WorksPhysics and Astronomy
Thermoelectric properties and modulation doping of ultrawide-bandgap nitride monolayers
Abstract
Two-dimensional III-nitride alloys provide a composition-tunable platform for exploring thermoelectric transport and intrinsic electric-field tolerance. We combine first-principles electronic-structure calculations with semiclassical transport for ordered monolayer BAlN and BGaN. State-dependent momentum relaxation times computed with AMSET include acoustic deformation-potential and piezoelectric scattering. Using a common, temperature-independent lattice thermal conductivity of 550 W/mK, we predict maximum thermoelectric figure of merit ZT>2.5 for selected p- and n-type monolayers and ZT>2 at 300~K for n-type BGaN. Composition-dependent band curvature, screening, and scattering help interpret the different p- and n-type responses. Vacuum-referenced band alignments identify candidate modulation-doping partners for both carrier types; explicit hGaN/VO calculations yield a hole surface density of approximately 1.6 cm. A separate phenomenological model using the band gap and maximum phonon frequency predicts intrinsic breakdown fields of 20.06 and 20.74 MVcm for the Al- and Ga-containing x=0.75 alloys, respectively. These results motivate further investigation for thermoelectric and high-field power-electronic applications. The ZT values apply to the specified scattering and lattice-conductivity models, while the breakdown estimates require validation for monolayers and do not resolve field direction.
manuscript.pdf
Thermoelectric properties and modulation doping of ultrawide-bandgap nitride monolayers
Abstract
Introduction
III-nitride semiconductors combine wide bandgaps with chemical, mechanical, and thermal stability [2] [1], supporting applications in high electron mobility transistors [4] [3], light-emitting diodes [5], lasers [7] [6], and quantum sensors [9] [8]. The high critical fields of wide-bandgap materials such as GaN motivate their use in power switching and RF electronics [11] [10]. Ultrawide-bandgap (UWBG) materials extend this search to gaps wider than that of GaN, including AlN, Al-rich AlGaN, -GaO, and diamond [13] [12]. Their prospects depend jointly on field tolerance, carrier transport, dopability, contacts, and heat removal. These challenges differ by material: low thermal conductivity is a particular limitation of -GaO, whereas diamond has exceptionally high thermal conductivity [12]. Monolayer III-nitride alloys offer an additional setting in which composition and reduced dimensionality can modify the electronic structure and intrinsic field tolerance.
Thermoelectric energy conversion provides a complementary motivation for studying these alloys. Waste-heat recovery and localized cooling require materials that balance electrical conduction and thermopower against heat conduction [14]. BiTe-based alloys are established low-temperature thermoelectrics [17], while device-integrated thermoelectric cooling offers a route to managing local heat generation [20]. Large bandgaps increase the energy required for thermal electron–hole pair generation and can help suppress bipolar transport at elevated temperatures. A wide gap alone, however, does not establish a large thermoelectric figure of merit: the band
dispersion, carrier scattering, and lattice heat transport must also be considered. Low-temperature transport is particularly relevant to wearable generators and sensors [21].
The tellurium-free composition of the III-nitrides also motivates examining their materials cost and suitability for low-toxicity applications. The historical laboratory-grade BiTe catalog prices cited here range from $6 to $17 per gram [28] [27], compared with constituent-material estimates of approximately $0.12/g for Ga-based nitrides and $0.003/g for Al-based nitrides [29]. These estimates motivate a possible material-cost advantage: they exclude monolayer synthesis, processing, and device fabrication. Reports of biocompatibility and low cytotoxicity for GaN surfaces [32] provide further motivation, while the biocompatibility of each alloy and device structure requires separate evaluation.
Compared with bulk, nanowire, and quantum-dot III-nitrides, their two-dimensional forms provide distinct bonding environments and opportunities for interface control [33]. In boron-based ordered monolayers, composition-induced puckering changes the band-edge orbital character, symmetry, and directional dispersion. Band curvature and dielectric screening therefore offer useful descriptors for interpreting composition-dependent transport, while vacuum-referenced band edges guide the selection of contacting materials. Two-dimensional interfaces also permit modulation doping, in which an adjacent layer supplies carriers to the conducting layer [35] [34]. Spatial separation of the dopant and channel can reduce ionized-impurity scattering, motivating its use for both electronic and thermoelectric transport [36].
Building on our studies of ordered nitride monolayers [33] and interfacial charge-transfer doping [35], we investigate monolayer BGaN and BAlN using first-principles electronic structure and semiclassical transport theory. Earlier calculations by Wines, Ersan, and Ataca examined transport coefficients of related two-dimensional III-nitride alloys using a constant relaxation-time approximation, but did not report [40]. Here, state-dependent AMSET momentum relaxation times [41] are used to evaluate thermoelectric response across temperatures and carrier concentrations, with particular attention to performance near 300 K under a stated common lattice-conductivity reference. We interpret the results using the composition-dependent masses, dielectric response, work functions, and effective polar-optical-phonon frequencies in Table I. Candidate p- and n-type modulation-doping partners are selected by band alignment, with an explicit hGaN/VO interface illustrating hole transfer. We also estimate intrinsic breakdown fields with the Kim–Pilania–Ramprasad band-gap/maximum-phonon-frequency model [42] and examine their relevance to high-field power electronics. This maximum-frequency input is distinct from the effective polar-optical frequency used to discuss carrier scattering. The thermoelectric predictions, proposed carrier-control routes, and intrinsic-field estimates are assessed within their respective modeling assumptions.
| Material | (eV) | (THz) | Estimated (MV/cm) | ||
| hAlN | 3.54 | 1.83 | 5.07 | — | 6.23 |
| BAlN | 4.09 | 1.76 | 5.40 | 23.98 | 8.90 |
| BAlN | 4.42 | 1.73 | 5.36 | 24.29 | 15.95 |
| BAlN | 4.11 | 1.85 | 5.30 | 29.25 | 20.06 |
| hGaN | 4.14 | 1.99 | 5.03 | — | 4.84 |
| BGaN | 3.94 | 2.28 | 5.22 | 28.98 | 10.99 |
| BGaN | 4.87 | 1.72 | 5.25 | 23.42 | 12.98 |
| BGaN | 4.22 | 6.55 | 5.34 | 28.07 | 20.74 |
| Material | Direction | Direction | ||
| hAlN | 0.52 | -1.47 | ||
| 0.52 | -1.55 | |||
| BAlN | 0.55 | -1.97 | ||
| 0.53 | — | — | ||
| BAlN | 0.68 | -0.93 | ||
| 0.65 | -4.71 | |||
| BAlN | 0.83 | -0.73 | ||
| — | — | -1.3 | ||
| hGaN | 0.27 | -1.10 | ||
| 0.27 | -1.17 | |||
| BGaN | 0.50 | -3.3 | ||
| 0.52 | — | — | ||
| BGaN | 0.54 | -0.51 | ||
| 0.56 | -0.51 | |||
| BGaN | 0.75 | -0.94 | ||
| 0.79 | — | — |
Table I. Electronic properties and estimated intrinsic dielectric breakdown fields of the monolayers. (a) Material-level dielectric response, work function, effective polar-optical-phonon frequency, and estimated breakdown field. (b) Directional band-curvature masses. The valence-band entries retain their negative curvature signs; hole transport is discussed using their magnitudes. Dashes denote values not reported.
Results and discussion
Crystal structure and electronic band structure
We considered monolayer boron-based III-nitride ordered alloys, BGaN and BAlN, for boron fractions ranging from to . Thermodynamic and dynamic stability of these materials has been demonstrated in a prior work of the author [33].
The evolution of the electronic structure with change in fractional B composition is shown in Figure 1(c). When boron is introduced into pristine hexagonal GaN and AlN, the crystal structures buckle in the out-of-plane direction, with N atoms shifting from a trigonal-planar to a pyramidal bonding configuration, as shown in Figure 1(a). This structure is induced by the large difference in atomic size between B and Al/Ga (atomic radii: 0.85 Å and 1.25/1.30 Å, respectively), which leads to a strained heterogeneous bonding environment that is relieved by adopting a lower energy puckered configuration. As a result, the lattice symmetry is reduced, permitting previously symmetry-forbidden mixing between and , orbitals and modifying the valence-band character and bandgap. The resulting valence bands can exhibit strongly anisotropic dispersion and extrema at low-symmetry points of the Brillouin zone. Table I quantifies the band curvatures and electrostatic response; their connection to transport and charge transfer is discussed in Sec. II E.

Figure 1. (a) Top and side views of the 2D III-nitride crystal structures, arranged left to right and top to bottom for ; ; ; and . (b) The Brillouin zone of the hexagonal materials (left) and the reduced symmetry Brillouin zone of the puckered materials (right). (c) Electronic structure of (top), and (bottom) computed using the hybrid DFT functional HSE06.
Thermoelectric performance
The thermoelectric performance of a material at a given temperature can be assessed by the dimensionless figure of merit
where is the electrical conductivity, the Seebeck coefficient , the temperature , the electronic thermal conductivity , and the lattice thermal conductivity .
Carrier transport is evaluated using state-dependent AMSET MRTA lifetimes for the stated acoustic deformation-potential and piezoelectric scattering model [41]. All nitride estimates use the common reference W m K, held fixed across compositions and sampled temperatures. The models and normalization conventions are detailed in Sec. III.
Figure 2 presents the maximum thermoelectric figure of merit of the III-nitride monolayers and compares these predictions with representative thermoelectric material families. At all B-fractions a is exhibited, and a majority display an average on the temperature range 300–900K. Three monolayers exhibit exceptional high-temperature , which compares favorably with both established TE materials such as BiTe and other emerging 2D materials such as transition metal dichalcogenides (TMDs). The composition-dependent balance between band dispersion, screening, and scattering underlying these results is examined using Table I in Sec. III E.

Figure 2. Maximum thermoelectric figure of merit of III-nitride monolayers compared with that of representative families of TE materials. The III-nitride estimates use throughout; this assumption does not apply to the literature benchmarks. Other 2D materials include: graphene nano-ribbons (GNR) [43, 44, 45], WS [46, 47, 48], MoSe [46, 48, 49], MoS [46, 47, 48, 50, 51], WSe [46, 47, 48, 49], PtTe [52, 53], graphene [54], while bulk materials include: InSb [55, 56], MgSb [57], BiCuSeO [58, 59, 60, 61], SiGe [62, 63, 64], PbSe [65, 66], Cu(I, Ga)Te [67, 68, 69], PbS [70, 71, 72], CuS [58, 59, 60, 61], Half-Heusler [73, 74, 75, 76, 77], BiTe [19, 78, 79], CoSb [80, 81, 82], SnSe [83, 84, 85, 86]. N-type (p-type) doping of the 2D III-nitrides is indicated by navy (pink) markers respectively.
Low-temperature thermoelectric performance
The low-temperature (300K) thermoelectric figure of merit for both n- and p-type doping of the 2D III-nitride materials was investigated and is presented in Figure 3. The performance of the monolayers is compared directly with other 2D materials such as TMDs in Figure 3(a), with the figure of merit at 300K plotted against the bandgap. While all of the semiconducting 2D materials exhibit significant , most feature bandgaps substantially smaller than those of the III-nitrides, which at increased temperatures can result in excitation of minority carriers and shift in chemical potential which suppresses the thermoelectric figure of merit. Moreover, unlike TMDs such as MoS, the III-nitride materials are not limited to low temperature TE applications due to onset of sublimation at K [94]. In Figure 3(b), we compare the 2D III-nitrides with established bulk low-temperature TE materials. Under the adopted lattice-conductivity reference, the predicted III-nitride values compare favorably with the selected bulk benchmarks at the sampled low temperatures, with obtained for BGaN.

Figure 3. Maximum low-temperature (300 K) thermoelectric figure of merit . The III-nitride estimates use throughout; the literature benchmarks retain their original thermal-conductivity models or measurements. (a) Comparison of monolayer 2D materials, including graphene [54], MoS [46, 47, 48, 50, 51], MoSe [46, 48, 49], WS [46, 47, 48], WSe [46, 47, 48, 49], GaS [87], GaSe [87], GaTe [87], and graphene nano-ribbons (GNR) [43, 44, 45]. corresponding to n-type (p-type) 2D III-nitrides is indicated by solid (hatched) blue markers respectively. (b) Comparison between of III-nitride monolayers and that of representative bulk low-temperature TE materials including BiSbTe [88, 89], BiTeSe [88, 89], BiSb [89, 90], CePd [89, 91], CsBiTe [89, 92], MgBiSb [89, 93], and YbAl [89, 91].
Modulation doping and interfacial charge transfer
Wide bandgap semiconductors such as hBN frequently exhibit a doping asymmetry problem (i.e. where only one type of doping – p- or n-type – of the material is easy due to band alignment, while the other is difficult). This presents a serious obstacle to realization of wide bandgap semiconductor devices. For wide bandgap nitrides like AlN or hBN, one of the principal difficulties for n-type doping is the close proximity of the CBM to the vacuum level [95], which results in a small intrinsic electron affinity . The selective alloying of elements that shift the CBM away from the vacuum level increases the electron affinity and dopability of the material [96]. As shown in Figure 5(a), the introduction of group-III elements (Al, Ga) into hBN results in a more favorable band alignment for n-type conduction. In Figure 5(a), band alignment with respect to the vacuum level is shown for the III-nitride monolayers (left). As with other III-nitrides, the N -derived states of the alloy valence band maximum (VBM) are localised and thus exhibit low dispersion (see Figure 1(c) where several flat bands are evident for , , ). Materials that would permit work-function-mediated charge transfer – i.e modulation doping – of the monolayers [35, 38, 97] [34] are shown on the right of Figure 5(a), and include transition metal oxides (TMOs) and perovskites.

Figure 5. The band alignment referenced to vacuum level of the valence band maximum (VBM) and conduction band minimum (CBM) energies computed using HSE06. Indirect (direct) bandgap given in units of eV with location in k-space of the conduction (valence) band extrema indicated on each respectively. Materials to engineer interfacial charge-transfer through modulation doping for p- and n-type conduction in the III-nitride monolayers are shown on the right. Proposed p-type modulation dopants include transition metal oxides (TMOs): VO [122, 123], CrO [123, 124], MoO [123, 125], WO [123, 126], as well as the metal AgF [127]; while n-type dopants include perovskites: BaMoO [127, 128], CaMoO [128], LaFeO [128, 129], SrMoO [128], SrVO [128, 129, 130], LaMnO [128, 129], LaRuO [129], alkali-metal-adsorbed hexagonal GaN monolayers: hGaN : Na, K, Cs [131], and hexaborides LaB/hBN [132], LaBaB [133].
Traditional methods of doping such as substitutional doping [98], plasma treatment[99], and ion implantation [100], introduce charge scattering centers which degrade thermoelectric performance. Modulation-doping [97, 101, 102, 103, 104, 105] [34] can reduce ionized-impurity scattering by spatially separating dopants from the conducting channel, as shown in Figure 4(b). This separation offers a route to increasing carrier concentration while preserving carrier mobility, with benefits for electronic and thermoelectric transport [106]. Modulation doping has been extensively employed in thin-films, III-V heterostructures, and 2D semiconductors, such as to improve silicon-germanium alloy electrical conductivity by 54% [63], and thermoelectric power factor by 40% [107]. Improvements in ZT have also been demonstrated: Cu-intercalation-mediated modulation doping in n-type yielded at 375 K, approximately 75% above the non-intercalated CuI-doped reference at the same temperature [38]. These results motivate the material-specific p- and n-type modulation-doping partners proposed here.

Figure 4. (a) Band diagram depicting p-type modulation doping of a 2D III-nitride layer using either another semiconductor that has a broken-gap (i.e type III) band alignment with or a metal with work function . The electronic bands are modified by the difference in work function between the two materials in the heterostructure and results in transfer of electrons from the 2D III-nitride to the CT material, which hole dopes the 2D III-nitride. (b) Comparison between ’standard’ substitutional doping in which impurities are introduced into the TE or channel material and modulation doping in which an encapsulation layer donates (accepts) electrons to (from) the pristine TE or channel material. [38]. In two-dimensional materials, scattering events from long-range Coulomb interactions are greatly amplified due to reduced dielectric screening in the out-of-plane direction. This effect can significantly degrade carrier transport and make substitutional doping problematic for atomically thin conduction channels. [121].
For p-type doping, materials with work function deeper than the valence band maximum of the III-nitride monolayers (AgF and TMOs) are proposed, which make electron transfer from the valence band of the III-nitrides to the conduction band of the adjacent material favorable and can result in efficient hole doping (see schematic of p-type modulation doping in Figure 4(a)). Recent works have shown massive interfacial charge transfer in graphene/-RuCl [108, 109, 110, 111, 112, 113, 114, 115, 116, 117] and WSe/-RuCl [102, 118, 119, 120] heterostructures with this method.
Conversely, for n-type doping, materials with work function closer to vacuum level than the III-nitride CBM are proposed, which enable electron transfer from the adjacent material to the III-nitride conduction band. The array of proposed p- and n-type modulation dopants in Figure 5 is a distinct contribution of this work: it identifies candidate charge-transfer partners for accessing both hole and electron conduction in the III-nitride monolayers. Figure 6(a) illustrates the p-type route with calculated charge transfer between hGaN and VO, giving a simulated hole surface charge density of . Favorable band alignment is used to select the broader candidate set; the explicit hGaN/VO example demonstrates interfacial charge transfer. The literature enhancement in motivates this strategy, while the values reported here are those of the stated monolayer transport model.

Figure 6. Charge-density difference isosurface of hGaN with (a) proposed p-type modulation dopant V₂O₅ and (b) hBN, which demonstrates lack of charge transfer between materials without necessary band alignment. The planar-averaged charge-density difference is shown on the left hand side of each subfigure with charge depletion (holes) and accumulation (additional electrons) indicated by blue and yellow respectively. For the hGaN/V₂O₅ heterostructure shown in (a) each hGaN primitive cell transfers 0.09 to V₂O₅, resulting in a hGaN hole surface charge density of . In contrast, an order of magnitude less charge (0.003) is transferred between layers with unfavorable type-I band alignment, such as demonstrated by the hGaN/hBN heterostructure.
Electronic properties and transport interpretation
Table I connects the composition-dependent electronic structure in Figure 1 to the transport and charge-transfer properties discussed above. Band curvature sets the carrier velocities and contributes to the density of states, dielectric polarization screens electrostatic perturbations, and the work function describes the Fermi-level position relative to vacuum. The effective optical-phonon frequency provides an energy scale for an additional carrier-scattering channel. These quantities therefore describe different contributions to the electronic response rather than interchangeable measures of thermoelectric performance.
The electron masses increase with boron content over the compositions reported in the table. In the Al-containing series, the values increase from in hAlN to – at , – at , and at . The corresponding Ga-containing values increase from in hGaN to –, –, and –. The similar masses along the two tabulated directions at and indicate relatively weak directional variation of
the local conduction-band curvature. In a parabolic-band picture, , where is the conductivity mass. Thus, the increasing electron mass would lower mobility at a fixed relaxation time. The conductivity comparison between the Al- and Ga-containing alloys cannot, however, be inferred from these masses alone: the different conductivities in Figure S1 of the Supplemental Material [134] also reflect their material-dependent MRTA lifetimes and the full occupied band dispersion.

Figure S1. Temperature-dependent transport properties of and monolayers as a function of (a) n-type and (b) p-type surface carrier concentration. The Seebeck coefficient and electrical conductivity are given as the magnitude of the in-plane tensor elements (i.e. ).
The hole masses show a more pronounced and nonmonotonic composition dependence. At , the reported magnitudes are for BAlN and for BGaN, larger than the corresponding electron masses and the hole masses of hAlN (–) and hGaN (–). These values quantify the flattening of the N--derived valence-band edge associated with puckering and orbital mixing in Figure 1. Within a parabolic-band interpretation, a larger density-of-states mass can increase at a fixed sheet carrier density, while a larger mass along the transport direction reduces carrier velocities. This competition helps interpret the generally larger thermopower and smaller conductivity for p-type than for n-type transport at in Figure S1. At , the reported hole masses of the Al-containing alloy, and , differ by approximately a factor of five, whereas the two reported Ga-containing hole masses are both . Consequently, a uniformly heavy or strongly anisotropic hole band is not a property of every alloy in these series.
The directional contrast in BAlN illustrates why density-of-states and conductivity masses must be distinguished. A heavy direction can supply a large density of states while a lighter direction supports faster transport, a combination that can improve the balance between thermopower and conductivity [135]. The tabulated line curvatures do not establish the complete mass tensor, valley multiplicity, or energy separation between pockets. These factors, and the increase in scattering phase space that can accompany a larger density of states, determine whether anisotropy improves the power factor [136]. In particular, the large hole mass of BGaN pertains to its p-type response; the strong n-type performance in Figure 3 involves its more dispersive conduction band, with masses of 0.50–0.52, together with the calculated electron lifetimes.
The static dielectric constants reveal a different composition dependence. The in-plane values lie between 3.54 and 4.87, while the out-of-plane values are 1.72–2.28 for all entries except BGaN. For this alloy, , exceeding its in-plane value of 4.22 and reversing the dielectric anisotropy of the other tabulated materials. This enhanced perpendicular response suggests stronger screening of slowly varying fields normal to the layer, relevant to remote charges and interfacial charge transfer. Static dielectric screening also enters the piezoelectric-scattering model, together with the piezoelectric and elastic tensors [41]; a larger individual dielectric component alone therefore does not establish a longer lifetime or higher conductivity. The tabulated permittivities include electronic and ionic polarization and use the effective-thickness treatment described in the Supplemental Material [134, 137].
The estimated intrinsic breakdown fields in Table I(a) rise from 6.23 (hAlN) and 4.84 (hGaN) MV/cm to 20.06 and 20.74 MV/cm at , respectively. These B-rich values indicate high predicted intrinsic field tolerance of interest for power electronics. For scale, a literature assessment gives bulk GaN and diamond critical fields of approximately 3.4 and 10 MV/cm, respectively, and identifies the commonly quoted 8 MV/cm for -Ga₂O₃ as an estimate [138]. Differences in dimensionality, field profiles, and methods make these comparisons indicative. The present values are screening estimates from the Kim–Pilanía–Ramprasad model [42], using the band gap and maximum phonon frequency ; their application to monolayers requires validation. The equation and full trend are given in Sec. SV and Figure S3 of the Supplemental Material [134].
Figure S3. Estimated intrinsic dielectric breakdown field of monolayer (circles) and (triangles) as a function of boron fraction , from the numerical outputs of Eq. (25) listed in Table S1. The color scale denotes the maximum phonon frequency in THz after wavenumber conversion. The two series share the hBN endmember at ; its two markers overlap.
The work functions span 5.03–5.40 eV and show that the vacuum-referenced electronic energies are tunable by alloying. Relative to hAlN (5.07 eV), the Al-containing alloys have higher work functions, reaching 5.40 eV at before decreasing to 5.30 eV at . The Ga-containing values increase from 5.03 eV in hGaN to 5.22, 5.25, and 5.34 eV with increasing . Since , these changes describe the Fermi-level offsets relevant to equilibration with a contacting material. For a semiconductor, the work function also depends on Fermi-level placement and does not specify the electron affinity or ionization energy by itself. The vacuum-referenced band edges in Figure 5 are therefore needed to select electron-donating and electron-accepting partners. The work-function trends complement that selection, while the explicit interface calculation in Figure 6 accounts for charge redistribution in the hGaN/V₂O₅ example.
The effective polar-optical-phonon frequencies range from 23.42 to 29.25 THz for the six mixed alloys. They increase with in the Al-containing series, whereas the Ga-containing series has its lowest value at . At a fixed temperature, a higher phonon energy reduces the Bose occupation relevant to absorption and raises the
Low-symmetry point coordinates: : [0.31, 0.35, 0.00]; : [0.66, −0.35, 0.00]; : [0.04, 0.04, 0.00].
carrier-energy threshold for emission. This motivates examining reduced polar-optical scattering, but the frequency alone cannot establish that this mechanism is negligible. Its strength also depends on the coupling matrix elements, electronic and ionic screening, and the dielectric environment [139, 140]. The reported transport values retain the ADP-plus-PIE approximation in Sec. III A. Moreover, a weighted zone-center optical frequency does not determine lattice thermal conductivity; no composition-dependent is inferred from this column of Table I.
Together, these results explain how alloying changes the ingredients of the thermoelectric response. The density of states, carrier velocities, and screened scattering jointly determine the energy-dependent transport distribution, and hence , , and . Reduced dimensionality further changes the density-of-states profile, providing the quantum-confinement context illustrated in Figure S2 of the Supplemental Material [134, 141, 142, 143]. The sensitivity of thermopower to these ingredients can be illustrated by the Mott relation in the degenerate limit [144],

Figure S2. Qualitative schematics of (a) the effect of quantum confinement on the density of states and band structure of semiconductor materials and (b) the dependence of the thermoelectric figure of merit , Seebeck coefficient , electrical conductivity , and thermal conductivity on carrier concentration.
where and is the transport distribution defined in Sec. III A. Its logarithmic derivative contains both band-dispersion and scattering contributions. The finite-temperature coefficients are obtained from the full transport moments, which retain the state-dependent lifetimes. Accordingly, the large values in Figure 2 and Figure 3 reflect a balance between the power factor and electronic heat conduction, evaluated with the common lattice-conductivity reference. Table I provides a physical interpretation of that balance, while the full transport calculation is needed to establish the ranking of compositions and carrier types.
Computational methods
Carrier scattering and electronic transport
Electronic structure and semiclassical transport were investigated using density functional theory, BoltzTraP2 [145], and AMSET [41]. Carrier scattering was treated within the momentum relaxation-time approximation (MRTA), with a momentum relaxation time evaluated for each electronic band and wave vector at temperature and chemical potential . The MRTA accounts for the change in carrier velocity associated with a scattering event when evaluating momentum relaxation. The state-dependent lifetimes were retained in the transport calculation rather than replaced by a common constant.
The scattering model includes acoustic deformation-potential (ADP) and piezoelectric (PIE) scattering. ADP scattering describes the coupling between carriers and the band-energy shifts induced by acoustic lattice deformations. PIE scattering arises from the electric fields generated by piezoelectric polarization accompanying these deformations. Scattering probabilities were evaluated within the Born approximation using AMSET, with first-principles deformation potentials, elastic properties, piezoelectric coefficients, and dielectric response as material-specific inputs [41]. The contributions of the included mechanisms were combined for each electronic state according to Matthiessen’s rule,
where is the momentum relaxation rate associated with scattering mechanism .
Only in-plane electronic transport is considered. Let denote the volumetric transport distribution along an in-plane principal direction . It includes the band velocities, electronic-state weights, and state-dependent MRTA lifetimes. With the Fermi–Dirac distribution, the transport moments are
For electron charge , with , the principal-direction transport coefficients are
Here, is the electronic thermal conductivity at zero electrical current. A lifetime common to all electronic states factors out of every moment and cancels from . State-dependent lifetimes generally do not cancel; the present calculation retains the AMSET MRTA lifetimes in the transport response [41, 145]. The temperature-dependent averages in Figure 7 summarize these lifetimes and are not substituted for them in the transport integrals.

Figure 7. Temperature-dependent average MRTA relaxation times. (a) Total averages for and . (b),(c) Acoustic deformation-potential (ADP), piezoelectric (PIE), and total averages for the Al- and Ga-containing alloys, respectively.
Calculations were performed at multiple temperatures and carrier concentrations within the rigid-band approximation: the electronic bands are held fixed while occupations change with temperature and chemical potential. The relation between volumetric and sheet transport quantities is defined in Sec. III D. Additional computational parameters are given in the Supplemental Material [134].
Polar-optical-phonon scattering was omitted from the reported calculations. This approximation was motivated by the high effective optical-phonon frequencies calculated for the mixed nitride monolayers, approximately 20–30 THz. Higher phonon energies reduce thermal phonon populations, while the magnitude of polar-optical scattering also depends on coupling strength and the availability of absorption and emission processes. In two-dimensional materials, the Fröhlich interaction depends on dielectric screening and the surrounding environment [139, 140]. The resulting transport predictions therefore apply to the stated ADP-plus-PIE scattering model.
Carrier scattering and electronic transport
Effective polar-optical-phonon frequency
The effective polar phonon frequency was determined from the phonon frequencies and eigenvectors generated by DFPT in VASP. Each -point phonon mode from the DFPT run was weighted by the associated dipole moment as
where is the Born effective charge, and then averaged over the unit sphere to capture dipole anisotropy and polar divergence as . The effective phonon frequency is then given by the weighted sum of -point phonon modes
Lattice heat transport: literature-based reference
We represent lattice heat transport using the common reference , held fixed across all compositions and sampled temperatures. This literature-motivated reference replaces an alloy-specific phonon-transport calculation. It is motivated by the high room-temperature in-plane conductivity discussed for bulk hBN [146]. For comparison, first-principles calculations report 14.93 W m K for planar monolayer GaN [147] and 306.5 W m K for unstrained monolayer AlN at 300 K [148]. Within a common normalization, using the higher reference gives smaller estimates than using either of these lower lattice conductivities with the same electronic transport coefficients.
For fixed , , and , the reference estimate is a lower bound with respect to the lattice contribution whenever
This comparison requires consistent normalization of all conductivities. Alloying and chemical ordering can modify phonon transport nonmonotonically [149]. The cited room-temperature values therefore motivate a high-conductivity reference scenario without establishing an upper bound for every alloy and temperature. Both the low-temperature results and the wider temperature survey use this same fixed lattice reference. The electronic transport coefficients and carrier relaxation times retain their calculated temperature and carrier-concentration dependence.
Two-dimensional normalization
For a monolayer, we distinguish the periodic supercell height normal to the layer from the effective material thickness . Following the relaxed-bilayer thickness convention of Laturia et al. [137], is defined by the interlayer separation in the relaxed bilayer. For the six mixed alloys (, 0.5, and 0.75 in the Al- and Ga-containing series), this value is 7 Å. For in-plane cell area , supercell-normalized coefficients refer to the volume , whereas effective volumetric coefficients refer to . The corresponding sheet electrical and electronic thermal conductances are
Thus, conversion from the supercell convention to an effective material thickness multiplies both electronic conductivities by . The Seebeck coefficient is unchanged by this common rescaling. Sheet conductances provide a thickness-independent description of monolayer transport [150]; has units of S and has units of W K.
The sheet carrier concentration is
where and are carrier concentrations normalized to and , respectively. Length and concentration units must be compatible; for example, in cm and in cm give in cm.
Defining the lattice sheet thermal conductance by , the principal-direction figure of merit is
For consistency with electronic coefficients expressed using thickness , the lattice reference must be represented by . Equation 14 is invariant under a change in when the physical sheet conductances are held fixed and all volumetric conductivities are rescaled consistently. Holding the numerical value 550 W m K fixed while changing instead changes the reference sheet heat conductance and defines a different model.
Summary and conclusions
We have investigated the electronic structure, thermoelectric response, and estimated intrinsic breakdown fields of ordered monolayer and . Composition-induced puckering changes the band-edge dispersion and anisotropy. The band-curvature masses, dielectric response, work functions, and optical-phonon descriptors connect these changes to carrier transport and interfacial charge transfer. The p- and n-type responses reflect their respective valence and conduction bands together with state-dependent carrier lifetimes.
Within the acoustic deformation-potential plus piezoelectric scattering model, selected p- and n-type monolayers have predicted maximum ; n-type reaches at 300 K. These results motivate further investigation of low-temperature thermoelectric applications. All estimates use the same temperature-independent lattice reference, 550 W m K, rather than alloy-specific phonon transport. With electronic coefficients held fixed and conductivities normalized consistently, this gives a lower bound with respect to the lattice contribution only when the actual lattice conductivity does not exceed the reference. The omission of polar-optical scattering and other unmodeled mechanisms remains a limitation of the quantitative transport predictions.
Vacuum-referenced band alignments identify candidate modulation-doping partners for both carrier types. The explicit hGaN/ calculation yields a hole surface density of approximately cm, illustrating the p-type charge-transfer route. Transport in these heterostructures remains to be evaluated to determine how the spatial separation of dopants and carriers affects attainable thermoelectric performance.
The Kim–Pilania–Ramprasad estimates increase from 6.23 and 4.84 MV/cm in hAlN and hGaN to 20.06 and 20.74 MV/cm in their respective alloys. These high predicted intrinsic fields motivate investigation of the B-rich alloys for power electronics alongside their thermoelectric response. Their assessment uses the band gap and maximum phonon frequency, distinct from the effective polar-optical-phonon frequency. Validation for monolayers requires attention to field direction, contacts, defects, and device geometry; the scalar estimates do not establish a device breakdown voltage.
The tellurium-free chemistry and preliminary constituent-cost comparison provide additional motivation, while fabrication costs and alloy-specific biocompatibility remain to be established. Alloy-resolved lattice heat transport, additional carrier-scattering mechanisms, doped-interface transport, and direction-resolved breakdown studies are needed to assess the experimentally attainable thermoelectric and high-field performance.
References
Supplemental Material for “Thermoelectric properties and modulation doping of ultrawide-bandgap nitride monolayers”
Ariel R. Barr
supplement.pdf
Supplemental Material for “Thermoelectric properties and modulation doping of ultrawide-bandgap nitride monolayers”
Computational methods
The Boltzmann transport coefficients were computed within the momentum relaxation-time approximation (MRTA), using state-dependent carrier scattering rates calculated with AMSET. BoltzTraP2 (BT2) [1] and AMSET [2] were used for the electronic-structure-based transport analysis. In this framework, relaxation times are evaluated for individual band and wave-vector states rather than imposed as a common constant. Uniform sampling of the monolayer electronic bands was performed using VASP [3] on a -centered grid with vacuum, and interpolated using BT2 under a rigid band approximation (i.e. the bands do not change with doping or temperature) with k-point sampling 30 times more dense than the ab initio mesh, which was approximately equivalent to a k-point mesh of .
For each band and wave vector , AMSET yields a momentum relaxation time . The transport distribution retains these state-dependent lifetimes together with band velocities and electronic-state weights. The main article defines the transport moments and the resulting , , and zero-current electronic thermal conductivity . A lifetime common to all states cancels from the Seebeck coefficient; state-dependent lifetimes generally do not [2, 4]. The temperature-dependent average lifetimes shown in the main article summarize the MRTA calculation and are not used as common lifetimes in place of the state-resolved values. Only in-plane transport is considered.
State- and temperature-dependent scattering rates were evaluated within the Born approximation using AMSET [2]. Inputs include contraction- and expansion-averaged acoustic deformation potentials, elastic properties, and the piezoelectric tensor obtained using density functional perturbation theory (DFPT) in VASP. The deformation-potential tensor describes band-energy shifts under homogeneous strain,
where is the strain tensor.
For the included ADP and PIE mechanisms, the state-dependent momentum relaxation rates are combined using Matthiessen’s rule,
Here, denotes the momentum relaxation rate for mechanism [4] [5, 6]. Polar-optical-phonon scattering was omitted from the reported transport calculation, as described in the main article. The effective optical-phonon frequencies are approximately 20–30 THz for the mixed alloys, compared with approximately 8 THz in GaAs. Related studies provide context for the high-frequency argument [7, 9] [8].
The two-dimensional Fröhlich interaction also depends on dimensionality, dielectric screening, and the surrounding environment [11] [10]. The omission of polar-optical scattering is therefore an approximation of the present transport model, whose applicability depends on these additional coupling and screening effects.
The effective polar phonon frequency was determined from the phonon frequencies and eigenvectors generated by DFPT in VASP. Each -point phonon mode from the DFPT run was weighted by the associated dipole moment as
where is the Born effective charge, and then averaged over the unit sphere to capture dipole anisotropy and polar divergence as . The effective phonon frequency is then given by the weighted sum of -point phonon modes
The static dielectric response was calculated using the method of Laturia et al. [12], with 20 Å vacuum in the out-of-plane direction. Both electronic and ionic contributions are included in the static dielectric constants. The supercell dielectric response contains a vacuum contribution, which is removed using the out-of-plane and in-plane rescaling relations given as Eqs. (1) and (2) in Ref. [12]:
Here, is the complete supercell height, including the vacuum, and is the effective monolayer thickness. Following the relaxed-bilayer convention, is obtained from the interlayer separation in the relaxed bilayer, rather than the vertical extent of one puckered monolayer. The reported value is 7 Å for each of the six mixed alloys (, with and , 0.5, 0.75). The in-plane dielectric response is reported as an in-plane average. These equations rescale dielectric constants; the conversion of electronic transport coefficients is defined separately in Sec. SIII.
Lattice heat transport: literature-based reference
The estimates use the common lattice reference at every sampled composition and temperature. Its literature rationale and conditional lower-bound interpretation are described in the main article: bulk hBN motivates the high reference [13], with lower published monolayer GaN and AlN values providing context [15] [14]. No alloy-specific lattice-conductivity curve was calculated. For a fixed common normalization, the reference figure of merit is
The fixed lattice reference is separate from the temperature-dependent MRTA lifetimes and electronic coefficients. At fixed , , and , it gives a lower bound with respect to the lattice contribution only when the actual does not exceed the reference in the same normalization. The room-temperature literature values do not establish this inequality for every alloy and temperature; alloying and ordering can change lattice heat transport nonmonotonically [16].
Two-dimensional transport normalization
For supercell height and effective material thickness , the normalization definitions in the main article give
Here, SC denotes quantities normalized to the complete periodic supercell volume. The height includes the material and vacuum; it is not the vacuum width alone. For in , must be expressed in cm to obtain in . The Seebeck coefficient requires no thickness rescaling.
The corresponding sheet conductances are and . The lattice reference must use the same thickness convention, , when combined with the electronic terms in [17]. Thickness invariance requires the physical sheet conductances to remain fixed. Changing while keeping the numerical lattice conductivity fixed at 550 W m K instead changes the reference model. The effective thickness is defined from the relaxed-bilayer interlayer separation as stated above; the complete supercell height remains a separate quantity.
Quantum-confinement schematic
Figure S2 provides qualitative context for the dimensionality and carrier-concentration effects discussed in the main article.
Estimated intrinsic dielectric breakdown field
The intrinsic dielectric breakdown fields of the monolayers were estimated using the phenomenological model of Kim, Pilania, and Ramprasad [18], which combines the electronic band gap and the maximum phonon frequency :
This is Eq. (3) of Ref. [18], converted from to ; the prefactor in the original unit convention is 24.442. Here, is the ordinary phonon cutoff frequency, corresponding to the quantity denoted in that reference. Its numerical value is entered in THz without a factor of . It is distinct from the dipole-weighted effective polar-optical-phonon frequency defined in Sec. SI. The band gap and maximum frequency describe electronic excitation and phonon-mediated energy relaxation, respectively. The supplied maximum-frequency data are interpreted as wavenumbers in and converted using . Table S1 lists the wavenumbers, converted frequencies, and breakdown fields to two decimal places, matching the field precision in main-article Table I(a).
| Material | (cm) | (THz) | Estimated (MV/cm) | |
| 0 | 862.01 | 25.84 | 6.23 | |
| 0.25 | 1026.86 | 30.78 | 8.90 | |
| 0.5 | 1231.13 | 36.91 | 15.95 | |
| 0.75 | 1255.40 | 37.64 | 20.06 | |
| 0 | 788.99 | 23.65 | 4.84 | |
| 0.25 | 1086.10 | 32.56 | 10.99 | |
| 0.5 | 1259.42 | 37.76 | 12.98 | |
| 0.75 | 1250.36 | 37.48 | 20.74 | |
| 1 | 1459.14 | 43.74 | 36.40 |
Table S1. Maximum phonon wavenumbers, corresponding frequencies, and estimated intrinsic dielectric breakdown fields, rounded to two decimal places.
Figure S3 shows an increase in the estimated breakdown field with boron content, from 6.23 and 4.84 for hAlN and hGaN, respectively, to 36.40 for the common hBN endpoint. The maximum phonon frequency likewise rises overall toward the B-rich end. Within Eq. (25), larger values of either or increase . The relative ordering of the two series changes at intermediate compositions. At the Ga-containing alloy has the higher cutoff frequency (37.76 versus 36.91 THz) but the lower estimated field (12.98 versus 15.95 ). Thus, the cutoff frequency alone is insufficient to rank the estimated breakdown fields.
The B-rich alloys therefore exhibit high predicted intrinsic breakdown fields. For context, Slobodyan et al. [19] assessed bulk-diode critical fields at a common drift-region doping of , obtaining approximately 3.4 for GaN and 10 for diamond; the widely quoted 8 for -GaO was identified as an estimate. These comparisons provide a scale for the present predictions; differences in dimensionality, field profiles, and methods preclude a direct ranking of device performance.
The Kim model was fitted to first-principles intrinsic-breakdown data for bulk crystalline insulators [18]. Its transfer to these monolayers requires validation, and its scalar form does not resolve field direction. The hBN endpoint illustrates this limitation: the present model gives 36.40 MV/cm, whereas Hattori et al. measured approximately 3 MV/cm in-plane () and 12 MV/cm out-of-plane () in single-crystal hBN under controlled humidity [20]. Those bulk-crystal measurements and the monolayer estimate describe different geometries. Direction-resolved calculations or measurements are needed to establish attainable fields; defects, contacts, and the device environment also affect experimental breakdown. The present results thus support further investigation of high-field applications. These estimates complement the main article’s thermoelectric results but do not enter or replace the state-dependent ADP-plus-PIE transport model.
References
References
- [2]Y. Zhang, F. Udrea, and H. Wang, Nature electronics 5, 723 (2022).
- [4]H. Chandrasekar, T. Razzaq, C. Wang, Z. Reyes, K. Majumdar, and S. Rajan, Advanced Electronic Materials 6, 2000007 (2020).
- [7]H. Yoshida, Y. Yamashita, M. Kuwabara, and H. Kan, Applied physics letters 93 (2008).
- [9]C. M. Gilardoni, S. E. Barker, C. L. Curtin, S. A. Fraser, O. Powell, D. K. Lewis, X. Deng, A. J. Ramsay, C. Li, I. Aharonovich, et al., arXiv preprint arXiv:2408.10348 (2024).
- [11]D. Tan, IEEE Power Electronics Magazine 2, 38 (2015).
- [13]M. H. Wong, O. Bierwagen, R. J. Kaplar, and H. Umezawa, Journal of Materials Research 36, 4601 (2021).
- [14][14–16]
- [15]G. Schierning, Nature Energy 3, 92 (2018).
- [16]M. Zebarjadi, K. Esfarjani, M. Dresselhaus, Z. Ren, and G. Chen, Energy & Environmental Science 5, 5147 (2012).
- [17][17–19]
- [18]C. Gayner and K. K. Kar, Progress in Materials Science 83, 330 (2016).
- [19]B. Poudel, Q. Hao, Y. Ma, Y. Lan, A. Minnich, B. Yu, X. Yan, D. Wang, A. Muto, D. Vashaee, et al., science 320, 634 (2008).
- [21][21–26]
- [28]Aldrich, S. (2025). Bismuth(III) telluride, powder, -325 mesh, 99.99% trace metals basis.
- [29][29–31]
- [35]R. Arora, A. R. Barr, D. T. Larson, M. Pizzochero, and E. Kaxiras, Phys. Rev. Materials 9, L021601 (2025).
- [36][36–39]
- [38]C.-L. Chen, T.-H. Wang, Z.-G. Yu, Y. Hutabalian, R. K. Vankayala, C.-C. Chen, W.-P. Hsieh, H.-T. Jeng, D.-H. Wei, and Y.-Y. Chen, Advanced Science 9, 2201353 (2022).
- [40]D. Wines, F. Ersan, and C. Ataca, ACS Applied Materials & Interfaces 12, 46416 (2020).
- [41]A. M. Ganose, J. Park, A. Faghaninia, R. Woods-Robinson, K. A. Persson, and A. Jain, Nat. Commun. 12, 2222 (2021).
- [42]C. Kim, G. Pilania, and R. Ramprasad, Chemistry of Materials 28, 1304 (2016).
- [43]V.-T. Tran, J. Saint-Martin, P. Dollfus, and S. Volz, Scientific reports 7, 2313 (2017).
- [44]H. Sevin¸cli, C. Sevik, T. C¸ a˘gın, and G. Cuniberti, Scientific reports 3, 1228 (2013).
- [45]P. Dollfus, V. H. Nguyen, and J. Saint-Martin, Journal of Physics: Condensed Matter 27, 133204 (2015).
- [46]D. Li, Y. Gong, Y. Chen, J. Lin, Q. Khan, Y. Zhang, Y. Li, H. Zhang, and H. Xie, Nano-Micro Letters 12, 1 (2020).
- [47]D. I. Bilc, D. Benea, V. Pop, P. Ghosez, and M. J. Verstraete, The Journal of Physical Chemistry C 125, 27084 (2021).
- [48]W. Huang, H. Da, and G. Liang, Journal of Applied Physics 113 (2013).
- [49]S. Kumar and U. Schwingenschlogl, Chemistry of Materials 27, 1278 (2015).
- [50]R. Liu, H. Lü, X. Zu, and Y. Zhang, Journal of Applied Physics 135 (2024).
- [51]D. Wickramaratne, F. Zahid, and R. K. Lake, The Journal of chemical physics 140 (2014).
- [52]J. Zhang, Y. Xie, Y. Hu, and H. Shao, Applied Surface Science 532, 147387 (2020).
- [53]B. Marfoua and J. Hong, ACS applied materials & interfaces 11, 38819 (2019).
- [54]R. D’Souza and S. Mukherjee, Journal of Applied Physics 124 (2018).
- [55]Y. Cheng, J. Yang, Q. Jiang, D. He, J. He, Y. Luo, D. Zhang, Z. Zhou, Y. Ren, and J. Xin, Journal of Materials Chemistry A 5, 5163 (2017).
- [56]S. Yamaguchi, T. Matsumoto, J. Yamazaki, N. Kaiwa, and A. Yamamoto, Applied Physics Letters 87 (2005).
- [57]J. Zhang, L. Song, H. Pedersen, H. Yin, L. T. Hung, and B. B. Iversen, Nature communications 8, 13901 (2017).
- [58]H. Liu, X. Shi, F. Xu, L. Zhang, W. Zhang, L. Chen, Q. Li, C. Uher, T. Day, and G. J. Snyder, Nature materials 11, 422 (2012).
- [59]H. Liu, X. Yuan, P. Lu, X. Shi, F. Xu, Y. He, Y. Tang, S. Bai, W. Zhang, L. Chen, et al., Advanced Materials 25, 6607 (2013).
- [60]B. Zhong, Y. Zhang, W. Li, Z. Chen, J. Cui, W. Li, Y. Xie, Q. Hao, and Q. He, Applied Physics Letters 105 (2014).
- [61]A. A. Olvera, N. A. Moroz, P. Sahoo, P. Ren, T. P. Bai-Leya, A. A. Page, C. Uher, and P. F. P. Poudeu, Energy Environ. Sci. 10, 1668 (2017).
- [62]X. Wang, H. Lee, Y. Lan, G. Zhu, G. Joshi, D. Wang, J. Yang, A. Muto, M. Tang, J. Klatsky, et al., Applied Physics Letters 93 (2008).
- [63]B. Yu, M. Zebarjadi, H. Wang, K. Lukas, H. Wang, D. Wang, C. Opeil, M. Dresselhaus, G. Chen, and Z. Ren, Nano letters 12, 2077 (2012).
- [64]S. Bathula, M. Jayasimhadri, N. Singh, A. Srivastava, J. Pulikkotil, A. Dhar, and R. Budhani, Applied Physics Letters 101 (2012).
- [65]L.-D. Zhao, S. Hao, S.-H. Lo, C.-I. Wu, X. Zhou, Y. Lee, H. Li, K. Biswas, T. P. Hogan, C. Uher, et al., Journal of the American Chemical Society 135, 7364 (2013).
- [66]H. Wang, Y. Pei, A. D. LaLonde, and G. J. Snyder, Proceedings of the National Academy of Sciences 109, 9705 (2012).
- [67]T. Plirdpring, K. Kurosaki, A. Kosuga, T. Day, S. Firdosy, V. Ravi, G. J. Snyder, A. Harnwunggmoung, T. Sugahara, Y. Ohishi, et al., Advanced Materials (Deerfield Beach, Fla.) 24, 3622 (2012).
- [68]Y. Luo, J. Yang, Q. Jiang, W. Li, D. Zhang, Z. Zhou, Y. Cheng, Y. Ren, and X. He, Advanced Energy Materials 6, 1600007 (2016).
- [69]J. Zhang, R. Liu, N. Cheng, Y. Zhang, J. Yang, C. Uher, X. Shi, L. Chen, and W. Zhang, Advanced Materials 26, 3848 (2014).
- [70]L.-D. Zhao, S.-H. Lo, J. He, H. Li, K. Biswas, J. Androulakis, C.-I. Wu, T. P. Hogan, D.-Y. Chung, V. P. Dravid, et al., Journal of the American Chemical Society 133, 20476 (2011).
- [71]H. Wang, E. Schechtel, Y. Pei, and G. J. Snyder, Advanced Energy Materials 3, 488 (2013).
- [72]M. Ibáñez, Z. Luo, A. Genc, L. Piveteau, S. Ortega, D. Cadavid, O. Dobrozhan, Y. Liu, M. Nachtegaal, M. Zebarjadi, et al., Nature communications 7, 10766 (2016).
- [73]B. Kozinsky and D. J. Singh, Annual Review of Mate- rials Research 51, 565 (2021).
- [74]G. Samsonidze and B. Kozinsky, Advanced Energy Materials 8, 1800246 (2018).
- [75]G. Joshi, X. Yan, H. Wang, W. Liu, G. Chen, and Z. Ren, Advanced Energy Materials 1, 643 (2011).
- [76]X. Yan, W. Liu, S. Chen, H. Wang, Q. Zhang, G. Chen, and Z. Ren, Advanced Energy Materials 3, 1195 (2013).
- [77]C. Fu, S. Bai, Y. Liu, Y. Tang, L. Chen, X. Zhao, and T. Zhu, Nature communications 6, 8144 (2015).
- [78]W. Xie, X. Tang, Y. Yan, Q. Zhang, and T. M. Tritt, Applied Physics Letters 94 (2009).
- [79]S. I. Kim, K. H. Lee, H. A. Mun, H. S. Kim, S. W. Hwang, J. W. Roh, D. J. Yang, W. H. Shin, X. S. Li, Y. H. Lee, et al., Science 348, 109 (2015).
- [80]B. Sales, D. Mandrus, and R. K. Williams, Science 272, 1325 (1996).
- [81]X. Shi, J. Yang, J. R. Salvador, M. Chi, J. Y. Cho, H. Wang, S. Bai, J. Yang, W. Zhang, and L. Chen, Journal of the American Chemical Society 133, 7837 (2011).
- [82]G. Rogl, A. Grytsiv, P. Rogl, N. Peranio, E. Bauer, M. Zehetbauer, and O. Eibl, Acta Materialia 63, 30 (2014).
- [83]L.-D. Zhao, S.-H. Lo, Y. Zhang, H. Sun, G. Tan, C. Uher, C. Wolverton, V. P. Dravid, and M. G. Kanatzidis, nature 508, 373 (2014).
- [84]L.-D. Zhao, G. Tan, S. Hao, J. He, Y. Pei, H. Chi, H. Wang, S. Gong, H. Xu, V. P. Dravid, et al., Science 351, 141 (2016).
- [85]K. Peng, X. Lu, H. Zhan, S. Hui, X. Tang, G. Wang, J. Dai, C. Uher, G. Wang, and X. Zhou, Energy & Environmental Science 9, 454 (2016).
- [86]A. T. Duong, V. Q. Nguyen, G. Duvjir, V. T. Duong, S. Kwon, J. Y. Song, J. K. Lee, J. E. Lee, S. Park, T. Min, et al., Nature communications 7, 13713 (2016).
- [87]B. P. Bahuguna, L. Saini, R. O. Sharma, and B. Tiwari, Physical Chemistry Chemical Physics 20, 28575 (2018).
- [88]J. Mao, H. Zhu, Z. Ding, Z. Liu, G. A. Gamage, G. Chen, and Z. Ren, Science 365, 495 (2019).
- [89]J. Mao, G. Chen, and Z. Ren, Nature Materials 20, 454 (2021).
- [90]B. Lenoir, M. Cassart, J.-P. Michenaud, H. Scherrer, and S. Scherrer, Journal of Physics and Chemistry of Solids 57, 89 (1996).
- [91]G. Mahan, B. Sales, and J. Sharp, Physics Today 50, 42 (1997).
- [92]D.-Y. Chung, T. Hogan, P. Brazis, M. Rocci-Lane, C. Kannewurf, M. Bastea, C. Uher, and M. G. Kanatzidis, Science 287, 1024 (2000).
- [93]Y. Pan, M. Yao, X. Hong, Y. Zhu, F. Fan, K. Imasato, Y. He, C. Hess, J. Fink, J. Yang, et al., Energy & Environmental Science 13, 1717 (2020).
- [94]ACS. (2020). Molecule of the Week: Molybdenum disulfide.
- [95]L. Weston, D. Wickramaratne, M. Mackoit, A. Alkauskas, and C. G. Van de Walle, Physical Review B 97, 214104 (2018).
- [96]A. Zunger, Applied Physics Letters 83, 57 (2003).
- [97]T. Zheng, E. Low, N. Rafizadeh, K. S. Burch, and H. Zhao, Nano Letters 25, 4054 (2025).
- [98]J. Suh, T.-E. Park, D.-Y. Lin, D. Fu, J. Park, H. J. Jung, Y. Chen, C. Ko, C. Jang, Y. Sun, et al., Nano letters 14, 6976 (2014).
- [99]M. Chen, H. Nam, S. Wi, L. Ji, X. Ren, L. Bian, S. Lu, and X. Liang, Applied Physics Letters 103 (2013).
- [100]A. Nipane, D. Karmakar, N. Kaushik, S. Karande, and S. Lodha, ACS nano 10, 2128 (2016).
- [101]S. Bai, D. Wan, Y. Tian, C. Chang, H. Xie, C. Zhou, I. Chung, J. He, P. Kang, and L.-D. Zhao, Nature Reviews Materials, 1 (2026).
- [102]J. Xie, Z. Zhang, H. Zhang, V. Nagarajan, W. Zhao, H.-L. Kim, C. Sanborn, R. Qi, S. Chen, S. Kahn, et al., Nano Letters 24, 5937 (2024).
- [103]R. Dingle, H. Störmer, A. Gossard, and W. Wiegmann, Applied Physics Letters 33, 665 (1978).
- [104]H. Daembkes et al., Modulation-doped field-effect transistors: principles, design, and technology (IEEE Press, 1991).
- [105]D. Lee, J. J. Lee, Y. S. Kim, Y. H. Kim, J. C. Kim, W. Huh, J. Lee, S. Park, H. Y. Jeong, Y. D. Kim, et al., Nature Electronics 4, 664 (2021).
- [106]Gossard, A. (1985). Molecular beam epitaxy and heterostructures. In Springer (pp. pp. 499–531).
- [107]M. Zebarjadi, G. Joshi, G. Zhu, B. Yu, A. Minnich, Y. Lan, X. Wang, M. Dresselhaus, Z. Ren, and G. Chen, Nano letters 11, 2225 (2011).
- [108]B. Zhou, J. Bagley, P. Lampen-Kelley, J.-Q. Yan, D. G. Mandrus, and E. A. Henriksen, Physical Review B 100, 165426 (2019).
- [109]S. Mashhadi, Y. Kim, J. Kim, D. Weber, T. Taniguchi, K. Watanabe, N. Park, B. Lotsch, J. H. Smet, M. Burghard, et al., Nano letters 19, 4659 (2019).
- [110]T. K. Chau, Y. Choi, K. Hwang, G. Oh, S. Akhtar, J. Kim, K.-Y. Choi, H.-Y. Lee, S. J. Hong, and D. Suh, Advanced Functional Materials 34, 2306964 (2024).
- [111]Y. Wang, J. Bagley, E. Gerber, M. Gray, N. Kumar, X. Lu, J.-Q. Yan, E. Fedorouini, R. Basnet, S. J. Yun, et al., Nano letters 20, 8446 (2020).
- [112]A. Rossi, C. Johnson, J. Bagley, J. C. Thomas, L. Francaviglia, R. Dettori, A. K. Schmidt, K. Watanabe, T. Taniguchi, M. Cothrine, et al., Nano Letters 23, 8000 (2023).
- [113]D. J. Rizzo, B. S. Jessen, F. L. Ruta, J. Zhang, J.-Q. Yan, L. Xian, A. S. McLeod, M. E. Berkowitz, K. Watanabe, et al., Nano letters 20, 8438 (2020).
- [114]R. A. Vitalone, B. S. Jessen, R. Jing, D. J. Rizzo, S. Xu, V. Hsieh, M. Cothrine, D. G. Mandrus, L. Wehmeier, G. L. Carr, et al., ACS nano 18, 29648 (2024).
- [115]J. Bagley, J. Butler, S. Biswas, Z. Ge, S. Lagasse, T. Taniguchi, K. Watanabe, M. Cothrine, D. G. Mandrus, J. Velasco Jr, et al., Nano Letters 22, 4124 (2022).
- [116]Z. Qiu, Y. Han, K. Noori, Z. Chen, M. Kashchenko, L. Lin, T. Olsen, J. Li, H. Fang, P. Lyu, et al., Nature Materials 23, 1055 (2024).
- [117]D. J. Rizzo, S. Shabani, B. S. Jessen, J. Zhang, A. S. McLeod, C. Rubio-Verdú, F. L. Ruta, M. Cothrine, J. Yan, D. G. Mandrus, et al., Nano letters 22, 1946 (2022).
- [118]A. J. Sternbach, R. A. Vitalone, S. Shabani, J. Zhang, T. P. Darlington, S. L. Moore, S. H. Chae, E. Seewald, X. Xu, C. R. Dean, et al., Nano Letters 23, 5070 (2023).
- [119]D. J. Rizzo, J. Zhang, B. S. Jessen, F. L. Ruta, M. Cothrine, J. Yan, D. G. Mandrus, S. E. Nagler, T. Taniguchi, K. Watanabe, et al., Nano Letters 23, 8426 (2023).
- [120]J. Pack, Y. Guo, Z. Liu, B. S. Jessen, L. Holtzman, S. Liu, M. Cothrine, K. Watanabe, T. Taniguchi, D. G. Mandrus, et al., Nature Nanotechnology 19, 948 (2024).
- [121]N. Ma and D. Jena, Physical Review X 4, 011043 (2014).
- [122]S. Laubach, P. C. Schmidt, A. Thißen, F. J. Fernandez-Madrigal, Q.-H. Wu, W. Jaegermann, M. Klemm, and S. Horn, Physical Chemistry Chemical Physics 9, 2564 (2007).
- [123]M. T. Greiner, M. G. Helander, W.-M. Tang, Z.-B. Wang, J. Qiu, and Z.-H. Lu, Nature materials 11, 76 (2012).
- [124]H.-J. Zhai, S. Li, D. A. Dixon, and L.-S. Wang, Journal of the American Chemical Society 130, 5167 (2008).
- [125]M. Kröger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky, and A. Kahn, Organic Electronics 10, 932 (2009).
- [126]J. Meyer, M. Kröger, S. Hamwi, F. Gnam, T. Riedl, W. Kowalsky, and A. Kahn, Applied Physics Letters 96 (2010).
- [127]P. Schindler, E. R. Antoniuk, G. Cheon, Y. Zhu, and E. J. Reed, Advanced Functional Materials 34, 2401764 (2024).
- [128]T. Ma, R. Jacobs, J. Booske, and D. Morgan, Journal of Materials Chemistry C 9, 12778 (2021).
- [129]R. Jacobs, J. Booske, and D. Morgan, Advanced Functional Materials 26, 5471 (2016).
- [130]L. Lin, R. Jacobs, D. Chen, V. Vlahos, O. Lu-Steffes, J. A. Alonso, D. Morgan, and J. Booske, Advanced Functional Materials 32, 2203703 (2022).
- [131]Z. Cui, X. Wang, E. Li, Y. Ding, C. Sun, and M. Sun, Nanoscale research letters 13, 1 (2018).
- [132]H. Yamaguchi, R. Yusa, G. Wang, M. T. Pettes, F. Liu, Y. Tsuda, A. Yoshigoe, T. Abukawa, N. A. Moody, and S. Ogawa, Applied Physics Letters 122 (2023).
- [133]T. Ma, R. Jacobs, J. Booske, and D. Morgan, The Journal of Physical Chemistry C 125, 17400 (2021).
- [134]See the Supplemental Material accompanying this article for computational methods, scattering times, dielectric rescaling, carrier-density-dependent transport coefficients, and intrinsic dielectric breakdown estimates.. (n.d.). See the Supplemental Material accompanying this article for computational methods, scattering times, dielectric rescaling, carrier-density-dependent transport coefficients, and intrinsic dielectric breakdown estimates..
- [135]D. S. Parker, A. F. May, and D. J. Singh, Physical Review Applied 3, 064003 (2015).
- [136]J. Park, Y. Xia, V. Ozoliņš, and A. Jain, npj Computational Materials 7, 43 (2021).
- [137]A. Laturia, M. L. Van de Put, and W. G. Vandenberghe, npj 2D Materials and Applications 2, 6 (2018).
- [138]O. Slobodyan, J. Flicker, J. Dickerson, J. Shoemaker, A. Binder, T. Smith, S. Goodnick, R. Kaplar, and M. Hollis, J. Mater. Res. 37, 849 (2022).
- [139]T. Sohier, M. Calandra, and F. Mauri, Physical Review B 94, 085415 (2016).
- [140]T. Sohier, M. Calandra, and F. Mauri, Physical Review B 96, 159904 (2017).
- [141]L. D. Hicks and M. S. Dresselhaus, Physical Review B 47, 12727 (1993).
- [142]L. D. Hicks and M. S. Dresselhaus, Physical review B 47, 16631 (1993).
- [143]J. Mao, Z. Liu, and Z. Ren, npj Quantum Materials 1, 1 (2016).
- [144]Nye, J. F. (1985). Physical properties of crystals: their representation by tensors and matrices. https://doi.org/Oxford University Press
- [145]G. K. H. Madsen, J. Carrete, and M. J. Verstraete, Comput. Phys. Commun. 231, 140 (2018).
- [146]C. Yuan, J. Li, L. Lindsay, D. Cherns, J. W. Pomeroy, S. Liu, J. H. Edgar, and M. Kuball, Commun. Phys. 2, 43 (2019).
- [147]Z. Qin, G. Qin, X. Zuo, Z. Xiong, and M. Hu, Nanoscale 9, 4295 (2017).
- [148]A. Banerjee, B. K. Das, and K. K. Chattopadhyay, Phys. Chem. Chem. Phys. 24, 16065 (2022).
- [149]H. Wang, D. Wei, J. Duan, Z. Qin, G. Qin, Y. Yao, and M. Hu, NanoTechnology 32, 135401 (2021).
- [150]X. Wu, V. Varshney, J. Lee, Y. Pang, A. K. Roy, and T. Luo, Chemical Physics Letters 669, 233 (2017).
