betaFeedback

WorksPhysics and Astronomy

Unraveling hidden Raman modes in αα-MnTe: Effects of long-range dipole-dipole interactions on phonon bands

Abstract

In this study, we address the seemingly open question of disparity between the experimentally and theoretically observed Raman spectra for the altermagnetic αα-MnTe. We establish that the consideration of long range (LR) dipole-dipole interaction is crucial to accurately estimate the phonon band structure and thereby capturing each of the experimentally observed Raman signatures. The LR correction brings about substantial changes to the high-frequency optical modes and their group-theoretical representations. Specifically, for the highly debated Raman peak around 175 cm−1^{-1}, alongside the Raman silent B1u_{1u} mode, the LR corrections introduce Raman-active longitudinal optical A′^{\prime} and A1_1, and transverse optical A1_1 modes. Overall, the calculated Raman spectra suggest that the intensity of these modes is weak in the pristine system, while certain prescribed distortions can increase their intensity. The Raman peak around 140 cm−1^{-1}, as observed experimentally, is identified to be emerging from a higher-order Raman process due to zone off-center contributions, while the other debated 120 cm−1^{-1} peak has both the zone center and off-center contributions.

Over the last few years the insulating α\alpha-MnTe (henceforth referred as MnTe) has been intensely investigated as it exhibits non-relativistic spin splitting of the bands, a characteristic feature of altermagnetism[1], and finite anomalous Hall conductivity[2, 3]. Alongside the electronic and magnetic structure[5] [4], the experimental Raman characterization, one of the fundamental sample characterization tools, has opened up a new fascinating research problem involving this compound. Broadly, the experimental results are segregated into two categories, one with the presence of a Raman peak at 175 cm−1\mathrm{cm}^{-1}[6, 7, 8, 9] and the others which do not observe such peaks[10, 11, 12].

The MnTe films that are grown on InP (111) substrate, as reported by Zhang et. al.[7] show Raman peaks at 122.5 cm−1\mathrm{cm}^{-1}, 140.6 cm−1\mathrm{cm}^{-1}, 175.7 cm−1\mathrm{cm}^{-1}, 268.8 cm−1\mathrm{cm}^{-1}, and 345.8 cm−1\mathrm{cm}^{-1}. There are past studies that did not report the Raman peak at around 175 cm−1\mathrm{cm}^{-1}. For example, two of the earlier Raman measurements[10, 13] on μ\mum thick MnTe films, carried out at low temperature (<30< 30 K) do not exhibit excitation at 175 cm−1\mathrm{cm}^{-1}. However, these studies report Raman excitation in the range 195 - 215 cm−1\mathrm{cm}^{-1}. Another recent experimental work reveals that the MnTe films grown on GaAs (111) substrate show Raman peaks at 90 cm−1\mathrm{cm}^{-1}, 120 cm−1\mathrm{cm}^{-1}, and 140 cm−1\mathrm{cm}^{-1}[11]. Interestingly, the air exposure samples in this study did not exhibit the 175 cm−1\mathrm{cm}^{-1} peak, even though one would expect oxidation and Te excess surface in them. In a recent experimental study on MBE grown MnTe film on Si(111) substrate, we noticed mixed observations. The pristine spot showed Raman intensity peaks around 90-105 cm−1\mathrm{cm}^{-1}, 120 cm−1\mathrm{cm}^{-1}, and 140 cm−1\mathrm{cm}^{-1}. The spot burned with laser irradiation showed these peaks as well as the peak around 175 cm−1\mathrm{cm}^{-1}[12]. We attribute the latter to the signature of distorted MnTe2_2 which has a well established Ag_g peak at 175 cm−1\mathrm{cm}^{-1}[14].

In the case of CVD grown bulk MnTe single crystals, polarization dependent Raman measurements show peaks at 120 cm−1\mathrm{cm}^{-1}, 140 cm−1\mathrm{cm}^{-1}, 175 cm−1\mathrm{cm}^{-1}, and 270 cm−1\mathrm{cm}^{-1}[8]. In the pristine system, the Γ\Gamma-centered optical phonon mode, B1u_{1u}, around 175 cm−1\mathrm{cm}^{-1} is Raman silent. The authors propose that a probable distortion, reducing the symmetry of the D6h_{6h} to D3h_{3h} point group, converts the B1u_{1u} mode to A1′'_1 and makes it Raman active. Similarly, another study[9] on bulk single crystal MnTe also reports the Raman peak at 175 cm−1\mathrm{cm}^{-1}. However, in the absence of any symmetry lowering, the authors here assigned the peak to the possible presence of residual MnTe2_2. Differing with the above, another report on bulk MnTe showed 175 cm−1\mathrm{cm}^{-1} Raman peak and assigned it to E2g_{2g} mode.[6]

Concurrent to the experimental studies, several first-principles calculations are carried out to obtain the phonon modes and thereby the Raman signatures in MnTe. Thapa et al. ruled out the possibility that the E2g_{2g} mode is responsible for Raman excitation at 175 cm−1\mathrm{cm}^{-1}. Instead, they proposed it as an outcome of a plasmonic (electronic excitations) origin, generated due to self hole doping. Specific to 175 cm−1\mathrm{cm}^{-1} peak, Wu et al.[8] proposed that, a plausible symmetry lowering distortion in highly symmetric MnTe can result in a leakage of 175 cm−1\mathrm{cm}^{-1} mode. The theoretical calculations of Raman active modes reported in [11] did not identify the concerned feature while accounting the in-plane tensile strain induced by the GaAs substrate and its temperature dependence. A DFT+U study by Uyukur et al.[9] in MnTe, with varying U values and different exchange-correlation functionals, reveals the presence of two closely spaced Raman active modes in the range ∼\sim86-102 cm−1\mathrm{cm}^{-1} and no other Raman active modes. Interestingly, the experimental study in this work reports the existence of two other strong modes at 120 cm−1120\ \mathrm{cm}^{-1} and 140 cm−1140\ \mathrm{cm}^{-1}. The latter are believed to be non Γ\Gamma-centered modes which are activated by higher order Raman processes. This feature has not been substantiated by any other work.

The contradictions among the varied reports, as mentioned above, have only increased the curiosity to address this seemingly open question. In this work, we demonstrate the presence of all the experimentally observed Raman peaks and proposed reasons on why some of the experiments may be missing them. We also show that the long-range dipole-dipole interaction, not employed in earlier studies, significantly modifies the phonon band structure and creates many low-intense phonon modes including the ones around 175 cm−1175\ \mathrm{cm}^{-1}. We further show that the higher-order Raman process with zone off-center contribution generates the rarely observed 140 cm−1140\ \mathrm{cm}^{-1} peak.

The harmonic forces required to analyze the vibrational modes of MnTe were calculated within density functional theory (DFT) as implemented in VASP. The interatomic forces were calculated using the PAW method with the PBE-GGA exchange-correlation functional. The crystal structure was fully relaxed while retaining the A type Néel order of this altermagnet. To account for the effect of correlations, the DFT+UU method is employed through the Dudarev approach[15]. A symmetric 2×2×22\times2\times2 supercell is used to incorporate the vibrations of the atoms. After calculating a set of converged forces using density-functional perturbation theory (DFPT), the dynamical matrices were obtained on a 9×9×39\times9\times3 kk-mesh and, the same set of convergence parameters were used as in the ground-state calculations. The dynamical matrix is solved using Phonopy[16] to obtain the phonon eigenvalues and eigenvectors.

The DFT+UU (U=4 eVU = 4\ \mathrm{eV}) calculated phonon band structure and atom-projected phonon density of states of MnTe are shown in Figure 1(a). This band structure is obtained by unfolding the phonon bands of the supercell onto the Brillouin zone (BZ) of the primitive unit cell and it is in agreement with the literature[11]. For comparison, the folded supercell phonon bands in the region of interest (150150–200 cm−1200\ \mathrm{cm}^{-1}) are shown in Figure 1(b), where we observe six high-frequency solutions of the dynamical matrix lying at Γ\Gamma. The highest optical mode (1), at 185.09 cm−1185.09\ \mathrm{cm}^{-1}, corresponds to a symmetric transverse optical (S-TO) vibration. The two doubly degenerate modes (2, 3), at 178.05 cm−1178.05\ \mathrm{cm}^{-1}, correspond to asymmetric transverse optical (A-TO) vibrations. The three degenerate longitudinal optical (LO) modes occur at 177.74 cm−1177.74\ \mathrm{cm}^{-1} and consist of two asymmetric LO (A-LO) and one symmetric LO (S-LO) modes. The displacement patterns of these modes are represented by vectors attached to the atomic sites in Figure 1(c). When these supercell phonon modes are unfolded onto the BZ of the primitive unit cell, the S-TO mode remained at Γ\Gamma, whereas the two A-TO modes are mapped onto the high-symmetry point A, as marked in Figure 1(a). Among the three LO modes, only the S-LO mode appears at the M point, while the two A-LO modes could not be discerned. Since all of them are located in the region of interest (175 cm−1175\ \mathrm{cm}^{-1}), it is important to analyze these vibrations along with the zone-center one to provide a broader understanding of the high frequency vibrational response of MnTe.

Phonon band structure, atom-projected phonon density of states, and displacement patterns of MnTe

Figure 1. (a) Phonon band structure and atom projected phonon density of states (PhDOS) of bulk MnTe without LR-correction. The high-symmetry points connecting the q-path is indicated in the Brillouin zone (right). The blue and red points on the spectra at Γ\Gamma, A and M points indicate three independent transverse optical (TO) and one longitudinal optical (LO) vibrations observed in MnTe system. (b) The phonon band structure for the 2×2×22\times2\times2 supercell of MnTe from which the bands are unfolded to represent the primitive unit cell band structure, as in Fig. (a). The blue and red bands are only shown for the high frequency TO and LO optical phonons, respectively. Once unfolded, the modes, 2-6 do not lie at Γ\Gamma and are mapped to other high symmetry points of the primitive unit of MnTe. We mapped (2, 3) and (6) belong to A and M, respectively. The vibration corresponds to these identified LO and TO modes are shown in panel (c). We have classified the vibrations to be symmetric if the neighboring Mn atoms exhibit identical displacement amplitudes and directions, and asymmetric when their displacement patterns differ.

There is a notable difference in the phonon band structure, particularly the optical phonon bands, of Figure 1(a) from the one shown by Wu et al. [8], which suggests that supercell convergence is a necessary sanity check. In Fig.S1 of the Supplementary Material (SM)-I, we have shown the evolution of the phonon band structure with the supercell size and this implies that the band structure corresponding to the 3×3×33\times3\times3 supercell fits well with this literature. The supercell convergence test also tells us that the flat band observed for the primitive unit be comes oscillatory as we increase the supercell. The other interesting observation is the presence of an unusually steep phonon dispersion along Γ\Gamma-A for supercells 2×2×22 \times2 \times2 and 3×3×33 \times3 \times3. Such an abrupt dispersion near Γ\Gamma can be linked with the finite size truncation of the vibration space. In fact, it is known that the estimation of accurate phonon modes requires very large supercells. Particularly, in the case of semiconductors and insulators, incomplete electronic screening in the case of smaller supercells can lead to inaccurate estimation of the interatomic force constants which become a prominent factor in the case of polar systems like NaCl, MgO, KCl etc.[17, 18]. In such systems, the finite size truncation can lead to Gibb’s oscillation [16, 19]. As it is practically impossible to compute an infinitely large supercell for the calculations, alternatively, one can consider the non-analytic correction (NAC) using the Born effective charges and dielectric tensor proposed by Gonze et al. [19] to account for the long-range (LR) behavior. The Born effective charges, listed in Table S1 and Table S2 of SM-II, are found to be substantial for MnTe, and therefore the LR dipole interaction correction is required to calculate the phonon band structure. The NAC incorporated phonon band structure of MnTe is shown in the lower panel of Figure S1 of SM, which not only implements significant corrections to the optical phonon modes, but also shows convergence with a 2×2×22 \times2 \times2 supercell. A detailed analysis of LR interaction induced changes in MnTe using mode dependent Born effective charges is provided in SM-II.

Phonon band structures of MnTe with varying supercell size

Figure S1. Phonon band structures of MnTe with varying supercell size. The upper and lower panels show the band structures with and without LR-corrections respectively. The long-range dipole-dipole contribution to the dynamical matrix, resulted in improved convergence with a 2×2×22 \times2 \times2 supercell, which is the minimum supercell size to retain all the operational symmetries for the space group.

AtomBorn effective charge Z∗Z^{*}
Mn(3.528≥Z∗≥2.7060.000.000.003.528≥Z∗≥2.7060.000.000.002.735≥Z∗≥2.571)\begin{pmatrix} 3.528 \ge Z^{*} \ge2.706 & 0.00 & 0.00 \\ 0.00 & 3.528 \ge Z^{*} \ge2.706 & 0.00 \\ 0.00 & 0.00 & 2.735 \ge Z^{*} \ge2.571 \end{pmatrix}
Te(−3.528≤Z∗≤−2.70570.000.000.00−3.528≤Z∗≤−2.70570.000.000.00−2.735≤Z∗≤−2.571)\begin{pmatrix} -3.528 \le Z^{*} \le-2.7057 & 0.00 & 0.00 \\ 0.00 & -3.528 \le Z^{*} \le-2.7057 & 0.00 \\ 0.00 & 0.00 & -2.735 \le Z^{*} \le-2.571 \end{pmatrix}
ϵ∞\epsilon_{\infty}(29.52≥ϵ≥10.070.000.000.0029.52≥ϵ≥10.070.000.000.0028.92≥ϵ≥10.32)\begin{pmatrix} 29.52 \ge\epsilon\ge10.07 & 0.00 & 0.00 \\ 0.00 & 29.52 \ge\epsilon\ge10.07 & 0.00 \\ 0.00 & 0.00 & 28.92 \ge\epsilon\ge10.32 \end{pmatrix}

Table S1. Range of Born effective charge, Z∗Z^{*}, and dielectric tensor, ϵ∞\epsilon_{\infty} for U values from 0 to 7 eV.

ω\omega (cm−1^{-1})ZMn∗Z^{*}_{\mathrm{Mn}}ZTe∗Z^{*}_{\mathrm{Te}}
Before LRAfter LRBefore LRAfter LRBefore LRAfter LR
128.74181.56 (Γ\Gamma-M) 128.74 (Γ\Gamma-M)(3.27,0,0);(3.27,0,0); (0,3.27,0)(0,3.27,0)(−2.84,−1.64,0)(-2.84,-1.64,0) (−1.64,2.84,0)(-1.64,2.84,0)(2.14,0,0);(2.14,0,0); (0,2.14,0)(0,2.14,0)(−1.86,−1.07,0)(-1.86,-1.07,0) (−1.07,1.86,0)(-1.07,1.86,0)
118.13170.42 (Γ\Gamma-A) 118.13 (Γ\Gamma-M)(0,0,3.16)(0,0,3.16)(0,0,3.17)(0,0,3.17) (0,0,3.16)(0,0,3.16)(0,0,2.08)(0,0,2.08)(0,0,2.08)(0,0,2.08)
128.74179.36 (Γ\Gamma-L) 128.74 (Γ\Gamma-L)(3.27,0,0);(3.27,0,0); (0,3.27,0)(0,3.27,0)(−2.59,−1.49,1.29)(-2.59,-1.49,1.29) (−1.64,2.84,0)(-1.64,2.84,0)(2.14,0,0);(2.14,0,0); (0,2.14,0)(0,2.14,0)(−1.70,−0.98,0.85)(-1.70,-0.98,0.85) (−1.07,1.86,0)(-1.07,1.86,0)
118.13120.22 (Γ\Gamma-L)(0,0,3.16)(0,0,3.16)(1.16,0.67,2.89)(1.16,0.67,2.89)(0,0,2.08)(0,0,2.08)(0.76,0.44,1.90)(0.76,0.44,1.90)

Table S2. Phonon frequencies and atom resolved mode effective charges (Zm,κ(q)Z_{m,\kappa}(\mathbf{q})) before and after the inclusion of LR interaction.

To further elucidate the distinct features introduced by the LR dipole-dipole interactions correction, in Figure 2, we represent the atom projected phonon band structure and compared them with the bands obtained without the LR corrections (red solid lines). We would also like to add that, since the corrected optical modes are not well-defined at Γ\Gamma due to directional dependence, we assigned their symmetry by comparing their characters with the appropriate irreducible representations along the corresponding high symmetry q-path in the BZ. The important changes by the LR correction, indicated by arrows in Figure 2, are as follows. (i) The steep dispersion of the TO phonons that rises from 118 cm−1118\ \mathrm{cm}^{-1} along Γ\Gamma-A disappears and instead a new branch of TO emerges from A at a degenerate point (178 cm−1178\ \mathrm{cm}^{-1}) and reaches 170.4 cm−1170.4\ \mathrm{cm}^{-1} at Γ\Gamma. This TO mode, marked by a blue circle, follows the C6vC_{6v} symmetry of the Γ\Gamma-A path and is assigned the A1A_1 mode. (ii) The LO branch that starts at 140 cm−1140\ \mathrm{cm}^{-1} at M and reaches 128.8 cm−1128.8\ \mathrm{cm}^{-1} rises to 181.6 cm−1181.6\ \mathrm{cm}^{-1}. A similar one is observed along K-Γ\Gamma. Both are marked by a pink rectangle and are also assigned the A1A_1 mode based on the C2vC_{2v} symmetry along the Γ\Gamma-M (K) path. (iii) Focusing on the path along L-Γ\Gamma-H, we find that a branch of LO in the range 128 cm−1128\ \mathrm{cm}^{-1} changes the curvature from downward to upward, as indicated by the arrows, and reaches 179±dcm−1179 \pm\mathrm{d} \mathrm{cm}^{-1} with a discontinuity at Γ\Gamma. This corrected LO mode (pink rectangle) follows the CsC_s symmetry of the Γ\Gamma-L (H) path and is assigned the A′A' mode. (iv) One of the TO modes at Γ\Gamma, marked with a blue circle and amplified in the inset, shifts from 118118 to 120 cm−1120\ \mathrm{cm}^{-1}. The LR correction lifts phonon degeneracy and modifies the harmonic vibrational energy landscapes around those above mentioned frequency values. Modes III and IV exhibit a significant increase in their effective force constants (see Figure S2, SM-III). This mode is assigned with A′A' and has a significant role in modifying the Raman spectra, as discussed later. Vibrational modes are assigned to appropriate irreducible representations by performing the point group dictated symmetry operations to their phonon eigenvectors and comparing them with the corresponding character tables (see SM-IV for details).

Comparison of phonon band structures without and with long-range interaction

Figure 2. The comparison of phonon band structure without LR interaction (red solid lines) and with LR interaction (color projected Mn and Te contributions). The inclusion of LR lifts the degeneracy in high frequency optical modes along certain symmetry paths and the changes are indicated through arrows.

Relative energy change as a function of phonon mode amplitude for four shifted modes

Figure S2. Relative energy change as a function of phonon mode amplitude before and after the inclusion of LR interactions for the four shifted modes labeled 1-4 [see also Figure 2(a)]. The energy profiles are fitted using a quadratic function to extract the corresponding harmonic force constants (CC). Notably, modes 3 and 4, associated with phonon propagation along the Γ\Gamma–L direction, exhibit a significant increase in CC upon inclusion of the LR correction.

Table I lists the existing and the new (bold) Raman modes after the LR correction. Their group theoretical representations and optical activities are also mentioned. Interestingly, we find that all the four new modes, A1A_1 (TO), A1A_1 (LO), A′A' (LO), A′A' (TO), denoted in Figure 2, are Raman active according to the selection rules of their respective path symmetries and they emphasize the necessity of the LR dipole-dipole interactions to explain the origin and robustness of the Raman response measured in MnTe.

The strong correlation has a profound effect on the electronic structure of MnTe[20, 21]. With a change in the electronic structure, the mean-field, responsible for the lattice vibration, also changes. Therefore, it is prudent to examine the effect of the correlation on the vibrational response of MnTe. To achieve this, we performed structural relaxation as a function of UU, and the results are summarized in Figure 3. From Figure 3(a), we see that both lattice volume and the c/ac/a ratio increase progressively with increasing UU, indicating an evolution of the equilibrium structure. We also note that with an increase in c/ac/a, the system achieves a saturated local magnetization consistent with previous reports[20, 22].

Normalized group-velocity maps overlaid with LR-corrected phonon band structures and schematic illustrations of TO and LO phonon guided distortions

Figure 3. (a) The c/ac/a ratio and the primitive-cell volume as a function of the effective Hubbard parameter (UU). Normalized group-velocity (vgv_g) maps overlaid with the LR-corrected phonon band structures for (b) U=1.5U = 1.5 eV, (c) 2 eV, and (d) 4 eV. The results indicate the UU-dependent behaviour of the high frequency optical phonon branches. Schematic illustrations of the structural distortions generated by displacements along the corresponding (e) TO, and (f) LO phonon eigenvectors of MnTe. The space groups of the resulting distorted structures are indicated.

Frequency (cm−1^{-1})Vibrational signatureMode representationActivity
185.03Mn: TO (Γ\Gamma)B2uB_{2u} (D6hD_{6h})S
181.56Mn: LO (Γ\Gamma-M)A1A_1 (C2vC_{2v})R
179.36Mn: LO (Γ\Gamma-L)A′A^{\prime} (CsC_s)R
170.42Mn: TO (Γ\Gamma-A)A1A_1 (C6vC_{6v})R
128.74 (2)Mn: LO (Γ\Gamma)E1uE_{1u} (D6hD_{6h})IR
120.35Te: TO (Γ\Gamma)B1gB_{1g} (D6hD_{6h})IR
120.22Mn: TO (Γ\Gamma-L)A′A^{\prime} (CsC_s)R
118.13Mn: TO (Γ\Gamma)A2uA_{2u} (D6hD_{6h})S
92.35 (2)Mn: LO (Γ\Gamma)E2gE_{2g} (D6hD_{6h})R
91.58 (2)Te: LO (Γ\Gamma)E2uE_{2u} (D6hD_{6h})S
Acoustic (3)Γ\Gamma––

Table 1. Raman modes with their group theoretical representations. The new Raman modes after the inclusion of LR corrections are highlighted in color. As observed, without LR interaction, phonon modes were either silent or, infrared active. After the LR driven splitting, new Raman active phonon modes are found and therefore additional peaks emerge in the frequency range of 120 -190 cm−1^{-1}.

Subsequently, the LR interaction included the phonon band structure for each value of UU, using the corresponding relaxed lattice parameters are calculated, and the results are shown in Figure 3(b). From the figure we gather that there are three major changes around the zone-center with increase in UU. First, the optical modes gradually shift toward higher frequencies with increasing UU. Second, concerning the bands along A-Γ\Gamma-M, we find that the relative ordering of the top LO and top TO bands changes with increase in UU, resulting in a LO–TO crossover (marked by a shaded region). Third, a distinct splitting occurs between Mn and Te dominated vibrational branches around 90 and 120 cm−1^{-1} (not shown here). These changes imply that the vibrational response of MnTe is strongly coupled to the UU-dependent structural changes. Based on the distinct structural and vibrational trends, UU dependent behavior can be broadly divided into three characteristic regimes, namely, R-I: 0 ≤U\leq U ≤\leq 1.8 eV; R-II: 2 ≤U\leq U ≤\leq 2.8 eV; and R-III: 3.0 ≤U\leq U ≤\leq 7.0 eV, as indicated in Figure 3(a).

As seen from Figure 3(b), for the top optical branches, the dispersion gradually steepens with increasing UU. To characterize it further and to shed light on the directional behavior of these high frequency vibrations, we examine the phonon group velocity, vgv_g, along the three independent high-symmetry directions, Γ\Gamma-A, Γ\Gamma-M, and Γ\Gamma-L, respectively (see SM-IV for more details). For the first and second correlation regimes, the highest LO, TO modes both exhibit negligible vgv_g along A-Γ-M. However, in the third regime, the vgv_g for TO along Γ\Gamma-M (25.51 km/s) becoming approximately five times larger than that along Γ\Gamma-A (5.97 km/s). In contrast, along Γ\Gamma-L, vgv_g is consistently the largest both for TO and LO modes across all UU ranges. In fact, we find that the vgv_g of the TO (255.73 km/s) and LO (2613.92 km/s) modes along Γ\Gamma-L are nearly two and three orders of magnitude larger, respectively, than those along Γ\Gamma-M. The directional dependence of vgv_g, reveals that the progression of the zone-center modes associated with specific distortions can be a prominent pathway to understand the vibrational fingerprint of MnTe.

Next, we examine the structural alterations arising from the high frequency vibrations and subsequent Raman response. This is achieved by “freezing" the phonon eigenvectors of the six modes (three LO and three TO) and analyzing the resulting symmetry lowered structures. Each mode produces a unique distortion in MnTe and lowers the parent space group (SG) symmetry P63/mmcP6_3/mmc (194) to a corresponding maximal subgroup symmetry, as indicated in Figure 3(c). The distortion directions can be visualized from the lower panel of Figure 3. The freezing of the S-TO mode produces a structure with SG P6ˉm2P\bar{6}m2 (187), whereas the A-TO mode produces a structure with SG P3m1P3m1 (156) structure. The A-LO and S-LO modes, in turn, lead to even lower symmetric structures with SG Amm2Amm2 (38) and CmCm (6), respectively.

The LR corrected phonon band structures for two representative LO and TO distortions, P6ˉm2P\bar{6}m2 and Amm2Amm2, are shown in Figure 4(a, Figure 4b), remaining two are shown in Figure S3,Figure S3). The P6ˉm2P\bar{6}m2 distortion produces only minor changes to the phonon bands of the pristine MnTe. On the other hand, when compared to the dispersion along the transverse Γ\Gamma-A and longitudinal Γ\Gamma-M for the pristine MnTe (see Figure 2), the Amm2Amm2 distorted structure with a different BZ (Figure 4(b)) exhibits contrasting phonon dispersion along the transverse Γ\Gamma-Z and longitudinal Γ\Gamma-Y directions (see the highlight in Figure 4(a)). As observed from Figure 4(b), Amm2Amm2 type distortion reduces the directional disparity of the modes around zone-center. The band structures corresponding to P3m1P3m1 and CmCm distorted structure are nearly identical to the case of P6ˉm2P\bar{6}m2 and Amm2Amm2, respectively and hence are not shown.

Phonon band structures and Raman spectra of pristine and distorted MnTe

Figure 4. (a, b) Phonon band structures for two distinct TO (P6ˉm2P\bar{6}m2) and LO (Amm2Amm2) distortion of MnTe. The atoms are displaced along the identified phonon modes to the tune of a mass-weighted mode amplitude of 0.1 Å·amu1/2^{1/2}. Their corresponding BZs are shown with right panel. (c, d) LR-corrected Raman spectra of pristine MnTe with all the plausible distorted analogues for two separate frequency ranges 80 - 100 cm−1^{-1}, and 100 - 190 cm−1^{-1}, respectively. The Raman peak positions without LR interaction are represented through colored dots in the spectra. (e) The off-center phonon contribution to the Raman spectra emerging out of the M and L centered phonons and are compared to that of the zone centered phonons.

Phonon band structures for the TO and LO distortion of MnTe

Figure S3. (a, b) Phonon band structures for the TO (P3m1P3m1) and LO (CmCm) distortion of MnTe. The atoms are displaced along the identified modes to the tune of a mass weighted mode amplitude of 0.1 amu1/20.1\ \mathrm{amu}^{1/2}.

To comprehend the overall changes in the vibrational landscape of MnTe, we calculated the first-order Raman spectra of pristine and distorted MnTe by taking into account the NAC. The detailed methodology for calculating Raman intensities and constructing Raman spectral envelope from LR corrected phonon frequencies and eigenvectors is provided in SM-V. The zone-center Raman spectra are presented in Figure 4(c) and (Figure 4d) for two separate frequency ranges of 80 - 100 cm−1^{-1} and 100 - 190 cm−1^{-1}, respectively. The positions of the peaks without LR interaction are also indicated by colored dots. We emphasize that, several experimental studies[11, 12] have reported four distinct Raman features below 200 cm−1^{-1}, centered around 90 cm−1^{-1}, 122 cm−1^{-1}, 140 cm−1^{-1}, and 175 cm−1^{-1}.

Although theoretical calculations ignoring the LR dipole-dipole interactions have correctly reproduced the 90 cm−1^{-1} peak[8, 9], they have failed to do so for the remaining three. Various mechanisms, including distortion-mediated and plasmon-related effects, have been proposed to account for the higher frequency Raman features in MnTe (see the first section). Raman spectra for a MnTe2_2, appearing as an impurity phase in MnTe is also calculated, see SM-VI. In this study, by incorporating the LR dipole-dipole interaction and by resolving the associated splitting in previously degenerate LO and TO modes, we reproduce all four experimentally observed Raman peaks under appropriate structural and vibrational conditions.

The prominent Raman peak near 90 cm−1^{-1} in Figure 4(c) corresponds to the robust E2gE_{2g} type vibration of pristine MnTe. This mode remains the most intense Raman peak irrespective of the structure being pristine or distorted. However, the distortions P6ˉm2P\bar{6}m2 and CmCm cause a small split of this mode by ∼\sim4 cm−1^{-1}, while preserving its characteristic vibrational pattern. Ukyur et al. in their experimental studies have reported the presence of close-spaced double peaks corresponding to this E2gE_{2g} mode[9]. As observed in Figure 4(d), for pristine MnTe, the peak near 120 cm−1120\ \mathrm{cm}^{-1} originates from the LO branch along Γ\Gamma-L with A′A^\prime symmetry and is revealed by the LR correction (see inset of Figure 2). Without the correction, this mode had the A2uA_{2u} symmetry and is Raman inactive. The Raman response also depends strongly on the distortion, such as, the TO-type distortions (P6ˉm2P\bar{6}m2, P3m1P3m1) suppress the 120 cm−1120\ \mathrm{cm}^{-1} feature, whereas the LO-type distortions (Amm2Amm2, CmCm) intensify it. LO type distortions also give rise to another peak at 129 cm−1129\ \mathrm{cm}^{-1} which was also silent (E1uE_{1u}) in pristine MnTe.

The LR dipole-dipole interactions introduce a new Raman peak in pristine MnTe that appears at 179 cm−1179\ \mathrm{cm}^{-1} (see Figure 4(d)). We traced its origin to the splitting of the S-LO phonon along the Γ\Gamma-L direction (see Figure 2) and we attribute it as one of the reasons for experimentally observing a peak around 175 cm−1175\ \mathrm{cm}^{-1}. The same feature is retained in the both LO-distorted structures, but disappears for the TO-type distortions. Instead, the TO-distorted structures develop another A1A_1-type Raman mode near 185 cm−1185\ \mathrm{cm}^{-1} both with and without LR correction. This is consistent with the discussions made by Wu et al., where the authors have pinned down the highest TO mode to be Raman active under this specific (S-TO distortion) as the symmetry of this mode changes from B2uB_{2u} to A′A^\prime [8]. Furthermore, S-LO distortion corresponding to CmCm symmetry shows a Raman active mode at 170 cm−1170\ \mathrm{cm}^{-1} (see the upper panel Figure 4(d)).

Next, we discuss the experimental Raman peak ∼140 cm−1\sim140\ \mathrm{cm}^{-1}. Previous reports have attributed it to the existence of zone off-center Raman modes [8, 9]. From our calculated Raman spectra, we indeed found a Raman peak at 140 cm−1140\ \mathrm{cm}^{-1} (see Figure 4(e)) and they originate from zone boundary L-phonons. Since this peak does not arise from zone-center phonons, but instead comes to existence from off-center contributions at L and M, the experimental detection of this particular peak may indicate the activation of phonons through higher-order Raman processes[23, 24]. In addition, we see two new peaks around 120 cm−1120\ \mathrm{cm}^{-1} arising from L and M phonons. Therefore, 120 cm−1120\ \mathrm{cm}^{-1} peak is collectively formed by zone-center and zone-boundary phonons, while the 120 cm−1120\ \mathrm{cm}^{-1} peak has contribution from both zone-center and zone boundary phonons. The highly debated Raman peak around 175 cm−1175\ \mathrm{cm}^{-1} seems to have many origins, including LR effects in pristine MnTe, and LO distortions. The Robust Raman peak around 175 cm−1175\ \mathrm{cm}^{-1} due to the presence of an impurity MnTe2\mathrm{MnTe}_{2} phase can be considered as an experimental anomaly.

Acknowledgement

SJ thanks IIT Madras for providing fellowship, HPCE, and Kamadhenu cluster for the required computational support.

In summary and conclusion, we provide the missing theoretical explanation of the experimentally observed landscape of the Raman response of MnTe. We demonstrate that the LR dipole-dipole interaction is the key to accurately calculate the phonon band structure of this altermagnetic insulator and it is due to the fact that MnTe possess large directional-dependent Born effective charges. The LR corrections to the phonon band structure introduce additional modes due to splitting, and subsequently some of the silent modes give rise to Raman Active modes. Our study reveals Raman active modes around 120 cm−1120\ \mathrm{cm}^{-1}, 140 cm−1140\ \mathrm{cm}^{-1}, and 179 cm−1179\ \mathrm{cm}^{-1} which are in addition to the existing 90 cm−190\ \mathrm{cm}^{-1} peak, which is reproduced even without LR correction. Further analysis shows that the 140 cm−1140\ \mathrm{cm}^{-1} peak originates from the

Supplementary Material

I: Supercell convergence test

II: Non analytical correction in long wavelength limit

The Born effective charge tensor (Z∗Z^*) quantifies the change in macroscopic polarization induced by an infinitesimal atomic displacement and is defined as Zκ,αβ∗=Ωe∂Pα∂uκβ∣E=0Z^*_{\kappa,\alpha\beta} = \left.\frac{\Omega}{e}\frac{\partial P_\alpha}{\partial u_{\kappa\beta}}\right|_{\mathbf{E}=0}, where Ω\Omega is the unit-cell volume, ee is the elementary charge, and PαP_\alpha denotes the α\alpha component of the macroscopic polarization. Here, uκβu_{\kappa\beta} represents the displacement of atom κ\kappa along the direction β\beta, while α\alpha specifies the polarization direction (α,β=x\alpha,\beta= x, yy, or zz) (in zero external field). Thus, Z∗Z^* characterizes the coupling between ionic displacements and electronic response as macroscopic polarization.

For a phonon characterized by the wave vector q\mathbf{q}, a nonzero projection of the Born effective charge, q⋅Zκ∗≠0\mathbf{q} \cdot\mathbf{Z}^*_{\kappa} \ne0, gives rise to a long-range electric field associated with lattice vibration. In the long-wavelength limit (q→0\mathbf{q} \to0), this field introduces a nonanalytic, direction-dependent dipole-dipole contribution to the dynamical matrix, which is screened by the high frequency dielectric tensor (ϵ∞\epsilon_\infty) of the material. This non-analytical correction (NAC) to the dynamical matrix, therefore, can be written as[17, 18, 25]

Dκα,κ′βNA(q→0)=4πe2Ω[q⋅Zκ∗]α[q⋅Zκ′∗]βMκMκ′(q⋅ϵ∞⋅q)(1)D^{\mathrm{NA}}_{\kappa\alpha,\kappa'\beta}(\mathbf{q}\to0)=\frac{4\pi e^2}{\Omega}\frac{[\mathbf{q}\cdot\mathbf{Z}^*_{\kappa}]_\alpha[\mathbf{q}\cdot\mathbf{Z}^*_{\kappa'}]_\beta}{\sqrt{M_\kappa M_{\kappa'}}(\mathbf{q}\cdot\epsilon_\infty\cdot\mathbf{q})} \tag*{(1)}

where MκM_\kappa and Mκ′M_{\kappa'} are the atomic masses corresponding to the dynamic charges, Zκ∗Z^*_\kappa and Zκ′∗Z^*_{\kappa'}. The total dynamical matrix around the point Γ\Gamma was, therefore, obtained by

D(q)=DA(q)+DNAC(q→0),(2)D(\mathbf{q})=D^{\mathrm{A}}(\mathbf{q})+D^{\mathrm{NAC}}(\mathbf{q}\to0), \tag*{(2)}

where DAD^{\mathrm{A}} represents the analytic part obtained from the short-range interatomic force constants.

We have obtained the Z∗Z^* and ϵ∞\epsilon_\infty tensors from first-principles calculations for the α\alpha-MnTe system. Incorporating these quantities into D(q)D(\mathbf{q}) accounts for the direction-dependent modifications to the phonon dispersion spectra, as shown in Figure 2(a). The optical phonon branches near the Γ\Gamma point experiences significant changes, including possible LO-LO and TO-TO splitting in MnTe. The magnitude of this splitting depends on the ratio of Z∗Z^* and ϵ∞\epsilon_\infty, and the propagation direction, q\mathbf{q}. Hence, we calculate the mode effective charge by projecting Z∗Z^* onto the corresponding phonon mode, eκ,me_{\kappa,m}. For this particular vibrational eigenvector, represented by the unit vector qm\mathbf{q}_m, the effective charge of the mode is expressed as[26, 27],

Zm,κ(q)=∑α,βqmZκ,αβ∗(3)Z_{m,\kappa}(\mathbf{q})=\sum_{\alpha,\beta}q_m Z^*_{\kappa,\alpha\beta} \tag*{(3)}

A nonzero value of Zm(q)Z_m(\mathbf{q}) for each sublattice vibration indicates that the corresponding phonon mode can create long-range order in the system, eventually creating macroscopic polarization.

As shown in Table I, Z∗Z^* of Mn and Te exhibit nearly equal magnitudes but opposite signs, with the in-plane components varying from approximately 3.528 to 2.706, and the out-of-plane components from 2.735 to 2.571 with the increase in UU. ϵ∞\epsilon_\infty is also anisotropic across the considered UU range, with in-plane components varying between 29.52 - 10.07, and the out-of-plane component between 28.92 - 10.32.

In MnTe, the direction dependent lifting of optical phonon degeneracy near the Γ\Gamma point can be understood through the calculation of mode effective charges, Zm(q)Z_m(\mathbf{q}), summarized in Table S2. Before including LR interactions, the optical modes at 128.74 cm−1^{-1} are doubly degenerate, with linear polarizations arising from the relative displacement of the Mn and Te sublattices. The effective charges in the corresponding mode are ZMn=(3.27,0,0)Z_{\mathrm{Mn}}=(3.27,0,0) and ZTe=(2.14,0,0)Z_{\mathrm{Te}}=(2.14,0,0). Their parallel orientation indicates that the cumulative contributions of Mn and Te reinforce the induced macroscopic polarization despite their opposite Z∗Z^* (see Table S1). Upon inclusion of the LR interaction, the split observed for a degenerate mode mm, can be quantitatively estimated as,

Δωm2=e2ϵ0Ω∣q⋅Zm∣2q⋅ϵ∞⋅q(4)\Delta\omega_m^2=\frac{e^2}{\epsilon_0\Omega}\frac{\lvert\mathbf{q}\cdot\mathbf{Z}_m\rvert^2}{\mathbf{q}\cdot\epsilon_\infty\cdot\mathbf{q}} \tag*{(4)}

ZmZ_m is the mass weighted Zm(q)Z_m(\mathbf{q}). Using the atomic masses of Mn (54.938 amu) and Te (127.60 amu), the mass weighted mode effective charge for propagation along q=(1,0,0)\mathbf{q}=(1,0,0) is, Zxx=3.2754.938+2.14127.60≈0.631Z_{xx}=\frac{3.27}{\sqrt{54.938}}+\frac{2.14}{\sqrt{127.60}}\approx0.631.

For the primitive cell of MnTe, Ω=103.9603\Omega=103.9603 Å3^{3} and ϵ∞,xx=11.5224\epsilon_{\infty,xx}=11.5224 (see Table S1), the squared frequency difference (ν~\tilde{\nu}) becomes

Δν~2=41078.17∣Zxx∣2.(5)\Delta\tilde{\nu}^{2}=41078.17\lvert Z_{xx}\rvert^{2}. \tag*{(5)}

Substituting the calculated ZxxZ_{xx} gives ν~new≈181.4\tilde{\nu}_{\mathrm{new}}\approx181.4 cm−1^{-1} from the initial value of 128.74 cm−1^{-1}, corresponding to an LO-LO split of approximately 52.7 cm−1^{-1}. This agrees well with the LR corrected phonon spectra, where the mode shifts from 128.74 to 181.56 cm−1^{-1} (see Table S2).

Similarly, along Γ\Gamma-L (q=(1,0,1)\mathbf{q}=(1,0,1)), the mode shift occurs at 179.36 from 128.74 cm−1^{-1}. The corresponding Zm,κ(q)Z_{m,\kappa}(\mathbf{q}) also change orientation and become plane-polarized as a result of this correction, indicating a redistribution of the eigenvector character (amplitude ∣Zm,κ(q)∣\lvert Z_{m,\kappa}(\mathbf{q})\rvert remains the same). Similar kind of behaviour is also observed along the in-plane Γ\Gamma-K direction (q=(1,1,0)\mathbf{q}=(1,1,0)) as well. Thus, the LR interaction modifies both the phonon frequencies and the polarization character of the optical modes in MnTe, depending on the propagation direction. Furthermore, along the Γ\Gamma-A direction, the optical mode at 118.13 cm−1^{-1} shifts to 170.42 cm−1^{-1} upon inclusion of the LR interaction. The atom resolved mode effective charges remain unchanged for the out-of-plane direction.

III: Energy landscape with and without LR interaction

For a phonon mode mm, characterized by the eigenvector eκαme^{m}_{\kappa\alpha}, the atomic displacement is expressed as,

ΔRκαm=ueκαmMκ,(6)\Delta R^{m}_{\kappa\alpha}=u\frac{e^{m}_{\kappa\alpha}}{\sqrt{M_{\kappa}}}, \tag*{(6)}

where uu is the mass-weighted mode amplitude and MκM_{\kappa} is the mass of atom κ\kappa (phase factor, ϕ=0\phi=0 for zone-center vibrations). The potential energy along this vibrational mode can be expanded as,

Um(u)=U0+(dUmdu)0u+12(d2Umdu2)0u2+O(u3).(7)U_{m}(u)=U_{0}+\left(\frac{dU_{m}}{du}\right)_{0}u+\frac{1}{2}\left(\frac{d^{2}U_{m}}{du^{2}}\right)_{0}u^{2}+\mathcal{O}(u^{3}). \tag*{(7)}

At equilibrium, the linear term vanishes. Within the harmonic approximation,

ΔUm(u)=Um(u)−U0=12Cmu2,Cm=(d2Umdu2)0.(8)\Delta U_{m}(u)=U_{m}(u)-U_{0}=\frac{1}{2}C_{m}u^{2},\qquad C_{m}=\left(\frac{d^{2}U_{m}}{du^{2}}\right)_{0}. \tag*{(8)}

Following this formalism, we employed a mode dependent frozen-phonon approach, in which atomic configurations are systematically distorted along the corresponding phonon eigenvectors. The distorted atomic positions are then obtained as,

Rκαm(u)=Rκα0+ΔRκαm(u),(9)R_{\kappa\alpha}^{m}(u) = R_{\kappa\alpha}^{0} + \Delta R_{\kappa\alpha}^{m}(u), \tag*{(9)}

where Rκα0R_{\kappa\alpha}^{0} denotes the equilibrium atomic position. For each selected mode, a series of distorted structures is generated by varying uu while keeping the eigenvector fixed. The corresponding total energies are evaluated without further atomic relaxation, allowing the potential energy surface to be mapped along the chosen vibrational coordinate. We quantify the effect of LR interactions on the vibrational potential of MnTe and examined the relative potential energy (UU) as a function of newly introduced phonon mode amplitude. Within the harmonic approximation, the energy retains its parabolic dependence on the mode amplitude, expressed as ΔU=12Cu2\Delta U = \frac{1}{2}Cu^{2}. We focus on four modes (1-4) highlighted in Figure 2 of the main manuscript, which exhibit distinct responses to the LR correction. Owing to the symmetry of the harmonic potential, only positive mode amplitudes are shown. As illustrated in Figure 2, modes 1 and 2, associated with vibrations along the Γ\Gamma-A and Γ\Gamma-M directions, respectively, exhibit negligible changes in their energy profiles and force constants upon inclusion of the LR interaction. In contrast, mode 3, propagating along Γ\Gamma-L, shows a pronounced increase in the curvature of the potential energy, indicating enhanced vibrational stiffness. Mode 4 exhibits the largest increase in the fitted force constant, from 3.76 to 4.47 eV/amu.A˚2\mathring{\mathrm{A}}^{2}. These results helped us quantifying the overall effects of LR correction on the vibrational restoring forces in a mode and direction dependent manner, leading to selective phonon hardening in MnTe around the zone-center.

Group-theoretical Analysis of Phonon Modes

The original group of MnTe points is D6hD_{6h} that contains 24 symmetry operations. At Γ\Gamma, all 24 symmetry operations are respected, but away from the the center, certain symmetry operations are violated. The reduction of these symmetry operations gives rise to several subgroups. After the inclusion of LR, there are four new direction dependent phonon modes around the center whose irreducible representations depend on the path symmetry along that direction, as seen in Figure 2. As a result, along Γ\Gamma-A path the C6vC_{6v}, Γ\Gamma-M path C2vC_{2v}, and Γ\Gamma-L CsC_{s} point group symmetry is maintained. These groups contain 12, 4, and 2 symmetry operations, respectively.

C6vC_{6v}EE2C62C_{6}2C32C_{3}C2C_{2}3σv3\sigma_{v}3σd3\sigma_{d}Basis functions
A1A_{1}111111zz; x2+y2,z2x^{2}+y^{2},z^{2}
A2A_{2}1111-1-1RzR_{z}
B1B_{1}1-11-11-1—
B2B_{2}1-11-1-11—
E1E_{1}21-1-200(x,y)(x,y); (xz,yz)(xz,yz)
E2E_{2}2-1-1200(x2−y2,xy)(x^{2}-y^{2},xy)

Table S3. Character table of the C6vC_{6v} point group.

The eigenvector for Mode 1 at 170.42 cm−1^{-1} along Γ\Gamma-A direction signifies an out of plane optical vibration of two atomic sublattices in opposite directions and is of A1A_{1} character with Raman tensor, R(A1)R(A_{1}) as,

R(A1)=(a000a000b),(10)R(A_{1}) = \begin{pmatrix} a & 0 & 0 \\ 0 & a & 0 \\ 0 & 0 & b \end{pmatrix}, \tag*{(10)}
C2vC_{2v}EEC2(z)C_{2}(z)σxz\sigma_{xz}σyz\sigma_{yz}Basis functions
A1A_{1}1111zz; x2,y2,z2x^{2},y^{2},z^{2}
A2A_{2}11-1-1xyxy; RzR_{z}
B1B_{1}1-11-1xzxz
B2B_{2}1-1-11yy; yzyz

Table S4. Character table of the C2vC_{2v} point group.

Along Γ\Gamma-M, Mode 2 at 181.56 cm−1^{-1} behaves as a LO mode where atomic displacements are strictly aligned at approximately 30∘30^\circ to the xx axis, with Mn and Te moving oppositely in-plane. This mode transforms according to the totally symmetric A1A_1 irreducible representation of the C2vC_{2v} group, with R(A1)R(A_1) as,

R(A1)=(a000b000c)(11)R(A_1)=\begin{pmatrix}a&0&0\\0&b&0\\0&0&c\end{pmatrix} \tag*{(11)}
CsC_sEEσxz\sigma_{xz}Basis functions
A′A'11xx, zz; x2x^2, y2y^2, z2z^2, xzxz
A′′A''1-1yy; xyxy, yzyz

Table S5. Character table of the CsC_s point group.

In both Modes 3 and 4, at 179.36 cm−1179.36\ \mathrm{cm}^{-1} and 120.22 cm−1120.22\ \mathrm{cm}^{-1}, respectively, along Γ\Gamma–L, the Mn and Te sublattices exhibit oppositely directed optical vibrations with coupled in-plane and out-of-plane components. For both modes, the in-plane displacement vectors are oriented at 30∘30^\circ to the xx-axis. These modes transform according to the irreducible representation A′A' of CsC_s with the corresponding Raman tensor, R(A′)R(A') as,

R(A′)=(a0d0b0d0c)(12)R(A')=\begin{pmatrix}a&0&d\\0&b&0\\d&0&c\end{pmatrix} \tag*{(12)}

All four modes lie close to 175 cm−1175\ \mathrm{cm}^{-1} and have only diagonal Raman tensor components, making them observable in parallel polarization but absent in cross-polarization. This is consistent with the nature of the experimentally detected reports mentioned earlier.

Distortion Design in MnTe

To investigate the previously proposed mode dependent lattice distortions, we first evaluated the phonon group velocities, vg,m(q)=∇qωm(q)v_{g,m}(q)=\nabla_q\omega_m(q), for the high-frequency optical branches along selected high-symmetry directions. Although the phonon group velocity does not directly represent the evolution of a structural distortion, it provides useful information about the direction of vibrational energy propagation. Based on the calculated vg,m(q)v_{g,m}(q) we identified four representative directions that exhibit appreciable group velocities, representing two predominantly transverse optical (TO) and two longitudinal optical (LO) modes. Subsequently, mode-dependent frozen-phonon distortions were constructed using the corresponding phonon eigenvectors, with a fixed displacement amplitude of 0.10.1 Å, to investigate the associated changes in phonon spectra. Two such distortion induced phonon band structures are given in Figure 4(a) of the main manuscript and the other two are shown here in Figure 3(a,Figure 3b).

Details of Raman Spectra Calculation

For subsequent estimation of Raman intensity, Γ\Gamma-center phonon frequencies ωj\omega_j and eigenvectors eje_j (mode, j=1,12j=1,12 for MnTe) are taken from the dynamical matrix included in LR interaction computed with Phonopy. For each mode, two structures are generated by displacing every atom along ±0.01\pm0.01 Å, as explained in SM-III. The macroscopic static dielectric tensor ε±\varepsilon^{\pm} of each displaced structure is then computed using VASP. The Raman tensor of mode jj then calculated from the finite difference of ε+\varepsilon^{+}, ε−\varepsilon^{-} with respect to the normal coordinate uju_j,

Rαβ(j)=V4π∂εαβ∂uj(13)R_{\alpha\beta}^{(j)}=\frac{V}{4\pi}\frac{\partial\varepsilon_{\alpha\beta}}{\partial u_j} \tag*{(13)}

where VV is the primitive cell volume. The calculated RαβR_{\alpha\beta} for each mode follows the symmetry constraints imposed by group theory, as discussed in SM-IV. For each vibrational mode, RαβR_{\alpha\beta} can be reduced to the two rotational invariants, the isotropic component αj\alpha_j and the anisotropic component βj2\beta_j^2, where

αj=13(Rxx(j)+Ryy(j)+Rzz(j)),(14)\alpha_j=\frac{1}{3}\left(R_{xx}^{(j)}+R_{yy}^{(j)}+R_{zz}^{(j)}\right), \tag*{(14)}
βj2=12[(Rxx−Ryy)2+(Rxx−Rzz)2+(Ryy−Rzz)2]+3(Rxy2+Rxz2+Ryz2).(15)\beta_j^2=\frac{1}{2}\left[\left(R_{xx}-R_{yy}\right)^2+\left(R_{xx}-R_{zz}\right)^2+\left(R_{yy}-R_{zz}\right)^2\right]+3\left(R_{xy}^2+R_{xz}^2+R_{yz}^2\right). \tag*{(15)}

The Raman activity of mode jj is then calculated as [28, 29],

Ij=45αj2+7βj2.(16)I_j=45\alpha_j^2+7\beta_j^2. \tag*{(16)}

This corresponds to the sum of the parallel and perpendicular components, Ij=I∥+I⊥I_j=I_{\parallel}+I_{\perp}, with I∥=45αj2+4βj2I_{\parallel}=45\alpha_j^2+4\beta_j^2 and I⊥=3βj2I_{\perp}=3\beta_j^2 [30]. Therefore, the Raman intensity calculated this way does not depend on the selective polarization environment.

VII: Phonon spectra of MnTe2

In Figure S4, we have plotted the phonon band structure and the Raman response. The latter reproduces the well-established 175 cm−1^{-1} peak. Some of the experimental studies on MnTe have reported this peak and they have attributed it to the impurity MnTe2_2 phase.

Calculated phonon dispersion, bulk crystal structure, and Raman spectrum of α-MnTe2

Figure S4. (a, b) Calculated phonon dispersion of α\alpha-MnTe2_2 and its corresponding bulk crystal structure, respectively. (c) Calculated Raman spectrum with vibrational mode assignments based on the Pa3‾Pa\overline{3} (No. 205) space group symmetry. A prominent Raman peak is observed near 175 cm−1^{-1}, which may serve as a characteristic signature of α\alpha-MnTe2_2 in experimental MnTe samples containing Te-rich impurity phases.

References

References

  1. [5]termagnetic candidate mnTe, Advanced Functional Materials 34, 2405829 (2024).DOI
  2. [6]K. Kluczyk, K. Gas, M. Grzybowski, P. Skupiński, M. Borysiewicz, T. Fąs, J. Suffczyński, J. Domagała, K. Grasza, A. Mycielski, et al., Coexistence of anomalous hall effect and weak magnetization in a nominally collinear antiferromagnet mnte, Physical Review B 110, 155201 (2024).DOI
  3. [7]J. Zhang, Q. Lian, Z. Pan, W. Bai, J. Yang, Y. Zhang, X. Tang, and J. Chu, Spin-phonon coupling and two-magnons scattering behaviors in hexagonal nias-type antiferromagnetic mnte epitaxial films, Journal of Raman Spectroscopy 51, 1383 (2020).DOI
  4. [8]A. Wu, D. Chen, J.-Y. Chung, X. Wang, M. Zeng, C. Liu, S. Gradečak-Garaj, and X. Li, Domains with noncentrosymmetric structural distortion in altermagnetic mnte, Physical Review B 114, 074436 (2026).DOI
  5. [9]E. Ukyur, M. V. Goncalves-Faria, S. Rössler, V. A. Ginga, M. Schmidt, S. Winnerl, M. Helm, and A. A. Tsirlin, Revisiting the symmetry and optical phonons of altermagnetic α-mnte, Physical Review B 113, 224107 (2026).
  6. [10]W. Szuszkiewicz, J. Mohrange, M. Jouanne, M. Kanehisa, R. Świrkowicz, E. Dynowska, E. Janik, T. Wojtowicz, and J. Kossut, Temperature dependence of raman scattering by magnons in bulk-like mbe-gwn mnte films, Acta Physica Polonica A 92, 1025 (1997).DOI
  7. [11]N. Azam, S. M. Shahed, L. T. Schmidt, S. Bey, O. Yastrubchak, M. F. Munoz, R. Torsi, T. H. Y. Pham, D. Balsundaram, R. B. Regami, et al., The origin of the observed raman peaks in {\alpha}-mnte, arXiv preprint arXiv:2609.19467 (2026).arxiv.org/abs/2609.19467
  8. [12]P. Bhardwaj, R. Sarkar, N. Shyaga, S. K. Negi, K. Biswas, S. Jana, P. Nukala, B. R. K. Nanda, and D. Suri, Nature of the 175-180 cm−1 raman feature in altermagnetic α-mnte (2026), arXiv:2609.39108 [cond-mat.mtrl-sci].arxiv.org/abs/2609.39108
  9. [13]W. Szuszkiewicz, M. Jouanne, J.-F. Morhange, M. Kanehisa, E. Dynowska, K. Gas, E. Janik, G. Karczewski, R. Kuna, and T. Wojtowicz, Raman scattering as a tool to characterize semiconductor crystals, thin layers, and low-dimensional structures containing transition metals, physica status solidi (b) 251, 1133 (2014).
  10. [14]T. Palasyuk, D. Kurzydłowski, A. S. Wadge, R. Bacewicz, M. H. Dalsaniya, M. Wierzbicki, D. Jastrzębski, A. Wiśniewski, W. Szuszkiewicz, I. Antoniazzi, et al., Paramagnetic phase of mnte2. Complete one phonon raman spectrum, Journal of Alloys and Compounds , 189349 (2026).DOI
  11. [15]S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P. Sutton, Electron-energy-loss spectra and the structural stability of nickel oxide: An lsda+u study, Physical Review B 57, 1505 (1998).DOI
  12. [16]A. Togo, L. Chaput, T. Tadano, and I. Tanaka, Implementation strategies in phonopy and phono3py, J. Phys. Condens. Matter 35, 535001 (2023).DOI
  13. [17]A. Togo, H. Hayashi, T. Tadano, S. Tsutsui, and I. Tanaka, Lo-mode phonon of kcl and nacl at 300 k by inelastic x-ray scattering measurements and first principles calculations, Journal of Physics: Condensed Matter 34, 365401 (2022).DOI
  14. [18]G. Coiana, J. Lischner, and P. Tängney, Breakdown of phonon band theory in mgo, Physical Review B 109, 014310 (2024).arxiv.org/abs/2307.10944
  15. [19]X. Gonze and C. Lee, Dynamical matrices, born effective charges, dielectric permittivity tensors, and interatomic force constants from density-functional perturbation theory, Physical Review B 55, 10355 (1997).
  16. [20]E. D. Chernov and A. V. Lukoyanov, Electronic correlations in altermagnet mnte in hexagonal crystal structure, Materials 18, 2637 (2025).
  17. [21]B. Kang, A. Janotti, D. Q. Ho, M.-H. Kim, C. H. Park, S. Choi, M. R. Pederson, and E. Kim, Strong electron correlations and ligand hybridization for altermagnetism, arXiv preprint arXiv:2605.14248 (2026).arxiv.org/abs/2605.14248
  18. [22]E. P. Carlisle, G. Yumnam, S. Calder, B. Haberl, J.-X. Xiong, M. A. McGuire, A. Zunger, R. P. Hermann, and B. A. Frandsen, Tuning the magnetic properties of the spin-split antiferromagnet mnte through pressure, Physical Review B 112, 014450 (2025).
  19. [23]H. Padmanabhan, M. Poore, P. K. Kim, N. Z. Koocher, V. A. Stoica, D. Puggioni, H. Wang, X. Shen, A. H. Reid, M. Gu, et al., Interlayer magnetophononic coupling in mnbi2te4, Nature communications 13, 1929 (2022).
  20. [24]G. Güntherodt, W. Bauhofer, and G. Benedek, Zone-boundary-phonon Raman scattering in v i 2 due to modulation of exchange interaction, Physical Review Letters 43, 1427 (1979).
  21. [25]X. Gonze and C. Lee, Dynamical matrices, born effective charges, dielectric permittivity tensors, and interatomic force constants from density-functional perturbation theory, Physical Review B 55, 10355 (1997).DOI
  22. [26]A. Molina-Sanchez and L. Wirtz, Phonons in single-layer and few-layer mos 2 and ws 2, Physical Review B—Condensed Matter and Materials Physics 84, 155413 (2011).
  23. [27]M. Srinivasan, K. Fujiwara, et al., Understanding the role of non-analytical corrections in phonon transport and thermal conductivity in half-heusler alloys, Journal of Applied Physics 137, 2025 (2025).
  24. [28]A. Fonari and S. Stauffer, vasp_raman.py, 2013, URL github. com/raman-sc/VASP.
  25. [29]D. Porezag and M. R. Pederson, Infrared intensities and raman-scattering activities within density-functional theory, Physical Review B 54, 7830 (1996).pubmed.ncbi.nlm.nih.gov/9984457
  26. [30]D. A. Long, The raman effect, (No Title) (2002).

Paper details

vlnc:W_qrxqgfjr
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents