[00:00:00] Speaker 03:
Final case for argument is 23-1726, ocean semiconductor versus applied materials.

[00:01:07] Speaker 01:
Good morning, Your Honors.

[00:01:10] Speaker 01:
I'm Joseph Zito from the Devlin law firm here on behalf of the petitioner on a patent from Advanced Micro Devices that was owned by the petitioner.

[00:01:21] Speaker 01:
We have four distinct bases upon which we've asked for either reversal or remand, but I'm going to give some background and then explain how these four bases interact with each other for which we should have reversal or remand.

[00:01:37] Speaker 01:
First, starting out with the patent at issue, it's a method and an apparatus providing fault detection in an advanced process control framework, an APC framework, which is mentioned throughout the brief.

[00:01:51] Speaker 01:
This framework existed, but this patent improved the way an APC framework could be used to control a process.

[00:01:58] Speaker 01:
The process is the fabrication of a chip through the many steps that the process goes through with the entire system.

[00:02:06] Speaker 01:
And as the board found, the process did, or the patent did receive secondary considerations because of widespread acceptance.

[00:02:29] Speaker 01:
And the board recognized that but gave it not as much weight as we think it should have.

[00:02:34] Speaker 01:
Part of the reason why it didn't give it the appropriate weight, we believe, is the board did not consider that this is considered other ways to advance processing.

[00:02:44] Speaker 02:
Are you asking us to re-weigh the considered weight?

[00:02:47] Speaker 02:
Yes.

[00:02:48] Speaker 01:
One of the four basis is it.

[00:02:50] Speaker 02:
So you're asking us to re-weigh the evidence and to see if the board was correct or not?

[00:02:57] Speaker 01:
Correct.

[00:02:58] Speaker 01:
We think that it should have been given greater weight.

[00:03:00] Speaker 02:
We can't do that, right?

[00:03:05] Speaker 01:
We're not fact-finders.

[00:03:07] Speaker 01:
Well, you can review facts for clearly erroneous, but you're not fact-finders.

[00:03:13] Speaker 01:
But the reason why this relates to... That's not going to be a winning argument for you, okay?

[00:03:18] Speaker 01:
Okay, but which is why it's our fourth argument.

[00:03:21] Speaker 02:
And the reason why it's related... You should strike it, because we're not fact-finders.

[00:03:25] Speaker 02:
We're not going to reweigh the evidence that the board heard.

[00:03:29] Speaker 02:
The typical standard to review in these cases is whether the board decision is supported by substantial evidence.

[00:03:39] Speaker 01:
The weight it's given without reweighing it is relevant because it relates to the improvement in an APC, which was not disclosed in either of the references.

[00:03:49] Speaker 01:
As the petition itself says, and it's repeated on page 14 of our opening brief,

[00:03:59] Speaker 01:
Cornell, the primary reference, lacks numerous components.

[00:04:03] Speaker 01:
A fault detection unit, a data collection unit, accumulating state data, translating, sending state data, performing the predetermined action.

[00:04:12] Speaker 01:
So all of those, according to the petition itself, are lacking for Cornell.

[00:04:17] Speaker 01:
Cornell, the primary reference, relates to gathering vast quantities of data about a chip as it goes through the entire process.

[00:04:25] Speaker 01:
and then making determinations about subsequent chips and subsequent processing to have greater acceptable output.

[00:04:35] Speaker 01:
It lacks, again, according to the petition, all of those elements, which we have contended are not provided by Fox.

[00:04:45] Speaker 01:
Fox doesn't teach an advanced process control framework.

[00:04:49] Speaker 01:
And that's the third argument, which is where we proved

[00:04:52] Speaker 01:
by cross-examination of their expert, that not all the elements are taught in the two combined references.

[00:04:58] Speaker 03:
This is the part the board found was waived, right?

[00:05:01] Speaker 03:
Was waived, yeah, the board... Right, and we're reviewing that for abuse of discretion?

[00:05:06] Speaker 03:
Correct.

[00:05:07] Speaker 03:
And I guess the question I have is,

[00:05:09] Speaker 03:
Why did you wait to depose the other side's expert until after the petitioner filed its reply?

[00:05:19] Speaker 03:
Why didn't you depose the person much earlier so that anything you learned from any such deposition could have been included in your patent on your response rather than having to delay it and then try to present it in a serve reply?

[00:05:36] Speaker 01:
because additional issues were raised in the reply to the patent owner's response that prompted the need to take the deposition.

[00:05:45] Speaker 01:
And those were the issues.

[00:05:46] Speaker 03:
I mean, it seems to me this whole question of what does an APC require?

[00:05:50] Speaker 03:
What are the elements of an APC?

[00:05:52] Speaker 03:
That could have been ferreted out earlier to make this particular argument that you did not raise until your CERB reply.

[00:06:01] Speaker 01:
Right.

[00:06:01] Speaker 01:
And we raised that issue.

[00:06:02] Speaker 01:
We had that discussion before the board.

[00:06:04] Speaker 01:
And the board, in their discretion, allowed us to.

[00:06:06] Speaker 01:
Take the deposition of the expert right, but then it also issued a warning.

[00:06:10] Speaker 03:
Did it not that be very careful, you know, we're gonna Allow you to depose this individual, but be very clear that you cannot all of a sudden then introduce New arguments that could have been raised earlier That that's correct.

[00:06:29] Speaker 03:
And and then it concluded the server why in fact did that and so that's why it barred you from

[00:06:35] Speaker 03:
from relying on various statements from the deposition.

[00:06:40] Speaker 01:
Correct.

[00:06:40] Speaker 01:
That's procedurally correct.

[00:06:41] Speaker 01:
We believe that procedure was in error, but that we did not raise new issues, raise issues that were raised in the responsive brief, not issues that could have been raised in our original petitioner patent owners.

[00:06:55] Speaker 03:
I'm still unclear why you waited so long to try to depose the other side's expert.

[00:07:03] Speaker 03:
Traditionally, it happens much sooner in the process.

[00:07:07] Speaker 01:
That's correct.

[00:07:08] Speaker 03:
So I don't understand what took so long.

[00:07:14] Speaker 01:
Without getting into privileged decision making, it's difficult to answer that question.

[00:07:20] Speaker 01:
But the issue became apparent after the

[00:07:25] Speaker 01:
petitioner's brief in reply to a patent lawyer's brief.

[00:07:29] Speaker 01:
The issue became apparent that the expert was not reading an APC as it properly should have been, based upon the arguments that were presented in the responsive brief.

[00:07:39] Speaker 01:
And that's when it became apparent that we should take the deposition and find out what the expert

[00:07:46] Speaker 01:
meant when he said APC in the report, which we thought was the same thing as an APC, but in the reply brief, they said it was not, it was something different.

[00:07:55] Speaker 01:
And that's what raised the issue where we wanted to ferret that out by taking the deposition.

[00:08:03] Speaker 01:
That's how it occurred.

[00:08:07] Speaker 01:
But again, that's a third basis.

[00:08:09] Speaker 01:
I think the first two basis are the strongest ones, which are that you can't combine Fox with Cornell because they teach opposite things.

[00:08:18] Speaker 01:
Cornell is missing a number of elements because Cornell does not do the, the

[00:08:23] Speaker 01:
evaluation and data gathering is needed for an APC to control the process.

[00:08:29] Speaker 01:
Cornell does lots of data gathering on the overall chip manufacturing to make improvements in the chip manufacturing.

[00:08:37] Speaker 01:
Fox also doesn't gather that data.

[00:08:40] Speaker 01:
Fox, as we argued in our brief, takes 11 factors, combines them into, using a telling T-squared statistical analysis, converts them into one number.

[00:08:51] Speaker 01:
those are 11 factors that are in the forge in one step of the process of making a chip.

[00:08:57] Speaker 01:
They are gas pressure, temperature, a lot of factors that are then put into signal analysis and come up with one number.

[00:09:05] Speaker 01:
Fox then teaches changing each of those parameters minutely until the number comes back within range.

[00:09:13] Speaker 01:
It doesn't teach analyzing whether or not the chip was manufactured properly.

[00:09:17] Speaker 01:
It doesn't teach

[00:09:19] Speaker 01:
doesn't gather any of the data that Hotelling needs.

[00:09:23] Speaker 01:
All it does is make one step in the overall process, one portion of the overall process in one machine and a series of machines operate a little better and a little more optimally by doing a real-time adjustment of many factors and not having to look at all 11 at the same time, reducing it to one.

[00:09:42] Speaker 01:
Whereas the opposite is taught in Cornell, where you want to gather as much data as you can, look at all that overall data,

[00:09:49] Speaker 01:
to make very detailed decisions about the entire process of manufacturing ships.

[00:09:54] Speaker 01:
And so, but I give the context of cross is for advanced process control.

[00:10:04] Speaker 01:
And it was found to be to have secondary considerations, although they were not more moderate.

[00:10:10] Speaker 02:
Doesn't Cornell operate to provide defect data in the production process at many different stages of the process itself?

[00:10:27] Speaker 01:
Correct.

[00:10:27] Speaker 01:
But it does that from the abstract of Cornell.

[00:10:33] Speaker 01:
The technique allows a total clusterized wafer fabrication process and prevents wafer rejection.

[00:10:41] Speaker 01:
Reject the wafers.

[00:10:43] Speaker 01:
What CrossTips talks about is stopping, there's a problem, the APC can now, which controls the entire process.

[00:10:49] Speaker 02:
Well, you don't want to, you don't want to reject wafers.

[00:10:54] Speaker 02:
You want to, you want to save them.

[00:10:58] Speaker 02:
I mean, that's what this is about.

[00:11:00] Speaker 02:
Correct.

[00:11:00] Speaker 02:
It's about improving the process so that you have better production, less rejection.

[00:11:06] Speaker 01:
They are both in the same area of technology.

[00:11:09] Speaker 01:
We don't deny that.

[00:11:12] Speaker 01:
What the argument is is that Fox only teaches how to optimize one particular piece of equipment by taking 11 factors, reducing to one, changing the parameters, and getting that one resultant statistical number in range so you can continue going.

[00:11:27] Speaker 03:
And Fox says that's advantageous to do that.

[00:11:31] Speaker 03:
Fox says it's advantageous to convert all of this data you vacuum up and then run this multivariate analysis in order to come up with the T2 calculation.

[00:11:44] Speaker 03:
And so the idea is,

[00:11:46] Speaker 03:
Because it's advantageous, you would likewise do it in other circumstances where you're vacuuming up a lot of data and you're comparing that data to a bunch of stored data for fault detection and the semiconductor processing.

[00:12:02] Speaker 01:
Yes.

[00:12:03] Speaker 01:
Being advantageous.

[00:12:04] Speaker 03:
So in that way, that's why the multivariate analysis of Fox lends itself to the Cornell process.

[00:12:13] Speaker 01:
It does if Coronado wanted to.

[00:12:16] Speaker 01:
gather up a bunch of data, reduce it to a parameter, and move on.

[00:12:20] Speaker 01:
Cornell wants to do the opposite of that.

[00:12:24] Speaker 01:
Cornell is, again, it does not want to inject any wafers.

[00:12:28] Speaker 01:
The patent in suit rejects wafers, and it does it quickly.

[00:12:32] Speaker 01:
The APC looks at the entire system and all the machines, and if one machine is not operating properly, rejects the wafers from that machine, doesn't send them to the next machine, shuts that machine down, and makes the whole system work.

[00:12:45] Speaker 01:
Cornell waits until the wafers go all the way through the machine, looks at the rejection of sections of the wafer in minute detail, and then determines what we should change in the process the next time to get even better yield.

[00:13:00] Speaker 01:
It doesn't reject wafers or shut down machines.

[00:13:03] Speaker 01:
Fox doesn't shut down machines or reject wafers.

[00:13:05] Speaker 01:
It looks at a bunch of parameters in one operation and makes minute adjustments.

[00:13:10] Speaker 01:
That would be.

[00:13:11] Speaker 03:
I thought Fox interrupts in undesirable processing.

[00:13:17] Speaker 03:
If it.

[00:13:17] Speaker 03:
Through its fault detection.

[00:13:19] Speaker 01:
Through fault detection it, no, it readjusts.

[00:13:26] Speaker 01:
Again, reading from just the abstract.

[00:13:28] Speaker 01:
The algorithm calculates a T squared value which is then used to generate a feedback signal.

[00:13:32] Speaker 01:
If the T squared value is out of range to indicate out of tolerance condition.

[00:13:37] Speaker 01:
It then goes on to change all of the 11 parameters

[00:13:41] Speaker 01:
to get the t-square value back into tolerance and continue the processing.

[00:13:44] Speaker 01:
It doesn't reject the wafers.

[00:13:48] Speaker 01:
Fox teaches an advantageous process that could help anyone making chips.

[00:13:55] Speaker 01:
But it doesn't flow into, and that's the reason you don't combine it, because that advantage doesn't flow into the purpose of Cornell.

[00:14:03] Speaker 01:
It could help in one step of Cornell, but it doesn't flow into the purpose of Cornell.

[00:14:06] Speaker 01:
And that's what you need to be able to combine the two references that your

[00:14:10] Speaker 01:
helping Cornell achieve what's in the patent.

[00:14:15] Speaker 01:
And you're not.

[00:14:16] Speaker 01:
You might help Cornell process one process better, but you're not helping it identify chips that are out of tolerance and rejecting those and closing down machines in an overall system.

[00:14:29] Speaker 01:
So Fox can't help Cornell do that.

[00:14:32] Speaker 01:
And as a petition admits, Cornell is missing many of the steps, many of the pieces of the system necessary to do that.

[00:14:40] Speaker 01:
The non-combination, not only do they teach away from each other, gather lots of data, reduce it to simple data, adding the two together still would not accomplish the missing pieces.

[00:14:51] Speaker 01:
It would simply make, possibly make one step operate a little more efficiently by not having to reject as many links.

[00:14:58] Speaker 03:
So patents already expired, is that right?

[00:15:00] Speaker 01:
Yes.

[00:15:01] Speaker 01:
One minute left.

[00:15:19] Speaker 00:
Good morning, your honors.

[00:15:20] Speaker 00:
May it please the court?

[00:15:21] Speaker 00:
My name is Jennifer Nock on behalf of Appellee Applied Materials.

[00:15:26] Speaker 00:
In a thorough and carefully considered final written decision, the board correctly concluded that the combination of Cornell and Fox renders every challenge claim obvious.

[00:15:36] Speaker 00:
And Ocean Semiconductor's appeal relies on an improper, narrow view of obviousness and motivation to combine that is directly contrary to KSR in this court's precedent.

[00:15:48] Speaker 00:
Ocean improperly asks this court to reweigh evidence that the board thoroughly considered and relies on arguments that were waived and are not supported by the record.

[00:15:59] Speaker 00:
So we respectfully request that the board's decision be affirmed.

[00:16:03] Speaker 00:
I'll turn to motivation to combine.

[00:16:06] Speaker 00:
The board thoroughly addressed the motivations to combine Coronel and Fox at appendix pages 16 through 20.

[00:16:15] Speaker 00:
That analysis is well-grounded in the evidence and the law.

[00:16:19] Speaker 00:
As the court has recognized, Fox teaches that the t-squared value, its statistical analysis, allows for improvement, and it's a way of handling the vast amounts of data that are being collected by systems like Cornell and Fox itself.

[00:16:36] Speaker 00:
They're both doing fault detection in a semiconductor process in the same type of tool as one another.

[00:16:43] Speaker 00:
The board found that there was ample motivation to combine and that these processes were compatible and were directly, it was easy to improve Cornell using Fox's methods.

[00:17:00] Speaker 00:
Strikingly, Ocean's briefs cite to counsel's oral argument in front of the board.

[00:17:09] Speaker 00:
There's no record evidence supporting that there was any problem or that a posto would have difficulty combining the Fox method with Cornell.

[00:17:20] Speaker 00:
And there's simply no record evidence to support the arguments that counsel is making today to this court.

[00:17:30] Speaker 00:
In contrast, Dr. Hatalis' testimony at appendix pages 880 to 881

[00:17:35] Speaker 00:
discusses and explains how APOSA would have been motivated to use FOX's multivariate analysis to improve the Cornell system.

[00:17:45] Speaker 00:
And the board's analysis goes through and demonstrates that that combination of FOX and Cornell meets every element of the challenge claims.

[00:18:00] Speaker 00:
As to the APC framework,

[00:18:03] Speaker 00:
Again, substantial evidence supports the board's finding that Cornell teaches the claimed APC framework.

[00:18:10] Speaker 00:
An APC framework was a known standard component.

[00:18:14] Speaker 00:
And the arguments that Ocean Semiconductor makes on appeal were raised only for the first time in the SIR reply.

[00:18:24] Speaker 00:
They're essentially untimely claim construction arguments about what does it mean to have an APC framework.

[00:18:30] Speaker 00:
And there's simply no reason that that argument

[00:18:33] Speaker 00:
if it had been based in evidence, that those arguments could have been made in the patent owner response.

[00:18:41] Speaker 00:
I think the fact that they weren't made in that context shows how weak those arguments are.

[00:18:46] Speaker 00:
And in fact, many of them would contradict the teachings of the 402 patent itself.

[00:18:53] Speaker 00:
So there's no abuse of discretion in excluding Ocean's late certify arguments.

[00:19:06] Speaker 00:
Council touched on the weighing of the objective and evidence of non-obviousness.

[00:19:15] Speaker 00:
As the court has recognized, Ocean is simply improperly asking this court to re-weigh the evidence and to assign undue importance to what the board found was an ambiguous teaching in an article from 1997 that just

[00:19:31] Speaker 00:
simply did not outweigh the extensive evidence of obviousness that was set forth in the petition and in the testimony of Dr. Attalas.

[00:19:41] Speaker 02:
Do we have cases that say that's a question of fact the way or whether it's a question of law?

[00:19:49] Speaker 00:
Page 49 of the red brief, we cited the NRAE Warsaw orthopedic case, which states we do not reweigh the evidence on appeal, but rather determine whether substantial evidence supports the board's fact findings.

[00:20:03] Speaker 00:
In Ray Gart's side, cited at page 44 in the red brief, says the possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency's finding from being supported by substantial evidence.

[00:20:18] Speaker 03:
I guess in the end, when it comes to this secondary consideration argument,

[00:20:22] Speaker 03:
what boils down to is did the board make a reasonable reading of that long passage in that one article?

[00:20:28] Speaker 03:
And what was that one article trying to get at when it was talking about protocols and standards or something like that, right?

[00:20:34] Speaker 00:
Exactly, Your Honor.

[00:20:35] Speaker 00:
And at most, Ocean has presented what we think is an implausible reading, but an alternative reading of that passage.

[00:20:44] Speaker 00:
The board

[00:20:46] Speaker 00:
dug into all of the details and is acting as a factor.

[00:20:50] Speaker 03:
If we think the board has a reasonable reading of that long passage, then that's the end of that issue.

[00:20:55] Speaker 00:
Exactly, your honor.

[00:21:00] Speaker 00:
If there are any further questions, otherwise I will see the rest of my time.

[00:21:04] Speaker 00:
Thank you.

[00:21:05] Speaker 00:
Thank you.

[00:21:13] Speaker 01:
I just go back to the,

[00:21:16] Speaker 01:
two major points we make, that Cornell, in the petition, misses many elements.

[00:21:23] Speaker 01:
Fox does not satisfy those elements.

[00:21:26] Speaker 01:
In addition, Fox does not combine them, which should not properly be combined with Cornell, because although they both relate to the same area of art and the same technology, they are very different references, and where one collects as much data as possible to make an overall analysis to change

[00:21:44] Speaker 01:
what is done in the entire process to make less rejection of chips, because that's good.

[00:21:49] Speaker 01:
And Fox is directed towards reducing the amount of data that's collected into a single statistical number, not to make overall changes to the process, but to change those 11 parameters that it's gathering in real time

[00:22:04] Speaker 01:
to adjust the statistical number to be within range.

[00:22:07] Speaker 01:
And so they're not the same objective.

[00:22:10] Speaker 01:
They're opposite objectives.

[00:22:11] Speaker 01:
One is collect more data to make much overall greater decisions.

[00:22:15] Speaker 01:
The other is to reduce the data to a statistical number to make immediate changes.

[00:22:20] Speaker 00:
Thank you.

[00:22:21] Speaker 00:
We thank both sides.

[00:22:22] Speaker 00:
The case is submitted.