Complete Report - University of New South Wales

More documents

Recommendations

Info

ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE2010/11ANNUAL REPORTElectronic density of states (DOS) of Si 10QD in 2 MLSiO 2(dotted black), fully oxidised nanocrystal (grey)and Si 10QD in 2 ML SiO 2with Sc replacing Si atom inthe SiO 2shell at the outermost position (top). DOS ofSc-doped approximants split into α (spin up, red) andβ (spin down, blue) DOS because of non-even numberof electrons within the approximant due to acceptorstate of Sc. The b-highest occupied molecular orbital(HOMO) 1193 indicated in magenta can take up 0.07electrons from the Si 10QD over two 2 ML (6 Å) of perfectSiO 2. The β-lowest unoccupied MO (LUMO) 1194 shownin green can take up a much bigger proportion of anelectron if the Si QD is slightly bigger (e.g. Si 35, d QD=11 Å). The Sc-doped nanocrystals are shown with theiso-density plots of 1.35×10 −3 e/Å 3 for β-HOMO 1193(lower left) and β-LUMO 1194 (lower right). The atomsof the Si 10QD and the Sc atom are highlighted in cyan.Si atoms in SiO 2are grey, O atoms are red, H-atomsterminating the outermost O-bonds are white. All DFTcomputations carried out with a Hartree-Fock 3-21G(d)optimisation and B3LYP/6-31G(d) electronic structurecalculation [4.5.23].Figure 4.5.42Spatial distribution of 4s and 3delectronic states of Sc (upper) andenergetic arrangement of 4s and 3dstates (lower).Figure 4.5.41material and being collected in the external circuitas if from an n-type material, with electrons injectedback into the device through the B doped materialwhere they would recombine with the immobileholes, thus making the B material appear p-type.This qualitatively mirrors the behaviour observed,but will require corroboration. As it relies on thedifference in electron and hole mobilities, it isessentially the same as the Dember effect, whichis known to produce a photovoltaic effect throughcarrier separation, albeit not very efficiently.It is also possible that both the sub-oxide dopingeffect and the mismatched QD size effects operatein the device. The fact that both mechanisms actin the same direction is fortuitous as both couldtherefore contribute to the observed rectifyingphotovoltaic behaviour. Differentiation betweenthese or a determination of the relative strengthswill require further study on the effects of doping,interlayers and band alignments. These will alsofurther optimisation of the devices and lead tobetter photovoltaic performance from these Si QDnanostructure devices.4.5.2.5.2 Ab-initio modelling ofmodulation doping possibilities inSi QDs nanostructuresResearchers:Dirk König, James Rudd, Daniel Hiller (IMTEK,University of Freiburg, Germany)As discussed in Section 4.5.2.5, direct doping ofQDs by dopants is thermodynamically very unlikely.The alternative of modulation doping of the SiO 2matrix is also not feasible due to the strong anionicnature of O leading to very deep ionisation energies[4.5.48]. We have concentrated recently on possiblecandidates for more efficienct acceptor dopingof Si QDs. Modulation acceptors must have anenergetic position below the confined hole level,they must not introduce defect levels in the QDband gap if located at the interface as an activedopant and as a completely saturated foreignatom, and they must have one single oxidationnumber. The transition metal scandium (Sc) fulfilsthe last two requirements. It also has a very suitablevalence state configuration which consists of afull (doubly occupied) 4s shell and one electronin the 3d 1 state located within the 4s shell. On theenergy scale, the 4s electrons have a higher bindingenergy as compared to the 3d 1 and 3d 2 states. Thisprovides the 3d 1 and 3d 2 states with an electrostaticscreening which results in a higher binding energyof these states. Thereby an unoccupied 3d 2 stateexists at a high binding energy which is able to takeup an electron from a nearby source such as a Si QD,see Fig. 4.5.41.Ab-initio Density Functional Theory calculationson a Si 10QD in 2 mono layers (MLs) of SiO 2havebeen carried out, with a Si atom at the outermostposition with full O termination replaced by a Scatom. The Si 10atom cluster is the smallest size whichstill behaves as a Si QD [4.5.23]. Such a nanoparticleallows us to directly observe the electronic structureof Sc as a modulation acceptor in SiO 2in theproximity of a Si QD, see Fig. 4.5.42.Figure 4.5.42 shows that 0.07 electrons stemmingfrom the Si 10QD (d QD= 7.3Å) are localized atthe Sc atom over two 2 ML (6 Å) of perfect SiO 2.Technological relevant QD sizes start at d QD≈ 20Å,with a much lesser degree of quantum confinementas compared to a 7.3Å QD. It is thus reasonable toassume that a Si QD with d QD≥ 20Å will be positivelyionized to a much greater extent, ensuring thatholes are majority carriers in a Si QD super lattice(SL) embedded in SiO 2. Experimental verificationof this doping mechanism is being carried out. Thedifference to conventional acceptor modulationdoping as used in III-V electronic devices is theway the Si or Ge QDs are ionised by Sc in SiO 2. AsSc induces a deep acceptor level in SiO 2it cannotbe thermally ionized from the SiO 2valence band.Instead, the Sc acceptor state takes up the electrondirectly from the initially occupied state at a lowerbinding energy presented by a nearby QD, see Fig.4.5.43. This requires the Sc acceptor to be within50Å of the Si QD in order to exploit field emission80
ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE2010/11ANNUAL REPORTAcceptor modulation doping:Conventional example with GaSQD in AlAs matrix and electronthermally emitted into acceptor(Mg), such that holes are capturedby the QD (left). In SiO 2, chargetransfer occurs directly as theSc state is too deep within theband gap to be thermally ionized.Instead, the acceptor state isionised directly from the QD byfield emission of an electron intothe Sc acceptor state.Figure 4.5.43which is a consequence of the energy differencebetween the QD HOMO and the Sc acceptor state.Acceptor modulation doping of SiO 2can beaccomplished by incorporation of a small amountof Sc, say one monolayer of Sc, into the SiO 2barriersbetween the SRO QD array layers. During thesegregation anneal, Sc is incorporated into theSiO 2matrix and thereby activated as an acceptor.Ab-initio calculation indicates that Sc attachedto the Si QD does not create any levels withinthe HOMO-LUMO gap of a Si 84(OH) 64QD. We cantherefore assume that a Sc atoms bonded onto aSi QD does not result in a detrimental change inits electronic structure. This concept is now beinginvestigated experimentally.4.5.2.6 Summary of Group IVnanostructures for tandem cell elementsIn 2010, significant progress has been made onunderstanding the transfer of growth parametersof Si QDs to other matrices. The use of anisotropicstructures both to provide diffusion barriersto excessive QD growth and to provide higherconductivity paths in the transport direction, hasshown the validity of these approaches both forSi 3N 4and for SiC matrices. Knowledge of Ge QDnanostructures has improved significantly withclear p-type behaviour of Ge QD s in SiO 2and verysuccessful growth of Ge QWs in Si 3N 4matrix. Theincrease in sophistication of modelling of confinedenergy in these structures now matches reasonablywell with experimental data.Improved modelling of confined levels inirregular shaped particles also matches wellwith experimental data of other groups. Higherlevel device modelling is now able to describethe anomalous effects due to high lateral seriesresistance in devices and consequent currentcrowding. This is now being used to optimisenanostructure growth and device design, so as topush beyond the current open circuit voltages of490mV. Part of this is an understanding of doping.This has progressed with greater knowledge andcontrol over the effects on QSD size of introductionof P and B dopants, and modelling of possibledoping mechanisms based on modified modulationdoping and these size effects. Alternativemodulation dopants are being modelled withscandium as an acceptor looking like a strongcandidate. Further work will build on this increasedunderstanding of anisoptropic structures, growth ofdifferent materials and understanding of doping toimprove materials and devices further.4.5.3 Hot Carrier cellsResearchers:Shujuan Huang, Santosh Shreshtha, DirkKönig, Robert Patterson, Pasquale Aliberti,Binesh Puthen Veettil, Lara Treiber, IvanPerez-Wurfl, Andy Hsieh, Yu Feng, JamesRudd, Stephan Michard, Martin Green,Gavin ConibeerHot carrier solar cells offer the possibility of veryhigh efficiencies (limiting efficiency above 65% forunconcentrated illumination) but with a structurethat could be conceptually simple compared toother very high efficiency PV devices – such asmulti-junction monolithic tandem cells. For thisreason, the approach lends itself to ‘thin film’deposition techniques, with their attendant lowcosts in materials and energy usage and facility touse abundant, non-toxic elements.An ideal Hot Carrier cell would absorb a wide rangeof photon energies and extract a large fractionof the energy to give very high efficiencies byextracting ‘hot’ carriers before they thermalise tothe band edges. Hence an important property of ahot carrier cell is to slow the rate of carrier coolingto allow hot carriers to be collected whilst they arestill at elevated energies (“hot”), and thus allowinghigher voltages to be achieved from the cell andhence higher efficiency. A Hot Carrier cell must alsoonly allow extraction of carriers from the devicethrough contacts which accept only a very narrowrange of energies (energy selective contacts or81
Page 4:
ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE
Page 8 and 9:
Page 10 and 11:
Page 12 and 13:
Page 14:
Page 17 and 18:
Efficiency and cost projections for
Page 19 and 20:
Page 21 and 22:
Page 23 and 24:
4.3 First GenerationWafer Based Pro
Page 25 and 26:
Jsc improvement afterapplying diele
Page 27 and 28:
Page 30 and 31:
Page 34 and 35: ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE
Page 36: ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE
Page 40 and 41: 4.4 Second Generation:Silicon, Orga
Page 47 and 48: I-V curves of the individual cells
Page 49 and 50: Voc [mV]2 m thick film480BSF deposi
Page 57 and 58: (a) The schematic band energystruct
Page 59 and 60: light intensity (Suns)100.0010.001.
Page 63 and 64: 4.5 Third Generation Strand -Advanc
Page 128 and 129: Z. Ouyang, X. Zhao, S. Varlamov, Y.
Page 130 and 131: J. Huang, J. Wong, M. Keevers and M
Page 132:
show all

Complete Report - University of New South Wales

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?