Arxiv:1911.01936V1 [Cond-Mat.Mes-Hall] 5 Nov 2019

Symmetry breaking and circular photogalvanic eﬀect in epitaxial Cdx Hg1 – x Te ﬁlms

S. Hubmann,1 G. V. Budkin,2 M. Otteneder,1 D. But,3 D. Sacré,1 I. Yahniuk,3 K. Diendorfer,1 V. V. Bel’kov,2 D. A. Kozlov,4 N. N. Mikhailov,4 S. A. Dvoretsky,4 V. S. Varavin,4 V. G. Remesnik,4 S. A. Tarasenko,2 W. Knap,3 and S. D. Ganichev1 1Terahertz Center, University of Regensburg, 93040 Regensburg, Germany 2Ioﬀe Institute, 194021 St. Petersburg, Russia 3International Research Centre CENTERA, Institute of High Pressure Physics, Polish Academy of Sciences PL-02346 Warsaw, Poland 4Rzhanov Institute of Semiconductor Physics, 630090 Novosibirsk, Russia

We report on the observation of symmetry breaking and the circular photogalvanic effect in Cdx Hg1–x Te alloys. We demonstrate that irradiation of bulk epitaxial films with circularly polarized terahertz radiation leads to the circular photogalvanic effect (CPGE) yielding a photocurrent whose direction reverses upon switching the photon helicity. This effect is forbidden in bulk zinc-blende crystals by symmetry arguments, therefore, its observation indicates either the symmetry reduction of bulk material or that the photocurrent is excited in the topological surface states formed in a material with low Cadmium concentration. We show that the bulk states play a crucial role because the CPGE was also clearly detected in samples with non-inverted band structure. We suggest that strain is a reason of the symmetry reduction. We develop a theory of the CPGE showing that the photocurrent results from the quantum interference of different pathways contributing to the free-carrier absorption (Drude-like) of monochromatic radiation.

I. INTRODUCTION helical surface states becomes possible due to the fact that the proper choice of the amount of Cadmium in the alloy yields an inversion of the band ordering, see Cdx Hg1–x Te alloy, also known as MCT (Mercury Cad- e.g. [17, 19, 34]. Furthermore, in materials with low x the mium Telluride), is one of the most leading materials band structure can be changed from normal to inverted used for sensitive and fast infrared detectors [1–7]. Wide band ordering simply by a variation of the temperature, bandgap tunability of these materials allows radiation T [19]. Such x- and T -driven band inversions give rise detection in an extremely wide frequency range, span- to a large variety of novel physical concepts including ning from near- to mid-infrared wavelength. Further- the Veselago lenses [35], development of long wavelength more, it has been used for the development of tera- lasers with suppression of the Auger processes [36, 37], hertz (THz) radiation detection, see e.g. Refs. [8–12]. etc. The introduction of the concept of topological insulators (TIs) [13–15] attracted great attention to novel as- In all publications discussing the band structure, trans- pects of Cdx Hg1–x Te compounds as well as low dimen- port phenomena, opto-electronic effects, and magneto- sional quantum heterostructures based on these mate- optical properties of Cdx Hg1–x Te films the point group rials. The reason for that is the inverted band struc- symmetry of the crystal is considered to be Td. This fol- ture in HgTe and Cdx Hg1–x Te with a Cadmium con- lows from the crystallographic structure of the system. centration not exceeding the critical value xc [16–20], Here, investigating terahertz radiation induced photo- which is a crucial condition for the formation of heli- galvanic currents in Cdx Hg1–x Te films with an inverted cal surface or edge states [14, 15, 20–25]. In compar- band structure we surprisingly observed a well pro- ison to other materials with a non-trivial band struc- nounced circular photogalvanic effect (CPGE) [38, 39], ture, CdHgTe-based compounds seem to be more promis- whose prerequisite is gyrotropy. Consequently, by sym- ing due to their very high carrier mobility and feasibil- metry arguments it is forbidden in the non-gyrotropic ity to suppress effects from three-dimensional carriers in Td group. Therefore, the CPGE observation indicates the sample volume. This is also supported by the well either (i) the symmetry reduction of the bulk material arXiv:1911.01936v1 [cond-mat.mes-hall] 5 Nov 2019 developed technological process originally motivated by or (ii) excitation of the photocurrent in the topological the fabrication of detectors which has been adapted for surface states formed in the material with low Cadmium the growth of TI materials on demand. The observation concentration. We performed a careful study of terahertz of Kane fermions in bulk Cdx Hg1–x Te crystals [17–19], radiation induced CPGE in bulk Cdx Hg1–x Te films with quantum spin Hall effect [25–27] and helical edge pho- structure compositions similar to that in most of the pre- tocurrents in CdHgTe/HgTe/CdHgTe quantum wells [28] vious studies of topological surface states in this material. as well as the demonstration of Dirac surface states in 3D We show that an attempt to ascribe the generation of the TI made of strained HgTe [29–33], are some important CPGE solely to the helical topological surface states fails, achievements in the physics of topological insulators. In because pronounced CPGE is also detected in crystals the case of bulk Cdx Hg1–x Te crystals the formation of with x above the critical one, which, correspondently, are 2

( a ) ( b ) were prepared, see Fig.1 (c)-(f). In three of these sam- z z y ℏ ω x θ ples (#A,#C, and #D) the active layers with constant C d x H g 1 - x T e x contents were sandwiched (surrounded) by regions with 0 gradually growing Cadmium concentration. In sample C d T e U #B, by contrast, the active layer was capped by 30 nm Z n T e R L ( 0 1 3 ) G a A s s u b s t r a t e Cd0.85Hg0.15Te. Note that due to the Fermi level position free carriers are located in the active layer and, in the ﬁlms with x < xc in the topological surface states. x x

, , The samples were supplied with several indium-soldered n ( c ) S a m p l e # A n 0 . 9 ( d ) S a m p l e # B o o

i 0 . 4 i

t t contacts bonded to a chip holder, see Fig.1 (b). a a r r

t t 0 . 6 As a radiation source for the photocurrent measure-

n 0 . 3 n e e

c c ments two types of molecular gas THz lasers were used:

n n 0 . 3 o 0 . 2 o x = 0 . 1 5 On the one hand, a continuous wave (cw) laser with a

c x = 0 . 1 5 c

d d power P of up to 60 mW [42], on the other hand a pulsed

C 0 . 1 C 0 . 0 0 1 2 3 4 5 0 1 2 3 4 5 laser with a pulse duration of about 100 ns, a repetition T h i c k n e s s , z ( µ m ) T h i c k n e s s , z ( µ m ) rate of 1 Hz and a peak power P of up to 60 kW [43–45]. x x

, , The lasers emitted frequency lines in the range between

n 0 . 5 ( e ) S a m p l e # C n ( f ) S a m p l e # D o o 0.6 and 2.6 THz. The nearly gaussian-shaped beam, con- i i 0 . 4 t t

a 0 . 4 a r r trolled by pyroelectric camera [46], was focused onto the t t n n

e 0 . 3 e sample using a parabolic mirror. The beam spot diam- c c 0 . 3 n n eter varied, depending on the radiation frequency, from o 0 . 2 x = 0 . 1 8 o c c x = 0 . 2 2 1.5 to 3 mm. The initially linearly polarized radiation d d

C 0 . 1 C 0 . 2 was modified applying lambda-quarter plates made from 0 2 4 6 8 1 0 1 2 0 2 4 6 8 crystal quartz. The rotation of these plates by the angle T h i c k n e s s , z ( µ m ) T h i c k n e s s , z ( µ m ) ϕ with respect to the initial laser polarization plane re- FIG. 1. (a) Sketch of the structure composition (not to scale) sulted in the controllable variation of the radiation polar- (b) Experimental setup. (c)-(f) Content of Cadmium in the ization state. By that the degree of circular polarization epitaxial layer as a function of the distance from the CdTe Pcirc was changed according to layer’s top. The values of the Cd content x in the active + − Cdx Hg1–x Te layers for samples #A-#D are given in the plots. Iσ − Iσ Pcirc = + − = sin (2ϕ) , (1) Iσ + Iσ characterized by a normal band ordering. To explain the + − origin of the observed CPGE, we suggest that the studied where Iσ (Iσ ) is the intensity of the right- (left-) epitaxial films are strained and the actual symmetry of handed circularly polarized radiation part. The Stokes the crystal is reduced. In strained zinc-blende-type crys- parameters defining the degrees of linear polarization tals, CPGE may emerge [40]. We develop a microscopic were varied according to [47] theory of the CPGE for the Drude-like indirect optical transitions in bulk crystals induced by terahertz radia- cos(4ϕ) + 1 sin(4ϕ) P L1 = ,P L2 = . (2) tion, which describes well the experimental findings. We 2 2 show that the radiation helicity sensitive photocurrent stems from the interference of virtual transitions via the THz radiation was applied at normal incidence or at conduction and valence bands contributing to the real oblique incidence with an angle of incidence θ. The pho- optical transitions. tocurrent was measured as a voltage drop across a load resistor, see Fig.1 (b), or as the voltage drop over the sample itself. The measurements were carried out in an II. SAMPLES AND METHODS optical cryostat which allowed us to access a temperature range from liquid helium to room temperature.

We studied Cdx Hg1–x Te layers with graded band gap layers at absorber boundaries grown by molecular beam epitaxy on semi-insulating (013)-oriented GaAs III. RESULTS substrates. ZnTn (30 nm thick) and CdTe (6 µm thick) buffer layers [41] were fabricated on top of the GaAs sub- The circular photocurrent was first observed at liquid strates. The structure composition is shown in Fig.1 (a). helium temperature in samples #A and #B with x = The samples were quadratically shaped with approximate 0.15 being characterized by the inverted band structure. dimensions of 5 × 5 mm2, see Fig.1 (b). Four samples Figure2 shows the data obtained at normally incident with different profiles of the Cadmium concentration x radiation of low power cw radiation with frequencies 1.63 3

centrosymmetric bulk Cdx Hg1–x Te crystals polarization 1 0 independent photocurrents as well as those proportional ( a ) x = 0 . 1 5 T = 4 . 2 K to the degree of linear polarization can stem from linear ) 8 s a m p l e # A

W photogalvanic or photon drag eﬀects [38, 39, 48]. These /

A eﬀects are well known for other non-centrosymmetric ma- µ 6 f = 2 . 5 4 T H z ( terials and are out of scope of our paper. By contrast,

P f = 1 . 6 3 T H z

in such crystals the circular photocurrent is forbidden

3 4 6 ) 1 0

J by symmetry arguments. Therefore, below we focus on

W t /

n 2 the origin and properties of this photocurrent. Analyz- A e 5 r µ r (

ing the circular photocurrent further we applied radia- u P c

o / 0 tion at oblique incidence, see inset in Fig.2 (a). This

c f = 2 . 5 4 T H z r o i

c measurements demonstrated that the circular current h J

P - 2 ◦ ◦ - 1 0 ° 0 ° 1 0 ° Jcirc = [J(ϕ = 45 ) − J(ϕ = 135 )]/2 varies with the a n g l e , θ - 4 angle of incidence θ as - 0 . 8 0 Jcirc = Jc cos(θ) + Jc sin(θ) , (4)

) ( b ) x = 0 . 1 5 T = 4 . 2 K W / s a m p l e # B f = 2 . 5 4 T H z Samples #A and #B at liquid helium temperature A

µ have a Cd concentration in the active layer lower than (

- 0 . 9 the critical one and, therefore, are characterized by an /

4 inverted band ordering. The latter results in the for- 1 J

, mation of topological surface states. These states, at t n

e least for samples with an abrupt increase of the Cad- r r

u mium concentration x, like in the top layer of sample

c - 1 . 0 o

t #B, are two-dimensional and, therefore, are character- o

h ized by reduced symmetry. Consequently, in such states P the circular photogalvanic effect becomes possible. To - 1 . 1 prove that formation of the surface states is an unam- 0 ° 4 5 ° 9 0 ° 1 3 5 ° 1 8 0 ° biguous requirement for the generation of the circular P h a s e a n g l e , φ photocurrent in the bulk Cdx Hg1–x Te crystals we carried out measurements at liquid helium temperature in sam- FIG. 2. Photocurrents between contacts 6 and 3 (a) and 1 ples with normal band ordering (sample #D, active layer and 4 (b) [see Fig.1 (b)] as a function of radiation helicity x = 0.22) and with almost linear dispersion (sample #C, measured at liquid helium temperature in Cdx Hg1–x Te crys- active layer with Cadmium concentration x = 0.18 being tals with a Cadmium concentration x ≈ 0.15. The photocur- close to the critical one). Figure3 shows the data ob- rent was excited by THz radiation from a low power cw laser tained applying normally incident radiation of a cw laser (P ≈ 47 mW) operating at frequencies 1.63 and 2.54 THz. Solid curves show fits after Eq. (3). Fitting parameters are: operating at frequencies 0.69, 1.63, and 2.54 THz. In (a): Black curve: Jc/P = 3.6 µA/W, J0/P = 5.1 µA/W, both cases the photocurrent can be well fitted by Eq. (3), JL1/P = −0.3 µA/W, JL2/P = 0.15 µA/W. Red curve: surprisingly, with substantial contribution of the circular Jc/P = 4.5 µA/W, J0/P = 2.2 µA/W, JL1/P = 1.4 µA/W, photocurrent. These results already rule out band inver- JL2/P = 0.26 µA/W. (b): Jc/P = 74 nA/W, J0/P = sion and topological states formation as a prerequisite of 980 nA/W, JL1/P = 13 nA/W, JL2/P = −12 nA/W. The the CPGE in bulk Cdx Hg1–x Te crystals. Moreover, ap- inset in panel (a) shows the dependence of the circular pho- plying the radiation of high power pulsed lasers, which tocurrent Jcirc on the angle of incidence θ. The solid curve is a fit according to Eq. (4). Along the top the polarization increases the sensitivity of the method, we observed that ellipses corresponding to key phase angles ϕ are sketched. the circular photocurrent can clearly be detected even at room temperature. This is shown in Fig.4 for samples and 2.54 THz. The overall polarization dependences can #A and #C with x ≈ 0.15 and x ≈ 0.18. Previous stud- be well fitted by ies of Cdx Hg1–x Te crystals demonstrated that at room temperature all our samples are characterized by a nor- cos(4ϕ) + 1 sin(4ϕ) J = Jc sin(2ϕ) + J0 + JL1 + JL2 . (3) mal band order and no topological states are present. 2 2 Summarizing the experimental part, our experiments In both samples the total photocurrent is dominated by provide clear evidence for the generation of the circular the circular photocurrent described by the first term in photogalvanic effect in bulk Cdx Hg1–x Te crystals with Eq. (3) proportional to the coefficient Jc, and the po- both inverted and normal band orderings, as well as for larization independent offset, J0. Contributions propor- samples with critical Cadmium concentration character- tional to the degrees of linear polarization, while being ized by an almost linear energy dispersion. The fact that present, are substantially smaller. Generally, in non- the CPGE is clearly detected for samples with x larger 4

1 . 4 1 2 ( a ) x = 0 . 2 2 ( a ) σ + x = 0 . 1 5 ) s a m p l e # D )

W s a m p l e # A W

/ + 1 . 3 σ / 1 0 A T = 4 . 2 K A T = 3 0 0 K µ n ( (

P P

/ σ - / - σ 1 . 2 8 3 4 6 1 J J

, , t t n n e e 1 . 1 6 r r r r u u c c o o t t o o 1 . 0 4 h h P P f = 1 . 6 3 T H z f = 2 . 0 2 T H z 0 . 9 2 0 + 5 ( b ) σ x = 0 . 1 8 ( b ) x = 0 . 1 8 ) )

W s a m p l e # C s a m p l e # C W / / 0 + A σ - 2 T = 4 . 2 K A T = 3 0 0 K µ n ( (

P P

/ - /

σ - 5 5 5 2 2 J J -

- 4 σ , , t t n n e e - 1 0 r r r r u u c c o o - 6 t t o o - 1 5 h h P P f = 0 . 6 9 T H z f = 0 . 7 8 T H z - 8 - 2 0 0 ° 4 5 ° 9 0 ° 1 3 5 ° 1 8 0 ° 0 ° 4 5 ° 9 0 ° 1 3 5 ° 1 8 0 ° φ P h a s e a n g l e , φ P h a s e a n g l e ,

FIG. 3. Photocurrents between contacts 1 and 4 (a) and 2 and FIG. 4. Photocurrents between contacts 6 and 3 (a) and 2 and 5 (b) [see Fig.1 (b)] as a function of radiation helicity mea- 5 (b) [see Fig.1 (b)] as a function of radiation helicity measured at liquid helium temperature in Cdx Hg1–x Te crystals sured at room temperature in Cdx Hg1–x Te crystals with an with active layer Cadmium concentrations x = 0.22 (sample active layer Cadmium concentration x = 0.15 (sample #A, #D, f = 1.63 THz) and 0.18 (sample #C, f = 0.69 THz). f = 2.02 THz) and 0.18 (sample #C, f = 0.78 THz). The The photocurrent was excited by THz radiation from a low photocurrent was excited by THz radiation from a high power power cw laser (P ≈ 20 mW for f = 0.69 THz; P ≈ 56 mW pulsed laser (P ≈ 4 kW). The solid curves show fits according for f = 1.63 THz). The solid curves show fits according to to Eq. (3). Fitting parameters are: (a): Jc/P = 1.8 nA/W, Eq. (3). Fitting parameters are: (a): Jc/P = 0.1 µA/W, J0/P = 7 nA/W, JL1/P = −2.3 nA/W , JL2/P = 1.3 nA/W; J0/P = 1.1 µA/W, JL1/P = 208 nA/W, JL2/P = 82 nA/W. (b): Jc/P = 3.2 nA/W, J0/P = −9.9 nA/W, JL1/P = (b): Jc/P = 1.2 µA/W, J0/P = −4.2 µA/W, JL1/P = 9 nA/W, JL2/P = 1 nA/W. −77 nA/W, JL2/P = −2.8 µA/W. The magenta dashed curve in panel (a) shows the corresponding contribution of the circular photocurrent Jc, the blue dashed line the one proportional to the degree of linear polarization PL2. Note that both dashed lines are shifted by an offset of J0. spite the fact that the group lacks the center of space than the critical one demonstrates that the CPGE gener- inversion, it is non-gyrotropic and does not support ation is not limited to the films with topological surface CPGE [38]. However, if the spatial symmetry of the states. crystal is reduced further, CPGE may emerge. The most likely origin of the symmetry reduction in our samples is strain stemming, e.g., from lattice mismatch at the IV. THEORY AND DISCUSSION Cdx Hg1–x Te film interfaces. [49].

A. Symmetry analysis In strained zinc-blende crystals, CPGE can occur [40]. Bulk Cdx Hg1–x Te has nominally zinc-blende crystal To ﬁrst order in strain, the dependence of the CPGE structure which is described by the Td point group. De- current density j on the static strain tensor u is described 5 by coordinate frame Eqs. (5) transforms into

jx = −χ1 sin θ cos ϑ[(uzz − uyy) cos 2φ + 2uyz sin 2φ]IPcirc , jx0 = [χ1(uy0y0 −uz0z0 )êx0 +χ2(ux0y0 eˆy0 −ux0z0 eˆz0 )]IPcirc, sin 2φ j 0 = [χ (u 0 0 −u 0 0 )ê 0 +χ (u 0 0 eˆ 0 −u 0 0 eˆ 0 )]IP , j = cos θ (χ + χ )(u − u ) + χ u cos 2φ y 1 z z x x y 2 y z z x y x circ y 1 2 zz yy 2 2 yz jz0 = [χ1(ux0x0 −uy0y0 )êz0 +χ2(ux0z0 eˆx0 −uy0z0 eˆy0 )]IPcirc, (5) + χ1 sin θ sin ϑ[(uzz − uyy) cos 2φ − uyz sin 2φ] IPcirc , where eˆ = q/q is the unit vector pointing along the pho- jz = − cos θ χ1uyz sin 2φ + sin θ sin ϑ χ2uyz cos 2φ+ ton wave vector q, I is the local intensity of radiation, x0 k [100] y0 k [010] z0 k [001] sin 2φ and , , and are the cubic − (χ1 − χ2)(uzz − uyy) IPcirc , (7) axes. The parameters χ1 and χ2 are linearly indepen- 2 dent and describe the contributions to the photocurrent here ϑ describes position of the plane of incidence with caused by normal and shear strain, respectively [40]. The respect to the crystallographic axes, which in the exper- CPGE current vanishes in the case of hydrostatic strain iment is unknown. One can clearly see that even for ar- that does not disturb the crystal symmetry. Phenomeno- bitrary ϑ contributions of the photocurrent proportional logical Eqs. (5) can be readily constructed using the the- to sin θ are present, which corresponds to J 0 in Eq. (4) ory of group representations. In (013)-grown structures c describing experimental data in the inset in Fig.2(a). experimentally studied in our work, the tensor of strain induced by the lattice mismatch has four non-zero components uxx, uyy, uzz, and uyz in the coordinate frame B. Microscopic theory x k [100], y k [031], and z k [013] relevant to the structure orientation. Typically, the components uxx and uyy at the interfaces are determined by the lattice mismatch between the film and the buffer layer and are equal to j v - each other. The other components, uzz and uyz, can be e found by minimizing the elastic energy, see, e.g., Ref. [33].

) - In the experiment on (013)-grown samples described s - -- t - -

i - above, we study the photocurrent excited by normally n u

θ = 0 eˆ k z . incident radiation ( ). For this geometry, and b ħ ω + ħ Ω r a

Eqs. (5) take the form (

k E

y ε

sin 2φ g F r

jy = (χ1+χ2)(uzz−uyy) +χ2uyz cos 2φ IPcirc, e

2 n E jz = −χ1uyz sin 2φIPcirc, (6)

where φ is the angle between [001] and [013], φ = arctan(1/3) 0 . A substantial photocurrent ex- - k ' - k 0 0 k 0 k ' cited at normal incidence and sensitive to the degree x x x x W a v e v e c t o r , k x ( a r b . u n i t s ) of circular polarization Pcirc has been detected in all samples for all frequencies and temperatures used, see FIG. 5. Photocurrent generation via Drude-like indirect op- ω Figs.2-4 and Jc in Eq. (4). This photocurrent corre- tical transitions. The radiation with frequency is absorbed with simultaneous absorption or emission of a phonon with sponds to the in-plane component of photocurrent jy in frequency Ω. For circularly polarized radiation, the transi- Eq. (6). Note that the geometry of the samples allows tions are asymmetric in the k-space (thin and thick arrows) us to measure only the in-plane component of the current. resulting in a photocurrent. The asymmetry comes from interference of diﬀerent pathways contributing to the transi- In addition to the helicity driven photocurrent at nor- tions, see Fig.6. mal incidence experiment shows that tilting the light by Now we turn to the microscopic mechanism of the pho- an angle of incidence θ results in an additional photocur- tocurrent generation. We consider Cdx Hg1–x Te with the rent proportional to the degree of circular polarization content of Hg below the critical value of the transition to Pcirc being proportional to sin θ, see inset in Fig.2(a). a 3D topological insulator. The samples have the con- To describe the photocurrent at oblique incidence of ra- ventional band structure with the Γ6 conduction band diation one should also take into account all components and the Γ8 valence band. The band gap is larger than of eˆ. For eˆ = (sin θ cos ϑ, sin θ sin ϑ, cos θ) in the x, y, z the photon energy of THz radiation and free carriers are 6 present in the sample. Therefore, the radiation is ab- ( a ) ( c ) ( e ) sorbed via indirect optical transitions (Drude-like) in the )

s + / - + / - + / - t σ σ σ conduction band, see Fig.5. These transitions are as- i n sisted by the scattering of electrons by phonons or static u

. ε

b F defects of the structure to simultaneously satisfy the laws r a ( of energy and quasi-momentum conservation. Indirect k

ε h h optical transitions are described by the second-order per- , y g turbation theory involving virtual processes via interme- r l h e diate states. The matrix element of the real transition n from the initial state i = (k, s), where k is the wave vec- E tor and s is the spin index, to the ﬁnal state f = (k0, s0) is given by the sum of the compound matrix elements of ( b ) ( d ) ( f ) the virtual transitions via all possible intermediate states j ) , s + / - + / - + / - t σ σ σ i n

X VfjRji RfjVji . ε Mfi = + , (8) b F E − E E − E r

i j i j a

j (

ε h h

, where Vfj and Rji are the matrix elements of the electron y g r l h scattering and electron-phonon interaction, respectively, e n and Ej is the total energy of the system in the j state. E The main contribution to radiation absorption comes from the virtual transitions with intermediate states in the conduction band, Fig.6. There are two types of such 0 0 0 processes: the processes where electron-photon interac- W a v e v e c t o r , k x ( a r b . u n i t s ) tion is followed by electron scattering [shown in Fig.6 (a)] FIG. 6. Panels (a) and (b): Virtual intraband optical tran- and the processes with the opposite order [shown in sitions with intermediate states in the conduction band. Red Fig.6 (b)]. The virtual transitions via the conduction- circles and red dashed arrows denote electron-photon interband states describe well the Drude absorption. How- action and electron scattering, respectively. Panel (a) cor- ever, they are not sensitive to the circular polarization responds to the process when photon absorption is followed of the radiation and do not introduce asymmetry in the by electron scattering, while panel (b) sketches the process electron distribution in the k space. in the opposite order. Panels (c)-(f): Virtual intraband optical transitions with intermediate states in the heavy-hole and To obtain the photocurrent, one should also take into light-hole subbands. Red solid and red dashed arrows denote account the virtual transitions with intermediate states interband electron-photon interaction and electron scattering, in the valence band [50]. Figures6 (c)-6 (f) sketch four respectively. The mixing of heavy-hole and light-hole states possible processes of such type via the heavy-hole and the by static strain, essential for the emergence of the CPGE in light-hole bands. Due to the selection rules for interband zinc-blende crystals, is shown as the distortion of the valence- optical transitions, these processes are sensitive to the band spectrum. radiation helicity [51]. Moreover, their contributions can be quite large because the band gap in CdHgTe samples, and correspondingly the denominator in Eq. (8), is small. the photocurrent vanishes in agreement with the symme- The probability of the real transitions (k, s) → (k0, s0) try analysis presented above. The strain-induced mixing is determined by the squared modulus of the matrix ele- of the heavy-hole and light-hole states is schematically ment shown in Fig.6 as a distortion of the valence-band spec- 2 trum. 2 (c) (v) |Mk0s0,ks| = M 0 0 + M 0 0 , (9) k s ,ks k s ,ks To summarize, the microscopic model takes into account two ingredients essential for the CPGE: (i) the lack M (n) where k0s0,ks are the matrix elements of the virtual of space inversion center in the crystal which enables the n transitions via the band. It contains the interference interference of the optical transition pathways via the 2Re[M (c)∗ M (v) ] term k0s0,ks k0s0,ks . The term does not vanish in conduction and valence bands and (ii) the strain-induced non-centrosymmetric crystals and is responsible for the mixing of the states. circular photogalvanic effect [50]. The circular photocurrent emerges only in strained We calculate the photocurrent in the 6-band Kane crystals. Therefore, in the calculation of the matrix ele- model considering the static strain of the crystal and ments of the virtual transitions we also take into account the electron scattering by acoustic phonons. In the basis the mixing of the states by the static strain. Otherwise, of the Γ6 and Γ8 states, the Kane Hamiltonian has the 7 form [52] is the part describing the strain-induced coupling of the Γ Γ ! 6 and 8 states in zinc-blende crystals [51, 53], 02 Hcv H = † , (10) Hcv −EgI4  i 0 0 + 0 0  − y z √ x z 0  2  where 02 is the 2 × 2 zero matrix, I4 is the 4 × 4 identity    r  matrix,  2 iy0z0 + x0z0   i x0y0 − √  †  3 6   k 0 − ik 0  Vcv = Ξcv   , (15) − x √ y 0  r   iy0z0 − x0z0 2   2   √ i 0 0     3 x y   r   6   2 kx0 − iky0   iy0z0 − x0z0   kz0 − √  0 √ †  3 6  2 Hcv = P   , (11)  r  and Ξcv is the interband deformation potential. The  kx0 + iky0 2   √ kz0  Hamiltonian (13) is used to calculate both the mixing  6 3    of the states by the static strain u and the electron scat-  kx0 + iky0  0 √ tering by longitudinal acoustic (LA) phonons. The tensor 2 of strain produced by the LA phonons is given by and P is the Kane parameter. The Hamiltonian (10) s 2 describes six eigenstates: the conduction-band states qαqβ q 2 2 ∗ X ~ iq·r −iq·r † |e, k, ±1/2i with the dispersion εc,k = k /(2m ), the αβ = i 2 e aq − e aq , (16) ~ q 2ρ Ωq states in the light-hole subband |lh, k, ±1/2i with the q 2 2 ∗ dispersion εlh,k = −Eg − ~ k /(2m ), and the disper- sionless states in the heavy-hole subband |hh, k, ±3/2i where q is the phonon wave vector, ρ is the crystal den- ∗ 2 2 with the energy εhh = −Eg. Here, m = 3~ Eg/(4P ) is sity, Ωq = csq is the photon frequency, cs is the speed of † the effective mass. The Hamiltonian of electron-photon longitudinal sound, and aq and aq are the operators of interaction is given by phonon annihilation and creation, respectively. We as- e sume that the photon frequency ω considerably exceeds R = − A · ∇kH, (12) both the frequency of phonons involved in scattering ~c −1 Ω|k0−k| and the scattering rate τ , and that the phonon where e is the electron charge, c is the speed of light, occupation numbers are large, i.e., kBT/(~Ω|k0−k|) 1, A is the vector potential of the electromagnetic field re- where T is the temperature. To the first order in the 2 2 lated to the radiation intensity I by I = A ω nω/(2πc), wave vector, the matrix elements of the virtual transi- and nω is the refractive index of the crystal. The strain tions via the conduction band with the absorption of a Hamiltonian in the 6-band model is given by photon and the simultaneous emission (+) or absorption ! (−) of a LA phonon has the form Ξc Tr () I2 Vcv V = † , (13) V VBP cv ∗ 0 s (c,±) e m Ξc A(k − k ) kBT 0 Mk0s0,ks = ±i 2 δs ,s . (17) where Ξc in the conduction-band deformation potential, c ω 2 ρ cs is the strain tensor, VBP is the Bir-Pikus Hamiltonian which, in the spherical approximation, has the form [52] In calculating the matrix elements of the virtual tran- 5 X sitions via the valence band we take into account the VBP = a + b I4 Tr () − b JαJβ αβ , (14) 4 mixing of the heavy-hole and light-hole stated by the αβ (v) static strain. The corresponding contribution to Mk0s0,ks a and b are the valence-band deformation potentials, Jα proportional to the static strain can be obtained by the are the matrices of the angular momentum 3/2, and Vcv third-order perturbation theory and is given by 8

(±) (±) (v,±) X Vk0s0,kmUkm,knRkn,ks Rk0s0,k0mUk0m,k0nVk0n,ks Mk0s0,ks = + (ε + ω − ε )(ε + ω − ε ) (ε − ε 0 )(ε − ε 0 ) m,n ck ~ mk ck ~ nk ck mk ck nk (±) (±) Vk0s0,kmRkm,knUkn,ks Uk0s,k0mVk0m,knRkn,ks + + 0 (εck + ~ω − εmk)(εck − εnk) (εck + ~ω − εmk)(εck + ~ω − εnk) (±) (±) Uk0s,k0mRk0m,k0nVk0n,kcs Rk0s,k0mVk0m,knUkn,ks + + , (18) (εck + ~ω − εnk0 )(εck − εmk0 ) (εck − εmk0 )(εck − εnk)

where the indexes m and n run over the valence sub- the model of the drifting electron gas is given by bands. The matrix elements of electron scattering with (±) 1 X εk d f(εk) X d f(εk) the emission or absorption of a LA phonon Vk0s0,km and = / εk , (22) τ τ(εk) d εk d εk the matrix elements of heavy-hole–light-hole mixing by k k the static strain Ukm,kn are calculated using the strain which gives Hamiltonian (13). The exact analytical expression for (v) 1 4m∗ k T Ξ2 Mk0s0,ks is too cumbersome to be printed. An estima- B c ¯ = 3 2 k , (23) tion capturing the dependence on the band-structure τ 3π~ ρ cs parameters and deformation potentials is the following (v) 3 p 2 where Mk0s,ks ∼ (e ω u b A P Ξcv/c Eg ) kBT/ρ cs. ¯ X X The photocurrent in the relaxation time approximation k = |k|f(εk)/ f(εk) (24) is given by the standard expression k k

X 2π (c,±) (v,±) 2 is the mean value of |k|. Taking into account Eq. (23) j = eτ |M 0 + M 0 | ∗ 2 2 k s,ks k s,ks and m = 3~ Eg/(4P ), we can rewrite Eq. (20) in the 0 ~ kk s,± form × (vk0 − vk)[f(εck) − f(εck0 )]δ(εck0 − εck − ~ω) , (19) 18 e3 Ξ b χ = − cv τ v = k/m∗ 1 3 2 ¯ 2 where is the momentum relaxation time, ~ is 35π ~ ω c nω k Ξc Eg P the electron velocity, and f(εck) is the equilibrium Fermi- Z p Dirac distribution function. × dε ε(ε + ~ω)(2ε + ~ω)[f(ε) − f(ε + ~ω)] . (25) Labour-consuming calculation of Eq. (19) with the matrix elements (17) and (18) yields the CPGE current (5) In the case when the photon energy ~ω is much less with the coeﬃcients than the mean electron energy ε¯, Eq. (25) takes the form

2 3 5 3 3 512π e τ kBT Ξcv Ξc b P 64π e Neb Ξcv P χ1 = − , χ1 = − 2 2 2 5 2 4 (26) 315 ~ c nω ρ cs (~ω) Eg 35 ~ ω c nω Ξc Eg X 02 2 × [f(ε ) − f(ε 0 )](k + k ) δ(ε 0 − ε − ω) P ck ck ck ck ~ where Ne = 2 k fεk is the electron density. kk0 Finally, the in-plane photocurrent induced at normal (20) incidence of radiation in (013)-oriented ﬁlms is given by and χ2 = −(5/3)χ1. Now using Eq. (6) one can already 3 3 64π e Neb Ξcv P obtain the equation for the in-plane photocurrent which jy = 2 4 × 21 ~ ωcnω Ξc Eg describes well all major experimental ﬁndings. To pro- ceed further, we assume that the momentum relaxation 2 sin 2φ (uzz − uyy) − uyz cos 2φ IPcirc . (27) of electrons is also determined by the electron-phonon 5 2 interaction. For quasi-elastic scattering by LA phonons, This equation describes the observed polarization de- the momentum relaxation time depends on energy and pendence of the photocurrent, its almost linear increase has the form with the frequency decrease as well as drastic increase of 1 m∗ k T Ξ2 r2m∗ε the photocurrent magnitude by cooling the sample from = B c k . 3 2 2 (21) room to liquid helium temperature. Indeed, the propor- τ(εk) π~ ρ cs ~ tionality of the photocurrent Jc to the radiation helic- The relaxation time of the average electron velocity in ity Pcirc has already been addressed above and is clearly 9

5 1 0 4 be formed and this case is relevant to the above derived ) 1 0 # A 3 W equation for the photocurrent. The current density jc / 1 0 T = 4 . 2 K A 2 J

n 1 0 was obtained from the photocurrent c measured in ex-

4 ( T = 3 0 0 K

1 W 1 0 # A

/ 1 0 P periment after jc/I = Jc/P · Sbeam/(d · dbeam), where / 0 c A 1 0 J

n d is the width of the conducting channel, dbeam is the - 1 , 1 0 ) 3 z 1 0 1 2 3 4 5 radiation beam diameter and Sbeam is the beam area. H f ( T H z ) Note that while for room temperature results this calcu- T 1

/ T = 4 . 2 K lation seems to be reasonable at helium temperatures it

# B

f 2

( 1 0

is not straightforward, because free carriers and, conse- × quently, the current may be distributed inhomogeneous P /

c 1 across the sample. The latter unknown factor makes a

J 1 0 # A quantitative comparison of 4.2 K data with the theory difficult. TableI presents the strain u obtained using the # B T = 3 0 0 K 0 2 1 0 parameters 2m0(P/~) = 18.8 eV [55], b = −1.4 eV, and 0 . 1 5 0 . 2 0 0 . 2 5 0 . 3 0 nω = 4.6 [56]. While the rate Ξcv/Ξc is unknown for C d c o n c e n t r a t i o n , x HgTe, we use Ξcv/Ξc = 0.3 for GaAs [53]. TableI shows FIG. 7. Dependence of the circular photocurrent contribu- that the strain u in samples with different compositions tion normalized to radiation frequency on the cadmium con- varies from 1.6×10−7 to 6.9×10−6 and is well below the centration x for temperatures T = 4.2 K (blue symbols) and −3 strain umax = 1.52 × 10 estimated by minimizing the T = 300 K (red symbols). The inset shows the dependence elastic energy for pure HgTe (x = 0) deposited on CdTe of the circular photocurrent on the frequency for sample #A at temperatures T = 4.2 K (blue symbols) and T = 300 K substrate. [33]. We note that a small value of the strain (red symbols). Dashed lines in the inset indicate a fit of Jc/P is not surprising because we deal with thick films. Un- according to Jc/P ∝ 1/f. der this condition strain is expected to be z-coordinate dependent and be strongest at the bottom boundary. Fi- seen in Figs.1-4 obtained for 4.2 and 300 K. The inset nally we note that other possible sources of the CPGE in Fig.7 presenting the circular photocurrent as a func- in the films made of non-gyrotropic crystals are (i) other tion of the radiation frequency shows that it varies after (unrelated to strain) mechanics of bulk symmetry reduc- Jc ∝ 1/ω, which agrees with Eq. 27. Results obtained for tion; (ii) interface related effects, such as 2D electron samples with different Cd contents x and temperatures, states, which may occur at the interfaces or anisotropic are characterized by substantially different energy gaps. scattering of bulk electrons at the interfaces; and (iii) According to Eq. 27 the latter should strongly affect the topological surface states for samples with x < xc.A circular photocurrent magnitude yielding drastic increase photocurrent sensitive to the helicity of incident photons of the current amplitude upon the energy gap reduction can also emerge as a result of the circular-to-linear po- 4 larization conversion in a birefringent medium and the (Jc ∝ 1/Eg ). Figure7 and Tab.I present the magnitude of the circular photocurrent obtained for different linear photogalvanic effect (LPGE). This scenario can be samples and temperatures. Note that to make compari- excluded since the birefringence in the THz range is very son of data obtained at different frequencies possible, we weak while the detected CPGE and LPGE currents are used the fact that Jc ∝ 1/ω and normalized the data to comparable. the radiation frequency. The figure reveals that at fixed temperature, either 300 K or 4.2 K, Jc/P substantially reduces with the band gap increase, see TableI. Further- V. SUMMARY more, comparing the photocurrent magnitudes in each in- dividual sample we see that the photocurrent is increased by more than two orders of magnitude at the temperature To summarize, we demonstrate in our work the symme- decrease from 300 to 4.2 K. This fact is in agreement with try breaking in CdHgTe structures resulting in a helicity- substantial reduction of the band gap at low tempera- sensitive photogalvanic current with opposite directions tures. The values of Eg for different sample compositions for excitation with right- and left-handed circularly po- x and temperatures are taken from Refs. [19, 34, 54]. We larized radiation. The circular photocurrent is present note that for samples #A and #B at 4.2 K, the band gap in films with different Cadmium concentrations as well becomes negative and applicability of Eq. 27 is not justi- as in a wide temperature range, which supports the con- fied. Now we estimate from the room temperature exper- clusion of the strain-induced symmetry reduction. The 2 sin 2φ developed theoretical model describes the experimental iment the strain u = (uzz − uyy) − uyz cos 2φ . 5 2 data in second-order perturbation theory considering the Room temperature data are used because, under these interference of matrix elements in the probability of indi- conditions, all the samples have a normal band disper- rect Drude-like optical transitions involving virtual pro- sion, thus no topologically protected surface states can cesses via intermediate states. 10

f Jc/P jc/I N Eg u Eg Sample x −3 −6 (THz) (nA/W) (µA/W) (cm ) (meV) (10 ) (meV) T = 300 K T = 4.2 K

#A 0.15 0.60 2.9 2.1 1.8 × 1017 83 0.4 -32 #B 0.15 2.03 1.4 1.0 1.3 × 1017 83 0.5 -32 #C 0.18 0.78 0.8 0.8 1.1 × 1017 123 1.6 20 #D 0.22 2.03 0.1 0.2 4.1 × 1016 182 6.9 94 TABLE I. Experimental data for normalized photocurrent, carrier densities, band gaps and the calculated strain for samples #A-#D at 300 K and additionally the band gap values for 4.2 K [54].

VI. ACKNOWLEDGMENTS Y. Wang, Nature Comm. 2, 574 (2011). [16] N. N. Berchenko and M. V. Pashkovskii, Sov. Phys. Usp. 19, 462 (1976). The support from the CENTERA, DFG priority pro- [17] M. Orlita, D. M. Basko, M. S. Zholudev, F. Teppe, gram SFB 1277 (project A04), the Elite Network of W. Knap, V. I. Gavrilenko, N. N. Mikhailov, S. A. Bavaria (K-NW-2013-247), the Volkswagen Stiftung Pro- Dvoretskii, P. Neugebauer, C. Faugeras, A.-L. Barra, gram, and the Program N13 of the Presidium of the RAS G. Martinez, and M. Potemski, Nat. Phys. 10, 233 is gratefully acknowledged. (2014). I. Y. thanks the National Science Centre, Poland, [18] J. D. Malcolm and E. J. Nicol, Phys. Rev. B 92, 035118 (grant No. UMO-2017/25/N/ST3/00408) for support. (2015). [19] F. Teppe, M. Marcinkiewicz, S. S. Krishtopenko, S. Ruf- G.V.B. also acknowledges support from the "BASIS" fenach, C. Consejo, A. M. Kadykov, W. Desrat, D. But, foundation. W. Knap, J. Ludwig, S. Moon, D. Smirnov, M. Orlita, Z. Jiang, S. V. Morozov, V. Gavrilenko, N. N. Mikhailov, and S. A. Dvoretskii, Nat. Commun. 7, 12576 (2016). [20] G. Tomaka, J. Grendysa, M. Marchewka, P. Śliż, C. Becker, A. Stadler, and E. Sheregii, Opto-Electron. [1] P. Capper, Narrow-gap II-VI Compounds for Optoelec- Rev. 25, 188 (2017). tronic and Electromagnetic Applications (Springer US, [21] M. I. Dyakonov and A. V. Khaetskii, JETP Lett. 33, 110 New York City, 1997). (1981). [2] P. Norton, Opto-Electronics review 10, 159 (2002). [22] M. Marchewka, J. Grendysa, D. Żak, G. Tomaka, P. Śliż, [3] M. Henini and M. Razeghi, Handbook of Infra-red Detec- and E. Sheregii, Solid State Commun. 250, 104 (2017). tion Technologies (Elsevier Science, Amsterdam, 2002). [23] G. M. Minkov, A. V. Germanenko, V. A. Larionova, and [4] A. Rogalski, Rep. Prog. Phys. 68, 2267 (2005). O. E. Rut, Phys. Rev. B 54, 1841 (1996). [5] C. Downs and T. Vandervelde, Sensors 13, 5054 (2013). [24] C. L. Kane and E. J. Mele, Phys. Rev. Lett. 95, 226801 [6] A. Rogalski, Infrared and Terahertz Detectors, Third Edi- (2005). tion (Taylor & Francis Ltd, Abingdon, 2018). [25] B. A. Bernevig, T. L. Hughes, and S.-C. Zhang, Science [7] N. Vanamala, K. C. Santiago, and N. C. Das, AIP Ad- 314, 1757 (2006). vances 9, 025113 (2019). [26] M. König, S. Wiedmann, C. Brüne, A. Roth, H. Buh- [8] S. A. Dvoretsky, N. N. Mikhailov, Y. G. Sidorov, V. A. mann, L. W. Molenkamp, X.-L. Qi, and S.-C. Zhang, Shvets, S. N. Danilov, B. Wittman, and S. D. Ganichev, Science 318, 766 (2007). J. Electron. Mater. 39, 918 (2010). [27] K. C. Nowack, E. M. Spanton, M. Baenninger, M. König, [9] V. V. Rumyantsev, D. V. Kozlov, S. V. Morozov, M. A. J. R. Kirtley, B. Kalisky, C. Ames, P. Leubner, C. Brüne, Fadeev, A. M. Kadykov, F. Teppe, V. S. Varavin, M. V. H. Buhmann, L. W. Molenkamp, D. Goldhaber-Gordon, Yakushev, N. N. Mikhailov, S. A. Dvoretskii, and V. I. and K. A. Moler, Nat. Mater. 12, 787 (2013). Gavrilenko, Semicond. Sci. Technol. 32, 095007 (2017). [28] K.-M. Dantscher, D. A. Kozlov, M. T. Scherr, S. Gebert, [10] S. Ruﬀenach, A. Kadykov, V. V. Rumyantsev, J. Torres, J. Bärenfänger, M. V. Durnev, S. A. Tarasenko, V. V. D. Coquillat, D. But, S. S. Krishtopenko, C. Consejo, Bel’kov, N. N. Mikhailov, S. A. Dvoretsky, Z. D. Kvon, W. Knap, S. Winnerl, M. Helm, M. A. Fadeev, N. N. J. Ziegler, D. Weiss, and S. D. Ganichev, Phys. Rev. B Mikhailov, S. A. Dvoretskii, V. I. Gavrilenko, S. V. Mo- 95, 201103 (2017). rozov, and F. Teppe, APL Materials 5, 035503 (2017). [29] C. Brüne, C. X. Liu, E. G. Novik, E. M. Hankiewicz, [11] D. Yavorskiy, K. Karpierz, M. Baj, M. Bąk, N. Mikhailov, H. Buhmann, Y. L. Chen, X. L. Qi, Z. X. Shen, S. C. S. Dvoretsky, V. Gavrilenko, W. Knap, F. Teppe, and Zhang, and L. W. Molenkamp, Phys. Rev. Lett. 106, J. Usakowski, Sensors 18, 4341 (2018). 126803 (2011). [12] M. Bąk, D. Yavorskiy, K. Karpierz, J. Łusakowski, [30] A. M. Shuvaev, G. V. Astakhov, C. Brüne, H. Buh- D. But, J. Przybytek, I. Yahniuk, G. Cywiński, W. Knap, mann, L. W. Molenkamp, and A. Pimenov, Semicond. F. Teppe, S. Krishtopenko, N. Mikhailov, S. Dvoretsky, Sci. Technol. 27, 124004 (2012). and V. Gavrilenko, Acta Phys. Pol. A 134, 973 (2018). [31] D. A. Kozlov, Z. D. Kvon, E. B. Olshanetsky, N. N. [13] J. E. Moore, Nature 464, 194 (2010). Mikhailov, S. A. Dvoretsky, and D. Weiss, Phys. Rev. [14] M. Z. Hasan and C. L. Kane, Rev. Mod. Phys. 82, 3045 Lett. 112, 196801 (2014). (2010). [32] C. Brüne, C. Thienel, M. Stuiber, J. Böttcher, H. Buh- [15] J. Zhang, C.-Z. Chang, Z. Zhang, J. Wen, X. Feng, K. Li, mann, E. G. Novik, C.-X. Liu, E. M. Hankiewicz, and M. Liu, K. He, L. Wang, X. Chen, Q.-K. Xue, X. Ma, and L. W. Molenkamp, Phys. Rev. X 4, 041045 (2014). 11

[33] K.-M. Dantscher, D. A. Kozlov, P. Olbrich, C. Zoth, S. D. Ganichev, Phys. Rev. B 77, 245304 (2008). P. Faltermeier, M. Lindner, G. V. Budkin, S. A. [45] C. Drexler, N. Dyakonova, P. Olbrich, J. Karch, Tarasenko, V. V. Bel’kov, Z. D. Kvon, N. N. Mikhailov, M. Schafberger, K. Karpierz, Y. Mityagin, M. B. Lif- S. A. Dvoretsky, D. Weiss, B. Jenichen, and S. D. shits, F. Teppe, O. Klimenko, Y. M. Meziani, W. Knap, Ganichev, Phys. Rev. B 92, 165314 (2015). and S. D. Ganichev, J. Appl. Phys. 111, 124504 (2012). [34] C. Rigaux, in Narrow Gap Semiconductors Physics and [46] S. Ganichev, Physica B 273-274, 737 (1999). Applications (Springer Berlin Heidelberg, 1980) pp. 110– [47] V. V. Bel’kov, S. D. Ganichev, E. L. Ivchenko, S. A. 124. Tarasenko, W. Weber, S. Giglberger, M. Olteanu, H. P. [35] Y. Betancur-Ocampo and V. Gupta, J. Phys.: Condens. Tranitz, S. N. Danilov, P. Schneider, W. Wegscheider, Matter 30, 035501 (2017). D. Weiss, and W. Prettl, J. Phys. Cond. Matt. 17, 3405 [36] S. V. Morozov, V. V. Rumyantsev, M. A. Fadeev, M. S. (2005). Zholudev, K. E. Kudryavtsev, A. V. Antonov, A. M. [48] B. I. Sturman and V. M. Fridkin, The Photovoltaic and Kadykov, A. A. Dubinov, N. N. Mikhailov, S. A. Dvoret- Photorefractive Eﬀects in Non-Centrosymmetric Materi- sky, and V. I. Gavrilenko, Appl. Phys. Lett. 111, 192101 als (Gordon and Breach Science Publishers, New York, (2017). 1992). [37] V. V. Utochkin, V. Y. Aleshkin, A. A. Dubinov, V. I. [49] Note that in samples with x < xc the CPGE caused Gavrilenko, N. S. Kulikov, M. A. Fadeev, V. V. Rumyant- by the optical transitions in the surface states may con- sev, N. N. Mikhailov, S. A. Dvoretskii, and S. V. Moro- tribute to the total current. A similar current has been zov, Semicond. 53, 1154 (2019). previously observed in HgTe 2D TIs [28] and mechanisms [38] E. L. Ivchenko, Optical Spectroscopy of Semiconductor are overviewed in Ref. [57]. Nanostructures (Alpha Sci. Int. Ltd., Harrow, 2005). [50] S. A. Tarasenko, JETP Lett. 85, 182 (2007). [39] S. D. Ganichev and W. Prettl, Intense Terahertz Exci- [51] E. L. Ivchenko and S. A. Tarasenko, J. Exp. Theor. Phys. tation of Semiconductors (Oxford University Press, Ox- 99, 379 (2004). ford, 2005). [52] G. Bir and G. Pikus, Symmetry and Strain-induced Ef- [40] Y. B. Lyanda-Geller and G. E. Pikus, Fiz. Tverd. Tela fects in Semiconductors (Wiley, New York, 1974). 31, 77 (1989), sov. Phys.-Solid State 31, 2068. [53] G. Pikus, V. Marushchak, and A. Titkov, Sov. Phys. [41] Y. G. Sidorov, S. A. Dvoretskii, V. S. Varavin, N. N. Semicond. 22, 115 (1988). Mikhailov, M. V. Yakushev, and I. V. Sabinina, Semi- [54] J. P. Laurenti, J. Camassel, A. Bouhemadou, conductors 35, 1045 (2001). B. Toulouse, R. Legros, and A. Lusson, J. Appl. [42] P. Olbrich, J. Karch, E. L. Ivchenko, J. Kamann, Phys. 67, 6454 (1990). B. März, M. Fehrenbacher, D. Weiss, and S. D. [55] E. G. Novik, A. Pfeuﬀer-Jeschke, T. Jungwirth, V. La- Ganichev, Phys. Rev. B 83, 165320 (2011). tussek, C. R. Becker, G. Landwehr, H. Buhmann, and [43] S. D. Ganichev, U. Rössler, W. Prettl, E. L. Ivchenko, L. W. Molenkamp, Phys. Rev. B 72, 035321 (2005). V. V. Bel’kov, R. Neumann, K. Brunner, and G. Ab- [56] S. Adachi, Handbook on Physical Properties of Semicon- streiter, Phys. Rev. B 66, 075328 (2002). ductors (Kluwer Academic Publishers, Boston, 2004). [44] W. Weber, L. E. Golub, S. N. Danilov, J. Karch, C. Reit- [57] M. V. Durnev and S. A. Tarasenko, Ann. Phys. 531, maier, B. Wittmann, V. V. Bel’kov, E. L. Ivchenko, Z. D. 1800418 (2019). Kvon, N. Q. Vinh, A. F. G. van der Meer, B. Murdin, and