The Basic Structure of a Heterojunction Cell

(a) Two Different Semiconductors

Heterojunction technology uses two materials with different energy bandgaps.

«Hetero»⁴means “different” in Greek. This is the big difference with the so-called

“homo⁵-junction cells” discussed in Chap.5(PERC cells). In the case of the hetero-junction silicon solar cell, the following two materials are used: crystalline silicon (c-Si) and amorphous silicon (a-Si: H). Amorphous silicon has a bandgap of 1.7 eV, in contrast to crystalline silicon with 1.12 eV. In this configuration, the bandgap of the c-Si (n) lieswithinthe bandgap of the a-Si:H silicon (see Fig.7.5. drawings A and B). Such a configuration is called a «heterojunction type 1⁶» contact. Other heterojunction contacts are:

1. those relating to amorphous and microcrystalline silicon tandem cells, 2. those between Cadmium telluride and Cadmium Sulfide (CdTe/ CdS),

Fig. 7.5 The bandgap energy of crystalline silicon is 1.12 eV and the bandgap energy of amorphous silicon is ~1.7 eV. The dashed violet line shows the Fermi level. The blue line shows the conduction band edge and the red line represents the valence band edge. Drawings A and B show the two semiconductors. “A” represents the amorphous part a-Si(p) and “B” the crystalline part c-Si(n).

The two individual materials A and B are joined together and the respective majority carriers of the materials A and B will diffuse into the other area via the concentration balance. Only the front side is shown

4Hetero derives from the Greek word Heteros meaning «different» or «other».

5Homo-, Greek prefix expressing the notion of “same, identical”.

6Configurations in which the bandgap of one of the materials is lying within the bandgap of the other material are called «heterostructures of Type 1». However, if the bandgaps are «staggered»

in the sense that the conduction band of the second material is lower than the conduction band of the first material, and at the same time, the valence band of the second material is also lower than the valence band of the first material, then we have a «heterostructure of Type 2».

170 S. Leu and D. Sontag 3. those between Copper-Indium-Gallium-Diselenide and Cadmium Sulfide

(CIGS/CdS),

4. those relating to Group⁷III–V tandem cells (GaAs, etc.).

These other contacts are not described in this chapter: (1) has been mentioned in Chap.6, (2) and (3) will be mentioned in Chap.8—whereas (4) will not at all be treated in this book. Indeed, although III–V tandem cells give the highest cell efficiencies ever obtained (over 40%), their fabrication cost is at present prohibitively high, so that they are only used in special situations (like concentrating photovoltaics, CPV, see Chap. 10, Sect.10.3.2) and in space applications.

(b) Band Diagram and Tunnelling

In Fig.7.5the construction of the band diagram of a heterojunction cell is shown [6]—for the sake of simplicity this is done only forthe front side, here for thep-side, where thepn-junction is.The band diagram is essential to understand the benefits of passivated contacts, which ultimately characterize the HJT cell. Let us first look only at the individual layers—the individual components of the heterojunction before joining them together. Individual layers: The bandgap of thep-doped, amorphous silicon a-Si:H(p) layer is shown hatched on the left in Fig.7.5, drawing A. Just next to it (drawing B), is the bandgap of then-doped silicon c-Si (n) layer (wafer). This bandgap, dotted in drawing B, is within the first bandgap («type 1» heterojunction).

When building the heterojunction, e.g. putting in contact the (p) amorphous silicon layer and the (n) crystalline silicon wafer, equilibration occurs, which will determine the band diagram as discussed below.

As in the dark—without sunlight—the potential in a material has to be bal-anced, the two Fermi levels will be adjusted to the same potential. Let us look at what happens in detail: When we bring the two individual materials A and B into contact,⁸ the respective majority charge carriers begin to diffuse to the other side.

In the n-doped silicon c-Si (n) layer, the electrons are the majority carriers. The electrons flow from the right side in Fig.7.6(position➁) to the left side (position

➀).This flow charges the right side (β) positively and the left side (α) negatively, see Fig.7.6. This charge carrier exchange flows until the two Fermi levels have become equal and an equilibrium state forEFprevails. In Fig.7.6the two Fermi levels are therefore on the same line.

Let us take a closer look to the region around the interface. The band bending and the gradient causedthereby in the Conduction Band Edge E_C and in the Valence Band Edge E_V constitute an electrical potential for the charge carriers that drives

7The term «Group» here refers to the numbering of the columns in the periodic system.

8The idea that we start with individual layers, which we afterwards join together is, of course, just a “Gedankenexperiment”, an artifice we imagine in our minds, so as to understand better what happens. In reality the two individual layers are joined together right from the beginning—as the HJT cell is fabricated.

7 Crystalline Silicon Solar Cells: Heterojunction Cells 171

Fig. 7.6 Bandgap diagram after joining the two materials A and B. Shown is thep-side of the heterojunction cell (here the front side, with thepn-junction) without metallization. The band-bending of the valence band and the conduction band is illustrated. The conduction band stretches aroundE_C(=E_C+band-bending), while the valence band stretches aroundE_V (=E_V− band-bending) and its offset is therefore smaller. One can also clearly see the peak that occurs when the two different potentials of the two semiconductors are joined together. For simplicity’s sake, we assume that the convergence takes place at line M. Above the band diagram the cross-section of the solar cell is illustrated. The light hits the solar cell on the left side; the back side of the cell is to the right

them in opposite directions to each other, electrons to the right hand side and holes to the left hand side in Fig.7.6.⁹

In the area of the junction (see Fig.7.6) the transition from the bandgap of amor-phous silicon (~1.7 eV) to the bandgap of crystalline silicon (1.12 eV) takes place.

The edgesECof the conduction band andEVof the valence band are bent. Initially, the valence band follows the conduction band at a distance of 1.7 eV, up to the transi-tion; and then, after the transition the distance becomes 1.12 eV. The transition takes place within a few atomic layers, i.e. within a width of a nanometre or less, whereas the junction itself with the charged regions described above extends over tens of nanometres in the highly doped amorphous silicon and hundreds of nanometres in the lowly doped crystalline silicon. There is therefore a band offset of ~0.6 eV at the interface between the two materials (line M in Fig.7.6), which is typically split into 0.45 eV as conduction-band offset and 0.15 eV as valance-band offset. The down-ward bending due to the junction formation followed by this band offset leads to the

9Those majority carriers that accumulate now on opposite sides constitute a concentration imbalance which is a chemical potential. The process is self-regulating with electrical and chemical potential neutralising each other at every point in space. In fact, this is the requirement which is postulated in the beginning, the combined electrochemical potential is identified with the Fermi level.

172 S. Leu and D. Sontag

Fig. 7.7 Schematic illustration of the flow of electrons and holes. The negatively charged electrons

“slide down” on the conduction band slope to the energetically lower, right side); on the other hand, the positively charged holes flow to the energetically higher side, the left side. They have to tunnel through the very narrow peak

spike in the valence band on the left of the Line M (1st peak). After that—shortly to the right of the vertical line M—the bandgap reduces completely to 1.12 eV. Again, the valence band follows the conduction band. The valence band makes a 2nd peak here, but this time upwards [6].

(c) Charge Transport

Now what is the effect of these features on the carrier currents in the solar cell?

Photogenerationof electrons and holes in the crystalline silicon c-Si(n) layer (wafer) leads to an excess of both electrons and holes in the wafer compared to the equilibrium state described above. Near the entry point of the light (on the left side, position➀ in Fig.7.7), the photogeneration will be at its highest value, and from their onward it will drop off exponentially to the right (see Chap. 4, Fig.4.2). At the back side of the wafer (position➂), the photogeneration will be substantially lower.

Now, we will look at the electrons and holes separately: Electrons will be diffusing from left to right, as their concentration is very high on the left side of the c-Si (n) layer, (position➀, strong photogeneration) and relatively low on the right side (position

➂, weak photogeneration); this is the point where the electrons leave the solar cell, see Fig.7.7.

There will be no electrons “going back” from the c-Si (n) layer to the a-Si:H(p) layer, as there is a potential difference of E_C (Fig.7.6) to be overcome. Holes, on the other hand, flow from the right side in Fig.7.7to the left side, because the valence band edge of c-Si is lower than that of a-Si. The holes can tunnel through the very narrow peak. The holes are prevented from flowing to the right by the peak,¹⁰ and additionally, by the potential difference inE_V(Fig.7.6). Hence, the peaks in

10The probability for charge carriers to tunnel through a peak depends on the height and width of the peak, and also on the density of charge carriers present just before the peak; now, in➁there is a high density of holes, so there will be many holes tunnelling through the peak from left to right,

7 Crystalline Silicon Solar Cells: Heterojunction Cells 173 the valence band help in obtaining a more effective separation of charge carriers.

Since holes can flow freely to the left-side contact, but electrons do not flow in that direction, an ideal contact—without any recombination at the electrode surface, is formed (selective contacts or passivated contacts).

However, the peak must not be too high, otherwise it constitutes a blockade and the holes cannot tunnel through. Conversely, if the peak is too small, cell efficiency and the fill factor are reduced [7]. Later in Sect.7.2.3, we will also look at the back side of the HJT cell and see then that there is a peak at the back side, as well, through which the electrons can tunnel through.

Thanks to the fact that almost no electrons flow through the (p) amorphous silicon layer (and similarly on the back side, no holes flow through the a-Si(n) amorphous silicon layer), recombination losses are avoided. The contacts created in this man-ner are passivated contacts. They are selective and not Ohmic. They let either only the holes pass through or else only the electrons pass through. [8] In addition, these selective barriers are unidirectional: The charge carriers, once they tunnel through the barrier, cannot return. And finally, they are not Ohmic contacts and reduce recombi-nation losses. This has an impact on the equivalent circuit diagram. In the equivalent circuit diagram of a homojunction cell with fired contacts (Al-BSF cell, PERC cell, Chap.5) we have a two-diode model with saturation currents¹¹:J₀₁andJ₀₂. With the passivated contacts of the HJT cell, the saturation currentJ₀₂is minimized, because recombination is strongly reduced.

The saturation currents, which have already been discussed in Chap.3have a very high priority in the evaluation of a solar cell, because they determine the open circuit voltageV_oc of the solar cell and, thus, also its efficiencyη. Figure7.8shows the equivalent circuit diagrams for homojunction and heterojunction cells.J01denotes the leakage current caused by the surface recombination and the losses in the bulk.

Fig. 7.8 aEquivalent circuit for homojunction cells with fired contacts (see Chap.5on PERC cell) andbequivalent circuit for HJT cells.J₀₁denotes the leakage current caused by the surface recombination and by the losses in the bulk.J₀₂denotes the losses caused by the space charge zones. In homojunction cells the losses in the space charge zones are higher than in HJT cells (The reason being that in the heterojunction cell, thep-region is separated from then-type silicon by the intrinsic layer)

whereas in➀there is a very low density of holes (most holes there are siphoned off towards the contact)—thus, there will be very few holes tunneling through the peak from left to right.

11Also called dark or leakage currents.

174 S. Leu and D. Sontag J02denotes the leakage current caused by the space charge zones. In homojunction cells a distinct space charge zone is present. In heterojunction cell, thep-region is separated from then-region by a layer of intrinsic amorphous silicon, and the space charge region is not so pronounced, because of lower doping efficiencies and thinner layers, as compared to homojunction cells.

(d) Structure of Amorphous Layer

Amorphous silicon consists of the same atoms as pure, crystalline silicon. However, it does not show a periodic crystalline structure. Although at the interface with the crystalline silicon it takes over the “crystalline” structure but already three to four bond lengths further, e.g. already after 0.5 nm, the bond angles deviate strongly, so that no periodic correlation as in crystalline silicon is recognizable. This leads to two effects: first, due to the absence of a crystalline structure, the electrical conductivity decreases and, secondly, many open bonds occur, e.g. many unsaturated defects, so-called “dangling bonds” (see also Chap.6). These lead to increased recombination of the charge carriers. The dangling bonds can be neutralized with hydrogen; this step is called “saturation” in technical jargon. This is why hydrogen is added to amorphous silicon. The electrical conductivity of the amorphous material can be increased by doping, but remains always relatively low, due to its amorphous character.

In HJT cells the low conductivity of the amorphous layer is not an obstacle pro-vided the amorphous layer is thin. The charge carriers can then cross this layer vertically. Since the layer is very thin, there is no appreciable loss due to electrical resistance. However, the low conductivity is not sufficient to enable the charge car-riers to move laterally to the metal contacts. It is therefore necessary to cover the amorphous layer with an additional, highly conductive layer. This layer must also be transparent to light.

(e) TCO Layer

Transparent Conductive Oxides (TCO) are suitable candidates, which incorporate all required properties. The charge carriers only have to flow now vertically through the two thin amorphous layersi-pori-nto the cell surface. The lateral current conduction to the metal fingers is then taken over by the TCO layer. In addition to its function as a conductive layer, the TCO layer also takes over the function of anAnti-Reflection Coating (ARC) layer. Such an ARC layer reduces reflection losses—more light can thereby be absorbed by the solar cell. The TCO layer must hence be highly transparent and should ideally not absorb any light itself (parasitic absorption). As explained in Chap.4, this layer has an optimal thickness of 80 nm, if it is used (as is the case here) as asingleanti-reflective layer.

The two objectives that the TCO layer has to meet contradict each other: Either the TCO layer is very transparent but does not conduct well electrically, or it conducts very well electrically and is less transparent because it is thicker or it is doped to a higher extent. In order to determine the optimum for the cell, a compromise has to be found in which the mutual distance of the metal fingers and their thickness must also be taken into account. An optimum is usually found with a relatively low conductivity of the TCO layer combined with fine and narrow metal fingers, which are, typically,

7 Crystalline Silicon Solar Cells: Heterojunction Cells 175 less than 40μm thick. These approximately 60–80 metal fingers consist of highly conductive silver. Usually the metal fingers are printed on the TCO using a silver paste. In most cases, a screen-printing process is used.

The most common TCO material currently used in heterojunction cells isIndium TinOxide (ITO). Depending on the application, the tin content varies between 5 and 10%. A high tin content increases conductivity, but—at the same time—it limits transparency. Further in ITO sputtering, one can have a whole range of transparency versus conductivity by just varying the oxygen content in the film. Production is about choosing the optimal parameters for the most suitable TCO layer. Since indium is a scarce raw material, efforts are currently being made to reduce its use—or in the medium term, even to replace it completely.