Magnetic ratchet effect in phosphorene
Narjes Kheirabadi Physics Department, Alzahra University, Vanak, Tehran 1993893973, Iran
Abstract: The magnetic ratchet effect has been studied in phosphorene by the use of the Boltz- mann kinetic equation that is a semi-classical approach. The Hamiltonian of phosphorene in a steady parallel magnetic field is derived using the tight–binding model. We consider the effect of the magnetic field on non–linear dynamics in the presence of an ac laser field and spatial inversion asymmetry. We have shown that for anisotropic 2D materials and phosphorene, the ratchet current has the response to three different light polarizations: linearly polarized light, circularly polarized light, and unpolarized light.
I. INTRODUCTION rials with parabolic dispersion, the circular ratchet effect and the unpolarized ratchet current under an in-plane In a ratchet machine, while a pawl moves upward and magnetic field is possible only if the momentum relax- downward, the wheel rotates in one direction. While the ation time depends on energy [6, 8, 14]. As an exam- magnetic ratchet effect is an effect accordingly a dc cur- ple, for bilayer graphene where Coulomb impurities act rent will be produced in a semiconductor where it is un- like short–range scatters, it is shown that only the linear der an alternating electric field of laser radiation and a magnetic ratchet current is possible [6]. Additionally, the steady magnetic field. Ratchet effects induced by the established circular polarized light detectors rely on chi- in-plane magnetic field were previously studied [1, 2]. It ral organic semiconductors and metal metamaterials [15], has been experimentally observed in graphene, where the however, based on this study we predict that inorganic symmetry is broken by adatoms [3] or superlattice [4] and phosphorene could act as a polarized light detector. This Si-MOSFET [5]. It has also been theoretically predicted point is also valuable to develop a polarization detection for gated bilayer graphene [6, 7] and quantum well [8]. and measurement based on anisotropic 2D materials. However, it is unclear how this effect would appear in an anisotropic material. In this paper, we solve this issue by the study of the magnetic ratchet effect in phosphorene, II. HAMILTONIAN a monolayer of black phosphorus. In phosphorene, each phosphorus atom is covalently The unit cell of phosphorene is depicted in Fig. 1 with bonded with three beside phosphorus atoms. Hence, each armchair (zigzag) edge structure at the x(y) direction. p orbital retains a lone pair of electrons. Because of the According to this figure, there are four atoms in the unit sp3 hybridization, phosphorene does not form an atom- cell, two atoms on the lower layer (A1 and B1), and ically flat sheet like graphene. This property generates two atoms on the upper layer (A2 and B2). Intralayer an intrinsic in-plane anisotropy that results in a specific coupling t1, interlayer couplings t2, t4 and t5, different angle-dependent conductivity [9]. The puckered struc- on-site energies (U1 and U2) and δ, interlayer potential ture of phosphorene deduces to a strong anisotropy in asymmetry, are depicted in Fig. 1. It is worthy to men- electric conducting and it is important to have novel de- tion that the on-site energy difference δ between the po- vices with anisotropic properties [10]. Phosphorene has sitions A1 and B1 (A2 and B2) is introduced and de- also been fabricated and manipulated in the lab [11, 12]. termined by infrared spectroscopy [16]. In addition, the The most remarkable properties of phosphorene are high interalayer hopping parameter t3 is the transfer energy carrier mobility, high optical and UV absorption, strong of B1 atom of one unit cell with A1 atom of the beside in-plane anisotropy, showing a direct bandgap, and other unit cell and it is not depicted in the Fig. 1. Further- attractive properties, which are of particular interest for more, the intralayer distance between atoms in one unit optoelectronic applications [12, 13]. cell is a, and for d as interlayer distance, d0 is the distance Here, we have considered the effect of an ac laser field between B1 and A2 atoms. We also assume that two an- ◦ and an in–plane steady magnetic field on the induced gles α and β are α = A\1B1A1 = B\2A2B2 = 96.5 and arXiv:2008.01561v3 [cond-mat.mes-hall] 9 Jan 2021 ◦ ◦ ◦ ◦ second–order dc current in phosphorene. We have con- β = A\2B1A1 − 90 = 108.04 − 90 = 18.04 . While, the sidered the effect of the spatial asymmetry caused by upper layer is located at d/2, the lower layer is located the disorder and an external gate on the ratchet cur- at −d/2 where d = 2.13A˚ [17, 18]. Finally, we assume rent. In this article, we will show that under asymmetric that phosphorene is under the effect of an in–plane steady spatial disorder, phosphorene produces a dc current that magnetic field, B = (Bx,By, 0), where its vector poten- includes the responses to linearly polarized light, circu- tial is A = z(By, −Bx, 0) chosen to preserve translation larly polarized light, and unpolarized light because the symmetry in the phosphorene plane; z is the Cartesian momentum relaxation time depends on energy. This ef- coordinate perpendicular to the phosphorene plane. fect is observable in monolayer graphene due to the linear Because there are four atoms in the unit cell of phos- dispersion of carriers, as well [3]. However, in 2D mate- phorene, phosphorene tight–binding Hamiltonian is a 2
ay(py + bx) f3= 2t3 cos 2~ i(p − b ) α × exp − x y (2d0 sin β + a cos ), (3) ~ 2 and for the upper layer, we have
0 ay(py − bx) i(px + by) α f1 = 2t1 cos exp a cos , (4) 2~ ~ 2
0 ay(py − bx) f3= 2t3 cos 2~ i(p + b ) α × exp − x y (2d0 sin β + a cos ). (5) ~ 2 Furthermore, we have
i 0 f2 = t2 exp[− pxd sin β], (6) ~
FIG. 1: The side view of four atoms in the unit cell of px 0 α py α f4 = 4t4 cos[ (d sin β + a cos )] cos[ a sin ], (7) phosphorene. A1 and B1 atoms on the lower layer, and ~ 2 ~ 2 A2 and B2 on the upper layer have been depicted. The px 0 intralayer distance between atoms in one unit cell is a, f5 = t5 exp[i (ax − d sin β)]. (8) 0 and d is the distance between B1 and A2 atoms in ~ different layers. Straight lines indicate intralayer Additionally, it is important to work in the low–energy coupling t1. Interlayer couplings t2, t4, t5 are depicted regime. To do so, we make a Taylor expansion of fi by dash lines. Parameters U1, U2, δ indicate different functions in the vicinity of the Γ point. Consequently, we 2 2 on–site energies, as described in the main text. Note can assume that cos x = 1−x /2 and exp x = 1+x+x /2. that x is armchair and y is zigzag direction. In addition, we neglect those terms that are quadratic or higher in the magnetic field.
4×4 matrix, and phosphorene has two conduction bands and two valence bands. To write the tight–binding III. RATCHET CURRENT IN A TWO DIMENSIONAL MATERIAL Hamiltonian of phosphorene in a parallel magnetic field, we use the Peierls substitution. For instance, to deter- mine the Hamiltonian element for a process of hopping According to the perturbation theory, the magnetic dependent valence band is between the in–plane A and B sublattices, HAB, we have determined the following summation over B sites at the h1|V |0i |0ip = |0i + |1i, (9) position RB j 1 − 0 p 3 RA ! where |0i and |0i = |pi are valence band and perturbed X ie Z HAB = t1 exp iK · (RBj − RA) − A.dl . valence band eigenstates, respectively. In this equation, R j=1 ~ Bj |1i is the conduction band eigenstate, and V is that part of the Hamiltonian which includes the magnetic field. Here, K = p/~ is the electron wave vector, −e is the Additionally, 1 and 0 are the conduction band and the charge of electron and dl is the length differential. Con- valence band energies, respectively. By the same method, sequently, we can show that the Hamiltonian of phos- the perturbed conduction band is derived, as well. phorene in the steady magnetic field and in the basis of T We assume that the two dimensional material, phos- (A1,B1,A2,B2) is phorene, is under an in–plane ac electric field with the an- gular frequency ω. It means that E (t) = E exp(−iωt)+ U1 f1 + f3 f4 f2 + f5 k k ∗ ∗ ∗ ∗ E∗exp(iωt), where E = (E ,E ) and E∗ = (E∗,E∗). f1 + f3 U1 + δ f2 + f5 f4 k k x y k x y H = ∗ 0 0 . (1) f4 f2 + f5 U2 + δ f1 + f3 This in–plane radiation changes the electron distribution ∗ ∗ ∗ 0 ∗ 0 ∗ function, so that the electron distribution function is de- f2 + f5 f4 f1 + f3 U2 pendent on the momentum p and t time; f(p, t). For We assume that b = edB/2, in–plane momentum p = homogeneous materials, we use Boltzmann kinetic equa- (px, py, 0) and ax and ay are the length of the unit cell tion, hence, the results of this paper are valid in the semi- into the x and y directions, respectively. For the lower classical regime, ~ω f ; f is the Fermi level [19]. As- layer, we have suming v · ∂f/∂r = 0, v is the electron velocity, we have
ay(py + bx) i(px − by) α ∂f(p, t) f1 = 2t1 cos exp a cos , (2) −eEk · ∇pf(p, t) + = S{f}. (10) 2~ ~ 2 ∂t 3
Collision integral S{f} is For an isotropic material in the absence of the magnetic field X 0 S{f} = [Wpp0 f(p , t) − Wp0pf(p, t)]. −1 X 0 p0 τ|j| ≡ Wp0p[1 − cos(j[φ − φ])] p0 For a perturbed electron gas, the scattering rate is (0) is the relaxation time of the jth angular harmonic of Wp0p = Wp0p + δWp0p, (11) the electron distribution function. However, for an (0) anisotropic 2DEG like phosphorene, it is [20, 21] where Wp0p is the rate of the electron scattering between unperturbed states, and δWp0p is the change of the scat- −1 2π X 0 2 τ (ξ, p)= |hp |δH| pi| δ ( − 0 ) tering rate because of the perturbation. |j|,p p p ~ p0 Additionally, according to the Fermi’s golden rule, the 0 0 [ξ.Vg(p )]τ 0 transition rate between p and p states under a scattering × 1 − |j|,p , potential, δH, is [ξ.Vg(p)]τ|j|,p
2π 0 2 where ξ is the unit matrix of the electric field and Vg is Wp0p = |hp |δH| pi| δ(p − p0 ). (12) ~ the group velocity. In addition, operators in Eq. 14 are where angular brackets indicate an average over impurity e(Ex − iEy) j ∂ positions. Considering static impurities, we can write the αj = − + , following equation for δH 2 p ∂p
Nimp X e(E∗ − iE∗) j ∂ δH = Yˆ u(r − Rj), (13) x y α˜j = − + , j=1 2 p ∂p where Nimp is the number of impurities, u(r − Rj) de- scribes the spatial dependence of the impurity potential, e(E + iE ) j ∂ η = x y + , and Yˆ is a dimensionless matrix describing the structure. j 2 p ∂p We also neglect the interference between different impu- rities, and we use the Fourier transform of the impurity ∗ ∗ potential e(Ex + iEy ) j ∂ η˜j = + , Z 2 p ∂p u(q) = d2ru(r)e−iq.r/~. e l where p = |p|. The factors δSj in Eq. 14 describe the In the scattering rate, we perform a harmonic expansion correction to the scattering caused by the magnetic field. of the impurity potential as described in the following To quantify the dc current caused by an ac elec- equation tric field, it is necessary to determine time–independent 0 asymmetric parts of the distribution function; f±1 har- 0 2 X im0(φ0−φ) | u˜(p − p) | = νm0 e , monics. We assume that electrons are trapped in a huge m0 box with length L and under a periodic potential. For δf = f 0 exp(iφ) + f 0 exp(−iφ), the current density is where φ is the polar angle of momentum and ν−m = νm 1 −1 0 because it is an even function of (φ − φ). g X J = − eV δf, (15) To determine the current by the Boltzmann kinetic L2 g equation (Eq. 10), we consider that f(p, t) is a series p with two indices n, m . So, the distribution function is where g is the spin degeneracy factor (g = 2). expanded in terms of φ and t harmonics with coefficients To solve Eq. 15, the summation over the momentum (n) fm that are functions of , the total energy of an electron vector could be written based on two integrals, one inte- gral over the electron energy, , and another one over the X (n) f(p, t) = fm exp(imφ − inωt), direction of the momentum, φ n,m X Z ∞ Z 2π dφ where m and n are integers and multiplying the Boltz- (...) = L2 Γ()d (...). 2π mann equation by a factor exp (−ijφ + ilωt), where j p 0 0 and l are integers and integrating over a period 2π of an- (l) gle φ and a period of time, t, lead to coupled equations The coupled equations (Eq. 14) for the harmonics fj of between different harmonic coefficients the distribution function are also used to express δf in (0) −1 l l−1 l+1 l−1 terms of the equilibrium distribution function f . In (τ − ilω)f = αj−1f + αj−1f + ηj+1f 0 |j|,p j j−1 e j−1 j+1 addition, we consider that we have a degenerate electron l+1 l +ηej+1fj+1 + δSj. gas, at low–temperature condition; kBT f . Hence, (0) (14) we can assume that ∂f0 /∂ ≈ −δ( − f ). 4
IV. PHOSPHORENE for the conduction and valence bands, we can show that considering δ = U2 = 0 will deduce to a zero ratchet cur- We assume that the band dispersion is equal to , den- rent. Besides, for any amount of δ factor, for a nonzero sity of states per spin per unit area is Γ(), and the U2 amount, C1 is nonzero. However for δ = 0 or U2 = 0, C is equal to zero. group velocity of trapped electrons is Vg = ∇p. To 2 Moreover, we can show that the relevant δSl factors in calculate the change of the scattering rate, δWp0p in j Eq. 11, we use a numerical approach. Accordingly, af- Eq. 14 that describe the correction to scattering caused ter deriving the perturbed eigenstates of the Hamilto- by the magnetic field are nian (Eq. 1) according to Eq. 9, we substitute values of δSl = 0, t (i = 1, 2...5) coupling parameters and lattice param- 0 i l l eters in derived perturbed eigenstate for the conduction δS1= Λ C1By + iC2Bx f2, and valence bands. We consider that t = −1.220eV , l l 1 δS−1= Λ C1By − iC2Bx f−2, t2 = 3.665eV , t3 = −0.205eV , t4 = −0.105eV and l l δS2= Λ C1By − iC2Bx f1, t5 = −0.055eV [17, 18, 22]. We also substitute numerical 0 l l values of ax, ay, α, β and d (Eqs. 2 to 8). To derive Wp0p, δS−2= Λ C1By + iC2Bx f−1, we also consider different forms of Yˆ matrix (Eqs. 12 and where 13). In the symmetric case, where the upper and lower layers are under the effect of disorder, with equal amounts edπn Λ= imp ΩΓ()p, of scattering on the two layers, the disorder matrix is a 2~3 unit matrix. Besides, according to the selected basis that Ω= −(ν0 − ν2). T is (A1,B1,A2,B2) , if we consider that scattering is lim- ited to the lower layer (ζ = 1) or upper layer (ζ = −1), Hence, we can show that the corresponding in–plane cur- the disorder matrix is rent is 1 0 2 2 0 ∗ ∗ Yˆ = Iˆ+ ζσˆ ⊗ Iˆ . Jx= M1,x[By(|Ex| − |Ey| ) + Bx(ExEy + EyEx)] 2 z 0 2 0 ∗ ∗ +M2,xBy |E| + M3,xBxi(ExEy − EyEx), (16) Hence, we derive a general form for Wp0p. This general form that is linear in magnetic field and momentum is 0 2 2 0 ∗ ∗ 2π nimp 0 2 Jy= M1,y[−Bx(|Ex| − |Ey| ) + By(ExEy + EyEx)] Wp0p(U1,U2, δ, ζ)= | u˜(p − p) | δ(p0 − p) ~ L2 +M B0 |E|2 − M B0 i(E E∗ − E E∗), (17) 2,y x 3,y y x y y x × C0(U1,U2, δ, ζ) 0 0 where By = C1By and Bx = C2Bx. Furthermore, M coefficients are the current responses to different light 1 0 + C1(U1,U2, δ, ζ)byp(cos φ + cos φ ) polarizations. M1 is the current response to the linearly ~2 polarized light, M2 is the current response to the unpo- 1 0 + C2(U1,U2, δ, ζ)bxp(sin φ + sin φ ) , larized light, and M3 is the current response to the circu- ~2 larly polarized light. We can show that for phosphorene 2 and anisotropic 2D materials where nimp = Nimp/L is the density of impurities and C0, C1 and C2 are three parameters that change by the 3 ge X −1 ∂ 2τ1,i ∂f0 change of on–site energies (U1, U2 and δ) and disorder M1,i = − 2 Vg,iτ1,iτ2,iΛ + 2 2 , 4L p ∂p 1 + ω τ1,i ∂p types. To estimate C0, C1 and C2 prefactors, we discuss p about the problem numerically. We select U1 = 0, and we consider different values for U and δ parameters. Hence, 3 2 ge X 2 ∂ U is a tunable factor that is defined as the difference in M2,i= − Vg,iτ1,i + 2 4L2 p ∂p the electrostatic potential on the two layers [6, 16]. We p 2 consider δ in the range of 0 to 20meV and U2 in the range 2τ τ Λ 1 − ω τ τ ∂f × 1,i 2,i 1,i 2,i 0 , of 0 to 40meV [6, 16]. Then, we calculate C0, C1 and 2 2 2 2 1 + ω τ1,i 1 + ω τ2,i ∂p C2 based on selected ranges of values for three different disorder types. For the case of symmetric disorder, Yˆ = Iˆ, we can ge3 X 2 ∂ M3,i= − Vg,iτ1,i + show that the ratchet current is equal to zero; C1 = C2 = 4L2 p ∂p 0. Consequently, the symmetry of the upper and lower p layer should be broken by disorder or substrate to have 2ωτ1,iτ2,iΛ τ1,i + τ2,i) ∂f × 0 , a nonzero ratchet current. 2 2 2 2 1 + ω τ1,i 1 + ω τ2,i ∂p For asymmetric disorder types, we can show that by the change of δ in the range of 0 to 20meV and U2 in where Vg,x = p/mxx, Vg,y = p/myy, and mxx and myy the range of 0 to 40meV , the C0 is equal to 1/4 in or- are the effective masses along x (armchair) and y (zigzag) der of 10−2. Additionally, for these two disorder types, directions. 5