Solar

Using ALD for improved efficiency crystalline silicon solar cells

Plasma ALD Al<sub>2</sub>O<sub>3</sub> shows superior surface passivation quality, and when implemented in a highly doped p-type emitter cell, has yielded cell efficiencies up to 1% absolute higher than current techniques.

Plasma ALD Al2O3 shows superior surface passivation quality, and when implemented in a highly doped p-type emitter cell, has yielded cell efficiencies up to 1% absolute higher than current techniques.

Chris Hodson, Oxford Instruments Plasma Technology, Bristol, UK, 
Erwin Kessels, Eindhoven University of Technology, Netherlands

Minimizing losses in solar cells, thereby improving their efficiency is a key focus of any organization involved in photovoltaic (PV) research. To the outside observer, what may seem like relatively small efficiency improvements can significantly reduce the cost/Watt of the cell and transform the economic situation.

Near exhaustive efforts have been employed in first generation crystalline silicon (c-Si) solar cell research to improve efficiencies and as a result ever more novel techniques must be considered to make further significant gains. One such recent development is the introduction of a relatively new technique to solar cell device structures, atomic layer deposition (ALD).

Figure 1. Schematic representation of one cycle of the plasma-assisted atomic layer deposition (ALD) process of Al<sub>2</sub>O<sub>3</sub>. The precursor is Al(CH3)3 and the plasma gas is O<sub>2</sub>. The cycle can be repeated until the film thickness projected is achieved.

ALD has recently been adopted in production by the mainstream semiconductor industry to answer the needs of downscaling. ALD is the method of choice for depositing high quality films with ultimate growth control and with excellent step coverage on very demanding high-aspect ratio features. ALD addresses the current boost in nanoscience and nanotechnology research with its demanding topography requirements.

Although there are many potential applications in PV that could benefit from layers produced by ALD, only recently has significant research taken place for c-Si solar cells. In a collaboration between the Technical University of Eindhoven and the leading Fraunhofer Institute for Solar Energy Research (ISE) in Germany, a then record efficiency of 23.2% was obtained for passivated emitter with rear locally diffused (PERL) solar cells based on n-type silicon by the application of an ultra-thin ALD aluminium oxide (Al2O3) surface passivation layer at the front of the solar cell [1] using the Oxford Instruments’ FlexAL reactor. At the time of the research, this represented a 1% absolute efficiency increase compared to more commonly used passivations, demonstrating the benefits of considering alternative techniques.

Atomic layer deposition

As shown in Fig. 1, the virtue of the ALD technique is that the deposition is controlled at the atomic level by self-limiting surface reactions by alternate exposure of the substrate surface to different gas-phase precursors. Each surface reaction occurs between a gas phase reactant (precursor) and a surface functional group creating a volatile product molecule that desorbs from the surface, and a new surface functional group that is not reactive with the precursor. After pumping away the first precursor and the volatile reaction products, a second precursor is introduced, which deposits a second element through reaction with the new surface functional group and then restores the initial surface functional group. This set of reactions form one ALD-cycle resulting in less than one atomic layer of film growth, typically 0.5–1.0Å per cycle. The ALD-cycle can be repeated until the desired film thickness is reached.

Figure 2. A schematic representation of a c-Si solar cell illustrating the application of ALD Al<sub>2</sub>O<sub>3</sub> for the passivation of both the front and rear surface.

Furthermore, unlike chemical vapor deposition (CVD), the deposition rate is not proportional to the flux on the surface. Therefore, the same amount of material is deposited everywhere on the surface, even in high-aspect ratio structures, when there is sufficient flux. Other benefits of ALD are the good uniformity that can be achieved on large substrates, the relatively low substrate temperatures used in the process (temperature window typically 150–350°C), and the fact that ALD can readily produce multilayer structures.

Al2O3 deposited by ALD is a well understood process and most commonly uses trimethyl aluminium (Al(CH3)3 or TMA) as the aluminium source and either water, ozone or oxygen radicals as the oxidant. Each cycle consists of TMA dosing followed by a purge, then oxidant exposure followed by a purge. Typically 0.9–1.2Å is deposited in each of these cycles depending on the deposition temperature and the choice of oxidant used.

The choice of oxidant depends on the application. In general oxygen radicals generated by a plasma source are more reactive than water relying on purely thermal energy. This increased reactivity can give improved film quality with lower impurity levels. The plasma radicals can also be used to treat/clean the surface prior to deposition.

ALD is fully vacuum compatible with other common deposition processes already in use in solar cell production, e.g., plasma-enhanced chemical vapor deposition (PECVD). The main challenge to overcome prior to wider adoption of ALD in PV manufacturing is the demanding throughput requirements and the comparatively slow deposition rate of ALD. These are discussed in relation to surface passivation below.

ALD in PV research

ALD has been researched for use in several solar cell applications dating from the early nineties to the present day. These applications include absorber films, buffer layers, interface layers, transparent front contacts, photoanodes and, and as shown in Fig. 2, most recently surface passivation layers [2].

As an example, ZnO is a transparent conducting oxide with a wide range of uses. The addition of aluminium doping can also improve the conductivity and combined with the high transmission make ZnO:Al an interesting alternative to indium tin oxide (ITO).

ZnO-based films with inclusions by S and Mg have been investigated as buffer layers between the CI(G)S absorber layer and the window layer [3]. ZnO is commonly applied in second generation thin film solar cells as a front contact and back reflector. Third generation dye-synthesized solar cells have been demonstrated using high surface area ZnO nanotubes as photoanodes. The ZnO nanotubes were synthesized by combining anodic aluminium oxide (AAO) templating and ALD [4].

Within first generation c-Si solar cell technology, ALD has only been employed very recently as a surface passivation layer.

Recent breakthroughs in c-Si PV by ALD Al2O3

In the battle to improve efficiency in c-Si solar cells, one of the critical factors is avoiding recombination losses of the charge carriers created in the silicon. One of the problems is that sooner or later, these carriers meet a surface or interface and the presence of any defects at the surface or interface can increase recombination rates. The c-Si surface can be well passivated using a thin film of either silicon dioxide or silicon nitride.

Silicon oxide (SiO2) would normally be thermally grown and undergoes additional chemical annealing to provide a very high quality interface with a very low defect density. The high temperature required to thermally grow the SiO2 however, generally limits its applicability to high purity c-Si material, successful on lab scale for high efficiency cells, but limited industrial use.

Figure 3. Measured emitter saturation current density of B-diffused p-type emitters as a function of sheet resistance. Surfaces are passivated by forming-gas-annealed thermal silicon oxide, PECVD silicon nitride, PECVD amorphous silicon and plasma assisted ALD-Al<sub>2</sub>O<sub>3</sub>. A lower the emitter saturation current density indicates a better passivated surface.

Silicon nitride (Si3N4) is the most commonly used passivation of industrial solar cells and is typically deposited by PECVD. This leads to a very good surface passivation due to a built in electrical field from positive fixed charges with a typical density of 1012cm-2 in the Si3N4. The engineering of electrical fields below the silicon surface causing field-effect passivation makes the application of Si3N4 particularly useful in the case of highly doped n-type c-Si emitters, where recombination is reduced by the minority charge carrying holes being shielded from the front surface.

Conversely to silicon nitride, aluminium oxide contains fixed negative charges. Recently Agostinelli et al. and Hoex et al. have shown that very thin ALD-Al2O3 films (5-30nm) yield an excellent level of surface passivation on both low resisitivity n-type and p-type Si after a post deposition anneal at 425°C [5–7]. The excellent surface passivation by Al2O3, outperforming any other thin film passivation layer (Fig. 3), can be related to a satisfactory low interface defect density in combination with a strong field-effect passivation. The electrical field is induced by a very high negative fixed charge density of up to 1013cm-2 present in the Al2O3 film at the interface with the underlying Si substrate [8]. Due to the difference in polarity between the charges in Si3N4 and Al2O3, the Al2O3 also works very well on highly doped p-type c-Si emitters.

The excellent passivation of lightly doped p-type Si by Al2O3 has recently been confirmed through the application of ALD-Al2O3 and ALD-Al2O3/PECVD-SiO2 stacks at the rear of PERC (passivated emitter and rear cell) solar cells [9]. An independently confirmed energy conversion efficiency of 20.6% and a rear surface recombination velocity of 70cm/s were reported (for the reference cell with thermally grown SiO2 these values were 20.5% and 90cm/s, respectively). No separate anneal was necessary since the thermal treatment within the solar cell process flow appeared sufficient to obtain the excellent surface passivation properties. Furthermore, it was confirmed that the solar cells with Al2O3 rear surface passivation did not suffer from the so-called parasitic shunting effect. The latter is observed when the positive charge dielectric Si3N4 is applied at the rear surface with metal contacts [13]. The reason is that the high fixed negative charge density in the Al2O3 induces an accumulation layer instead of an inversion layer in the c-Si underneath the rear surface.

The presence of a high negative charge density makes Al2O3 also an ideal candidate for the passivation of highly-doped p-type silicon. The latter is important for n-type solar cells that have an enormous economical potential but for which the passivation of p-type emitters (either B- or Al-doped) has proved challenging as it would require a high density of negative charge rather than positive. Hoex et al. demonstrated experimentally that Al2O3 yielded a lower surface recombination velocity on highly-doped p-type silicon than thermal SiO2, as-deposited PECVD-Si3N4 and amorphous silicon [10].

The fact that the high surface passivation performance can also be obtained at the device level has recently been demonstrated by a passivated emitter with rear locally diffused (PERL) solar cells in collaboration with the Fraunhofer ISE Institute. These n-type PERL solar cells in which Al2O3 was applied to passivate the p-type emitter yielded an independently confirmed solar cell efficiency of 23.2% [1]. In these studies plasma assisted Al2O3 films have been deposited in the FlexAL reactor [10, 11]. The surface passivation of the plasma assisted ALD films was found be to excellent and required little optimization. Thermal ALD and plasma assisted ALD films have been grown in the Oxford Instruments OpAL reactor. Thermal ALD films have also shown promise, but their route to integration has proven to be less trivial. Initial results prior to significant optimisation showed the plasma assisted film had a much greater negative fixed charge density that in turn can explain significantly higher measured effective lifetimes of the charge carriers of p-type c-Si and a better surface passivation.

These results clearly demonstrate the potential of ALD-Al2O3 film within c-Si solar cell technology. Besides the presence of a high fixed negative charge density and the low interface defect density the merits of the combination ALD and Al2O3 are the absence of absorption in the visible part of the spectrum (Al2O3 has a band gap of ~8.8eV), the high UV stability, the very thin film thickness, the low temperature processing (avoiding bulk lifetime degradation of c-Si), the gentle deposition conditions (virtually no surface damage by the impact of energetic ions), and the excellent uniformity on large substrates.

Up-scaling the process

As demonstrated in this paper, there are many potential technical advantages in terms of device performance and design possibilities in adopting ALD layers in the solar cell fabrication sequence. Additionally, it has recently been demonstrated that Al2O3 and stacks of Al2O3 and Si3N4 are thermally stable against high-temperature firing processes as applied for screen printed solar cells, the technology which is the work horse of the c-Si photovoltaics industry [12]. However, as previously mentioned, achieving economically viable throughputs of thousands of wafers per hour by ALD presents challenges. Oxford Instruments and the Technical University of Eindhoven are working on calculations and solutions to reach the efficiency increase versus manufacturing costs of implementing the surface passivation per cell produced. Factors that must be considered include; capital equipment depreciation, precursor costs and deposition rate (inc. cycle time) per cell. Calculation models show the passivation film thickness to be a key parameter in the economic viability and research is on-going using well-passivating ultra-thin Al2O3 films with only a few nanometer thickness.

Conclusion

The introduction of ALD-Al2O3 as a surface passivation shows great promise. Plasma ALD shows consistently superior surface passivation quality, and when implemented in a highly doped p-type emitter cell, has yielded cell efficiencies up to 1% absolute (6% relative) higher than current passivation techniques.

Acknowledgments

FlexAL is a registered trademark and OpAL is a trademark of Oxford Instruments.

References

  1. 1. J. Benick, B. Hoex, M.C.M. van de Sanden, W.M.M. Kessels, O. Schultz, S. Glunz, Appl. Phys. Lett. 92, 253504 (2008).
  2. 2. W.M.M. Kessels, B. Hoex, M.C.M. van de Sanden, Proc. of the 33rd IEEE Photovoltaics Specialists Conference, San Diego, U.S.A. (2008).
  3. 3. U. Malm, J. Malmström, C. Platzer-Björkman, L. Stolt, Thin Solid Films 480, 208 (2005).
  4. 4. A.B.F. Martinson, J.W. Elam, J.T. Hupp, M.J. Pellin, Nano Lett. 7, 2183 (2007).
  5. 5. G. Agostinelli, A. Delabie, P. Vitanov, Z. Alexieva, H.F.W. Dekkers, S. de Wolf, G. Beaucarne, Sol. Energy Mater. Sol. Cells 90, 3438 (2006).
  6. 6. B. Hoex, S.B.S. Heil, E. Langereis, M.C.M. van de Sanden, W.M.M. Kessels, Appl. Phys. Lett. 89, 042112 (2006).
  7. 7. B. Hoex, J. Schmidt, P. Pohl, M.C.M. van de Sanden, W.M.M. Kessels, J. Appl. Phys. 104, 044903 (2008).
  8. 8. B. Hoex, J.J.H. Gielis, M.C.M. van de Sanden, W.M.M. Kessels, J. Appl. Phys. 104, 113703 (2008).
  9. 9. J. Schmidt, A. Merkle, R. Brendel, B. Hoex, M.C.M. van de Sanden, W.M.M. Kessels, Prog. Photovoltaics 16, 461 (2008).
  10. 10. B. Hoex, J. Schmidt, R. Bock, P.P Altermatt, M.C.M. van de Sanden, W.M.M. Kessels, Appl. Phys. Lett. 91, 112107 (2007).
  11. 11. S.B.S. Heil, J.L. van Hemmen, C.J Hodson, N. Singh, J.H. Klootwijk, F. Roozeboom, M.C.M. van de Sanden, W.M.M. Kessels, J. Vac. Sci. Technol. A 25, 1357 (2007).
  12. 12. G. Dingemans, P. Engelhart, R. Seguin, B. Hoex, M.C.M. van de Sanden, W.M.M. Kessels,, Proc. of the 34th IEEE Photovoltaics Specialists Conf., Philidelphia, U.S.A. (2009).
  13. 13. S. Dauwe, L. Mittelstadt, A. Metz, R. Hezel, Prog. Photovoltaics 10, 271 (2002).

Chris Hodson received his degree in applied physics at Durham U. in the north east of England and is the ALD Product Manager at Oxford Instruments Plasma Technology, North End, Yatton, Bristol, BS49 4AP UK; ph.: +44 1934 837000; email [email protected] Erwin Kessels received his MSc and PhD degree at the Eindhoven U. of Technology and is currently an associate professor at the Eindhoven University of Technology, Department of Applied Physics, PO Box 513, 5600 MB Eindhoven, Netherlands; ph,.: +31 40 2473477; email [email protected]