Buffer engineering

The fine tuning of the buffer layer properties to adapt it in the best way to the CIGS absorber is known as “buffer engineering”. When absorber and buffer are faced together, the properties of the resulting heterojunction are mainly influenced by the energy-bands structure and the physical interface. Each of these factors deserves detailed analysis since they are combinations of various aspects and may affect fill factor as well current and voltage output of the solar cell. Consequently, the deepest understanding of the materials and the finest control in their depositions are necessary to reach higher efficiency of alternative-buffered devices.

7.2.1. Band alignment at the pn-junction

The energy-band structure of a junction strictly depends on its materials. Despite silicon, CIGS is a direct band-gap semiconductor: no phonons are involved in the charge-carriers transitions and conduction and valence bands extremes are related to the same reciprocal-lattice orientation. The main phenomena to be considered in the study of the band line-up are the band bending, typical of pn-junction, and the band discontinuities, which are unavoidable in heterojunction solar cells for both conduction and valence band. Since absorber, buffer and TCO are not the same material, the energy gap and electron affinity of p-type and n-type semiconductors are obviously different, thus bringing to energy discontinuities at the interface. The CIGS absorber is the smaller bang-gap material of the solar cell, while the TCO has the wider energy gap. Between them, the buffer layer plays a central role in the device behaviour creating, towards the absorber, the so-called Conduction Band Offset (CBO):

discrepancy in the conduction band of two facing materials. CBO is probably the most important energetic aspect of the junction, managing the electron transfer through it and hence affecting the nature of the solar cell, i.e. its efficiency. The offset may be either positive or negative, in some cases even zero. Figure 7.1 shows the band alignment for positive and negative CBO.

Figure 7.1 - Examples of a positive (a) and a negative (b) conduction band offset between absorber and buffer layer

A large positive CBO would be detrimental for the performance, introducing a high barrier for the electrons and consequently reducing their probability to overcome the obstacle, if not providing

additional energy to drive them towards the n-type TCO. On the other side, for negative CBO, theory predicts an increment of recombination probability due to the junction moved towards the physical interface.

However, theory suggests that excellent performance of solar cells can be achieved with a zero or small positive CBO between the CIGS and the buffer. Voc and FF are maximized for 0<CBO<+0.4eV. This CBO values are extremely healthy since CBO pushes the electrical pn-junction slightly into the CIGS, away from the physical material interface, without blocking the current. Moving the electrical junction below the surface of the absorber means turning the CIGS upper part into n-type, as pointed out by the position of the Fermi level. The new situation forming a CIGS homojunction is called “buried junction”

and it is supposed to lower the interface recombination since the hole concentration at the interface is reduced. Actually, the CIGS surface has been observed to be already turned into n-type. Two different mechanisms have been proposed: Cd diffusion into the CIGS surface layer and Se vacancies (VSe+) at the CIGS surface. Se vacancies in the topmost layer could pin the Fermi level close to the conduction band. The CIGS surface would be turned n-type and the built-in electric field could be able to drive the migration of Cu atoms towards the CIGS bulk, generating Cu-poor phases called order vacancy or order defect compound (OVC or ODC). The inverted top-layer is expected to be thinner than 15nm and with doping level of 10¹¹-10¹² cm^-3, orders of magnitude lower than the typical CIGS doping density of 10¹⁶ cm^-3, but comparable with a reasonable surface states density of 10¹² cm^-2 necessary for the surface inversion. However, the CIGS exposure to air induces the passivation of VSe by oxygen which reduces the inversion. (112) (113) (114) (115) (116) (117) (118) (119)

In conclusion, for low positive or zero barrier height in the conduction band, the energy provided by the space-charge-region electric field to the electrons is enough to guarantee negligible losses during their extraction. At the same time, the further sink of electrical junction minimizes recombination at the interface. The bands alignment in CIGS solar cell is illustrated in Figure 7.2. It is remarkable to point out that CdS buffer layer provides a small positive CBO towards CIGS absorber, for a wide range of Ga content. The success of CIGS/CdS solar cells can be attributed to both favourable CBO and good interface properties. Unfortunately, enhancing the GGI ratio results in larger band gap, raising the EC of CIGS only, which brings to non-favourable CBO and lower efficiency devices. This might be the main reason for which CIGS solar cells buffered with CdS work so well for all CIGS compositions except the Ga-richer ones. (120)

Figure 7.2 - Typical CIGS solar-cell bands line-up with CdS or Zn(O,S) buffer.

Zn(O1-x,Sx) is a promising candidate to substitute CdS thanks to its tuneable band gap. Changing the composition, it is possible to vary the CBO (and energy gap) and find the most fitting one for the CIGS absorber. Considering the sulphur content x as the variable, one can range over from 3.2eV for x=0 (ZnO) to 3.6 eV for x=1 (ZnS), with a minimum of 2.6 eV at about x=0.45 due to the bowing. But the most interesting characteristic is the trends of valence band and conduction band, which are not linear with the sulphur content. Remarkably, the conduction band value is almost constant from x=0 to about x=0.45 and then starts to increase, while the valence band has the opposite trend increasing with the S content until x=0.5 and then remaining constant, as shown by Figure 6.3. (121) (122)

Figure 7.3 – Schematic representation of Zn(O1-x,Sx) valence and conduction bands (a) and energy gap (b) in function of its composition x (sulphur content), as found in literature: (121) (122)

The CBO formed by Zn(O,S) towards CIGS depends both from the buffer composition (S content) and the absorber composition (Ga content) at the interface. Table 2 gives an overview of Zn(O1-x,Sx)-CBOs in function of its composition for a CIGS-GGI of 30%, compared with CdS buffer. ZnO has a negative

CBO towards CIGS. Increasing the S content, CBO turns positive only for x>0.5. The same CBO height of CdS (+0.3 eV) is reached for S content slightly higher than 0.7. Later, the CBO abruptly rises, following the increasing of the Zn(O,S) conduction band and causing the current block for CBO >0.5 eV. This phenomenon is highlighted in the JV measurements by the so called “roll-over-effect”, which is tremendously detrimental for the device behaviour, since causes the FF dropping and may even result in Voc and Jsc loss. (121) (122) (123)

Table 7.2 – Energy gap and relative CBO formed towards CIGS-GGI=30% of ZnO, CdS and Zn(O,S) buffer layers.

Buffer S content (x) Buffer EG (eV) CBO (eV)

CdS - 2.4 +0.3

ZnO - 3.3 -0.2

Zn(O1-x,Sx) 0.3 2.7 -0.2

Zn(O1-x,Sx) 0.5 2.6 -0.1

Zn(O1-x,Sx) 0.7 3.0 +0.2

Zn(O1-x,Sx) 0.8-0.9 3.2 +0.7

ZnS 1 3.6 +1.2

7.2.2. Physical interface

The conduction band offset is probably the most important parameter to be taken in account when studying the improvement of buffer layer in CIGS-based solar cells, but it is not the only one. Several other factors should affect the CIGS/buffer interface formation and the role they play is strictly connected to the material used as buffer and the deposition technique and process applied.

Among the huge variety of interface features, the key parameters to obtain high-efficiency CIGS-based solar cells are:

 post-deposition treatment with alkaline compounds;

 CIGS surface sulfurization to passivate Se vacancies and lower the position of the absorber EV;

 enhanced inversion of the absorber interface region due to the doping (by Cd or Zn) of CIGS surface;

 low Ga content in the uppermost CIGS layer to reduce the point-defect concentrations;

 lattice matching between CIGS and buffer layer;

 alkaline (Na or K) doping of CIGS grains close to the surface to control the CIGS doping in the SCR;

 CIGS roughness, to be reduce to avoid shunt paths due to non-conformal growth of the buffer.

All of them are suggested to play a crucial part in the success of CIGS/CdS-junction solar cells. For this reason, it is highly recommended to understand how their role changes when using alternative buffer layers and processes different then chemical bath deposition.