By Topic

• Abstract

SECTION I

## INTRODUCTION

MATERIAL technology innovations to exchange hazardous, rare, and costly substances for clean and sustainable resources need to be developed for all commoditized mass-produced devices that include the silicon (Si) solar cells of photovoltaic (PV) industries. Silver (Ag) application films (AFs) are widely used in Si solar cells as front-side contacts and backside soldering tabs/pads [1], [2], [3], [4], [5], [6], [7], [8], [9], [10]. The first attempt to make a crystalline metallic copper (Cu) film rather than an Ag AF used an electroplating technique developed for front-side contacts of a Si solar cell [11]. However, producing Cu AFs by atmospheric sintering, which is currently used for most Ag AFs [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], would be industrially more attractive because this technique is anticipated to not only exchange rare Ag with more sustainable Cu but to achieve excellent cost performance in both producing the material and processing the crystalline Si PV as well. Sintering is the heat treatment that is applied to AFs to make them electrical conductors; however, Cu oxidizes easily in an atmospheric environment at temperatures of $\geq 200 ^{\circ}$C [12], [13], [14], [15], and the cuprous oxide (Cu2O) that is formed on heating essentially cannot self-deoxidize in this environment at elevated temperatures because temperatures above its boiling point are required [16], [17]. This has, therefore, restricted the range of temperature to conduct Cu AF studies to $< 200 ^{\circ}$C [18], and no Cu AF has yet been produced by an atmospheric sintering process at elevated temperatures $\geq 200 ^{\circ}$ C.

Here, we report the first production of a crystalline metallic Cu AF in an oxidizing environment during a high-temperature atmospheric sintering process. By using a paste consisting of a copper–phosphorus (Cu–P) alloy with a mixed microstructure of Cu and Cu phosphide (Cu3P) as a starting material for the sintering, we demonstrated the preferential growth of a crystalline metallic Cu network in air at temperatures $\geq 450 ^{\circ}$C and, thus, the formation of a Cu AF with low-electrical resistivity that satisfies the criteria for backside soldering tabs/pads used in crystalline Si PVs. We show the Cu–P alloy has the role that governs deoxidization of Cu 2O, which is formed during the sintering, by virtue of a concurrent oxidation of the Cu3P at an elevated temperature. The produced Cu AF also has a Cu glass hybrid microstructure in which Cu–P–O glasses (Cu 2(PO4) and Cu2P2O7) are self-assembled and embedded into the opening of the Cu network. Our results may open the way to the widespread use of atmospherically sintered Cu AFs in next-generation crystalline Si PVs.

SECTION II

## METHODOLOGY

Our method to fabricate Cu AFs consists of the atomizing Cu–P alloy particles, forming a Cu–P alloy paste by blending the particles with a solvent and binder, screen-printing the paste onto a substrate, and atmospherically sintering the screen-printed AF. Chemical and physical properties of the atmospherically sintered AFs at various temperatures are finally characterized to clarify the chemical mechanism of the Cu AF creation.

### A. Atomization of Copper–Phosphorus Alloy Particles

A high-pressure water atomization furnace with a swirl water jet [19] (Fukuda Metal Foil & Company, Ltd.) was used to make Cu–P alloy particles that had a chemical composition of Cu–6.57 mass%P (99.80% purity) (hereafter referred to as Cu–7 mass%P alloy particles), a tap density of 3.85 g/cm3, a Brunauer–Emmett–Teller specific surface area of 0.3 m2/g, and a laser 50% diameter of $4.408 \mu$ m, as determined by a Nisshin Engineering Inc. turbo-classifier TC-15.

### B. Paste Blending

The Cu–P alloy paste was prepared by blending 87.0 mass% Cu–P alloy particles with 12.61 mass% solvent and 0.39 mass% binder, with no glass frit, which is generally used in most Ag AFs [1], [2], [3], [4], [5], [6], [7], [8], [9], [10] in order to provide good adhesion and electrical contact with the substrate. The solvent was terpineol with an 88.9% purity, 0.932 specific gravity, 1.4834 refractive index, and 0.02 mass% water. The binder was Grade STD200 ethyl cellulose (ethoxized percentage 49%) with a molecular mass of 190.0.

### C. Screen-Printing

The Cu–P alloy paste was screen-printed in the form of a 30-$\mu$ m-thick layer (measured by cross-sectional scanning electron microscopy (SEM) after sintering) onto a 45 mm $\times 45$ mm area on a Si substrate using a printing press (Microtec Inc., MT-320T) with a printing mask (Sonocom Company, Ltd., KH250-30-60). The substrates were 185- $\mu$ m-thick P-type single-crystalline Si wafers. The wafer surface was textured using an etching solution of sodium hydroxide and isopropyl alcohol. An n+-Si layer on the surface of the wafer was formed using phosphorus oxychloride (POCl3). The sheet resistance of the n+-Si layer was adjusted to $40 \Omega$/sq. A 90-nm-thick SiNX layer was deposited onto the front side of the wafer using a PECVD process [20], [21]. The screen-printing was performed on the SiN X layer-deposited side of the wafer, and the screen-printed AF was then air-dried for 60 s on a conventional hot plate at a temperature of $\leq 150 ^{\circ}$ C.

### D. Atmospheric Sintering

The sintering was performed in an atmospheric environment using a tunnel kiln (Noritake Company Ltd., solar cell heat treatment machine-baking furnace for research and development) that was 507 mm wide internally, 907 mm long, and 523 mm high. The kiln was equipped with a walking beam system consisting of a quartz beam with a stainless pin that was used to transport the Si-wafer through the centre of the kiln. Sintering was performed in an atmospheric environment over the temperature range of 300–$500 ^{\circ}$ C, while the temperature was increased in a stepwise fashion at $50 ^{\circ}$C intervals. The temperature was controlled in three sequential heating zones, the first and second of which were used as preheating zones and the third of which was set to the desired sintering temperature. The first and second heating zones were held at $300 ^{\circ}$C unless the sintering temperature was $500 ^{\circ}$ C, in which case, the first and second preheating temperatures were 300 and $305 ^{\circ}$ C, respectively. Each zone was heated using ten near-infrared heaters, each 15 kW in capacity, placed above and below the beam position of the Si-wafer transportation system. The screen-printed film samples were continuously transported through the three zones, spending 10 s in each zone. The airflow inside the tunnel kiln was controlled using air intakes and exhaust ejectors mounted in each temperature zone. For the first and last zones, the flow through the air intakes was 50 l/min, and the flow through the exhausts was 80 l/min. The flow was zero for the intake and exhaust in the second zone. An 80-l/min air curtain was situated in a duct located between the sample entrance and the first preheating zone.

### E. Characterization of Atmospherically Sintered Application Films

The atmospherically sintered AFs were characterized using field-emission SEMs (Hitachi S-4800 and SU-70) observation, energy dispersive X-ray spectroscopy (EDX) (Horiba EX-350) analysis, electron backscattering pattern (EBSP) (TSL MSC-2200) measurements using a 20-kV acceleration voltage and a 0.13-$\mu$m step size, X-ray diffraction (XRD) (Rigaku RINT2500HL) using Cu K $\alpha$ radiation, and electrical conductivity measurements using an ac four-probe method (Mitsubishi Chemical Loresta-EP MCP-360 and Hiresta-UP MCP-HT450). The XRD measurements were performed using the focusing beam method $(\hbox{2}\theta - \theta$ rotation mechanism) with an X-ray output of 50 kV–250 mA, which allowed the whole thickness of the AFs to be probed. Continuous scans were performed over the $\hbox{2}\theta$ range of 5– $100 ^{\circ}$ with a scanning rate of $0.5 ^{\circ}$/min. A $5 ^{\circ}$ soller slit, $0.5 ^{\circ}$ divergence slit, $0.5 ^{\circ}$ scattering slit, and 0.15-mm receiving slit were incorporated into the XRD system. Data were sampled at steps of $0.01 ^{\circ}$. The integrated peak intensity associated with a given crystalline phase was analyzed as follows: First, peaks were identified and indexed on the basis of the crystal structures listed in the International Centre for Diffraction Data database; second, the data for the instrumental peak width were corrected using a standard sample (NIST-Si); third, the integrated intensities and widths of peaks were estimated by profile fitting; and fourth, the integrated intensity ratio for each phase was calculated by summing the integrated intensities of peaks associated with that phase and dividing them by the total sum of peak intensities for all phases.

Thermogravimetry-differential thermal analysis (TG-DTA) (Shimazu DTG60H) was also performed in an atmospheric environment with an airflow of 50 ml/min and a heating rate of $\hbox{0.167}\,^\circ\hbox{C/s} (\hbox{10}\,^\circ\hbox{C/min})$ in order to understand oxidation behavior of Cu3P phase. Samples of 30 mg in mass were loaded into an alumina cell as follows: as-atomized Cu-7 mass%P alloy particles with an average diameter of $\sim 1 \mu$m and pure Cu3P particles with diameters of $<300 \mu$m and a purity of $>99.0\%$ (Kojundo Chemical Laboratory Company, Ltd., Cu3P powder sieved through a $300 \mu$m mesh) were used for the TG-DTA.

SECTION III

## RESULTS

### A. Producing Copper Application Films by Atmospheric Sintering

Fig. 1(a) shows spherical Cu–P alloy particles produced using a high-pressure water atomization furnace with a swirl water jet that had a chemical composition of Cu–7 mass%P. Fig. 1(b) shows a cross-sectional view of a typical as-atomized Cu–7 mass%P particle. A proeutectic microstructure is evident, which is consistent with the Cu–P binary alloy phase diagram [22].

Fig. 1. Atmospheric sintering forms a crystalline metallic Cu network. (a) SEM secondary electron image of as-atomized starting material (Cu–P alloy particles) showing spherical shape. (b) Cross-sectional SEM backscattered electron image of an individual particle showing the pro-eutectic microstructure. Primary Cu crystals are evident, between which the eutectic consisting of Cu and Cu3P is visible as a lamellar structure. (c) Cross-sectional SEM secondary electron images showing the shapes of the alloy particles, with the elemental distribution (determined by EDX) mapped onto the images, in screen-printed AF samples atmospherically sintered at temperatures from 300 to $500 ^{\circ}$ C.

The screen-printed and dried Cu–P alloy paste AF were sintered in an atmospheric environment by increasing the temperature from 300 to $500 ^{\circ}$C at intervals of $50 ^{\circ}$ C, with a holding time of 10 s at each temperature. The sintering drastically changed both the shape of the Cu–P alloy particles in the AF and the distribution of the elements within the particles. Fig. 1(c) shows micrographs depicting the distribution of Cu (red), P (green), and oxygen (O; blue) determined by EDX analysis, mapped onto an SEM secondary electron image with the same cross-sectional field. The yellowish particles were not oxidized at $300 ^{\circ}$ C, whereas Cu2O rings appeared around the particles, visible as light pink fringes, at $350 ^{\circ}$ C. At temperatures of $\geq 450 ^{\circ}$ C, both the microstructure and the elemental distribution drastically changed: A crystalline metallic Cu network was formed, and a Cu–P–O glass was self-assembled in the openings of the network. Evidence is given later of the formation of the crystalline metallic Cu network by EBSP measurement. The Cu network appears pink in the SEM images because it contains a small amount of P (between 0.31 and 0.44 mass% as determined by EDX analysis) in the form of a dilute solid solution.

### B. Indicating Crystalline Metallic Copper Network

Fig. 2(a) shows inverse pole figure (IPF) maps (derived from EBSP measurements) of crystalline metallic Cu with the face-centered cubic lattice formed during the atmospheric sintering; Fig. 2(b) shows the corresponding SEM micrographs. At temperatures $\leq 400 ^{\circ}$ C, metallic Cu was found only in the form of small crystals within the Cu–P alloy particles; at temperatures $\geq 450 ^{\circ}$ C, a crystalline metallic Cu network with larger, unoriented grains was formed. The grains were completely interconnected to form a network, and some of the grains exhibited stripes, which were probably twin structures. We can thus conclude that the key reactions involved in the formation of the metallic Cu network—that is, the annihilation of fine-grained Cu formed by oxidation of the starting material and the subsequent generation of large-grained crystalline metallic Cu network by deoxidizing Cu2O—must occur at temperatures between 400 and $450 ^{\circ}$ C.

Fig. 2. Crystalline metallic Cu network generated by oxidizing small Cu crystals in Cu–P alloy particles followed by deoxidizing Cu 2O. (a) IPF maps (derived from EBSP measurements) showing the evolution of crystalline Cu distribution in cross sections of screen-printed AFs atmospherically sintered at the indicated temperatures. (b) SEM backscattered electron images corresponding to the IPF maps in (a). (c) Color code for Cu crystal orientations in the IPF maps shown in (a).

### C. Chemical Mechanism of Copper Application Film Formation

We used XRD to evaluate the fractions of phases present in AFs formed at different sintering temperatures, and the results are shown in Fig. 3 (XRD patterns) and Fig. 4 (integrated XRD peak intensity ratios) along with the corresponding electrical resistivities. The XRD patterns of all the examined AFs showed crystalline peaks (see Fig. 3); no halo patterns arising from amorphous substances were evident. No Cu oxide species were observed at $300 ^{\circ}$ C, which is consistent with the microstructure [see Fig. 1(c)]. When the sintering temperature was raised to $350 ^{\circ}$C and then to $400 ^{\circ}$ C, Cu2O appeared and its fraction increased, whereas that of crystalline metallic Cu decreased and that of Cu3P remained almost unchanged [see Fig. 4(a)]. No P–O compounds were detected. Evidently, Cu2O is deoxidized at $450 ^{\circ}$C because a rapid increase in the fraction of crystalline Cu occurred at that temperature (reaching an integrated intensity ratio of 66.9%), along with a corresponding decrease in the ratio of Cu2O to 12.0%. This reaction takes place in conjunction with the oxidation of Cu3P; the Cu3P disappeared, and the new compounds Cu 2(PO4 and Cu2P2O 7 appeared at $450 ^{\circ}$ C. We can thus conclude that the Cu–P alloy governs the deoxidization of Cu 2O by virtue of the concurrent oxidation of Cu 3P, and that Cu3P acts as a suitable reducing agent for Cu2O in the atmospherically sintered AF. Our XRD results imply that the following two reactions take place simultaneously and that reaction (2) largely contributes to the rapid increase in metallic Cu at $450 ^{\circ}$C TeX Source \eqalignno{{\rm 2\hbox {Cu}} + {1\over 2}\,{\hbox {O}}_{\rm 2} &\to {\hbox {Cu}}_{\rm 2} {\hbox {O}} & {\hbox{(1)}} \cr {\hbox {Cu}}_{\rm 2} {\hbox {O}} + {\rm 3\hbox {Cu}}_{\rm 3} {\hbox {P}} + {\rm 5\hbox {O}}_{\rm 2} &\to {\rm 7\hbox {Cu}} + {\hbox {Cu}}_{\rm 2} ({\hbox {PO}}_{\rm 4}) + {\hbox {Cu}}_{\rm 2} {\hbox {P}}_{\rm 2} {\hbox {O}}_{\rm 7}.\cr & & {\hbox{(2)}} }

Fig. 3. XRD patterns for the screen-printed and atmospherically sintered AFs characterized in Figs. 1 and 2.
Fig. 4. Deoxidizing Cu 2O leads to growth of crystalline Cu and dramatic reduction in electrical resistivity. (a) Integrated XRD peak intensity ratios for all the crystalline phases in the screen-printed AFs characterized in Figs. 1 and 2. (b) Corresponding electrical resistivity for the AFs in (a).

Reaction (3), shown below, is also likely to be accelerated at higher temperatures because, as shown in Fig. 4(a), the fraction of Cu2(PO4) increased and the fractions of both Cu and Cu 2P2O 7 decreased at $500 ^{\circ}$C TeX Source $${\rm 2\hbox {Cu}} + {\hbox {Cu}}_{\rm 2} {\hbox {P}}_{\rm 2} {\hbox {O}}_{\rm 7} + {1\over 2}\,{\hbox {O}}_{\rm 2} \to {\rm 2\hbox {Cu}}_{\rm 2} ({\hbox {PO}}_{\rm 4}). \eqno{\hbox{(3)}}$$ Furthermore, the XRD results indicate that the Cu–P–O glass, which is self-assembled and embedded into the openings of the Cu network [see Fig. 1(c)], consists of two crystalline phases: Cu2(PO4) and Cu2P2O7. As shown in Fig. 4(b), the electrical resistivity of the AF was dramatically reduced at sintering temperatures of $\hbox{450} ^\circ\hbox{C}$ and $\hbox{500}^\circ\hbox{C} (\hbox{2.40} \pm \hbox{0.58} \times \hbox{10}^{-5} \Omega\cdot\hbox{cm}$ and $\hbox{2.46} \pm \hbox{0.78} \times \hbox{10}^{-5}\,\Omega\cdot\hbox{cm}$, respectively), at which a network of crystalline metallic Cu (containing between 0.31 and 0.44 mass% P) was formed. The AFs with such electrical resistivity satisfied the criteria for backside soldering tabs/pads used in crystalline Si PVs.

SECTION IV

## DISCUSSION

The results shown in Figs. 1 4 suggest that Cu2O should deoxidize due to Cu3P oxidizing [expressed by reaction (2)] at temperatures between 400 and $450 ^{\circ}$ C. We confirmed this by performing TG-DTA measurements on the as-atomized Cu-7 mass%P alloy particles and on pure Cu3P particles (see Fig. 5), in the absence of a solvent, binder, or glass frit. The heat-flow plots for both the Cu–7 mass%P alloy and the Cu3P particles sharply peaked at $\sim 430 ^{\circ}$C [see Fig. 5(a) and (b)].

Fig. 5. Oxidizing Cu 3P probed by TG-DTA. (a) Heat flow (purple), weight (green), and temperature (blue) plots for as-atomized Cu-7 mass%P alloy particles with an average diameter of $\sim 1 \mu$m showing exothermic heat flow peak at $\sim 430 ^{\circ}$ C. (b) Corresponding heat flow, weight, and temperature plots for pure Cu3P particles with diameters of $<300 \mu$ m.

The sharp exothermic peak observed at $\sim 200 ^{\circ}$C for the Cu–7 mass%P alloy particles is thought to be caused by Cu oxidizing within the particles. That Cu rapidly oxidizes at temperatures between 200 and $250 ^{\circ}$C is an established phenomenon [14], [15], although the AF sample that contained the solvent (terpineol) and binder (ethyl cellulose) and was sintered at $300 ^{\circ}$C showed no evidence of Cu oxidizing inside the Cu–7 mass%P alloy particles [see Figs. 1(c) and 4(a)]. The terpineol is vaporized before sintering so that the binder seems to prevent Cu oxidizing at temperatures up to $300 ^{\circ}$ C, at which point, it starts to burn (as confirmed by TG-DTA). Further, efforts should, therefore, be made to optimize the properties of the atmospherically sintered Cu AFs by exploring the potential of alternative solvent, binder, and glass-frit materials.

More advanced studies on the Cu AFs developed in this paper will also provide various solutions for practical requirements of backside soldering tabs/pads used in various types of Si PVs, for example, forming a self-assembled conductive barrier between the Cu AF and Si wafer, joining tab leads to the Cu AFs, and improving the Cu AFs so that they can be formed through a sintering process at higher temperatures similar to those of aluminium paste or Ag paste sintering processes used for present crystalline Si PVs.

SECTION V

## CONCLUSION

We reported the first production of a Cu AF consisting of a novel network of crystalline metallic Cu embedded with self-assembled Cu–P–O glasses (Cu2(PO4) and Cu2P 2O7) that are grown in the sintering process using air at temperatures of $\geq 450 ^{\circ}$ C. For the Cu AF production, we developed a starting paste material composed of Cu–P alloy particles with a mixed microstructure of Cu and Cu3P phases. The Cu AF has electrical resistivity that satisfies the criteria for backside soldering tabs/pads used in crystalline Si PVs. Our study clarifies in detail the production method used for the Cu AFs and the chemical mechanism of their creation, as well as characterizing in detail the chemical and physical properties of the films. Our results may open the way to the widespread use of atmospherically sintered Cu AFs in next-generation crystalline Si PVs.

### ACKNOWLEDGMENT

The authors acknowledge the valuable scientific discussions with S. Watanabe, N. Sakaguchi, S. Miura (Hokkaido University), M. Iwamuro, S. Takeda, K. Okaniwa, M. Kim, M. Katayose (Hitachi Chemical Company, Ltd.), Y. Sawai, and N. Tanaka (Hitachi, Ltd.).

## Footnotes

T. Kato is with the Hitachi Research Laboratory, Hitachi, Ltd., Ibaraki 319-1292, Japan, and also with the Hokkaido University, Sapporo 060-8628, Japan (e-mail: takahiko.kato.ts@hitachi.com).

S. Adachi, T. Nojiri, and M. Yoshida are with the Tsukuba Research Laboratory, Hitachi Chemical Company, Ltd., Tsukuba 300-4247, Japan (e-mail: s-adachi@hitachi-chem.co.jp; t-nojiri@hitachi-chem.co.jp; masa-yoshida@hitachi-chem.co.jp).

T. Aoyagi, T. Naito, and H. Yamamoto are with the Hitachi Research Laboratory, Hitachi, Ltd., Ibaraki 319-1292, Japan (e-mail: takuya.aoyagi.ga@hitachi.com; takashi.naito.gb@hitachi.com; hiroki.yamamoto.gz@hitachi.com).

Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.

## References

No Data Available

## Cited By

No Data Available

None

## Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available