDTR thermal model to measured thermoreflectance ratios; this is a parameter-estimation measurement, not a derivation in which an output is equivalent to an input by construction. The sensitivity analysis that motivates the isotropic model uses the extracted conductivity as an illustrative input, but it is a diagnostic and does not generate the fitted value. The isotropic assumption itself is a modeling choice that could be questioned (the paper does not quantify the sensitivity magnitudes or test an anisotropic model against Sood et al.'s observed anisotropy), but an insufficiently justified assumption is a correctness/uncertainty concern, not circularity. The manuscript itself cautions, in the Sensitivity analysis section before the Summary, that with low sensitivity 'similar agreement between the experimental data and the thermal model may be obtained using different combinations of thermal properties'; this is weighed here as an honest uncertainty caveat, not as evidence that the reported values reduce to the model inputs, because the authors do not claim the values are derived from that assumption. The paper cites prior work by the corresponding author for TDTR experimental details and for context, but these citations are not load-bearing: the TDTR method is standard, the extracted parameters are compared with external literature values (Sood, Anaya, Malakoutian, Tzeng, etc.), and no uniqueness theorem or ansatz is imported from the authors' own prior work to force the result. Accordingly, no circular step is exhibited and the score is 0.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim is a measured thermal conductivity value, so the model inputs are free parameters in the TDTR fit. The most important free parameters are the effective conductivity itself and the thermal boundary conductances. The assumptions include the isotropic film model and the validity of the multilayer heat conduction model. No new physical entities are introduced.

free parameters (7)
  • Effective thermal conductivity of 760 nm film = 73 W/mK
    Fitted from TDTR data using the isotropic model; the central reported quantity.
  • Effective thermal conductivity of 1000 nm film = 86 W/mK (listed as 85 in Fig. 3b caption)
    Fitted from TDTR data using the isotropic model; the central reported quantity.
  • TBC Al/diamond for 760 nm film = 77.8 MW/m2K
    Used in the TDTR model; likely fitted along with the conductivity.
  • TBC Al/diamond for 1000 nm film = 52.5 MW/m2K
    Used in the TDTR model; likely fitted along with the conductivity.
  • TBC diamond/tape for 760 nm film = 18 MW/m2K
    Used in the TDTR model; likely fitted to the data.
  • TBC diamond/tape for 1000 nm film = 20 MW/m2K
    Used in the TDTR model; likely fitted to the data.
  • Kapton tape thermal conductivity = 0.15 W/mK
    Assumed value for the tape substrate in the TDTR model; not measured in this work.
assumptions (6)
  • domain assumption The diamond film can be modeled as a homogeneous, isotropic thermal conductor (kappa_r = kappa_z = kappa_eff).
    The paper adopts an isotropic model because the TDTR sensitivities to in-plane and out-of-plane conductivity are comparable, but it does not establish that the film is actually isotropic. This assumption directly determines the reported values.
  • domain assumption The standard multilayer TDTR heat conduction solution applies to the Al/diamond/tape stack.
    The TDTR analysis relies on the usual cylindrical multilayer thermal model; the paper does not validate this model for the transferred film geometry.
  • domain assumption The Kapton tape is semi-infinite with a known thermal conductivity of 0.15 W/mK.
    The tape is the bottom layer in the model; its properties are assumed from typical Kapton values rather than measured.
  • ad hoc to paper Mechanical release with Scotch tape and transfer to Kapton does not alter the diamond film's thermal properties.
    The film is removed from Si and flipped before measurement; any strain, cracking, or contamination introduced by the peel is not characterized.
  • domain assumption The Raman FWHM values (6.7 and 8.7 cm-1) indicate sufficient diamond phase purity for the thermal analysis to be meaningful.
    The paper uses Raman as evidence of quality but does not quantify the sp2 fraction or its impact on thermal transport.
  • domain assumption The aluminum transducer's thermal conductivity (200 W/mK) derived from four-probe and Wiedemann-Franz is correct.
    The Al layer is the heat sink; an error in its conductivity would propagate into the extracted diamond thermal conductivity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of High Thermal Conductivity of Back-End-of-Line Compatible Diamond Films." pith.science (2026). https://pith.science/paper/SUVUL47B

@misc{pith2026260808534,
  author       = {Pith},
  title        = {Pith review of: High Thermal Conductivity of Back-End-of-Line Compatible Diamond Films},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SUVUL47B}},
  note         = {Machine review of arXiv:2608.08534}
}
read the original abstract

Back-end-of-line (BEOL) thermal management requires electrically insulating heat-spreading dielectric that can be integrated within thermal budgets below 400 C. Here, we report polycrystalline diamond films grown directly on Si at a substrate temperature below 400C. Two films with average thickness of 760 and 1000 nm were characterized by Raman spectroscopy, scanning electron microscopy (SEM), and time-domain thermoreflectance (TDTR). Raman spectra show a sharp diamond peak with minor signatures of non-diamond carbon, while SEM reveals lateral growth and large grain size. Temperature dependent TDTR measurements were performed from room temperature to 100C. Sensitivity analysis indicates that the sensitivity of cross-plane thermal conductivity is comparative to the in-plane thermal conductivity. Accordingly, the films were analyzed using an isotropic thermal model by considering the nearly-isotropic grain structure, yielding room temperature effective thermal conductivity of 73 and 86 W m-1 K-1, respectively. These values are about two orders of magnitude higher than those of conventional dielectric materials and demonstrate the potential of diamond films grown at low temperatures as dielectric heat-spreading layers.

Figures

Figures reproduced from arXiv: 2608.08534 by the authors.

Figure 1
Figure 1. TDTR measurement configuration and Raman characterization. (a) Schematic of the TDTR measurement geometry after mechanical transfer of the diamond film and deposition of the Al transducer on the nucleation side. (b) Raman spectra of the diamond films. As shown in Fig. 1b, Raman results were used to verify the diamond phase and the crystalline quality of our samples. Both spectra show a clear diamond peak near 1332 c… view at source ↗
Figure 2
Figure 2. Representative morphology of the diamond films. (a, b) Plan-view and (c, d) cross-sectional SEM images acquired from separate specimens grown under the same conditions as the films used for TDTR measurements. Representative plane-view and cross sectional SEM images of diamond films grown under the same conditions are shown in [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Thermal conductivity of diamond films. (a) Effective thermal conductivity extracted from TDTR measurement of the 760 nm and 1000 nm samples. (b) Sensitivity coefficients of in-plane and out-of-plane thermal conductivity, calculated using = 200 W m-1 K-1 which was derived from four probe measurements and the Wiedemann￾Franz law, = 73 W m-1 K-1 , TBCAl/diamond = 77.8 MW m-2 K-1 , = 78 nm, = 760 nm, TBCdiamond/tape = 1… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

1 extracted references · 1 canonical work pages

  1. [1]

    Thermal management materials for 3D-stacked integrated circuits,

    High Thermal Conductivity of Back-End-of-Line Compatible Diamond Films Jinwen Liu1, Chufei Cheng2, Feifei Tan3, Jinquan Zhang1, Di Lu3, *, Bing Dai2, *, Jiaqi Zhu2, Runsheng Wang1, 4, Zhe Cheng1, 4, * 1 School of Software & Microelectronics, Peking University, Beijing 100871, China 2 National Key Laboratory of Science and Technology on Advanced Composites...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.