Effect of Grain Areas on Minority-carrier Lifetime in Undoped N-type BaSi 2 on Si ( 111 )

We have grown undoped n-BaSi2 epitaxial films with different grain sizes on Si(111) and characterized their minority-carrier lifetime, τ. We found that τ value in undoped n-BaSi2 did not depend on average grain area, but on surface condition. The samples with mirror surfaces had large τ of about 0.4 μs and those with cloudy surface small τ of about 8 μs. We tried to cap the sample surface in situ with a 3 nm Ba or Si layer in order to control the surface of BaSi2, and succeeded to intentionally form BaSi2 with large τ.


Introduction
Because of outstanding properties of semiconducting BaSi2, we have studied extensively its fundamental properties.BaSi2 has a suitable band gap for solar cell of approximately 1.3 eV, matching the solar spectrum [1].In addition, it has a large absorption coefficient (ca.α=3×10 4 cm -1 at 1.5 eV) and long minority-carrier diffusion length (ca.L=10 μm) [1,2].Undoped BaSi2 shows n-type conductivity with the electron concentration of about 10 16 cm -3 [3].In our previous work, we achieved high hole concentrations exceeding 10 20 cm -3 in B-doped p-type BaSi2 [4].Therefore, we aim to fabricate a BaSi2 pn diode using the undoped n-and B-doped p-type BaSi2.a-axis-oriented BaSi2 can be grown on a Si(111) surface.Since these films have three epitaxial variant rotated by 120° with each other, a large number of grain boundaries (GBs) exist in the BaSi2 epitaxial films [5].According to our previous research using Kelvin probe force microscopy (KFM), the potential at GBs became higher than those in grain interior in undoped n-BaSi2 [6].We thus expect that photogenerated minority carriers, holes, are repelled back from the GBs and the carrier recombination are not likely to occur at the GBs.In order to clarify this assumption, we fabricated approximately 0.5-μm-thick BaSi2 films with various grain sizes, namely, various densities of GBs and characterized their minority-carrier lifetime, τ.

Experiment
We used an ion-pumped molecular beam epitaxy (MBE) system equipped with an electron-beam evaporation source for Si and a standard Knudsen cell for Ba.We employed two-step growth techniques.After the thermal cleaning of the Si substrate, Ba was deposited on the hot floating-zone n-Si(111) substrate (ρ>1000 Ω•cm) in order to fabricate the template layer, which works as seed crystals for subsequent layers [7].Then, Ba and Si were co-deposited on the template layer at optimum growth temperature, 580 °C, by MBE.The layer thickness was about 0.5 μm.We prepared 8 samples with various grain sizes by changing the growth temperature and Ba deposition rate during reactive deposition epitaxy (RDE) [8].This is because various RDE growth conditions change the migration of Ba atoms on the Si substrate.We also prepared the films capped in situ with a 3 nm Ba or Si.The details of sample preparation are shown in Table I.The grain size was characterized by electron backscatter diffraction (EBSD).The minority-carrier lifetime was measured by microwave-detected photoconductivity decay (μ-PCD) method.Electron-hole pairs were generated by a 349 nm laser pulse, and then, the photoconductivity decay curves were measured by the reflectivity of microwave with the frequency of 26 GHz [9].We measured the decay curves with various excitation laser intensity ranging between 1.1×10 2 and 1.3×10 5 W/cm 2 .The crystalline quality was characterized by reflection high-energy electron diffraction (RHEED).In order to evaluate the composition of the surface, x-ray photoelectron spectroscopy (XPS) were performed.Al Kα radiation (1486.6 eV) was used for x-ray radiation source.The probing depth was about 3 nm.

Results and discussions
The 12×12-μm 2 EBSD crystal orientation maps and distribution histograms of grain area fraction in samples A-H are shown in Fig. 1.Average grain areas calculated from the histograms are also shown.Red, blue, green in the EBSD crystal orientation maps indicate three epitaxial variants of  a-axis-oriented BaSi2 on Si(111).As shown in Fig. 1, we fabricated BaSi2 epitaxial films with various grain areas.Thus, it is safe to state that the density of BaSi2 GBs decreases with increasing average grain area.We next measured the minority-carrier lifetime in these samples by μ-PCD.Figure 2(a) shows the example of typical decay curve of BaSi2, which was measured in sample H.We also presented the normalized μ-PCD decay curves of samples A and C in Fig. 2(b), and sample B and D-F in Fig. 2  (c).These curves were measured when excess carrier density, Δn, was 2.4×10 18 cm -3 , calculated from the absorption coefficient of BaSi2 at a wavelength of 349 nm, and irradiated laser intensity.
The decay rates of sample A and C became faster than the others.According to our previous studies, decay curve can be divided in two three modes from the viewpoint of decay rate, and was fitted well using the following equation.
Here, I1, I2 and I3 are the coefficients and τAuger, τSRH and τSRH-trapping the time constants for each decay mode.In the initial mode, Auger recombination is dominant.This is because BaSi2 has a large absorption coefficient, and high excitation occurs in the μ-PCD measurement.Thus, it is difficult to avoid the Auger recombination in this work.However, this high excitation is not likely to happen in the actual AM 1.5 irradiation.Therefore, we used the second decay curve due to Shockley-Read-Hall recombination to investigate the minority-carrier lifetime.
Figure 3 shows the correlation between τSRH and average grain area.τSRH value can be separated into two groups, that is, about 0.4 and 8 μs.However, τSRH values did not depend on the average grain areas in each group.This result indicates that the GBs do not act as recombination centers for minority-carriers, holes.This is consistent with our KFM results [6].On the contrary, the samples with large τSRH had cloudy surfaces and those with small τSRH mirror surfaces.

011401-3
The photographs of mirror-and cloudy surface of BaSi2 are shown in Fig. 4(a) and (b), respectively.In order to clarify the reason why the surface became cloudy, we next investigated the crystalline quality of these samples.Figure 4(c) and (d) show the RHEED patterns of sample A and B taken just after the MBE growth.We see sharp streaky pattern in both samples and do not see any clear difference.Therefore, the surface condition did not become cloudy just after MBE growth in ultrahigh vacuum chamber, but after air exposure.We speculated that the sample surfaces were oxidized when the samples were exposed to air, and their oxidation layers may passivate the defects exist in the BaSi2 surface.However, the problem was that we were not able to control the surface of BaSi2.This means that the formation of the sample with large τSRH was out of control.
In order to solve this problem, we next grew two samples G and H.These samples were capped in situ with a 3 nm Ba or Si layer after MBE growth so that we could control the surfaces of the samples.The correlation between τSRH and average grain area of samples G and H are also presented in Fig. 3.The important point is τSRH value was approximately 10 and 7 μs, and almost the same as those with cloudy surfaces.This result indicates that we succeeded to intentionally form the BaSi2 films with large τSRH by capping with a Ba or Si layer.In order to clarify the compositions of the surfaces, we performed XPS measurement.Figure 5 shows the normalized XPS spectra of (a) Ba 3d5/2, (b) O 1s, (c) Si 2p, and (d) C 1s states for sample G and H.The clear peaks in Figs. 5 (a), (b) and (d) imply that oxides such as BaCO3 and BaO, and others like BaOH exist in the sample capped with a Ba layer.On the other hand, the formation of SiO2 was promoted in sample H capped with a Si layer.However, we were not able to identify the origin of large τSRH only in these results.Table II shows the atomic percentages of O in the surface regions of BaSi2 in sample A, B, G, and H, calculated from XPS spectrum.We can see here that the O ratio is smaller only in sample A than the others.Therefore, the samples with smaller τSRH had a small ratio of O. On the contrary, those with large τSRH had a large ratio of O.This result suggests that O atom may play an important role to achieve large τSRH.Therefore, we conclude that τSRH values were not determined by GBs, but surface conditions, and the capping with a Ba or Si layer leads us to fabricate BaSi2 films with large τSRH.However, there are few data about passivation of BaSi2.Thus, further studies are required.

Fig. 4 .Fig. 5 .
Fig. 4. Photographs of (a) mirror-and (b) cloudy surface of BaSi2.RHEED patterns for (c) sample A and (d) sample B just after the MBE growth of BaSi2 are also shown[10]

Table I .
Sample preparation: BaSi2 layer thickness, surface condition, and BaSi2 average grain area by EBSD are specified

Table II .
Atomic percentages of O in the surface regions of BaSi2 in sample A, B, G, H.