Bulk single crystals of the layered chalcophosphate SnP2_2S6_6 exhibit an exceptionally large non-resonant second-harmonic coefficient, d33d_{33} ≈ 53 pm/V at a 1550 nm fundamental, together with wide mid-infrared transparency (0.54–8.5 µm), a ~2.3 eV band gap, a femtosecond laser damage threshold of 350 GW/cm², and both type-I and type-II phase-matchability. The first-principles study that established these numbers (He et al., 2022) left two explicit issues open, and no resolution of either is known in the literature to date. First, the standard protocol — density-functional-theory (DFT) structures relaxed with LDA or van der Waals (vdW) corrections, a scissor shift, and an independent-particle evaluation of χ(2)\chi^{(2)} — underpredicts d33d_{33} by roughly 3.5× (−15 vs −53 ± 6.4 pm/V); the authors hypothesized that vdW corrections acting on the electronic density, not merely on the geometry, are needed and deferred the investigation. Second, the proposed origin of the giant response — a proximate double resonance enabled by an isolated intermediate conduction band — remains an unverified hypothesis. The problem is to deliver a quantitative, mechanism-resolved first-principles account of the giant non-resonant SHG of bulk SnP2_2S6_6: reproduce the measured tensor within experimental uncertainties with the responsible physical effect identified, and quantify or refute the isolated-band double-resonance enhancement. A 2025 study of the sister compound SnP2_2Se6_6 reproduces the same theory–experiment gap and adds a competing excitonic hypothesis, sharpening the attribution question.

Background

Second-harmonic generation (SHG) is the nonlinear optical (NLO) process in which a crystal without inversion symmetry converts two photons of frequency ω\omega into one photon of frequency 2ω2\omega; its strength is quantified by the second-order susceptibility tensor χ(2)\chi^{(2)}, conventionally reported as SHG coefficients dijd_{ij} in units of pm/V. Infrared NLO crystals must combine a large non-resonant dijd_{ij} (both fundamental and harmonic photons below the band gap, so absorption is negligible), a high laser-induced damage threshold, a wide transparency window, and phase-matchability (keeping the generated wave in step with the driving polarization along the crystal); the commercial benchmarks are AgGaS2_2 and AgGaSe2_2, whose coefficients are ~13 and ~33 pm/V.

SnP2_2S6_6 is a layered chalcophosphate (space group R3, point group 3, chiral and polar) built from SnS6_6 octahedra and covalent phosphorus dimers, structurally derived from the ferroelectric Sn2_2P2_2S6_6 by replacing half of the Sn2+^{2+} ions with Sn4+^{4+} and leaving the other half of the sites vacant. He et al. (ACS Photonics 9, 1724, 2022) grew bulk single crystals and measured the complete SHG tensor, finding d33d_{33} ≈ 53 pm/V at 1550 nm — larger than AgGaS2_2 and AgGaSe2_2 — with type-I and type-II phase-matchable effective coefficients of roughly 20 and 15 pm/V.

Their first-principles protocol: Kohn–Sham density-functional theory (DFT) in the local-density approximation (LDA) or with a dispersion-corrected (van der Waals, vdW) relaxation of the layered structure; the severely underestimated band gap (1.11 eV LDA, 0.97 eV vdW, vs ~2.3 eV measured) is patched by a rigid upward shift of the conduction bands (a 'scissor shift' of 1.22 or 1.36 eV); and the frequency-dependent χ(2)\chi^{(2)} is evaluated in the independent-particle approximation, i.e., as a sum over independent vertical transitions between Kohn–Sham bands in the length-gauge formalism. Two outcomes of that study frame the open problem. First, whereas most tensor components agree reasonably with experiment, the out-of-plane component d33d_{33} is underpredicted by about a factor of 3.5 (−15 vs −53 ± 6.4 pm/V at 1550 nm) and d14d_{14} even has the wrong calculated sign; the computed values are sensitive to the lattice relaxation (LDA vs vdW) and to the scissor shift. Relaxing the structure with vdW corrections alone was found insufficient, and the authors hypothesized that vdW corrections that also act on the electronic density — as in non-local vdW density functionals, rather than additive dispersion schemes that change only the geometry — are needed to reconcile theory and experiment, explicitly leaving this for future work. Second, the computed band structure contains an isolated, weakly dispersive first conduction band (bandwidth ≈ 0.7 eV), separated by gaps from the valence band below and the upper conduction bands above. Using a sum rule that reconstructs the static non-resonant χ(2)\chi^{(2)} from the spectrum of Im[χ(2)\chi^{(2)}], the authors showed that low-energy one-photon transitions (≈ 2.4 eV) from the valence band to this isolated band dominate the response, and proposed that the giant non-resonant SHG arises from a proximate double resonance: virtual two-photon paths (valence → isolated band → upper conduction bands) in which the fundamental and second-harmonic photon energies lie just below the two resonant one-photon transition energies — the bulk-crystal analogue of double-resonance enhancement known from asymmetric semiconductor quantum wells. This mechanism, however, was presented as a hypothesis whose quantitative verification was deferred to future work.

The sister compound SnP2_2Se6_6 measured in 2025 shows an even larger non-resonant d33d_{33} (−222 ± 30 pm/V at 2 µm) and the same ~3.7× underprediction by an independent-particle-plus-scissor calculation, this time on the experimental crystal structure, leading that study to propose planar exciton confinement as the missing ingredient.

Problem Statement

The problem left open by He et al. (2022) is to convert the proximate explanation of SnP2_2S6_6's record non-resonant SHG into a quantitative, verified account. One scientific objective — a mechanism-resolved first-principles understanding of why bulk SnP2_2S6_6, and especially its out-of-plane component d33d_{33}, has such a large non-resonant second-harmonic response — with two linked questions whose answers must be mutually consistent.

(1) Protocol/accountability question: identify which approximation of the standard protocol (DFT geometry, exchange–correlation treatment, scissor shift, independent-particle χ(2)\chi^{(2)}) is responsible for the ~3.5× underprediction of d33d_{33}, and demonstrate a documented first-principles protocol that reproduces the complete measured SHG tensor of bulk SnP2_2S6_6 within the experimental uncertainties — in particular d33d_{33} = 53 ± 6.4 pm/V in magnitude at a 1550 nm fundamental (the sign of dijd_{ij} is convention-dependent; the source tabulates the measured value as ∓53 ± 6.4 pm/V against a calculated ∓15 pm/V, so agreement is to be assessed within the stated uncertainties on a single consistent sign convention). The source paper's specific hypothesis is that van der Waals corrections acting on the electronic density (non-local vdW functionals, not merely dispersion-corrected geometry) are required; competing hypotheses raised by later work, notably excitonic and local-field effects beyond the independent-particle approximation, must be weighed alongside it in the attribution.

(2) Mechanism question: quantitatively verify or refute the proximate double-resonance mechanism — determine how much of the non-resonant SHG response is contributed by virtual transitions through the isolated intermediate conduction band, for instance by band- or transition-resolved decomposition of χ(2)\chi^{(2)}, a model-Hamiltonian analysis with and without the isolated band, or targeted experiments (e.g., SHG dispersion measured toward the resonant range at ≈2.4 eV), and reconcile the result with the measured tensor.

Completion boundary: the problem is solved when a first-principles account reproduces the measured SHG tensor of bulk SnP2_2S6_6 within experimental error bars with the responsible physical effect(s) identified, and the isolated-band double-resonance contribution to the enhancement is quantified — or the mechanism is quantitatively refuted and replaced by a verified alternative account of the same measurements.

References

  1. Jingyang He, Seng Huat Lee, Francesco Naccarato, Guillaume Brunin, Rui Zu, Yuanxi Wang, Leixin Miao, Huaiyu Wang, Nasim Alem, Geoffroy Hautier, Gian-Marco Rignanese, Zhiqiang Mao, Venkatraman Gopalan (2022). SnP2_2S6_6: A Promising Infrared Nonlinear Optical Crystal with Strong Nonresonant Second Harmonic Generation and Phase-Matchability. ACS Photonics 9(5), 1724–1732. DOI: 10.1021/acsphotonics.2c00131 (preprint arXiv:2204.04298). https://doi.org/10.1021/acsphotonics.2c00131
  2. Jadupati Nag, Saugata Sarker, Safdar Imam, Abishek Iyer, Michael J. Waters, Albert Suceava, James M. Rondinelli, Mercouri G. Kanatzidis, Venkatraman Gopalan (2025). Large Non-Resonant Infrared Optical Second Harmonic Generation in Bulk Crystals of Van der Waals Semiconductor, SnP2_2Se6_6. Advanced Optical Materials 13(9), 2402649. DOI: 10.1002/adom.202402649. https://doi.org/10.1002/adom.202402649
  3. Jingyang He, Yingdong Guan, Victor Trinquet, Guillaume Brunin, Ke Wang, Robert Robinson, Rui Zu, Suguru Yoshida, Seng Huat Lee, Yu Wang, Yanglin Zhu, Gian-Marco Rignanese, Zhiqiang Mao, Venkatraman Gopalan (2023). MgSiP2_2: An Infrared Nonlinear Optical Crystal with a Large Non-Resonant Phase-Matchable Second Harmonic Coefficient and High Laser Damage Threshold. Advanced Optical Materials 11(24), 2301060. DOI: 10.1002/adom.202301060. https://doi.org/10.1002/adom.202301060
  4. Yue Zhang, Fakun Wang, Xin Feng, Zongdong Sun, Jianwei Su, Mei Zhao, Shuzhe Wang, Xiaozong Hu, Tianyou Zhai (2022). Inversion symmetry broken 2D SnP2_2S6_6 with strong nonlinear optical response. Nano Research 15(3), 2391–2398. DOI: 10.1007/s12274-021-3806-0. https://doi.org/10.1007/s12274-021-3806-0
  5. Peng-Yuan Chen, Zhen Quan, Shu-Dong Wang (2024). Excitonic optical properties in monolayer SnP2_2S6_6. Chinese Physics B 33(10), 107105. DOI: 10.1088/1674-1056/ad73b5. https://doi.org/10.1088/1674-1056/ad73b5