Lightwave Logic Inc (LWLG)

[tedpeele]

post 104887, Kerrisdale addressing poling being scaleable in foundries, and yields in the foundries. TIerprathy:

2. Surely part of integrating a polymer into a PDK is doing pilot runs. Don't you believe successful pilot runs (which the company has stated it's doing) at a foundry is proof poling isn't an issue at scale? If not, why not?

Kerrisdale:

Again, a pilot run is by definition not scalable. Every single wafer that goes through the process has to be poled. You can have a process that encapsulates that but that also fails most of the time, or results in heterogeneous devices across different wafers, or has poor yields, etc Our entire point with poling is to assert that individual runs cannot in any way be extrapolated to scalable production. A PDK can be developed after perfecting the process of actually producing chips, but on its own it says nothing about whether the result of that process is useful and scalable production.

Tierprathy

2. " What do you make of the CEO's claim that wafer yields have been excellent at the foundries? Doesn't that imply that foundry processes have been engaged to a sufficient level to determine poling is no longer an issue?"

Kerrisdale:

Our response: No. "yields" can mean different things in different contexts. It's possible Lebby meant that yields on wafers where the poling worked were high. Meanwhile, the poling doesn't work on all wafers, and shorts out at other times. Maybe yields were high, but device parameters were heterogeneous, which is just the same as...low yields. This fits very well within our model of a vague statement that can be interpreted in enough ways as to make it meaningless....

It's not nonsensical at all. You can have all the wafers in your chip pass testing, but then only have 1 wafer out of 10 result in homogeneous chips. You can then point to the one wafer as having excellent yield, but ignore the fact that all the other wafers were fried.

Additionally, "functionality" is a lot more fuzzy than you're making it out to be. If you have - on the same wafer - many functional chips, but they're heterogeneous on, say, a single parameter - optical loss - you can't sell those chips, because you need to be able to state their parameters on a spec sheet. If you're going to sell all those chips and they all "function" that's not good enough.

....Koog, this is a good point as well. Our point is that we're not even sure exactly in what sense LWLG is using the term "yield" because they don't attach it to the word "wafer" as is normally standard in semiconductor terminology. It's possible that they mean something else.

....Xena, we're tempted to say that this discussion proves our point in the sense that we're all arguing about what the term "yield" means and LWLG was purposely vague about it.