`· · · · · · · · · · · · · · · ·PAGES:· 1-159
`·2· · · · · · · · · · · · · · ·EXHIBITS:· See Index
`
`1
`
`·3
`· · ·UNITED STATES PATENT AND TRADEMARK OFFICE
`·4· · · · · · · ·_________________
`
`·5· ·BEFORE THE PATENT TRIAL AND APPEAL BOARD
`· · · · · · · · ·_________________
`·6
`· · · · · · ·MICRON TECHNOLOGY, INC.,
`·7· · · · · · · · · Petitioner,
`
`·8· · · · · · · · · · · v.
`
`·9· · · · · · · · ·VERVAIN, LLC,
`· · · · · · · · · ·Patent Owner
`10· · · · · · · _________________
`
`11· · · · · · · Case IPR2021-01550
`
`12· · · · · U.S. Patent No. 10,950,300
`
`13· ·Title:· LIFETIME MIXED LEVEL NON-VOLATILE
`· · · · · · · · · MEMORY SYSTEM
`14· ·________________________________________
`
`15
`
`16· · ·DEPOSITION of SUNIL P. KHATRI, PH.D.
`
`17· · · ·- CONDUCTED BY VIDEOCONFERENCE -
`
`18· · · · · ·Friday, September 2, 2022
`
`19· · · ·10:01 a.m. Central Daylight Time
`
`20
`· · · · · · ·Michelle Keegan, RMR, CRR
`21· · · · · · · · · · Lexitas
`· · · · ·508-478-9795 ~ 508-478.0595 (Fax)
`22· · · · · · ·www.LexitasLegal.com
`
`Micron Ex. 1060, p. 1
`Micron v. Vervain
`IPR2021-01550
`
`
`
`2
`
`·1· ·A P P E A R A N C E S:
`
`·2
`
`·3· · ·ORRICK, HERRINGTON & SUTCLIFFE LLP
`· · · ·By:· Jason Lang, Esq.
`·4· · ·1000 Marsh Road
`· · · ·Menlo Park, California 94025-1015
`·5· · ·Phone:· (650) 614-7400
`· · · ·Email:· jlang@orrick.com
`·6· · ·Counsel for Petitioner Micron Technology, Inc.
`
`·7
`· · · ·ORRICK, HERRINGTON & SUTCLIFFE LLP
`·8· · ·By:· Parth Sagdeo, Esq.
`· · · ·222 Berkeley Street, Suite 2000
`·9· · ·Boston, Massachusetts 02116
`· · · ·Phone:· (617) 880-1800
`10· · ·Email:· psagdeo@orrick.com
`· · · ·Counsel for Petitioner Micron Technology, Inc.
`11
`
`12· · ·MCKOOL SMITH
`· · · ·By:· Arvind Jairam, Esq.
`13· · ·1999 K Street, NW, Suite 600
`· · · ·Washington, D.C. 20006
`14· · ·Phone:· (202) 370-8300
`· · · ·Email:· ajairam@mckoolsmith.com
`15· · ·Counsel for Patent Owner Vervain, LLC
`
`16
`· · ·Also Present:
`17· · · Caylob Suarez, Video Monitor
`
`18
`
`19
`
`20
`
`21
`
`22
`
`Micron Ex. 1060, p. 2
`Micron v. Vervain
`IPR2021-01550
`
`
`
`3
`
`·1· · · · · · · · · · · ·I N D E X
`
`·2
`· · ·Deposition of:· · · · · · · · · · · · · · · · Page
`·3
`· · ·SUNIL P. KHATRI, PH.D.
`·4
`· · · · By Mr. Sagdeo· · · · · · · · · · · · · · · · ·4
`·5
`· · · · By Mr. Jairam· · · · · · · · · · · · · · · ·154
`·6
`
`·7
`· · · · · · · · · · · ·E X H I B I T S
`·8
`· · ·No.· · · · · · · · · · · · · · · · · · · · · ·Page
`·9
`· · ·Exhibit 1057· ·US Patent Application· · · · · · 80
`10· · · · · · · · · Publication, Moshayedi, Pub
`· · · · · · · · · · No US 2009/0327591 A1,
`11· · · · · · · · · 12/31/2009, 18 pages without
`· · · · · · · · · · Bates numbering
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`Micron Ex. 1060, p. 3
`Micron v. Vervain
`IPR2021-01550
`
`
`
`4
`
`·1· · · · · · · · · P R O C E E D I N G S
`
`·2· · · · · · · · ·SUNIL P. KHATRI, PH.D.,
`
`·3· ·having been satisfactorily identified and duly
`
`·4· ·sworn by the Notary Public, was examined and
`
`·5· ·testified as follows:
`
`·6· · · · · · · ·EXAMINATION BY COUNSEL FOR
`
`·7· · · · · ·PETITIONER MICRON TECHNOLOGY, INC.
`
`·8· ·BY MR. SAGDEO:
`
`·9· · · ·Q. All right.· Good morning, Dr. Khatri.
`
`10· ·Thanks for coming back for another day.
`
`11· · · · · So let's start with a few housekeeping
`
`12· ·questions.· Where are you at the moment?
`
`13· · · ·A. I am in College Station, Texas.
`
`14· · · ·Q. Is there anybody in the room with you?
`
`15· · · ·A. No.
`
`16· · · ·Q. Other than the computer that you're using
`
`17· ·for the videoconference, do you have any computers
`
`18· ·or documents or other information in your presence
`
`19· ·right now?
`
`20· · · ·A. I have other computers in the house, but
`
`21· ·they are far away.· I'm not looking at them.· They
`
`22· ·are not accessible to me.· There's this one
`
`Micron Ex. 1060, p. 4
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·desktop which I'm using which is hosting the Zoom
`
`·2· ·call.· And that's the only desktop -- that's the
`
`·3· ·only computer that's on at this point.
`
`·4· · · · · And in terms of documents, I think you
`
`5
`
`·5· ·asked --
`
`·6· · · ·Q. Yeah.
`
`·7· · · ·A. -- I have a printout which was agreed upon
`
`·8· ·I think by both sides.· I have a printout of the
`
`·9· ·'300 patent with me, my declaration for the '300
`
`10· ·patent -- these are unmarked copies -- and also
`
`11· ·the printout of Dusija.· And actually Moshayedi
`
`12· ·from last night as well.
`
`13· · · · · So I have a printout of these patents --
`
`14· ·the '300 patent as well as my declaration.· And
`
`15· ·that's about it.· None of these are marked.
`
`16· · · ·Q. Okay.· Sounds good.· So because I'm not
`
`17· ·there with you, I'll ask that you not look at any
`
`18· ·documents unless I ask you to specifically.· Does
`
`19· ·that sound okay?
`
`20· · · ·A. And if I need to look at a document, I'll
`
`21· ·let you know that I'm looking at it, if that's
`
`22· ·okay.
`
`Micron Ex. 1060, p. 5
`Micron v. Vervain
`IPR2021-01550
`
`
`
`6
`
`·1· · · ·Q. Okay.· Sounds good.· So you've been
`
`·2· ·administered an oath, I guess, or perhaps the oath
`
`·3· ·from yesterday has continued.· And so you
`
`·4· ·understand that you're under the obligation to
`
`·5· ·testify fully and truthfully.· Correct?
`
`·6· · · ·A. I do understand that.
`
`·7· · · ·Q. Is there anything that would influence
`
`·8· ·your ability to testify fully and truthfully
`
`·9· ·today?
`
`10· · · ·A. No.
`
`11· · · ·Q. Great.· So you told my colleague,
`
`12· ·Mr. Lang, yesterday about your experience with
`
`13· ·flash memory.· I'd like to drill down on that a
`
`14· ·bit.
`
`15· · · · · Could you describe your experience with
`
`16· ·flash memory controllers?
`
`17· · · ·A. You know, so with flash memory
`
`18· ·controllers, some of the work that I've done with
`
`19· ·controllers are -- it started early on with my
`
`20· ·master's thesis where I was talking about the --
`
`21· ·there was a new computer system that was called
`
`22· ·"metric," which was a multithreaded RISC
`
`Micron Ex. 1060, p. 6
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·microprocessor, which is actually today they would
`
`·2· ·call it hyperthreaded.· But it was one of the
`
`7
`
`·3· ·first.
`
`·4· · · · · And I was -- my master's thesis was to
`
`·5· ·design the memory interface and the memory
`
`·6· ·subsystem including the memory controllers.· So
`
`·7· ·that was basically, you know, design of a memory
`
`·8· ·controller that I did early on.
`
`·9· · · · · And then after that, as I was mentioning
`
`10· ·yesterday, I do teach all kinds of memory in my
`
`11· ·circuits classes.· And this is also mentioned in
`
`12· ·my declaration.
`
`13· · · · · I teach undergraduates -- I have taught
`
`14· ·many times an undergraduate circuit design class.
`
`15· ·I've also taught a graduate circuit design class.
`
`16· · · · · And in these courses, basically memory is
`
`17· ·an important part of those courses.· So we drill
`
`18· ·down to all aspects of memory in that, including
`
`19· ·the memory cell, how cells are written, how cells
`
`20· ·are erased, for example, in the case of flash, and
`
`21· ·so on.
`
`22· · · · · So this course covers flash at all levels,
`
`Micron Ex. 1060, p. 7
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·in some sense.· It covers memories at all levels,
`
`·2· ·like architectures of the memories and such as
`
`8
`
`·3· ·well.
`
`·4· · · · · And then I have another course, which is
`
`·5· ·sort of an embedded systems, which it is actually
`
`·6· ·a hardware, software codesign course.· Popularly
`
`·7· ·it might be called -- in other places it might be
`
`·8· ·called an embedded systems course.
`
`·9· · · · · And in that course I teach about flash
`
`10· ·memory and I teach about, you know, what are the
`
`11· ·design constrictions when you have flash and, for
`
`12· ·example, what the limitations of flash are based
`
`13· ·on its controllers, based on the read/write speed
`
`14· ·differences and such.
`
`15· · · · · So that's part of what I do in my
`
`16· ·coursework.
`
`17· · · · · And then also I have -- in terms of my
`
`18· ·expert witness work, I've done around the early
`
`19· ·2010-on time frame a case on flash memory, which
`
`20· ·did talk about -- which did have -- involve
`
`21· ·looking at the controllers of commercial flash
`
`22· ·chips and stuff.
`
`Micron Ex. 1060, p. 8
`Micron v. Vervain
`IPR2021-01550
`
`
`
`9
`
`·1· · · · · So this is essentially -- some of this
`
`·2· ·work now in terms of research papers, again,
`
`·3· ·memory has been something that I've focused on
`
`·4· ·extensively over my career.
`
`·5· · · · · I've had papers where we talk about
`
`·6· ·adaptive body bias to speed up memory as well as
`
`·7· ·to improve reliability.· Papers where -- this was
`
`·8· ·in the 2008, '09-on time frame, I think.
`
`·9· · · · · And then also papers where -- there was a
`
`10· ·couple of papers, I think, for the adaptive body
`
`11· ·bias.
`
`12· · · · · And then there was papers where we were
`
`13· ·talking about stability of RAMs.· Stability
`
`14· ·because, you know, like, it's important that when
`
`15· ·we think we're storing a 1, we're storing a 1.· Or
`
`16· ·we think we're storing a 0, we're storing a 0.
`
`17· · · · · So how is that ability to store data
`
`18· ·susceptible or altered by alignment of conditions,
`
`19· ·power supply, processing variations, temperature
`
`20· ·variations, all kinds of stuff like that.· So we
`
`21· ·had a paper like that.
`
`22· · · · · And let's see.· What else?· So I've done a
`
`Micron Ex. 1060, p. 9
`Micron v. Vervain
`IPR2021-01550
`
`
`
`10
`
`·1· ·lot of work on flash-based memory.
`
`·2· · · · · And this basically would be -- this has
`
`·3· ·actually culminated in some -- in a short summer
`
`·4· ·faculty fellowship at the U.S. Air Force in Rome,
`
`·5· ·New York, in 2017 and then a longer sabbatical for
`
`·6· ·six months with them, through COVID, which was
`
`·7· ·done from home, actually, but I was working with
`
`·8· ·them very closely in the last couple of years.
`
`·9· ·And again, possible research funding soon to
`
`10· ·continue that effort as well.
`
`11· · · · · So that's been a kind of a research effort
`
`12· ·which entails flash memory and the use of flash
`
`13· ·memory.
`
`14· · · · · And some of the things we're going to
`
`15· ·study in that is, it's secure flash.· I won't get
`
`16· ·into the details, but it's secure flash memory,
`
`17· ·which will then entail modifying controllers,
`
`18· ·automating and altering controllers in different
`
`19· ·new ways.
`
`20· · · · · And so the U.S. government is very
`
`21· ·interested.· I mean, the Air Force in particular,
`
`22· ·the U.S. Air Force is very interested in some of
`
`Micron Ex. 1060, p. 10
`Micron v. Vervain
`IPR2021-01550
`
`
`
`11
`
`·1· ·this work that I'm doing.
`
`·2· · · · · And this is -- I'm not sure exactly on
`
`·3· ·their side how classified it is, so I won't go
`
`·4· ·into any more details on that.
`
`·5· · · · · There's work that's going on with the
`
`·6· ·U.S. government.· So this is basically some more
`
`·7· ·work now.
`
`·8· · · · · And also -- I don't remember exactly when,
`
`·9· ·but there was -- my group had the first sequence
`
`10· ·of papers that uses flash memory in different ways
`
`11· ·to do -- to implement logic circuits and such. I
`
`12· ·had a Ph.D. student whose thesis was entirely on
`
`13· ·that.
`
`14· · · · · And then subsequently I've had work --
`
`15· ·collaborative work with the Arizona State
`
`16· ·University, with a professor there, on, again,
`
`17· ·using flash for -- in applications, in sort of
`
`18· ·embedded applications, along with traditional
`
`19· ·CMOS, to implement circuits that are fast, that
`
`20· ·are secure, easier to test, and easier to
`
`21· ·obfuscate and such, which is an important feature
`
`22· ·for security.· So we have multiple papers on
`
`Micron Ex. 1060, p. 11
`Micron v. Vervain
`IPR2021-01550
`
`
`
`12
`
`·1· ·that -- in that -- on that front.
`
`·2· · · · · And also using, again, flash memory
`
`·3· ·with -- for machine learning and have
`
`·4· ·automated artificial intelligence-type
`
`·5· ·applications using convolutional neural networks.
`
`·6· · · · · So basically, we've published a paper on
`
`·7· ·that recently.· So that gives somebody an idea of
`
`·8· ·how you would use flash cleverly, if you will, to
`
`·9· ·come up with artificial neural networks.
`
`10· · · · · There are realizations in VLSI in
`
`11· ·integrated circuits that are dramatically smaller
`
`12· ·in area than anything else we can do.· In other
`
`13· ·words, like, tens of times smaller.· I don't
`
`14· ·remember the exact ratio.· So that's some more
`
`15· ·work that we're doing.
`
`16· · · · · And also, more recently exploring
`
`17· ·collaboration with people who are working on
`
`18· ·3D NAND flash in academia.· Basically, the idea
`
`19· ·there is to use 3D NAND flash ideas and flash
`
`20· ·memory ideas and basically minimally preserve them
`
`21· ·to do other alternative applications.
`
`22· · · · · Basically, that would mean modifying the
`
`Micron Ex. 1060, p. 12
`Micron v. Vervain
`IPR2021-01550
`
`
`
`13
`
`·1· ·controller and augmenting the controller.
`
`·2· · · · · So this is basically -- and I'm sure I'm
`
`·3· ·missing something, you know.· But I'm just trying
`
`·4· ·to recall things that I've been doing on flash for
`
`·5· ·some time.
`
`·6· · · · · But it's been extensive.· Flash is, you
`
`·7· ·know, in some sense to me it's one of my favorite
`
`·8· ·technologies.
`
`·9· · · · · I've also, by the way, I've rewritten a
`
`10· ·position paper, which will be coming out in
`
`11· ·December.· It's actually a chapter in a book, in
`
`12· ·sort of a research monograph book.
`
`13· · · · · And our position is, you know, that flash
`
`14· ·is in some sense a forgotten technology in some
`
`15· ·ways.· And so basically that chapter describes
`
`16· ·flash and our work on flash and basically how it
`
`17· ·fits into the grand ecosystem of flash technology.
`
`18· · · · · So this is basically some of the work.
`
`19· ·And some of this is -- as I've mentioned, right,
`
`20· ·the work that I've been doing with the Air Force,
`
`21· ·which some of the stuff I could talk about
`
`22· ·basically does include some very fast memory
`
`Micron Ex. 1060, p. 13
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·approaches, fast memory approaches.· This has been
`
`14
`
`·2· ·actually published for dynamic grams, for
`
`·3· ·ring-based resident clock for memories.
`
`·4· · · · · That's actually shown to be better than
`
`·5· ·any of the types of memories that you have
`
`·6· ·available today, even the faster types of
`
`·7· ·memories.
`
`·8· · · · · So that's some of the work I've done with
`
`·9· ·the Air Force, but that's led into a great
`
`10· ·interest in the flash-based work and the
`
`11· ·flash-based, you know, memory work where we would
`
`12· ·have secure memory for national -- sort of
`
`13· ·national applications.· I mean, for applications
`
`14· ·that are sort of secure and secret or whatever.
`
`15· ·Whatever the government wants.· I just provide the
`
`16· ·technology.
`
`17· · · · · So this experience that I've had for
`
`18· ·memory and for flash over all these years has led
`
`19· ·to a great interest from the U.S. Air Force.· It's
`
`20· ·their Air Force research lab in New York, in Rome.
`
`21· ·It's a town called Rome in New York.
`
`22· · · · · And there may be more topics that I could
`
`Micron Ex. 1060, p. 14
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·respond to.· I mean, there might be some more
`
`·2· ·things in my -- in fact, I can look at my resume
`
`·3· ·and just see if there was anything I missed in the
`
`15
`
`·4· ·way of flash technologies.
`
`·5· · · · · Actually, not my resume.· Sorry.· I'm
`
`·6· ·looking at my -- what's it called?· My
`
`·7· ·declaration.· Right?
`
`·8· · · · · And in my declaration, I have listed --
`
`·9· ·let's see, on page 8, which would be Paragraph 13,
`
`10· ·a series of papers and technologies and such about
`
`11· ·flash.
`
`12· · · · · So one of the topics that we are working
`
`13· ·on is computing in memory.· Actually, computing in
`
`14· ·memory using flash.
`
`15· · · · · So there is basically aspects of the
`
`16· ·controller that need to be modified because this
`
`17· ·is sort of a hybrid memory now.· It's actually a
`
`18· ·memory that needs to talk to DRAM and to flash as
`
`19· ·well.
`
`20· · · · · And then I did talk about the neural
`
`21· ·networks paper.· I did talk about that.· Let's see
`
`22· ·if I missed anything else.
`
`Micron Ex. 1060, p. 15
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· · · · · I did talk about the stability of RAMs and
`
`·2· ·improving the power and speed using selective body
`
`16
`
`·3· ·bias.
`
`·4· · · · · And body bias is one of the key techniques
`
`·5· ·that's used in flash, actually, to do the
`
`·6· ·programming.· So that's basically one other way
`
`·7· ·that -- that's one of the ways that we can program
`
`·8· ·a flash memory.· So this has implications for
`
`·9· ·flash as well.
`
`10· · · · · And then a sequence of papers on
`
`11· ·flash-based circuits for logic.· So that's
`
`12· ·something that you can see on page 9 as well in
`
`13· ·this list.· And design flow for flash.· Let's see.
`
`14· ·Optimization of the necklace of a flash-based
`
`15· ·design as well.· What else?· Low-power flash
`
`16· ·circuits as well.
`
`17· · · · · So that's several of the papers.· Oh,
`
`18· ·yeah.· And then there was one more which is listed
`
`19· ·here, so I'm glad I looked at this.
`
`20· · · · · There was a sort of novel idea that we
`
`21· ·presented, which was -- it's called a ternary CAM.
`
`22· ·A ternary CAM in and of itself is a well-known
`
`Micron Ex. 1060, p. 16
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·idea from the past.· It's something that's used
`
`17
`
`·2· ·for -- in networking.
`
`·3· · · · · So for example, when you do routing of
`
`·4· ·packets on the Internet, these large routers have
`
`·5· ·something on them called a ternary CAM, which is a
`
`·6· ·fast circuit to route packets on the Internet.
`
`·7· · · · · Now, historically TCAMs were designed
`
`·8· ·using traditional logic.· But since they involve a
`
`·9· ·lot of memory, and since we could use them using
`
`10· ·flash, and since we've been working on flash
`
`11· ·extensively as well at the time and even before,
`
`12· ·so basically we came up with a way to design the
`
`13· ·ternary CAM using flash.· So that's this F TCAM
`
`14· ·that you see there.
`
`15· · · · · So it's an efficient flash-based ternary
`
`16· ·CAM design.· And in this ternary CAM, basically
`
`17· ·there's -- I think there's two or three papers
`
`18· ·that you see.
`
`19· · · · · Sorry.· There was another ternary CAM
`
`20· ·paper we had in 2004, which was -- actually, it
`
`21· ·says awarded best paper.· That should be
`
`22· ·corrected.· I think it was nominated to be awarded
`
`Micron Ex. 1060, p. 17
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·best paper.· And I believe I should verify that
`
`18
`
`·2· ·and modify this.
`
`·3· · · · · So if that's the case, this is an error in
`
`·4· ·my CV.· I do know that it was nominated to be
`
`·5· ·awarded.· I'm not sure if it finally got awarded.
`
`·6· ·That's the paper that is a 2003 paper which was
`
`·7· ·presented in Dallas that you can see on page 9.
`
`·8· · · · · In any case, the ternary CAM paper,
`
`·9· ·because it consists of -- the ternary CAM is
`
`10· ·implemented in flash.· And basically, it uses the
`
`11· ·flash memory properties, the properties -- the
`
`12· ·memory properties of flash.
`
`13· · · · · And so in this paper we did present
`
`14· ·basically a controller that would basically manage
`
`15· ·the data flow, if you will, or the data back and
`
`16· ·forth between -- there was an onboard program as
`
`17· ·well as the flash.· So we had to manage those.
`
`18· · · · · So there was this controller involved in
`
`19· ·that -- how do you say it?· In that transferring
`
`20· ·of data.
`
`21· · · · · So this was all part of what was involved
`
`22· ·in this flash ternary CAM paper.
`
`Micron Ex. 1060, p. 18
`Micron v. Vervain
`IPR2021-01550
`
`
`
`19
`
`·1· · · · · And of course, I don't have my CV in front
`
`·2· ·of me.· But, you know, there's -- there's possibly
`
`·3· ·something else I might have missed.· But mostly
`
`·4· ·this is basically what I would have -- what I've
`
`·5· ·talked about in terms of flash-based approaches.
`
`·6· · · ·Q. Okay.· So it sounds like you are certainly
`
`·7· ·very experienced with flash and controllers that
`
`·8· ·would interface with flash memory.· Would you
`
`·9· ·agree with that?
`
`10· · · ·A. I think that, you know, like I said, flash
`
`11· ·has been something that has been -- all kinds of
`
`12· ·memories have been a love of mine, if you will. I
`
`13· ·don't know if that sounds a little funny to say
`
`14· ·that.· But as a technologist, I enjoy flash. I
`
`15· ·enjoy even circuit design.
`
`16· · · · · As a technologist, I enjoy and I'd say I
`
`17· ·have expectation in many areas including flash.
`
`18· ·That's something that I could say.
`
`19· · · ·Q. Were you familiar with the design of solid
`
`20· ·state drives as of 2011?
`
`21· · · ·A. Again, in particular, like I said, I've
`
`22· ·talked about these.· I've talked about the -- one
`
`Micron Ex. 1060, p. 19
`Micron v. Vervain
`IPR2021-01550
`
`
`
`20
`
`·1· ·of the things I tell my class about the solid
`
`·2· ·state drives and sort of the evolution that
`
`·3· ·happened from spinning disk drives to solid state
`
`·4· ·drives, one of the things I teach my students --
`
`·5· ·so the short answer is yes, I was very familiar.
`
`·6· · · · · But my take on this -- and that's what I
`
`·7· ·try to teach my students -- is that in some sense
`
`·8· ·flash is a marvelous technology.
`
`·9· · · · · And why do I say this?· Because imagine
`
`10· ·this.· There's a solid state drive that's been
`
`11· ·around for tens of years.· Tons of investment in
`
`12· ·terms of technology and stuff that's been devoted
`
`13· ·to it.· Right?· And it has a fatal flaw because
`
`14· ·it's spinning, you know.· The disk is spinning.
`
`15· · · · · So if you want to read it, there's a
`
`16· ·spot -- there's a single spot where you read it.
`
`17· ·You read the disk.
`
`18· · · · · Now, if the disk happens to be not where
`
`19· ·you want it and it's, let's say, 360 degrees away,
`
`20· ·you have to wait for a whole rotation before the
`
`21· ·disk comes to your spot.· So it has built in
`
`22· ·inherent native latencies.· Right?
`
`Micron Ex. 1060, p. 20
`Micron v. Vervain
`IPR2021-01550
`
`
`
`21
`
`·1· · · · · But the marvelous thing about flash is the
`
`·2· ·observation that -- two marvelous things.· One is
`
`·3· ·that flash memory was -- like SSD-based flash
`
`·4· ·memories and solid state disk space memories, they
`
`·5· ·were designed simply to beat this slow-spinning
`
`·6· ·disk.· And they beat it by about a factor of 10.
`
`·7· ·10 or a little bit better.
`
`·8· · · · · How do they beat it by a factor of 10?
`
`·9· ·Well, despite the fact that in a solid state disk
`
`10· ·you have flash transistors, you'd see like 64 of
`
`11· ·them or 128 of them, which actually increases the
`
`12· ·resistance of the stack in a block.
`
`13· · · · · Despite that, despite that increased
`
`14· ·resistance which slows things down, flash
`
`15· ·transistors are fast enough.· The important thing
`
`16· ·is fast enough when used in that context with
`
`17· ·these large, long CD stacks.· They're fast enough
`
`18· ·to beat the SSD, which is a marvelous engineering
`
`19· ·sort of observation or an engineering design, you
`
`20· ·know.
`
`21· · · · · So the idea of doing this is -- one of the
`
`22· ·things I teach is this is a technology that
`
`Micron Ex. 1060, p. 21
`Micron v. Vervain
`IPR2021-01550
`
`
`
`22
`
`·1· ·despite using 128 transistor series, it actually
`
`·2· ·gives you a faster design than the competition.
`
`·3· · · · · And the next thing about it that's
`
`·4· ·marvelous, which is what I teach my students --
`
`·5· ·and this really inspires students when they come
`
`·6· ·and do research with me on these topics and such.
`
`·7· · · · · But the other thing that is marvelous is
`
`·8· ·flash, if you think about it, is an extremely
`
`·9· ·limiting -- limited technology.
`
`10· · · · · I mean, if you look -- and this is well
`
`11· ·agreed upon by many sources; in fact, even some of
`
`12· ·the patents that are in suit here.
`
`13· · · · · The number of writes it takes before,
`
`14· ·like, an MLC memory, is something like a thousand
`
`15· ·to 10,000-something.· SLC memory, something like
`
`16· ·maybe 100,000 plus or minus.
`
`17· · · · · So for an enterprise application,
`
`18· ·something like 100,000 application -- 100,000
`
`19· ·writes before your disk dies is completely
`
`20· ·unacceptable.· Right?
`
`21· · · · · Like, if you have things on Wall Street
`
`22· ·that are running on computers which die after
`
`Micron Ex. 1060, p. 22
`Micron v. Vervain
`IPR2021-01550
`
`
`
`23
`
`·1· ·100,000 writes, that would simply be unacceptable.
`
`·2· · · · · But again, the marvelous thing about
`
`·3· ·flash -- and this is what I teach my students when
`
`·4· ·I talk about controllers, when I talk about wear
`
`·5· ·leveling and such -- is that in flash the
`
`·6· ·observation is that, yes, even though we have a
`
`·7· ·technology that's so limited because the
`
`·8· ·transistors in flash will die after at most
`
`·9· ·100,000 writes, despite that, they're able to --
`
`10· ·in SSDs by using controllers and by clever wear
`
`11· ·leveling and by block management, they're able to
`
`12· ·actually extend the life of these to actually
`
`13· ·something that becomes acceptable for use in even
`
`14· ·enterprise, even high security, even basically
`
`15· ·fail-safe types of operations.· And that's a
`
`16· ·marvelous, you know, stroke of engineering.
`
`17· · · · · And this is basically what I teach my
`
`18· ·students when we talk about the -- like the
`
`19· ·inspiration, if you will, for flash and why flash
`
`20· ·is so important.· And this is also one of the
`
`21· ·reasons why I say flash is this sort of marvelous
`
`22· ·technology, in some ways forgotten.· Right?
`
`Micron Ex. 1060, p. 23
`Micron v. Vervain
`IPR2021-01550
`
`
`
`24
`
`·1· · · · · Imagine that if you have something that's
`
`·2· ·not very reliable in the individual transistor
`
`·3· ·level but somehow you're able to make from it an
`
`·4· ·SSD by clever block, bad block management or
`
`·5· ·garbage collection or wear leveling.
`
`·6· · · · · From that bad transistor, from that
`
`·7· ·unreliable transistor, you can build on this and
`
`·8· ·create an SSD that's just so reliable, that's
`
`·9· ·corporate -- I mean, that's enterprise-level
`
`10· ·reliable.· And that's to me, you know, the
`
`11· ·fascinating appeal of flash.
`
`12· · · · · And that's what I convey to my students.
`
`13· ·And they find it inspiring.· And that's why I have
`
`14· ·many students working on this topic with me.
`
`15· · · · · At this moment I have one, two, and a new
`
`16· ·Ph.D. student as well.· So three Ph.D. students
`
`17· ·whose work is just flash, you know.· It's just
`
`18· ·flash-based stuff.
`
`19· · · · · Depending on this Air Force award, that
`
`20· ·number might increase as well.· And it would
`
`21· ·increase and extend into secure flash-based
`
`22· ·memories.· The whole gamut of it.
`
`Micron Ex. 1060, p. 24
`Micron v. Vervain
`IPR2021-01550
`
`
`
`25
`
`·1· · · · · So it's basically the SSD -- all the way
`
`·2· ·from the SSD design down to the controllers down
`
`·3· ·to the cells down to the block management and
`
`·4· ·stuff.
`
`·5· · · · · So it would basically be a sort of· top-
`
`·6· ·down sort of research effort, I guess, you know.
`
`·7· · · ·Q. So one of the things you mentioned was how
`
`·8· ·these flash memories are composed of 128 series
`
`·9· ·transistors, they're stacked.
`
`10· · · · · Are you familiar with how these devices
`
`11· ·are manufactured, the fab process?
`
`12· · · ·A. Yeah.· So in general, in my research I
`
`13· ·don't get into the actual manufacturing, but I am
`
`14· ·familiar with the manufacture.
`
`15· · · · · There's various -- how do you say? --
`
`16· ·recipes for the flash transistor.· There's various
`
`17· ·recipes for programming as well.
`
`18· · · · · Let's talk first about the recipes for,
`
`19· ·you know, for the floating gate manufacture.
`
`20· · · · · Well, and there's something else I should
`
`21· ·remember to say about this, something called
`
`22· ·embedded flash.· So I'll keep it -- I'll make a
`
`Micron Ex. 1060, p. 25
`Micron v. Vervain
`IPR2021-01550
`
`
`
`·1· ·note of it in my head to remember to say this in
`
`26
`
`·2· ·the end.
`
`·3· · · · · But in normal flash, the lovely part of
`
`·4· ·it -- and again, this is one part of flash that
`
`·5· ·makes it very exciting for me -- is most of the
`
`·6· ·fab steps, or fabrication steps, are common with
`
`·7· ·CMOS in a normal flash technology.
`
`·8· · · · · But then just a few more steps makes a
`
`·9· ·transistor a flash transistor.· So what that --
`
`10· ·essentially in a normal transistor, you have a
`
`11· ·gate which is separating -- which is separated
`
`12· ·by -- let's see.· Let me go bottom up.
`
`13· · · · · There's a surface of the silicon that's --
`
`14· ·we call that the bulk, the bulk layer.· On top of
`
`15· ·the bulk we put a thin layer of oxide.· So this is
`
`16· ·an insulator.· Oxide is an insulator.· So we put a
`
`17· ·little oxide.· And then we put a gate on top. A
`
`18· ·gate is a conductor.
`
`19· · · · · This typically results in something called
`
`20· ·a MOSFET.· Lots of details left out because on
`
`21· ·both sides of the bulk we've got to have a
`
`22· ·diffusion, right, which is basically a source and
`
`Micron Ex. 1060, p. 26
`Micron v. Vervain
`IPR2021-01550
`
`
`
`27
`
`·1· ·a drain.
`
`·2· · · · · But assuming you had that, a source and a
`
`·3· ·drain and a bulk.· On top of the bulk, we put
`
`·4· ·something -- we put an oxide, which is the --
`
`·5· ·which is an insulator.· And then on top of the
`
`·6· ·insulator we put a conductor, which is the gate.
`
`·7· ·Right?
`
`·8· · · · · This conductor might be nitrided silicon
`
`·9· ·dioxide.· This basically makes it more conductive.
`
`10· ·Or it could be metal.· More recently we've been
`
`11· ·using metal, but it can also be nitrided silicon.
`
`12· ·So that's also called both nitrided or salicided.
`
`13· ·That's called a conductive polysilicon.
`
`14· · · · · So now once we have this, a voltage change
`
`15· ·on the gate modulates charge in the bulk and that
`
`16· ·causes conduction.· So this is a normal
`
`17· ·transistor.
`
`18· · · · · Fundamentally, a flash transistor does
`
`19· ·very little more than this.· That's why it's so
`
`20· ·beautiful to me.· You just put on top of this gate
`
`21· ·that we talked about, let's first make sure that
`
`22· ·this gate is not connected to anything.· So it's a
`
`Micron Ex. 1060, p. 27
`Micron v. Vervain
`IPR2021-01550
`
`
`
`28
`
`·1· ·floating gate.· So that's why it's called floating
`
`·2· ·gate sometimes.· Floating as in electrically
`
`·3· ·floating or electrically insulated or electrically
`
`·4· ·connected to nothing.
`
`·5· · · · · So we have this floating gate on top of
`
`·6· ·which we're going to put another layer of oxide.
`
`·7· ·On top of this other layer of oxide, we're going
`
`·8· ·to put another gate, and that's going to be called
`
`·9· ·the actual gate that we control.
`
`10· · · · · So now we have this structure which is
`
`11· ·gate -- a contacted gate or a gate we control,
`
`12· ·oxide, floating gate, oxide, bulk.
`
`13· · · · · So it's like a club sandwich now.· On the
`
`14· ·sandwich we added one more slice of bread, or
`
`15· ·something like this, and we made it a three slice
`
`16· ·of bread sandwich or something like this with
`
`17· ·oxides in between.
`
`18· · · · · Of course, the middle piece of bread is a
`
`19· ·floating piece of bread in this sort of weird
`
`20· ·analogy.
`
`21· · · · · But now this caused us to get a floating
`
`22· ·gate.· So this is basically classically what a
`
`Micron Ex. 1060, p. 28
`Micron v. Vervain
`IPR2021-01550
`
`
`
`29
`
`·1· ·floating gate is.
`
`·2· · · · · From a fabrication point of view, there's
`
`·3· ·very little extra that needs to happen from a
`
`·4· ·traditional CMOS gate or a traditional MOSFET to
`
`·5· ·actually get to the floating gate.
`
`·6· · · · · Now, of course, there's technology steps
`
`·7· ·and such involved.· So therefore, these companies
`
`·8· ·that make flash transistors generally tend to
`
`·9· ·specialize in them, the fabrication companies.
`
`10· · · · · There's several of these companies.
`
`11· ·Right?· So there's various companies in the world,
`
`12· ·some in the U.S., many in the Far East.
`
`13· · · · · So this is basically -- this is
`
`14· ·essentially the fabrication process that you have.
`
`15· · · · · Now, one type of flash is where this
`
`16· ·floating gate is actually a conducting gate.
`
`17· ·Basically, this would -- the layers now would
`
`18· ·be -- the floating gate layer would actually be
`
`19· ·salicided or nitrided poly to make it conductive.
`
`20· · · · · A different version of the fabrication of
`
`21· ·flash is actually something they call SONOS,
`
`22· ·S-O-N-O-S, which is basically sort of the same as
`
`Micron Ex. 1060, p. 29
`Micron v. Vervain
`IPR2021-01550
`
`
`
`30
`
`·1· ·the traditional flash except that this floating
`
`·2· ·layer is now an insulator.· It's known to be a
`
`·3· ·little more reliable than the traditional flash.
`
`·4· · · · · So the insulating layer that's used as a
`
`·5· ·floating layer is silicon nitride, typically.· And
`
`·6· ·that's why it has an N in the middle of it.· So
`
`·7· ·that's the nitride.· N for nitride, es