Graphenic-pyrolyzed polymer thin films for gas sensors

January 23, 2026 – Researchers from the University of the Bundeswehr Munich and Ulm University (Germany) have presented a new study of ultrathin conductive carbon films fabricated from pyrolyzed photoresist. TEM and Raman spectroscopy revealed nanocrystalline sp2 domains within the films. The scalable fabrication via standard lithography enabled their use as highly sensitive NO2 gas sensors on full 4-inch wafers, marking a significant step toward a new generation of ultrathin sensor platforms.

Glassy carbon was originally described by Jenkins and Kawamura as a dense, isotropic, and conductive form of carbon that readily forms from polymeric precursors under inert conditions.[1] Decades later, two-dimensional carbon materials such as graphene – first produced and identified in 2004 by Novoselov and Geim – and single-layer amorphous carbon (MAC) gained significant importance due to their exceptional electronic and mechanical properties. Yet fabricating devices from single-layer graphene remains challenging: the process demands high temperatures, catalytic substrates, and complex multi-step transfer techniques. To overcome these limitations, recent studies have turned to scalable, process-compatible carbon layers derived from pyrolyzed polymer films (PPF).[2],[3],[4] PPF-based devices can be manufactured using highly scalable processes compatible with standard semiconductor production lines – an approach already demonstrated in the fabrication of gate layers for planar high-power hot electron emitters.[5]

The fabrication sequence is summarized in the left panel of Fig. 1. In the first step, AZ nLOF 2070 resist is applied via spin coating at varying degrees of dilution. This thin layer is patterned through UV lithography and a photomask, followed by a post-exposure bake that ensures crosslinking of the resist and stabilization of the polymer framework. To define the contact electrodes (100PPF), a second layer of undiluted AZ nLOF 2070 is spin-coated directly on top of the first and re-patterned. Owing to their greater thickness, these contact paths exhibit significantly higher conductance than the thinner channel regions.

Both layers are then developed simultaneously in AZ 2026 MIF. As illustrated in the right panel of Fig. 1, the researchers processed full 4-inch Si/SiO2 wafers into devices featuring a transfer length method (TLM) layout. The inset at the bottom right of this panel shows a single die, representing one of 61 total units on the wafer. The final PPF thickness is determined by the dilution and spin speed used during fabrication, and subsequently by pyrolytic shrinkage. A higher dilution of the resist leads to proportionally greater shrinkage during pyrolysis, as a larger fraction of the initial material consists of volatile components such as the dilutor.[5],[6]

The left panel of Fig. 2 displays a representative AFM scan of an outer die, centered on the edge of the ultrathin 3PPF channel, with a corresponding line profile superimposed. Between the SiO2 substrate and the continuous PPF film, the team observed a disordered transition zone with increased roughness, likely originating from lithographic residues not fully removed by the developer. Similar measurements conducted on the central die for 5PPF and 100PPF yielded average thicknesses of 1.1 nm and 73 nm, respectively, in good agreement with prior reports.[5]

As shown in the right panel of Fig. 2, nanocrystalline domains with a lateral size of approximately 5 nm are visible, featuring hexagonal ordering in the corresponding fast Fourier transform (FFT).[7] This sixfold modulation resembles patterns reported for Ni-assisted ‘glassy graphene’ films,[8] yet the authors observe it here without a catalyst. Such quasi-periodic contrast may also arise from Moiré effects due to local overlapping or bilayer regions, implying localized higher crystallinity. This structural motif is consistent with nanographitic sp2 domains embedded within a disordered matrix.[9],[10] While these regions might extend over larger areas, they could be partially obscured by remaining carbonaceous impurities.[11] The film also exhibits filamentous structures reminiscent of classic glassy carbon and pyrolyzed resins,[1],[10] consistent with previous TEM studies describing the coexistence of ordered and disordered regions in pyrolyzed carbon systems.[12]

The scientists employed X-ray photoelectron spectroscopy (XPS) to assess and quantify the sp2/sp3 hybridization ratio across PPFs of varying thicknesses. C 1s core-level spectra (Fig. 3, left panel) were deconvoluted into sp2 carbon, sp3 carbon (C–C, ~285.0 eV), and minor components corresponding to C–O, C=O, as well as shake-up satellites originating from π–π* transitions. All samples were dominated by sp2 bonding, indicating efficient graphitization during pyrolysis.

Raman spectroscopy provides further insights into structural order, defect density, and lateral domain size (La). The right panel of Fig. 3 displays Raman spectra for 3PPF and 100PPF films. The G-band – associated with E2g in-plane stretching vibrations of sp2 atoms in both rings and chains – consistently appeared above 1600 cm–1. This position is typical for nanocrystalline graphite and marks the upper limit of ‘Stage 1’ in the Ferrari–Robertson disorder model.[13] In this regime, nanocrystalline graphite with moderate disorder is characterized by an upward G-shift and a dispersive D-peak following the Tuinstra–Koenig (TK) relation.[14] The highest G-band frequency was observed for the 5PPF film (1604 cm–1), which – together with its larger domain size and higher conductivity compared to the thinner 3PPF – reflects an optimized short-range order, confirming previous observations.[5] The G-band width (FWHM) remained nearly identical for all samples (~61 cm–1), suggesting that bond-length fluctuations do not depend on film thickness. These results are in good agreement with previous studies on pyrolyzed, sp2-rich carbon networks.[1],[9],[10]

The disorder-activated D-peak (~1350 cm–1), originating from the breathing modes of six-membered rings, appeared in all spectra, confirming that aromatic character is retained despite structural disorder.[1],[13] The ID/IG ratio was lowest for 5PPF (1.08) and slightly higher for the 100PPF layers (1.15). Notably, the 3PPF film exhibited the highest ratio (1.19), which – combined with the slight reduction in sp2 content observed via XPS – points to a higher defect density in the ultra-thin limit.[15] Since all films lie at the end of ‘Stage 1’ (above the 2 nm threshold), the TK relation[14] remains applicable, yielding calculated crystallite sizes of 3.7 nm for 3PPF and 4.1 nm for 5PPF – in excellent agreement with the TEM findings.

All samples displayed a weak but consistent 2D band between 2670 and 2685 cm–1. The I2D/IG ratios remained below 0.09, significantly lower than values for monolayer graphene. Despite the monolayer thickness of the 3PPF film, its weak and broadened 2D signal – shifted to 2670 cm–1 – indicates that the formation of a coherent π-electron band structure is suppressed by small domain sizes and significant structural distortions. These second-order features are characteristic of disordered, few-layer sp2 carbon,[9],[15] where strong structural distortions and small lateral domains impede the resonance process required for a strong 2D signal.[10] In summary, the ID/IG ratios, the consistent G-peak width, and the suppressed 2D band collectively confirm the nanocrystalline, glassy character of the investigated PPFs.[1],[10]

To assess the electrical properties and process uniformity of the ultrathin PPFs, the team performed wafer-scale sheet resistance (RS) measurements using the Transmission Line Method (TLM). The left panel of Fig. 4 shows the determined sheet resistance for 61 chips across a 4-inch wafer. Remarkably, RS could be extracted for every single chip – a 100% yield that is striking given the near-monolayer thickness of the channels. A slight, off-center radial gradient is evident, with values decreasing from the periphery toward the center. This aligns with expected thickness variations from manual spin-coating and potential temperature gradients during radiation-heated, cold-wall pyrolysis. Although the measured sheet resistance varies by approximately one order of magnitude across the wafer,[5] this variation is primarily confined to the outermost chips. The overall distribution remains continuous and uniform, suggesting that such fluctuations could be readily mitigated through industrial automation.

Within the central 3×3 chip array (Fig. 4, left panel) the mean sheet resistance is (1.99 ± 0.40) × 106 Ω/□. Based on a film thickness of 0.34 ± 0.05 nm (determined by AFM on the central chip), the corresponding conductivity is estimated at (1.5 ± 0.3) × 103 S/m. Additional conductivity values for selected positions across the wafer are shown in the top-middle panel of Fig. 4. While these results agree well with reports on thicker pyrolyzed films,[6] the values are reduced by a factor of 2–3, reflecting the incomplete aromatization and restricted charge transport pathways in the ultra-thin 3PPF. By comparison, 5PPF films previously exhibited conductivities approximately one order of magnitude higher.[5]

To evaluate the gas-sensing performance of the 3PPF films, the researchers conducted feasibility studies using a narrow-channel (100 μm) TLM device. The sensor was manually wire-bonded with silver paste onto a leaded substrate and mounted in a sealed test chamber. Resistance changes were monitored under atmospheric conditions during exposure to 22.5 ppm NO2 in N2 for intervals of 1 to 10 minutes. To accelerate sensor recovery, UV illumination (375 nm) was employed.[16]

As shown in the upper-right panel of Fig. 4, the 3PPF films behave as reliable p-type materials, showing a characteristic decrease in resistance upon exposure to electron-withdrawing NO2 molecules. Based on a signal-to-noise ratio (SNR) of 3, an empirical limit of detection (LOD) of 2.99 ppm was determined. Although some literature reports NO2 sensitivity in the ppb range, this detection limit is highly promising given the exposure time, reliability, and stability of these ultra-thin films.[17],[18] The dynamic response and typical relative percentage change are further detailed in the lower-right panel of Fig. 4, highlighting the rapid resistance variation during gas injection and the subsequent recovery phase.

TEM and Raman spectroscopy confirm a stable, nanocrystalline sp2 structure for the 3PPF type, despite its sub-nanometer thickness. This structural robustness enabled successful feasibility studies for gas sensing on 4-inch wafers, where the films functioned as reliable p-type semiconductors in response to NO2. With a demonstrated 100% device yield across the entire wafer, these findings underscore the potential of pyrolyzed polymer films as a scalable platform for a new generation of highly sensitive, large-scale carbon-based sensors.

Resource:
Galfe, N., Herdl, F., Klenk, S., Quincke, M., Ó Coileáin, C., Lee, K., Kaiser, U., Duesberg, G. S. (2026).
Approaching the Monolayer Limit of Carbon Layers by Pyrolysis of Polymer Films.
Adv. Electron. Mater. 12(3), e00525.
https://doi.org/10.1002/aelm.202500525

References
  1. Uskoković, V. (2021). A historical review of glassy carbon: Synthesis, structure, properties and applications. Carbon Trends, 5, 100116. https://doi.org/10.1016/j.cartre.2021.100116

  2. Angelova, P. and Gölzhäuser, A. (2017). Carbon nanomembranes. Physical Sciences Reviews, 2, 20160105. https://doi.org/10.1515/psr-2016-0105

  3. Qi, Y., Westphal, M., Khayya, N. et al. (2022). Thickness-varied carbon nanomembranes from polycyclic aromatic hydrocarbons. ACS Applied Materials & Interfaces, 14, 9433–9441. https://doi.org/10.1021/acsami.1c22406

  4. Yao, Z., Meyerbröker, N., Qi, Y. et al. (2023). Scalable synthesis of carbon nanomembranes from amorphous molecular layers. ACS Applied Materials & Interfaces, 15, 41101–41108. https://doi.org/10.1021/acsami.3c07369

  5. Herdl, F., Galfe, N., Klenk, S. et al. (2025). Highly efficient planar hot electron emitters based on ultrathin pyrolyzed polymer films. ACS Applied Materials & Interfaces, 17, 34637–34646. https://doi.org/10.1021/acsami.4c19809

  6. Schreiber, M., Lutz, T., Keeley, G. P. et al. (2010). Transparent ultrathin conducting carbon films. Applied Surface Science, 256, 6186–6190. https://doi.org/10.1016/j.apsusc.2010.03.138

  7. Kalita, G., Wakita, K., and Umeno, M. (2012). Low temperature growth of graphene film by microwave assisted surface wave plasma CVD for transparent electrode application. RSC Advances, 2, 2815. https://doi.org/10.1039/c2ra20648k

  8. Dai, X., Wu, J., Qian, Z. et al. (2016). Ultra-smooth glassy graphene thin films for flexible transparent circuits. Science Advances, 2, 1601574. https://doi.org/10.1126/sciadv.1601574

  9. Kaplas, T. and Kuzhir, P. (2017). Ultra-thin pyrocarbon films as a versatile coating material. Nanoscale Research Letters, 12, 121. https://doi.org/10.1186/s11671-017-1896-0

  10. Jurkiewicz, K., Pawlyta, M., Zygadło, D. et al. (2018). Evolution of glassy carbon under heat treatment: correlation structure–mechanical properties. Journal of Materials Science, 53, 3509–3523. https://doi.org/10.1007/s10853-017-1753-7

  11. Linck, M., Hartel, P., Uhlemann, S. et al. (2016). Chromatic aberration correction for atomic resolution TEM imaging from 20 to 80 kV. Physical Review Letters, 117, 76101. https://doi.org/10.1103/PhysRevLett.117.076101

  12. CN, S. K., Possel, C., Dehm, S. et al. (2024). Graphitizability of polymer thin films: an in situ TEM study of thickness effects on nanocrystalline graphene/glassy carbon formation. Macromolecular Materials and Engineering, 309, 2300230. https://doi.org/10.1002/mame.202300230

  13. Ferrari, A. C. and Robertson, J. (2000). Interpretation of Raman spectra of disordered and amorphous carbon. Physical Review B, 61, 14095–14107. https://doi.org/10.1103/PhysRevB.61.14095

  14. Tuinstra, F. and Koenig, J. L. (1970). Raman spectrum of graphite. The Journal of Chemical Physics, 53, 1126–1130. https://doi.org/10.1063/1.1674108

  15. Yuan, R., Guo, Y., Gurgan, I. et al. (2025). Raman spectroscopy analysis of disordered and amorphous carbon materials: a review of empirical correlations. Carbon, 238, 120214. https://doi.org/10.1016/j.carbon.2025.120214

  16. Ricciardella, F., Lee, K., Stelz, T. et al. (2020). Calibration of nonstationary gas sensors based on two-dimensional materials. ACS Omega, 5, 5959–5963. https://doi.org/10.1021/acsomega.9b04325

  17. Ko, G., Kim, H.-Y., Ahn, J., Park, Y.-M., Lee, K.-Y., and Kim, J. (2010). Graphene-based nitrogen dioxide gas sensors. Current and Applied Physics, 10, 1002–1004. https://doi.org/10.1016/j.cap.2009.12.024

  18. Novikov, S., Lebedeva, N., Satrapinski, A., Walden, J., Davydov, V., and Lebedev, A. (2016). Graphene based sensor for environmental monitoring of NO2. Sensors and Actuators B: Chemical, 236, 1054–1060. https://doi.org/10.1016/j.snb.2016.05.114