COfiF-96OLtOl--HIGH PERFORMANCE TRANSPARENT CONDUCTING FILMS OF CADMIUM INDATE

Transcription

1 KAPL-P-44 (K9639) COfiF-96OLtOl--HIGH PERFORMANCE TRANSPARENT CONDUCTING FILMS OF CADMIUM INDATE RF Sputtering, TJ Coutts, WP Mulligan April 996 NOTICE This report was prepared as an aountof work sponsoredby the United States Government. Neither the United States,nor the United StatesDepartmentof Energy, nor any of their employees, nor any of their ontrators, subontrators,or their employees,makes any warranty, express or implied, or assumesany legal liabilityor responsibility for the auray, ompletenessor usefulness of any information, apparatus, produt or proess dislosed, or represents that its use would not M h g e privately owned rights. KAPL ATOMIC POWER LABORATORY SCHENECTADY, NEW YORK 3 Operated for the U. S. Department of Energy by KAPL, In. a Lokheed Martin ompany

2 DISCLAIMER This report was prepared as an aount of work sponsored by an ageny of the United States Government. Neither the United States Government nor any ageny thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liabiiity or responsibility for the auray, ompleteness, or usefulness of any information, apparatus, produt, or proess dislosed, or represents that its use would not infringe privately owned rights. Referene herein to any speifi ommerial produt, proess, or servie by trade name, trademark, manufaturer, or otherwise does not neessarily onstitute or imply its endorsement, mommendation. or favoring by the United States Government or any ageny thereof. The views and opinions of authors expnssed herein do not neessarily state or reflet thos of the United States Government or any ageny thereof.

3 DISC LAlM ER Portions of this doument may be illegible in eletroni image produts. Images are produed from the best available original doument.

4 , z HIGH PERFORMANCE TRANSPARENT CONDUCTING FILMS OF CADMIUM INDATE PREPARED BY RF SPUTTERING T. J. COUTTS, X. WU, and W. P. MULLIGAN National Renewable Energy Laboratory, Golden, CO 84 ABSTRACT We are examining various spinel-strutured thin films ( e g, CdSn4, ZnSn4) to develop higher-quality transparent onduting oxides (TCO) than more onventional materials suh as indium tin oxide. Here, we report on admium indate (CdIn4,CIO), whih is another member of this family. Thin films of CIO were deposited by radio-frequeny (RF)magnetron sputtering, from an oxide target, onto borosiliate glass substrates. The variables inluded the substrate temperature, sputtering gas omposition, and pressure. Film properties were measured before and after heat treatment. Charaterization involved Hall effet measurements, optial and infrared spetrophotometry, X-ray diffration, and atomi-fore mirosopy. - ~m were ahieved for a film thikness of.55 pm. The Film resistivities as low as. 3 ~ C transmittane was 9% in the visible region of the spetrum, without orretion for substrate losses and without an anti-refletion oating. The plasma resonane ourred at longer wavelengths than for other materials and this, with a bandgap of approximately 3. ev, presents a wide window for optial transmittane. The highest mobility was 54 m V-I s--l and the highest arrier onentration was 7.5~m-3. INTRODUCTION Transparent onduting oxides (TCOs) are an essential omponent of several important appliations. These inlude photovoltai panels, flat-panel displays, smart windows based on eletrohromi phenomena, and heat-refleting windows. In addition, there are smaller appliations, inluding the oating of light-emitting diodes, the properties of whih are greatly improved by the use of a TCO oating. In eah of these ases, there is a trend towards larger area devies, that is beginning to plae exessive demands on the traditional TCOs suh as tin oxide, indium tin oxide, and zin oxide. As is very well known, the essential properties required of TCOs are high eletrial ondutane and a high optial transmittane. It an readily be demonstrated from elementary eletromagneti theory [I] that these are somewhat exlusive requirements, beause the free eletrons needed for a high ondutane also lead to optial absorbane. However, traditional TCOs have demonstrated adequate properties until the reent introdution of larger-area devies (display panels are expeted to be more than 8 square feet in area, and photovoltai panels of this area are already being manufatured). Larger areas ause exessive parasiti losses, unless the series resistane introdued by the TCO an be redued below the values typially ahieved with traditional TCOs. The minimum resistivity ahieved in prodution for the traditional TCOs is about 5x-4 R m, and the sheet resistane is typially - R per square. To redue the sheet resistane below this range, it is vital to redue the resistivity: inreasing the thikness is not viable beause of problems of exessive optial absorption and the possibility of mehanial problems during proessing. There appears to be little prospet of ahieving lower resistivities with the materials mentioned above. Many papers have reported efforts to do so over the last forty years; however, no signifiant fundamental progress has been made towards reduing resistivity and absorbane. The essential requirement for progress is the development of novel materials with wellunderstood properties. At the National Renewable Energy Laboratory (NREL), we have been performing both fundamental and applied researh into ternary oxide strutures, mainly based on the spinel struture. Among these, we have already reported on admium stannate (CdSn4) [, and we are presently developing other materials in the family. Even though the prerequisites of an industrially aeptable TCO inlude non-toxiity and plentiful elements, we are also interested in gaining a fundamental understanding of the new family of materials, in general, whih is why

5 I 'r CdIn,O, is inluded in our study. in addition, there are other appliations are less ost-sensitive, and the same demanding riteria do not apply []. Cadmium indate, whih onflits with both of the above requirements, has been reported as having promising properties [3, 4. Despite its toxiity and In ontent, it may have signifiant value in ertain solar ells, that already ontain a signifiant amount of Cd beause of its exellent transmittane in the visible portion of the spetrum. In any event, it is also likely to reveal additional fundamental information, whih may lead to a better understanding of the family of materials as a whole. In turn, this ould lead to the desired pratial result: namely, an improved, industrially aeptable, TCO. SUMMARY OF MODELLING RESULTS We have previously disussed the modelling of TCO films in detail [, and only a summary of the approah and key results will be given here. The optial properties of materials an, to a first approximation, be desribed by the Lorentz osillator model. This is a seond-order differential equation in whih there is a damping term (giving rise to absorption by the free eletrons) and a restoring fore (giving rise to a natural frequeny of osillation). The equation also ontains the onentration of free eletrons. Usually, one assumes that the restoring fore is too small to be relevant, and a solution of the differential equation is obtained that involves only the damping term. The damping term is simply related to the mobility (stritly, the a mobility) of the eletrons. The solution gives the displaement of the eletron as a funtion of time, whih leads to its veloity and a ondutivity. Maxwell's equation, relates the omplex permittivity to the a ondutivity, and hene, to the eletron onentration and mobility. The optial onstants (the refrative index and the extintion oeffiient) an then be obained from the omplex permittivity. The refletane and transmittane an then be alulated for a f i m of arbitrary thikness on an a substrate with known optial onstants, using Fresnel's equations. Hene, eletrially measurable quantities (the eletron onentration and mobility) may be related to optially measurable quantities (the refletane and transmittane). All the modelling data assume an effetive mass of. times the free-eletron mass, a high-frequeny permittivity of 8, and a film thikness of.5 pm. Figure shows the shift of the plasma edge to shorter wavelengths as the arrier onentration is inreased. The mobility is assumed to remain onstant at mv's-'. This is ertainly not a reasonable expetation for the higher arrier onentrations, but it has been obtained for admium stannate at lower onentrations [SI. Note that the urves also steepen as the onentration is inreased. This is beause the parameter atually ontrolling the optial properties is the ondutivity, rather than either the onentration or the mobility alone. Figure shows the free-arrier absorbane for the film desribed in Figure. The absorbane is obtained from onservation of energy (i.e., A = - R - T), and it is very large for higher arrier onentrations, even though the mobility is maintained at mv's-'. The height of the absorbane peak inreases with arrier onentration simply beause there are more osillators to absorb energy at the resonane ondition. The absorbane peak narrows with inreasing arrier onentration beause the ondutivity inreases, and this fator ontrols the seletivity of the filter and the long-wave refletane. In Figure 3, the refletane is modelled with the mobility being treated parametrially and the arrier onentration being kept onstant at lom~ m - Although ~. the urves steepen with inreasing mobility, and the long-wave refletane inreases signifiantly, the plasma wavelength is not influened muh. The absorbane is shown modelled in Figure 4, again with the arrier onentration being kept ~, the mobility being treated parametrially. In both figures 3 and 4, the onstant at lo' ~ m - and upper limits of the mobility are unrealistially large. Suh a high eletron onentration would inevitably prelude mobilities of this magnitude. Nevertheless, the model makes the interesting point that the peak height and width both derease with inreasing mobility. If it is neessary to adjust the position of the plasma edge, then this must be done through the arrier onentration; the mobility having only a seond order effet on the plasma wavelength.

6 .9 I I d a, e 5:.4 II.... ==.5 a Wavelength (pm) 8 9 Figure : Variation of the refletane of a film of.5 pm thikness, on a glass substrate, with wavelength. The mobility was taken as m V-' s-', and the arrier onentration was treated parametrially Wavelength (pm) 8 9 Figure : Variation of the absorbane of a film of.5 pm thikness, on a glass substrate, with wavelength. The mobility was taken as m V-' s-', and the arrier onentration was treated parametrially a OC..6.5! Wavelength (pm) Figure 3: Variation of the refletane of a film of.5 pm thikness, on a glass substrate, with wavelength. The arrier onentration was taken as IO' ~ m -and ~, the mobility was treated parametrially Wavelength (pm) 9 Figure 4: Variation of the absorbane of a film of.5 pm thikness, on a glass substrate, with wavelength. The arrier onentration was taken as IO' ~ m - and ~, the mobility was treated parametrially.

7 The essential features to emerge from this treatment are: I) The absorbane dereases with mobility but inreases with arrier onentration; ii) There is a transition from a high transmittane to a high refletane in the near4.r. region, at a wavelength known as the plasma wavelength (usually about - pm for a TCO); iii) The steepness of the transition (important for some appliations) inreases with the mobility; this also leads to dereasing free-arrier absorbane in this region; and iv) The long-wave refletane inreases with both mobility and arrier onentration These results are summarized in Figures -4, whih are part of a modelling study originally published in referene. Many assumptions are involved in this elementary modelling, amongst whih is that the relaxation time of the eletrons under the very-high frequeny eletri field of the eletromagneti radiation is the same as that inferred from eletrial measurement of the mobility. It is by no means apparent that this is realisti. We have also assumed that a single-osillator model is adequate to desribe the infra-red optial properties of the materials. More ompliated models have been used [6]to aount for the optial properties in the visible, but we have restrited our attention to the region in the immediate viinity of and beyond the plasma wavelength. We see from this summary that a high eletron mobility is essential for optimum films, and it is this quality that attrated us originally to the spinel family of materials. The evidene seems to be that these have higher mobilities than onventional TCOs [ 5 ], whih has been onfmed by a number of workers for admium stannate [ and, to a lesser extent, for admium indate [3]. Possible reasons for this inlude a smaller effetive mass, a longer relaxation time, or a ombination of these. We are interested in determining whih of these is more important, and then in using the knowledge to help design new, improved TCOs. The relaxation time may be ontrollable through the deposition parameters, while the effetive mass may only be ontrolled though a areful hoie of materials. Therefore, the researh is aimed at both of these: optimization of the deposition parameters, and working with a variety of materials that appear to have the potential of having lower effetive mass. We are also interested in determining the speies responsible for the donor ation, sine it is almost ertain (for these materials) that the usually-postulated oxygen vaanies annot be the ause. This will be explained in further detail later. EXPERIMENTAL The films were deposited by RF sputtering in a CVC SC3 system, and the targets onsisted of equal molar frations of CdO and InzO3. The purity of the powders was 99.5%. The powder mix was pre-reated and the reated material was hot-pressed into " diameter sputtering targets, supplied by Cera In. The separation between the substrate and the target was maiptained at 6 m, and the r.f. power was watts. This typially resulted in a deposition rate of A s-. We note that this rate is muh lower than that for CdSn4,using the same deposition onditions. The films were deposited in pure argon, pure oxygen, or argodoxygen mixtures, and the omposition of the mixture was ontrolled using two mass flow ontrollers, with an auray of k. sm. The pressure of the sputtering gas was mtorr M. mtorr. A range of deposition temperatures was used, and annealing (of some samples) was performed in pure argon up to a temperature of 68 C. The substrates were Corning glass 759 or polished silion wafers, and were leaned using standard proedures. The films were between 3-6 A in thikness and they were haraterized by Hall effet measurements, ultravioletvisible spetrophotometry up to pm (using a Cary 3), Fourier transform infrared spetrophotometry at longer wavelengths, X-ray diffration using a Rigaku mahine with Cu-Ka radiation, and atomi-fore mirosopy. We note that, in our work on admium stannate, we were also able to shed some light on the effet of the deposition and annealing parameters using Mossbauer spetrosopy [. This is not appliable to admium indate beause indium does not have a signifiant Mossbauer signature.

8 RESULTS Figure 5 shows the variation of the arrier onentration and mobility as a funtion of the gas omposition in the sputtering system. The film was deposited at a substrate temperature of OC, and no annealing was used. The arrier onentration (and ondutivity) peaked at an oxygen:argon flow-rate ratio of.3%, i.e., 3 parts per thousand. Note that the vauum system had been pumped down to a base pressure of about lo-' Torr, so it is reasonable to laim ontrol to this level of auray. The arrier onentration is quite signifiantly influened by the gas omposition although this is not true of the mobility, whih hanges minimally. The arrier onentration passes through a maximum and the resistivity through a minimum. Hene, we should expet to see this refleted in the optial properties. Figure 6 shows the optial properties of three films deposited as desribed for Figure 5, but for three different xygen:argon ratios. This demonstrates that the position of the plasma edge ould be adjusted by hangiig this ratio: an effet that is simply related to the arrier onentration, as desribed in the modelling results above. For 4 ratios of.,.3, and.47%, the plasma 4 wavelengths were.,.86, and.97 pm, reu J spetively. The arrier onentration peaks at 5 4.3%, and this orresponds to the shortest v plasma wavelength Figure 7 shows the X-ray diffration spetra of 38 films deposited at various substrate temperatures a from room temperature up to 4 C. The atmos 37 phere was a.3% mixture of oxygen:argon, established as giving the highest arrier onentra36 tion. The films were not annealed after deposi tion. The key feature is that the only evidene 7 v Ratio of xygen:argon flow-rates (%).9 L Figure 5: Variation of the mobility and arrier onentration for CIO films deposited at different oxygen: argon flow-rate ratios. of any rystallinity was the () peak of In,O,, although amorphous material was presumably also present. The maximum mobility for these films was 45 m VI ssl and the minimum resistivity was 4. 5 ~ i-~m, both ouring for a substrate temperature of C. Although the resistivity is quite low, signifiantly superior results have been obtained on more onventional materials system. It is evident from these data that the spinel phase annot be formed for films deposited in relatively low partial pressures of oxygen. The data in Figure 8 are for a film deposited in pure oxygen, at room-temperature and then annealed in pure argon at progressively higher temperatures. This figure shows that these are 5 5 Wavelength (pm) Figure 6: Variation of the speular refletane of three films deposited at different xygen:argon flow-rate rate ratios. The substrate was not heated and annealing was not used.

9 superior deposition onditions. The mobility inreases from to 5 m V-' s-', although there is a less pronouned effet on the arrier onentration. The highest 6 mobility for an film deposited in t h s >I4 way was 54 my V- s-. Figure 9 shows.3 the X-ray diffration spetra of CIO films TS= 35 C 5 deposited in pure oxygen on substrates at various temperatures. They were not annealed after deposition, and only the () peak of In, was observed.. Figure illustrates the benefits of postts = 3 C deposition annealing for films deposited TS = C in pure oxygen at several substrate temperatures. Annealing was performed in pure argon at 68 C, after whih all the peaks assoiated with the spinel phase were observed. Hene, we onlude that post-deposition annealing is essential for the formation of the spinel phase. Figure 7: X-ray diffration spetra of CIO films The data shown in Figure apply to deposited in a.3% xygen:argon mixture on another film deposited in pure oxygen on substrates at various temperatures, and annealed an unheated substrate, but annealed in in pure argon. CdS/Ar at 68 C, that had previously been shown to have benefiial effets on 5 CdSn, films [ l, 4. The arrier on.7 entration attained a value of nearly m 45 'E 6 ~ ~~ 'm -the ~,highest for any film fab.5, " riated being 7. 5 ~ ~ ~m~- ~Even. with u suh a high arrier onentration, the v, 4 was relatively large,.3-5 mobility > N 3 m V s-'. The resistivity of this film was. 3 ~ ' S ~ m, the lowest to our. knowledge attained for admium indate..% The preise nature of the doping by the CdS is unknown, although it is tempting I z to speulate that it is due to interstitial.7 admium. This, however, is an open issue. Figure shows the effet on the X.5 ray diffration spetrum after the CdS/Ar annealing proedure. It is lear that there Annealing temperature ("C) is extensive rystallization of the film, ln3 () h v '5 Figure 8: Variation of mobility and arrier onentration with annealing temperature. The films had been deposited in pure oxygen, on an unheated substrate, and then were annealed in pure argon. with all the peaks assoiated with the spinel phase being present. This probably aounts for the large inrease in mobility. An as-deposited film is also shown, and this is learly amorphous. This was also suggested in Figure 9. It should also be mentioned that a simulation of the X-ray diffration spetrum of admium indate has been performed. A omparison between the simulation and the atual data shows that we are justified in laiming that the CIO is single-phase spinel after annealing.

10 This has also been onfirmed by referene to the JCPDS file. 6-4 C A Figure 9: X-ray diffration of CIO films deposited in pure oxygen on substrates at various temperatures. The films were not annealed Figure : X-ray diffration of a CIO film deposited in pure oxygen at various substrate temperatures, after annealing in pure argon at 68 C Figure 3 shows the optial properties of a CIO film deposited and annealed aording to the optimal proedures desribed above. The mobility of this film was 36.5 m V-' s-' and its arrier onentration was 6. 8 ~ ~~' r n - giving ~, a resistivity of. 3 ~!- ~m. The film thikness was approximately 55 A, thus giving a sheet resistane of 4. L per square. Notie that the free-arrier absorbane is approximately 7% at its peak, i.e., at the plasma wavelength, as predited by the model desribed earlier. The plasma wavelength for this film was approximately.5 pm, and the film does not ut off in the short-wave region until about.4 pm, giving a very wide wavelength range of high transmittane. Absorbane in the visible portion of the spetrum is less than 5% for most of the wavelength range, implying that this material ould have signifiant use for a vaziety of visiblelight appliations. It ould also be valuable for infra-red refleting appliations, as is illustrated in Figure 4. This shows that the long-wave speular refletane is approximately 95%. Aording to the theory of eletromagnetism, the long-wave limit refletane should only be ontrolled by the sheet resistane of the refleting film [7]. Based on a value of 4.! per square, a refletane of 95% is expeted, again making the point that a simple modelling approah appears adequate to explain the properties of these films. Figure 5 shows an atomi-fore mirograph of a film deposited and annealed under the optimized onditions desribed in this setion. The area examined is.5x.5 pm and, although the surfae appears rough, the rootomeansquare roughness is only + A. At this level, there is very little sattered light, whih is why the speular properties remain so high. For some appliations, a totally speular film is essential, while for others, a high degree of texturing is required. We have not yet attempted to develop texturing tehniques.

11 r E DISCUSSION AND SUMMARY The modelling work makes a number of interesting, if fairly well4.5 known, points. These are onsex quenes of the Drude theory of ele4 r.- trons in ondutors, and the theory 3.5 E of eletromagnetism. In partiular, it indiates that a high mobility is es.3 S8 sential if low-absorbane films are $ lo.5 8 required. If the arrier onentration L alone is inreased, then the extintion 5 oeffiient (and, hene, the absorp.5 tion oeffiient) inreases. Consequently, the absorption peak narrows (beause of the higher ondutivity), but inreases in height. On the other Temperature ( C) hand, if the arrier onentration is kept onstant and the mobility is inreased, then both the peak height Figure : Variation of mobility and arrier onand width derease. We hose to entration with annealing tempertaure. The work on the spinel family beause films were deposited in oxygen on an unheated high mobilities had been demonsubstrate and annealed in CdS/Ar 68 C strated in earlier work by Nozik [ 5 ], who suggested that higher mobility was due to a lower effetive-mass of ondution eletrons, although this 8 remains to be onfirmed. This is a After annealing at somewhat anomalous proposition 6 68 C in CdS/Ar and Haake [4] suggested that there >\ 4 may have been additional phases in.-v) his films. We have shown that sinf+ gle-phase films an b made by are ful hoie of deposition onditions,.- and have also demonstrated high?! mobilities. This perhaps supports the notion of lower effetive-mass, ) (5. (44) although a longer relaxation time would also aount for the superior mobility. Further analysis of the 4 spinel-based materials is needed to establish whih of these is responsible. Films deposited in low partial pressures of oxygen have relatively high mobility after annealing but do 8 not onsist of the spinel phase. To Figure : X-ray diffration of a CIO film deposited in produe this, it is essential to deposit pure oxygen on an unheated substrate and annealed in the films at room temperature in pure CdS/Ar at 68 C. Crystallization in CdS/Ar atually o- oxygen, and then anneal them at urred at a lower temperature that in Ar alone. high temperature in an inert atmosphere. Perhaps, the film phaseseparates when high-temperature deposition is used. In any event, oly indium oxide peaks are ob8 v 4-l /

12 , * served in the XRD spetra, and the resistivities are not speially low. In addition, the optial absorbane of these films is muh larger than that of those made using the optimal proedures. It appears that some spinel materials form the inverse phase more ommonly than the normal spinel ghase. Skribljak et al. [ 8 determined that CdIn,O, had the spinel struture (ubi) with a = 9. 5 A. They onluded that it was probably inverse spinel, but noted the diffiulty in distinguishing the normal from inverse struture due to the similarity of the Cd+and In3+sattering fators, More reently, Shannon et al. [9], based on ioni radii onsiderations, suggested that CdIn,Oq was probably a normal spinel. Normal spinel would have divalent admium on the tetrahedral sites and trivalent indium on the otahedral sites. For the inverse spinel, half of the indium atoms would be loated on the tetrahedral sites, while the admium atoms plus the other half of the indium atomwould be randomly distributed over the otahedral sites. The doping mehanisms are also not well-understood and may be related to an inter-play between the normal and the inverse forms. There are additional onsiderations, however. Films annealed in argon, rather than CdWAr, have a higher arrier mobility and the reasons for this are unknown, exept to suggest that ionized-impurity sattering is responsible for the redution in the CdS/Ar ase, beause of the higher arrier onentration. For admium stannate, it has been suggested that there may be a self-doping mehanism whereby group IV-tin atoms oupy group II-admium sites, thus ating as double-donors. For admium indate, if the equivalent were to our, lower arrier onentrations would be expeted beause the antiwavelength (pm) site defet would at as a singly ionized donor. This appears to be our, Figure 3: Transmittane, refletane, and absoralthough this ertainly does not onbane of a CIO film deposited at room-temperature, firm the self-doping (anti-site) model. in pure oxygen on an unheated substrate, and then The doping mehanism ould also be annealed in CdS/Ar at 68 C. due to interstitial admium although no proof of this exists, of whih we are aware. Thus, for both CdS annealing and non-admium annealing, there are several questions about the nature of the doping mehanism whih must be answered. We note that, even films deposited in pure oxygen retain quite high arrier onentrations, as-deposited. The usual argument for oxygen vaanies being the soure of the arriers does not, therefore, appear satisfatory. The spinel family is reeiving onsiderable attention beause of its apparent ability to provide a novel TCO material with superior properties to the onventional materials. CIO is one of the lesswell-investigated members of this family, presumably beause it ontains both toxi and ostly elements, and we have shown it to have onsiderable promise for various appliations requiring either extremely high transmittane in the visible region, or extremely high refletane in the infrared region. However, there are some appliations in whih these undesirable properties may not be problemati. We have also shown that it must be deposited under arefully ontrolled onditions, to optimize its properties. The single-phase spinel is essential for the best properties and this an only be ahieved by depositing the films at room-temperature in pure oxygen, and then annealing them in pure argon, or CdS/Ar, at as high a temperature as possible. This ertainly exludes the material as a TCO on temperature-sensitive substrates. This is an early point in our investigation of this material and there are many issues whih are not well-understood.

13 , t We have ahieved films with very low absorbane in the visible region of.9? the spetrum (<5%), very high refle.8 5 tanes in the infra-red region (>95%), (u and sheet resistanes as low as 4. SZ.7 + a per square. The plasma edge for the.6 > highest arrier onentrations lies at G= about. pm and the fundamental ab.5 + z sorption edge is at approximately pm, again easily adequate for visible go.3 light appliations. The films may be (I) ethed and patterned quite readily in. : either HC or HF, whih is an advan. tage over some materials like SnO,. F The films are very smooth even after heat-treatment, whih ensures that they 5 5 reflet and transmit light speularly, Wavelength (pm) rather than diffusely. However, the fat that they an be ethed so readily Figure 4: Refletane of a CIO film deposited in nieans that they ould probably be texpure oxygen on an unheated substrate, and antured if required. This, however, is nealed in CdS/Ar at 68 C. usually ahieved during growth, by areful hoie of deposition parameters. l L Q) 3... Figure 5: Atomi fore mirograph of the surfae of a CIO film showing the relative smoothness, on an atomi sale. The film was deposited and annealed under the optimized onditions desribed above.

14 - REFERENCES [ T. J. Coutts, X. Wu, and W. P. Mulligan., Eletroni Materials Conferene, Charlottesville, VA, 995; to be published in J. Ele. Mats. [] G. E. Guazzoni, and M. F. Rose, nd. NREL Conferene on Thermophotovoltai Generation ofeletriity, Colorado Springs, July 995, p. 6, Eds. J. P. Benner, T. J. Coutts, and D. S. Ginley, Amerian Institute of Physis Conferene Proeedings, No. 358 [3] K. Budzynska, E. Leja, and S. Skrzypek, Solar Energy Mats.,, 57, (985) [4] G. Haake, J. Apply Phys., 47, 486, (976) [SI A. J. Nozik, Phys. Rev., B6, 453, (97) [6]Proeedings of the 35th. National Symposium of the Amerian Vauum Soiety, Minneapolis, Otober, 995 [7] M. T. Mohammad, and W. A. S. Abdul Ghafor, Solid St. C o r n s., 88, 7, (993) [8] M. Skribljak, S. Dasgupta, and A. B. Biswas, Ata Cryst.,, 49, (959) [9] R. D. Shannon, J. L. Gillson, and R. J. Bouhard, J. Phys.. Chem. Solid., 38, 877, (977)