Photoluminescence intensity change of GaP_1−xN_x alloys by laser irradiation Open Access

Dilute nitride semiconductors have attracted much attention because they have unique properties, such as the large bandgap bowing^1–7 and the splitting of conduction bands,^8,9 which are beneficial for improving the performance of the next-generation optoelectronic devices. Intermediate-band solar cells (IBSCs)¹⁰ are promising candidates for the next generation photovoltaic devices that seek to improve the conversion efficiency.

Among dilute nitride semiconductors, GaP_1−xN_x alloys are known for their large bandgap bowing along with other modifications to the band structure and have attractive potential applications in optical devices such as light-emitting diodes in monolithic optoelectronic integrated circuits^11,12 and high efficiency IBSC.^13,14 Apart from a general expectation for the intermediate band, up-conversion photoluminescence (PL)¹⁵ via deep levels is also useful for the enhancement of the conversion efficiency of solar cells.

Owing to the difference in the crystal structure and lattice parameter between GaP and GaN, there is an extreme miscibility gap so that the growth of GaP_1−xN_x alloys requires nonequilibrium conditions, and several works concerning the nonequilibrium epitaxial growth of GaP_1−xN_x alloys by metalorganic vapor phase epitaxy (MOVPE)^4,16 and molecular beam epitaxy (MBE)^1,3,17 have been reported. During the growth process, the required lower temperatures for incorporating nitrogen atoms together with the disparity between N and the replaced group-V atoms are known to favor the formation of various defects in GaP_1−xN_x alloys that act as nonradiative recombination centers leading to the performance degradation of optoelectronic devices. Over the past few years, considerable theoretical and experimental efforts have been directed toward the identification and understanding of defects that are major obstacles for wide-spread device applications. In particular, it is of great importance to understand relevant defects generated or multiplied while the device is operating.

In this study, our aim is focused on the nonradiative recombination centers that cause the change in the luminescence efficiency in GaP_1−xN_x. To obtain this goal, we have measured in real time the PL intensity change in GaP_1−xN_x with various N concentrations under laser irradiation.

The samples used in this study were GaP_1−xN_x alloys grown on GaP (001) substrates by low-pressure MOVPE. Trimethylgallium, phosphine, and 1,1-dimethylhydrazine were used for the Ga, P, and N sources, respectively. Details concerning the MOVPE growth procedure have been described elsewhere.⁴ The samples are comprised of a 200 nm thick GaP buffer layer and a 450 nm thick GaP_1−xN_x layer. The growth temperature was 630 °C–700 °C. The nitrogen concentration was determined by x-ray diffraction, assuming that GaP_1−xN_x obeys Vegard’s law.

Micro-PL measurements were performed to obtain high excitation power density at room temperature. A diode pumped solid state laser (λ = 532 nm), focused to ∼1 µm in diameter, was used as an excitation light source. To inspect the change in the PL intensity, each sample was exposed to the laser with power densities ranging from 0.035 MW/cm² to 11.5 MW/cm². PL spectra were measured with a 30-cm monochromator and an intensified charge coupled device camera.

In contrast to our previous results^18–20 that photoexcitation with high excitation power density at low temperatures improves the luminescence efficiency of GaAsN alloys, we found that the laser irradiation results in a noticeable decrease in the PL intensity of GaP_1−xN_x.

To investigate whether such degradation phenomena are temporarily or permanently caused, PL mapping measurements were carried out across the area irradiated by the laser. First, GaP_1−xN_x (x = 0.75%) was irradiated ∼1 min by laser with a power density of 3.6 MW/cm², and immediately after that, we took the PL intensity mapping image shown in Fig. 1(a) by using the laser with a lower power density of 0.035 MW/cm², which slightly affects the degradation. Then, we waited for 24 h after the strong irradiation and retook the PL mapping image shown in Fig. 1(b). A dark area is observed in both images, where the PL degradation occurs due to the strong laser irradiation. The intensity variation between the dark and bright areas, representing the quantity of laser-induced PL degradation, was found to remain the same in both images. This clearly confirms that the degradation phenomena are permanently caused, indicating that some defects to act as nonradiative recombination centers are formed by laser irradiation.

Figures 2(a) and 2(b) show the time evolution of the PL spectra observed from the samples with N concentrations of x = 0.26% and 1.4%, respectively, during the laser irradiation with a power density of 11.5 MW/cm². It is found from Fig. 2(a) that the PL intensity steeply decreases to less than 70% of the initial value in 50 s and reaches about 60% after 200 s of the irradiation. As seen for the sample with x = 1.4% in Fig. 2(b), the PL intensity gradually decreases to about 90% of the initial value in 50 s and reaches about 85% after 200 s of the laser irradiation. It should be noted that the shape of the PL spectra hardly changed for both samples.

The laser irradiation may lead to three effects that are important in the process of the formation of defects, i.e., heating, the deformation of the subsurface layer, and the local excitation of the centers.²¹ Thus, we estimated the temperature of GaP_1−xN_x layers during strong laser irradiation from the slope of the higher energy side of PL spectra and found that the rise of the temperature was less than 20 K even if the laser power density was 11.5 MW/cm². Therefore, the heating effect can be almost neglected in this study. In our study using a continuous wave laser, another possible effect, subsurface deformation may possibly be caused by the local temperature increase due to laser irradiation. As we found, however, the temperature rise from room temperature was less than 20 K for the maximum laser power density in our study, which leads to only a small thermal expansion. Moreover, we observed no changes in the intensity of PL from deep levels in S-doped GaP substrates under the GaP_1−xN_x layers during measurements. This indicates that the penetration depth of the laser exceeds the layer thickness of GaP_1−xN_x and that only GaP_1−xN_x deteriorated by the laser irradiation. Thus, the subsurface deformation can be excluded in this study.

The surface recombination rate may possibly change and affect the PL intensity with an increase in the surface defect density or surface morphology change due to laser irradiation other than the subsurface deformation. Thus, we performed PL measurements of GaP_1−xN_x (x = 1.4%) covered with a 10 nm thick GaP cap layer and found that there are no differences in PL degradation between the samples with and without a GaP layer. This result indicates that the influence of the changes in the surface recombination rate is negligibly small and probably suggests that laser irradiation does not increase the surface defect density or change the surface morphology.

In addition, laser-induced oxidation on the top surface should be taken into account as a possible cause to affect the PL intensity because the PL measurements were carried out in the atmosphere. In order to examine the influence of laser-induced surface oxidation, we carried out the PL measurements of GaP_1−xN_x (x = 1.4%) without a GaP cap layer in a vacuum. As a result, the PL degradation was observed also in a vacuum in the same manner as in the atmosphere, and we can neglect laser-induced surface oxidation or the influence of surface oxidation to the surface recombination rate even if laser irradiation induced the surface oxidation.

Therefore, the third effect of laser irradiation is the dominant reason for the PL degradation in GaP_1−xN_x. To evaluate the degradation processes in GaP_1−xN_x during the laser irradiation, we have performed real-time PL measurements.

Figure 3 shows the irradiation time dependence of the integrated PL intensity of GaP_1−xN_x (x = 1.4%) normalized at the start of laser irradiation for various laser power densities. As can be found from this figure, the normalized PL intensity more rapidly and largely decrease to a saturation value with increasing laser power density. This tendency was observed for all the samples used in this study.

Figure 4 shows the laser irradiation time dependence of the PL decay profile of GaP_1−xN_x alloys with different N concentrations. With increasing nitrogen concentration, the PL intensity degradation is found to become smaller and slower under the same irradiation power density of 11.5 MW/cm². The red lines in Figs. 3 and 4 indicate the fitted curves using the following expression based on the stretched exponential function:^22,23

where the parameter A represents the magnitude of the PL intensity reduction normalized by the initial PL intensity and τ is a characteristic decay time constant that defines the inverse of the reduction speed of the PL intensity. The fit can explain well the temporal change in the PL intensity.

Figure 5(a) shows the fitting parameters A and τ as a function of the laser power density. As can be seen in this figure, the parameter A increases, while τ decreases with increasing laser power density. It is found that A and τ are roughly proportional to log(S/S₀), where S is the laser power density and S₀ is a certain normalization constant, although the proportional coefficient is positive and negative for A and τ, respectively. The cause of such logarithmic dependence has remained unclear but needs further consideration. The parameter β is not shown here but is in the range between 0.40 and 0.72, indicating the broad probability distribution of the decay time constant.^22,23

In addition, the nitrogen concentration dependence of the fitting parameters A and τ is shown in Fig. 5(b). The parameter A has a tendency to become smaller, while τ becomes longer with increasing nitrogen concentration. In particular, the parameter A representing the normalized reduction in the PL intensity is larger than 0.4 in the N concentration range less than around 0.1%. However, we found that the PL intensity does not change for GaP by the laser irradiation even with the maximum power density in this study. Therefore, the decrease in the PL intensity by laser irradiation is a unique phenomenon of GaP_1−xN_x alloys containing N atoms. In other words, the PL intensity degradation is involved with the existence of N atoms, probably N-related defects, as discussed later.

Time-resolved PL measurements²⁴ revealed that nonradiative recombination centers increase with increasing N concentration in GaP_1−xN_x. Since there already exist a large number of nonradiative centers in samples with higher N concentrations before the laser irradiation, the normalized reduction in the PL intensity for x > 1% is smaller than 0.2 even if quite a few defects to act as nonradiative recombination centers are formed by laser irradiation. For lower N concentrations, defects acting as nonradiative recombination centers are very few in the as-grown samples, and consequently, the laser irradiation leads to the larger normalized PL intensity reduction, which shows that GaP_1−xN_x alloys with even lower N concentrations are abound with hidden nonradiative defects induced by the incident laser beam.

In previous reports, Ga interstitials,^25,26 N clusters,^27,28 N interstitials,²⁸ and defect complexes composed of multiple nitrogen atoms²⁸ have been identified as defects forming deep-levels^29,30 in GaP_1−xN_x alloys. In GaAsN, another dilute nitride semiconductor, N–N split interstitials,^31–35 and N–As split interstitials,^{31–33,35,36} as well as Ga vacancies³⁷ and Ga interstitials^38,39 have been recognized as deep-level defects. Bouzazi et al.⁴⁰ reported that electron irradiation enhanced the density of the nitrogen-related nonradiative recombination center (E1), which is possibly associated with N–As split interstitials in GaAsN. This type of defect is found from first-principles calculations to be one of the lowest energy interstitial defects under various experimental conditions.⁴¹ As being a dilute nitride semiconductor, the GaPN alloy presents many similar characteristics with the GaAsN alloy. Considering the similarity between laser irradiation and electron irradiation, N–P split interstitials are possible candidates for the defects formed by laser irradiation in GaP_1−xN_x, though the combination with other measurements should be performed to identify the defects.

The changes in the PL intensity were investigated in detail to study the evolution of nonradiative recombination centers in GaP_1−xN_x alloys upon the laser irradiation. The PL mapping measurements confirmed that the degradation of the PL intensity induced by laser irradiation is an irreversible change due to the generation of defects. Real-time PL measurements revealed that the decay of the PL intensity is much more pronounced by strong laser irradiation. For samples with N concentrations as low as around 0.1%, larger and faster PL degradation occurs, indicating that GaP_1−xN_x with even lower nitrogen concentrations are abound with hidden defects to act as nonradiative recombination centers by laser irradiation. The PL measurements using high-power density photoexcitation proposed in this study would be very useful to understand and evaluate the impact of irradiation-induced nonradiative defect generation or multiplication, which causes the deterioration of optoelectronic devices based on dilute nitrides during operation.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

This work was supported by the JSPS KAKENHI (Grant No. JP19H02612).