CdHgTe-based quantum well (QWs) heterostructures have a number of remarkable properties, first one of which is the possibility of realization of a two-dimensional (2D) topological insulator (TI)^[1]. In this case, conducting edge states arise at the sample edge. The backscattering of charge carriers in these states is suppressed due to the time reversal symmetry^[2]. At present, the physics of TIs is experiencing rapid development: new classes of TIs are predicted (topological crystalline insulators^{[3, 4]}, magnetic Tis^[5], higher-order Tis^[6], etc.), practical applications of TIs are proposed^{[2, 5]}, new materials are discovered^{[7, 8]}, etc. At the same time, despite theoretical predictions of the existence of the 2D TI state^[9−11], and even its experimental implementation both in semiconductor heterostructures^{[1, 12]} and in more exotic systems, such as germanene^[13], monolayer WTe₂^{[14, 15]}, ultrathin Na₃Bi^[16], HgTe quantum wells still remain the most studied 2D TI. This is due to both a more reproducible technology for fabricating structures compared to monolayer/ultrathin materials, and the largest edge mean free path in such objects (up to 10 μm)^{[17, 18]}. The latter allows studying edge transport in fairly large objects. In addition to topological properties, HgTe/CdHgTe heterostructures have other noteworthy features. For example, when varying the thickness of the quantum well, or under external influence^[19], the band spectrum can qualitatively change from normal through a gapless state with a linear dispersion relation to an inverted one (in which the electron-like subbands lie below the hole-like ones) and further to a 2D semimetal^[20].

Both when studying the edge topological states and, in general, the band spectrum, the problem of controlling the concentration of charge carriers and, accordingly, the position of the Fermi level often arises. For example, for the experimental observation of quantized conductivity by edge states (quantum spin Hall effect^[21]), the Fermi level should be inside the bandgap. Otherwise, the edge conductivity will be shunted by the bulk conductivity. The traditional method of controlling the concentration of charge carriers is the fabrication of gated structures^{[1, 17, 20]}. At the same time, the concentration can be controlled in an alternative way—by using the effect of persistent photoconductivity (PPC), i.e., photoinduced conductivity that persists after the termination of the photoexcitation. Such an effect is observed in many heterosystems^[22−31], including CdHgTe-based QW heterostructures^[32−35]. From the practical point of view, the most interesting is the effect of bipolar PPC, in which the concentration of charge carriers can either increase or decrease when the structure is illuminated, depending on the wavelength (see, for example, Refs. [28−32, 34]). Finally, it is necessary to note separately the possibility of changing the type of conductivity due to the PPC effect observed in HgTe/CdHgTe heterostructures^{[32−34, 36, 37]}. The latter is necessary to have the Fermi level located in the bandgap, which, as noted above, is required for studying topological edge states.

Nevertheless, despite the simplicity of using the PPC effect to control the Fermi level position, questions about the efficiency of this method and its 'equivalence' to gate-controlling the charge carrier concentration remain open. In most cases, researchers use a single LED with a fixed wavelength, and the steady-state concentration is determined by the duration of illumination^[36−39]. However, this can lead to completely different results. For example, in normal band structure HgTe/CdHgTe double quantum wells (DQW) with the bandgap (E_g) of ~80 meV, simultaneous coexistence of electrons and holes was observed under short-term (<1 s) illumination with blue light^[38]. The authors of Ref. [38] associated this with the appearance of photoinduced inhomogeneity of charge distribution. At the same time, in another 2D TI—InAs/GaSb QW—the use of a gate resulted in the simultaneous coexistence of the normal and topological phases, while UV illumination resulted in the appearance of only one homogeneous topological phase^[39]. Finally, in a gapless HgTe/CdHgTe QW, green light illumination did not lead to a qualitative difference compared to the use of a gate^[37]. Thus, the use of the PPC effect can lead to different scenarios, so for its proper use, it is desirable to understand the mechanisms of the effect, as well as to know the optimal wavelengths for specific heterostructures.

The most complete information on the mechanisms of the effect can be obtained from spectral studies of the PPC. In this case, it is possible to identify common features of the PPC spectra in heterostructures with different parameters and determine the influence of certain mechanisms by comparing positions of the features with the energy diagrams of the structures. For example, for HgTe/CdHgTe heterostructures, the key influence of the CdTe cap layer on the PPC effect was identified^[32−35]. This made it possible to specify the wavelengths which are most effective for controlling the concentration of charge carriers^[34].

In this paper, detailed studies of the PPC spectra at different temperatures in CdHgTe-based single QW heterostructures with different conductivity types are performed in a wide range of wavelengths. The decisive role of the CdTe cap layer on the PPC effect in these heterostructures is confirmed. A spectral feature previously not discussed in the literature is detected. The emergence of this feature is associated with the presence of a deep state in the CdTe cap layer, located at 0.3 eV above the top of the valence band.

The structures under study were grown by molecular beam epitaxy on a 400-μm-thick semi-insulating GaAs(013) substrate^{[40, 41]}. A buffer consisting of a 30-nm-thick ZnTe layer and a 5-μm-thick relaxed CdTe layer was grown on the substrate. Then, the active part of the structures was grown, including a lower Cd_xHg_1–xTe barrier with the thickness d_bar, an Hg_1–yCd_yTe QW with the thickness d_QW, and an upper Cd_xHg_1–xTe barrier with the thickness d_bar. Some structures were doped with indium, which led to the electron type of conductivity. Also, a number of structures were annealed in an inert gas, which, in general, should increase the concentration of mercury vacancies and, accordingly, increase the concentration of holes^[42]. A CdTe cap layer of thickness d_cap was grown on top of the structure.

The structure diagram is shown in Fig. 1, while the parameters of all the studied structures are given in Table 1. The nominal parameters of the structures were determined during growth by ellipsometry^[40]. For Cd_xHg_1–xTe barriers, in which the cadmium content is sufficiently high (x > 0.5), the error in determining the composition by this method does not exceed 0.03. For a number of structures, the parameters were confirmed or refined by other methods, such as magnetospectroscopy^{[37, 38, 43−45]} or photoconductivity spectroscopy^{[38, 46]}. The type of band structure was either calculated using growth parameters or determined experimentally using the above-mentioned methods.

Therefore, in this work, structures with different types of band spectrum and conductivity, with different thicknesses and compositions of QWs, as well as with different compositions of barrier layers were studied.

To study the PPC, samples with the typical dimensions of 2 × 4 mm² were prepared from the structures. Point indium contacts in the Hall geometry were soldered to the samples. The PPC spectra were recorded using an MDR-206 grating monochromator in the wavelength range of 400–2000 nm (0.62–3.1 eV) at T = 4.2 K or T = 77 K. To eliminate the influence of higher-order diffraction, optical filters made of colored glass were used. The light from the monochromator was guided using polished metal tubes and a reflective mirror. It was focused onto a sample using a cone. The sample itself was fixed at the end of the cone on a DIP8 panel, and the entire assembly was placed in the dark in a helium or nitrogen Dewar transport vessel. KU-1 quartz was used as an input window. The sample current was 1–10 μA, the voltage drop was taken across the potential leads, i.e. the resistance was measured using the four-probe method. The spectra were recorded both with increasing and decreasing wavelengths at a rate of 10 or 5 nm/min.

In addition to recording spectra during continuous scanning with illumination turned on, the resistance kinetics was measured when switching on and off the illumination with a fixed wavelength (point-by-point scanning). In this measurement mode, steady-state resistance values were determined both with the illumination on and after the light was switched off.

Moreover, for some samples, measurements of the quantum Hall effect were carried out at a given illumination wavelength at T = 4.2 K, which made it possible to determine the type and concentration of charge carriers. For this purpose, the samples were placed in a cryostat in the center of a superconducting solenoid. The maximum magnetic field was 4.5–5 T.

The concentration and type of charge carriers listed in Table 1, were determined using the Hall effect in a weak magnetic field of 0.05 T. The measurements were carried out in a metal chamber that completely shielded the sample from background radiation. The values listed were calculated assuming that there was only one type of carriers.

To interpret the spectral features of the PPC, energy diagrams were calculated for all structures. The Laurenti formula^[47] was used for the bandgap values of the Cd_xHg_1–xTe layers. The values of the bandgap offset at the heterointerface and the position of the spin-split band for binary materials CdTe and HgTe were taken from Refs. [19, 48]. Linear interpolation was used for Cd_xHg_1–xTe. The calculation results are presented in the Discussion Section.

Fig. 1 shows the typical kinetics of the structure resistance when switching on and off the illumination for samples with different characteristic resistance values. In the case of the sample 170301_p (Fig. 1(a)), which has a fairly high 'dark' resistance of ~80 kΩ, switching on light with a radiation quantum energy of 2.76 eV leads to an increase in resistance by more than five times (~430 kΩ), while subsequent switching off and on of the illumination causes a change in resistance by no more than 2.5% (Fig. 1(a)). That is, in this case, the negative PPC is observed.

For another sample 091225_n (Fig. 1(b)) with a lower 'dark' resistance of 2.3 kΩ, the PPC effect also takes place. The sample after exposure to light with a quantum energy of 1.32 eV and its subsequent switching off has a steady-state resistance of 1.82 kΩ. Switching on the light with a lower quantum energy of 1.13 eV leads to the establishment of a new resistance value of 0.8 kΩ. Switching off the light leads to a slight 'recovery' of the resistance to 1.1 kΩ. Nevertheless, this value differs significantly from the initial one (1.82 kΩ), which clearly demonstrates the presence of a positive PPC.

Similar behavior for other wavelengths occurs for the sample 100708^*_n, which was additionally annealed (Fig. 1(c)). However, for this particular sample, annealing did not lead to an increase in the hole concentration (Table 1). In this structure, as in the structure 091225_n, turning off the illumination leads to a certain 'recovery' of the resistance. At the same time, this new value (0.56 kΩ) differs significantly from the initial one (0.69 kΩ), which indicates a pronounced effect of the PPC.

As one can see from Fig. 1, the kinetics of resistance changes are complex, they are clearly not described by a simple exponential decay. Therefore, this issue requires separate consideration. At the same time, it can be noted that the times of setting the new values ​​are no more than a couple of tens of seconds, which allows recording the PPC spectra continuously with a fairly slow change in the wavelength of light.

Fig. 2 shows the typical PPC spectra in a structure with the hole-type conductivity, which has a high dark resistance (Fig. 3(a)), and n-type structures with a significantly lower dark resistance (Fig. 3(b) and 3(c)). As can be seen from the figures, the spectra obtained via continuos scanning from lower energies to higher ones and vice versa coincide well. Moreover, the positions of the spectral features always coincided. Therefore, only one curve will be presented in the figures below. The point-by-point scanning more or less replicates continuous recording, and all the characteristic spectral features are reproduced in both cases. This indicates that the sufficiently 'slow' continuous scanning used makes it possible to record the spectra of persistent photoconductivity. Some difference in the resistance values determined with the illumination on (yellow symbols in Fig. 3) and after switching it off (black symbols) indicates that in addition to the PPC, the structures under study also demonstrate 'ordinary' photoconductivity. This also follows from the resistance kinetics (Fig. 1). However, this difference is not critical for detecting and analyzing the main spectral features.

Fig. 3 also shows the values of the electron concentration for a given radiation quantum energy (asterisks in Fig. 3) for the n-type structures. Both structures 091225_n and 100708^*_n retain their electron type of conductivity over the entire energy range. As can be seen from Fig. 3(b) and 3(c), the dependences of the concentration on the energy actually mirror the dependences of the resistance―an increase in the resistance is accompanied by a decrease in the electron concentration, and vice versa. This indicates that the main contribution to the PPC is achieved due to a change in the concentration of charge carriers in the HgTe QW. A similar conclusion was made in Ref. [34] in relation to the DQWs HgTe/CdHgTe heterostructures.

In order to be able to clearly compare the spectra of the PPC structures with very different resistances, one can calculate the relative change in conductivity (1/R – 1/R_dark)/(1/R_dark), where R_dark is the 'dark' value of the structure’s resistance obtained after cooling.

The results of such a comparison for a number of structures are shown in the Fig. 4. As can be seen from the figure, there is a clear correlation between the spectra of structures with the n-type dark conductivity (structures 091217-1_n, 101109_n, 100708^*_n, 091225_n, 100707-1_n) and between the spectra with the p-type dark conductivity (structures 130410^*_p, 170301_p). A number of common features can be identified in the spectra.

Feature 1 is a sharp increase in conductivity in the case of the n-type structures, or a sharp decrease in the case of p-type structures at the energy of about 1.6 eV (vertical dotted line in Fig. 4). The position of this feature is completely independent of the parameters of structures under study.

The second feature is the minimum (maximum) conductivity for n-type (p-type) structures in the energy range of 2.2–2.5 eV (Fig. 4).

Feature 3—a sharp increase (decline) in conductivity for n-type (p-type) structures in the region of 'low' (<1.1 eV) energies. This feature is indicated by solid arrows in Fig. 4. The position of this feature for most structures corresponds to energies of 0.85–0.95 eV, but for some structures (e.g., 100708^*_n) this energy is noticeably lower.

In addition to the above-mentioned features, other, less pronounced features are also observed in the PPC spectra, for example, conductivity oscillations in the range from 0.8–1.1 eV to 1.2–1.5 eV. The same oscillations were also observed in HgTe/CdHgTe DQWs with different cap layers^[35]. Such features will be considered further in the Discussion Section.

The PPC spectra were also measured at T = 77 K. The dependences of the relative change in conductivity are shown in Fig. 5 for a number of structures. As Fig. 5 shows, an increase in temperature does not lead to a qualitative change in the spectra compared to the spectra obtained at T = 4.2 K—there are the same spectral features: a sharp increase (or decrease) in conductivity (features 1 and 3) and a local minimum (or maximum) in conductivity (feature 2). In addition, feature 4 should be separately noted—a fairly narrow local peak (or dip for n-type samples) of conductivity at the radiation quantum energy of ~2.8 eV, which is clearly expressed for almost all structures. The same feature is also present in the spectra obtained at T = 4.2 K (Fig. 4), but it is less pronounced there.

As noted above, the change in conductivity under the illumination in the studied structures is associated, first of all, with a change in the concentration of charge carriers. An increase in the concentration of electrons in the QW leads to an increase in conductivity in n-type samples and, similarly, an increase in the concentration of holes leads to an increase in conductivity in p-type samples. At the same time, it is evident from Fig. 4 and Fig. 5 that the same spectral feature manifests itself differently in n- and p-type structures. For example, features 1 and 3, which represent a sharp increase in conductivity in n-type structures, represent a sharp decline for p-type structures. Similarly, features 2 and 4, the minima of conductivity in n-type structures, are maxima in p-type structures. From this we can conclude that features 1 and 3 are associated with the 'supply' of electrons to the QW, which leads to an increase in the electron concentration in n-type structures and a decrease in the hole concentration in p-type structures, and features 2 and 4 are caused by the 'supply' of holes to the QW.

Previous studies of the PPC effect in HgTe/CdHgTe heterostructures^[32−35] clearly show that the establishment of a new carrier concentration in the QW is determined by the balance of many processes occurring when the structure is illuminated. These processes include the generation of electron−hole pairs or the injection of carriers into the band from deep levels, drift and diffusion, carrier capture in various traps, recombination, etc. In addition to that, one of the main factors is the built-in electric field present in HgTe/CdHgTe heterostructures^{[45, 49, 50]}, which causes carrier separation and determines the value of the steady-state concentration when illuminated with light with a given quantum energy. In fact, it turns out that the concentration and conductivity type of charge carriers in the QW are determined only by the quantum energy of the incident radiation and do not depend on the previous state of the structure.

Since illumination of the structure with light of a given energy causes many different processes to appear, at last leading to a change in the carrier concentration in the QW, we will not consider how exactly this or that concentration is established, but will focus on the 'initial' reason. For this, the spectral position of the observed features will be compared with the energy diagrams of the structures. This will allow us to determine the key influence of certain parameters of the structures on the PPC spectra, which, among other things, has a practical significance.

Let us start with the feature 1—a sharp increase in conductivity for n-type structures at the incident radiation quantum energy of ~1.6 eV (Fig. 4 and Fig. 5). This feature in the PPC spectra of HgTe/CdHgTe heterostructures is well known^[32−35] and is associated with the 'switching on' of electron−hole pair generation in the CdTe cap layer. The band gap of CdTe at T = 4.2 K is 1.605 eV, and at T = 77 K it is 1.593 eV^[47]. As can be seen from the Fig. 4 and Fig. 5, a sharp increase in conductivity begins at energies somewhat lower (~20 meV) than E_g^CdTe, which may be associated with a lower real value of E_g of the cap layer caused by the penetration of mercury atoms from the upper barrier of CdHgTe during growth. To achieve such a reduction in E_g, only 2% mercury penetration is sufficient. An unambiguous connection of the feature 1 to the cap layer material was also shown directly in Ref. [35], where HgTe/CdHgTe heterostructures with different cap layers were studied.

It is necessary to consider the PPC spectra of the structure 170301_p separately. This structure has a dark conductivity of the p-type (Table 1), accordingly, the feature 1 is a conductivity drop. However, this drop is observed at significantly lower energies than E_g^CdTe (the difference is about 90 meV, Fig. 4 and Fig. 5). Apparently, in this particular structure, additional mercury atoms penetrated into the cap layer during the growth process, which led to the actual 'replacement' of the CdTe layer with Cd_0.95Hg_0.05Te.

The feature 2, which appears in the energy range of 2.2–2.5 eV in the form of a wide minimum for n-type structures and in the form of a wide maximum for p-type structures (Fig. 4 and Fig. 5), is caused by a shift in equilibrium to the 'hole' side and is associated with the transition from the spin-split band of the upper barrier of CdHgTe to the conduction band of CdTe (Fig. 6). Such an indirect transition in real space causes generation of an electron in the CdTe layer and a hole in the upper barrier layer of CdHgTe. This may result in the hole having more chances to reach the QW than in the case of 'direct' generation of an electron–hole pair in the CdTe cap layer. This leads to a shift in the balance towards the 'hole' side and the emergence of the feature 2. The hypothesis under consideration was proposed in Ref. [34], where HgTe/CdHgTe heterostructures with a DQW were studied. In that work, an ideal agreement was observed between the spectral position of the feature 2 and the calculated transition from the spin-split band of the upper barrier of CdHgTe to the conduction band of CdTe. The data presented in this work do not contradict this hypothesis, although the agreement between the calculations and the experimental data is somewhat worse (Fig. 7). Nevertheless, from Fig. 7, it is evident that there is a clear correlation between the position of the feature 2 and the cadmium fraction x in the Cd_1–xHg_xTe barrier, which indicates the direct influence of the latter to the formation of the feature 2.

The feature 3 observed in the energy range of 0.65–0.95 eV and not discussed in the literature previously, is a sharp increase in conductivity for structures with the n-type dark conductivity and a decrease for structures with the p-type dark conductivity (Fig. 4 and Fig. 5). Accordingly, this feature is caused by a shift in the equilibrium towards the 'electron' side. The position of this feature, like the feature 2, correlates with the cadmium fraction in the barriers (Fig. 8), but the dependence on x is much stronger. This correlation suggests that CdHgTe barrier layers are also involved in the formation of the feature 3. Several possible transitions can be considered (Fig. 6): a―band-to-band transition in the barrier layer, b―valence band–deep level (or deep level–conduction band) transition in the barrier layer; c—indirect in real space transition from the deep level of the CdTe cap layer to the conduction band of the CdHgTe barrier, d—transition from valence band of the CdHgTe barrier to deep level of the CdTe cap layer. The key point for identifying the desired transition is the slope of the dependence of the position of feature 3 on the cadmium fraction of the barrier.

The energy of transition a turns out to be significantly higher than the position of the feature 3, especially at large x values. At the same time, the steepness of this dependence is stronger than the dependence of the position of the feature 3 on x (Fig. 8(a), straight line a). This discrepancy could be partially compensated by assuming the presence of a transition to some deep level in the CdHgTe barrier layer, lying below the bottom of the conduction band or the top of the valence band by ~0.13 eV (Fig. 8(a), straight-line b). On the other hand this does not solve the 'slope problem'. Of course, one can additionally assume that there is some dependence of the position of this deep level on the barrier composition, which would just compensate the difference in steepness. However, using such a large number of poorly substantiated assumptions seems to be a wrong way. There is a simpler explanation. The steeper dependence of the bandgap in the CdHgTe layer is very well compensated by the dependence of the valence band offset (VBO) at the CdTe–CdHgTe heterointerface. Therefore, the dependence of the transition energy from the deep level lying 0.315 eV above the top of the CdTe valence band to the conduction band of the CdHgTe barrier very well describes the dependences of the position of the feature 3 on x (Fig. 8(a), straight-line c). In this case, the 'mirror' transition from the valence band of the CdHgTe barrier to some deep level of the cap layer lying ~0.6 eV below the bottom of the CdTe conduction band cannot adequately describe the experimental dependence, since the steepness of such a transition is significantly less than necessary (Fig. 8(a), straight-line d).

Thus, the transition to the conduction band of the CdHgTe barrier layer from the deep level of the CdTe cap layer, located 0.315 eV above the top of the valence band and capable of 'supplying' electrons, is the most probable reason for the feature 3. This is also confirmed by the analysis of the position of feature 3 in the PPC spectra measured at T = 77 K (Fig. 8(b)). The above-mentioned deep level energy can be estimated as 0.24–0.39 eV (solid areas in Fig. 8).

In the presence of such a deep level, which manifests itself in photoexcitation processes, it seems reasonable to expect a possible transition from this level directly to the conduction band of CdTe. The energy of such a transition is 1.29 eV at T = 4.2 K and 1.278 eV at T = 77 K. Indeed, in most of the PPC spectra at the designated energies of the incident quantum of radiation, an inflexion is observed, indicating a change in the balance in the 'electron' side (see dash–dotted line on the Fig. 4 and Fig. 5).

Finally, with the active influence of indirect transitions in real space on the spectra of the PPC, it is natural to assume the possibility of a transition from the valence band of the cap layer to the conduction band of the CdHgTe barrier. Such a transition should manifest itself as a feature located 0.315 eV away from the feature 3 toward higher energies. The energy of such transition is indicated in Fig. 4 and Fig. 5 by dashed-dotted arrows. It is evident that in the spectra of most samples, they correspond to small features corresponding to an increase in conductivity.

As for the exact nature of the deep level being discussed, this question remains open and is beyond the scope of this paper. It can only be noted that various sources indicate the presence of states in CdTe with close energies: E_v + 0.28 eV^[51], E_v + 0.32 eV and E_v + 0.4 eV^[52]. However, it remains unclear why exactly this state manifested itself in the PPC spectra. At the same time, the presence of feature 3 in the PPC spectra allows us to claim that the PPC spectroscopy can be used to probe some deep states.

The feature 4 is most pronounced in the PPC spectra measured at T = 77 K (Fig. 5). It is a small minimum of conductivity in n-type structures and a maximum in p-type structures at the radiation quantum energy of 2.8 eV. The position of this feature does not depend on the composition of the barrier layers or on the QW parameters, which indicates that it is somehow related to the CdTe cap layer. This is separately confirmed by the PPC spectra of the structure 170301_p, in which the position of feature 1, unambiguously related to the cap layer, is shifted by 50 meV toward lower energies relative to the feature 1 in other structures (Fig. 4 and Fig. 5). The position of feature 4 is 30 meV lower than the position of feature 4 in the other structures. That is, the feature 4, one way or another, relies to the CdTe cap layer. However, experimental and theoretical studies of the band structure of CdTe^{[53, 54]} do not indicate the presence of any features of the band spectrum corresponding to the energy of 2.8 eV at low temperatures. Therefore, the question of the nature of feature 4 also remains open.

The last but not the least spectral feature is the conductivity oscillations with a change in the quantum energy in the range from 0.8–1.1 eV to 1.2–1.5 eV (Fig. 4, Fig. 5, and Fig. 9). For some structures, these features are strongly pronounced (see, for example, the PPC spectrum of the structure 100708^*_n in Fig. 4 and the inset in Fig. 9). For other structures, these oscillations appear only after 'subtracting' the main trend of the spectral line. Nevertheless, for all structures as a whole, general dependences of the oscillation period on the quantum energy of the incident radiation are observed, regardless of the barrier composition and the QW parameters: in the long-wavelength region, the oscillation period remains almost unchanged and equal to about 40 meV, then, starting from energy values of ~1.0–1.2 eV, it begins to decrease, reaching 15–20 meV in the short-wavelength region. Moreover, the same dependence also takes place for structures with DQWs with different cap layers^[35], which makes it possible to exclude the relation of oscillations to the cap layer (Fig. 9).

Similar oscillations were observed in other heterosystems^{[25, 30]} and in epitaxial films^{[55, 56]}, not only in the PPC spectra, but also in 'traditional' photoconductivity spectra. In most cases, the oscillations were associated with cascade emission of longitudinal optical phonons in polar semiconductors^[57]. In our case, apparently, a similar mechanism takes place. The oscillations begin in the energy range of 0.75–1.1 eV (depending on the structure). These values are very close to the bandgaps of the CdHgTe barrier of the structures under study (from 0.741 eV at T = 4.2 K for the structure 100708_n with x = 0.56 to 1.143 eV for the structure 101109_n with x = 0.77). That is, oscillations of the PPC are observed in the region where quantum energies exceed the bandgaps of the CdHgTe barriers. In this energy range, electron–hole pairs are generated in the barrier layer. As one can see from Fig. 4, this region corresponds to a decrease in conductivity in n-type structures, i.e. the dynamic equilibrium in this region shifts toward the 'hole' side. Obviously, both light and heavy holes can be generated, but only light holes can be generated with energies exceeding the energy of the longitudinal optical phonon, so oscillations in the PPC spectra appear only as a small contribution against the background of the general decline. An increase in the radiation quantum energy leads to an increase in the kinetic energy of the photoexcited hole, which raises its chances of reaching the QW. As soon as its energy reaches the energy of the optical phonon, rapid relaxation occurs and the hole appears at the top of valence band with low kinetic energy. This leads to some change in the dynamic equilibrium, which manifests itself in the form of PPC oscillations. The period of such oscillations should be equal to $\hslash {\omega }_{\text{LO}}\left(1+\frac{m_{\text{e}}}{m_{\text{lh}}}\right)$, where $\hslash {\omega }_{\text{LO}}$—energy of a longitudinal optical phonon (for compositions from 0.55 to 0.77 it is 20–20.5 meV^[58]), m_e and m_lh—effective masses of electrons and light holes, respectively. For CdHgTe, within the specified range of compositions, the effective masses of electrons and light holes near the Г-point of the Brillouin zone are practically the same^[59]. Therefore, the oscillation period should initially be about 40 meV, which is what is observed (Fig. 9). Finally, with the raise of the quantum energy of the incident radiation, the bands nonparabolicity in CdHgTe^[59−61] begins to manifest itself, which leads to a change in the effective masses in the formula above, and, ultimately, to a decrease in the oscillation period (cf. Ref. [25]). We calculated, within the framework of the k·p method (see details, for example, in Ref. [62]), the dependences of the effective masses on the wave vector (and, accordingly, energy) at T = 0 K in bulk Cd_1–xHg_xTe for different values of composition x (solid lines in Fig. 9). As can be seen from Fig. 9, the calculated dependences of the period of the PPC oscillations are in good agreement with the experimental data, which allows us to consider the cascade emission of longitudinal optical phonons as the main process responsible for the PPC oscillations.

Thus, the work shows that the PPC spectra of single HgTe/CdHgTe QWs exhibit common characteristic features, the spectral position of which do not depend on either the growth parameters, or the type of band structure, or the type of dark conductivity. These features are determined by the equilibrium between various competing processes that arise when the structures are illuminated with light with different photon energies. Nevertheless, the spectral position of the features is determined primarily by the presence of a CdTe cap layer in the structures under consideration, as well as by the difference in the composition of the CdHgTe barriers. In addition, spectral studies of the PPC made it possible to identify a transition from a certain deep state in the CdTe cap layer and to determine the energy of this state, which makes it possible to consider the technique used for probing deep levels in the cap layers of various heterostructures in which the PPC effect is present.