The 2003 Nature report by Möller, Perlov, Jackson, Taussig and Forrest introduced a write-once read-many-times (WORM) memory pixel that integrates a poly(3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) conductive-polymer fuse in series with a low-temperature thin-film silicon p–i–n diode on flexible stainless-steel foil, writing by a current-controlled, thermally activated un-doping of the polymer. The authors explicitly left open the quantitative physics and deployment validation of this architecture: the temperature-rise estimates neglected film–substrate thermal coupling; the multi-stage switching was tied to uncharacterized morphologically distinct domains and an inferred second species of chains; the pulse-duration and duty-cycle dependence was mapped only qualitatively; the passive-matrix scaling estimates rested on unvalidated assumptions about shunt removal and written-to-forward current ratios; and long-term retention and environmental robustness were unestablished. Later work has built adjacent WORM devices (electron-injection dedoping of PEDOT:PSS with ZnO nanoparticles; redox and proton-motion attributions; silver-nanoparticle and silver-nanowire fuse arrays; sol-gel silica antifuse cells) without closing the architecture's open cores: no time-resolved electrothermal model or direct temperature measurement of the switching transient including film–substrate impedance exists, the morphological origin of the multi-stage switching is unidentified, and no fabricated dense passive-matrix array of the polymer-fuse/diode pixel has been reported.

Background

A write-once read-many-times (WORM) memory is a nonvolatile memory in which each cell can be irreversibly switched from one resistance state to another a single time and read repeatedly; it targets ultralow-cost permanent archival storage, where erasability and the transistor-per-cell architecture of conventional Flash are unnecessary. Passive-matrix addressing connects two-terminal cells at row–column cross-points, so each cell must be strongly nonlinear (typically a series diode) to prevent 'sneak' currents through half-selected and unselected cells from corrupting reads; this is what allows a two-terminal cell rather than a three-transistor cell, shrinking the cell area from about 9F29F^{2} to 4F24F^{2}, where FF is the minimum feature size. PEDOT:PSS (also written PEDT/PSS) is a two-component blend of the conducting, p-doped polymer poly(3,4-ethylenedioxythiophene) with the polyanion polystyrene sulfonic acid serving as counterion; it is electrochromic (its optical absorption changes with oxidation state) and, in its as-deposited conducting form, has a conductivity that drops irreversibly when it is thermally or electrochemically de-doped — the positively charged PEDT chains are reduced to the neutral, insulating form.

Möller, Perlov, Jackson, Taussig and Forrest (Nature 426, 166–169, 2003; companion paper in Journal of Applied Physics 94, 7811–7819, 2003) demonstrated a WORM pixel consisting of a 25–60 nm PEDOT:PSS fuse layer on a low-temperature thin-film silicon p–i–n diode (15 nm n⁺-Si / 400 nm intrinsic Si / 25 nm p⁺-Si) deposited on flexible Al-coated stainless-steel foil. Above an injection threshold of about 4.54.5 V, double injection of holes and electrons (injected electrons neutralizing the hole space charge) produces a transient current peak with a delay scaling as t=d2/[μn(V−Voffset)]t = d^{2}/[\mu_{\mathrm{n}}(V - V_{\mathrm{offset}})], from which an electron mobility μn≈1.7×10−6 cm2V−1s−1\mu_{\mathrm{n}} \approx 1.7 \times 10^{-6}\ \mathrm{cm^{2}V^{-1}s^{-1}} was extracted; current densities near 103 A cm−210^{3}\ \mathrm{A\,cm^{-2}} reached within roughly a microsecond were estimated to heat the film to about 200∘C200^\circ\mathrm{C}, driving the de-doping that permanently reduced the pixel conductivity by up to 10310^{3}–10410^{4}. Switching times as short as about 2 μs2\ \mu\mathrm{s} were observed, roughly 200 isolated films switched to the non-conducting state without a single short, and from the measured current contrast the authors estimated that a 1-Mbit passive-matrix block written in about 1 s on roughly 1 mm21\ \mathrm{mm^{2}} at an assumed 500 nm pixel pitch was feasible. The switching itself was measured on isolated PEDOT:PSS films between Au and indium-tin-oxide (ITO, a transparent conducting oxide) contacts, not on the integrated diode pixel.

The paper's own discussion acknowledged what such a device program still lacked, and this is the question: (i) the temperature-rise calculations neglected the thermal impedance between the polymer film and the substrate, holding the film–substrate interface at room temperature for the first ~100 ns, so the actual transient temperatures that drive un-doping in realistic device stacks were unquantified and no direct temperature measurement existed; (ii) the PEDT/PSS film is morphologically heterogeneous, and the observation of multiple switching peaks with distinct pulse-duration (τ\tau) dependence suggested a putative second species of chains whose precise chemical or structural identity, and hence the origin of the multi-stage switching, was unresolved; (iii) the dependence of the permanent conductivity change on pulse duration and duty cycle was described only qualitatively, leaving optimized minimal-energy write pulses undetermined; (iv) the array estimates assumed the parasitic conducting-polymer shunt could be removed in patterning (demonstrated only by an oxygen-plasma etch on a separate device without a diode) and that written-pixel currents would be far below forward currents, assumptions untested in fabricated, densely patterned passive-matrix arrays with realistic interconnects; and (v) devices were characterized electrically in air but long-term retention, humidity/oxygen stability, and cycling robustness of written and unwritten states in array contexts were unestablished. Reviews of organic memory devices in 2009 and 2014 still describe the charging mechanisms, stability, and industrial readiness of organic memories as unresolved, and the subsequent literature developed WORM memories based on different fuse materials and switching mechanisms rather than closing these gaps for the polymer-fuse/diode architecture.

Problem Statement

For the hybrid WORM memory architecture of Möller et al. (2003) — a 25–60 nm PEDOT:PSS (PEDT/PSS) polymer fuse in series with a low-temperature thin-film silicon p–i–n diode on flexible metal foil, written by current-controlled, thermally activated un-doping of the polymer — establish the predictive physical understanding and the array-level validation that the source paper assumed but did not possess. Concretely, the question asks for five jointly linked results that together make this architecture a quantitatively understood and deployable memory technology: (1) a time-resolved electrothermal description of the write event that includes the thermal impedance between the polymer film and the substrate (rather than holding the film–substrate interface at room temperature for the first ~100 ns), quantifying the actual transient temperatures that drive un-doping in the realistic diode-integrated stack and anchored, if possible, in a direct measurement of the film temperature during switching; (2) identification of the chemical and microstructural origin of the multi-stage switching, i.e. which morphologically distinct PEDT/PSS domains or chain species (including the inferred second species) produce the several switching peaks and their pulse-duration dependence, or a demonstration that no such distinct species is required; (3) an experimental map, over the tested range of pulse duration τ\tau and duty cycle, of how the magnitude and rapidity of the permanent conductivity change depend on those parameters, sufficient to determine the write-pulse schedule of minimum energy that achieves a specified switching yield; (4) fabrication and electrical characterization of densely patterned passive-matrix arrays of this pixel that test the scaling assumptions of the source paper — removal of the conducting-polymer shunt in patterning, written-pixel currents far below forward currents (on the order of 10−410^{-4} or less), and freedom from sneak-path, interconnect, and yield failures at realistic array sizes and addressing schemes; and (5) long-term retention, humidity/oxygen stability, and cycling robustness of both written and unwritten states in array contexts. The five results are jointly required and individually refutable: (1)–(3) constitute the predictive write-physics account and (4)–(5) the array-level validation, so the completion boundary is met only when all five are established — a quantitative, experimentally validated account of the write physics of this pixel (1)–(3) together with a demonstrated dense passive-matrix array meeting the assumed current ratios and stability requirements (4)–(5). No proper subset of (1)–(5) constitutes a resolution.

References

  1. Sven Möller, Craig Perlov, Warren Jackson, Carl Taussig, Stephen R. Forrest, "A polymer/semiconductor write-once read-many-times memory", Nature 426, 166–169 (2003). DOI: 10.1038/nature02070, https://doi.org/10.1038/nature02070
  2. Sven Möller, Stephen R. Forrest, Craig Perlov, Warren Jackson, Carl Taussig, "Electrochromic conductive polymer fuses for hybrid organic/inorganic semiconductor memories", Journal of Applied Physics 94, 7811–7819 (2003). DOI: 10.1063/1.1627482, https://doi.org/10.1063/1.1627482
  3. Jian Hu, Howard M. Branz, P. Stradins, S. Ward, A. Duda, Qi Wang, Craig Perlov, W. Jackson, Carl Taussig, "Write-once diode/antifuse memory element with a sol-gel silica antifuse cured at low temperature", IEEE Electron Device Letters 26, 17–19 (2005). DOI: 10.1109/LED.2004.839218, https://doi.org/10.1109/LED.2004.839218
  4. W. B. Jackson, R. Elder, W. Hamburgen, A. Jeans, H.-J. Kim, H. Luo, P. Mei, C. Perlov, C. Taussig, H. Branz, P. Stradins, Q. Wang, S. Ward, S. Braymen, F. Jeffery, "Amorphous silicon memory arrays", Journal of Non-Crystalline Solids 352, 859–862 (2006). DOI: 10.1016/j.jnoncrysol.2005.11.139, https://doi.org/10.1016/j.jnoncrysol.2005.11.139
  5. Jianpu Wang, Feng Gao, Neil C. Greenham, "Low-power write-once-read-many-times memory devices", Applied Physics Letters 97, 053301 (2010). DOI: 10.1063/1.3473775, https://doi.org/10.1063/1.3473775
  6. Viet Cuong Nguyen, Pooi See Lee, "Resistive Switching Memory Phenomena in PEDOT PSS: Coexistence of Switchable Diode Effect and Write Once Read Many Memory", Scientific Reports 6, 19594 (2016). DOI: 10.1038/srep19594, https://doi.org/10.1038/srep19594
  7. Robert A. Nawrocki, Erin M. Galiger, David P. Ostrowski, Brian A. Bailey, Xin Jiang, Richard M. Voyles, Nikos Kopidakis, Dana C. Olson, Sean E. Shaheen, "An inverted, organic WORM device based on PEDOT:PSS with very low turn-on voltage", Organic Electronics 15, 1791–1798 (2014). DOI: 10.1016/j.orgel.2014.05.003, https://doi.org/10.1016/j.orgel.2014.05.003
  8. Jaakko Leppäniemi, Mikko Aronniemi, Tomi Mattila, Ari Alastalo, Mark Allen, Heikki Seppä, "Printed WORM Memory on a Flexible Substrate Based on Rapid Electrical Sintering of Nanoparticles", IEEE Transactions on Electron Devices 58, 151–159 (2011). DOI: 10.1109/TED.2010.2088402, https://doi.org/10.1109/TED.2010.2088402
  9. Jaakko Leppäniemi, Tomi Mattila, Terho K. Kololuoma, Mika Suhonen, Ari T. Alastalo, "Roll-to-roll printed resistive WORM memory on a flexible substrate", Nanotechnology 23, 305204 (2012). DOI: 10.1088/0957-4484/23/30/305204, https://doi.org/10.1088/0957-4484/23/30/305204
  10. Jaakko Leppäniemi, Tomi Mattila, Kim Eiroma, Toshihiko Miyakawa, Kazuhiro Murata, Ari Alastalo, "Printed Low-Voltage Fuse Memory on Paper", IEEE Electron Device Letters 35, 354–356 (2014). DOI: 10.1109/LED.2014.2300413, https://doi.org/10.1109/LED.2014.2300413
  11. D. Prime, S. Paul, "Overview of organic memory devices", Philosophical Transactions of the Royal Society A 367, 4141–4157 (2009). DOI: 10.1098/rsta.2009.0165, https://doi.org/10.1098/rsta.2009.0165
  12. Jagan Singh Meena, Simon Min Sze, Umesh Chand, Tseung-Yuen Tseng, "Overview of emerging nonvolatile memory technologies", Nanoscale Research Letters 9, 526 (2014). DOI: 10.1186/1556-276X-9-526, https://doi.org/10.1186/1556-276X-9-526
  13. Viet Cuong Nguyen, "Solution-processed PEDOT: PSS/PVA polymer blend write-once-read-many memory devices", E3S Web of Conferences 496, 02001 (2024). DOI: 10.1051/e3sconf/202449602001, https://doi.org/10.1051/e3sconf/202449602001
  14. Ruolin Wang, Qiaofeng Li, Linrun Feng, Wei Hu, Wenjiang Liu, Xiaojun Guo, "Inkjet-Printed Multi-Bit Low-Voltage Fuse-Type Write-Once-Read-Many Memory Cell", IEEE Electron Device Letters 37, 862–865 (2016). DOI: 10.1109/LED.2016.2564988, https://doi.org/10.1109/LED.2016.2564988
  15. Luyao Zhu, Sujie Chen, Haoyu Zhou, Sai Peng, Qiuqi Zhang, Xiachang Zhang, Xiaojun Guo, "Silver Nanowire Mesh-Based Fuse Type Write-Once-Read-Many Memory", IEEE Electron Device Letters 39, 347–350 (2018). DOI: 10.1109/LED.2018.2794999, https://doi.org/10.1109/LED.2018.2794999