betaFeedback

WorksPhysics and Astronomy

Origin of Rashba spin-splitting and strain tunability in ferroelectric bulk CsPbF3_3

Abstract

Spin-orbit coupling (SOC) in conjunction with broken inversion symmetry acts as a key ingredient for several intriguing quantum phenomena viz. persistent spin textures, topological surface states and Rashba-Dresselhaus (RD) effects. The coexistence of spontaneous polarization and the RD effect in ferroelectric materials enables the electrical control of spin degree of freedom. In light of this, we explore here the ferroelectric lead halide perovskite viz. CsPbF3_3 as a potential candidate in the field of spintronics by employing state-of-the-art first-principles based methodologies viz. density functional theory (DFT) with semi-local and hybrid functional (HSE06) combined with spin-orbit coupling (SOC) and many-body perturbation theory (G0_0W0_0). For a deeper understanding of the observed spin-splitting, we have analyzed the spin textures within the combined framework of DFT and k.p\textbf{k.p} model Hamiltonian. The latter confirms that there is no out of plane spin component indicating that the Rashba splitting dominates over Dresselhaus splitting in this system. Owing to the presence of Pb-6pp orbital in conduction band, the large value of Rashba coefficient (αRα_R) at conduction band minimum (CBm) is noticed in comparison to that of at the valence band maximum (VBM). Notably, we also observe that strength of Rashba spin-splitting can be substancially tuned on application of uniaxial strain (±5%\pm5\%). This finding will further pave the path for perovskite-based spintronics devices.

E-mail: [email protected][PB]; [email protected][SB]

Phone: +91-11-2659 1359

Graphical TOC Entry

Image not extracted

Figure 1.

Lead halide perovskites have evolved as an excellent choice in optoelectronics owing to their exotic properties viz. suitable optical band gap, high absorption coefficient, low trap density and reasonable manufacturing cost[1, 2, 3, 4] [5, 6, 7, 8, 9]. These alluring materials exhibit great applications as an absorber for high power conversion efficiency solar cells10–12. Since last decade, consistent research efforts have led to the revolutionary growth of power conversion efficiency to over 25%13,14. Additionally, the large spin-orbit coupling (SOC) tied to the presence of heavy element Pb plays a decisive role in determining the electronic properties of lead halide perovskites15,16. Over the years, the influence of SOC is well documented on the band structures of these perovskites16. Intriguingly, SOC in conjunction with broken inversion symmetry acts as a key ingredient for several exotic phenomena such as persistent spin textures17,18, topological surface states19 and Rashba-Dresselhaus (RD) effects20–24.

In absence of the inversion symmetry, the crystal feels an effective magnetic field due to SOC. This field coupled with spin moment leads to the momentum dependent splitting of bands, known as Rashba-Dresselhaus (RD) splitting. Dresselhaus25 and Rashba26 effects were originally reported in zinc-blende and wurtzite structures, respectively. The primary difference between both these effects resides in the origin of the noncentrosymmetry. In Rashba effect, there is site inversion asymmetry, whereas bulk inversion asymmetry is present in Dresselhaus effect. During the past decade, RD effect has been the subject of intense research due to its potential applications in the emergent field of spintronics27,28. In view of this, lead halide perovskites have been studied as promising functional materials for spin based applications29–33. Recently, a significant Rashba effect has been reported in tetragonal phase of MAPbI3\mathrm{MAPbI}_{3} (MA = CH3NH3+\mathrm{CH}_{3}\mathrm{NH}_{3}^{+}) due to the rotations of MA ion34,35. The effect has also been predicted in CsPbBr3\mathrm{CsPbBr}_{3} and MAPbBr3\mathrm{MAPbBr}_{3} perovskites36,37. Lately, there are several reports on ferroelectric-coupled Rashba effect in halide perovskites (viz. β\beta-MAPbI3\mathrm{MAPbI}_{3}, β\beta-MASnI3\mathrm{MASnI}_{3}, ortho-MASnBr3\mathrm{MASnBr}_{3}, FASnI3\mathrm{FASnI}_{3} (FA = HC(NH2)2+\mathrm{HC(NH}_{2})_{2}^{+}), opening new possibilities for perovskite based spin devices29,38. However, due to the volatile nature of organic molecules, these perovskites suffer from poor stability towards heat and moisture39.

Therefore, in our present work, we intend to explore inorganic FE perovskites in terms of RD splitting. Note that, the interplay between ferroelectricity and Rashba effect give rise to electric control of bulk Rashba effect, leading to an exciting class of Ferroelectric Rashba Semiconductors (FERSCs)40–42. This effect can be theoretically quantified by employing k.pk.p perturbation theory. According to this theory, the lowest-order RD Hamiltonian is given by43,44

HRD(k)=αR(σxky−σykx)+αD(σxkx−σyky)(1)H_{\mathrm{RD}}(k)=\alpha_{\mathrm{R}}(\sigma_x k_y-\sigma_y k_x)+\alpha_{\mathrm{D}}(\sigma_x k_x-\sigma_y k_y) \tag*{(1)}

Here, σx\sigma_x and σy\sigma_y are Pauli spin matrices, kik_i is the crystal momentum (i=x,y,zi=x,y,z), αR\alpha_{\mathrm{R}} and αD\alpha_{\mathrm{D}} represent the Rashba and Dresselhaus coefficients, respectively. On solving the free-electron Hamiltonian with these terms, we get two spin-split polarized states having opposite spin polarization. Though Rashba and Dresselhaus exhibit a similar type of splitting, projection of spin in momentum space i.e., spin texture gives the idea on the nature of the splitting. Striving towards sizable Rashba-type splitting, CsPbF3_3 perovskite may be a desirable candidate. To the best of our knowledge, the available literature does not encompass the detailed quantitative study of RD effect in this material. The presence of Pb and ferroelectricity in noncentrosymmetric phase of CsPbF3_3 indicates the possibility of RD effect in this material.

Motivated by this idea, we have undertaken a theoretical study based on the perturbative k.pk.p formalism and backed by the first-principles calculations. In the present work, we have studied Pm3ˉmPm\bar{3}m and R3cR3c phases of CsPbF3_3. Firstly, we have examined the electronic band structures in the above mentioned phases. After that, we have estimated the band gap of these phases using first-principles based approaches combined with SOC, viz. density functional theory (DFT)45,46 with semi-local exchange-correlation (εxc\varepsilon_{xc}) functional (PBE47), hybrid DFT with HSE0648,49 and single-shot GW (G0W0G_0W_0)50,51 under the many body perturbation theory (MBPT) and Wannier model approach. Subsequently, we have analyzed the electronic band structure of R3cR3c phase in terms of Rashba splitting under the combined framework of DFT and perturbative k.pk.p formalism. We have also investigated the effect of strain on electronic band gap and Rashba parameters of R3cR3c phase. Finally, we have determined the minimum energy pathway of the FE transition using climbing image nudged elastic band (CINEB) method.

Image not extracted

Figure 1 (PDF p. 5). (a) Optimized crystal structure of CsPbF3\mathrm{CsPbF}_{3} in rhombohedral R3cR3c phase. Cs, Pb and F atoms are indicated by red, grey and yellow colors, respectively. (b) The first hexagonal Brillouin zone showing the high symmetry path for band structure calculations in R3cR3c phase of CsPbF3\mathrm{CsPbF}_{3}. Electronic band structure of CsPbF3\mathrm{CsPbF}_{3} for R3cR3c phase, calculated using (c) PBE (d) PBE+SOC. The conduction and valence bands considered in the discussion are indicated by orange color. (e) Schematic representation of bands showing Rashba splitting. (f) Splitting of conduction band minimum (CBm) and valence band maximum (VBM) of the chosen bands along the M-Γ\Gamma-K path. The inset shows the enlarged view of CBm. (g) Projected density of states (pDOS) in R3cR3c phase of CsPbF3\mathrm{CsPbF}_{3} calculated using HSE06+SOC. The Fermi energy is set to zero in the energy axis.

CsPbF3\mathrm{CsPbF}_{3} mainly manifests in cubic (Pm3‾mPm\overline{3}m) and rhombohedral (R3cR3c) phases52,53. The Pm3‾mPm\overline{3}m phase is centrosymmetric i.e., contains an inversion center. On the other hand, the rhombohedral R3cR3c phase (see Fig. 1(a)) is noncentrosymmetric and exhibits FE behavior due to distortion of cations away from anionic polyhedra53. The calculated change in FE polarization of the rhombohedral (R3cR3c) phase relative to the centrosymmetric structure is 34 μC/cm2\mu\mathrm{C}/\mathrm{cm}^{2} along the [0001][0001] direction in hexagonal setting (along the [111][111] direction in rhombohedral setting). The details of optimized crystal structures are provided in Sec. I of Supplemental Information (SI). Next, we have investigated the electronic band structures of Pm3‾mPm\overline{3}m and R3cR3c phases. The band structure of Pm3‾mPm\overline{3}m phase in presence of SOC reveals that there is no momentum-dependent splitting owing to its centrosymmetric structure (band structure and pDOS of Pm3‾mPm\overline{3}m phase are given in Sec. II of SI).

In addition, the cubic Pm3ˉmPm\bar{3}m phase is not dynamically stable (see § 8 for phonon band structures). Therefore, we have explored RD splitting in the latter phase (i.e. R3cR3c). We have plotted the band structure of FE R3cR3c phase along the high symmetry path using hexagonal setting (as shown in Fig. 1(b)). Firstly, we have performed non-spin polarized calculations. After that we have considered SOC in the calculation of electronic band structure. Fig. 1(c) and 1(d) show the electronic band structures calculated using PBE and PBE+SOC, respectively. A direct band gap of 3.26 eV is observed without SOC, whereas on including SOC, the band gap is reduced to 2.42 eV (indirect) around the Γ\Gamma point. This change in band gap is attributed to SOC arising from the presence of Pb-6p6p orbitals in the conduction band (see pDOS in Fig. 1(g)). Hence, SOC is indispensable and duly considered in all further calculations. Note that PBE εxc\varepsilon_{\mathrm{xc}} functional is well known to underestimate the band gap due to its inability to capture the electron’s self-interaction error. Therefore, we have also calculated the band gap with hybrid εxc\varepsilon_{\mathrm{xc}} functional HSE06 (with SOC), which comes out to be 3.57 eV. Until date, there is no theoretical or experimental report on the band gap of R3cR3c phase. In view of this, for the better estimation of the band gap, we have employed G0W0\mathrm{G}_{0}\mathrm{W}_{0} approximation on top of HSE06+SOC orbitals, that results in the band gap of 5.01 eV (see Figure 5 (a)).

Image not extracted

Figure 5. (a) Climbing image nudged elastic band (CINEB) calculation for the polarization switching process in CsPbF₃ perovskite. Two ferroelectric (FE) structures in the ground state with opposite direction of electric polarization are shown. EbE_b is the activation barrier energy for the polarization switching process. Reversible in-plane spin textures calculated at constant energy E=EF+2.5E=E_F+2.5 eV with opposite spin polarization: (b) -P (c) +P.

Image not extracted

Figure 2. (a) Band structure of R3cR3c phase using G0W0\mathrm{G}_{0}\mathrm{W}_{0}@HSE06+SOC. (b) Imaginary part (Im (ε\varepsilon)) of the dielectric function for R3cR3c phase of CsPbF3_{3} calculated using HSE06+SOC and G0W0\mathrm{G}_{0}\mathrm{W}_{0}@HSE06+SOC.

sequently, we have also plotted the imaginary part (Im (ε\varepsilon)) of dielectric function with converged kk-grid (for kk-grid convergence, see Sec. IV in SI) for R3cR3c phase of CsPbF3_{3} using HSE06+SOC and

G0W0G_0W_0@HSE06+SOC (see Figure 5 (b)). The first peak of Im (ε\varepsilon) corresponds to the electronic band gap, which is the same as we get from band structure calculations. For imaginary part (Im (ε\varepsilon)) of dielectric function of Pm3‾mPm\overline{3}m phase, see Sec. V in SI. The values of band gaps are listed in Table 1.

StructurePBEPBE+SOCHSE06HSE06+SOCG0_0W0_0@HSE06+SOC
Pm3‾mPm\overline{3}m2.621.693.6154,552.714.38
R3cR3c3.262.424.343.575.01

Table 1. Band gap (in eV) of Pm3‾mPm\overline{3}m and R3cR3c phases using different εxc\varepsilon_{\mathrm{xc}} functionals.

It is worth quoting that HSE06+SOC/G0W0G_0W_0@HSE06+SOC only enhances the band gap without any notable change in the nature of band structure and strength of Rashba splitting27,34. To confirm this, we have compared the band structures of Pm3‾mPm\overline{3}m phase using PBE+SOC, HSE06+SOC, G0W0G_0W_0@HSE06+SOC and have found that the band profile remains same (see Sec. VI in SI for details). Therefore, in view of computational cost PBE+SOC is considered to compute RD parameters.

Inclusion of SOC leads to the splitting of bands in kz=πck_z=\frac{\pi}{c} plane, which is perpendicular to the direction of polarization (see Fig. 1(d)). On the contrary, splitting is completely absent along Γ\Gamma-A i.e., in direction parallel to polarization axis, indicating that the momentum dependent splitting around Γ\Gamma point is the Rashba-type splitting. As a consequence, the CBm and VBM are located slightly off the Γ\Gamma point. The CBm shifts from Γ\Gamma towards M and K by 0.0750.075 Å−1^{-1} and 0.0610.061 Å−1^{-1}, respectively. The VBM shifts from Γ\Gamma towards M and K by 0.0220.022 Å−1^{-1} and 0.0120.012 Å−1^{-1}, respectively (see Fig. 1(f)). This shift from Γ\Gamma point in either direction is known as offset momentum (δk\delta k) (see Fig. 1(e)). The difference of energies at Γ\Gamma and extremum point is known as Rashba spin-splitting energy (δE\delta E).

To grasp the overall nature of splitting, the constant energy 2D contour plots of spin texture are plotted in kxk_x-kyk_y plane centered at Γ\Gamma point (see Figure 3(a) for schematic representation of spin textures). Figure 3(b) and 3(c) show xx, yy and zz components of spin at constant energies around CBm and VBM, respectively. As we can see from spin textures of CBm and VBM, the in-plane spin components (SxS_x and SyS_y) are perpendicular to the crystal momentum and out-of-plane component (SzS_z) is completely absent. This results in the helical shape of spin texture with inner and outer bands having opposite orientation, which confirms the existence of dominant Rashba splitting. In order to have a quantitative study of the RD effect, we have considered the k.pk.p Hamiltonian.

Image not extracted

Figure 3. (a) Schematic representation of spin textures in Rashba splitting. Spin projected constant energy contour plots of spin texture calculated in kxk_x-kyk_y plane centered at Γ\Gamma point. The upper and lower panels represent the spin textures calculated at constant energy: (b) E=EFE_\mathrm{F} + 2.5 eV and (c) E=EFE_\mathrm{F} − 0.2 eV, respectively. Electronic band structures showing spin-splitting of (d) CBm and (e) VBM around Γ\Gamma point, respectively. Band structure is plotted along (2πa0.25,0,0)(\frac{2\pi}{a}0.25,0,0) − (0,0,0)(0,0,0) − (2πa0.16,2πb0.16,0)(\frac{2\pi}{a}0.16,\frac{2\pi}{b}0.16,0) direction of momentum space, which is M − Γ\Gamma − K direction. DFT and k.pk.p band structures are plotted with dashed lines and dots, respectively. In the color scale, red depicts spin-up while blue depicts spin-down states.

The model Hamiltonian of R3cR3c structure possessing C3vC_{3v} point group symmetry near Γ\Gamma point in the presence of SOC can be written as56–58:

HΓ(k)=H0(k)+ασykx+βσxky+γσz[(kx3+ky3)−3(kxky2+kykx2)](2)H_{\Gamma}(k) = H_0(k) + \alpha\sigma_y k_x + \beta\sigma_x k_y + \gamma\sigma_z \left[(k_x^3 + k_y^3) - 3(k_x k_y^2 + k_y k_x^2)\right] \tag*{(2)}

Here Ho(k)H_o(k) is the Hamiltonian of the free electrons with eigenvalues Eo(k)E_o(k). α\alpha, β\beta are the coefficients of the linear term and γ\gamma is the coefficient of the higher order term in the Hamiltonian.

Since we have not noticed out-of-plane spin component in the spin texture, therefore, we have neglected the higher order terms in the Hamiltonian (see § 12 in SI for more details). Hence, the Hamiltonian for Γ\Gamma point (considering only linear terms) can now be written as:

HΓ(k)=Ho(k)+αkxσy+βkyσx=Ho(k)+αR(σxky−σykx)+αD(σxkx−σyky)(3)\begin{aligned} H_{\Gamma}(k) &= H_o(k) + \alpha k_x \sigma_y + \beta k_y \sigma_x \\ &= H_o(k) + \alpha_R(\sigma_x k_y - \sigma_y k_x) \\ &\quad+ \alpha_D(\sigma_x k_x - \sigma_y k_y) \tag*{(3)} \end{aligned}

The energy eigen values corresponding to this Hamiltonian is given as, Eo(k)=ℏ2kx22mx∗+ℏ2ky22my∗E_o(k) = \frac{\hbar^2 k_x^2}{2m_x^*} + \frac{\hbar^2 k_y^2}{2m_y^*} and mx∗m_x^*, my∗m_y^* are the effective masses in xx and yy directions, respectively. Rashba and Dresselhaus coefficients are defined as αR=(α−β)/2\alpha_R = (\alpha- \beta)/2 and αD=(α+β)/2\alpha_D = (\alpha+ \beta)/2, respectively59. On diagonalizing Eq. 3, we get energy eigenvalues as:

E±(k)=Eo(k)±α2kx2+β2ky2(4)E_{\pm}(k) = E_o(k) \pm\sqrt{\alpha^2 k_x^2 + \beta^2 k_y^2} \tag*{(4)}

The band structures obtained from DFT and the model Hamiltonian around Γ\Gamma point are shown with dashed lines and dots in Figure 3(d) and 3(e), respectively. The good agreement between both the band structures justifies the reliability of the chosen Hamiltonian. The DFT calculations for CBm give δE=43.5\delta E = 43.5 meV, δk=0.075\delta k = 0.075 Å−1^{-1} along Γ\Gamma-M direction, where ky=0k_y = 0. Hence, the value of α=2δEδk=1.16\alpha= \frac{2\delta E}{\delta k} = 1.16 eVÅ. Along Γ\Gamma-K direction, where kx=kyk_x = k_y, δE=43.5\delta E = 43.5 meV, δk=0.061\delta k = 0.061 Å−1^{-1} and the calculated value of α2+β2=2δEδk=1.43\sqrt{\alpha^2 + \beta^2} = \frac{2\delta E}{\delta k} = 1.43 eVÅ. On putting the value of α\alpha, we get β=−0.84\beta= -0.84 eVÅ, which results in αR=1.00\alpha_R = 1.00 eVÅ and αD=0.16\alpha_D = 0.16 eVÅ. Similarly, for VBM, δE=3.4\delta E = 3.4 meV, δk=0.022\delta k = 0.022 Å−1^{-1} along Γ\Gamma-M direction. So, the value of α=2δEδk=0.31\alpha= \frac{2\delta E}{\delta k} = 0.31 eVÅ. Along Γ\Gamma-K direction, δE=3.4\delta E = 3.4 meV and δk=0.012\delta k = 0.012 Å−1^{-1}. Therefore, α2+β2=2δEδk=0.57\sqrt{\alpha^2 + \beta^2} = \frac{2\delta E}{\delta k} = 0.57 eVÅ, which gives β=−0.48\beta= -0.48 eVÅ. Using α\alpha and β\beta parameters, the estimated values of αR\alpha_R and αD\alpha_D are 0.40 eVÅ and αD=0.09\alpha_D = 0.09 eVÅ, respectively. The values are summed up in Table 2. As we can see from the values of RD parameters, the Rashba effect dominates in the conduction band. The origin of large Rashba effect in CBm can be attributed to the stronger SOC stemming from higher contribution of Pb-6p orbital in conduction band (see Fig. 1(g)). This large contribution of SOC in CBm compared to VBM is in well agreement with the previous findings60. On fitting the DFT band structure around Γ\Gamma point for CBm, we find α=−0.90\alpha= -0.90 eVÅ and β=1.20\beta= 1.20 eVÅ. This gives αR=1.05\alpha_{R} = 1.05 eVÅ and αD=0.15\alpha_{D} = 0.15 eVÅ. Similarly, fitting the band structure for VBM, gives α=−0.33\alpha= -0.33 eVÅ and β=0.49\beta= 0.49 eVÅ and hence αR=0.41\alpha_{R} = 0.41 eVÅ and αD=0.08\alpha_{D} = 0.08 eVÅ. These values are in well agreement with the predicted values based on DFT calculation. The calculated values of RD parameters also confirm that the splitting around Γ\Gamma point is mainly dominated by the Rashba effect.

PositionδE\delta E (meV)δkΓ−M\delta k_{\Gamma-\mathrm{M}} (Å−1^{-1})δkΓ−K\delta k_{\Gamma-\mathrm{K}} (Å−1^{-1})αR\alpha_{\mathrm{R}} (eVÅ)αD\alpha_{\mathrm{D}} (eVÅ)
CBm43.50.0750.0611.050.15
VBM3.40.0220.0120.410.08

Table 2. Rashba parameters for band-splitting at Γ\Gamma point in R3cR3c phase.

After thorough analysis of RD spin-splitting, we have investigated the effect of strain on the band structure of R3cR3c phase. For this, we have applied uniaxial strain in zz-direction, which is defined as

x=c−c0c0×100%(5)x = \frac{c-c_{0}}{c_{0}} \times100\% \tag*{(5)}

where c0c_{0} is the equilibrium lattice constant and cc is the strained lattice constant. The lattice vector cc is varied from -5% to +5%, where “+” and “−” are used to denote tensile and compressive strain, respectively. After optimizing the structures at a given strain, we have plotted the band structures along high symmetry path. Figure 4(a) shows the band structure of R3cR3c phase along Γ\Gamma-M-K-Γ\Gamma-A-L path, under the uniaxial strain of ±5%. To clearly examine the shift, we have also plotted the band structures along M-Γ\Gamma-K (see Figure 4(b)). Here, we have focused on CBm as there is large Rashba splitting in comparison to VBM. From the band structures in Figure 4(b), we infer that strain causes a shift in momentum (δk\delta k) on either side of Γ\Gamma point. The arrows represent the direction of shift. We have taken the equilibrium band structure (i.e., with 0% strain) as the reference for all the strained band structures. Interestingly, under compressive strain, the bands shift off from Γ\Gamma on either side towards M and K. On the contrary, under tensile strain, the bands shift towards Γ\Gamma from either side. This in turn changes δk\delta k and δE\delta E as a function of strain, and the overall effect of these parameters will change αR\alpha_{\mathrm{R}} and αD\alpha_{\mathrm{D}} accordingly. To quantify the effect of strain on Rashba parameters, we have calculated their values at a given strain within the framework of DFT. Figure 4(c-Figure 4e) show δE\delta E, δk\delta k, αR\alpha_{\mathrm{R}} and αD\alpha_{\mathrm{D}} as a function of strain (for details, see § 13 in SI). From Figure 4(c), we have observed that under compressive strain αR\alpha_{\mathrm{R}} is significantly enhanced from 1.05 to 1.48 eVÅ making the material tunable for spintronics application. Also, we have seen notable change in electronic band gap on the application of strain. The band gap values are mentioned in Figure 4(b) (for more quantitative details, see Sec. IX in SI).

Image not extracted

Figure 4. Band structures of CsPbF3_3 (R3cR3c phase) under uniaxial strain of ±5%\pm5\%, calculated using PBE+SOC for path (a) Γ\Gamma-M-K-Γ\Gamma-A-L and (b) M-Γ\Gamma-K. (c) Rashba spin-splitting energy (δE\delta E) (d) offset momentum (δk\delta k) (e) Rashba parameters (αR\alpha_R and αD\alpha_D) as a function of strain. The values are calculated for CBm in R3cR3c phase. Note that δE\delta E is same along both M-Γ\Gamma and Γ\Gamma-K directions.

Lastly, we have explored the possibility of polarization switching in FE CsPbF₃ using strain effects. To predict the feasibility of this phenomenon, we have analyzed the minimum energy pathway of the FE transition using CINEB method61,62. For this, we have chosen the structure deformation path between two FE states with opposite spontaneous polarization through a centrosymmetric reference. As shown in Figure 5, the EbE_b denotes the activation barrier for the polarization switching, which comes out to be 0.75 eV. This suggests that a switchable FE polarization is plausible in the material. The strain tunability and reversible spin textures owing to FE switching in CsPbF3\mathrm{CsPbF}_{3} give electrical control of spins and provide an all-semiconductor design for spintronic devices such as in spin-field effect transistor63,64.

In summary, using first-principles based methodologies viz. PBE, HSE06 and many-body perturbation theory (G0W0G_{0}W_{0}), we have systematically studied the electronic structure of Pm3‾mPm\overline{3}m and R3cR3c phases of CsPbF3\mathrm{CsPbF}_{3}. Note that SOC is duly considered in all the calculations. The R3cR3c phase being noncentrosymmetric is investigated in terms of RD splitting. The DFT calculations are well corroborated by symmetry adapted k⋅p\mathbf{k}\cdot\mathbf{p} model Hamiltonian. The helical nature of in-plane spin components in spin textures indicates the Rashba-type splitting. The calculated values of RD parameters suggest that the Rashba effect is predominant in our material while Dresselhaus effect has a negligible contribution. In addition, a larger Rashba effect is observed in CBm than VBM of R3cR3c phase owing to the larger contribution of Pb-6p orbital in conduction band. Further, we have noticed that the Rashba parameters viz. δk\delta k, δE\delta E, αR\alpha_{\mathrm{R}}, and αD\alpha_{\mathrm{D}} almost linearly increase under compressive strain. The band gap of CsPbF3\mathrm{CsPbF}_{3} is significantly tuned with the application of strain. Also, we have observed reversible spin textures via FE switching. These results indicate that FE CsPbF3\mathrm{CsPbF}_{3} is having enough potential as an excellent amenable material in the field of spintronics.

Computational Methods

The DFT based first-principles calculations have been performed using the Vienna ab initio simulation package (VASP)65,66. The ion-electron interactions in all the elemental constituents are described using projector augmented wave (PAW)67,68 method as implemented in VASP. We have considered pseudopotentials with the following valence states: Cs, 5s25p66s15s^{2}5p^{6}6s^{1}; Pb, 6s25d106p26s^{2}5d^{10}6p^{2}; F, 2s22p52s^{2}2p^{5}. The structures are optimized using Perdew-Burke-Ernzerhof generalized gradient approximation (PBE-GGA), relaxing all ions until Hellmann-Feynman forces are less than 0.001 eV/Å. The cutoff energy of 520 eV is used for the plane-wave basis set such that the total energy calculations are converged within 10−510^{-5} eV. The Γ\Gamma-centered 6×6×66\times6\times6 and 9×9×49\times9\times4 kk-grid are used to sample the irreducible Brillouin zones of cubic phase with Pm3‾mPm\overline{3}m space group and rhombohedral phase with R3‾cR\overline{3}c space group of bulk CsPbF3\mathrm{CsPbF}_{3}, respectively. The phonon calculations are performed for 3×3×33 \times3 \times3 and 2×2×22 \times2 \times2 supercells in Pm3‾mPm\overline{3}m and R3‾cR\overline{3}c phases using PHONOPY package69,70. In order to predict band gap, single-shot GW (G0W0G_{0}W_{0}) calculations have been performed on top of the orbitals obtained from HSE06+SOC εxc\varepsilon_{\mathrm{xc}} functional [G0_{0}W0_{0}@HSE06+SOC]. For this, we have used 6×6×66 \times6 \times6 and 4×4×24 \times4 \times2 kk-grids in Pm3‾mPm\overline{3}m and R3‾cR\overline{3}c phases, respectively. The number of bands is set to four times the number of occupied bands. The polarizability calculations are performed on a grid of 50 frequency points. FE polarization is evaluated in the framework of Berry-phase theory of polarization71,72. The minimum energy pathways of FE transitions are determined through the climbing image nudged elastic band (CINEB) method73. Spin textures are plotted by calculating expectation values of spin operators SiS_i (i=x,y,zi = x, y, z), given by

⟨Si⟩=12⟨Ψk∣σi∣Ψk⟩(6)\langle S_i \rangle= \frac{1}{2}\langle\Psi_k | \sigma_i | \Psi_k \rangle \tag*{(6)}

where σi\sigma_i are the pauli matrices and Ψk\Psi_k is the spinor eigenfunction obtained from noncollinear spin calculations. The spin texture is calculated with closely spaced 12×1212 \times12 kk-grid around high symmetry points. The PyProcar code is used to calculate the constant energy contour plots of the spin texture74.

Acknowledgement

P.B. acknowledges UGC, India, for the senior research fellowship [grant no. 1392/(CSIR-UGC NET JUNE 2018)]. D.G. acknowledges UGC, India, for the senior research fellowship [grant no. 1268/(CSIR-UGC NET JUNE 2018)]. S.S. acknowledges CSIR, India, for the junior research fellowship [grant no. 09/086(1432)/2019-EMR-I]. S.B. acknowledges the financial support from SERB under core research grant (grant no. CRG/2019/000647). We acknowledge the High Performance Computing (HPC) facility at IIT Delhi for computational resources.

Supporting Information Available

Optimized structures of Pm3ˉmPm\bar{3}m and R3cR3c phases; Band structure and pDOS (projected density of states) of cubic Pm3ˉmPm\bar{3}m phase; kk-grid convergence in Pm3ˉmPm\bar{3}m and R3cR3c phases; Imaginary part (Im⁡(ε)\operatorname{Im}(\varepsilon)) of the dielectric function for cubic Pm3ˉmPm\bar{3}m phase; Comparison of PBE+SOC and HSE06+SOC band structures of cubic Pm3ˉmPm\bar{3}m phase; RD parameters as a function of strain in R3cR3c phase; Band gap as a function of strain in R3cR3c phase.

Supplemental Material

I. Optimized structures of Pm3ˉmPm\bar{3}m and R3cR3c phases.

II. Band structure and projected density of states (pDOS) of cubic Pm3ˉmPm\bar{3}m phase.

III. Phonon band structures of Pm3ˉmPm\bar{3}m and R3cR3c phases.

IV. kk-grid convergence in Pm3ˉmPm\bar{3}m and R3cR3c phases.

V. Imaginary part (Im(ε\varepsilon)) of the dielectric function for cubic Pm3ˉmPm\bar{3}m phase.

VI. Comparison of PBE+SOC, HSE06+SOC and G0W0G_{0}W_{0}@HSE06+SOC band structures of cubic Pm3ˉmPm\bar{3}m phase.

VII. Higher order term contribution in SOC Hamiltonian of R3cR3c phase.

VIII. RD parameters as a function of strain in R3cR3c phase.

IX. Band gap as a function of strain in R3cR3c phase.

Optimized structures of Pm3ˉmPm\bar{3}m and R3cR3c phases

Atomxyz
Cs (1a)000
Pb (1b)0.50.50.5
F (3c)00.50.5

Table S1. Structure of optimized Pm3ˉmPm\bar{3}m CsPbF3_3, a=4.80652a=4.80652 Å

Atomxyz
Cs (6a)000
Pb (6b)00.50.5
F (18e)0.4407200.25

Table S2. Structure of optimized R3cR3c CsPbF3_3, a=6.80864a=6.80864 Å, c=16.30620c=16.30620 Å. Hexagonal settings are used.

Band structure and projected density of states (pDOS) of cubic Pmˉ3mP\bar{m}3m

Image not extracted

Figure S1. Band structure of cubic Pmˉ3mP\bar{m}3m phase of CsPbF3_3, calculated using (a) PBE (b) PBE+SOC. (c) Projected density of states (pDOS), calculated using HSE06+SOC.

Phonon band structures of Pm3ˉmPm\bar{3}m and R3cR3c phases

The optimized crystal structures of Pm3ˉmPm\bar{3}m and R3cR3c phases are shown in Figure S2 and S2(b). To analyze the dynamical stability, we have plotted the phonon band structures for optimized Pm3ˉmPm\bar{3}m and R3cR3c phases, respectively. Figure S2 and S2(d) show the phonon band structures calculated along the high-symmetry points of the Brillouin zone. Notably, the negative frequencies can be seen in phonon band structure of cubic phase (see Figure S2) leading to the instability of this phase. The absence of negative frequencies in Figure S2 confirms the dynamical stability of R3cR3c phase.

Image not extracted

Figure S2. Crystal structure of CsPbF3_3 (a) cubic Pm3ˉmPm\bar{3}m phase and (b) rhombohedral R3cR3c phase. Cs, Pb and F atoms are indicated by red, grey and yellow colors, respectively. Phonon band structure of CsPbF3_3 (c) cubic Pm3ˉmPm\bar{3}m phase and (d) rhombohedral R3cR3c phase.

kk-grid convergence in Pm3‾mPm\overline{3}m and R3cR3c phases

Figure S3 (a) and (b) show the variation of imaginary (Im⁡(ε)\operatorname{Im}(\varepsilon)) and real part (Re⁡(ε)\operatorname{Re}(\varepsilon)) of dielectric function for cubic Pm3‾mPm\overline{3}m phase while Figure S3 (c) and (d) show the imaginary (Im⁡(ε)\operatorname{Im}(\varepsilon)) and real part (Re⁡(ε)\operatorname{Re}(\varepsilon)) of dielectric function for R3cR3c phase. On increasing kk-grid, no shift is observed in first peak position of Im⁡(ε)\operatorname{Im}(\varepsilon) part of dielectric function. Hence, 6×6×66 \times6 \times6 and 4×4×24 \times4 \times2 kk-grids are sufficient to compute quasi particle (G0W0G_{0}W_{0}) band gap in Pm3‾mPm\overline{3}m and R3cR3c phases, respectively.

Real and imaginary parts of the dielectric function for $Pm\overline{3}m$ and $R3c$ phases

Figure S3. Real (Re⁡(ε)\operatorname{Re}(\varepsilon)) and imaginary (Im⁡(ε)\operatorname{Im}(\varepsilon)) part of dielectric function for Pm3‾mPm\overline{3}m and R3cR3c phases using PBE εxc\varepsilon_{xc} functional.

Imaginary part (Im⁡(ε))(\operatorname{Im}(\varepsilon)) of the dielectric function for cubic Pm3ˉmPm\bar{3}m phase

Imaginary part of the dielectric function for cubic Pm̄3m phase

Figure S4. Imaginary part (Im⁡(ε))(\operatorname{Im}(\varepsilon)) of the dielectric function for Pm3ˉmPm\bar{3}m phase of CsPbF3_3 calculated using HSE06+SOC and G0_0W0_0@HSE06+SOC fxcf_{\mathrm{xc}} functional, respectively.

Comparison of PBE+SOC, HSE06+SOC and G0_0W0_0@HSE+S band structures of cubic Pm3ˉmPm\bar{3}m phase

Band structures of cubic $Pm\bar{3}m$ phase calculated using PBE+SOC, HSE06+SOC, and G$_0$W$_0$@HSE06+SOC

Figure S5. Band structure of cubic Pm3ˉmPm\bar{3}m phase of CsPbF3_3, calculated using (a) PBE+SOC (b) HSE06+SOC (c) G0_0W0_0@HSE06+SOC, respectively.

Higher order term contribution in SOC Hamiltonian of R3cR3c phase

The model Hamiltonian of R3cR3c structure possessing C3vC_{3v} point group symmetry near Γ\Gamma point in the presence of SOC can be written as:[1, 2, 3, 4]

HΓ(k)=Ho(k)+HSO(7)H_{\Gamma}(\mathbf{k}) = H_o(\mathbf{k}) + H_{SO} \tag*{(7)}

where,

HSO=ασykx+βσxky+γσz[(kx3+ky3)−3(kxky2+kykx2)](8)H_{SO} = \alpha\sigma_y k_x + \beta\sigma_x k_y + \gamma\sigma_z \left[(k_x^3+k_y^3)-3(k_xk_y^2+k_yk_x^2)\right] \tag*{(8)}

and Ho(k)H_o(\mathbf{k}) is free particle Hamiltonian. α\alpha, β\beta are the coefficients of linear terms and γ\gamma is the coefficient of cubic term in SOC Hamiltonian. The energy eigenvalues of Hamiltonian are given as:

E(k)±=ℏ2kx22mx+ℏ2ky22my±ESO(9)E(\mathbf{k})^{\pm} = \frac{\hbar^2 k_x^2}{2m_x} + \frac{\hbar^2 k_y^2}{2m_y} \pm E_{SO} \tag*{(9)}

where, mxm_x and mym_y represent the effective masses in xx and yy directions, respectively. ESOE_{SO} is the energy eigenvalue of SOC Hamiltonian given by ESO(k)=α2kx2+β2ky2+γ2f2(kx,ky)E_{SO}(k) = \sqrt{\alpha^2 k_x^2+\beta^2 k_y^2+\gamma^2 f^2(k_x,k_y)}, where f(kx,ky)=(kx3+ky3)−3(kxky2+kykx2)f(k_x,k_y)=(k_x^3+k_y^3)-3(k_xk_y^2+k_yk_x^2). The normalized spinor wavefunctions corresponding to these energy eigenvalues are given by

Ψk±=eik⋅r2π(ρ±2+1)(iαkx−βkyγf(kx,ky)∓ESO1)(10)\Psi_{\mathbf{k}}^{\pm} = \frac{e^{i\mathbf{k}\cdot\mathbf{r}}}{\sqrt{2\pi(\rho_{\pm}^2+1)}}\begin{pmatrix} \dfrac{i\alpha k_x-\beta k_y}{\gamma f(k_x,k_y)\mp E_{SO}} \\ 1 \end{pmatrix} \tag*{(10)}

where ρ±2=α2kx2+β2ky2(γf(kx,ky)∓ESO)2\rho_{\pm}^2 = \dfrac{\alpha^2 k_x^2+\beta^2 k_y^2}{(\gamma f(k_x,k_y)\mp E_{SO})^2}. The expectation values of spin operators are given by

{sx,sy,sz}±=±1ESO{βky,αkx,γf(kx,ky)}(11)\{s_x,s_y,s_z\}^{\pm} = \pm\frac{1}{E_{SO}}\{\beta k_y,\alpha k_x,\gamma f(k_x,k_y)\} \tag*{(11)}

The in-plane spin components are given by α\alpha and β\beta, whereas out-of-plane spin component can be explained using γ\gamma. The out-of-plane spin component is negligible if γ\gamma is much smaller than α\alpha and β\beta. Since we have not noticed significant out-of-plane spin component in the spin texture, therefore, we had neglected the higher order terms in the Hamiltonian.To confirm the same, we have calculated the value of γ\gamma by fitting the model Hamiltonian. The value of γ\gamma came out to be 0.012 eV A˚\mathrm{eV}\,\AA, which is negligible in comparison to α\alpha and β\beta.

RD parameters as a function of strain in R3cR3c phase

Strain (%)δE\delta E (meV)δkΓ−M\delta k_{\Gamma-\mathrm{M}} (Å−1^{-1})δkΓ−K\delta k_{\Gamma-\mathrm{K}} (Å−1^{-1})αR\alpha_{\mathrm{R}} (eVÅ)αD\alpha_{\mathrm{D}} (eVÅ)
-5109.20.1260.1031.480.25
-496.90.1180.0961.410.23
-383.50.1080.0871.350.21
-270.20.0980.0791.250.19
-156.80.0860.0701.130.18
043.50.0750.0611.000.16
131.00.0620.0500.870.13
221.60.0510.0410.740.11
311.90.0370.0300.550.09
48.70.0320.0250.490.05
53.90.0220.0170.320.02

Table S3. RD parameters as a function of strain for CBm in R3cR3c phase.

Band gap as a function of strain in R3cR3c phase.

Bar chart showing band gap as a function of uniaxial strain

Figure S6. Band gap as a function of uniaxial strain (±5%\pm5\%), calculated using PBE+SOC and HSE06+SOC.

References

  1. [1]Bandyopadhyay, S.; Paul, A.; Dasgupta, I. Origin of Rashba-Dresselhaus effect in the ferroelectric nitride perovskite LaWN₃. Phys. Rev. B 2020, 101, 014109.DOI
  2. [2]Tao, L. L.; Paudel, T. R.; Kovalev, A. A.; Tsymbal, E. Y. Reversible spin texture in ferroelectric HfO₂. Phys. Rev. B 2017, 95, 245141.DOI
  3. [3]Absor, M. A. U.; Santoso, I.; Harsojo,; Abraha, K.; Kotaka, H.; Ishii, F.; Saito, M. Strong Rashba effect in the localized impurity states of halogen-doped monolayer PtSe₂. Phys. Rev. B 2018, 97, 205138.DOI
  4. [4]Sheoran, S.; Kumar, M.; Bhumla, P.; Bhattacharya, S. Rashba dominated spin-splitting in the bulk ferroelectric oxide perovskite KIO₃. arXiv preprint arXiv:2105.08452 2021,DOI

Paper details

vlnc:W_dnkrm3mn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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