Solar PV cell production output for 2009 reveals an ongoing segmentation in both crystalline silicon (c-Si) and thin-film manufacturing approaches, with clear distinctions emerging regarding low cost and high efficiency. Coherent’s Finlay Colville explains these trends and the direct implications for production-line tooling down through the supply chain.
by Finlay Colville, Coherent Inc.
February 8, 2010 – Solar PV cell production output for 2009 reveals an ongoing segmentation in manufacturing approaches used within each of the crystalline silicon (c-Si) and thin-film segments. Clear distinctions are now emerging regarding low-cost and high-efficiency mass production, with direct implications for production line tooling down through the supply chain.
Going back over 20 years into annual solar PV output tables, the breakdown for cell production has nearly always been based on the absorber materials employed (c-Si, a:Si/μc-Si, CIS/CIGS, and CdS/CdTe) and further, for c-Si cells, looking at the silicon feedstock type (mono, poly, ribbon, etc.). The origin of this categorization has somewhat evolved out of the various phases of (predominantly early US-based) funded R&D within the industry; ribbon silicon for example dating back to Tyco Labs in the early 1970s. The basic assumption is that fundamental distinctions can be made regarding the efficiency, and raw material or production (mainly opex) costs.
This type of breakdown remains a valid approach to track technology trends, but obscures some subtle changes afoot since 2007, and certainly diverts attention from the not-so-subtle conclusions drawn from recent year-end PV league tables. Also, within each of the legacy categories above, there can be rather profound differences, other than the absorber material or feedstock type. These include:
- Capex costs for production lines from under $10M to greater than $100M;
- Ramp-up times to achieve volume production from a few months to over a year;
- Operator skill levels to run production equipment from routine to highly intensive;
- Qualified equipment availability from a single source to a group of competing vendors;
- Process flows requiring either off-the-shelf or custom designed tooling; and
- Varying revenue returns specific to the cell production step, rendering installed lines anything from shuttered to over utilized.
A fresh look at production trends in PV
Given this new set of dynamics, a modified analysis seems more appropriate for the production tooling supply chain today: one which better captures the issues at large, helps forecast capex levels, while at the same time providing a fresh take on solar PV production trends. To start with, c-Si cells get split, not by silicon feedstock, but into standard and high-efficiency cell concepts. High-efficiency types are those that demand alternate process flows (and by consequence new production-line tooling) compared to standard c-Si cells — analogous to the back-junction and heterojunction types within the market for some time. For thin-film, categories that have a profound impact down through the equipment supply chain are the module size, and whether manufacturing is based upon a glass/rigid substrate or a flexible/roll-to-roll (R2R) process.
Figure 1 highlights production trends using these specific criteria. Again, just to be clear, two distinct c-Si types are compared: standard (crucially, incorporating the new low-cost variant) vs. high efficiency. And for thin-film, panels nominally close to the “display” Gen 4 glass size are compared with all other variants lumped together (larger panel sizes, flexible substrates, third-generation solar cell concepts, etc.). Therefore, each comparison is largely independent of the absorber materials, layers, or number of junctions employed.
|Figure 1. Solar cell production (c-Si cells and thin-film panels) broken down by key factors directly affecting production-line equipment tooling.|
Anomalies in production output data for 2009 will no doubt continue for several months. (Debate is still ongoing over 2008 production data, ranging from under 6GW to nearly 8GW). Much of this confusion relates simply to how ‘cell production’ is interpreted and subsequently counted. Cell production (for c-Si cells) has traditionally been defined as transforming (typically boron) doped wafers into fully functioning PV cells, with the underlying premise that any given wafer can only be made into a PV cell once. But there are at least four ways to get inflated production data:
- Including numbers from pure reselling (either cell or module), for example by an agent or a distributor;
- The simple mistake of counting c-Si module output as cell output (e.g., the manufacturer has no production equipment to make cells — just modules);
- The not-so-simple issue of third party outsourced toll manufacturing more prevalent within vertically integrated players (e.g., the supplier has the tooling, can make cells, but has decided not to lose margin there for the time being); and
- The even trickier case of trying to decipher what everyone means by the phrase ‘cell production’ – this has got somewhat blurred of recent.
These are not new issues, but with a pool of cell/panel producers numbering several hundred (and growing still) and spread globally, there is simply far more scope for error than before.
High-efficiency cells offer market differentiation
Regardless of the absolute numbers for 2009 production — which will undoubtedly get tweaked in the forthcoming months, maybe years — segmentation within c-Si cell production appears to be well underway. First, there is the cost-driven approach enabled by a) the lowest route for cell capex (dropping to the $7M-$8M level for a home-built 25MW line), and b) accessing lowest manufacturing costs highlighted by the growing trend towards outsourced manufacturing. And second, there is the approach by way of the high-efficiency route. High-efficiency (or premium performance cells) fall under several categories that have been known for ages: back-junction, heterojunction, selective emitter, and a final (“other”) subset comprising back-contact (or wrap-through) cells, new front surface texturing methods, and rear side enhancements including improved surface passivation layers, localized openings with selective diffusion, or simply increasing long wavelength internal reflectivity.
Now, with nearly every top-tier c-Si cell manufacturer having announced a high-efficiency cell product during 2009, how can their relative merits be benchmarked? Look no further than process flows and equipment tooling. It is often sufficient to ask one question only: Is high efficiency enabled by new equipment and a changed process flow (a true indicator of high efficiency production)? Or are the cells in question simply the result of small tweaks to existing equipment designed to produce standard cells (somewhat analogous to sorting higher efficiency cells from a typical production run)?
Figure 2 illustrates the different types of high efficiency cells, including estimates for 2008-2009, and an approximate forecast out to 2012 (purely as a trending exercise). Looking at the figure, one might be tempted to say, “déjà vu, or just plain good sense?” Either way, selective emitters look set to dominate high efficiency c-Si cell production (again). Crucially, each cell type above requires: a different process flow to implement; the use of dedicated production equipment; and possible changes during module interconnection downstream.
|Figure 2. The different types of high-efficiency c-Si cells by share of total high-efficiency production output.|
While it is true that high-efficiency cells have not taken off as much as predicted over the last 12-18 months — due somewhat to the stronger pull towards low cost cell manufacturing of standard cells during 2009 — they remain part of almost all roadmaps outlined by leading suppliers. And the more cell production is dominated in the short term by the lowest-cost cell manufacturing approach, the more differentiation ultimately is secured for high-efficiency cell types. Of all the high-efficiency options, selective emitters are probably still the hot topic in c-Si cell manufacturing and, more recently, down through the equipment supply chain.
Selective emitters remain so popular because they provide a direct route to efficiency enhancement by simple modification to the process flow used to manufacture the market dominant (standard) cell type used today. So, let’s quickly explain this seemingly new term being used nearly everywhere. There are four basic types of selective emitter to watch out for in the press, even although the actual terminology may be hidden for competitive positioning: diffusion-masked (sometimes called double-diffusion), laser-doped (also known as LDSE), etched-back, and printed (doped) paste. In fact, sometimes new cell concepts use more than one of these steps together. Additionally, introducing selective emitters may be complemented by improvements to front finger metallization (alternatives to legacy screen printers) or improvements to the cell’s rear surface.
But wait! Wasn’t this hot news over 10 years ago? Why didn’t all this happen then?
The (laser-grooved) buried contact cell, manufactured for years by the technology leader during PV’s first growth phase (BP-Solar), was indeed the first example of a selective emitter taken into volume production. By 2003, then-Shell Solar ranked #4 (at the time) in production tables, muted adding selective emitters to their high efficiency boron-BSF cells. Researchers in Europe also highlighted their potential, with the ENEA in Italy among the first to investigate experimentally several selective emitter concepts. BP-Solar’s buried contacts represent one of several diffusion-masked variants possible, in which two (thermal) diffusion process steps (the double-diffusion) optimize phosphorous concentration levels beneath and between the front fingers. From 1995 for the next 10 years, the efficiency levels of BP-Solar’s ‘Saturn’ selective emitter cells outperformed nearly everything else in production, and in many ways they aptly set the scene for today’s resurgence in high-efficiency cells.
As a consequence, R&D effort within top-tier cell makers has never been so focused on transferring lab results to manufacturing, and installing new production tooling to enable this. The clear front-runners are those with long standing collaborative relationships with leading PV R&D institutes, or well-staffed with highly focused future technology groups. For nostalgia (but also to grasp how much effort is needed to move from standard to high efficiency), just re-read the trailblazing output from UNSW during the 1980s, or the excitement conveyed within many of BP-Solar’s commercialization announcements some 10 years later. What’s different today though is having a (maturing) market-pull, access to production equipment that can enable the new processes into mass production, and a tangible revenue return from market share gained by premium performance c-Si cells.
Turn-key production lines or in-house custom built expansion?
Directly following on from the above split into low-cost and high-efficiency c-Si cells, decision-making is evolving also with regard to turn-key production lines vs. in-house customized tool design. (Similar issues impact thin-film tooling, but are easier to track in today’s climate where one player accounts for most of the production and where there is more polarity in approaches pursued.) Returning to c-Si tooling, key questions emerge for tool suppliers:
- What is the split in production output when comparing complete turn-key lines with modular build approaches?
- Whenever you read a top-10 list of cell producers, again ask yourself about the breakdown. How many of these cells were made by turn-key lines and how many by customer owned tool design and sourcing?
- Moving forward, will production-line tooling simply end up being prescribed proactively by the cell makers themselves down to tool builders, thereby allowing the cell makers to retain their unique IP/know-how and competitiveness through the use of custom designed processes?
At first glance, and based solely on production numbers for 2009, the trend for turn-key lines probably looks set to proceed as business-as-usual. For example, new entrants are the prime candidates for turn-key line buy-in to fast track initial growth. Conversely, there is increased ownership of in-house tooling and intellectual property by the market leading players (our top-10 set) as they pursue lowest costs and highest efficiencies on their own terms.
Just to be clear though, for every tool manufacturer within the equipment supply chain today, tracking customer preferences for turn-key or in-house customized production lines has a direct impact on product strategy and market share. And the trends towards high efficiency and low cost manufacturing very much dominate proceedings here.
Getting a better handle on capex numbers
Next, by categorizing actual production output by manufacturing trends (Fig. 1), a more realistic assessment of capex can be obtained. Again, historically, capex trends were easier to track by adding up the predicted expansion plans of (fewer) well-known cell makers or established thin film players, and then factoring in projected market demand over the coming years. Recently, capex forecasting has become much harder to predict, with different levels of risk assessment applied. Not to mention an abundance of prevailing large (often unfunded) capacity announcements. To a first approximation, capex trends can be based upon actual production output today and its likely evolution over the next couple of years. By default, this removes pilot line spend, much of the VC-funded tooling pull of recent: in fact, just about anything purchased which is not contributing to real production output yet.
Finally, moving down through the supply chains (for each of the tracks outlined above) enables capex forecasting for key equipment required across the board. As an example, let’s look at laser-based tooling used within the PV industry to illustrate what emerges when doing technology-specific capex forecasting based on total available market (TAM) and actual PV production output. This comparison (Figure 3) shows estimates of annual solar PV capex for laser sources used within laser based tools. By restricting the total capex spend (large bars, or TAM) to individual production lines currently producing >10MWpa (lower bars), it is evident most of the lasers today are at R&D evaluation, inside pilot lines, or in fabs either at installation/qualification phase or simply awaiting volume production ramp up.
|Figure 3. Laser source capex trends observed within the PV industry today.|
Indeed, similar trends are observed when reviewing most equipment types called upon by the full portfolio of technologies being pursued within the PV industry today, characterized by: 1) heavy utilization of equipment by those currently running in volume production (with service revenue potential); 2) a large fraction of tooling awaiting ramp up; 3) a diverse range of applications sold to technology differentiated (more wildcard) manufacturing approaches, often at the R&D or pilot line phase.
Looking forward, watch out for new technologies emerging within cell production lines, with press releases dominated more by related volume production announcements and less on research activities or demonstrator cells for trade show promotion. Also, with reference to Fig. 3, expect increased laser based process tool deployment emerging among thin film and high-efficiency c-Si producers; especially as market dynamics create the demand necessary to convert pilot lines to full production status or to ramp up some of the dormant fabs.