Multi has stopped at nothing.
In less than ten years, multi wafer/cell capacity increased by more than ten times (over 50GW China mainland plus around 10GW in TW). First wave was around 2008-2010, stopped a while by wide market collapse, then came second round and lasted till now. If first wave was a simple sprint of growing capacity, then the second was a mix of expansion spree, cost cut, and efficiency, in a timing fashion from early phase until present. Looking to mono, first wave was no different. But second has evidently gone through an order of cost cut and expansion spree, only efficiency part missing. This stark discrepancy around EFFICIENCY should raise some scrutiny while checking what’s more meaningful and, what’s more sustainable.
By 2015, top ten Si-c based panel makers in the world became almost 100% multi concentrated. They were mostly located in China mainland, taking over 50% of WW capacity. People were talking about PV production, Si-c based especially, become a less tech-driven cycle, and more commodity goods driven, in which cost is the only way to win out. A wide campaign of slashing cost started to sweep the whole industry.
Looking to multi casting process. The common poly charging load was 200-400kg per one run, so called size G4 and G5, with original design. But almost all furnaces were big enough to hold size G6, some even G7. G6 furnace could hold poly up to 900kg, raising furnace productivity up from 4kg/h up to 12kg/h, power consumption from 15KWh/kg down to 8KWh/h, therefore cutting casting cost down by a whopping 25%. First trial of resize might be tough. But soon people found it easier than many other prospects and, along the way crystallization was also optimized. Around 2014 there were nearly 4000 pieces multi furnaces coming to Hotzone life cycle (mainly graphite parts and insulation), most factories updated them to G6 or G7.
Consumable supplies made their fair contribution. Pushed by all sorts of investment and government subsidies, China became dominant nation of not only PV products at main supply (wafer, cell, panel, inverter), but also all sorts of consumable materials. There is a long list of consumable materials which went through from zero local maker to severely oversupply by local just in matter of 3 to 4 years. Names are quartz crucible, graphite, insulation, wire, cleaning chemical, SiC powder, paste, glass, backsheet, and on and on. For wafer factories, some materials cut price by more than half just in two years, like crucible, insulation and wires, giving large support to wafer cost down.
As expanding frenzy continued, competition started to fuel and price increasingly corrupted. Many factories started to run production at ice thin profit or even negative. Naturally differentiation strategy became next clutch. The sensation of high efficiency wafer/cells seemed to start somewhere early 2015, sweeping all along the chain with increasing heat. Cell makers did the best, as most of them were WW top makers and IPOs, with deep pocket and good engineers. And they had more appetite to innovate and stay in front. Cell efficiency, mostly multi type, went up by 10%(relative) every year since 2013. Benchmark of multi cell efficiency jumped by 0.1% absolute across the industry by every quarter and lasting that way for almost two years. No sign seen yet of any slowing down till now. A wide variety of improvement at process steps was widely applied, like, double printing, 4BB, back contact, optimized passivation, and so on.
Regarding wafer production, probably it’s made progress no less marvelous. Though cell efficiency largely comes from cell process itself, but no cell could get good efficiency out of wafer of poor crystallization. Cell factories pushed wafer suppliers vehemently for better crystal, wafer factories spared no efforts of their own. Wafer related efficiency increased by 3.0% absolute in 4 years from 15.5% to 18.5%, a nearly 15Wp output gain from one single piece of panel (60cells). HE (high efficiency) casting technology might originally come along from Taiwan institutes, but it got wild growth among Chinese factories. And great number of improvements were added on top of initial technology. From the mainstream seed-based half melting process, to cost effective full melting, from revised temperature control to various designs to tackle red zones, Chinese engineers found their way rushing wildly in getting better crystal.
And again, consumables made no less support to boost all the way up. We have to say Chinese PV industry has got the most comprehensive supply community of PV production, and with most dynamic innovation. Multi crucibles were made by quartz of better purity and density, preventing impurity from getting into wafer body. Better Si3N4 coating recipes were applied to assure better crystal forming. Robot coating machine was used to assure uniformity. With almost zero tech background, consumable factories are earnest to learn and starving to improve. Only except IP issues were rampant and foreign supplier were pushed out of business unfairly (a long list of names: crucible, Si3N4, wire, graphite, paste, connector, etc.)
That fact was, multi makers made great improvement in quality and efficiency. While in the same time mono was busy in cutting cost, and as it seemed, only.
Only they (multi) did little in slicing. Or, not enought. This is a blunder that could cost life. Process cost of multi wafer slicing did drop substantially, along with fast falling of consumable prices. If anything near to innovation, it was application of structural wire, which tightly entangles couple of wires together to carry off slurry residue more effectively, leading to higher cutting speed and productivity. Compared to disruptive move of mono toward DW slicing, multi made progress of almost negligible.
Only they were stuck by tremendous challenges. Since 2014 GCL quietly installed seven units of DW saws for its DW wafer slicing experiment. Enormous efforts and failure were experienced, before it was realized that cutting multi wafer, a stuff brittle and laden with inclusions, was a long pull trudge. In short, there are two major obstacles, one at wafer slicing process, one at cell.
1, Breakage: Multicyrstal is a composite of countless tiny mono crystals, so called grains, with jumbled crystal orientation (aka amorphous orientation).
Clustering around are stress points caused by inclusions (metals, Sic, etc.), which are prone to break itself or wire when it cuts through. Though it’s much lower cost to grow and easy to operate, the haphazard grains cause problems in many downstream process positions. The ultimate one is high breakage. It occurs at wafer side (yield down), at wire side (wire broken often and machine down in wire entanglements), at cell process (texturing, printing, etc.), and further to panel making (handling, encapsulation). In addition, this prevents wafers from going down to thinner size, an effective way to cut cost.
2 Cell Texturing. This determines acceptance of multi wafer as a qualified component for cell lines. Again due to grains and stress points, surface of DW sliced multi wafer has saw marks of much shallower and more intense than usual. The traditional alkaline process (HF/HNO3) would make less porous feature at cell surface, which leads to higher reflection of light input (more sunlight is reflected). As result, DW multi panels got efficiency lose about 0.3%, unaffordable for any cell maker.
Problems apart, the pilot production back to 2015 showed DW slicing could bring down slicing cost by near 30%, or 5% on wafer whole cost. It was a temptation big enough for any factory to push forward, given growing strain of overcapacity and slashed margin cross the industry. By 2016 the cost saving came to 50% for slicing and 10% for whole wafer, thanks to process optimization and sharp drop of DW wire price (another big story to tell). Therefore, DW multi slicing started to speed up through the chain, bringing pretty considerable profits to those achieving mass production.
There were still lots of doubts around whether or not this could repeat paradigm of mono DW slicing. For quite long time multi DW capacity grew in a quiet way, even clandestine, among a number of in-house vertical factories. With in-house QC rules, they got much less quality issues so that they could push through production quickly. Whatever problems encountered internally, they just present market nice and quality panels. And it seems nobody doubted they made more money than before and, it was huge. At least some market hints could prove this. Some DW saws were air shipped from Europe (over 8tons per machines), and the buyer multi wafer factory claimed additional fright cost can be recovered within one month. Large number of local DW saw makers, long time suffering from lack of orders (lower price, lower quality, and laden with IP scams), suddenly found themselves faced with explosion of equipment orders. DW wire supply, which was highly occupied by Japanese companies until 2015, was suddenly taken by great number of local factories, and quality improved rapidly (another IP scam). Wire price was cut by 60% within two years. By early 2017, DW wire capacity has overcome mono consumption by 100%, only waiting for orders coming from multi.
Is this march real and lasting? People was asking with same question back to 2013 when mono factories started DW adventure. It ended up with remarkable achievements. No doubt it was splendid work mono made, but they also had got some original advantages. Fox example, mono wafer is stronger, less prone to breakage. Saw marks are in better shape, almost free from texturing issues. Easy to go thinner. And so on.
Most of following events happened after 2016, with competition going intense than ever, and mono knocking on the door.
For breakage. Wire quality has improved substantially, firstly pushed by mono factories, then multi. Core wire got better to deal with wire tension at cutting. Diamond particles got stronger, more wire attaching, and better shaped.
For cell texturing, some old fashioned surface processes have been brought to the table again, like RIE (Reactive Ion Etching) and MCCE (Metal Catalyzed Chemical Etching), and input with new ideas. These so called black silicon techniques (wafer appears to be black) were applied back to 1960s by Japanese PV factories, to make cell surface more porous, less light reflective, and therefore more efficiency gains. After decades of slumbering, they are now picked up to deal with inferior saw marks. They seemed to work very well, and efficiency gains still considerable, a rise at range of 0.2-0.6% (absolute). Moreover, they provide good foundation to further advanced cell processes, like PERC, PERL, IBC, and so on. People are expecting, and working hard than ever, to get a standard perpetuating multi community:
DW slicing + Black Silicon Process + PERC
With a 18.4% of basic multi cell efficiency, plus average gain of 0.5% from black silicon, plus 0.2% from PERC, here comes 20.1%. An efficiency 0.4% behind mono (PERC), while cell cost 10-15% lower, it could be a winning point. Though multi grain feature decides that fact that multi DW slicing is less impacting than mono, it’s still a quantum leap once mass production achieved. Plus, pressure from mono increases exponentially since China government’s FRP project came along. And more panels are shutting down multi lines and switching to mono.
It’s a battle with scale growing by ten times, retreat is immense lose, failure is catastrophic.
Finally, a financial issue needs to be addressed before multi factories get to full speed on DW. It’s likely no less impacting. Data shows there were around eight thousand pieces of slurry based wire saws in China until mid 2016, all serving for multi slicing. Moving to DW would request delicate works dealing with such a huge asset. Mono guys were dealing with this issue five years ago in a way much less painful, having only few hundreds of saws in hand. They firstly tried manual converting based on old machines. Soon they found it was difficult and machine performance was unrepairably lower than original ones. The huge cost saving brought by DW having been recognized, time was then more precious than anything. They quickly decided to scrape old slurry saws ( or sell to multi factories with very low price) and purchase brand new. Reports indicate that only the top multi wafer maker G company (~20GW capacity) has got slurry saws with booking value near to $500mil, the whole industry up to $1.5B. Converting those saws would face same issues as mono did, and time consumption is enormous. Replacing with new machines would request huge fresh investment, plus huge lose of assets if putting old ones to wreckage.
What we are seeing is, in spite of all obstacles, DW slicing is marching on persistently. Newly added capacity are mostly DW based. Old machines are being converted with more efforts than ever. Good news is, a lot of improvements have been made in regard of DW saw converting in last years. Great number of engineering companies, and thousands of skilled engineers in traditional machine tool factories, have come up to serve this huge demand.
Author:Xiaodong(Bond) Wang Former Director Int'l Business Development