Emitter depth

In a first investigation, the influence of the emitter/p-n-junction depth on the specific contact resistance and the contact formation of aluminium-containing silver screen-printed contacts to boron emitters was examined. For that purpose, four different BBr₃ -based emitters with similar boron surface concentration but different emitter depthde

were designed. The doping profiles of the different emitters measured by electrochem-ical capacitance voltage (ECV) measurements (compare section 2.1.4) are shown in figure 3.22. Relevant emitter parameters are summarized in table 3.3. The emitters D1, D2 and D4 show high boron surface concentrations NB,s between 8.4·10¹⁹cm⁻³ and 8.6·10¹⁹cm⁻³ and feature no boron-rich layer. NB,sof emitter D3 is slightly lower, however, as the progression of the emitter profile is very similar to the other emitters in the surface near region, the emitter was included in the investigation. The red line in figure 3.22 indicates the doping density above which field emission is the dominating current transport mechanism over the metal-semiconductor contact. In figure 3.22 the

3.5 Influence of emitter properties on contact formation

Figure 3.22:Emitter profiles of emitters featuring different emitter depth but similar surface concentrations measured by ECV. The unfilled symbols represent the first values measured in the base, representing the density of electrically active phosphorous. The red line marks the doping density above which current transport via field emission dominates for the metal-semiconductor contact.

first measured values in the base, representing the density of electrically active phos-phorous, are depicted as unfilled symbols. The depth of the emitterdeis then evaluated as the mean value of the last point in the emitter and the first point in the base. The measurement inaccuracy ofde is estimated as half the distance between the two points.

The determination of de and the measurement error according to this procedure may be imprecise as the emitter depth depends on surface properties and etch velocity. Ad-ditionally, for the determination of the emitter depth, one has to keep in mind that ECV measurements close to the p-n-junction get more inaccurate [23, 114]. However, these parameters and their inaccuracies are not further considered here.

For the experiment n-type Cz-Si wafers with a resistivity of≈ 1.4 Ωcm were used. The wafers were alkaline textured and cleaned. Then the different emitters were diffused in a BBr₃-based process. After the removal of the borosilicate glass and the deposition of

Table 3.3: Relevant emitter properties of the emitter D1-D4 with different depth.

emitter

75 nm SiN_x:H by PECVD two different commercially available aluminium-containing silver pastes, called Ag/Al-A and Ag/Al-B, were screen-printed.

Prior to firing the samples, the impact of the doping density on the wafer temperature was checked. This was done as the absorption of IR radiation from the belt furnace is doping density dependent, as free carrier absorption increases with increasing doping as Bende et al. reported [150]. For the samples in the actual experiment the most pronounced temperature difference is expected for samples with the biggest difference in emitter sheet resistance. Therefore temperature profiles of metallized wafers with emitter D1 and D4 fired in the standard process were measured with a thermo-couple on the wafers by a Datapaq-profiler. The samples were measured two times each. In figure 3.23 the two profiles with the biggest deviation in peak firing temperature are depicted. The wafer with lower total doping density shows a slightly higher firing temperature in the peak region (≈6°C) . Taking the measurement accuracy of the mea-surement with the temperature profiler (<20°C) into consideration, this deviation can be neglected. The variation of the wafer temperature with doping density reported of by Bende et al.[150] could therefore not be confirmed for the used emitters. However, in their work base material with considerably lower doping densities/ lower total con-ductance (<31 mS/) was used. Their results additionally suggest that the influence of the total conductance on the wafer temperature decreases for higher doping densities.

In the actual experiment the conductance of the base material is≈714 mS/. Adding a 30 Ω/emitter leads to a total conductance of 780 mS/which is an increase of around 9 %. Comparing wafers with different emitters the difference in total conductance is even smaller. The relative differences in conductance are therefore considerably smaller than the ones Bendelet al. reported. This could explain why the reported trend cannot be observed for the emitters developed in this experiment. As no significant deviation

0 10 20 30 40 50

0 200 400 600 800

D1 (120_Ω/sq) T_peak= 812°C D4 (33_Ω/sq) T_peak= 806°C

Temperature(°C)

Time (s)

Figure 3.23: Measured temperature profiles for emitters D4 and D1.

3.5 Influence of emitter properties on contact formation

0.2 0.4 0.6 0.8 1.0

0 5 10

15 Ag/Al-A

Ag/Al-B

ρ c(mΩcm2 )

d_e(μm)

Figure 3.24: Specific contact resistance in dependence of emitter depth.

in the temperature profiles could be observed, all wafers were fired in the standard firing process introduced in section 3.3.

The specific contact resistance of the different samples is shown in figure 3.24. The re-sults confirm the relation between emitter depth and specific contact resistance reported in [151]: a deeper emitter results in a reduced contact resistance. Additionally, it can be observed that for shallow emitters small variations of the emitter depth show a large influence on the specific contact resistances. For deeper emitters the curve becomes flatter and tends to saturation.

To examine whether the reduction of the specific contact resistance can be attributed to differences in the contact microstructure, contacts were etched back in hydrofluoric acid and analyzed by means of SEM. In figure 3.25 SEM overview images for samples with emitter D1 and D4 are shown. For better visibility some groups of contact spots grown into the surface are marked by orange circles. Comparing the number of contact spots on both interfaces, no large difference can be observed. However, small differences cannot be ruled out, as the number of contact spots cannot be evaluated quantitatively for large areas due to the limited contrast of the SEM images. For all emitters, contact spots occur in groups. The number of spikes differs from group to group for all emit-ters. Higher magnification SEM micrographs show no difference in size and shape of the contact spots for the different emitters.

The SEM analysis carried out revealed no significant difference in the interface structure and properties of contact spots for varying emitter depth. Although small differences in the number of contact spots cannot be ruled out, they could not account for the large influence of emitter depth on the specific contact resistance.

The observed reduced contact resistances for deeper emitters is therefore attributed to the reduction in Schottky barrier width (WB) of the metal-semiconductor contact with

increasing doping concentration. The most effective current transport mechanism over metal-semiconductor contacts is field emission. As discussed in section 1.7.2, field emis-sion dominates at doping densities>1·10²⁰cm⁻³. In figure 3.22 this value is indicated by the red horizontal line. With increasing emitter depth the width of the region where the doping density exceeds this value, increases as well. In section 3.5.3 the specific contact resistance of a single contact spot on the different emitters will be evaluated using an analytical model based on the different current transport mechanisms.

contact spots

D1 50 μm

(a)

50 μm

contact spots D4

(b)

Figure 3.25: SEM overview images of contacts etched back in HF: (a) emitter D1 (shallow) (b) emitter D4 (deep).