pith. sign in

arxiv: 2604.14680 · v1 · submitted 2026-04-16 · ❄️ cond-mat.mtrl-sci

Anomalous Platinum and Oxygen Transport during Electroforming of NbOx Memristors

Shimul Kanti Nath , Sanjoy Kumar Nandi , Xiao Sun , Sujan Kumar Das , Bin Gong , Nicholas J. Ekins-Daukes , Deepak Mishra , Mahesh P. Suryawanshi
show 4 more authors
William D. A. Rickard Songyan Yin Michael P. Nielsen Robert G. Elliman
This is my paper

Pith reviewed 2026-05-10 11:35 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords memristorelectroformingNbOxfilamentoxygen transportplatinum diffusionnegative differential resistancesecondary ion mass spectrometry
0
0 comments X

The pith

Electroforming NbOx memristors moves platinum and oxygen atoms along the same filamentary path.

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

The paper demonstrates that in Pt/NbOx/Nb2O5/Pt memristors, electroforming produces a filament where oxygen enriches from the oxide deep into the platinum electrode while platinum forms a rich region penetrating back into the oxide. This contradicts the usual assumption that only oxygen vacancies create the filament and that platinum electrodes remain inert. Measurements using time-of-flight secondary ion mass spectrometry map the distributions, and models connect the transport to localized heating and rapid thermal cycling during negative differential resistance operation. Understanding this metal transport is important because it affects how the filament forms, how stable the device is, and how it performs over time.

Core claim

Electroforming and subsequent operation of Pt/NbOx/Nb2O5/Pt devices can induce an unexpected and highly correlated redistribution of both oxygen and platinum. Time-of-flight secondary ion mass spectrometry reveals a filamentary pathway characterized by micrometer-scale oxygen enrichment extending from the Nb2O5 layer through NbOx and deep into the Pt top electrode, accompanied by the formation of a Pt-rich filament penetrating the oxide stack along the same filamentary path. Finite-element and lumped-element modelling show that current-controlled negative-differential-resistance operation produces localized Joule heating and high-frequency thermal cycling, which strongly enhances oxygen and

What carries the argument

Thermally assisted Pt diffusion along vacancy-rich pathways produced by Joule heating and high-frequency thermal cycling during current-controlled negative differential resistance.

If this is right

  • Filaments in these memristors contain platinum ions from the electrode in addition to oxygen vacancies.
  • Post-forming electrical dynamics play a critical role in setting filament chemistry and device reliability.
  • Negative differential resistance operation enhances oxygen migration through thermal effects.
  • Device stability depends on the thermal history during and after forming.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar anomalous metal transport may occur in other metal-oxide memristors with noble metal electrodes when thermal cycling is present.
  • Models of filament formation should include the possibility of electrode atom incorporation when Joule heating is significant.
  • Suppressing thermal cycling during electroforming could prevent platinum diffusion and lead to more predictable filaments.
  • This mechanism might be exploited to tune filament properties by selecting appropriate electrode materials.

Load-bearing premise

That the Pt-rich filament forms via thermally assisted diffusion of platinum atoms enabled by the vacancy-rich pathways and the high-frequency thermal cycling from NDR operation.

What would settle it

Absence of a Pt-rich filament in devices electroformed under conditions that avoid current-controlled NDR or that minimize Joule heating, such as voltage-controlled forming or with enhanced heat dissipation.

Figures

Figures reproduced from arXiv: 2604.14680 by Bin Gong, Deepak Mishra, Mahesh P. Suryawanshi, Michael P. Nielsen, Nicholas J. Ekins-Daukes, Robert G. Elliman, Sanjoy Kumar Nandi, Shimul Kanti Nath, Songyan Yin, Sujan Kumar Das, William D. A. Rickard, Xiao Sun.

Figure 1
Figure 1. Figure 1: (a) TEM image and EDX maps of a device cross-section. (b) Electroforming step of a 20 μm device with Pt/Nb/Nb₂O₅/Pt structure under current-controlled testing with a bidirectional current sweep from 0 to 800 μA, and (c) SEM image of the top-view of the active device area (indicated by the dashed line) after electroforming, identifying filament location as a bright circular spot (TE and BE represent top and… view at source ↗
Figure 2
Figure 2. Figure 2: ToF-SIMS images showing the distribution of different ions in a Pt/Nb(O)/Nb₂O₅/Pt device (maximum current applied through the device = 800 μA as shown in Fig. S1). The top panel shows 3D images of the selected device area with signals from Pt- , O- , Nb-, NbO- and NbO2 - . The Pt- and O- 3D maps observe a distinct filamentary area, as marked by the dashed ellipses. The middle and bottom panels of [PITH_FU… view at source ↗
Figure 3
Figure 3. Figure 3: (a) Schematic of the device structure used in the COMSOL model. (b) 3D maps of temperature distribution within the device area before (left) and after filament formation (right), (c-d) radial temperature distribution after forming. (e) Diffusion coefficient (D) of O and Pt as a function of temperature calculated from Arrhenius type diffusion equation. To quantify the role of Joule heating on the transport … view at source ↗
Figure 4
Figure 4. Figure 4: (a) Schematic of the measurement circuit for LTspice modeling, (b) current￾controlled I-V characteristic of a device with Cp=20 pF, and (c) associated average filament temperature. (d) Current-controlled I-V characteristic of a device with varying Cp= 0 to 100 pF, (e) filament temperature when Cp=30 pF, and (f) magnified view of filament current and temperature when Cp=30 pF, which clearly indicates the pe… view at source ↗
read the original abstract

Electroforming of metal-oxide-metal memristors is generally attributed to the creation of oxygen-vacancy filaments within the oxide, with noble metal electrodes such as Pt and Au remaining chemically inert. Here, we demonstrate that electroforming and subsequent operation of Pt/NbOx/Nb2O5/Pt devices can induce an unexpected and highly correlated redistribution of both oxygen and platinum. Time-of-flight secondary ion mass spectrometry reveals a filamentary pathway characterized by micrometer-scale oxygen enrichment extending from the Nb2O5 layer through NbOx and deep into the Pt top electrode. Surprisingly, this is accompanied by the formation of a Pt-rich filament penetrating the oxide stack along the same filamentary path. Finite-element and lumped-element modelling show that current-controlled negative-differential-resistance operation produces localized Joule heating and high-frequency thermal cycling, which strongly enhances oxygen migration and enables thermally assisted Pt diffusion along vacancy-rich pathways. These findings reveal a previously unrecognized metal-ion transport mechanism in NbOx memristors and highlight the critical role of post-forming electrical dynamics in determining filament chemistry, stability, and device reliability.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript reports that electroforming of Pt/NbOx/Nb2O5/Pt memristors produces correlated micrometer-scale redistribution of both oxygen and platinum along filamentary paths, as directly mapped by time-of-flight secondary ion mass spectrometry (ToF-SIMS). Oxygen enrichment extends from the Nb2O5 layer through NbOx into the Pt electrode, accompanied by a Pt-rich filament penetrating the oxide stack. The authors attribute this to current-controlled negative-differential-resistance operation generating localized Joule heating and high-frequency thermal cycling that enhances oxygen migration and enables thermally assisted Pt diffusion along vacancy-rich pathways, as demonstrated by finite-element and lumped-element thermal models.

Significance. If the central observations and mechanism hold, the work is significant because it identifies a previously unrecognized Pt-ion transport process in NbOx memristors, contradicting the standard assumption that noble-metal electrodes remain chemically inert. The direct ToF-SIMS chemical mapping of correlated filaments provides concrete experimental evidence, and the linkage to post-forming electrical dynamics via standard thermal-transport modeling has clear implications for filament stability, device reliability, and endurance. The experimental approach and use of established physical modeling constitute clear strengths.

major comments (1)
  1. [Finite-element and lumped-element modelling section] Finite-element and lumped-element modelling section: the models are invoked to establish that NDR-driven thermal cycling produces temperatures and timescales sufficient for micrometer-scale Pt diffusion, yet no material parameters (thermal conductivity, specific heat, or Pt diffusivity in NbOx/Nb2O5), no integrated diffusion-length calculation, and no comparison against literature Arrhenius data are supplied. This leaves the quantitative causal link between predicted heating and the observed Pt-rich filament unverified and is load-bearing for the proposed mechanism.
minor comments (2)
  1. [ToF-SIMS results] The ToF-SIMS composition profiles and filament dimensions are presented without reported error bars, standard deviations, or the number of devices examined, which would strengthen the statistical robustness of the correlation claim.
  2. [Figures] Figure captions and axis labels in the SIMS imaging panels would benefit from explicit indication of the probed ion species and spatial scale to improve immediate readability.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their positive assessment of the work and for the constructive major comment, which we address below. We will revise the manuscript to strengthen the modeling section as suggested.

read point-by-point responses

  1. Referee: Finite-element and lumped-element modelling section: the models are invoked to establish that NDR-driven thermal cycling produces temperatures and timescales sufficient for micrometer-scale Pt diffusion, yet no material parameters (thermal conductivity, specific heat, or Pt diffusivity in NbOx/Nb2O5), no integrated diffusion-length calculation, and no comparison against literature Arrhenius data are supplied. This leaves the quantitative causal link between predicted heating and the observed Pt-rich filament unverified and is load-bearing for the proposed mechanism.

    Authors: We agree that the modeling section requires additional quantitative detail to fully substantiate the proposed mechanism. In the revised manuscript we will explicitly list the thermal conductivity, specific heat capacity, and Pt diffusivity values employed in both the finite-element and lumped-element models (sourced from established literature for NbOx, Nb2O5, and Pt), include the integrated diffusion-length calculations performed using the simulated temperature profiles and thermal-cycling timescales, and provide direct comparisons of these results against relevant literature Arrhenius parameters for Pt diffusion in oxide matrices. These additions will verify that the modeled conditions are sufficient to account for the observed micrometer-scale Pt transport. revision: yes

Circularity Check

0 steps flagged

No circularity: claims rest on direct TOF-SIMS observations plus standard thermal modeling

full rationale

The paper's central chain is experimental (TOF-SIMS mapping of correlated Pt and O filaments) followed by invocation of finite-element/lumped-element models to attribute the redistribution to Joule heating and thermal cycling during NDR. No equations, fitted parameters, or self-citations are presented such that any 'prediction' or uniqueness claim reduces by construction to the input data or prior author work. The modeling is described as explanatory rather than predictive in a fitted sense, and the manuscript does not rename known results or smuggle ansatzes via citation. This is the normal non-circular case for an observation-driven materials paper.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on direct experimental imaging and standard assumptions from device physics; no new entities are postulated and no free parameters are explicitly fitted in the abstract.

axioms (2)
  • domain assumption Noble metal electrodes such as Pt remain chemically inert during electroforming
    Standard assumption in memristor literature that the findings directly challenge.
  • domain assumption Current-controlled negative-differential-resistance operation produces localized Joule heating and high-frequency thermal cycling
    Invoked to explain enhanced migration of both oxygen and platinum.

pith-pipeline@v0.9.0 · 5550 in / 1350 out tokens · 40007 ms · 2026-05-10T11:35:51.498387+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

11 extracted references · 11 canonical work pages

  1. [1]

    When subjected to electrical stress, MOM devices typically exhibit non-volatile, volatile, or combined resistive switching responses

    Introduction Memristive switching in two -terminal metal- oxide-metal (MOM) devices has attracted extensive research interest due to its promising applications in non- volatile memory technologies and neuromorphic computing 1-5. When subjected to electrical stress, MOM devices typically exhibit non-volatile, volatile, or combined resistive switching respo...

    work page

  2. [2]

    Experimental Studies were conducted on 20 μm × 20 μm MOM cross- point device s comprising Pt/ Nb2O5Nb(O)/Pt structures grown on thermally oxidi zed Si substrates . The bottom metal electrodes consisted of a 25 nm Cr adhesion layer and a 40 nm Pt layer that were e-beam evaporated and patterned using a standard photolithographic lift-off process. A thin Nb2...

    work page

  3. [3]

    1a shows a TEM image and EDX maps of a representative device cross -section, confirming the thickness and composition of the layered structure

    Results and Discussions 6 3.1 Electroforming Behaviour and Localization of Conduction Fig. 1a shows a TEM image and EDX maps of a representative device cross -section, confirming the thickness and composition of the layered structure. The Pt/ NbOx/Nb2O5/Pt cross-point devices were initially in a high -resistance state and required electroforming to activa...

    work page arXiv 2020

  4. [4]

    Electroforming Figure S1: Bidirectional current sweep showing the electroforming step followed by an immediate NDR response in a 20 μm cross-point device that was further analyzed using ToF- SIMS (Fig. 3). 23 Figure S2: (a) Bidirectional voltage sweep showing a representative electroforming step (black line) followed by an immediate threshold switching re...

    work page

  5. [5]

    Figure S3: (a) Electroforming step of a 20 μm device with Pt/Nb/Nb₂O₅/Pt structure under current-controlled testing with a bidirectional current sweep from 0 to 800μA

    Structural and compositional analyses of the filamentary area in Pt/Nb/Nb₂O₅/Pt structures revealed by SEM, TEM, and EDX. Figure S3: (a) Electroforming step of a 20 μm device with Pt/Nb/Nb₂O₅/Pt structure under current-controlled testing with a bidirectional current sweep from 0 to 800μA. (b) SEM image of the top-view of active device area (indicated by t...

    work page

  6. [6]

    Figure S4: (a) ToF-SIMS images showing oxygen (O-) distribution in a Pt/Nb/Nb₂O₅/Pt device (maximum current applied through the device = 800 μA)

    ToF-SIMS analysis of post -electroforming devices with sub -stoichiometric NbOx/Nb2O5 films. Figure S4: (a) ToF-SIMS images showing oxygen (O-) distribution in a Pt/Nb/Nb₂O₅/Pt device (maximum current applied through the device = 800 μA). The selected device area was cropped and rotated to visualize the extent of the filamentary area

    work page

  7. [7]

    The finite element simulations employed the electric currents and heat transfer modules in COMSOL to simultaneously solve the coupled electrical and thermal transport equations

    COMSOL Parameters Figure S5: (a) Schematic of the device structure (b) Simulated I-V curve (c-d) temperature in different layers as a function of vertical and radial distance, respectively. The finite element simulations employed the electric currents and heat transfer modules in COMSOL to simultaneously solve the coupled electrical and thermal transport ...

    work page

  8. [8]

    LTSPICE Parameters In the LTSpice model, we consider ed a threshold- switching memristor in which electrical conduction is assumed to follow a Poole-Frenkel transport mechanism. Under this assumption, the resistance of the memristive region can be expressed as37: 𝑅𝑅𝑚𝑚= 𝑅𝑅0exp ⎝ ⎛ 𝐸𝐸𝑎𝑎− 𝑞𝑞� 𝑞𝑞 𝐸𝐸 𝜋𝜋𝜀𝜀0𝜀𝜀𝑟𝑟 𝑘𝑘𝐵𝐵𝑇𝑇 ⎠ ⎞ where 𝑘𝑘𝐵𝐵 is the Boltzmann constant, 𝐸...

    work page

  9. [9]

    ToF-SIMS analysis of post-electroforming devices with sub-stoichiometric NbOx films. To explore the nature of ion migration in sub- stoichiometric NbO x films, additional measurements were carried out on devices with Pt/Nb (5 nm)/NbO x/Pt structures, which predominantly exhibited threshold switching or NDR behaviour after electroforming. In all cases, the...

    work page

  10. [10]

    Figure S7: Electroforming and TEM imaging of the filamentary area in devices with (a) 10 μm Pt/Nb(5 nm)/NbOx/Pt and (b) 10 μm Pt/Nb(10 nm)/NbOx/Pt structures

    Structural and compositional analyses of the filamentary area under extreme biasing conditions and repeated cycling. Figure S7: Electroforming and TEM imaging of the filamentary area in devices with (a) 10 μm Pt/Nb(5 nm)/NbOx/Pt and (b) 10 μm Pt/Nb(10 nm)/NbOx/Pt structures. Inset (b -i) shows an enlarged view of the selected device area showing contrast ...

    work page

  11. [11]

    (c) Cycle- to-cycle variability of threshold and hold voltages in a Pt/Nb/Nb2O5/Pt device with a 30 nm Nb electrode

    Temperature-dependent I-V characteristics of a post-forming device Figure S8: (a) Post-forming NDR in a device with Pt/Nb/NbO x/Pt structure as a function of stage temperature, inset shows a temperature map showing filamentary conduction path after electroforming (further in-situ thermal mapping of NbO x devices can be found in our earlier publications32,...

    work page doi:10.1002/aelm.201800954 2020