Comparison of the contact formation with and without anti-reflection

Al2O3/SiNx:H 50 μm

(b)

SiO2/SiNx:H 50 μm

(c)

Figure 3.45:SEM overview images of contacts on different passivation layers etched back in HF. (a) SiN_x:H, (b) Al₂O₃/SiN_x:H, (c) SiO_x/SiN_x:H. The bright regions at the contact area are regions with silver-aluminium contact spots.

printing on phosphorous and boron emitters [165]. On boron emitters they found lower specific contact resistances for samples with SiO_x/SiN_x compared to SiN_x:H layer and a reducedρc on Al₂O₃/SiN_x:H-stacks compared to wafers with SiN_x:H. They additionally presented differential scanning calorimetry measurements of PbO glass with powders of the different dielectrics. A different behaviour of the dielectric powders compared to deposited dielectric layers can, however, not be ruled out. Considering the dielectrics investigated in this section, the mixture of PbO with SiO_x showed the lowest melting point, followed by SiN_x:H. For the PbO-Al₂O₃ mixture no melting was observed in the examined temperature range<850°C. Therefore, they attributed the low contact resis-tance on wafers with Al₂O₃/SiN_x:H-stack to a blistering of the Al₂O₃.

The results of these DSC measurements can explain the specific contact resistances observed in the present experiment by different starting temperatures for the etching of the dielectric layers by the lead glass contained in the paste. In contrast to the results of Kalio et al., in this experiment the highest specific contact resistance was found for the wafers with a Al₂O₃/SiN_x:H-stack. This can first be due to the different paste components. Additionally, differences in the deposition temperature of the ALD could result in a reduced occurrence of blistering [166]. The intact Al₂O₃-layer must be etched by the glass frit for facilitating the contact formation.

3.7.2 Comparison of the contact formation with and without anti-reflection coating

The influence of the presence of an anti-reflection layer between screen-printed contact and silicon surface on the contact formation process was investigated in an additional experiment. Alkaline textured wafers with a 50 Ω/BBr₃-based emitter were divided in two groups. The wafers of the first group ”SiN_x:H” received 75 nm PECVD SiN_x:H. The second group ”no SiN_x:H” did not receive any anti-reflection coating. In the following,

3.7 Influence of passivation/anti-reflection layer

two different aluminium-containing silver pastes were screen-printed on the wafers. All samples were fired in the standard firing process in an IR belt furnace. The temperature profiles on the wafers were comparable for wafers with and without dielectric layer.

The specific contact resistance for the different samples is depicted in figure 3.46. For both pastes, AgAl1 and AgAl2, the contact resistances of the samples without silicon nitride layer are reduced by half compared to the samples with SiN_x:H.

0.5 5

1 10

Dielectric layer

no SiN_x:H SiN_x:H

ρ

Ω

Paste AgAl1₁ AgAl2

Figure 3.46: Specific contact resistance of contacts of two different silver-aluminium pastes on wafers with and without silicon nitride layer.

The SEM analysis presented in the following only deals with samples of paste AgAl2, however, the observations are the same for both pastes.

The difference in contact resistance for wafers with and without silicon nitride layer can be explained by the distribution of contact spots on the silicon surface, as can be seen in figure 3.47. The images on the left show SEM micrographs of a sample with silicon nitride. On the right contacts on wafers without passivation layer can be seen. The white spots in the images represent silver-aluminium contact spots. The overview images in figure 3.47 (a) and (b) show that the frequency of the occurrence of areas with contact spots is comparable for both samples. On the wafers with silicon nitride groups of separated contact spots are visible, as observed in previous sections.

However, on wafers without silicon nitride these groups of contact spots appear grown together to larger spikes with diameters of more than 10μm. This can be seen better in magnified images of these areas in the lower part of figure 3.47. On wafers with silicon nitride layer (figure 3.47 (c)) the separated contact spots are surrounded by a silicon surface with a brighter contrast compared to the surface further away from the contact spots. For better visibility two of these regions are exemplarily encircled with green dotted lines. In section 3.6.2 it was shown that these areas of bright contrast represent regions with intact silicon surface that are covered by the inhomogeneous

aluminium-50 μm

contact spots SiNx:H

(a)

50 μm

contact spots no SiNx:H

(b)

10 μm

contact spots

intact pyramids SiNx:H

(c)

10 μm

contact spots no SiNx:H

(d)

Figure 3.47: SEM images of screen-printed contacts on wafers with (a,c) and without (b,d) SiN_x:H layer etched back in HF.

containing finger parts. The size of these bright areas is comparable with the diameter of the grown together contact spots on the wafers without silicon nitride that can be seen in figure 3.47 (d).

In figure 3.48 the shape of the contact spots on wafers with and without silicon nitride is compared. On wafers with SiN_x:H the metal spikes show the surface structure of the textured silicon surface. Sharp pyramids are visible as presented in section 3.6.2. On wafers without silicon nitride, however, the sharp morphology of the silicon texture is no longer visible at the surface of the contact spots. The shape of the contact spots resembles rounded silicon pyramids. In the metal spikes dark regions can be seen. The dark regions contain a high amount of silicon compared to the silver-aluminium phase surrounding these regions. As the spatial resolution of the SEM EDX is around 1μm, the composition of the different structures cannot be detected exactly and is therefore not quantitatively given here. A closer look at these silicon-rich inclusions is taken in section 3.9. Figure 3.49 shows a polished cross-section of the silicon-contact interface of a sample without silicon nitride. For some contact spots (marked by red arrow) the shape of an inverted pyramid, observed on wafers with dielectric layer, is lost probably caused by the fusion of several contact spots. Although no dielectric layer exists between

3.7 Influence of passivation/anti-reflection layer

2 μm

contact spots

intact pyramids SiNx:H

(a)

2 μm

contact spot

silicon rich inclusions no SiNx:H

(b)

Figure 3.48: Magnified SEM images of silver-aluminium contact spots on wafers with (a) and without (b) SiN_x:H layer. (b) The dark regions in the contact spot of sample ”no SiN_x:H”

are silicon-rich inclusions.

the contact spots and the metal of the bulk contact, the position of the former silicon surface can be seen at some contact spots. In figure 3.49 for the contact spot in the middle the position of the former silicon surface is marked by the two blue arrows. The layer that is visible between the arrows contains a high amount of silicon. A direct connection between the bulk contact and the silver-aluminium contact spots again only exists locally. This observation can be explained by a local material exchange from the silicon surface to the bulk contact and vice versa. Only at areas where particles of the silver-aluminium phase get in touch with the silicon surface, silicon can be dissolved and diffuses into the bulk contact. Silver-aluminium contact spots grow below the silicon surface which is partially removed during contact formation. At some areas, however, part of the surface silicon remains. The silicon spots and lamellas in the contact spots and the silver-aluminium phase of the bulk contact that are not part of the former silicon surface, however, are assumed to precipitate during cooling down the wafers. As larger contact spots grow for wafers without silicon nitride, more silicon is solved and therefore more silicon precipitates can be observed in the silver-aluminium phase.

The contact formation process for wafers without silicon nitride layer does differ in two main aspects from the contact formation with an anti-reflection layer: As no silicon nitride layer is present, the exchange of material is not limited to holes in the anti-reflection coating. However, material exchange occurs only at contact points where the silver-aluminium phase and silicon get in direct touch. The material exchange can start earlier in the firing process, as no dielectric layer, except possibly the thin native oxide layer, needs to be opened. Therefore, for wafers without silicon nitride layer, contact spots are larger than in the case with silicon nitride. The size of the grown together contact spots is comparable with the size of the areas where the aluminium-containing part of the contact get in touch with the silicon nitride/silicon of the wafer (bright areas

3 μm no SiN_x:H

Si

AgAl Al containing glass

Ag

contact spots

Figure 3.49: SEM image of cross-section of screen-printed contact on sample without silicon nitride.

in figure 3.47 (a)).

In section 3.6.2 it was shown that the silicon nitride layer acts as a mould that gives the contact spots the shape of the former silicon surface. For wafers without silicon nitride, no such mould exists. Therefore, the contact spots do not take the shape of the silicon texture. However, the contact spots resemble rounded silicon pyramids due to the fact that they grow via local contact points below the silicon wafer surface.

3.7.3 Summary: influence of passivation layers on the contact