1. Introduction

Multi-junction solar cells (MJSC) has been the most promising options to improve photoelectric conversion efficiency because of their better adjustment to the sun spectrum [1–3]. III-V compound semiconductor MJSC are used in terrestrial concentrator systerms or space power generation [1–5]. GaInP alloy which has tunable bandgap and can be grown lattice-matched to GaAs plays an extremely important role in the high efficiency MJSC [1–7]. GaInP has been proved to be the best option as the high energy absorption material within the multi-junction structure.

As one of the most important epitaxial techniques, the research of SCs grown by molecular-beam-epitaxy (MBE) has not been extensively investigated, due to the low growth temperature and the presence of isolated defects [8,9]. Solid state MBE growth avoids toxicity of the hydrides and metalorganics used in the deposition process. The strong properties of MBE, good homogeneity, sharp interface and relative simplicity of the growth of even complicated structures, become more important in SC growth. Recent reports have shown that the MBE-grown SCs are comparable to the Metal-organic chemical vapour deposition (MOCVD) growth, which might be owing to the wonderful controllable of phosphorous-containing material growth as well as the beneficial from the high purity material [10–13]. Taking the material quality into account, we hope for a longer diffusion length and a less recombination loss. Therefore, the electrical and optical properties of GaInP are of fundamental interest and practical use. For a typical GaInP SC, which was grown on the p-type GaAs substrate, the absorption of the sun light is mainly by the p-type GaInP base layer. However, the correlation between the electrical and optical properties of p-type GaInP with beryllium (Be) doping for different growth temperatures is scarce.

In this paper, we report the optical and electrical properties of Be-doped p-type GaInP grown by all-solid MBE under different growth temperatures. In contrast to the material grown at high temperature, an abnormal hole mobility of the material grown at low temperature is observed, which is resulting from the two dimensional (2D) holes accumulation at the GaInP/GaAs heterointerface. The PL experimental results have shown that the optical properties of the materials were affected by the holes accumulation at the heterointerface. Two PL peaks and two decay time of the material with 2D holes accumulation was observed when measured at low temperature, while only one PL peak was observed in the material grown at high temperature. Temperature-dependent PL and time-resolved photoluminescence (TRPL) behavior of the materials grown at different growth conditions have shown that holes accumulation would result in the exciton localization effect at the GaInP/GaAs heterointerface.

2. Experiment

The samples of Be doped GaInP on semi-insulating GaAs were grown by Veeco GEN20A dual-chamber all solid-state MBE machine equipped with a valved phosphorous cracker cell and a valved arsenic cracker cell. The growth rate was 1um/h with the V/III ratio of 50. The growth temperatures of samples with GaAs as the buffer layer were 530°C (hereafter high Tg) and 500°C (hereafter low Tg), respectively. During the GaInP growth, a typical 2 × 4 RHEED patterns was clearly observed in all the growth conditions. After growth, the X-ray diffraction (XRD) was performed to ascertain the mole fraction of In. The reflection from the (004) direction was detected and used for the analysis of the ω/2θ peak separation from which the lattice mismatch and composition were deduced. The GaInP samples were lattice matched to GaAs substrate. The mole fraction of In were 0.4795 and 0.4789 for sample A and sample B, respectively. And the lattice-mismatch were 6.55 × 10⁻⁴ and 7.4 × 10⁻⁴ for sample A and sample B, respectively. The result of (004) XRD ω/2θ scans of sample A and B were shown in the inset of Fig. 1. Sample A was grown at high Tg with the doping density of 4.5 × 10¹⁶ cm⁻³. Doping densities for sample B and C grown at low Tg were 5.0 × 10¹⁶ cm⁻³ and 2.2 × 10¹⁷ cm⁻³, respectively. Sample D was also grown at low Tg, but a 200 nm-thick undoped GaInP buffer layer was grown prior to the Be-doped GaInP layer. The thickness of p-type GaInP is 600 nm. The temperature-dependent mobility was performed using the van der Pauw method with He cryostats to permit temperature from 5 K to 300 K in a magnetic field of 5 kG, and the transient PL evolution was measured using a synchroscan streak camera with a time resolution of 15 ps. The continuous wave (CW) PL was excited by the 532 nm line of a semiconductor laser. Transmission electron microscope (TEM) specimens were prepared in cross section [100] orientation.

 

Fig. 1 (a) Room-temperature hole mobility as a function of carrier concentration for Be doped GaInP by MBE growth and Zn-doped Ga_0.5In_0.5P by MOCVD growth; (b) room temperature TRPL spectra of the materials for different growth temperature, which are sample A at high Tg and sample B at low Tg. The inset shows (004) XRD ω/2θ scans of sample A and B.

Download Full Size | PDF

3. Results and discussion

Figure 1(a) shows the relationship between mobility and carrier concentration for a number of GaInP samples at room temperature. For comparison, the mobility of GaInP grown by MOCVD with zinc doping reported in Ref. 14 was also presented. The mobilities of the material grown at high temperature are at the same level as that of MOCVD growth. While, in the case of low Tg the mobilities of the material are much higher than that of the high Tg materials, as well as the materials grown by MOCVD. However, the high mobility of the material grown at low Tg is contradictory to its TRPL results. Figure 1(b) shows the room temperature TRPL spectra of the materials at different growth temperatures. Both of the samples have a single exponential time constant. The decay time of 100 ps is obtained for sample A, while 30 ps is observed for sample B. The decay time of sample C is 20 ps (not shown here), which is similar to sample B (sample B and C are grown at low temperature). The mobility of the material grown at low temperature is high but PL decay time is short, which is a puzzle.

Figure 2 shows the hall mobility and the carrier concentration as a function of temperature. The holes concentration of both samples decreased monotonically with decreasing temperature from 300 K. Sample A and B became depleted at the temperature of 135 K and 160 K, respectively. The variations of the holes concentration of samples A is smaller than that of sample B. And the variation of mobility with temperature of sample A is rather slower. It varies with T^-0.8 while the other two samples vary with T^-2.2 (sample C has similar behavior with sample B, which is not shown here). In III-V compound semiconductors, the Hall mobility is mainly dominated by three scattering processes: The ionized impurity scattering which is proportional to T^3/2 usually at very low temperature, lattice scattering which to T^-3/2 at around room temperature, and alloy scattering which limits the mobility in the form of T^1/2 [15].

 

Fig. 2 (a) Hall mobility and (b) carrier concentration as a function of temperature for samples with different Tg.

Download Full Size | PDF

Since P-type III-V compound semiconductors with zincblende structure have an asymmetric p-like valence band, the heavy holes and the light holes must be taken into account for understanding of the holes mobility. However, in our case, mobility of the three samples does not exactly obey the law. The variation of T^-0.8 is comparable to the reported p-type GaInP where a combination of all the scattering mechanisms are included [16]. In addition, for the material of GaInP, carrier mobility is also affected by the degree of order [17–19]. We performed the TEM measurements to confirm the degree of ordering of the samples. The absence of 1/2{111} superspots in the samples grown at high and low temperatures indicates that disorder matrix dominates in the samples. Furthermore, the Raman spectrum also shows that all the samples are disorder structure, which is in consistent with the fact that GaInP material grown by solid state MBE has very little ordering [20].

The fact that the Hall mobility of the sample grown at the low temperature is about two times larger than that of the material grown at high temperature suggested the existence of two dimensional holes accumulation in the interface, which results in this parasitic conduction. For n-(Al)GaInP and AlGaAs Hall measurement, an unintentionally doped buffer layer of similar composition was usually grown to inhibit the two dimensional electron gas. Based on this idea, sample D with a 200 nm-thick undoped GaInP buffer layer was grown prior to the Be-doped GaInP layer. The undoped GaInP layer was used to eliminate the parasitic conduction at the heterointerface between the semi-GaAs and the p-typed GaInP. And the growth temperature of Be-doped GaInP layer was also 500°C (low Tg). The carrier concentration of 3 × 10¹⁷ /cm³ and the mobility of 20 cm²/vs of sample D were observed, which is in the range of MBE growth at high temperature. This fact indicates that the parasitic conductions observed in samples B and C are due to the effect of holes accumulation within the interface.

The PL experimental results have shown that the optical properties of the materials were also affected by the holes accumulation at the heterointerface. For the TRPL measurement, the short PL decay time indicates the existence of nonradiative center and/or defects in the materials grown at low temperatures. This is in consistent with our previous research in double heterostructures of GaInP with AlInP or AlGaInP barrier layers in Ref. 21, where a fast PL decay time in the case of low growth temperature and then a worse device performance. The holes accumulation at the interface can be reflected from the PL results. Figure 3 shows the temperature-dependent PL for the low growth temperature samples, sample B (a) and C (b). For both of the two samples, there are two PL peaks appearing in the spectra. With an increasing measuring temperature, the intensity of the low-energy PL peak decreases while the high-energy PL peak increases. With an increasing Be doping concentration in sample C at low temperature, the emission from the low-energy PL peak dominates the PL peak. Above 120 K, only the high-energy peak dominates the emission. The inset in Fig. 3 (a) shows the excitation power dependent PL spectra at 3 K. With the increasing excitation power, the low-energy PL peak shows an obvious blue shift to the high energy. The high-energy PL peak shows a small but an abnormal variation, i.e., the PL peak energy first shifts to high energy and then to low energy with increasing measuring temperature. We believe that the PL intensity competition process between localized state excited state leads to the apparently S-shaped variable-temperature PL behavior.

 

Fig. 3 The temperature-dependent PL for samples B (a) and C (b). The inset shows the excitation power dependent PL spectra of sample B at 3 K.

Download Full Size | PDF

For comparison, we performed the temperature-dependent PL measurement on samples A and D of which the mobility is normal, respectively, as shown in Figs. 4(a) and 4(b). In contrast to samples B and C of low-temperature growth, only a single PL peak is observed for both samples A and D. With increasing temperature, the PL peak shows a small but an abnormal “S” shape variation, i.e., the PL peak energy first shifts to low energy then to high energy, as is shown in the inset of Fig. 4(b). Since a similar PL behavior is observed for sample D which was grown at low temperature but with an undoped GaInP buffer, it is reasonable to consider that the low-energy PL appeared in sample B and C is attributed to the interface related localized exciton emission.

 

Fig. 4 The temperature-dependent PL measurement of samples A (a) and D (b).

Download Full Size | PDF

The TRPL spectra of sample A and B at 10 K are shown in Figs. 5(a) and 5(b), respectively. For sample A, only one PL peak is observed, while for sample B a weak PL peak lies in the low energy side of the main PL peak. With the time delay, the dominated PL peaks shift to the low energy side for both the samples, which indicates that the PL peak results from different emissions. The inset in Fig. 5(b) presents PL decay curves of sample B for different emission energies. Different from the room temperature PL decay, the PL decay characteristic can be approximated to be double exponential curve as $I=I_1{e}^{-t/{\tau }_1}+I_2{e}^{-t/{\tau }_2}$, illustrating that two radiative recombination mechanisms exist in the decay process, where $I_1$ and $I_2$ are the PL intensities at t = 0, ${\tau }_1$ and ${\tau }_2$ are two PL decay times, i.e. fast and slow ones, respectively. The PL decay times of the PL peak energy in the range of 1.97-1.99 eV are τ₁ = 380 ps and τ₂ = 770 ps. The PL decay times of τ₁ = 710 ps and τ₂ = 2 ns for the PL peak energy lying in the range of 1.94-1.96 eV are obtained.

 

Fig. 5 TRPL spectra for samples A (a) and B (b), the inset shows the PL decay curves for different emission peaks.

Download Full Size | PDF

The low-energy PL peaks shown in samples B and C which were grown at low temperature together with the abnormal holes mobility indicate the two dimensional holes accumulation at the interface of GaInP/GaAs. Figure 6 presents the calculated band diagram of GaInP/GaAs heterointerface. In the case of the low temperature growth, the transition of the electrons from the conduction band to the localized two dimensional holes well becomes possible due to the holes accumulation (as is shown by the dash line). Holes confined to the barrier exhibit higher mobilities by mitigating the deleterious effect of ionized impurity scattering of the bulk. The holes are free to move at the GaInP/GaAs interface. While in the case of high temperature growth, the enhanced growth temperature results in a rapid diffusion of Be [22]. As a result, no or less effect of the existence of holes accumulation on the electrical and optical properties is observed. The insert of an undoped GaInP buffer also can effectively release the holes accumulation at the interface and result in a normal holes mobility and PL behavior.

 

Fig. 6 The calculated band diagram of GaInP/GaAs heterinterface.

Download Full Size | PDF

4. Conclusions

We report the abnormal carriers’ mobilities of p-GaInP grown on GaAs by MBE at different growth temperature. Samples grown at low temperature have a relatively higher mobility but a shorter PL decay time, which was resulting from the 2D holes accumulation at the GaInP/GaAs heterointerface. Two PL peaks was observed in the spectra of the sample with 2D holes accumulation at low temperature. For the TRPL spectrum at 10 K, the PL decay characteristic was double exponential curve and two radiative recombination mechanisms exist in the decay process. Temperature-dependent PL and TRPL behavior of the materials grown at different growth conditions have shown that holes accumulation would result in the exciton localization effect at the GaInP/GaAs heterointerface.