In Ref. 1, Wierer et al. reported on three different structures of nitride-based p-i-n solar cells. A p-i-n diode consisting of 170 nm thick intrinsic layer sandwiched by GaN layers was fabricated. Alternative structures with intrinsic InGaN layers sandwiched by n-type InGaN or graded layer and a p-type layer were found to have improved short-circuit current densities. The impact of piezoelectric polarization and nonradiative recombination on the short-circuit current densities of GaN/InGaN photovoltaic devices was analyzed. The performance enhancement of the improved structures (samples B and C) was partially ascribed to the different carrier collection efficiency which is influenced by the polarization-induced band tilting. That is, the effect of polarization will tilt the energy band to the detrimental direction for carrier collection in the original structure (sample A) and to the favorable direction in the improved structures.
The experimental result reported in Ref. 1 is certainly a very useful reference in this research field; however, we have some different viewpoints regarding the explanation about the physical mechanisms which are relevant to the improvement of device performance. In this comment, similar to the study of Ref. 1, we investigate the energy band diagrams of the same structures with polarization effect by means of theoretical simulation. The simulation program used for this study is APSYS (Ref. 2) and the material parameters of semiconductors used in the simulation can be found in Ref. 3. The parameters used in this simulation could be different from those used in the simulation of Ref. 1, which might result in different values of the simulation results; however, the tendency and the qualitative comparison of the results would be similar
Figure 1 shows the energy band diagrams of sample A under different degrees of relaxation. The factor “degree of relaxation,” which varies from zero (with piezoelectric polarization) to unity (without piezoelectric polarization), is employed to adjust the value of piezoelectric polarization while the spontaneous polarization is maintained to be equal to the value of theoretical prediction.4 It is evident that the situation of band tilting shown in Fig. 1(a) of present simulation is similar to that shown in Fig. 2(a) of Ref. 1. However, the figures show dissimilar tendency under different degrees of relaxation. It means that whether the band can be tilted from “right” direction to “wrong” direction or not depends on the degree of polarization and, thus, it is closely related to the degree of relaxation. The information of the degree of relaxation, as well as the degree of polarization, for sample A is a lack in Ref. 1.
For sample B, it is expected that the polarization difference of adjacent layers is smaller due to the closer indium composition. As a result, the situation of polarization-induced band tilting would be slighter. Figure 2 shows the energy band diagrams of sample B under different degrees of relaxation. Under these circumstances, the situation of band tilting shown in Fig. 2(c) of present simulation is similar to that shown in Fig. 2(b) of Ref. 1. Note that the authors used the band diagram with serious polarization effect for sample A and the band diagram with slight polarization effect for sample B in Ref. 1. The statement “the InGaN layers are coherently strained to GaN for all the structures” in Ref. 1 seems unreasonable. Similar problem existed also for sample C.
From the above discussions, it seems that the evidence of the relation between the situation of band tilting (or the degree of polarization) and the performance variation in all structures under study is insufficient if the degree of relaxation is not determined experimentally in the original research.
This work is supported by the National Science Council of Taiwan under Grant No. NSC 99-2119-M-018-002-MY3.