Inorganic Nanocomposite Solar Cells

Document Sample
Inorganic Nanocomposite Solar Cells Powered By Docstoc
					Inorganic Nanocomposite Solar Cells by ALD                                               Bent, et al.

                         Inorganic Nanocomposite Solar Cells
                          by Atomic Layer Deposition (ALD)

   Stacey Bent, Professor, Department of Chemical Engineering; James S. Harris, Professor,
   Department of Electrical Engineering; Michael McGehee, Assistant Professor,
   Department of Materials Science and Engineering; Jeffrey King, Post-Doctoral
   Researcher; David Jackrel, Post-Doctoral Researcher; Evan Pickett, Graduate Researcher

       This project is a fundamental study into the development of low-cost, thin film solar
   cells. It explores the fabrication of semiconductor nanocomposites for photovoltaics
   using nanostructured inorganic materials and atomic layer deposition (ALD). The focus
   is on cells built by high-throughput techniques where nanoporous structures, ultrathin
   layers, and multiple junctions are used to achieve good energy conversion efficiencies at
   low cost.

       There is a strong need for the development of photovoltaic cells with low cost, high
   efficiency, and good stability. In thin film technologies, there exists a common problem
   with conversion efficiency due to poor materials quality; the photogenerated electrons
   and holes cannot travel very far before recombination (short free-carrier diffusion lengths)
   and are hence lost for power conversion. If the solar cell can be made using nanoscale
   heterojunctions, then every photogenerated carrier will have less distance to travel, and
   the problem of recombination can be greatly reduced. ALD is particularly well suited for
   this application since it can allow for highly uniform deposition on complex non-planar
   nanostructures with controllable thickness. With nanoscale diffusion lengths for the
   photogenerated carriers, the materials constraints can be relaxed, and low cost deposition
   routes become acceptable.

       A basic proposed structure for a nanostructured single-junction solar cell is illustrated
   in Figure 1, and consists of a nanostructured substrate that is coated with semiconducting
   layers through ALD and possibly other wet chemistry deposition techniques, such as
   chemical bath deposition or spray pyrolysis. Another related design involves first
   depositing the semiconductor absorber material in a planar geometry, then etching a
   periodic array of nanopores, and finally using ALD or other wet techniques to fill in the
   nanopores with an opposite polarity window layer to form the p/n junction. The absorber
   material could be composed of silicon, GaAs, CdTe, CIGS, or virtually any other solar
   cell material, since this layer can be deposited on a planar substrate and subsequently

GCEP Technical Report 2006                                                                         1
Inorganic Nanocomposite Solar Cells by ALD                                                Bent, et al.

            Figure 1: Schematic illustration of a single-junction nanostructured
           interpenetrated p/n junction solar cell, with an anodic alumina substrate.

       There are three major issues which must be addressed in the proposed solar cells.
   First is the development of the nanostructured substrates (or absorbers) with variable pore
   size and morphology. Second is the issue of deposition of the other material forming the
   p/n junction and the subsequent growth of additional layers into the nanostructured
   substrate, such as GaAs and other III-V materials, or CdTe, CdSe, and other II-VI
   materials. The third challenge is the electrical connection and current collection from all
   of the nanostructured p/n junctions.

       The nanostructured substrates (or absorbers) could potentially be fabricated by three
   techniques: nanosphere lithography, laser interference (or holographic) lithography, or
   anodization of a metal such as aluminum.[1-3] The primary reason for nanostructuring is
   to create a device that is optically thick to efficiently absorb the incident sunlight, and to
   simultaneously have every photogenerated electron-hole pair be very close to the p/n
   junction with respect to the minority carrier diffusion length, to minimize recombination
   losses. The minority carrier diffusion length in high quality silicon can be as long as 1
   mm, and only a few hundred microns of material are needed to absorb over 95% of the
   incident light. As a result, high quality silicon devices have achieved efficiencies near
   their theoretical limit. With poor quality (i.e. cheaply deposited) inorganic
   semiconductors, the diffusion length can be on the order of only 100 nm, while the
   absorption length is a micron at best. This leads to poor efficiencies in planar solar cells
   made from these materials, since only a small fraction of the sunlight is absorbed within a
   minority carrier diffusion length of the p/n junction, and many of the photogenerated
   minority carriers recombine before reaching the junction. Nanostructuring addresses this
   problem, provided the pore radius of the nanostructures is on the order of the minority
   carrier diffusion length, or about 100 nm. Nanosphere lithography is well-established for
   the 100 – 500 nm length scales, and interference lithography for length scales above
   about 200 nm. Some preliminary studies using 100 nm and 200 nm carboxylated poly-
   styrene nanospheres to pattern crystalline and amorphous silicon substrates are underway.
   The procedure involves spin-coating the nanospheres, followed by a reactive-ion etch
   (RIE) to reduce the sphere size, as illustrated in Figure 2. The spheres used for this
   preliminary study have poor size dispersity, which will be addressed by switching

GCEP Technical Report 2006                                                                          2
Inorganic Nanocomposite Solar Cells by ALD                                              Bent, et al.

   vendors for future studies. The next step involves depositing a few nanometers of Cr,
   and subsequently dissolving the spheres in a lift-off process, which leaves a film of metal
   protecting the silicon, with holes in the metal film where spheres previously were. Then
   the silicon can be reactive-ion etched to form pores with aspect ratios greater than 10:1.
   Figure 3 shows a nanostructured silicon substrate formed by this method.

           Figure 2: Carboxylated polystyrene nanospheres on a silicon substrate,
           after 90 sec RIE etch in O2 plasma (originally ~100 nm in diameter).
           Note the large polydispersity in the sphere sizes.

               Figure 3: Nanostructured p-type monocrystalline silicon substrate.

GCEP Technical Report 2006                                                                        3
Inorganic Nanocomposite Solar Cells by ALD                                              Bent, et al.

   Nanostructured Solar Cell Materials
        The second major challenge is the deposition of the other material forming the p/n
   junction into the nanostructured substrate, and the larger concern of what solar cell
   materials are appropriate to nanostructure in general. A major thrust of the project work
   thus far has been focused on theoretical modeling in order to determine materials design
   parameters. We have performed initial modeling studies on single and multijunction
   device geometries, buffer layers, pore dimensions, and materials electrical and optical
   properties, such as bandgap, free-carrier mobility, free-carrier lifetime, doping density,
   and dielectric constant. Thus far the electrical properties have been modeled in a 1-D
   geometry (using the modeling program AMPS), or evaluated using models from the
   literature, but work is beginning on a 2-D and 3-D modeling program (DESSIS) in order
   to more fully investigate the effects of different nanostructure geometries and the effects
   of materials parameters on nanostructured devices.[4-6] Buffer layers that could be used
   to reduce the recombination at the p/n junction were 1-D modeled to evaluate different
   buffer layer/device materials combinations.

        Since the project is focused on novel device architectures, it was determined that it
   would simplify the development if well-established solar cell semiconductors could be
   used. Furthermore, the device performance of a candidate material should be limited by
   the minority carrier diffusion length in planar structures, since this is the problem that
   nanostructuring addresses. There are many inorganic solar cell materials that have short
   diffusion lengths when deposited cheaply; cheap deposition often leads to small grain
   poly-crystalline material, or large impurity concentrations (point defects). These short
   diffusion lengths, however, are often the result of recombination through traps which
   leads to short minority-carrier lifetimes. Unfortunately, the negative effects of short
   minority-carrier lifetimes, namely high dark current, will be greatly exacerbated by the
   large increase in junction area that occurs with nanostructuring. As a result, it appears
   that materials with short diffusion lengths caused primarily by short minority-carrier
   lifetimes may not be good candidates for nanostructuring. Instead, what would be
   desired is a material with a short diffusion length and a relatively long minority-carrier
   lifetime. This can only occur in materials with fairly low free-carrier mobilities, on the
   order of 1~10 cm2/V-s, such as hydrogenated amorphous silicon (a-Si:H). Kayes, et al.
   pointed out in 2005 that, somewhat counter intuitively, a material that has a high free-
   carrier mobility, such as crystalline silicon or GaAs, would have excessive leakage
   current in a nanostructured interpenetrated p/n junction geometry; this would cause the
   open-circuit voltage, and thus the efficiency, to drop almost to zero.[6] From the
   combined results of all of the modeling, we have chosen hydrogenated amorphous silicon
   to be our first test material. We will have a-Si:H films deposited on planar substrates,
   and then nanostructure them, and finally deposit a conformal layer of opposite polarity to
   form the interpenetrated p/n junction.

       The opposite polarity material will be deposited by both ALD and by chemical bath
   deposition. An ALD reactor is being built to deposit II-VI materials. CdTe and CdSe are
   common absorber layers used in thin film solar cells, and CdS is a commonly used
   heterojunction material for thin film solar cells. Furthermore, zinc compounds (ZnSe,
   ZnTe and ZnS) are possible alternative window materials (Eg > 2.0 eV), and lead

GCEP Technical Report 2006                                                                        4
Inorganic Nanocomposite Solar Cells by ALD                                              Bent, et al.

   compounds (PbSe, PbTe, and PbS) are excellent narrow bandgap materials (Eg < 0.5 eV)
   useful in third-generation multijunction and ‘multiple electrons per photon’ solar cells.
   Thus, II-VI-like compounds of Cd, Zn and Pb, with Se, Te and S, cover the whole solar
   spectrum, and, being compound in nature, are readily depositable by ALD.

        Chemical bath deposition (CBD) is a solution-based deposition technique used for
   rapid, economical growth of thin films on a variety of substrates. The method is
   applicable for many different materials, including oxides and sulfides.[7] Specifically,
   CBD is a well-established technique for growth of CdS films in solar cell applications,
   and our first test structures will therefore be fabricated using this method. For CBD
   growth of CdS, a substrate is typically immersed in a hot (60 – 100 °C), well-stirred
   aqueous solution containing cadmium salt, a sulfur source (e.g. thiourea), an ammonia
   salt, and ammonia hydroxide. For our preliminary experiments, the deposition bath was
   maintained at 85 °C and contained cadmium sulfate (CdSO4), ammonium sulfate
   (NH3)2SO4, thiourea ((NH2)2CS), ammonium hydroxide (NH4OH), and deionized water.
   Conditions were optimized to ensure that film growth occurs heterogeneously on the
   substrate and beaker walls, instead of homogeneously as colloidal particles in the bath.[8]
   For this to occur, NH3 must be present in sufficient amounts to bind Cd in a Cd
   tetraamine complex. The use of ammonium sulfate helps stabilize this complex as well
   as control the pH of the bath. An SEM image of a planar CdS film grown via CBD is
   shown in Figure 4 below. Test structures of monocrystalline p-type silicon were
   nanostructured via nanosphere lithography, and a ~ 100 nm thick n-type CdS film was
   then deposited on the structure, as shown in Figures 5 and 6 below.

           Figure 4. Cross sectional view of CdS film on planar silicon substrate

GCEP Technical Report 2006                                                                        5
Inorganic Nanocomposite Solar Cells by ALD                                              Bent, et al.

          Figure 5. Cross sectional SEM image of CdS/Si nanostructured p-n junction.

           Figure 6. Cross sectional SEM image of CdS/Si nanostructured p-n junction at
           higher magnification.

   Electrical Connection and Current Collection
       The third challenge is the electrical connection and current collection from all of the
   nanostructured p/n junctions. The electrical connection will be made much simpler if the
   device structures can be planarized, either during or after growth. However, if the

GCEP Technical Report 2006                                                                        6
Inorganic Nanocomposite Solar Cells by ALD                                               Bent, et al.

   surface roughness is too great and standard metal evaporation is not sufficient to create a
   good electrical contact then alternative contact deposition techniques, such as screen-
   printing, will be investigated.

       One of the biggest fundamental electrical challenges facing interpenetrated p/n
   junction nanostructured solar cells is the recombination at the p/n junction interface.
   Recombination scales linearly with the junction area, and in these cells the junction area
   can be orders of magnitude larger than a planar design with the same solar cross-section.
   There has been some work done depositing thin buffer layers (~10 nm) in between the p-
   and n-type materials in nanostructured cells to suppress interfacial recombination.[9, 10]
   1-D modeling was performed to evaluate the effects of different buffer layer/device
   materials combinations. It has been suggested that a buffer layer with appropriate
   conduction and valence band energies could reduce interfacial recombination by reducing
   free-carrier concentrations at the interface between the n- and p-type materials. This will
   increase the open-circuit voltage, but modeling shows it comes at the cost of the device
   current; the lower carrier concentrations near the junction interface also reduce the
   electric field strength which is needed to collect the photogenerated carriers. We have
   therefore determined that much of the benefit of buffer layers observed in nanostructured
   interpenetrated p/n junction solar cells is likely the result of improved interface chemistry
   (passivation) at the junction, rather than an altering of electron and hole concentrations
   through band-level engineering.

       The versatility and film thickness control afforded by ALD will be invaluable to
   experimentally investigate the effects of different buffer layers of various thicknesses in
   our cells (1~10 nm). We plan to investigate various materials combinations involving
   a-Si:H as the absorber, such as n-CdS/p-CdSe/p-a-Si:H. The p-type CdSe buffer layer
   moves the p/n junction interface to between the CdS and the CdSe, two very similar
   materials, and away from the CdS/a-Si:H interface, which will likely suffer from poorer
   interface chemistry and more recombination centers.

       The current world record power conversion efficiency for single-junction
   hydrogenated amorphous silicon solar cells is just below 10%.[11] The world record for
   a multijunction hydrogenated amorphous silicon solar cell is below 13%.[11]
   Hydrogenated amorphous silicon has a bandgap of about 1.7 eV and the theoretical limit
   of efficiency for a solar cell with a 1.7 eV bandgap is above 25%. Some of the loss in
   efficiency in state of the art a-Si:H cells is due to low open-circuit voltages and low fill-
   factors, neither of which could be improved by nanostructuring. However, our modeling
   has shown that if nanostructuring enabled the cells to achieve 100% quantum efficiency,
   then the current could be increased from 16 mA/cm2 up to 21.5 mA/cm2. Device
   modeling indicates that the increase in current would significantly increase the open-
   circuit voltage as well, and such a device could achieve a power conversion efficiency as
   high as 15%. These devices would be single-junction cells, manufactured largely without
   vacuum processing. We are therefore moving towards the design and fabrication of a
   nanostructured solar cell that could be made using low-cost production techniques at
   efficiencies matching current thin-film devices.

GCEP Technical Report 2006                                                                         7
Inorganic Nanocomposite Solar Cells by ALD                                                   Bent, et al.

   Future Plans
       Nanostructured interpenetrated p/n junction solar cells will be investigated. The
   nanostructured substrates (or absorbers) could potentially be fabricated by three
   techniques: nanosphere lithography, laser interference (or holographic) lithography, or
   anodization of a metal such as aluminum. From the results of our modeling, we have
   chosen hydrogenated amorphous silicon to be our first test material. The opposite
   polarity material will be deposited by both ALD and by chemical bath deposition. An
   ALD reactor is being built to deposit II-VI materials. ALD will be invaluable to
   investigate the effects of different buffer layers in our cells. Our initial test structures will
   be fabricated using a p-type a-Si:H absorber, a thin (~10 nm) p-type II-VI buffer layer
   deposited by ALD, and a chemical bath deposited n-type CdS window. Future device
   generations will potentially incorporate alternate absorber materials, such as thin films of
   CIGS and CdTe, fabricated by standard techniques as well as more inexpensive, lower
   quality techniques, such as chemical bath deposition. The free-carrier mobilites in many
   of the thin film materials are somewhat lower than bulk crystalline Si or GaAs and thus
   could be good candidates for nanostucturing as well. Finally, sub-cells with different
   bandgap absorber layers will be integrated monolithically to form a multijunction cell.

        The original proposal described a multijunction nanostructured cell comprised of
   conformally deposited semiconducting layers. Figure 7 illustrates two additional
   possibilities for multijunction cell structures. The wide bandgap material needs to filter
   the light before reaching the narrow bandgap material in order to reap all of the benefits
   of the multijunction design. In Figure 7a two separately nanostructured junctions are
   formed. The tunnel junction and the ‘transparent material’ in the figure, forming the
   nanostructured substrate for the upper wide bandgap junction, would be deposited onto a
   finished narrow bandgap junction. In both sub-cells the light gray nanostructured pillars
   could be n-type semiconductors, or insulators (such as anodic alumina). For the structure
   in Figure 7b, only one nanostructuring process is needed. The narrow bandgap junction
   can be fabricated similarly to the single junction devices (growth #1 in the figure), but
   then a selective wet etch process would be performed to selectively remove half of each
   of the narrow bandgap layers, and leave the nanostructured scaffold intact. In a
   commercial process the etched materials may be recovered and reused. Finally, the
   tunnel junction and the wide bandgap layers would be grown in a re-growth step (growth
   #2 in the figure). Note the narrow bandgap material needs to be the first material grown
   in the re-growth, in order to form a single polarity layer (p-type in this example) on
   which to grow the tunnel junction and subsequent device layers. This creates some areas
   of the cell where the incoming light is not effectively filtered by the wide bandgap top
   sub-cell. Other possible disadvantages of the second design (Figure 7b) are that the tunnel
   junction is non-planar, and the process requires the complexity of added etching steps.
   On the other hand, it may be difficult in practice to nanostructure on top of a
   nanostructure, as in Figure 7a, and thus the structures in Figure 3b could be easier to
   fabricate with good uniformity.

GCEP Technical Report 2006                                                                             8
Inorganic Nanocomposite Solar Cells by ALD                                              Bent, et al.

         Figure 7: Two multijunction nanostructured interpenetrated p/n junction
         solar cell designs; a.) two independently nanostructured junctions, b.) an
         integrated nanostructured design, in which only one nanostructuring process
         is needed. The second junction is formed by a selective etch that leaves the
         nanostructured scaffolding, but removes two conformally grown p- and n-
         type materials (growth #1), followed by a re-growth (growth #2)that forms
         the upper junction.

       The materials properties and electrical device properties of each round of test
   structures will be characterized and modeled, and improvements in the materials and
   geometry will be incorporated into subsequent test structures. Future modeling studies
   will investigate nanostructure geometries and materials properties using the 2-D software
   DESSIS. Particularly, the effect of minority-carrier mobilities and lifetimes on the
   performance of solar cells with different nanostructures and interface properties will be
   investigated. Materials characterization will be performed using techniques such as SEM
   and TEM to determine the degree of pore-filling and film uniformity, and XRD to
   determine crystal structure and crystallinity. Scanning probe microscopy will be used to
   determine grain boundary and junction properties such as surface photovoltage at the
   nanometer length scale. XPS studies could be useful to determine the composition and
   bonding character of the surface species present before and after etching processes and
   buffer layer deposition. Optical absorption and photoluminescence will be performed to
   determine absorption coefficients and bandgaps. DLTS, deep-level transient
   spectroscopy could also be utilized to measure the non-radiative recombination centers
   (performed at Accent Optical, San Jose, CA), and spectral CL mapping could be used to
   determine uniformity and threading dislocation density (performed at NREL, Golden,
   CO). The device characterization will include photocurrent-voltage under simulated
   AM1.5 solar spectrum, spectral quantum efficiency, and capacitance-voltage

GCEP Technical Report 2006                                                                        9
Inorganic Nanocomposite Solar Cells by ALD                                                     Bent, et al.


       1.    Hulteen, J. C., Van Duyne, R. P. J. Vac. Sci. Technol. A 13, 1553-1558 (1995)
       2.    Berger, V., Gauthier-Lafaye, O. & Costard, E. J. Appl. Phys. 82, 60-64 (1997)
       3.    Diggle, J. W. ,Downie, T. C. & Goulding, C. W. Chem. Rev. 69, 365 (1969)
       6.    Kayes, B.M. and Atwater, H.A., J. Appl. Phys 97, 114302 (2005)
       7..   Mane, R. S. and Lokhande, C. D., Mat. Chem. and Phys. 65, 1-31 (2000).
       8.    Oladeji, I. O. and Chow, L., J. Electrochem. Soc. 144, 2342-2346 (1997)
       9.    Nanu, M., Schoonman, J., and Goossens, A., Nano Lett. 5(9), 1716-1719 (2005)
       10.   Nanu, M., Schoonman, J., and Goossens, A., Adv. Func. Mat. 15(1), 95-100 (2005)
       11.   Green, M.A., et al., Prog. Photovolt: Res. Appl. 14, 45-51 (2006)

      Stacey Bent:
      James S. Harris:
      Michael McGehee:

GCEP Technical Report 2006                                                                              10