n- and p-Type Wafers

The entire development of standard solar cells is based on the use ofp-type, e.g.

boron-doped crystalline silicon. This development is historically due to the fact that in the early days of photovoltaics, wafers were manufactured fromp-doped waste products of the semiconductor Industry. Over the years, the amount of waste was no longer sufficient and much effort was made to produce high-qualityp-type silicon specifically for photovoltaics. In the pastp-type material was cheaper to produce than n-type material. Today, both kinds of materials have more or less the same production costs. In principle, the question naturally arises why one generally uses n-type crystalline silicon for heterojunction (HJT) solar cells and notp-type silicon.

There are several reasons which are listed in the next sections:

(a) Capture Cross-sections In the comparison of n-type and p-type silicon, the key quantity “capture cross-section” plays a decisive role. The capture cross-section indicates the size of the capture radius for electrons or holes, for capture by foreign atoms (impurities).¹⁹In the following example, we assume that iron is the dominant impurity in silicon. We will investigate the cross section capture of iron for electrons and holes more closely.

As an example we consider capture cross section of iron [13] for electrons (e) and holes (h):

σe=3.5×10⁻¹¹cm⁻² σh=4.5×10⁻¹⁶cm⁻²

19Actually the term «capture cross-section» is used not only in the case of impurities, but in the case ofallrecombination centres or defects—as an example it is also used in the case of “dangling bonds”, which appear in amorphous silicon layers (see Chap. 6).

7 Crystalline Silicon Solar Cells: Heterojunction Cells 181 The capture cross section is 10⁵times larger for electrons than for holes regardless of whether we haven-type material orp-type material. However inn-type material the holes are the minority carriers and their density is less influenced by the capture process than the density of majority carriers (electrons).

Thanks to the smaller capture cross-section of the holes, the latter are captured less frequently by the impurities (here as an example, by the iron impurities) and have longer lifetimes. At the same time, they are the minority carriers, which contribute significantly to the open circuit voltageV_oc. This is one reason whyn-type silicon leads to better cell properties thanp-type silicon.

So, even in the hypothetical case wheren-type material andp-type silicon would have the same number (e.g. the same density) of impurities, and the same impurities like iron (Fe), these impurities are less harmful inn-type silicon than inp-type silicon.

Thus,n-type silicon does not have to be gettered (see section e below), at all, in order to obtain the material quality needed for solar cells.

(b) Carrier Lifetimes The lifetime of minority carriers is generally taken as a marker for the quality of a solar cell material. The lifetime of minority carriers indicates how long the minority charge carriers will, on an average, exist before they recombine with carriers of opposite charge (e.g. with the majority carriers). If the lifetime is too short, charge carriers cannot reach the junction from their point of origin in bulk silicon and recombine before. For n-type material the lifetime of holes (minority carriers) is up to approximately 10 ms, forp-type material the lifetime of electrons (minority carriers) is only up to approximately 2 ms. The probability that the minority charge carriers reach the junction and contribute to the solar cell current is, thus, two times higher forn-type material.²⁰

(c) Carrier Mobilities Inn-type material, the holes are the minority carriers and their mobility is only 450 cm²/Vs. The mobility of the electrons is three times higher at 1400 cm²/Vs. This then is the case in a cell with a backpn-junction: in the best case, the charge carrier generation takes place in a cell near the back surface and the hole diffuses to the back side by the shortest route. In the less favourable case, the hole has to take a longer path through the bulk to reach the junction. But since the mobility of the holes is relatively small, a high-quality bulk material is crucial, so that even in unfavourable cases the charge carriers have a high probability to reach the junction.n-type material satisfies these requirements, because the capture cross-section²¹σpfor holes inn-type silicon (minority carriers) is smaller than the capture cross-sectionσnfor electrons inp-type silicon (minority carriers).

(d) Boron Oxygen Complex and Degradation Effects Furthermore, boron in sili-con tends to form together with oxygen so-called boron-oxygen complexes,²²which

20Diffusion lengthL=(τ*D)^1/2.

21Explanation of capture cross-sections: Low capture cross-section means that activity radius around impurities is small and recombination activity is low. High capture cross-section means that activity radius around impurities is large, and recombination activity is high.

22In addition to boron, iron and copper in combination with oxygen can also produce interference effects.

182 S. Leu and D. Sontag lead to theLID (light induced degradation, see Chap.10) effect, which in its turn causes degradation in the performance of a solar cell of 5–15%. LID effects occur above all in monocrystallinep-type silicon, which is doped with boron and only to a lesser extent in multicrystalline solar cells; they hardly occur, at all inn-type silicon.

When we bring more hydrogen into the bulk, we can reduce the LID effect and we have less oxygen in silicon.²³ LID has according to recent investigations [14] been observed to occur especially in solar cells that have undergone high temperature cell processes (see Chap.10). Another way to reduce the LID effect is to dope with gallium (Ga). Because of patent restrictions,p-type gallium-doped wafers could not be offered commercially until 2018; it is at the moment not clear whether Ga-doped material will ever come onto the market.²⁴

With hydrogen we can reduce the LID effect. But the higher the hydrogen content, the higher the LETID (light and elevated Temperature Induced Degradation) effect. LETID occurs mainly in multicrystalline material when solar cells are exposed to ambient temperatures higher than 50 °C and is less pronounced in monocrystalline solar cells. It is enhanced because of the high firing temperature (800–1000 °C) used for the metallisation paste during cell manufacturing (see Chap.5). Due to these high temperatures, more hydrogen penetrates into the cell and, thus, the higher the LETID effect will be. The thinner the wafer, the less pronounced are LID and LETID.

If both effects, LID and LETID, occur simultaneously, this can lead to degradation, which is as high as 15%. Since HJT cells are manufactured withn-type material and additionally very low process temperatures are applied, LID and LETID effects are negligible.

(e) Gettering According to the above explanations a–d,n-type material has signifi-cant advantages overp-type material when both have the same density of impurities.

This is also the reason why n-type material does in general not require gettering processes. On the other hand, it is not at all possible to apply gettering processes to HJT cells: Due to the temperature sensitivity of the amorphous layers, process temperatures for HJT cells must not reach more than approximately 200 °C, whereas for gettering one requires temperatures of more than 500 °C. Because of this limita-tion, gettering²⁵cannot be used during the production of HJT cells, as it is used for the production of “standard” (homojunction) p-type crystalline silicon solar cells.

Indeed, gettering is not at all necessary for HJT cells withn-type material. This is because today’sn-type silicon is so pure that a high temperature gettering step (even if it could be applied) would no longer significantly contribute to cell improvement.

23Oxygen penetrates the silicon during crystallization and cannot be completely avoided.

24This is because it is very difficult to obtain homogeneous doping of silicon with Gallium (see segredation coefficient Chap.5).

25When silicon is cooled after the crystal pulling process, thermal donors (TD) may be formed by oxygen clusters. These are negatively charged. Thermal donors influence the resistivity. In the n-type material the resistivity decreases, in thep-type material it increases. TD dissolve at over 500 °C during a gettering process. However, the influence on the improvement of the cell efficiency is not economically meaningful (0.2% abs.); this is the reason why gettering processes are not carried out withn-type material.

7 Crystalline Silicon Solar Cells: Heterojunction Cells 183 (f) Thin Wafer With a share of more than 50% in the total production costs, the production of wafer material is the biggest cost driver in the manufacturing of a solar cell. Therefore, wafer production also offers the greatest potential for savings. Due to the low process temperatures and the symmetrical cell structure of heterojunction cells, the use of very thin wafers becomes possible here. A thick metal layer (Alu-minium) on the back side, as is usual with monofacial homojunction cells, leads to a warping of the cells—this is due to the different thermal expansion coefficientsαof the metal and silicon; these differences in the values ofαhave a negative effect—in conventional cell production—because of the high firing temperatures (>800 °C) and the subsequent rapid cooling. This can lead to major problems during cell production such as warping and cell breakage during screen printing. In the case of the hetero-junction cell, the metal layer is completely omitted, so that thinner wafers can be used for cell production. This leads to two opposite effects: A thinner wafer means that more light passes through the solar cell without being absorbed, so less light contributes to carrier generation. This reduces the short-circuit current densityJscof the solar cell. At the same time, passivation of the wafer surface is becoming increas-ingly important, as more charge carriers reach the surfaces and do not recombine in the bulk. A high degree of passivation, as is the case with heterojunction cells, leads to an increase in the open circuit voltageV_oc. To a certain extent, these two effects (lowerJ_scand higherV_oc) cancel each other out and cell efficiency remains the same.

Nevertheless, in addition to the cost advantage, there is also a performance advantage for thin wafers. When the solar cells are integrated into a module, series resistance losses occur in accordance with (7.4).

Ploss=I²R (7.4)

Equation (7.4) means that the power lossPlossincreases quadratically with the total current in the module. A lower current means, thus, a reduction in module losses.

It can be seen that with the same power class of individual cells, the power of the modules from thin cells is higher than that of the modules from thicker cells [15].

According to Fig.7.12the optimum wafer thickness for a heterojunction cell is 80–100μm whereas for homojunction cells the minimum cell thickness is about

>140μm.

For the reasons a–f given above, we may conclude that the design of the silicon heterojunction cell optimally complements the properties ofn-type material, and that n-type wafers are ideally suited for the production of highly efficient HJT cells.