{"id":"a9e56ee9-58a0-47b2-b950-dd3e91eb77b3","arxiv_id":"2501.03348","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"Erbium crystal field splittings in five oxides are reproduced from DFT only after a per-material fitted radial scaling of the 4f wavefunction.","lead":"This paper computes the crystal field splittings of erbium ions in five oxide crystals by combining density functional theory with a crystal field model. The match to measured spectra only holds after the size of the erbium 4f orbital is rescaled with a fitted constant for each material.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The per-host fitted epsilon is the load-bearing parameter: fitted to the same experimental splittings used for validation, and inconsistent with the Wannier/ionization estimate, so the ab initio predictive claim is not yet demonstrated.","rationale":"I focused on the fitting of epsilon because it is the gatekeeper for the whole central claim. The DFT part supplies V_CF, but the 4f electrons are frozen in the core and the radial function is not obtained from DFT; it is a hydrogenic Ansatz whose only free part is epsilon. The CFC integrals are weighted by R_4f^2, so epsilon multiplies the effective length scale of all moments, and the k=6 coefficients are especially sensitive. Thus the agreement in Fig. 2 is not primarily a test of the DFT charge density; it is a test of whether a one-parameter radial scaling can mimic the true CFCs. The paper's own evidence undermines the physical identification of the parameter: maximally localized Wannier functions match the hydrogenic form for epsilon between 3 and 4, and the Er3+ ionization energy gives epsilon=3.96, while the fitted values are near 2. This factor-of-two discrepancy is exactly the radial reduction advertised in the abstract, so the method is internally consistent if read as 'reduce relative to the Wannier value', but it means the fitted epsilon is not the physical dielectric constant. Rather, it absorbs all errors in the radial model, the DFT potential, and the frozen-core approximation. Because the parameter is fitted per host to the experimental levels, the reported agreement is not an independent confirmation. I do not call for rejection: the symmetry-constrained nonzero CFC patterns, the use of the independently developed qlanth Hamiltonian, and the qualitative ordering of multiplets are genuine successes. The condition for accepting the central claim as predictive is an out-of-sample test or a first-principles derivation of the scaling factor. The reader's CONDITIONAL verdict already captures this, so I leave the verdict unchanged.","tokens_in":10763,"tokens_out":6554,"duration_ms":63570,"concrete_test":"Run a strict out-of-sample test on all five hosts: (i) fit epsilon for each host using only the lowest observed Z-levels (e.g., Z1-Z4), then predict the remaining Z-levels and all Y-manifold splittings; report RMS errors with the fixed free-ion parameters. (ii) Repeat the entire calculation with a single global epsilon = 2.0 across all hosts, and again with epsilon = 3.96 from the Wannier/ionization estimate, without any re-fitting. If fixed-global or Wannier-epsilon runs reproduce measured splittings with RMS error below about 15 cm^-1, the per-host fit is not carrying the result; if errors grow to tens of cm^-1, the fitted epsilon is the load-bearing assumption. Either outcome settles whether the method is predictive or semi-empirical.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of a predictive ab initio method hinges on epsilon_{omega,q}, which controls the 4f radial extent through R_4f(r) = A r^3 exp(-r Z_eff/(4 a0 epsilon_{omega,q})). A larger epsilon makes a more diffuse orbital, so the fitted values (MgO 1.96, ZnO 1.90, TiO2 2.00, CaWO4 2.24, PbWO4 2.24) shrink the 4f shell by roughly a factor of two relative to the Wannier/ionization estimate of epsilon ~ 3.96 reported in the same paper. The paper states that epsilon is 'carefully optimized by fitting the calculated and the experimental energy levels' separately for each host. Because B_kq = (2k+1)/4pi ∫ V_CF(r) R_4f^2 C^k* r^2 dr, and the radial moments scale approximately as (a0 epsilon/Z_eff)^k, this single parameter strongly controls all CFCs, especially B6. Therefore the excellent agreement in Fig. 2 is substantially in-sample: for each of five hosts, one free parameter is adjusted to the same experimental spectra that are then quoted as validation. The paper does not state which experimental levels entered the fit, how many levels constrained each epsilon, or the fit residuals, making it impossible to separate real predictive power from parameter flexibility. A single scalar cannot by itself force agreement of an entire multiplet pattern if the symmetry and Hamiltonian are wrong, so the method may retain value; however, the 'ab initio with one physically justified parameter' claim requires out-of-sample evidence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an effective method for computing crystal-field coefficients (CFCs) B_k^q of Er3+ in five wide-band-gap oxide hosts (MgO, ZnO, TiO2, CaWO4, PbWO4) by combining DFT calculations with the 4f electrons placed in the core and a hydrogenic 4f radial wavefunction whose extent is controlled by a per-host dielectric constant epsilon. These CFCs are then fed into the qlanth Hamiltonian to compute crystal-field splittings of the ground 4I15/2 and first excited 4I13/2 manifolds. The authors report good agreement with low-temperature measurements, provided the 4f radial wavefunction is contracted by about a factor of two. The central claim is that this constitutes an ab initio method with a single physically justifiable adjustable parameter.","tokens_in":11125,"tokens_out":5308,"duration_ms":50687,"significance":"If the predictive claim were established, the method would provide a practical DFT-based route to crystal-field splittings of Er3+ in oxides, which is valuable for quantum communication and memory applications. The paper has clear strengths: it treats four distinct site symmetries, respects the symmetry-dictated form of the CFCs (e.g., the MgO relations B4±4 = sqrt(5/14) B4_0 and B6±4 = -sqrt(7/2) B6_0 are satisfied), and uses an open-source Hamiltonian code with documented improvements. However, the central claim is not yet demonstrated because the per-host dielectric constant is fitted to the same experimental levels used for validation and because the fitted values contradict the paper's own independent Wannier- and ionization-energy-based estimates. The significance is therefore conditional on resolving these load-bearing issues.","major_comments":[{"comment":"The central result depends on the per-host dielectric constant epsilon, which the text says is 'carefully optimized by fitting the calculated and the experimental energy levels.' The same experimental splittings are then displayed as validation in Fig. 2. Since B_k^q is computed from integrals involving R_4f^2 and the radial moments scale approximately as (a0 epsilon/Z_eff)^k, fitting epsilon per host can absorb a large part of the systematic error. The paper should report the fit protocol explicitly: which Z and Y levels entered the fit, how the residual was minimized, and the per-host residuals after fitting. It should also provide an out-of-sample test, such as a single epsilon fixed across all hosts or a leave-one-host-out prediction, before claiming predictive ab initio accuracy.","section":"Methods, R4f paragraph"},{"comment":"The fitted epsilon values (1.96 for MgO, 1.90 for ZnO, 2.00 for TiO2, 2.24 for CaWO4 and PbWO4) directly contradict the paper's own independent estimates. The text states that the Wannier-extracted 4f function agrees with the hydrogenic form when epsilon is between 3 and 4, and that the ionization-energy estimate gives epsilon = 3.96. With the hydrogenic form R4f(r) = A r^3 exp(-r Zeff/(4 a0 epsilon)), a smaller epsilon produces a more contracted orbital, and the CFC integrals scale as powers of (a0 epsilon/Zeff). This discrepancy is therefore not cosmetic: it undermines the claim that epsilon is the physically justified dielectric constant. Either the fitted values must be reconciled with the independent estimates, or the paper should explicitly describe epsilon as an empirical scaling parameter and remove the claim of physical justification.","section":"Methods, R4f paragraph and Wannier/ionization comparison"},{"comment":"The comparison in Fig. 2 is restricted to splittings within each manifold because, for every host, the calculated Y manifold is shifted so that Y1 aligns with the experimental Y1 value (e.g., -31 cm^-1 for Er:MgO). The stated Y1-Z1 transition energies differ from experiment by tens of wavenumbers in several cases (e.g., +63.32 cm^-1 for ZnO and +64.30 cm^-1 for CaWO4). The paper should state clearly that absolute multiplet energies are not predicted and should tabulate these Y1-Z1 residuals for all hosts. As it stands, the excellent agreement for the excited manifold is partly a consequence of this alignment convention.","section":"Fig. 2 and Y-manifold alignment"},{"comment":"PbWO4 is included in Table I and in the abstract, but no level-by-level comparison of the Z and Y manifolds is presented for PbWO4. The text reports only the Y1-Z1 transition (6570.05 cm^-1) and states that it shows good agreement. A figure or table analogous to Fig. 2 for PbWO4 is needed to support the claimed agreement for this host.","section":"PbWO4, section 3 and Table I"}],"minor_comments":[{"comment":"The phrase 'reducing the radial extent of the 4f wavefunctions by approximately a factor of 2' is directionally confusing when read against the formula R4f(r) = A r^3 exp(-r Zeff/(4 a0 epsilon)), because increasing epsilon in this formula spreads the orbital. Please specify the baseline (e.g., epsilon ~ 4 from Wannier/ionization versus the fitted epsilon ~ 2) and state the contraction factor explicitly.","section":"Abstract and Introduction"},{"comment":"The free-ion parameters (F^k, zeta_4f, alpha, beta, gamma, T^i, M^h, P^f) are taken from Ref. [17], which is a LaF3 analysis. The paper does not assess whether these parameters are transferable to the five oxide hosts; a sensitivity test with respect to these parameters would strengthen the conclusions.","section":"Methods, atomic parameters"},{"comment":"Table I uses entries with real and imaginary parts indicated by signs such as '∓' and '±' without fully defining the convention for B_k,±q. Please state the convention explicitly (e.g., column lists both B_k, q and B_k, -q with the stated sign relations) and ensure the table is self-explanatory.","section":"Table I"},{"comment":"The choice to keep 4f electrons in the core and to treat 6s2, 5p6, and 5d1 as valence electrons is central to the method, but the paper does not justify this choice or test its sensitivity. A brief discussion or a supplementary test would help establish the robustness of the CFCs.","section":"Methods, DFT setup"},{"comment":"Several details of the calculation are deferred to the simultaneously submitted companion works (Refs. [36] and [64]). The core method should be reproducible from the present manuscript alone, so please include the essential procedural details and either provide explicit data or clearly indicate where they will be permanently available.","section":"Reproducibility and companion paper"}],"recommendation":"major_revision","confidential_remarks":"The paper would be suitable for publication if reframed as a semi-empirical method with a fitted per-host scaling parameter, or if the authors can demonstrate genuine out-of-sample predictive power using an epsilon derived from Wannier/ionization arguments. As written, the 'ab initio with one physically justifiable parameter' claim is overstated because the parameter is fitted to the very spectra used for validation and disagrees with the independent physical estimate reported in the same paper. The companion manuscript should be consulted to verify whether additional details alter this assessment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a plausible semi-empirical route to Er crystal field splittings in oxides, but the central 'ab initio with one physically justified parameter' claim does not hold as stated. The single parameter is fit per host to the very spectra used for validation, and there's an internal inconsistency about whether it expands or contracts the 4f radial function.\n\nWhat's actually new: they compute CFCs for Er in MgO, ZnO, TiO2, CaWO4, and PbWO4 within the 4f-in-core DFT framework, using a hydrogenic 4f radial wavefunction screened by a host dielectric constant. Those specific CFC tables and splittings are new. The symmetry analysis for each site is careful and consistent with group theory, and the DFT setup is standard. They also report improvements to the qlanth code and show it can reproduce Carnall's LaF3 results to within about 30 cm^-1. That part is solid.\n\nThe soft spots are not minor. The dielectric constant epsilon_omega,q is 'carefully optimized by fitting the calculated and the experimental energy levels' — that is, fit to the same splittings they then display as agreement. The paper doesn't state which levels entered the fit or how many, so we can't separate predictive power from parameter flexibility. More damning, the formula R_4f ~ exp(-r Zeff/(n a0 epsilon)) makes a larger epsilon produce a more diffuse orbital, while the abstract says the fit reduces the radial extent by a factor of two. The fitted epsilons are ~2, whereas their own Wannier and ionization-energy estimates give ~4. So the scaling parameter contradicts the paper's own physical justification. The Y manifold is also shifted to align Y1, and remaining deviations are 20-65 cm^-1. These are not fatal to the method's usefulness as a compact empirical description, but they undercut 'excellent agreement' and the ab initio claim.\n\nWho should read it: anyone working on Er-based quantum memories or spin-photon interfaces who wants a quick way to get approximate CFCs for oxide hosts, and who understands they're semi-empirical. The paper deserves peer review because the issues are fixable — an out-of-sample test (compute epsilon from Wannier/ionization or a different host, then predict) would change the picture. I'd recommend sending it to review with the expectation of major revision, not desk rejection.","headline":"A useful semi-empirical scheme for Er crystal field splittings in oxides, but the ab initio claim is undercut by per-host fitting and an internal inconsistency in the radial wavefunction screening.","tokens_in":11681,"tokens_out":2692,"would_cite":false,"duration_ms":25351,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"One fitted parameter predicts erbium splittings in oxides","keywords":["erbium","crystal field splitting","density functional theory","4f core approximation","rare earth","oxide hosts","quantum memory","dielectric screening"],"falsifier":"Compute the crystal field coefficients for a new oxide host using the same workflow with an epsilon near 2, then compare the predicted Z- and Y-manifold splittings to low-temperature spectra; disagreement beyond about 30 cm-1 in the relative splittings would falsify the single-parameter claim.","tokens_in":10569,"feed_emoji":"🔬","tokens_out":7093,"duration_ms":56995,"temperature":0.7,"pith_summary":"This paper claims that the crystal field coefficients of Er3+ ions in five wide-band-gap oxide hosts can be computed from density functional theory (DFT) within the 4f-core approximation, where the 4f electrons are frozen in the core. The only adjustable input is a per-host dielectric constant that sets the radial extent of a hydrogenic 4f wavefunction; fitted values cluster near 2, which shrinks the orbital by about a factor of two relative to the extent inferred from Wannier functions or the Er3+ ionization energy. With these coefficients, an effective Hamiltonian reproduces the ground and excited manifold splittings measured at low temperature to within tens of wavenumbers across four different local site symmetries. If the method holds, it offers a cheap, single-parameter route to predicting erbium crystal-field splittings in candidate quantum-memory hosts.","feed_headline":"One fitted parameter predicts erbium splittings in oxides","feed_subtitle":"DFT-based crystal field coefficients match low-temperature spectra to within tens of wavenumbers across four site symmetries.","key_machinery":"The working object is the hydrogenic 4f radial wavefunction $R_{4f}(r) = A r^3 \\exp(-r Z_{\\mathrm{eff}}/(n a_0 \\epsilon))$, with $Z_{\\mathrm{eff}}$ obtained from atomic screening constants and $\\epsilon$ a per-host dielectric constant fitted to experiment. This wavefunction is used to evaluate the radial integrals that convert the DFT crystal-field potential into the coefficients $B^k_q$; the coefficients are then diagonalized together with the free-ion Hamiltonian (electrostatic, spin-orbit, two- and three-body interactions) to produce the crystal-field splittings.","core_discovery":"The central discovery is that the crystal field coefficients $B^k_q$ of Er3+ in oxide hosts can be obtained from the self-consistent DFT charge density and local potential after freezing the 4f electrons in the core, provided the radial part of the 4f orbital is taken to be hydrogenic with an exponent controlled by a fitted dielectric constant near 2. With these coefficients, the effective Hamiltonian reproduces the measured splittings of the ground ($^4I_{15/2}$) and first excited ($^4I_{13/2}$) manifolds for all five hosts, spanning local site symmetries $O_h$, $C_{3v}$, $D_{2h}$, and $S_4$. The authors argue that the single fitted factor of about two accounts for the known inadequacy of DFT in describing the strongly correlated, partially filled 4f shell.","pith_inferences":["Because the fitted epsilon differs from both the Wannier-derived value (~3-4) and the bulk static dielectric constants of these oxides, the parameter likely absorbs several distinct errors—DFT self-interaction, the hydrogenic approximation to the 4f orbital, and neglect of lattice relaxation around the dopant. A direct test would be to compare CFCs computed with a self-consistently relaxed 4f orbi","The near-constancy of the fitted epsilon (1.90-2.24) across hosts with very different dielectric properties suggests the correction is ionic rather than host-specific; if so, a universal epsilon near 2 may work for other wide-band-gap oxides and even for other rare-earth ions, a claim the paper does not make.","The paper aligns calculated Y-manifold centers to experiment, so the method predicts splittings within a manifold, not the absolute 4f-4f transition energy. A stronger test would be to predict both and check whether the residual ~30-60 cm-1 discrepancies in Y1-Z1 gaps are systematic."],"forward_implications":["The method gives a practical workflow: one DFT calculation with 4f in the core, one fitted epsilon, then an effective-Hamiltonian diagonalization, yielding CFCs and splittings for a new oxide host without iterative fitting to spectroscopy.","Agreement across four local site symmetries suggests the single-parameter correction captures the dominant screening error, so the approach may transfer to other trivalent rare-earth ions.","Predicted CFCs can be used to design erbium-based quantum memories and spin-photon interfaces, since the Y1-Z1 transition energy and manifold structure are reproduced.","The Wannier and ionization-energy estimates of epsilon (around 4) bracketing the fitted value (around 2) indicate the parameter is an effective screening correction rather than the true bulk dielectric constant."],"supporting_citations":[{"why":"Supplies the low-temperature experimental splittings for Er in MgO, ZnO, TiO2, CaWO4, and PbWO4 that the calculated values are compared against.","marker":"[4]"},{"why":"Provides the free-ion Hamiltonian parameters (electrostatic, spin-orbit, and configuration-interaction terms) used in the effective Hamiltonian.","marker":"[17]"},{"why":"Introduces the 4f-core approximation for computing crystal field parameters of rare-earth ions from DFT, the basis of the present method.","marker":"[19]"},{"why":"Extends the 4f-core approach with Wannier functions to rare-earth aluminates, providing the comparison for the 4f radial wavefunction.","marker":"[20]"},{"why":"Supplies the lanthanide Hamiltonian code used to diagonalize the free-ion plus crystal field Hamiltonian and produce the manifold splittings.","marker":"[23]"},{"why":"Provides the atomic screening constants from which the effective nuclear charge Z_eff of the 4f orbital is derived.","marker":"[32]"},{"why":"Used to extract maximally localized Wannier 4f functions for comparison with the hydrogenic radial model, yielding the epsilon estimate of 3-4.","marker":"[33]"}],"fun_headline_variants":["Single tweak makes ab initio erbium splittings match","Fitted dielectric constant nails erbium crystal fields","Erbium in oxides: ab initio plus one fudge factor","One scaling factor fixes erbium crystal field theory","DFT plus one constant reproduces erbium splittings"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire calculation rests on representing the 4f orbital as hydrogenic with a single fitted dielectric constant per host; if that functional form or the fitted epsilon value is wrong, all computed splittings change.","fun_headline_variants_meta":{"raw":{"variants":["Single tweak makes ab initio erbium splittings match","Fitted dielectric constant nails erbium crystal fields","Erbium in oxides: ab initio plus one fudge factor","One scaling factor fixes erbium crystal field theory","DFT plus one constant reproduces erbium splittings"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000781,"raw_usage":{"total_tokens":3401,"prompt_tokens":847,"completion_tokens":2554,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":463,"completion_tokens_details":{"reasoning_tokens":2469}},"tokens_in":463,"tokens_out":2554,"duration_ms":17241,"temperature":1.0,"reasoning_tokens":2469,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:52:48.303390+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the crystal field coefficients for a new oxide host using the same workflow with an epsilon near 2, then compare the predicted Z- and Y-manifold splittings to low-temperature spectra; disagreement beyond about 30 cm-1 in the relative splittings would falsify the single-parameter claim.","supporting_citations":[{"cited_title":"Rare-earth-doped materials for applications in quantum information storage and signal processing,","cited_arxiv_id":null,"evidence_quote":"Supplies the low-temperature experimental splittings for Er in MgO, ZnO, TiO2, CaWO4, and PbWO4 that the calculated values are compared against."},{"cited_title":"Synthesis and lumi- nescence properties of erbium-doped Y2O3 nanotubes,","cited_arxiv_id":null,"evidence_quote":"Provides the free-ion Hamiltonian parameters (electrostatic, spin-orbit, and configuration-interaction terms) used in the effective Hamiltonian."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the atomic screening constants from which the effective nuclear charge Z_eff of the 4f orbital is derived."}],"review_version":1}