Solid phase crystallized polycrystalline thin-films on glass from evaporated silicon for photovoltaic applications

Thin Solid Films 513 (2006) 356 – 363 www.elsevier.com/locate/tsf

Solid phase crystallized polycrystalline thin-films on glass from evaporated silicon for photovoltaic applications Dengyuan Song ⁎, Daniel Inns, Axel Straub, Mason L. Terry, Patrick Campbell, Armin G. Aberle Centre of Excellence for Advanced Silicon Photovoltaics and Photonics, University of New South Wales, UNSW Sydney NSW 2052, Australia Received 28 June 2005; received in revised form 12 December 2005; accepted 10 January 2006 Available online 20 February 2006

Abstract Polycrystalline silicon (poly-Si) thin-films are made on planar and textured glass substrates by solid phase crystallization (SPC) of in situ doped amorphous silicon (a-Si) deposited by electron-beam evaporation. These materials are referred to by us as EVA materials (SPC of evaporated a-Si). The properties of EVA poly-Si films are characterised by Raman microscopy, transmission electron microscopy, and X-ray diffraction. A narrow and symmetrical Raman peak at a wave number of about 520 cm− 1 is observed for all samples, showing that the films are fully crystallized. X-ray diffraction (XRD) reveals that the films are preferentially (111)-oriented. Furthermore, the full width at half maximum of the dominant (111) XRD peaks indicates that the structural quality of the films is affected by the a-Si deposition temperature and the surface morphology of the glass substrates. A-Si deposition at 200 instead of 400 °C leads to an enhanced poly-Si grain size. On textured glass, the addition of a SiN barrier layer between the glass and the Si improves the poly-Si material quality. No such effect occurs on planar glass. Mesa-type solar cells are made from these EVA films on planar and textured glass. A strong correlation between the cells' current–voltage characteristics and their crystalline material quality is observed. © 2006 Elsevier B.V. All rights reserved. Keywords: Evaporation; Silicon; Crystallization; Solar cell

1. Introduction Thin-film photovoltaics hugely reduces the semiconductor material content of the finished product, with 150–200 times less material used than in conventional Si wafer based cells. Crystalline Si thin-film solar cells are considered to be promising [1] due to their non-toxic constituents, the abundance of Si in the earth's crust, and the huge technological experience with Si in the microelectronics industry. In recent years, there has been strong interest in polycrystalline silicon (poly-Si) thin-film solar cells on foreign substrates [1–5]. Note that poly-Si, in contrast to microcrystalline Si, has no amorphous tissue in it. One of the most promising substrate materials for low-cost photovoltaics is glass. However, glass cannot tolerate lengthy high-temperature steps, and hence a number of low-temperature crystallization techniques for amorphous Si (a-Si), such as solid phase crystallization (SPC), rapid thermal annealing (RTA) and laser ⁎ Corresponding author. Tel.: +61 2 9385 5426; fax: +61 2 9662 4240. E-mail address: [email protected] (D. Song). 0040-6090/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2006.01.010

annealing crystallization, are currently being explored. Of these, the SPC technology appears to be the simplest approach, giving relatively good crystalline material quality while sufficient throughput can potentially be achieved. Most publications on SPC of a-Si are related to a-Si formation by plasma-enhanced chemical vapour deposition (PECVD) [2] or sputtering [6,7]. As a new route, in this paper poly-Si thinfilms are made on glass by SPC of electron-beam (e-beam) evaporated unhydrogenated a-Si. These poly-Si materials are referred to by us as EVA poly-Si material (SPC of evaporated aSi), and the resulting photovoltaic devices as EVA solar cells. Compared to the established SPC poly-Si on glass photovoltaic technology [3], e-beam evaporation in a non-ultra-high vacuum environment appears to have the potential to be a less expensive and simpler process for depositing a-Si material. It offers specific advantages such as very high Si deposition rate (up to 1 μm/ min), excellent Si source material usage, avoidance of toxic gases, and simple sample preparation conditions. In this study, EVA poly-Si thin-film diodes are fabricated on planar and textured glass substrates. We examine their structural

D. Song et al. / Thin Solid Films 513 (2006) 356–363

material quality by means of transmission electron microscopy (TEM), Raman spectroscopy, and X-ray diffraction (XRD). The impact of the substrate topography on the poly-Si material quality is investigated by means of the FWHM (full width at half maximum) of the dominant (111) diffraction peak in the XRD spectra. The open-circuit voltage Voc of these thin-film diodes is evaluated prior to the metallisation process by quasi-steady-state open-circuit voltage measurements. Finally, mesa-type EVA poly-Si solar cells are made and their illuminated current– voltage (I–V) characteristics are measured. A strong relationship between the structural quality of EVA poly-Si films and the electrical characteristics of the corresponding solar cells is found. 2. Experimental procedure Planar borosilicate glass panes (Borofloat33, 5 × 5 cm2, 3 mm thick) from Schott AG, Germany are used as substrates. Compared to the commonly used soda lime glass, Borofloat33 glass has a significantly enhanced thermal stability and hence is well suited to the SPC process. The key features of this glass are: (i) the transformation of the glass from the elastic to the viscous state takes place at 525 °C, and the softening point is at 820 °C [8], which allows the fabrication of thin-films by SPC at temperatures of up to about 650 °C; (ii) the coefficient of mean linear thermal expansion is 3.25 × 10− 6 K− 1 [8], which very closely matches that of crystalline silicon of 2.6 × 10− 6 K−1 [9]. This property avoids excessive stress from different contraction during the cooling down phase after SPC; and (iii) high transmittance (∼91%) in the visible and near-infrared range (i.e., low optical absorption loss). The glass substrates used in this work have four different topographies: (i) bare planar glass [type I], (ii) SiN-coated planar glass [type II], (iii) bare textured glass [type III], and (iv) SiNcoated textured glass [type IV]. The textured glass surface is realised using a novel method (aluminium induced texture; AIT) developed by us. Details of the AIT glass texturing method can be found in Refs. [10,11]. The AIT process starts with the deposition of a ∼600 nm thick layer of aluminum onto one surface of the glass pane. The Al-coated glass is then annealed at 500–630 °C for a period of 2–8 h to allow Al to reduce silicon dioxide in the interfacial region. This reaction occurs spatially non-uniformly. Finally, the resulting layer of Al oxide and traces of Al and Si in this layer are removed in a 2-step chemical etch (10 min in hot H3PO4, then about 30 s in a 1 : 1 HF : HNO3 solution at room temperature), giving a textured glass surface. The applied annealing temperatures, times and etching processes determine the topography of the textured glass surface. By selecting particular process parameters (annealing temperature and time, etching conditions), specific surface textures can be realised with respect to the root mean square roughness as well as the feature sizes. An approximately 80 nm thick silicon nitride (SiN) film is then deposited at 350 °C by PECVD (13.56-MHz parallel-plate system) onto one surface of some of the glass substrates. The purpose of the SiN film is primarily to act as a barrier layer between the glass and the poly-Si film, minimizing diffusion of

357

contaminants from the glass into the Si film during thermal sample treatments (such as SPC and RTA). Using a glass temperature of about 200–400 °C, a-Si films with an n+ n− p+ doping structure and a total thickness of 2– 2.5 μm are then evaporated with an e-beam. The in situ doping of the a-Si film is realized using gallium (Ga) and phosphorus (P) effusion cells. The n+ and p+ layers are each about 100 nm thick and have doping concentrations of 1 × 1020 cm− 3 (P) and 2 × 1018 cm− 3 (Ga), respectively. The n− base layer has a doping concentration of about 2 × 1016 cm− 3. Upon deposition of the aSi films, the SPC process is performed in a tube furnace in N2 ambient at 600 °C for a period of 48 h. Then the EVA thin-films receive an RTA process to activate dopants and improve the structural material quality, followed by a PECVD plasma hydrogenation process to improve their electronic properties. For the RTA, the samples are pushed into a conventional quartz tube furnace held at a constant temperature of 900 °C and removed after a few minutes. The hydrogenation is performed at 350 °C glass temperature in an ammonia plasma, using a standard 13.56-MHz parallel-plate PECVD system. Finally, mesa-type EVA solar cells are made from the hydrogenated material and their electrical performance under 1-Sun illumination (1 Sun = 1 kW/m2, sample temperature = 25 °C) is characterized. The film thickness is measured with a stylus surface roughness detector (Dektak IIA). The topographical properties of the films are characterized by atomic force microscopy (AFM; Digital Instruments, Dimension 3000 Nanoscope). The AFM images were taken in the contact mode, using a standard silicon nitride probe from Digital Instruments and a scan frequency of 1.0 Hz. The collected data consisted of the height information on a square sample area (216 × 216 pixels). The crystallinity and structure of the films are characterized by transmission electron microscopy (Philips, CM-200), Raman spectroscopy (Renishaw, RM-2000), and X-ray diffraction (Siemens, D5000). TEM samples were prepared by cutting cross-sections using a diamond saw, gluing and then thinning them using conventional techniques of mechanical polishing and argon ion milling. The XRD machine was typically operated at a voltage of 30 kV and a current of 30 mA, using Cu Kα radiation (λ = 1.540562 Å). A standard setting for continuous scans is an angular speed of 1°/ min and a step size of 0.02° in a 2θ range of 20–80°. The FWHM of the (111) XRD peaks is determined with an XRD data processing software (Trances 6, Diffraction Tech. Pty Ltd). The used Raman system is a micro-Raman spectrometer in backscattering configuration, with a 50× optical microscope objective. The laser light comes from an Ar ion laser, has a wavelength of 514.4 nm, and is applied to the air-side surface of the poly-Si films. Nominally, the laser-illuminated spot on the sample surface has a diameter of about 5 μm. The system is calibrated with a c-Si wafer grown by the float-zone (FZ) technique, which has a Raman peak at around 521 cm− 1 at room temperature. The open-circuit voltage Voc of the nonmetallised thin-film diodes is measured by means of the quasisteady-state Voc method (abbreviated “Suns-Voc“ in this work; see Ref. [12] for a description of the method.). The illuminated 1-sun I–V curves of the mesa-type EVA solar cells are measured under approximated standard test conditions (AM1.5G

358

D. Song et al. / Thin Solid Films 513 (2006) 356–363

spectrum, 100 mW/cm2, 25 °C junction temperature), using an array of halogen lamps for illumination, a temperature-controlled sample holder, and a computer-controlled data acquisition system.

3.1. Topography Fig. 1 shows the surface profiles obtained by AFM imaging of as-grown EVA poly-Si films made on (a) planar glass and (b) textured glass. The scanned area is 3 × 3 μm2 for the planar sample and 10 × 10 μm2 for the textured sample. A distinctly different surface topography is evident. The planar poly-Si films have a root-mean-square (RMS) roughness of 1–2 nm, depending on the a-Si deposition temperature and the presence of a SiN barrier layer. On planar glass we found that the SiN barrier layer increases the surface roughness of the poly-Si films, an effect we attribute to the topography of the PECVD-deposited SiN film. Evaporation of the a-Si at 200 °C instead of 400 °C resulted in a slightly lower surface roughness of the poly-Si films. One possible reason is that, during the SPC process, Si atoms are more mobile in a-Si films with higher degree of disorder (corresponding to a lower deposition temperature), resulting in a smoother surface. The measured surface topography of the textured samples shows that the AIT glass texture process forms concave dimples in the glass surface. The RMS roughness is totally dominated by the textured glass surface, which in turn strongly depends on the chosen AIT glass texturing process parameters. The RMS roughness of the poly-Si films on textured glass is 100–200 nm. 10

(b) (a) (e) 400

450

500

550

Raman shift

600

650

(cm-1)

Fig. 2. Raman spectra of EVA cells on four different glass substrates: (a) planar glass, (b) SiN-coated planar glass, (c) textured glass, and (d) SiN-coated textured glass. Curve (e) is the Raman spectrum of an a-Si film evaporated at 200 °C.

The difference between the highest and lowest points is about 800 nm. In this case, the surface roughness caused during the aSi deposition is negligible compared to the roughness due to the glass texture. The textured glass substrates used in the present work have a feature size of approximately 1–2 μm. 3.2. Structural material quality 3.2.1. Raman spectra and TEM Raman spectroscopy is used as a fast and convenient method for the evaluation of the structural quality of the fabricated polySi films. Fig. 2 shows the Raman spectra of four as-grown poly-

data

0

fitted Peak 1

-10 0

1

2

3

X Axis (μm) 500

(b) Z Axis (nm)

(c)

Peak 2

c-Si (a) (b) (c) (d) Poly-Si

Intensity (arb. units)

Z Axis (nm)

(a)

(d) Intensity (arb. units)

3. Properties of as-grown EVA poly-Si films

~520 cm-1

480

0

500

520

540

560

FZ c-Si 440

480

520

560

600

Raman shift (cm-1) -500 0

2

4 6 X Axis (μm)

8

10

Fig. 1. Surface profiles obtained by AFM imaging of as-grown EVA poly-Si films made on (a) planar glass and (b) textured glass.

Fig. 3. Comparison of the Raman spectra of a FZ c-Si wafer (solid line) and EVA poly-Si films prepared on (a) planar glass, (b) SiN-coated planar glass, (c) textured glass, and (d) SiN-coated textured glass. The inset shows that the measured Raman curves can be well described by two superimposed peaks located at about 520 and about 510 cm− 1.

D. Song et al. / Thin Solid Films 513 (2006) 356–363

Si films as a function of the substrate topography (curves a–d). For comparative purposes, the Raman spectrum measured on an evaporated a-Si film (curve e) is included. It can be seen that the as-deposited Si film exhibits a completely amorphous phase, as is evidenced by a broad transverse optical (TO) peak at about 480 cm− 1. After SPC, the crystalline silicon phase is observed for all samples at wave numbers of about 520 cm− 1 (c-Si TO phonon mode). The narrow and symmetrical Raman peaks are very similar to that of the FZ c-Si reference sample, indicating good crystalline silicon material quality for each of the four investigated substrate topographies. Fig. 3 shows a detailed comparison of the Raman spectra of these EVA poly-Si films with that of the FZ c-Si reference sample. It is found that, in the wave number range 490– 510 cm− 1, a small shoulder appears in the Raman spectra of all EVA films. By means of curve fitting it can be shown (see inset of Fig. 3) that the measured Raman signals can be well described by two superimposed peaks centred at about 520 cm− 1 (the main peak) and about 510 cm− 1 (a weak and broad peak). According to the literature, the latter peak is attributed to grain size related effects in small-grained (b 10 nm) crystalline silicon [13]. This suggests that our SPC samples contain some regions with very small grains (nanocrystalline Si). A further explanation for the Raman peak at 510 cm− 1 are grain boundary defects. Fig. 4 shows bright-field cross-sectional transmission electron microscopy (XTEM) images of representative EVA

359

poly-Si films made on planar glass (plot a), SiN-coated planar glass (plot b), textured glass (plot c), and SiN-coated textured glass (plot d). The precursor a-Si film of plots a and c was deposited at 200 °C, while the a-Si film for plots b and d was deposited at 400 °C. From these TEM results it follows that the poly-Si films are composed of a number of differently sized grains. The reason is the absence of a crystalline seed layer on the substrate, and thus the phase transformation starts with the spontaneous formation of microcrystallites in the amorphous matrix. A few of them will become large enough (i.e., nucleation) for their growth to continue, resulting in a nonuniform grain size distribution. In addition, comparing films with and without a SiN barrier layer, the SiN does not seem to have an effect on the random nucleation and the grain growth. Furthermore, these XTEM micrographs indicate that our poly-Si films are fully crystallized. No amorphous volume fraction can be seen. However, there exist small-grained (nanocrystalline) Si regions and microtwins that reside within the crystal grains and/ or grain boundaries, which is consistent with the Raman results of Fig. 3. The TEM images show that the grain size of EVA material is in the range 0.8–1.5 μm. No evident difference in the poly-Si grain size is observed for the four different types of glass substrate. It should be noted that the vertical grain sizes of most grains are 1 / 3 to 2 / 3 of the film thickness, it is therefore expected that these grain boundaries do not severely affect the

Fig. 4. Cross-sectional TEM images of EVA poly-Si films made on four types of substrate: (a) planar glass, (b) SiN-coated planar glass, (c) textured glass, and (d) SiNcoated textured glass. The precursor a-Si films were deposited at 200 °C for plots (a, c) and at 400 °C for plots (b, d).

D. Song et al. / Thin Solid Films 513 (2006) 356–363

current collection of photovoltaic cells made from these films. A detailed look at Fig. 4 reveals that an improved grain size (almost the entire film thickness) occurs for poly-Si films that were made using a low a-Si deposition temperature (200 °C, see plots a and c). Generally, the grain size is mainly controlled by the nucleation rate and the grain growth rate [14], which are strongly related to the structural disorder in the a-Si network. An increase in structural disorder of the a-Si network deposited at 200 °C compared to 400 °C results in an improved poly-Si grain size.

0.20 0.19 (111) FWHM (deg.)

360

a-Si deposition temperature: 400 °C

0.18

200 °C

0.17 0.16 0.15

(111) (220)

Intensity (arb. units)

(311)

(d) (c) (b)

(a) (e) 20

30

40

50

60

70

80

2 θ (degree) Fig. 5. X-ray diffraction patterns of EVA poly-Si films prepared on (a) planar glass, (b) SiN-coated planar glass, (c) textured glass, and (d) SiN-coated textured glass. Curve (e) is the XRD pattern of the a-Si film. The precursor a-Si films were deposited at 200 °C.

Pl an /p ar ol gl y- as Si s

Pl /S ana iN r /p gla ol s y- s Si

0.14 Te x /S ture iN d /p g o l la s ySi s Te xt u /p red ol y- gla Si ss

3.2.2. X-ray diffraction Fig. 5 shows the XRD spectra of our as-grown films as a function of the glass substrate topography (curves a–d). The XRD spectrum of an evaporated a-Si film is included as a reference (curve e). It is noted that the XRD spectrum of the evaporated a-Si film does not show any crystalline peaks, in agreement with the above Raman measurement results. After SPC all poly-Si samples exhibit diffraction peaks at 2θ values of about 28.4°, 47.3° and 56.2°, which are due to the {111}, {220} and {311} Si planes (see the XRD reference data for standard Si powder from the Joint Committee on Powder Diffraction Standards [15]). These well-developed peaks indicate that structurally good-quality crystalline silicon has been formed during SPC and that the films are of a polycrystalline nature. The broad structure around 2θ = 22° is due to the glass substrate, which was confirmed by performing an XRD scan on a bare borosilicate glass pane. The orientation factors Ohkl are calculated by comparing the intensities of XRD diffraction peaks with those obtained on a randomly oriented standard powder [16]. Considering that only the three major diffraction peaks (i.e., {111}, {220}, and {311}) are included in our calculation, the samples have randomly oriented grains when the three orientation factors are close to

Fig. 6. FWHM of the (111) diffraction peaks of EVA poly-Si films as a function of the glass substrate topography. The precursor a-Si layers were deposited at 400 and 200 °C, respectively.

33%. The orientation factor O111 of our poly-Si films has a value of about 38–42% for all investigated substrate types, showing that the crystallites in our EVA poly-Si films are preferably b111N oriented. The other two important orientations, O220 and O311, have values in the range 27–32%. Fig. 6 shows the FWHM of the dominant (111) diffraction peaks of EVA poly-Si films as a function of the substrate type. The a-Si deposition temperature was either 400 °C (squares) or 200 °C (circles), while all other process parameters were kept the same. This allows to investigate the effect of the substrate type on the film's crystallinity for the two a-Si deposition temperatures. It can be seen from Fig. 6 that the glass substrate topography affects the crystallinity of the poly-Si film. The trends in the FWHM are the same for both a-Si deposition temperatures. Comparing the crystalline quality for a given a-Si deposition temperature, poly-Si films crystallized directly on bare textured glass have the largest FWHM value of all investigated samples, indicating a degraded crystalline material quality. An explanation for this behaviour might be that a rough surface, especially some steep steps, disturbs both the nucleation process and the crystallization process of the films during annealing [17]. It is found that the use of a SiN film on the AITtextured glass greatly improves the crystallinity of the poly-Si films, leading to a reduced FWHM. The reason is the smoothening of the glass by the SiN. In contrast, a beneficial effect of a SiN barrier layer on planar glass on the crystallinity is not evident. With respect to the a-Si deposition temperature it is found that smaller FWHM values (i.e., a larger grain size according to Scherrer's formula [18]) are obtained for the lower deposition temperature (200 vs. 400 °C), regardless of the substrate type. These results are consistent with the results obtained from the TEM investigations. The enhanced grain size results from the increased structural disorder in the a-Si films deposited at lower temperature [19]. The higher degree of structural disorder in low-temperature deposited a-Si films causes a lower grain

D. Song et al. / Thin Solid Films 513 (2006) 356–363

360 320

(a)

280 Voc(mV)

nucleation rate in the initial stage of the SPC process [14], resulting in a larger grain size. We also tested a-Si deposition temperatures below 200 °C, however, the resulting poly-Si films had severe cracks, making them useless for solar cell fabrication. This indicates that an a-Si deposition temperature of 200 °C seems to be ideal for our SPC process.

361

240 200 160 120

4. Electrical properties of EVA solar cells

80 40

4.1. Quasi-steady-state Voc of non-metallised cells

after SPC after RTA after hyd. Light intensity (kW/m2)

FF (%)

50 40

(b)

30 20 10 0 8 7

(c)

Jsc(mA/cm2)

6 5 4 3

Type

2

Ave. Eff (%)

IV

III

0.98 0.52

II

I

1.02 1.12

1

an a ol r gl y- as Si s

Pl

/p

Pl /S ana iN r /p gla ol s y- s Si

xt u /p red ol y- gla Si ss

Te

x /S ture iN d /p g ol las ySi s

0

Fig. 8. 1-Sun parameters Voc (a), FF (b), Jsc (c) and Eff (inset in c) of EVA polySi solar cells as a function of the glass substrate topography. The data represent the average values of 4 mesa cells, respectively. The error bars are the standard deviations.

10

n=2

60

Te

EVA poly-Si thin-film solar cells were fabricated for the purpose of exploring the photovoltaic potential of this new polySi material and characterizing the diodes' electrical properties. To improve the electronic material quality of as-crystallized EVA films, the samples receive an RTA process and then a plasma hydrogenation process [20]. The open-circuit voltage of these thin-film diodes is evaluated with the Suns-Voc method because this does not require metal contacts, saving a lot of processing time during the initial stage of the development of a new material. Fig. 7 shows the Suns-Voc characteristics of a nonmetallised planar EVA solar cell at three different processing stages: after SPC (triangles), after RTA (squares), and after hydrogenation (circles). Also shown in the plot are fits of the Suns-Voc curves with a 2-diode model (using fixed diode ideality factors n of 1.0 and 2.0) plus a shunt resistor. It can be seen that the two post-deposition treatments shift the entire Suns-Voc curves to the right-hand side, resulting in a large improvement of the 1-Sun Voc (i.e., the Voc under 1-Sun illumination) from 121 mV (after SPC) to 338 mV (after hydrogenation). An optimisation of the RTA and hydrogenation processes of EVA poly-Si films can be found in Ref. [21]. The Voc increase due to the RTA and hydrogenation processes is largely caused by an improved crystalline material quality, an increase of the electrically active dopant fraction in the base region, and the passivation of defects by atomic hydrogen [20,21].

0 70

n=1 n=1

n=1

1

n=2

SPC RTA Hyd. fit

n=2

shunt

The 2-diode model fits in Fig. 7 reveal that, at each process stage (SPC, RTA, hydrogenation), the 1-Sun Voc is limited by the n = 2 diode. This suggests that the 1-Sun Voc is dominated by space charge region recombination. It is so far unclear whether this recombination occurs in the junction depletion region, the grain boundary depletion regions, or both.

0.1 0.0

0.1

0.2

0.3

0.4

0.5

4.2. I–V characteristics of mesa-type cells

Open-circuit voltage (V) Fig. 7. Measured Suns-Voc characteristics of a planar EVA solar cell after SPC (triangles), after RTA (squares), and after hydrogenation (circles). 1-Sun = 1 kW/ m2 and sample temperature = 25 °C. The solid lines are fits to the measured data using a 2-diode model with fixed ideality factors of 1.0 and 2.0 and a shunt resistance. The dashed and dashed-dotted lines are the n = 1 and n = 2 diodes of the solid-line fits, respectively.

The RTA-treated and hydrogenated materials were then processed into mesa-type solar cells, enabling the investigation of the cells' I–V characteristics. The cells have a circular active area (diameter = 4 mm, area = 0.126 cm2) and are illuminated from the air side (glass = substrate). The top electrode is a 80 nm thick sputter-deposited aluminium-doped ZnO film that

362

D. Song et al. / Thin Solid Films 513 (2006) 356–363

simultaneously acts as the cell's antireflection coating. The rear electrode is a 800 nm thick evaporated Al film that covers the entire periphery of the cell. To avoid Al evaporation onto the active area of the cell, a shading mask is attached to the samples during Al evaporation. The illuminated 1-Sun I–V curves of the fabricated EVA solar cells were measured under approximated standard test conditions (AM1.5G spectrum, 100 mW/cm2, 25 °C junction temperature), using a computer-controlled I–V tester. Fig. 8 shows the averaged 1-Sun parameters Voc, FF, Jsc and Eff of mesa-type EVA cells as a function of the glass substrate type. In each case the presented data is the average of 4 samples, whereby the samples were jointly RTA processed and hydrogenated to ensure similar material quality. The error bars represents the standard deviations. It is found that the cells' open-circuit voltage correlates very well with their crystalline material quality (see Fig. 6). The highest Voc is observed on planar glass, whereby inclusion of a SiN barrier layer has no impact on Voc. On textured glass the voltages are substantially reduced, particularly for samples that do not have a SiN barrier layer. The glass substrate topography also has an impact on Jsc (plot c). The cells on the textured substrates exhibit a slightly higher Jsc than those on planar glass, which is due to a light trapping effect. Cells on bare textured glass not only show a low Voc but also have a poor fill factor (plot b). This behaviour can be understood from the XRD FWHM results of Fig. 6. The large FWHM of the (111) XRD peaks of the films made on bare textured glass indicates that these films are small-grained and hence have a large grain boundary area per cm2 of sample surface area. We therefore attribute the reduced Voc of these samples to increased carrier recombination at grain boundaries. The reduced fill factor probably largely results from the additional series resistance due to the increased number of grain boundaries with potential barriers. Fig. 8 demonstrates that both the Voc and the fill factor can be improved on textured glass by depositing a SiN barrier layer onto the glass. It should be noted that the two types of textured samples have a slightly different short-circuit current density. We believe that this is mainly due to process-related variations of the AIT glass texturing method, resulting in slightly different light trapping properties. 5. Conclusions We have fabricated and characterized about 2 μm thick EVA poly-Si thin-films on both planar and textured glass substrates. The films are made by SPC of evaporated a-Si and are intended for photovoltaic applications. The films have a grain size in the range 0.8–1.5 μm. Raman spectra of the films show a sharp peak at a wave number of about 520 cm− 1, revealing that the poly-Si films have a high crystalline material quality. XRD analysis showed that lower FWHM values (i.e., an improved grain size) are obtained if the a-Si is evaporated at low substrate temperature (200 vs. 400 °C). Furthermore, FWHM analysis of the (111) XRD peaks reveals the impact of the glass substrate topography on the crystallinity of the films. Poly-Si films grown directly on bare textured glass have the largest FWHM. It was demonstrated

that the use of a SiN coating on AIT-textured glass greatly improves the crystallinity of the poly-Si films, leading to a reduced FWHM. No such effect occurs on planar glass substrates. The 2-diode model fits of the Suns-Voc curves reveal that the 1-Sun Voc is limited by n = 2 recombination. This suggests that the 1-Sun Voc of our cells is dominated by recombination in the p–n junction space charge region and/or at grain boundaries. I–V characteristics of mesa-type EVA cells show that the open-circuit voltage correlates well with the crystalline material quality. Compared to planar samples, EVA cells on textured glass exhibit a slightly higher Jsc due to a light trapping effect, however, their Voc is lower. A SiN barrier layer on the textured glass substrates was found to significantly improve both the crystalline quality of the poly-Si films and the energy conversion efficiency of the resulting solar cells. Since EVA cells are still at an early stage of their development, and the cell structure is not yet optimised, the prepared cells do not have a high efficiency. Both Jsc and Voc of these preliminary devices need further improvement, primarily via better minority carrier lifetime properties of the absorber layer. Experimental optimization of the cell properties is in progress. Acknowledgements This work has been supported by the Australian Research Council via its Centres of Excellence scheme. The authors thank Yidan Huang for the TEM measurements. DS, DI, AS and MLT acknowledge Ph.D. scholarships from The University of New South Wales. MLT additionally acknowledges a National Science Foundation Graduate Research Fellowship (USA). References [1] Z. Shi, S.R. Wenham, Prog. Photovolt. 2 (1994) 153. [2] T. Matsuyama, N. Terada, T. Baba, T. Sawada, S. Tsuge, K. Wakisaka, S. Tsuda, J. Non-Cryst. Solids 198-200 (1996) 940. [3] P.A. Basore, Proceedings of 29th IEEE Photovoltaic Specialists Conference, New Orleans, U.S.A., May 20-24, 2002, p. 49. [4] K.R. Catchpole, M.J. McCann, K.J. Weber, A.W. Blakers, Sol. Energy Mater. Sol. Cells 68 (2001) 173. [5] R.B. Bergmann, Appl. Phys. A 69 (1999) 187. [6] R. Rüther, J. Livingstone, N. Dytlewski, Thin Solid Films 310 (1997) 67. [7] G. Farhi, M. Aoucher, T. Mohammed-Brahim, Sol. Energy Mater. Sol. Cells 72 (2002) 551. [8] Schott Inc., Product Properties, (http://www.us.schott.com/whitegoods/ english/products/borofloat/attribute). [9] S.M. Sze, Physics of Semiconductor Devices, John Wiley & Sons, New York NY, 1981. [10] A.G. Aberle, P.I. Widenborg, N. Chuangsuwanich, International PCT patent application PCT/AU2004/000339, 19 March 2004. [11] N. Chuangsuwanich, P.I. Widenborg, P. Campbell, A.G. Aberle, Tech. Digest 14th International Photovoltaic Science and Engineering Conference, Bangkok, Thailand, Jan. 26–30, 2004, p. 325. [12] N.P. Harder, A.B. Sproul, T. Brammer, A.G. Aberle, J. Appl. Phys. 94 (2003) 2473. [13] S. Vepřek, F.A. Sarott, Phys. Rev. B 36 (1987) 3344. [14] K. Nakazawa, K. Tanaka, J. Appl. Phys. 68 (1990) 1029. [15] Powder Diffraction File, Joint Committee on Power Diffraction StandardsInternational Center for Diffraction Data, Swarthmore, PA, 1989, Card 271402.

D. Song et al. / Thin Solid Films 513 (2006) 356–363 [16] P. Joubert, L. Loisel, Y. Chouan, L. Haji, J. Electrochem. Soc. 134 (1987) 2541. [17] K. Roth, J.C. Goldschmidt, T. Puzzer, N. Chuangsuwanich, B. Vogl, A.G. Aberle, Proceedings of 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, May 11–18, 2003, p. 1202. [18] B.D. Cullity, Elements of X-ray Diffraction, 2nd Ed, Addison-Wesley, Reading, 1968, p. 102. [19] T. Baba, T. Matsuyama, T. Sawada, T. Takahama, K. Wakisaka, S. Tsuda, S. Nakano, Proceedings of 24th IEEE Photovoltaic Specialists Conference, Hawaii, U.S.A, Dec. 5–9, 1994, p. 1315.

363

[20] D. Song, A. Straub, P.I. Widenborg, B. Vogl, P. Campbell, Y. Huang, A.G. Aberle, Proceedings of 19th European Photovoltaic Solar Energy Conference, Paris, France, June 7–11, 2004, p. 1193. [21] M.L. Terry, A. Straub, D. Inns, D. Song, A.G. Aberle, Appl. Phys. Lett. 86 (2005) 172108.