Improved two-layer emitters

As seen in the previous section, the cell performance of the epitaxial wafer-equivalents with in-situ emitter is limited by low fill factors. A further increase of the phosphine flow during the cooling step was investigated to increase the surface doping concentration. The bulk emitter thickness was kept at approximately 1 µm and the phosphine gas flow during cooling was varied from 35 to 100 sccm, corresponding to phosphine concentrations of 7 to 19 ppm.

Figure 5-17 shows the two doping profiles, both grown with 5 ppm phosphine concentration, but one cooled with 13 and the other with 19 ppm phosphine concentration. Numerical calculations showed that the corresponding sheet resistances have values of 110 and 97 Ω/sq., respectively. A slightly higher surface concentration of 3x10¹⁹ cm^-3 and deeper diffusion is reached with the 19 ppm phosphine concentration.

A new solar cell batch was then produced with these emitter profiles. Table 5-5 shows the illuminated and dark cell parameters of cSiTF solar cells with epitaxial emitters, as they depend on the phosphine concentration during cooling. All cells with epitaxial emitters show excellent open circuit voltages of nearly 650 mV. A similar open circuit voltage is found for the reference cell with POCl₃ emitter grown on Cz. Even though there was a slight change in the surface concentration with higher phosphine concentration, no clear improvement in the fill factor was found. In contrast to the previous solar cell batch, all fill factors are increased and a maximum value of 78.6% was reached for a sample treated with 13 ppm phosphine concentration (not shown) similar to the sample with POCl₃ diffused emitter.

The series resistances fitted from the dark I-V curves of the cells show constant values of approximately 0.5 Ωcm². This value suggests a low contact resistivity of the titanium-silicon contact, which was confirmed by measurements of Transfer Length Model (TLM) structures [114]. The measured values of approximately 1x10^-4 Ωcm² are similar to the theoretical value of 8x10^-5 Ωcm² for a surface doping concentration of 3x10¹⁹ cm^-3 [75]. All dark saturation current densities J₀₂ of the cells with epitaxial emitters are low, indicating once more that only a little recombination occurs within the space charge region.

0 50 100 150 200 10¹⁸

10¹⁹ 10²⁰

[PH₃]_Cooling [ppm]

119 19 13

Phosphorus [cm-3 ]

Depth [nm]

Figure 5-17: SIMS profiles of epitaxial emitters cooled with 13 ppm (triangles), 19 ppm (dots) and 119 ppm (squares) PH3 in H2. Table 5-5: Solar cell parameters of EpiWEs grown on Cz with epitaxial emitters. The phosphine concentration was varied from 7 to 19 ppm during

cooling. All cells are confirmed by ISE CalLab for a cell size of 21 cm².

n° HC702B HC708B HC670B HC513

Emitter process Epi Epi Epi POCl3

[PH3]cooling [ppm] 19 13 7 -

demitter [µm] 0.9 0.8 0.7 0.4

dbase [µm] 18.6 17.6 17 22

VOC [mV] 643 646 649 645

JSC [mA/cm²] 29.1 29.5 28.9 29.4

FF [%] 77.8 77.8 77.9 78.7

η [%] 14.6 14.8 14.6 14.9

J01 [A/cm²] 3x10^-13 3x10^-13 2x10^-13 3x10^-13 J02 [A/cm²] 3x10^-8 2x10^-8 2x10^-8 1x10^-6

RS [Ωcm²] 0.5 0.7 0.6 1.0

RSh [Ωcm²] 1.4x10⁵ 8x10⁵ 8.9x10³ 5.2x10³

While epitaxial emitters on small area cells showed good performance, the applicability to larger area cells needed to be verified, as a large deviation of the sheet resistance is found on 10x10 cm² wafers (see Figure 5-1). Additionally, a further increase of the surface doping was investigated. Increasing the phosphine concentration to 119 ppm during cooling, causes a surface phosphorus concentration of 5x10¹⁹ cm^-3 (see Figure 5-17).

Table 5-6 shows the results of cSiTF cells with epitaxial emitters on mc and Cz substrates, as well as the corresponding reference cSiTF cells with POCl₃

emitters. Open circuit voltages up to 634 mV on mc and up to 655 mV on Cz for the cells with epitaxial emitters were achieved. These open circuit voltages are clearly higher than those for the reference cells on the corresponding substrates processed with a POCl3 emitter. The benefit of the moderately-doped thick emitter with a highly-doped surface is evident. Figure 5-19 shows the internal quantum efficienies of the samples with epitaxial emitter on mc and Cz subtrates. The high performance of the emitters is evident at short wavelengths.

Despite the large sheet resistance deviation up to ±50 Ω/sq. after the phosphine diffusion, high efficiencies of 13.6% on mc and 14.9% on Cz were reached with epitaxial emitters.

Table 5-6: Best solar cells with evaporated contacts on large cell area (92 cm²). All cells are confirmed by ISE CalLab.

n° OC-06-87 OC-09-192 HC758 HC664

Substrate mc mc Cz Cz

Emitter process Epi POCl3 Epi POCl3

Rsheet [Ω/sq.] 80 120 85 120

demitter [µm] 1.0 0.4 0.9 0.4

dbase [µm] 23 28 17 19

VOC [mV] 634 627 655 648

JSC [mA/cm²] 28.7 29.4 28.4 29.9

FF [%] 74.6 76.2 79.9 77.6

η [%] 13.6 14.1 14.9 15.0

J01 [A/cm²] 4x10^-13 5x10^-13 2x10^-13 2x10^-13 J02 [A/cm²] 6x10^-8 2x10^-7 3x10^-7 2x10^-8

RS [Ωcm²] 0.6 0.3 0.7 0.3

RSh [Ωcm²] 6.8x10⁵ 2.4x10⁴ 3.3x10⁴ 5.4x10⁵

Moreover, very high fill factors up to 79.9% on Cz substrates were achieved.

An increase of the fill factor is found for the epitaxial emitter compared with the POCl3 on Cz and all series resistances had low values between 0.3 and 0.7 Ωcm². However, the fill factors of the cSiTF cells on mc with epitaxial emitters did not match the high values of the reference samples with POCl3. The difference may arise from impurities, which diffuse into the epitaxial layer during the deposition process. A subsequent POCl₃ diffusion acts as a gettering step for the impurities, and the phosphorus glass, together with the impurities, is then removed. For the epitaxial deposition, no gettering occurs and the impurities coming from the substrate cannot be collected. However, traps in the space charge region would increase the dark saturation current density J₀₂, which

is not the case, as dark saturation current density is lower for the cell with the epitaxial emitter (Table 5-6).

200 µm Substrate

Epitaxy 16 µm 42 µm

200 µm Substrate

Epitaxy 16 µm 42 µm

Figure 5-18: Layer thickness inhomogeneity on mc substrates.

Another possible reason for the discrepancy between the results of the epitaxial and POCl3 diffused emitters on mc is that the growth of the epitaxial layer depends on the grain orientation. In Figure 5-18 the thickness inhomogeneity of the epitaxial layer on a mc substrate is shown. It was already reported that <111> facets are slower growing grains [115]. Furthermore, at the grain boundaries the thickness is often reduced compared to the intra-grain thickness, because the higher energy associated with the defects suppresses the layer growth nearby. The highly-doped n-type and p-type regions are then in direct contact, resulting in leakages and low fill factors [14]. However, as the shunt resistances are over 10⁴ Ωcm², this is not the dominant fill-factor limitation. SunsV_OC measurements were performed, which is a good method to measure the maximum possible fill factor without series resistance contribution [113]. The measurements show similar mismatch-corrected [116] pseudo fill factors of 78.0%. This indicates that the discrepancy of the real fill factors is induced by series resistance deviations. The large thickness deviation of the emitter results in a strong sheet resistance deviation, as shown in Figure 5-20.

The contribution of the emitter to the series resistance is directly proportional to the sheet resistance and therefore the total series resistance is increased for thinner emitters with the same doping level [78]. A lower doping on <111> than on <100> surfaces is reported [117], which further increases the sheet resistance inhomogeneity. This is confirmed by the dark I-V characteristics, as the series resistance for the Epi emitter is 0.6 in contrast to 0.3 Ωcm² for the POCl₃ emitter (Table 5-6, Figure 5-25).

400 500 600 700 800 900 1000 1100

Figure 5-19: Internal quantum efficiencies of EpiWE solar cells with epitaxial emitters on Cz (HC758) and mc (OC-06-87).

Figure 5-20: Sheet resistance mapping on an mc wafer with epitaxial emitter.