Re: Yields: from what we've been told, it is as Tickertype says, that for Valence 'yields' are representative of what percentage of raw material input to the entire process ends up in 'acceptable' cells.
This is a much more stringent definition than in the liquid-electrolyte lithium-ion cell business, where only at the final stage are the completed cells tested, and a certain percentage which pass determine the 'yield.' In the Valence case, even material 'lost' at startup and due to trimming of the laminate etc. all contribute to the 'yield' calculation, so it is a more complete representation.
Now, given the nature of the Valence manufacturing process, and the types of QC in place, my guess is that the dominant controlling factor for Valence yields is in the 'front end' of the process, which is the manufacture of the films and coating of the anode/cathode 'grids,' and in the construction of the bicells by means of lamination of anode/separator/cathode films.
If the films are nearly perfectly homogeneous, with minimal defects across the surface area, and with high tolerances for thickness across the surface area, AND if the grid coatings are uniform, AND if the bicell lamination yields a 'sandwich' with uniform contact properties across the surface area, THEN you ought to be well on your way to very high final yields (possibly over 90% ?). This is my guess from an engineering perspective, because the remaining back-end processing is not something that will likely result in cell-to-cell differences in performance. Rather, the back-end processing, while extremely critical, can be performed nearly identically for any two input cells. By this I mean that any two cells with identical properties as they enter the back-end processing are extremely likely to exit the process with similarly identical properties.
Of the back-end processes, the extraction and electrolyte injection seem next most sensitive to cell-to-cell variation, but much less so than that which results from non-uniform film and lamination properties as outlined above. The sealing and resealing (after de-gassing) of the packaged cells would have binary failures (ie, a poor seal 'fails' the cell, wasting all the effort prior to that point), but it would seem reasonable that once properly adjusted, the automated equipment would be capable of almost never making bad seals. Similarly, failures occurring at the conditioning step are likely a result of an incomplete extraction process or poor films or lamination, and hence are corrected by controlling those other processes.
Given this, while Zeev is correct to point out that the 'losses' at each stage multiply together to result in the final 'yield,' on the other hand it is quite possible that certain stages of manufacture have few 'losses' (e.g., attaching electrodes, or stacking of bicells into cells, or the packaging/sealing) once the processes are designed and controlled well. The result is that the loss factors are not uniformly distributed between the 15-20 or so 'steps' of manufacture that have been identified to us before (i.e., some of the steps really do have 99.999% 'good' yields). Hence in this case it is misleading to construct a multiplication of (0.99)**20 to represent the overall yields.
It is also true that as they exercise these processes with the commercial production equipment in place, they will continually identify particularly sensitive steps in the process, and then discover ways to control them better, so that yields can continually be improved.
From the above discussion, perhaps folks can appreciate the investment of time and capital that Valence had to make in order to get control of ALL aspects of the process, even to the extent of performing ALL the chemical mixing in-house. It all begins with 'the recipe.' They had many setbacks due to not being able to rely on suppliers performing ANY part of this, in fact even having to replace suppliers at times, as they ramped up volume.
In this context, Dennis also made a very good point that since these production steps have mostly been in place for the better part of a year, this is in fact what the company has been doing: Walking, walking, walking, then jogging (forever, it seemed...), and soon now to be running steadily. So in fact many of the yield issues may have been identified and resolved, as Lev has indicated on prior conference calls. He has always been very cautious in making comments regarding current status, so if he's said they've gotten good control of their yields, I believe him. From the above, we can all appreciate how much lies behind that seemingly innocuous statement, too.
All this is just my conjecture and opinion, FWIW.
Regards, Rich
PS re: yields, and potential for 'reworking' cells: my WAG here (and that's all it is) is that IF this would even be feasible, it would be limited to the thin film materials, not any part of the process from the lamination onwards. That is, if some part of the quality control determined that cells being made were failing because of a correctable error in the films, then the material in films already made but not yet laminated could potentially be reprocessed. However, I would think it unlikely that 'failed' completed cells could be re-worked in a cost-effective manner, if at all. Again, just my speculation. |
