Structural analyses of seeded thin film microcrystallinesilicon solar cellS. N. Agbo1*, S. Dobrovolskiy1, G. Wegh1, R. A. C. M. M. van Swaaij1, F. D. Tichelaar2,P. Sutta3 and M. Zeman1
1 Department of Electrical Sustainable Energy, Delft University of Technology, P. O. Box 5053, 2600GB Delft, The Netherlands2 Kavli Institute of Nanoscience Delft, Delft University of Technology, Lorentzweg 1, 2628 CJ, Delft, The Netherlands3 New Technologies Research Centre, University of West Bohemia, 306 14 Plzen, Czech Republic
thin film microcrystalline silicon; seed layers; crystallinity profile; spectral response
*Correspondence
S. N. Agbo, Department of Electrical Sustainable Energy, Delft University of Technology, The Netherlands.E-mail: [email protected]
Received 14 July 2011; Revised 16 December 2011; Accepted 17 July 2012
1. INTRODUCTION
In conventional radio-frequency plasma enhanced chemicalvapor deposition (rf PECVD), the growth of hydrogenatedthin film microcrystalline silicon (mc-Si :H) begins with anamorphous silicon incubation layer (ASIL), then continuesthrough the amorphous-to-microcrystalline transition intothe crystalline regime. Crystal growth nucleates on theunderlying amorphous tissue, and the crystal size andorientation depend on deposition parameters. The effectof this incubation layer on the performance of mc-Si : Hsolar cells has been widely reported [1–3]. In a p-i-n solarcell, for instance, absorption losses in the ASIL layer ofthe p-layer result in a reduced short-wavelength responseof the solar cell [4]. Reducing the thickness of the ASIL
layer during growth and obtaining high crystalline devicequality mc-Si : H material has been a major research chal-lenge till date.
Various approaches have been implemented to controlthe ASIL layer and thereby increase the crystalline massfraction, f, in the mc-Si : H layer. Van den Donker et al.[5] demonstrated that by means of controlling the transientdepletion of silane during rf PECVD, the ASIL layer canbe reduced. This they did by allowing for a delay betweenthe time of introduction of silane into the process chamberand the starting of the plasma. Agbo et al. [4] implementedthis technique in mc-Si : H p-i-n solar cell and observed anincrease in the blue response of the solar cell. A majorchallenge of this technique is the very narrow processwindow required for this process. Another approach often
PROGRESS IN PHOTOVOLTAICS: RESEARCH AND APPLICATIONSProg. Photovolt: Res. Appl. 2014; 22:346–355
Published online 26 August 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/pip.2274
employed is to grow a very thick mc-Si : H layer such thatthe incubation layer effect is reduced [6]. For example,up to 50-nm-thick mc-Si : H p-layer and 2000-nm absorberlayer are deposited. The difficulty here is coping with theincreased absorption with the increased film thicknessand the increased deposition time and material cost.
In recent years, depositing a seed layer is one of themostly used options to suppress the ASIL layer andenhance crystallization in the growth direction during thedeposition of both doped and intrinsic microcrystallinesilicon layers [7–9]. It involves the deposition of very thinlayers of mainly intrinsic mc-Si : H usually under very highhydrogen dilution. The seed layer serves as the nucleationlayer upon which the active layers of the solar cells aredeposited. By seeding the p-layer deposition, the p-i interfacein the solar cell is more transparent and the materialgrowth evolves with a more uniform f. Better charge carrierextraction has been reported [9].
The relationship between solar cell performance and thecrystallinity profile in the intrinsic mc-Si : H layer in growthdirection has not been demonstrated so far. This profilewill give an indication of the phase changes that occur asmc-Si :H is formed because the structure of mc-Si :H generallyevolves as a function of thickness due to the kinetics ofcrystalline nucleation and grain growth [10]. In this con-tribution, we report results of the effect of seeding on thef development profile of mc-Si : H and on the performanceof p-i-n and n-i-p mc-Si : H solar cells. For this, we inves-tigate the crystalline mass fraction in growth direction andshow that by seeding the f profile in the active layers, thetwo solar cell configurations are modified. Using a Ramansetup and Dektak step profiler, we present a crystallinitydepth profile of the solar cells. From transmission electronmicroscopy (TEM) image analyses and x-ray diffraction(XRD) measurements, we probe the structural propertiesof the bulk of the solar cells. Properties such as crystallinemass fraction, crystal size and orientation, and micro-strainare investigated.
2. EXPERIMENTAL DETAILS
For this study, we used the rf PECVD set-up in our labora-tory. First, the seed layers, which in our case are highlycrystalline intrinsic mc-Si : H layers, were optimized. Theoptimization focused on the thickness and the f and theireffect on material properties of mc-Si :H p-layer and i-layer.Details of the optimization results are reported elsewhere[11]. The optimized seed layers were implemented in p-i-nand n-i-p mc-Si : H solar cells. In the p-i-n series, first areference solar cell (herein after called sample A) was depos-ited without applying any seeding. Next, a p-i-n solar cell(sample B) of similar structure was deposited except thata 40-nm-thick intrinsic seed layer was deposited after thep-layer. In another solar cell (sample C), a seed layer wasintroduced before the growth of both the p-layer andi-layer. Sample A had the following structure: Corning(E2000) glass/800 nm ZnO :Al/20 nm mc-Si : H p/~1100nm
mc-Si : H i/20 nm a-Si :H n-layer/100 nm Ag/200 nm Al.For the n-i-p solar cells, a reference unseeded (sample D)and a seeded (sample E)were deposited using the same depo-sition conditions as the p-i-n cells except for the sequence ofdeposition. For sample E, the 40-nm-thick intrinsic seed layerwas deposited on top of the a-Si :H n-layer to precede thegrowth of the i-layer. The n-i-p cells had 250-nm-thick flatZnO :Al layer as back contact with a 300-nm-thick Ag layerdeposited on Corning glass (E2000). The structure of the un-seeded p-i-n and n-i-p solar cells is shown in Figure 1. For allexperiments, the thickness of the i-layer in the solar cells wasabout 1000–1200 nm. The solar cells were characterized bycarrying out external quantum-efficiency measurements andcurrent–voltage measurements under AM1.5 illumination.
The crystallinity development profile in the solar cellswas determined by a combination of reactive ion etching(RIE) and Raman spectroscopy. The crystallinity develop-ment profile here refers to the depth profile of fmeasured ingrowth direction. For this purpose, we used the RIE system(Alcatel CL GIR 300) in our group to etch through thedepths of the solar cells beginning from the top n-layerfor p-i-n (and top p-layer for n-i-p) solar cells down tothe front (back) TCO layer. The RIE plasma is generatedfrom a gas mixture composed of 70 sccm of CF4, 10 sccmof SF6 and 10 sccm of O2 at a pressure of 0.05mbar. RIE iscarried out at 50 �C. The etching time is within 50 s and solimits the possibility of plasma-induced crystallization ofthe films. Part of the sample was protected from the etchingby an Al bar. Depending on the pre-determined etch rate,the depths were determined using the Dektak step profiler(Veeco Dektak 150 with 12.5mm stylus). Each RIEprocess step etched 20–80 nm of the test material with aresolution of ~20nm. Raman measurements were performedat various depths in the solar cells after each etching step.Note that the exact penetration depth of the Raman varies
Figure 1. The schematic structure (not to scale) of the referenceunseeded (a) p-i-n and (b) n-i-p microcrystalline silicon solar cells.
Seeded thin film microcrystalline silicon solar cellS. N. Agbo et al.
as the crystalline mass fraction changes with depth. Thischange in crystalline mass fraction was also confirmed bythe change in the RIE etching rate. The Raman spectra weremeasured using a Ramanmicroscope (Renishaw, Ramascopesystem 2000, grating 1800 lines/mm) in a back scatteringgeometry with a 2-mW Ar laser at a wavelength of514.5 nm focused on a spot of about 1mm in diameter. Thepenetration depth of the 514.5-nm laser in mc-Si : H isestimated to be around 300 nm [12]. The crystalline massfraction was extracted from the Raman spectroscopy mea-surement by implementing peak-fitting approach followingthe procedure of Smit et al. [13].
Transmission electron microscopy measurements werecarried out on samples A to E. A FEI Tecnai F20ST/STEMtransmission electron microscope was used for imaging.Cross-sections of the samples were prepared by gluing aprotective glass on the deposited film and cutting a cross-section of thickness ~0.5mm. Mechanical grinding andpolishing of the cross-section to ~10 mm and a subsequentthinning down to electron transparency with a Gatan Arion mill PIPS model 691 followed. Bright and dark fieldimages were taken as well as lattice TEM high resolutionimages (HREM) without any objective aperture. Fast Fouriertransforms (FFTs) of the HREM images were carriedout. Because bright and dark field contrast of the samegrain changes a lot in the growth direction because ofdefects and bending of the TEM foil, only the graindiameter parallel to the substrate interface was deter-mined. Dark field images are not shown because theydid not reveal any additional information with respect tothe bright-field images.
Further structural characterization of the solar cells forcrystallite size and their crystallographic orientation andmicro-strains was investigated by XRD analysis. An auto-matic powder x-ray diffractometer X’Pert Pro equippedwith an ultra-fast linear semiconductor detector PIXceladjusted as a point detector was used for this purpose.Copper Ka radiation (l=0.154 nm) served as an x-ray source.The ceramic alumina from the National Institute of Standardsand Technology was used as instrumental standard.
3. RESULTS AND DISCUSSION
3.1. Seeding effects on p-type and intrinsicmc-Si :H
In Figure 2(a), the Raman spectra of 25-nm-thick p-layersdeposited on seed layers grown using varying silaneconcentration, Sc (Sc is the ratio between the silane flow tothe sum of the flow of silane and hydrogen) are comparedwith the reference case with no seed layer. We observethat for the p-layers with no seeding applied (none) andthe p-layer with a seed layer that has a large amorphousfraction (Sc=0.25%), the Raman spectra show a strongcontribution of the 480 cm�1 peak; an indication of a signif-icant amorphous fraction in the material from the onset ofgrowth. For low values of Sc (0.05% and 0.15%), the
Raman spectrum shows a more pronounced crystalline peakat 520 cm�1, indicating that there is virtually no amorphousfraction in the film. The Raman spectra of the intrinsic layersof different Sc are shown in Figure 2(b). The thickness of thei-layers ranged from 402 to 483 nm. The thicknesses of thei-layers were limited to this range so as to be able to probethe entire film. Higher values for f are observed in i-layers(although with high Sc) than in p-layers because thesei-layers are thicker and the crystalline mass fraction increaseswith film thickness. Without a seed layer, the Raman spec-trum clearly shows a broad peak at 480 cm�1. With seeding,f increases and this relates directly to other material prop-erties of the film: the dark conductivity increases and theactivation energy decreases [11].
3.2. Effectsof seed layerson theperformanceof p-i-n and n-i-p mc-Si :H solar cells
In Figure 3(a), the external quantum efficiencies (EQE) ofp-i-n solar cells A, B and C are presented. The figure shows
Figure 2. Raman spectra of (a) 25-nm-thick p-layers and (b)402–483-nm-thick intrinsic mc-Si : H layers deposited on seedlayers grown at varying Sc. The measurements were takenfrom the glass side to show the crystallinity structure of thelayers from the onset of growth. The legends indicate the
silane concentration of the seed layers.
Seeded thin film microcrystalline silicon solar cell S. N. Agbo et al.
that implementing seed layers in the growth of mc-Si : Hp-layer and i-layer of mc-Si : H p-i-n solar cell leads to alarge increase in short-circuit current density of the solarcell. In sample B, the increase in Jsc is observed to arisemainly from the increase in the blue/green region spectralresponse. Seeding of i-layer facilitates crystal nucleationhence resulting in a more crystalline and transparent material[14]. A more crystalline and transparent mc-Si : H materialimplies less absorption and hence lower solar cell spectralresponse. The EQE, however, increases because of bettercharge carrier collection from a more favorable electric fielddistribution in the cell. With seeding of both the p-layer andi-layer, as in sample C, the solar cell’s overall performanceeven increases further because of extended spectral responseof the solar cell to the infrared region. From the Jsc valuesobtained from EQE curves as shown in Figure 4, the Jscincreases by 10% for sample B and by 37% for sample C.On the other hand, the Voc and the fill factor drop by 11%and 2.9%, respectively.
Figure 3(b) shows a comparison between the EQE ofseeded (sample D) and unseeded (sample E) n-i-p solarcells. Unlike the p-i-n cells, the EQE show a slight decreasein the blue on seeding (sample E) and no significant effect in
the red region of the spectrum with respect to the referenceunseeded cell (sample D). Overall, the Jsc decreases by2%. The Voc increases slightly by 5% and the fill factordrops by 7.9% upon seeding. The seed layer used hereaffects the i-n interface, which is less critical to chargecarrier generation and collection than the p-i interface. Eventhough the value of f is increased from onset of growth inthis region, little contribution is made to the overall perfor-mance of the solar cell.
In conclusion, the use of seed layers leads to a gain inJsc and a drop in Voc in p-i-n cells, whereas the oppositeis observed in n-i-p cells. The different effects observedin p-i-n and n-i-p solar cells can be linked to the differentpositions of the seed layers in the solar cells and the differ-ent underlying substrates to the seed layer growth. Due tothe deposition sequence in n-i-p cells, the seed layer isdeposited on top of the a-Si : H n-layer. To grow mc-Si : Hon an amorphous silicon under-layer will result in a defec-tive transition layer at or near the interface [9]. This willlead to poor device performance and explains why then-i-p solar cell performance degrades upon seeding. Thedrop in the Voc can be linked to the increased porosity ofthe bulk material with increase in f [15].
3.3. Crystallinity development profiles ofmc-Si : H solar cells
The crystalline mass fraction as a function of depth in theexperimental solar cells is shown in Figure 5. In Figure 5(a), we observe that the crystallinity profile of the seededp-i-n solar cell (sample C) taken through the cross sectionof the solar cell is more uniform and follows from rapidnucleation. A gradual build up of the crystallinity in growthdirection is observed without seeding as in sample A.
Figure 3. External quantum efficiencies of unseeded and seeded(a) p-i-n and (b) n-i-p mc-Si : H solar cells. Note that the solar cellsare not fully optimized and the p-i-n cells had no back reflectors.
Figure 4. The I–V characteristics of the unseeded (sample A:p-i-n and sample D: n-i-p) and seeded (sample B and C: p-i-nand sample E: n-i-p) mc-Si : H solar cells. The external parameters(inset table): Voc, Jsc, and � have their units as V, mA/cm2 and %,respectively. The parameter, Jsc (EQE) is obtained by multiplyingthe EQE and the photon flux from AM1.5 spectrum and thenintegrating over the wavelength range. The value of � is based
on Jsc (AM1.5) obtained from solar simulator.
Seeded thin film microcrystalline silicon solar cellS. N. Agbo et al.
In seeded solar cells, crystallinity evolution is morerapid in the region close to the seed layers. In allsamples, the crystallinity increases in growth directionand saturates after attaining steady-state condition [10]at a film thickness of about 500 nm. At this thicknessthe film growth is at the full crystalline phase andhence no further increase in f is observed beyond thispoint. Notice that f begins with a value above 40% insample C, which is an indication that the ASIL iscompletely suppressed. The thickness of ASIL can beestimated from the depth profile of Figure 5(a) and hasa thickness of about 80 nm. However, ASIL can vary inthickness depending on the underlying substrate and thedeposition conditions [16]. The use of seed layer for therf PECVD growth of thin film microcrystalline siliconfacilitates crystal nucleation and eliminates the adverseeffect of amorphous silicon incubation layer. This showsthe possibility to grow microcrystalline silicon directly froma crystalline growth phase without the growth evolvingthrough the intermediate phases.
Crystallinity depth profiles for n-i-p mc-Si : H solar cellsare presented in Figure 5(b). Just as in the p-i-n cells, the
crystallinity is enhanced with seed layers and is alsoobserved to lead to rapid crystallinity build up or nucle-ation in growth direction and provides constant f of about60% afterwards. The increase in crystallinity observed forboth seeded p-i-n and n-i-p cells does not yield similarresults in the performance of the solar cells. The differ-ence can be attributed to the differences in the underlyingsubstrates and the sequence of deposition between p-i-nand n-i-p solar cells.
3.4. Microstructure analysis of p-i-n and n-i-pmc-Si :H solar cells
3.4.1. Analysis by transmission electronmicroscopy.Figure 6 shows the (i) TEM high resolution images(HREM), (ii) bright field (BF) images under focus, and(iii) the bright field images of the unseeded (sample A)and seeded (samples B and C) p-i-n solar cells. The highresolution images (HREM) of the samples are presentedwith insets showing the FFTs from a 15 nm� 60 nm areaclose to the ZnO :Al/mc-Si : H interface. The wide ringsare caused by the amorphous Si in this area. The ringpatterns consisting of individual spots indicate the pres-ence of silicon crystallites within the amorphous phase.The crystal density is higher in samples B and C than insample A, as estimated from inspection of a number ofimages as in Figure 6 (A(i), B(i) and C(i)) and their FFTs.From the BF images of all samples, isolated and elongatedmicrocrystals parallel to the growth direction and typicalfor microcrystalline growth are observed. The elongatedcrystallites are irregular in shape and have an estimatedaverage diameter of about 30 nm and a length that extendsthroughout almost the entire thickness of the solar cell. Thegrain size was estimated from bright field images at largemagnification. Columnar and cone-shaped crystals are alsoobserved. In bright-field images with higher magnificationand taken slightly under focus we observe elongated verti-cal features that correspond to voids, and cracks at grainboundaries that sometimes also extend throughout theentire thickness of the device. Narrow cracks (indicatedby arrows in Figure 6) (ii) appear as white lines in bright-field images that are slightly under-focused [17]. Fromthe BF images horizontal cracks are seen everywherewithin the first 35 nm of the mc-Si : H layer in sample A.This observation is in line with an earlier report of Roschek[18] where these cracks are found in regions close to thesubstrate and are linked to internal stress in the film duringthe onset of growth. Similarly, in sample B, horizontalcracks are also observed but within ~15–25 nm from thesubstrate. No horizontal cracks are observed in sample Cwithin 35 nm from the substrate. We attribute the reductionof cracks by seeding within the substrate-film interface tothe reduced stress from ASIL on the evolving material.Here the seed layer also serves as a buffer. The contactregion between the denser ASIL and the less dense overly-ing mc-Si : H experiences sheer stress and the attendantstrain effect is the horizontal cracks.
Figure 5. Depth profile of the crystalline mass fraction in (a) p-i-nand (b) n-i-p mc-Si : H solar cells. The arrows (not drawn to scale)are guides to the eye and indicate the starting and ending pointof RIE experiment. The error bars are the standard deviationvalues of the crystalline mass fraction obtained for different
Raman measurements taken at each given depth.
Seeded thin film microcrystalline silicon solar cell S. N. Agbo et al.
Transmission electron microscopy images of the n-i-pcells are presented in Figure 7. BF images of the completesolar cells are shown in Figure 7 D(i) and E(i); BF imagesof the top and the bottom regions of the absorber layer areshown in D(ii) and E(ii) and D(iii) and E(iii), respectively.Elongated grains are observed within the bulk of theabsorber layers, which are thinner than in the case of thep-i-n cells. A crack-rich layer with horizontal cracks was
detected 20–30 nm from the front ZnO :Al in unseededsample D. Also vertical cracks are evident. At the regionclose to the back contact such a layer was not observedas can be seen in Figure 7 D(iii) and E(iii). Similar imagesof seeded sample E also reveal both vertical and horizontalcracks. This trend is opposite to what we observed for thep-i-n cells. Note that in the n-i-p configuration the cracksappear in that part of the layer that is deposited at the end
Figure 6. TEM images showing (i) TEM high resolution images (HREM), (ii) bright field (BF) images under focus, and (iii) the bright fieldimages of unseeded (sample A) and seeded (samples B and C) p-i-n solar cells. (111), (220) and (113) lattice planes of Si are visible and
are reflected as spots in the FFTs. The scale bars are indicated for each image.
Figure 7. Transmission electron microscopy images of the entire structure of the n-i-p mc-Si : H solar cells for unseeded Sample D (Di)and seeded sample E (Ei). Columns (ii) and (iii) show the images of the top of the intrinsic layer and directly above the front ZnO:Al layer,
respectively for the two solar cells.
Seeded thin film microcrystalline silicon solar cellS. N. Agbo et al.
of the i-layer growth, whereas in the p-i-n the cracksappear in the region that is grown first. The differencebetween sample D and E from the HREM images lies inE having a larger crystallinity than sample D.
With seeding, f increases in both p-i-n and n-i-p solarcells. For p-i-n cells the increase in f is observed withoutcracks especially at regions close to the p-i interface. Onthe other hand, for n-i-p cells, cracks are observed nearthe interface. These cracks inhibit charge carrier transporthence a reduction of the performance of these n-i-p solarcells. Similar effects have been reported for solar cells withseed layers grown on an amorphous under-layer [9].
Grain size analyses of both p-i-n and n-i-p solar cells arepresented in Figure 8. The average diameter of the grainswas determined from high magnification bright fieldimages, at five different distances from the substrate withinthe first 300 nm of the silicon layer. Number of grains mea-sured for each position varied from 20 to 65 and the plotteddata in Figure 8 are the average values of the grain diame-ter determined from the number of grains counted for eachposition. The error bars represent the spread of the mea-sured values in the diameter distribution. The grain size(diameter) distribution of crystals in all samples show ageneral increase in growth direction as has also beenobserved by Hwang et al. [19]. An average grain size of30 nm is obtained. Seeding does not affect crystal sizebut affects the crystalline mass fraction in mc-Si : H forp-i-n cells. On the other hand, n-i-p cells show a decreasein the crystal size with seeding and this difference incrystal sizes gets bigger in growth direction. Figure 8 alsoindicates that n-i-p cells have smaller crystal sizes than thep-i-n cells.
3.4.2. Seed layer effect on the x-ray diffractionpattern of p-i-n and n-i-p mc-Si : H solar cells.In this study we use a procedure utilizing an integral widthof a diffraction line [20,21]. Equation 1 characterizes the
integral breadth b that includes two parameters, namelythe height and the integrated intensity.
b ¼ 1I0
ZI 2#ð Þ�d 2#ð Þ (1)
where Iint =RI(2#) � d(2#) is the integrated intensity (area
below the line) and I0 is the maximal intensity of thediffraction line. The instrumental resolution of the equip-ment was taken into account in order to obtain a physical(depending only on the properties of the material) compo-nent of the broadening of the diffraction line. Furthermore,the physical component of the integral breadth of thediffraction line is a convolution of Cauchy and Gaussiancomponents and so it is necessary to deconvolute into a
Cauchy part bfC and a Gaussian part bfG before the structural
parameters are extracted. The Cauchy and Gaussian partsof the integral breadth of the line represent the size of thecrystallites and the micro-strain, respectively. The averagecrystallite size is determined from the relation [21]:
Dh i ¼ l
bfC cos#(2)
Here, <D> is the average crystallite size in the directionperpendicular to the diffracting lattice planes, whereas lis the x-ray wavelength used and # is the Bragg’s angle.
eh i ¼ bfG4tg#
(3)
<e> as expressed in Equation 3 is the average micro-strainin the diffracting volume.
The XRD patterns of the test p-i-n and n-i-p solar cellsare presented in Figure 9(a) and (b), respectively. The pat-tern as shown in Figure 9(a) is typical of highly crystallinemc-Si : H material with the strong visible lines representingthe (111), (220) and (311) planes. The solar cells seem toshow preferential crystal orientation in the [111] directionirrespective of seeding. The intensity of the (111) siliconcrystal plane increases when seeding is used for the i-layer.For the more crystalline sample C having both p-layer andi-layer seeded, the line increases even further, an indicationthat crystal orientation is affected by seeding. The (220)and the (311) planes vary only slightly with seeding. Thebroadening of the first broad scattering line at the 2θposition of about 27.5� with seeding can be linked to theordered region of tetragonal silicon hydride [22]. For then-i-p cells, a contrary effect is observed. The (111) siliconplane intensity drops slightly with seeding, and the scatter-ing line at 27.5� 2θ position shows no variation as pre-sented in Figure 9(b). In both solar cell configurations,dominant peak intensity is mainly affected on seedingalthough the identity of the dominant peak (Si(111)) staysthe same.
Details of the microstructure analysis of the XRD resultsare presented in Table I. The crystalline mass fraction of the
Figure 8. Average grain sizes of mc-Si :H crystals calculated fromthe different positions in the bulk of both the p-i-n and n-i-p cells.The error margins are standard deviations from the average values
and lie between 10% and 30% of the plotted values.
Seeded thin film microcrystalline silicon solar cell S. N. Agbo et al.
p-i-n solar cells stays the same even with seeding. On theother hand, the f of sample E increases by 16% in compari-son with the unseeded sample D. These f values differ fromvalues obtained by Raman and this may be attributed todifferences in the probing technique and the probing depthbetween Raman and XRD. The two techniques probe differ-ent depths and regions of the samples. The probing depth of
XRD depends on wavelength (energy) of x-ray radiationused, and on the absorption of the material. This depth canextend up to 35mm in silicon for CuK-alpha x-ray radiation.Raman spectroscopy using a green laser on the other handhas a much lower probing depth, which is limited also to asmaller volume than the XRD. This can explain the differ-ences in f values obtained from the two techniques.
Average grain sizes of 20, 30 and 19 nm are obtainedfor samples A, B and C, respectively, and these sizes arein agreement with the grain sizes obtained from TEM.The grain size as well as the micro-strain in the materialsshows no dependence upon seeding. An increase in grainsize results in an increase in the micro-strain for both solarcell configurations. N-i-p cells show much smaller grainsizes in line with the TEM results, and a stronger depen-dence of the micro-strain on seeding is observed as itincreases by ~52% upon seeding. The general increase inthe micro-strain with grain size can be due to the increasedstructural stress in the films arising from increase in inter-action of grains as the grain sizes become bigger [23].
4. CONCLUSIONS
The use of seed layers to enhance crystal nucleation andcrystallinity in mc-Si : H material has been investigated.For conventional rf PECVD growth of microcrystallinesilicon film under high hydrogen dilution, we show thatnucleation starts on the underlying amorphous incubationlayer and crystallinity increases gradually in growth direc-tion. With the aid of seed layers, this process is modified.In this case, crystallization is rapid and starts almost fromonset of growth. In both p-i-n and n-i-p cells, we observethat the crystalline mass fraction saturates at the full crys-talline regime. The use of seed layers for the rf PECVDgrowth of thin film microcrystalline silicon facilitatescrystal nucleation and suppresses the amorphous siliconincubation layer. This way, it is possible to grow micro-crystalline silicon of uniform crystalline mass fractiondirectly from a crystalline growth phase without the growthevolving through the intermediate phases.
Optimized seed layers were implemented in p-i-n andn-i-p mc-Si : H solar cells. The use of seed layers leads toan increase in Jsc and a decrease in Voc in p-i-n mc-Si :H cells
Figure 9. X-ray diffraction patterns of the unseeded and seeded(a) p-i-n and (b) n-i-p mc-Si : H solar cells. The integrated intensityof the line is proportional to the amount of material (atoms,lattice planes) having the same distances from each other where
diffraction of x-rays takes place.
Table I. Orientation of lattice planes (hkl ) and micro-phase parameters in thin film mc-Si : H solar cells.
The intensity here refers to the integrated intensity of the line, which is the area below the diffraction line profile. Standard values represent the relative
integrated intensity values of the lines for completely randomly oriented crystallites (lattice planes) in space. The standard values were calculated from
the theory of diffraction using APX 63 Struc software.
Seeded thin film microcrystalline silicon solar cellS. N. Agbo et al.
and a decrease in Jsc and increase in Voc in n-i-p cells. Weobtained 37% increase in the Jsc with seeding in p-i-n solarcells relative to the reference case with no seeded layer.
In n-i-p cells, the efficiencies of the solar cells drop onseeding. The EQE of n-i-p cells decrease with seeding,whereas a strong increase is observed for the p-i-n cells.TEM measurements have revealed that seed layers canreduce the crack density in p-i-n mc-Si : H. The crystal sizeincreases in growth direction in both p-i-n and n-i-p solarcells and showed no dependence on seeding. However, inp-i-n cells, slightly larger grain sizes were found than inn-i-p cells. XRD results indicate that the peak intensity ofthe dominant crystal orientation is affected by the use ofseed layers, whereas its identity (in this case, Si (111))stays the same. In p-i-n solar cells, dominant [111] orienta-tion is observed in all solar cells and an increase in thecontribution to the spectrum of the (111) silicon plane isobserved with seeding. For the n-i-p cells, the dominanceof the (111) silicon plane is also observed, however itshows a slight drop on seeding. Crystallite sizes calculatedfrom XRD results corroborate the results from TEM andshows crystallites sizes in the range of 13–30 nm for allthe cells investigated, with the n-i-p cells having smallercrystallite sizes than the p-i-n cells.
ACKNOWLEDGEMENTS
The research of Solomon Agbo is facilitated by the Nether-lands Organization for International Cooperation in HigherEducation (NUFFIC). The technical support of MartijnTijssen and Kasper Zwetsloot during the deposition andcharacterization of the films and devices is highly appreci-ated. This work was partially supported by the CENTEMproject, reg. no.CZ.1.05/2.1.00/03.0088, which is co-fundedfrom the ERDF within the OP RDI Programme of theMinistry of Education, Youth and Sports and within theproject no. 1M06031.
REFERENCES
1. Koh J, Ferlauto AS, Rovira PI, Wronski CR, CollinsRW. Evolutionary phase diagrams for plasma-enhancedchemical vapor deposition of silicon thin films fromhydrogen-diluted silane. Applied Physics Letters 1999;75: 2286–2288.
2. Ferlauto AS, Koval RJ, Wronski CR, Collins RW.Extended phase diagrams for guiding plasma-enhancedchemical vapor deposition of silicon thin films forphotovoltaic applications. Applied Physics Letters2002; 80: 2666–2668.
3. Fujiwara H, Kondo M, Matsuda A, Real-time spectro-scopic ellipsometry studies of the nucleation and graingrowth processes in microcrystalline silicon thin films.Physical Review B 2001; 63: 115306–115314.
4. Agbo SN, Van Swaaij RACMM, Zeman M. Effect ofdeposition parameters of mc-Si : H solar cell absorberlayer on spectral response. Proceedings of the EleventhAnnual Workshop on Semiconductor Advances forFuture Electronic and SENSORS (SAFE), Veldhoven,Netherlands, 2008;. 440–443.
5. Van den Donker MN, Rech B, Finger F, Kessels WMM,Van de Sanden MCM. Highly efficient microcrystallinesilicon solar cells deposited from a pure SiH4 flow.Applied Physics Letters 2005; 87: 263503–263505.
6. Sarker A, Banerjee C, Barua AK. Lowering ofthickness of boron-doped microcrystalline hydroge-nated silicon film by seeding technique. Solar EnergyMaterials and Solar Cells 2005; 86: 365–371.
7. Zhou J, Ikuta K, Yasuda T, Umeda T, Yamasaki S,Tanaka K. Growth of amorphous-layer-free microcrys-talline silicon on insulating glass substrates by plasma-enhanced chemical vapor deposition. Applied PhysicsLetters 1997; 71: 1534–1536.
8. Vetterl O, Hulsbeck M, Wolff J, Carius R, Finger F.Preparation of microcrystalline silicon seed-layerswith defined structural properties. Thin Solid Films2003; 427: 46–50.
9. Li Y, Li L, Anna Selvan J, Delahoy A, Levy R. Effectsof seeding methods on the fabrication of microcrystal-line silicon solar cells using radio frequency plasmaenhanced chemical vapor deposition. Thin Solid Films2005; 483: 84–88.
10. Roca i Cabarrocas P, Fontcuberta A, Kalache B,Kasouit S. Microcrystalline silicon thin films grown byplasma enhanced chemical vapor deposition—growthmechanisms and grain size control. Solid StatePhenomena 2003; 93: 257–268.
11. Agbo SN, Van Swaaij RACMM, Zeman M. Effect ofseeding on the material property of mc-Si : H layer andthe spectral response of mc-Si :H solar cell. Proceedingsof the Twenty-fifth European Photovoltaic Solar EnergyConference, Valencia 2010; 3288–3291.
12. Droz C, Vallat-Sauvain E, Bailat J, Feitknecht L,Meier J, Shah A. Relationship between Raman crystal-linity and open circuit voltage in microcrystalline siliconsolar cells. Solar Energy Materials and Solar Cells2004; 81: 61–71.
13. Smit C, Van Swaaij RACMM, Donker H, PetitAMHN, Kessels WMM, Van de Sanden MCM.Determining the material structure of microcrystallinesilicon from Raman spectra. Journal of Applied Physics2003; 94: 3582–3588.
14. Huang Y, Chen T, Gordijn A, Dasgupta A, Finger F,Carius R. Preparation of microcrystalline silicon solarcells on microcrystalline silicon carbide window layersgrown with HWCVD at low temperature. Journal ofNon-Crystalline Solids 2008; 354: 2430–2434.
Seeded thin film microcrystalline silicon solar cell S. N. Agbo et al.
15. Madhumita N, Roca i Cabarrocas P, Johnson EV,Abramov A, Chatterjee P. The open-circuit voltage inmicrocrystalline silicon solar cells of different degreesof crystallinity. Thin Solid Films 2008; 516: 6974–6978.
16. Roca i Cabarrocas P, Layadi N, Heitz T, Drevillon B,Solomon I. Substrate selectivity in the formationof microcrystalline silicon: mechanisms and techno-logical consequences. Applied Physics Letters 1995;66: 3609–3611.
17. Python M, Domine D, Soderstrom T, Meillaud F,Ballif C. Microcrystalline silicon solar cells: effect ofsubstrate temperature on cracks and their role in post-oxidation. Progress in Photovoltaics: Research andApplications 2010; 18: 491–499.
18. Roschek T. Microcrystalline silicon solar cell preparedby 13.56MHz PECVD. PhD Thesis, Institute forPhotovoltaics, Forschungzentrum Julich, 2003.
19. Hwang HL, Wang KC, Hsu KC, Yew TR, Loferski JJ,Microstructure evolution of hydrogenated silicon thin
films. Progress in Photovoltaics: Research and Appli-cations 1996; 4: 165–192.
20. Langford JI. A rapid method for analyzing thebreadths of diffraction and spectral lines using theVoigt function. Journal of Applied Crystallography1978; 11: 10–14.
21. Delhez R, Keijser TH, Mittemeijer EJ. Determinationof crystallite size and lattice distortions through x-raydiffraction line profile analysis. Fresenius’ Journal ofAnalytical Chemistry 1982; 312: 1–16.
22. Mullerova J, Prusakova L, Netrvalova M, VavrunkovaV, Sutta P. A study of optical absorption inamorphous hydrogenated silicon thin films of var-ied thickness. Applied Surface Science 2010; 256:5667–5671.
23. Zhao Y, Zhang J. Microstrain and grain-size analysisfrom diffraction peak width and graphical derivationof high-pressure thermomechanics. Journal of AppliedCrystallography 2008; 41: 1095–1108.
Seeded thin film microcrystalline silicon solar cellS. N. Agbo et al.
