Nitrogen Concentration Dependence of Two‐Step Photocurrent Generation by Below‐Gap Excitation in GaPN Alloys

Two‐step photocurrent generation and its dependence on nitrogen concentration in GaPN alloys using two‐wavelength excited photocurrent (TWEPC) measurements are investigated. External quantum efficiency (EQE) measurements show an increase in the two‐step absorption of below‐gap excitation light through tail states at longer wavelengths and that the quasi‐direct gap redshifts as the nitrogen concentration increases. The improvement in the EQE at longer wavelengths is explained by the increased density of the tail states in GaPN with higher nitrogen concentrations. In contrast, the EQE decreases at shorter wavelengths with increasing nitrogen content. TWEPC results show that differential photocurrent density ΔJ$\cdot J$ , which shows the synergy effect of multiple‐wavelength light, reduces with the addition of nitrogen. It is found from rate–equation analysis that the decrease in the EQE at shorter wavelengths and in ΔJ$\cdot J$ is mainly due to the increased nonradiative recombination in GaPN with higher nitrogen concentration. The use of lattice‐matched alloys and a p‐type GaP capping layer as well as the optimization of the growth conditions can improve the EQE at shorter wavelengths and ΔJ$\cdot J$ .


Introduction
The intermediate band solar cell (IBSC) concept utilizes an intermediate band (IB) to produce additional photocarriers across the bandgap of semiconductor material by two-step absorption of below-gap excitation (BGE). [1,2]This concept has the potential to overcome the efficiency limit of conventional single junction solar cells, [3] and the maximum efficiency of IBSC is expected to be 63% under concentrated sunlight. [1]The device formation based on this concept has been realized by using various materials systems, highly mismatched alloys (HMAs), [4][5][6][7] quantum dots, [8][9][10][11] and dilute magnetic semiconductors. [12]e HMA family contains dilute nitride semiconductors and oxygen-doped ZnTe, [13] where the addition of nitrogen or oxygen results in splitting the conduction band (CB) of the host material into two subbands called E þ and E À bands, and the E þ and E À bands serve as the CB and the IB, respectively.
GaP 1Àx N x alloy, one of the dilute nitride semiconductors, has shown the CB splitting into the E þ and E À bands similar to other dilute nitride semiconductors. [14,15]n addition to this feature, the formation of band tail states has also been reported in GaP 1Àx N x alloys due to the nitrogen clusters and/or different local arrangements of nitrogen atoms in the crystal. [16,17][21] In our previous study, it was found that photocurrent generation is markedly enhanced by BGE through the band tail states.In the present study, we focus on the effect of nitrogen concentration on the two-step photocurrent generation using dual-wavelength excitation measurements.The purpose of this study is to unveil the effects of nitrogen concentration on the two-step photocurrent generation in the GaPN alloy.Our results present that nitrogen incorporation significantly alters the two-step photocurrent generation process due to changes in the density of tail states and the quality of the sample.

Results and Discussion
Figure 1 shows room-temperature external quantum efficiency (EQE) spectra of GaP 1Àx N x (x = 0.17%, 0.61%, and 1.87%) samples at bias voltages of 0 V (solid lines) and À2 V (dashed lines), respectively.These spectra are found to exhibit two peaks.The first peak is at around 450 nm (2.7 eV), referred to as the E þ bandgap originating from the direct bandgap of GaP.The second peak is observed in the range between about 560 nm (2.2 eV) and 600 nm (2.1 eV) for the nitrogen concentrations investigated, i.e., from 0.17% to 1.87%, and is referred to as the E À bandgap giving the quasi-direct gap transition.The shift in the second peak with DOI: 10.1002/pssb.202300369Two-step photocurrent generation and its dependence on nitrogen concentration in GaPN alloys using two-wavelength excited photocurrent (TWEPC) measurements are investigated.External quantum efficiency (EQE) measurements show an increase in the two-step absorption of below-gap excitation light through tail states at longer wavelengths and that the quasi-direct gap redshifts as the nitrogen concentration increases.The improvement in the EQE at longer wavelengths is explained by the increased density of the tail states in GaPN with higher nitrogen concentrations.In contrast, the EQE decreases at shorter wavelengths with increasing nitrogen content.TWEPC results show that differential photocurrent density ΔJ, which shows the synergy effect of multiplewavelength light, reduces with the addition of nitrogen.It is found from rate-equation analysis that the decrease in the EQE at shorter wavelengths and in ΔJ is mainly due to the increased nonradiative recombination in GaPN with higher nitrogen concentration.The use of lattice-matched alloys and a p-type GaP capping layer as well as the optimization of the growth conditions can improve the EQE at shorter wavelengths and ΔJ.
increasing nitrogen content to longer wavelengths (lower energies) corresponds to the large bandgap bowing which is typical in dilute nitride semiconductors and similar behavior was reproduced by the photoluminescence (PL) measurements Figure S1 (see Section S1, Supporting Information).The bandgap bowing and formation of tail states by the nitrogen addition led us to study the influence of nitrogen concentration on the two-step photocurrent generation.The EQE spectra at a bias voltage of À2 V are larger than those at a bias voltage of 0 V.This is because of effective carrier collection due to the additional electric field.All the samples show visible absorption by tail states at longer wavelengths though the absorption by tail states for the sample with the lowest nitrogen concentration of 0.17% can hardly be seen at a bias voltage of 0 V.The clear absorption by the tail states at longer wavelengths drove us to study the two-step photocurrent generation and quantify the increment in nitrogen concentration by the BGE excitation and or dual wavelength excitation for the realization of real-time application of devices based on this concept.
In addition, a monotonical increase in the EQE by tail states can be seen with increasing nitrogen concentration revealing the enhanced formation of tail states.In contrast, the two peaks in the EQE spectra decrease as the nitrogen content increases, which is believed to be partly due to the degradation of sample quality by larger lattice mismatch between GaP 1Àx N x and GaP.It is also important to note that the steep rise of the second peak at the longer wavelength side shifts to longer wavelengths with an increase in nitrogen concentration.This reveals that the E À -band edge becomes lower in energy with increasing nitrogen concentration.
Following the EQE measurements, the photocurrent measurements were carried out for the combination of above-gap excitation (AGE) and BGE light illumination on the samples.Section S2, Supporting Information shows the J-V curves under the dark conditions shown in Figure S2. Figure 2 shows photocurrent densities by BGE light illumination only, J BGE as a function of BGE photon flux for various wavelengths of BGE light at the short-circuit condition.The symbols and solid lines show the experimental and simulation (discussed later) results, respectively.First, J BGE increases almost linearly as the BGE photon flux increases for all the samples and is explained as two-step transitions [18] from the valance band (VB) to the tail states as the IB (step 1) and from the tail states to the E À and/or E þ (step 2) by the BGE light illumination shown in Figure 3.The monotonical increment in J BGE with shorter BGE wavelength, in other words, higher BGE phonon energy is well matched to the behavior seen in the EQE spectra, which is due to increased density of tail states.
Furthermore, J BGE tends to increase as the nitrogen content increases from 0.17% to 1.87%.This can also be explained by higher density of tail states.Meanwhile, almost no increment in the J BGE by the sample of 0.17% for 905 nm (1.37 eV), 980 nm (1.26 eV), and 1342 nm (0.92 eV) indicates that the tail states in this sample do not extend down to 1.37 eV.In contrast, reasonable increment in J BGE is observed for 785 nm (1.58 eV).Therefore, the tail states of 0.17% are shown to extend down to ≈1.5 eV in Figure 3.In the other samples, a reasonable response in J BGE is not observed when the BGE light of 1342 nm is illuminated.Thus, it is shown in Figure 3 that the tail states of 1.87% extend to around 1.1 eV.
We illustrate a schematic model of the density of states for the E þ and E À bands and tail states in the case of the lowest (0.17%) and the highest (1.87%) nitrogen content in Figure 3 based on the experimental results of the EQE, PL spectra (Section S1, Supporting Information), and TWEPC measurements.In addition, to discuss TWEPC results later, corresponding optical transitions are shown in this figure.
To provide a quantitative and qualitative understanding of the TWEPC results for the simultaneous illumination, we define differential photocurrent density ΔJ as and use it as the basis of our discussion about obtained experimental results.Incidentally, ΔJ is an important index expressing the synergy effect of simultaneous illumination of AGE and BGE and improving the efficiency of solar cells, which cannot be found in the EQE measurements.
The results of ΔJ derived from the experiments together with the simulation results are shown in Figure 4.The results show an increase in ΔJ as the BGE photon flux increases and saturates at higher photon fluxes for all the samples.The transition mechanism of larger ΔJ obtained by the simultaneous illumination than the J BGE with only the BGE illumination is explained in detail in our previous study about GaP 1Àx N x (x = 0.75%) as efficient re-excitation of relaxed carriers from the IB back to the CB by BGE light. [18]As shown in Figure 4, ΔJ is larger for the sample with lower nitrogen concentration.
It is more likely that higher N content affects factors such as absorption coefficient, nonradiative recombination rate, and carrier drift velocity.Thus, we have attempted to investigate these factors together with the quantitative interpretation of experimental results by simulation based on the governing rate equations.The model for carrier generation and relaxation for the transitions is shown in Figure 5.The rate equations are given as: [23][24][25][26] dn dt where P AGE and P BGE are the photon flux of AGE and BGE, respectively, r AGE and r BGE are the surface reflectance for the AGE and BGE light, β is the recombination coefficient including both radiative and nonradiative components, N t is the density of tail states, α is the absorption coefficient for AGE, f is the electron occupancy of tail states, d is the active layer thickness, C n1 and C p1 are the absorption cross-section of the tail states for electron and hole, respectively, while C n2 and C p2 are capture crosssection rate parameters of the tail states for electron and hole, respectively, and v dn and v dp are the drift velocities for electrons at the CB and holes at the VB, respectively.It should be noted that the mobility of carriers in the tail states is found to be negligibly small, [18] and consequently the current extraction from the tail states remains insignificant.In this study, therefore, we do not consider the contribution of the tail states to the current density.Thus, current extraction from the samples is given by The dependency of change in the current density with respect to applied BGE photon flux was obtained for all cases similar to that of experimental measurements solving the given rate equations, and the corresponding parameters are summarized in Table 1 and 2.
The simulated results are found to be in good agreement with the experimental data for all the BGE sources and all the measured cases, as shown in Figure 3 and 4.
Table 2 shows a decrease in the parameters v dn , v dp , and α while an increase in the parameter β with increasing nitrogen concentration, which are the prime reasons for the decrease in the EQE at short wavelengths, as shown in Figure 1.First, the drift velocities of electrons and holes v dn , and v dp are strongly affected by alloy fluctuation and are believed to become smaller with increasing nitrogen concentration.Next, the derived absorption coefficients of GaPN for the AGE light (λ = 450 nm) are rather smaller than the absorption coefficient of GaP.However, they are believed to be not real but apparent absorption coefficients affected by surface recombination because the samples have no capping layer, and thus the values became smaller.And finally, the recombination coefficient β, which is found to be significantly changed with increasing nitrogen concentration and the main factor for the decrease in the EQE, becomes larger with increasing nonradiative recombination centers due to nitrogenrelated defects or misfit dislocations.Thus, it is expected to be possible to increase the EQE if the formation of misfit dislocations is suppressed by the use of GaPAsN or InGaPN alloys lattice-matched to GaP instead of GaPN alloys.Figure 6 shows the parameters for tail states-related absorption for BGE, i.e., C p1 N t for absorption between VB and tail states and C n1 N t for absorption between tail states and CB.It is found that the tail states-related absorption between tail states and CB is larger than the absorption between VB and tail states for all the samples at any of the BGE light applied.The energy difference between the tail states and VB is larger than that between CB and the tail states, as shown in Figure 3. Thus, the joint density of states corresponding to the BGE energy between the tail states and VB is expected to be less than that between CB and the tail states.In addition, we found from our theoretical calculations that the matrix elements are not much different for the two optical  0.17% 0.61% 1.87% transitions, which will be described elsewhere.Therefore, the absorption between the tail states and CB is expected to be larger than that between VB and the tail states.It is also found that the parameters for absorption decrease as the BGE light wavelength increases and that higher nitrogen concentration leads to larger absorption for BGE, which is consistent with the EQE spectra at the band tail states.Figure 7 shows the nitrogen concentration dependence of the carrier capture rates by the tail states for electron and hole, C n2 N t and C p2 N t , obtained from the fitting to the experimental data.The carrier capture rates, C n2 N t and C p2 N t , which happen to be the same, are found to be in the order of 10 10 s À1 and to increase slightly with increasing nitrogen content, which is partly due to the increase in the density of tail states N t .
As shown in Figure 8, our extended numerical study demonstrates that the AGE sufficiently fills up the tail states, and the electron occupancy approaches around 0.5 and decreases as the BGE light is added due to G CI transitions, and tends to saturate at higher photon fluxes.We found that the electron occupancy is smaller for higher nitrogen content, indicating that the relaxation of carriers generated by AGE from the CB to the tail states is faster in the samples with higher nitrogen concentration, as shown in Figure 7.
Moreover, the electron occupancy is found to remain extremely as low as 6 Â 10 À3 , 1 Â 10 À3 , and 3 Â 10 À4 for the samples with nitrogen concentrations of 1.87%, 0.61%, and 0.17% under BGE light illumination only, respectively.Such a low occupancy is attributed to the large difference in the tail states-related absorption between C n1 N t and C p1 N t , as shown in Figure 6.
In this study, we have found that parameters associated with two-step photocurrent generation are dependent on nitrogen concentration in GaP 1Àx N x .It is true that the EQE at the tail states improves for higher nitrogen content due to the increase in the tail states-related absorption, but the EQE reduces at shorter wavelengths with increasing nitrogen concentration mainly because the addition of more amount of nitrogen degrades the crystal quality of GaPN, and thus increases the density of nonradiative recombination centers.However, the quality of samples with higher nitrogen content can be improved using lattice-matched alloys, such GaAsPN or InGaPN, and optimizing the growth conditions.

Conclusion
In summary, the two-step photocurrent generation process and its dependence on nitrogen concentration were studied in GaPN alloys using TWEPC measurements.First, the EQE results showed a clear existence of the formed tail states below the quasi-direct gap for the samples with higher nitrogen content.The density of tail states is found to increase and extend to lower energies, and the quasi-direct gap band edge redshifts with increasing nitrogen concentration.TWEPC measurements showed that two-step photocurrent generation by only BGE light increases while ΔJ decreases with nitrogen addition.Based on the numerical analysis, the reasons for the decrease in ΔJ are found to be increased nonradiative recombination as well as lower apparent absorption coefficient affected by the surface recombination for the short wavelength lights as the nitrogen concentration is raised.The simulation results also revealed that the tail states-related absorption increases with an increase in the nitrogen concentration, likewise, the capture rates of electrons and holes by the tail states become elevated slightly.Finally, by using lattice-matched alloys to suppress the generation of misfit dislocations, using a p-type GaP capping layer to suppress the surface recombination, and optimizing the growth conditions so that better-quality crystals are grown; the EQE and ΔJ expressing the synergy effect of AGE and BGE are expected to be improved significantly.

Experimental Section
This study used the samples grown by metal-organic vapor phase epitaxy with the sequential growth of about 300 nm GaP buffer layer and about 500 nm GaP 1Àx N x epitaxial layer on a (001) oriented 400 μm thick n-type GaP substrate at 750 and 670 °C, respectively. [22]0 0.5 1.0 1.5 2.0 0.9  Trimethylgallium and phosphine were utilized as the sources for Ga and P, respectively, while dimethylhydrazine was used as the nitrogen source.Despite unintentional doping, Hall measurements confirmed p-type conduction in the GaP 1Àx N x layers and that the hole concentration was as low as 10 15 cm À3 and was almost the same for various N compositions.In contrast, the electron concentration in n-type GaP substrates is on the order of 10 18 cm À3 Thus, the GaPN layers were almost entirely depleted.The front and back electrodes of AuZn and AuGe were deposited on the surface of the p-type GaP 1Àx N x and n-type GaP substrates, respectively.Two-step photocurrent generation measurements by the BGE light absorption were carried out at room temperature.First, we measured EQE spectra of GaP 1Àx N x alloys.Next, TWEPC measurements were performed by using a chopped light at 450 nm (2.76 eV) with a photon flux of 1 Â 10 16 cm À2 s À1 from the Xe lamp sourced monochromator as the AGE source whereas three semiconductor lasers of 785 nm (1.58 eV), 905 nm (1.37 eV), 980 nm (1.27 eV), and a diode-pumped-solid-state (DPSS) laser of 1342 nm (0.92 eV) was used for the BGE light sources.
Our measurements were taken for three cases, i.e., only BGE illumination, only AGE illumination, and simultaneous illumination of AGE and BGE, and resulting photogenerated current densities relative to the illumination were measured and are termed J BGE , J AGE , and J AGEþBGE , respectively.

Figure 1 .
Figure1.Room-temperature EQE spectra of GaP 1Àx N x samples measured at bias voltages of 0 V (solid lines) and À2 V (dashed lines).

Figure 2 .
Figure 2. BGE light-generated photocurrent density for three samples with a nitrogen concentration of a) 0.17%, b) 0.61%, and c) 1.87% as a function of BGE photon flux.

Figure 3 .
Figure 3. Schematic model of the density of states for E þ and E À bands and tail states based on the results of the EQE spectra.To discuss TWEPC results, corresponding optical transitions are added in this figure.

Figure 4 .Figure 5 .
Figure 4. Shows the differential photocurrent density as function of BGE photon flux for three samples with a nitrogen concentration of 0.17%, 0.61%, and 1.87% in figures labeled a-c) respectively.

Figure 6 .
Figure 6.Nitrogen concentration and wavelength dependence of tail states-related absorption for BGE.

CFigure 7 .
Figure 7. Nitrogen concentration dependence of the carrier capture rates by tail states for electron and hole, C n2 N t and C p2 N t .

Figure 8 .
Figure 8. Electron occupancy at the tail states as a function of BGE photon flux.

Table 1 .
The common parameters for the rate-equation analysis.