Skin-inspired in-sensor encoding of strain vector using tunable quantum geometry

Bin Cheng; Cong Xiao; Fanqiang Chen; Feng Miao; Jin Cao; Jingwen Shi; Lizheng Wang; Moyu Chen; Pengfei Wang; Shengyuan A. Yang

REVIEW 4 major objections 5 minor 70 references

Skin-inspired in-sensor encoding of strain vector using tunable quantum geometry

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that a single few-layer Td-WTe2 device can encode a strain vector's magnitude and direction by reading two independent nonlinear Hall signals governed by quantum geometry.

desk verdict Plausible proof-of-concept that two nonlinear Hall channels in WTe2 can encode strain vectors, but the quantitative strain calibration and the 'NJU' demo are weaker than the abstract implies. read the letter →

arxiv 2501.04215 v1 pith:FHRINI4A submitted 2025-01-08 cond-mat.mes-hall

Zenglin Liu , Jingwen Shi , Jin Cao , Zecheng Ma , Zaizheng Yang , Yanwei Cui , Lizheng Wang , Yudi Dai

show 10 more authors

Moyu Chen Pengfei Wang Yongqin Xie Fanqiang Chen Youguo Shi Cong Xiao Shengyuan A. Yang Bin Cheng Shi-Jun Liang Feng Miao

This is my paper · ORCID

classification cond-mat.mes-hall

keywords quantumgeometrynonlinearHalleffectstrainengineeringTd-WTe2BerrycurvatureBerry-connectionpolarizabilitytensorin-sensorencodingflexibleelectronics

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

Human skin encodes both the size and the direction of a deformation in a single touch; this paper claims that a single electronic device can do the same. The device is a few-layer flake of the topological semimetal Td-WTe2 on a flexible polyimide strip. When the strip is bent, strain changes the momentum-space distribution of two quantum-geometric quantities—the Berry curvature and the Berry-connection polarizability tensor—and these changes are read out as the second-order and third-order nonlinear Hall voltages. Because the two harmonic responses move independently with strain angle and magnitude, one device can simultaneously sense and encode the full strain vector, which ordinary piezoresistive sensors cannot do. A neural network trained on the two harmonic voltages reconstructs the strain pattern of an embossed character, demonstrating in-sensor encoding.

What carries the argument

The machinery is the strain-tunable quantum geometry of the electronic wave functions, probed by nonlinear Hall conductivities. The second-order signal is governed by the Berry-curvature dipole $\partial_x\Omega_z$; the third-order signal is governed by the Berry-connection polarizability tensor $G_{ab}(\mathbf{k})=2\sum_{n\neq0}\mathrm{Re}[(\mathcal{A}_a)_{0n}(\mathcal{A}_b)_{n0}]/(\varepsilon_0-\varepsilon_n)$, through terms like $\partial_x^2 G_{xy}$. Strain enters a tilted-Dirac model by deforming reciprocal space as $k'=(1-\mathcal{E}^T)k$ and by adding a pseudogauge field from electron-phonon coupling, which redistributes $\Omega_z$ and $G$ while preserving their independence; the measured harmonic voltages then serve as two independent readout channels for the strain vector.

What would settle it

Measure the strain directly on the WTe2 flake during bending—for example by tracking a Raman-active mode of WTe2 or by performing micro-X-ray diffraction over the channel—and compare it with Eq. (10). If the flake's actual strain differs from the bending-height estimate, or if the extracted scaling-law intercepts $\eta^{2\omega}$ and $\eta^{3\omega}$ do not follow the strain-angle pattern predicted by the strained tilted-Dirac model, the central claim that quantum geometry encodes the strain vector would be disproved.

Watch

Extended reading notes

Core claim

The central claim is that strain vectors can be encoded inside the sensor itself: in Td-WTe2, the strain dependence of the second-order nonlinear Hall conductivity $\chi^{2\omega}_{yxx}\propto -\int_k f_0\,\partial_x\Omega_z$ and that of the third-order channel $\chi^{3\omega}_{yxxx}\propto -\int_k f_0\,\partial_x^2 G_{xy}+\frac{1}{2}\int_k \partial_\varepsilon^2 f_0\, v_x v_y G_{xx}$ are independent, so the pair $(V^{2\omega}_{xy},V^{3\omega}_{xy})$ carries both the magnitude and direction of the applied strain. The authors establish that these strain responses are robust across devices and that a scaling-law analysis $E^{2\omega}_{xy}/E_{xx}^2=\xi^{2\omega}\sigma^2+\eta^{2\omega}$, $E^{3\omega}_{xy}/E_{xx}^3=\xi^{3\omega}\sigma^2+\eta^{3\omega}$ shows the strain mainly changes the intercepts $\eta$ associated with the intrinsic Berry-curvature and BPT distributions rather than the scattering slopes $\xi$. On this basis a single device, read at the second and third harmonics, is trained through an artificial neural network to output strain magnitude and angle, and the outputs reproduce the strain map of an embossed three-letter pattern.

Load-bearing premise

The load-bearing assumption is that the strain computed from the bending height of the polyimide substrate is exactly the uniform strain felt by the WTe2 flake; if the flake slips, clamps unevenly, or sees inhomogeneous strain, the reported strain magnitudes and the mapping learned by the ANN would not describe the device.

Editorial extensions

If this is right

A single WTe2 device can replace arrays of unidirectional strain gauges for tasks that need both angle and magnitude, because the two harmonic channels respond independently.
The encoding is intrinsic to quantum geometry, not a piezoresistive artifact: the scaling-law intercepts, tied to Berry curvature and BPT, carry most of the strain response.
Because the response survives over a temperature range up to at least 260 K, the sensing scheme is not restricted to cryogenic operation.
The independently varying second- and third-harmonic channels provide a natural two-dimensional code for a strain vector, which the trained ANN decodes with accuracy above 90%.
The nonlinear Hall channels respond to strain in all tested directions (0°, 45°, 90°), whereas the fundamental longitudinal voltage is sensitive mainly near 90°, so the harmonic readout extends directional coverage.

Reading between the lines

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

Extension: the same two-channel readout could be applied to other quantum materials with large Berry curvature and BPT, but the paper only demonstrates Td-WTe2.
Extension: the ANN was trained on interpolated data from 39 measured signals; a deployed sensor would need to handle hysteresis, drift, and temperature variation, which the paper does not quantify outside the measured range.
Extension: a direct local strain measurement on the flake, rather than the substrate bending formula, would test whether the reported strain magnitudes are quantitatively correct; this is an implicit next step.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Short version: the genuinely new thing here is using two independent nonlinear Hall channels in a single WTe2 device to encode both magnitude and direction of strain. That is a useful idea and the data are consistent with it. The strain-tuning of the second-order NLH in WTe2 was already reported by Ye et al. (ref 46) — the paper cites it but only for skew scattering — so the novelty is incremental, not the 'first time' claimed in the ToC.

What the paper does well: clean linear harmonic responses, repeatable across several devices, and the two harmonics show visibly different strain-angle dependences — enough to justify the claim of independent channels. The temperature dependence and the scaling-law analysis are standard and handled adequately; they do support the intrinsic (BC/BPT) contribution being the one that varies with strain. The model calculation is simple but useful for motivating the effect. I also note the paper itself includes a limitation: the NJU demonstration is called 'simulation results' in Section 4 yet the abstract presents it as demonstrated; that needs to be aligned.

Soft spots: first, the strain magnitude is computed from the beam-bending formula Eq. 10, with no independent measurement of the local strain in the flake. Slip or shear lag would scale all the reported strain values and could mix the angle labels. That does not kill the proof-of-concept, but it does mean the quoted 0.2%/0.4%/0.51% values and the ANN training labels are only as good as the transfer assumption. Second, the main strain-response figures (2h,i; 3a,b) lack error bars; the reader has to trust the reproducibility claims from Fig S6 alone. Third, the scanning of 'NJU' is not a physical scan; it is a simulation using measured harmonic signals. That is fine as an extrapolation, but should be labeled as such in the title/abstract.

The circularity concern is mild: the eta coefficients are extracted from scaling-law fits and then interpreted as intrinsic BC/BPT contributions. That is how the field usually does it, and the fitted values are consistent with prior reports, so I do not see it as a fatal flaw.

Who is it for: people working on nonlinear Hall effects, strain engineering of 2D materials, and flexible quantum-material sensors. It deserves a serious referee: send it out with a request for error bars, a local strain check or explicit caveat, and a clearer statement of prior work.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes and experimentally demonstrates a skin-inspired in-sensor encoding scheme in which a single few-layer Td-WTe2 device uses its second-order and third-order nonlinear Hall responses to encode the magnitude and direction of an applied strain vector. The authors measure linear harmonic responses under various strain magnitudes and directions, extract scaling-law coefficients to argue that the strain response originates from Berry curvature and Berry-connection polarizability, and train an artificial neural network to decode the strain vector from the two harmonic signals. They also demonstrate the encoding of the embossed characters 'NJU' as a proof of concept.

Significance. If the strain calibration and error analysis are strengthened, this would be a notable advance: it uses intrinsic quantum geometric responses in a topological semimetal for multi-parameter tactile sensing, offering a path beyond conventional piezoresistive strain sensors. The experiment is direct and reproducible across several devices (Fig. S6), and the proof-of-concept ANN decoding, despite being trained on interpolated data, demonstrates the encoding concept. The machine-checked data and the availability of multiple device results are strengths.

major comments (4)

[Section 4, Eq. (10)] The strain magnitude applied to the WTe2 flake is derived from a global beam-bending formula for the polyimide substrate, assuming perfect, slip-free, uniform strain transfer across the van der Waals interface. No local strain measurement on the device (e.g., Raman or X-ray) is reported; Fig. S4g only demonstrates repeatability of the four-probe resistance under cycling, which does not calibrate the absolute strain. If shear lag, clamping, or electrode pinning causes the local strain to differ from the nominal value, then all reported strain magnitudes (0.2%, 0.4%, 0.51%) and the ANN target labels are systematically wrong. The authors should provide a local strain calibration or explicitly quantify the uncertainty in the strain transfer.
[Fig. 2h,i and Fig. 3a-d] These central figures present the extracted second- and third-order NLH signals as functions of strain magnitude and temperature without any error bars, although the raw voltage signals in Fig. S9 do include error bars. The claim that the two channels respond 'independently' and 'robustly' to strain vectors is based on visual inspection of these curves. Please provide error bars on the extracted intensive quantities (propagated from at least forward/backward scans or repeated devices) and, if possible, a quantitative test of independence (e.g., comparing the strain-response slopes).
[Section 2.3] The scaling laws E_xy^{2ω}/E_xx^2 = ξ_{2ω} σ^2 + η_{2ω} and E_xy^{3ω}/E_xx^3 = ξ_{3ω} σ^2 + η_{3ω} are fitted, and the coefficients η_{2ω} and η_{3ω} are identified as 'intrinsic' and interpreted as evidence that the strain responses 'primarily stem from variations in the distribution of BC and BPT'. However, this attribution is not directly tested: the same scaling-law fit could be consistent with strain-induced changes in the Fermi surface or band structure that affect other intrinsic contributions. The model in Fig. S1 uses generic parameters (ω=0.05 eV, v=0.1 eV·Å, Δ=0.1 eV, β/a=1.1) and yields only qualitative distributions, not a quantitative comparison with the experimental η values. Please provide a concrete, falsifiable prediction (e.g., a predicted ratio η_{2ω}(θ)/η_{2ω}(0) as a function of strain angle, with parameter ranges) and compare it with the measured values.
[Section 4, 'Construction of ANN'] The ANN is trained on 4290 samples generated by cubic interpolation from 39 measured nonlinear Hall voltage signals, and the validation set is 21 experimental data. This raises the risk that the reported decoding accuracy is inflated by the interpolation. The five strain combinations used in the 'NJU' demonstration are exactly among the training values, so the demonstration does not test generalization to unseen strain states. Please report the ANN error on held-out experimental data without interpolation, and clarify the role of interpolation in the reported accuracy.

minor comments (5)

[Main text and Eq. (7)] The tensor notation is inconsistent: the main text defines the measured response as j_y^{2ω}=χ_yxx^{2ω} E_x E_x, but Eq. (7) gives χ_xyy^(2) = -(τ/2)∫ f0 ∂_y Ω_z, which is a different tensor component. Please unify the notation and explain how the symmetry argument for χ_xyy applies to the measured χ_yxx.
[Section 4, 'Construction of ANN'] In the ANN description, the sigmoid activation function for the output layer would constrain outputs to [0,1]; since the output includes an angle θ (which should be in [0°,360°) or [0,π)), please specify how the output layer encodes the angle.
[Supporting Information, Table 1] The Supporting Information table with the scaling-law coefficients is labeled 'Table 1' but is referred to as Table S1 in the main text; please correct the cross-reference.
[Section 4, '2D tilted Dirac cone model'] The model parameters in Eq. (3) are presented without discussion of their relevance to Td-WTe2; please add a sentence justifying the parameter choices or note that the model is illustrative only.
[Section 2.3] The sentence 'the strain responses of the second-order and third-order NLH signals primarily stem from the variations in the distribution of BC and BPT modulated by strain' appears before the scaling-law analysis is fully described; consider moving it after the discussion of Table S1.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the strain response of nonlinear Hall signals is directly measured, and the quantum-geometry interpretation rests on an externally established scaling-law decomposition rather than on the paper's own definitions.

full rationale

The central experimental claim is direct transport data: the second- and third-order nonlinear Hall voltages Vxy2ω and Vxy3ω are measured under bending-imposed strain (Fig. 2b–i), and their strain dependence is not derived from the toy model or from the ANN. The 2D tilted Dirac model (Fig. 1) is explicitly illustrative, with stated parameters and no fitting to the transport data; it therefore does not inject the experimental conclusion. The mechanism attribution uses the scaling law E_xy^nω/E_xx^n = ξσ^2 + η taken from refs [44,45,54], with η interpreted as the intrinsic Berry-curvature/Berry-connection-polarizability contribution. Although some of these references include co-authors, the scaling law is an externally established result and is not constructed from the present data; the temperature-dependent extraction of η is a standard decomposition, not a circular definition. The strain response of η is then compared with, rather than used to generate, the direct strain response in Fig. 2h,i, making it a consistency check. The ANN is a supervised decoder trained on measured voltage signals labeled by the geometric bending formula (Eq. 10); it demonstrates the encoding application but does not feed back into the quantum-geometry derivation. Eq. 10's assumption of uniform, slip-free strain transfer to the WTe2 flake is a calibration/validity concern, not a circular step. No fitted parameter is renamed as a prediction of the same quantity, and no claimed result reduces by construction to its own input.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The experimental strain values depend on the assumption of perfect strain transfer from the substrate. The mechanism interpretation relies on the standard scaling-law decomposition. The illustrative Dirac model uses hand-picked parameters and is not quantitatively connected to the experiment. No invented physical entities are introduced.

free parameters (4)

Dirac model tilt parameter omega = 0.05 eV
Hand-chosen parameter in Eq 3 of Sec 4; used for the qualitative Fig 1 calculation, not fitted to data.
Fermi velocity upsilon = 0.1 eV Angstrom
Hand-chosen model parameter in Eq 3; not experimentally constrained.
Mass gap Delta = 0.1 eV
Hand-chosen model parameter in Eq 3; not experimentally constrained.
Phonon coupling beta/a = 1.1
Hand-chosen parameter in the strain gauge field A in Sec 4; no independent experimental determination is provided.

assumptions (4)

domain assumption Scaling law: E_xy^{2omega}/E_xx^2 = xi_{2omega} sigma^2 + eta_{2omega} and E_xy^{3omega}/E_xx^3 = xi_{3omega} sigma^2 + eta_{3omega} holds over the measured temperature range.
Invoked in Sec 2.3 (Fig 3c,d) with the claim that eta terms represent intrinsic BC/BPT contributions and xi terms represent skew scattering. This decomposition is standard (refs 44,45,54) but is assumed, not derived.
domain assumption The strain computed from the PI substrate bending geometry (Eq 10) equals the actual strain experienced by the WTe2 flake.
Sec 4 'Determination of strain vector' derives epsilon from the substrate height h and assumes perfect, uniform strain transfer into the WTe2; no local strain measurement on the flake is provided.
domain assumption The nonlinear Hall conductivities in Eqs 7 and 8 relate the measured harmonic voltages to the Berry curvature dipole and Berry-connection polarizability within the relaxation-time approximation.
Sec 4 states these relations citing refs 43-45,51,52. The measured signal is assumed to be dominated by these intrinsic terms plus the skew-scattering terms captured by xi.
domain assumption The model Hamiltonian Eq 3 with the chosen parameters captures the essential strain response of WTe2, and the time-reversal partner of the Dirac cone contributes equally to the time-reversal-invariant nonlinear Hall conductivity.
Sec 4 uses a 2D tilted Dirac cone with hand-picked parameters and assumes the TR partner contributes equally; the model is not quantitatively benchmarked against the experiment.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Skin-inspired in-sensor encoding of strain vector using tunable quantum geometry." pith.science (2026). https://pith.science/paper/FHRINI4A

@misc{pith2026250104215,
  author       = {Pith},
  title        = {Pith review of: Skin-inspired in-sensor encoding of strain vector using tunable quantum geometry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FHRINI4A}},
  note         = {Machine review of arXiv:2501.04215}
}

read the original abstract

Human skin provides crucial tactile feedback, allowing us to skillfully perceive various objects by sensing and encoding complex deformations through multiple parameters in each tactile receptor. However, replicating this high-dimensional tactile perception with conventional materials' electronic properties remains a daunting challenge. Here, we present a skin-inspired method to encode strain vectors directly within a sensor. This is achieved by leveraging the strain-tunable quantum properties of electronic bands in the van der Waals topological semimetal Td -WTe2. We observe robust and independent responses from the second-order and third-order nonlinear Hall signals in Td -WTe2 when subjected to variations in both the magnitude and direction of strain. Through rigorous temperature-dependent measurements and scaling law analysis, we establish that these strain responses primarily stem from quantum geometry-related phenomena, including the Berry curvature and Berry-connection polarizability tensor. Furthermore, our study demonstrates that the strain-dependent nonlinear Hall signals can efficiently encode high-dimensional strain information using a single device. This capability enables accurate and comprehensive sensing of complex strain patterns in the embossed character "NJU". Our findings highlight the promising application of topological quantum materials in advancing next-generation, bio-inspired flexible electronics.

Figures

Figures reproduced from arXiv: 2501.04215 by the authors.

**Figure 1.** Encoding of strain vector using strain-tunable quantum geometry. a) Conceptual illustration of biological encoding scheme. Tactile receptors generate encoded bioelectrical signals conveying information about the direction and magnitude of skin deformation, [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 70 canonical work pages

[1]

K. O. Johnson, Curr. Opin. Neurobiol. 2001, 11, 455

work page 2001
[2]

R. S. Johansson, J. R. Flanagan, Nat. Rev. Neurosci. 2009, 10, 345

work page 2009
[3]

J. A. Pruszynski, R. S. Johansson, Nat. Neurosci. 2014, 17, 1404

work page 2014
[4]

J. D. Lieber, S. J. Bensmaia, Proc. Natl. Acad. Sci. U.S.A. 2019, 116, 3268

work page 2019
[5]

K. H. Long, J. D. Lieber, S. J. Bensmaia, Nat. Commun. 2022, 13, 1311

work page 2022
[6]

V. E. Abraira, D. D. Ginty, Neuron. 2013, 79, 618

work page 2013
[7]

C. Wang, D. Hwang, Z. Yu, K. Takei, J. Park, T. Chen, B. Ma, A. Javey, Nat. Mater. 2013, 12, 899

work page 2013
[8]

Subramanian, P

S. Subramanian, P. Kellnhofer, Y. Li, J. -Y. Zhu, A. Torralba, W. Matusik, Nature. 2019, 569, 698

work page 2019

Show all 70 references

[9]

S. N. Flesher, J. E. Downey, J. M. Weiss, C. L. Hughes, A. J. Herrera, E. C. Tyler - Kabara, M. L. Boninger, J. L. Collinger, R. A. Gaunt, Science. 2021, 372, 831

2021
[10]

H. Cho, I. Lee, J. Jang, J. -H. Kim, H. Lee, S. Park, G. Wang, Nat. Electron. 2023, 6, 619

2023
[11]

X. Yu, Z. Xie, Y. Yu, J. Lee, A. Vazquez-Guardado, H. Luan, J. Ruban, X. Ning, A. Akhtar, D. Li, B. Ji, Y. Liu, R. Sun, J. Cao, Q. Huo, Y. Zhong, C. Lee, S. Kim, P. Gutruf, C. Zhang, Y. Xue, Q. Guo, A. Chempakasseril, P. Tian, W. Lu, J. Jeong, Y. Yu, J. Cornman, C. Tan, B. Kim...

2019
[12]

M. Zhu, Z. Sun, Z. Zhang, Q. Shi, T. He, H. Liu, T. Chen, C. Lee, Sci. Adv. 2020, 6, eaaz8693

2020
[13]

H. U. Chung, B. H. Kim, J. Y. Lee, J. Lee, Z. Xie, E. M. Ibler, K. Lee, A. Banks, J. Y. Jeong, J. Kim, C. Ogle, D. Grande, Y. Yu, H. Jang, P. Assem, D. Ryu, J. W. Kwak, M. Namkoong, J. B. Park, Y. Lee, D . H. Kim, A. Ryu, J. Jeong, K. You, B. Ji, Z. Liu, Q. Huo, X. Feng, Y. De...

2019
[14]

L. E. Osborn, A. Dragomir, J. L. Betthauser, C. L. Hunt, H. H. Nguyen, R. R. Kaliki, N. V. Thakor, Sci. Robot. 2018, 3, eaat3818. 18

2018
[15]

Y. Wu, Y. Liu, Y. Zhou, Q. Man, C. Hu, W. Asghar, F. Li, Z. Yu, J. Shang, G. Liu, M. Liao, R. -W. Li, Sci. Robot. 2018, 3, eaat0429

2018
[16]

F. Liu, S. Deswal, A. Christou, M. Shojaei Baghini, R. Chirila, D. Shakthivel, M. Chakraborty, R. Dahiya, Sci. Robot. 2022, 7, eabl7286

2022
[17]

Sundaram, Science

S. Sundaram, Science. 2020, 370, 768

2020
[18]

C. Pang, G. -Y. Lee, T. Kim, S. M. Kim, H. N. Kim, S. -H. Ahn, K. -Y. Suh, Nat. Mater. 2012, 11, 795

2012
[19]

D. Kang, P. V. Pikhitsa, Y. W. Choi, C. Lee, S. S. Shin, L. Piao, B. Park, K. -Y. Suh, T. Kim, M. Choi, Nature. 2014, 516, 222

2014
[20]

S. Ota, A. Akira, D. Chiba, Nat. Electron. 2018, 1, 124

2018
[21]

O. A. Araromi, M. A. Graule, K. L. Dorsey, S. Castellanos, J. R. Foster, W. -H. Hsu, A. E. Passy, J. J. Vlassak, J. C. Weaver, C. J. Walsh, R. J. Wood, Nature. 2020, 587, 219

2020
[22]

Q. Yu, R. Ge, J. Wen, T. Du, J. Zhai, S. Liu, L. Wang, Y. Qin, Nat. Commun. 2022, 13, 778

2022
[23]

W. J. Yan, H. -R. Fuh, Y. Lv, K. -Q. Chen, T. -Y. Tsai, Y. -R. Wu, T. -H. Shieh, K. -M. Hung, J. Li, D. Zhang, C. Ó. Coileá in, S. K. Arora, Z. Wang, Z. Jiang, C. -R. Chang, H. -C. Wu, Nat. Commun. 2021, 12, 2018

2021
[24]

Y. Kim, A. Chortos, W. Xu, Y. Liu, J. Y. Oh, D. Son, J. Kang, A. M. Foudeh, C. Zhu, Y. Lee, S. Niu, J. Liu, R. Pfattner, Z. Bao, T.-W. Lee, Science. 2018, 360, 998

2018
[25]

J. Ge, X. Wang, M. Drack, O. Volkov, M. Liang, G. S. C. Bermú dez, R. Illing, C. Wang, S. Zhou, J. Fassbender, M. Kaltenbrunner, D. Makarov, Nat. Commun. 2019, 10, 4405

2019
[26]

S. Chun, J. -S. Kim, Y. Yoo, Y. Choi, S. J. Jung, D. Jang, G. Lee, K. -I. Song, K. S. Nam, I. Youn, D. Son, C. Pang, Y. Jeong, H. Jung, Y. -J. Kim, B. -D. Choi, J. Kim, S. -P. Kim, W. Park, S. Park, Nat. Electron. 2021, 4, 429

2021
[27]

F. Li, R. Wang, C. Song, M. Zhao, H. Ren, S. Wang, K. Liang, D. Li, X. Ma, B. Zhu, H. Wang, Y. Hao, ACS Nano. 2021, 15, 16422

2021
[28]

K. Yao, J. Zhou, Q. Huang, M. Wu, C. K. Yiu, J. Li, X. Huang, D. Li, J. Su, S. Hou, Y. Liu, Y. Huang, Z. Tian, J. Li, H. Li, R. Shi, B. Zhang, J. Zhu, T. H. Wong, H. Jia, Z. Gao, Y. Gao, Y. Zhou, W. Park, E. Song, M. Han, H. Zhang, J. Yu, L. Wang, W. J. Li, X. Yu, Nat. Mach. I...

2022
[29]

F. Liu, S. Deswal, A. Christou, Y. Sandamirskaya, M. Kaboli, R. Dahiya, Sci. Robot. 2022, 7, eabl7344

2022
[30]

I. You, D. G. Mackanic, N. Matsuhisa, J. Kang, J. Kwon, L. Beker, J. Mun, W. Suh, T. Y. Kim, J. B. -H. Tok, Z. Bao, U. Jeong, Science. 2020, 370, 961. 19

2020
[31]

Huang, J

Y. Huang, J. Zhou, P. Ke, X. Guo, C. K. Yiu, K. Yao, S. Cai, D. Li, Y. Zhou, J. Li, T. H. Wong, Y. Liu, L. Li, Y. Gao, X. Huang, H. Li, J. Li, B. Zhang, Z. Chen, H. Zheng, X. Yang, H. Gao, Z. Zhao, X. Guo, E. Song, H. Wu, Z. Wang, Z. Xie, K. Zhu, X. Yu, Nat. Electron. 2023, 6, 1020

2023
[32]

F. Miao, S. -J. Liang, and B, Cheng, NPJ Quantum Mater. 2021, 6, 59

2021
[33]

Zhang, L

Z. Zhang, L. Li, J. Horng, N. Z. Wang, F. Yang, Y. Yu, Y. Zhang, G. Chen, K. Watanabe, T. Taniguchi, X. H. Chen, F. Wang, Y. Zhang, Nano Lett. 2017, 17, 6097

2017
[34]

H. Kim, S. Z. Uddin, D. -H. Lien, M. Yeh, N. S. Azar, S. Balendhran, T. Kim, N. Gupta, Y. Rho, C. P. Grigoropoulos, K. B. Crozier, A. Javey, Nature. 2021, 596, 232

2021
[35]

Q. H. Wang, K. Kalantar-Zadeh, A. Kis, J. N. Coleman, M. S. Strano, Nat. Nanotechnol. 2012, 7, 699

2012
[36]

C. Ma, S. Yuan, P. Cheung, K. Watanabe, T. Taniguchi, F. Zhang, F. Xia, Nature. 2022, 604, 266

2022
[37]

Tokura, N

Y. Tokura, N. Nagaosa, Nat. Commun. 2018, 9, 3740

2018
[38]

Gianfrate, O

A. Gianfrate, O. Bleu, L. Dominici, V. Ardizzone, M. D. Giorgi, D. Ballarini, G. Lerario, K. W. West, L. N. Pfeiffer, D. D. Solnyshkov, D. Sanvitto, G. Malpuech, Nature. 2020, 578, 381

2020
[39]

H. Tian, X. Gao, Y. Zhang, S. Che, T. Xu, P. Cheung, K. Watanabe, T. Taniguchi, M. Randeria, F. Zhang, C. N. Lau, M. W. Bockrath, Nature. 2023, 614, 440

2023
[40]

N. Wang, D. Kaplan, Z. Zhang, T. Holder, N. Cao, A. Wang, X. Zhou, F. Zhou, Z. Jiang, C. Zhang, S. Ru, H. Cai, K. Watanabe, T. Taniguchi, B. Yan, W. Gao, Nature. 2023, 621, 487

2023
[41]

A. Gao, Y. -F. Liu, J. -X. Qiu, B. Ghosh, T. V. Trevisan, Y. Onishi, C. Hu, T. Qian, H. -J. Tien, S. -W. Chen, M. Huang, D. Bé rubé , H. Li, C. Tzschaschel, T. Dinh, Z. Sun, S. -C. Ho, S. -W. Lien, B. Singh, K. Watanabe, T. Taniguchi, D. C. Bell, H. Lin, T. -R. Chang, C. R. Du...

2023
[42]

J. Xiao, Y. Wang, H. Wang, C. D. Pemmaraju, S. Wang, P. Muscher, E. J. Sie, C. M. Nyby, T. P. Devereaux, X. Qian, X. Zhang, A. M. Lindenberg, Nat. Phys. 2020, 16, 1028

2020
[43]

Q. Ma, S. -Y. Xu, H. Shen, D. MacNeill, V. Fatemi, T. R. Chang, A. M. M . Valdivia, S. Wu, Z. Du, C. -H. Hsu, S. Fang, Q. D. Gibson, K. Watanabe, T. Taniguchi, R. J. Cava, E. Kaxiras, H. -Z. Lu, H. Lin, L. Fu, N. Gedik, P. Jarillo -Herrero, Nature. 2019, 565, 337. 20

2019
[44]

K. Kang, T. Li, E. Sohn, J. Shan, K. F. Mak, Nat. Mater. 2019, 18, 324

2019
[45]

S. Lai, H. Liu, Z. Zhang, J. Zhao, X. Feng, N. Wang, C. Tang, Y. Liu, K. S. Novoselov, S. A. Yang, W. -B. Gao, Nat. Nanotechnol. 2021, 16, 869

2021
[46]

X. -G. Ye, H. Liu, P. -F. Zhu, W. -Z. Xu, S. A. Yang, N. Shang, K. Liu, Z. -M. Liao, Phys. Rev. Lett. 2023, 130, 016301

2023
[47]

L. Min, H. Tan, Z. Xie, L. Miao, R. Zhang, S. H. Lee, V. Gopalan, C. -X. Liu, N. Alem, B. Yan, Z. Mao, Nat. Commun. 2023, 14, 364

2023
[48]

C. -P. Zhang, X. -J. Gao, Y. -M. Xie, H. C. Po, K. T. Law , Phys. Rev. B . 2023, 107, 115142

2023
[49]

Sodemann, L

I. Sodemann, L. Fu, Phys. Rev. Lett. 2015, 115, 216806

2015
[50]

C. Zeng, S. Nandy, S. Tewari, Phys. Rev. B. 2021, 103, 245119

2021
[51]

Y. Gao, S. A. Yang, Q. Niu, Phys. Rev. Lett. 2014, 112, 166601

2014
[52]

H. Liu, J. Zhao, Y. -X. Huang, X. Feng, C. Xiao, W. Wu, S. Lai, W. -B. Gao, S. A. Yang, Phys. Rev. B. 2022, 105, 045118

2022
[53]

A. A. Soluyanov, D. Gresch, Z. Wang, Q. Wu, M. Troyer, X. Dai, B. A. Bernevig, Nature. 2015, 27, 495

2015
[54]

C. Xiao, H. Zhou, Q. Niu, Phys. Rev. B. 2019, 100, 161403

2019
[55]

Z. Z. Du, C. M. Wang, S. Li, H. -Z. Lu, X. C. Xie, Nat. Commun. 2019, 10, 3047

2019
[56]

Suzuura, T

H. Suzuura, T. Ando, Phys. Rev. B. 2002, 65, 235412

2002
[57]

J. L. Mañ es, Phys. Rev. B. 2007, 76, 045430

2007
[58]

Y. Wang, C. Wang, S. -J. Liang, Z. Ma, K. Xu, X. Liu, L. Zhang, A. S. Admasu, S. -W. Cheong, L. Wang, M. Chen, Z. Liu, B. Cheng, W. Ji, F. Miao, Adv. Mater. 2020, 32, 2004533. 21 Table of contents ToC text: This manuscript for the first time proposes and experimentally demonst...

2020
[59]

The band geometric quantities are calculated from the two-dimensional tilted Dirac model, with the purple dashed line indicating the position of the Fermi surface

Calculation results of the two-dimensional tilted Dirac model Figure S1. The band geometric quantities are calculated from the two-dimensional tilted Dirac model, with the purple dashed line indicating the position of the Fermi surface. a) The distribution of the partial deriv...
[60]

a) Optical image of Device 01 after transport measurement

Sample thickness of Device 01 determined by atomic force microscope Figure S2. a) Optical image of Device 01 after transport measurement. b) Optical image of the exposed Device 01. The thick -layer h-BN, used for surface protection, was removed from the device surface using th...
[61]

a) The optical image of the sample for Device 01 before being transferred to the polyimide substrate

Crystal orientation of the sample determined by polarized Raman spectroscopy Figure S3. a) The optical image of the sample for Device 01 before being transferred to the polyimide substrate. b) Raman spectrum of the WTe2 sample. The areas of Raman peak 1 (P1) and Raman peak 4 (...
[62]

a) A structural diagram of the homemade strain setup

Homemade strain device used to apply strain vectors Figure S4. a) A structural diagram of the homemade strain setup. b,c,d) Optical photographs depicting the configuration of the homemade strain setup and the device when applying 0° (b), 45° (c), and 90° (d) strain vectors. e)...
[63]

Strain response of harmonic signals Figure S5. a,b,c) Second-harmonic transverse electric field 𝐸𝑥𝑦 2𝜔 = 𝑉𝑥𝑦 2𝜔 𝑊⁄ is plotted as a function of the square of longitudinal electric field 𝐸𝑥𝑥 = 𝑉𝑥𝑥 𝜔 𝐿⁄ when different magnitudes of strain vectors are applied at 0° (a), 45° (b) an...
[64]

a) The variation of the third-harmonic voltage signal from Device 02 is plotted as a function of the magnitude of the strain vector applied in different directions

Strain response measurements performed on other devices Figure S6. a) The variation of the third-harmonic voltage signal from Device 02 is plotted as a function of the magnitude of the strain vector applied in different directions. b) The variation of the third-harmonic voltag...
[65]

Temperature dependence of harmonic signals Figure S7. a,b,c) Second-harmonic transverse electric field 𝐸𝑥𝑦 2𝜔 = 𝑉𝑥𝑦 2𝜔 𝑊⁄ is plotted as a function of the square of longitudinal electric field 𝐸𝑥𝑥 = 𝑉𝑥𝑥 𝜔 𝐿⁄ at different temperatures when the 0.51% strain vectors are applied at...
[66]

a) Second-harmonic transverse voltage signal 𝑉𝑥𝑦 2𝜔 depends linearly on the square of AC current amplitude 𝐼 at 90 K when different magnitudes of strain vectors are applied at 0°

Supplementary measurement data for Device 01 Figure S8. a) Second-harmonic transverse voltage signal 𝑉𝑥𝑦 2𝜔 depends linearly on the square of AC current amplitude 𝐼 at 90 K when different magnitudes of strain vectors are applied at 0° . b) Third-harmonic transverse voltage sig...
[67]

a) The longitudinal voltage signal 𝑉𝑥𝑥 𝜔 is plotted as functions of the magnitude of strain vector applied in different directions

Harmonic voltage signals in response to strain vectors Figure S9. a) The longitudinal voltage signal 𝑉𝑥𝑥 𝜔 is plotted as functions of the magnitude of strain vector applied in different directions. b) Second-harmonic transverse voltage signal 𝑉𝑥𝑦 2𝜔 is plotted as functions of ...
[68]

a) The accuracy surpasses the 90% with an increment in the number of epochs

Extracting accuracy of high-dimensional information over the epoch Figure S10. a) The accuracy surpasses the 90% with an increment in the number of epochs. b) The mean square error decreases below 0.14 with an increment in the number of epochs. 33
[69]

Each character in "NJU" is composed of 25 (5× 5) pixels

The encoding of embossed characters Figure S11. Each character in "NJU" is composed of 25 (5× 5) pixels. a) Each pixel corresponds to a specific strain vector. b) Each pixel corresponds to the longitudinal voltage signal associated with the piezoresistive effect. c,d) Each pix...
[70]

Coefficients extracted from the scaling law Table 1. Quantity and symbol Strain vector Value with unit The slope derived from the fitted 𝐸𝑥𝑦 2𝜔 𝐸𝑥𝑥 2⁄ versus 𝜎2 curve: 𝜉2𝜔 None 1.951 × 10−8 𝜇𝑚3 ∙ 𝑉−1 ∙ 𝑆−2 (0.51%, 0° ) 9.772 × 10−9 𝜇𝑚3 ∙ 𝑉−1 ∙ 𝑆−2 (0.51%, 45° ) 6.214 × 10−8 𝜇𝑚...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

K. O. Johnson, Curr. Opin. Neurobiol. 2001, 11, 455

work page 2001

[2] [2]

R. S. Johansson, J. R. Flanagan, Nat. Rev. Neurosci. 2009, 10, 345

work page 2009

[3] [3]

J. A. Pruszynski, R. S. Johansson, Nat. Neurosci. 2014, 17, 1404

work page 2014

[4] [4]

J. D. Lieber, S. J. Bensmaia, Proc. Natl. Acad. Sci. U.S.A. 2019, 116, 3268

work page 2019

[5] [5]

K. H. Long, J. D. Lieber, S. J. Bensmaia, Nat. Commun. 2022, 13, 1311

work page 2022

[6] [6]

V. E. Abraira, D. D. Ginty, Neuron. 2013, 79, 618

work page 2013

[7] [7]

C. Wang, D. Hwang, Z. Yu, K. Takei, J. Park, T. Chen, B. Ma, A. Javey, Nat. Mater. 2013, 12, 899

work page 2013

[8] [8]

Subramanian, P

S. Subramanian, P. Kellnhofer, Y. Li, J. -Y. Zhu, A. Torralba, W. Matusik, Nature. 2019, 569, 698

work page 2019