# EPIC Online Technology Meeting on Integrated Photonics Manufacturing with TOSIA

https://www.youtube.com/watch?v=xRfc41ReQE0

[00:00] For all right, so I hope everyone is coming.
[00:37] So let's we start in 30 seconds.
[00:40] I see Barry has already question and what is it?
[00:44] Barry, okay, never mind.
[01:26] Hey, hey.
[01:36] Heyy.
[02:00] All right, uh, good time of the day, everyone.
[02:02] This is the series of Epic online technology meetings, uh, provided to you by the European photonic industry Consortium.
[02:10] And today we talk about the integrated photonics manufacturing with our partners from Tosia and Taiwan.
[02:18] Let me remind you quickly that Epic is the largest photonic industry Association in the world.
[02:23] We include over 800 corporate members, and we're there to
[02:28] support you with technology and marketing expertise.
[02:32] so today we have a very exciting agenda um of the meeting.
[02:37] we have three speakers representing European industry and three more representing the Taiwanese manufacturing industry.
[02:45] so we'll hear the talks from alpal Loom apak opto Electronics focus light uh wave splitter Technologies spyon Tech and irch.
[02:53] I'm glad to welcome all the speakers in the room with us and also my co- moderator elain Chen so nice to have you and uh so this is the first uh meeting we have and uh together and uh hopefully we will continue in the future the same way.
[03:15] so uh without further delay I'd like to start our meeting with a keynote presentation from a very exciting speaker Alexa kof.
[03:22] he's a CEO of alphal loom and alexe will talk about the quantum do lasers step toward easier
[03:30] integration with photonic integrated circuits.
[03:33] Alex are you with us?
[03:33] Yep, yep.
[03:37] Great, so whenever you're ready.
[03:42] Yeah, I'm ready.
[03:42] Please go ahead, share your slides.
[03:47] and uh we have to stop sharing yours.
[03:53] Okay.
[03:55] Okay, thank.
[04:01] Do you see my screen now?
[04:01] Yes, very good.
[04:05] Okay, yeah, I want, thank you very much and I'm especially excited, you know, to present it for uh Taiwanese Association.
[04:09] Because I spent like several years in it, many, many years ago in Shinu and I should say War I Taiwan, you know, it's it were those who were my favorite years, you know, in my life.
[04:18] I went everywhere, you know, Kinan Sunon Lake, you know, Alisan, all of those places.
[04:30] Okay, so um, let me go to, let me see where.
[04:34] I can okay so the outline you know that indeed today uh development of cost effective and reliable fonic integration technology is like really the key to unlock the huge volume potential for photonics actually if you look into all what's happening with thei right now and I'll show a couple of slides on that that there are billions of 200 gbit per second equivalent Optical channels will be needed you know per per year and the number will keep growing and the major key here in order to really unlock this potential is to make cost effective and reliable photonic integration then I will briefly touch like technology landscape and competition how do we see it at in alone and uh uh say a few words about the company and then we'll cover the topic why we believe Quantum do technology can really support such an effective integration because of those unique features which we have compared
[05:34] to traditional indium fast fight uh cwdb technology.
[05:40] U this is this is picture you know from actually father of mod data.
[05:43] Cent Andy beckish time found over and it's a good point stated here that really like AI change everything.
[05:52] so uh the demand and it's not a bubble you know it's not a telecom bubble like 20 years ago it's real and uh the demand of the power of those models is so big that actually what we got what we do this physical layer we are not in time you know to catch up even if we keep growing like crazy like all those Market prediction show the potential growth we still you know cannot support that growth of what is envisioned for overall uh model model power.
[06:27] it's like you know silicon is improving only two times for two years and like
[06:35] fonics is it's still much worse till now.
[06:38] and uh it's interesting to see you know that this is most reputed uh like company in marketing like counting in the United States and these guys are changing their forecast every half a year you know to higher value of to high number of channels basically to high number of optical transceivers to be built.
[07:00] uh they keep growing every half a year they keep growing their predictions and now it's interesting to see where they own present uh the comparison of what they predicted in March and what they predict now in October 24.
[07:15] so uh also what we expect uh that the transition from 800 gig let's say G transceiver to 1.6 terabit will happen much faster than any prediction even from such an optimistic agency as light counting so we expect that really and
[07:36] also when we calculate this numbers we calculate not like only bottom up like like counting is doing but also top down.
[07:44] when we look into the demand of those switch manufacturers those guys who build those switches where you have to plug your transceivers and we expect that really 1.6 terabit will start to dominate the market you know by middle of 2026.
[08:00] and that's why all the Technologies we're looking at they have to support such a speed.
[08:05] uh interesting slide from Malaga uh Forum on photonics which took place about one month ago.
[08:12] now it's like pict getting famous because if you look into the left side uh in integrated uh photonics you have only 20% of value for your fonic chips and 80% comes from Packaging testing whereas for electronics for silicon you know it's vice versa so the main value is in the
[08:37] chip and packaging cantastic occupies much smaller percentage of that value.
[08:43] and now you know recently invia even bypassed Apple in the market valuation.
[08:48] and this is unbelievable things you know.
[08:51] company which is making just silicon chips you know they are now uh worth of much more than such a company as Apple.
[09:00] so um if you look today into transceiver manufacturers right and majority of them are based in China uh and uh the value inside of optical transceiver is distributed between three things.
[09:14] it's like DP photonics components and assembly whereas once Optical engine it's developed it means where you have a good coupling of lasers to Silicon photonic modulators then coupling of peak into fiber and once LP is taken over and you know like there is a like plugable using DP and
[09:40] then there is a very big push actually
[09:42] from Arista uh they demonstrated that
[09:46] transceivers without DP but with much
[09:49] better uh design of PCB with good signal
[09:52] in Tey can be driven directly from asek
[09:56] from good Asic and in this sense the big
[09:59] with the big value of overall
[10:01] transceiver world will come from laser
[10:05] silicon photonic and integration and
[10:07] today whatever people say what we really
[10:10] see only Intel has a true integration
[10:13] technology in a volume you know when
[10:15] they do wave bonding of indium phosphide
[10:18] to silicon fonics and there are tons of
[10:20] companies which are well funded and
[10:23] which are developing you know this type
[10:24] of integration of laser to Peak and
[10:28] fiber to Peak there are different
[10:30] Technologies I will skip it you all know
[10:32] it like Edge coupling vertical coupling
[10:34] essent coupling and uh there are pros
[10:37] and cons to be considered in different
[10:40] applications uh there are again many activities to do slip chip on pistal or to couple light butt coupling or microt transfer printing interesting technology of phonic wire born from Vanguard and uh also let's say if you talk about coupling of speak to fiber there are a lot of developments of vertical coupling through grating and so on uh and uh all of them well you you will have slides right uh all of them have pros and cons and obviously if one can really mount laser on Silicon fonic chip and do butt coupling then potentially it might be uh the best approach however the challenge for standard Quantum well devices it's soall reflection back if you have uh reflection back on a level ofus 25 minus 30 DB traditional indium fi DB laser will suffer from uh losing its coherency you know your line your dfb line start
[11:42] to get noisy and wider uh there are a lot of work going on around coupling fiber into Peak.
[11:48] different activities of different companies at least what really C the ey today what Intel developed using blast.
[11:58] you know coupla and holder and then connecting it with a fiber in passive alignment by the way you know I have been I have been hearing you know like about like VR for years now people talking about coupling fiber through VR.
[12:14] but at least what we know that till now you know final active alignment is still needed so somehow this issue is not resolved either either if you look into different Technologies which will compete in order to win uh this race for 200 gig per Lambda or actually now like a few biggest companies they decided to develop shes which can support 448 gbit per second pump 4 or 228 gab they.
[12:42] started to develop the silicon and now.
[12:45] there is another challenge for photonics.
[12:46] we even cannot do today in the volume.
[12:48] 200 gig per Lane but people are pushing.
[12:51] you know to go to 400 gig per Lane so.
[12:54] there are different Technologies.
[12:55] competing it's obviously silicon fonic.
[12:57] and thin fil lium near.
[13:00] uh which is like external modulation and.
[13:02] in this case you need cwdb laser to support it or array of those or individual lasers and then indan fight.
[13:09] is Emil email by the way now they claim they can do even 400 gig per Lane uh but.
[13:16] the problem you have eight lasers for dr8 transceiver and you have to couple individually each one of them and the question also temperature stability.
[13:24] because when you push the speed you you know always you push one parameter you start to sacrify on others so it's not yet clear whether uh Emil can compete with silicon photonic and maintain their shares so we hope as we build CW dfb laser and hopefully there are good.
[13:43] solutions for integrated photonics that silicon photonic plus CW dfb laser will win uh this competition and actually Market forecast are going inide the bigger and bigger share will be taken by silicon photonic or and libat plus CW lasers uh here there is a prediction of because SR8 is based on dixel GR8 based on let's say silicon photonic and CW or EML and even though it says like 500 met it's a standard but majority of the link especially to support a clusters it's like you know below 100 m and another one is f fr4 or F fr8 when you have cwdm uh lasers split it is by 20 nanometer or by 10 nomer if you talk about 10 of them so that's why I hope again that silicon photonic will win against of vixel so uh few words about inalum you
[14:44] Know Inalum is a company which is spinoff from Yof Institute and our professor J he got Nobel prize for semiconductor laser.
[14:50] Company was started in 2003 and uh really made some achievements in development of quantum technology.
[14:59] Actually we were the first team who managed to bring Quantum dots from research to commercial applications.
[15:06] And the reason you know people very often ask me now when we present our results and they say okay you have been talking about Quantum dots now for decades why the progress is so slow.
[15:18] And the reason for that that Quantum dots are very slow by themselves because of damping of relaxation isolation and you cannot modulate them fast.
[15:28] That's why Quantum dot technology had to wait until external modulation technology is well-developed and now we expect that after some peck of innovations excitement like 20 years ago and uh now productivity phase is coming.
[15:45] So, uh, in Alum, it's a company in Germany.
[15:47] You know, we run our own manufacturing facilities.
[15:49] Uh, recently we started to invest significant amount of money in order to bring CICS to high volume.
[15:57] And, uh, we will be also working with contract manufacturers, and most likely in Taiwan in order to bring lasers to high volume manufacturing.
[16:06] There are the markets, you know, there are many of them where Alland, uh, lasers in SOA can support.
[16:10] And mainly, of course, it's cloud networking, uh, clusters, and some other very interesting application of MCW lasers where narrow line DB is needed, which quantum dot can provide.
[16:24] And very high power, uh, booster optical amplifiers are also needed to substitute DFAs which are used for 1.5 micron.
[16:29] And surprisingly, 1.3 micron turned out to be better, I.E. safety, and compared to 1.5 micron.
[16:36] So here's some record performance to demonstrate that in Gite, Gite, quantum dots DFB can.
[16:45] generate up to 125 degree C you know.
[16:49] with a pretty good well plug efficiency.
[16:51] actually nobody needed you know elect.
[16:54] Electronics doesn't work at this temperature.
[16:55] but 85 105 degree you can see pretty high power unachievable uh in Indian fast fight.
[17:03] technology so though again you know for pluggable transceivers dr8.
[17:06] F4 uh nobody needs power per Channel more.
[17:08] let's say than 150 uh mwatt and this is the product which we are ramping up right now.
[17:16] uh there was a nice demonstration at last where where two major transceiver companies showed uh performance.
[17:25] showed transceivers LP it means PL uh linear drive without DSP.
[17:28] and Quantum instead of four dfb lasers for eight channels they used like two DB Quantum do lasers for all four channels.
[17:42] and the main interesting things indeed you know it's proven first time in the.
[17:46] really production mode that we don't need Optical isolator maybe it's not such a big deal you know for active alignment for assembled transceivers but once it comes to integration it becomes very crucial very crucial parameter so yeah and I will send you the link and very nice you know very nice video recorded by Jose uh about this technology so here's a measurement from our transer manufacturer partner which demonstrates that uh quantum dot can stay in like minus 15 DB reflection back and we expect it's about like 15 DB better compared to Quantum well devices uh there is an interesting work going on also uh at Intel you know they started you know they do bonding of indium fosite material to Silicon photonic and build so-called hybrid lasers but now they started to publish results claiming all those advantages of quantum dots uh and there recent Publications on that that if you B yasite Quantum wafer to your silicon
[18:48] platform you might get certain Advantage.
[18:51] though if I look into performance so far.
[18:53] it's worse compared to what they achieved with indan fight technology.
[18:56] so we also work of course on high power dvb lasers uh for El actually.
[18:59] for cap package Optics because cap package Optics uh is envisioned in two flavors uh.
[19:03] one is like having uh fully integrated lasers and other one having uh external light source where where you have let's say array of dfb and then you intermix all lambdas into different fibers in order to bring one fiber with multiple Lambda to support Peak which sits next to H and then to dist distribute the signal.
[19:28] the estimation one would need about like six to eight such external light source for one AI chip in the future.
[19:34] and then you can easily calculate how many lasers are needed based on prediction how many AI chips will be built in the world in the future.
[19:43] so if talk about arrays uh we also have
[19:48] to say something here compared to Indian fast fight technology but I should say you know the main advantage of this quantum dot technology not just Quantum dots but gas platform because ganite is much more manufacturable compared to inum fosite if you think about silicon you know silicon is here in terms of manufacturability Eng fosite here G not far from Eng fosite compared to Silicon but still you know you can do 6inch Wafers you can uh which is really much much lower cost compared to Indian fi technology and the best solution of course to have C package objects with a single fiber bringing you several wavelength instead of intermixing TB array because it gives you additional uh packaging challenges uh and losses in your AWG uh mixers and this is the work we started to do many many years ago and now we feel it's like you know like a
[20:49] second win for comp laser because
[20:51] Quantum dots can also generate multiple
[20:53] wavelengths being coherent precisely
[20:56] space because these are fa otes and then
[21:00] you have a single chip which can support
[21:01] multiple wavel
[21:04] operation real ability and uh this is a
[21:07] picture you know if I guess none of us
[21:10] remember but uh if you think about it
[21:13] you know 70 years ago those vacuum tube
[21:15] computers they couldn't finish their
[21:17] simulation before some tubal fail now if
[21:20] you look into AI clusters where people
[21:24] train the model let say you know keep
[21:26] going for two days and they cannot
[21:29] afford failure of a single link of a
[21:31] single
[21:33] link uh and estimation that laser
[21:37] failure is a root cause for about 80
[21:40] plus% of Hardware failur main failure
[21:43] today by the way uh link slab and it's
[21:46] not yet clear what is the reason uh but
[21:49] if you talk about Hardware then uh it's
[21:55] lasers Quantum dos you know this is the
[21:57] picture from my PhD work in 99 before I
[22:00] went to Taiwan in Quantum dots they have
[22:02] so-called like inito window structure
[22:05] because the typical uh reason for facet
[22:10] degradation that you have overheating on
[22:12] the facet and your bang Gap shrinks and
[22:14] then start to absorb photons which are
[22:16] coming from the uh bulk material however
[22:21] Quantum dots ID slide at longer wavel
[22:24] mean shorter bigger energy sorry smaller
[22:28] energy and that's why they're not
[22:29] absorbed on this facet of course if you
[22:32] C your laser between Quantum dots uh and
[22:35] they're much less sensitive to
[22:37] dislocations you know so if you have
[22:39] Quantum well ideal Quantum well electron
[22:41] here electron here it's all the same so
[22:45] all electrons will fill dislocations but
[22:47] in Quantum dots of course you have 3D
[22:49] localization and that's why you are not
[22:52] that sensitive to it some relability
[22:55] some preliminary study from our
[22:57] customers and pretty good results we
[22:59] don't see basically any degradation and
[23:02] these are our recent results we're
[23:04] building now our machines for relability
[23:06] so here you can see even 105 degree C uh
[23:11] statistics not big enough uh but okay
[23:14] we're building more and these lasers
[23:17] they have 135° C pin Junction
[23:20] temperature we see some burning effect
[23:22] for 5% and then it stabilized so far we
[23:26] never observed like sudden failure of
[23:28] Quantum do laser you know if you do
[23:30] something wrong they will degrade but we
[23:32] never observe like in Quantum well you
[23:34] might have you know sudden failure and
[23:36] that's why people do Bing in order to uh
[23:39] select uh instant mortality of the LA so
[23:42] far we don't see it and if you can prove
[23:44] it in a volume then once you guys build
[23:46] your Integrated Solutions because the
[23:49] challenge is okay you want to put many
[23:50] lasers in your into your silicon phonic
[23:52] chip but how do you burn it you will
[23:54] burn the whole Optical engine it might
[23:56] be too expensive you know but if your
[23:58] laser can have very very very very low
[24:02] infant mortality then situation becomes
[24:04] much easier so you Bond them to Silicon
[24:07] photonic and then you bu in your overall
[24:11] uh Peak with lasers without being scared
[24:13] that if one laser fail uh the whole
[24:16] module will fail and like few words
[24:19] about our recent really development
[24:21] which is really interesting so we
[24:22] develop semiconduct Optical amplifier
[24:24] based on Quantum dots which can support
[24:27] several mod at lambdas traveling through
[24:30] it in one regime it's similar to edfa
[24:34] like people use edfa in tform to amplify
[24:37] many modulated lines simultaneously
[24:41] which is not the case for Quantum Wells
[24:43] because you will have two strong cross
[24:44] gain modulation and lines will start
[24:47] talking to each other on another side if
[24:49] you run
[24:50] your quantum dot specially designed in
[24:54] high power in High Gain saturation mode
[24:56] you can get very high power from a
[24:59] single stripe without taper design with
[25:02] a very high efficiency even at a high
[25:04] temperature so that's why we say that
[25:06] without Quantum dots we build kind of
[25:07] semiconductor n of a d which opens way
[25:10] to many other markets and this is what I
[25:13] say about economics of gas andite you
[25:15] can use multiple wave multiple wafer
[25:18] reactors people sometimes scared are
[25:20] scared by MBE but if you know how to run
[25:22] it nothing to be scared and there are
[25:24] big systems available and to summarize
[25:27] you know we we would say that quantum
[25:30] dot lasers can make life of silicon
[25:31] photonics easier because of high
[25:34] temperature operation especially if you
[25:35] want to put your you know integrated
[25:37] engine inside of the box or close to
[25:41] your ax and temperature is going to be
[25:43] high we expect we have to prove it in a
[25:45] volume but we expect fundamentally
[25:47] better relability
[25:48] and obviously gum asite is better
[25:52] material for manufacturing it's less
[25:53] fragile it's cheaper it's available on
[25:56] uh big wafers
[25:58] uh low Alpha Factor this is a really
[26:01] cool Advantage for integration because
[26:04] otherwise how do you integrate your
[26:05] laser in your silicon photonic you don't
[26:07] want to put you know some Fair magnetic
[26:09] like Optical isolator in between and
[26:11] then single chip wdm capability where
[26:15] either this is like single SOA with wdm
[26:18] function or comp laser or you know high
[26:22] he DP arrays thank you very much for
[26:25] your
[26:26] attention wow really great talk Alex uh
[26:30] thank you very much for this uh for the
[26:33] very clear introduction into the realm
[26:35] of Integrative phonics and also this
[26:37] comprehensive uh analysis of the market
[26:41] and the boom we're all facing in this
[26:44] market it's of course also very
[26:46] encouraging to to hear the story of
[26:49] inalum that uh your Works originate
[26:51] already from uh your work together with
[26:54] the main Nobel Prize winner in the area
[26:57] and uh the the performance of the lasers
[26:59] is amazing um I'd say U the proposition
[27:04] of inal loom is very clear and But Here
[27:06] Comes the question so where uh us uh and
[27:11] the Taiwanese op electronic
[27:14] manufacturing association can help you
[27:17] uh in your
[27:19] mission we need to ramp up high volume
[27:22] in Gite number one and we are constantly
[27:25] monitoring all integration Technologies
[27:28] which one will win you know there are
[27:30] like I don't know 20 companies now
[27:32] getting funded uh for that and it varies
[27:35] from few million in Europe and from 400
[27:37] million in the United
[27:39] States the volume of Finance being
[27:43] directed to
[27:44] this and when you say ramp up the volume
[27:47] can you maybe suggest some numbers in
[27:51] your head
[27:53] for so it's again how you look at this
[27:56] and who will take the major market
[27:59] shares we don't expect vixel to compete
[28:02] for 200 gig but EML or silicon photonics
[28:06] integrated so if you take if silicon
[28:10] photonics takes 100% let's make it
[28:13] simple to calculate the numbers then I
[28:16] would say you know like very soon like
[28:18] 500 million GB lasers will be needed on
[28:21] our side you know 500 million GB lers
[28:24] wow and of course we have to compete
[28:26] with engine F fighting indan fast fight
[28:28] is doing tremendously good job you know
[28:31] like if you look into performance of CW
[28:33] of D Indian fast fight D be like four
[28:35] years ago now it's a big difference and
[28:38] it's well established companies
[28:40] multi-billion dollar companies and we're
[28:42] a little you know company in Germany
[28:45] with a big
[28:47] ambitious okay okay wow yeah that's
[28:49] fascinating all right so yeah let me
[28:52] mention to all of you that yeah so the
[28:55] our epic online meetings are very active
[28:58] so you're all very welcome to ask
[29:00] questions to our speakers I mean right
[29:02] now to alexe and also in the in the
[29:04] upcoming talks so yeah please just push
[29:07] the raise hand button at the bottom of
[29:09] the zoom window and then I can pass the
[29:12] word to you and uh so you're welcome to
[29:14] ask the questions especially I encourage
[29:16] our Taiwanese friends since it's the
[29:18] first time for you than our meeting so
[29:20] please uh uh yeah feel free to also turn
[29:23] on the video camera so we can see you
[29:25] and uh get to know you a little bit uh
[29:28] all right let me start with the question
[29:30] from Thomas Le from bizlink please you
[29:33] can
[29:41] unmute you need to unmute yes go
[29:45] ahead hey hello
[29:50] hello hello Thomas yeah please uh say
[29:54] where you come from and uh your
[29:56] question uh I'm from B link I'm located
[29:59] in Taiwan Shinu so I just heard about
[30:03] the speaker he he was also working at
[30:05] Shu yeah looks in quaning Village you
[30:09] know on K ker Road quing Village yes yes
[30:12] yes yeah yeah and so far we have some uh
[30:17] um projector with some uh Netherland uh
[30:23] companies
[30:28] okay and I'm very happy to join this uh
[30:34] webinar and maybe we can talk talk more
[30:36] later later
[30:38] on okay okay that's a good question all
[30:41] right so indeed that's another story
[30:44] about the Epic meeting so we're very
[30:46] happy to also bring you in touch uh with
[30:49] our speakers or actually anyone else in
[30:51] the room who might uh be of interest for
[30:54] you so please feel free to also reach
[30:56] out to me or direct it to the people in
[30:58] the room you can also use the the chat
[31:00] of the of this meeting to to set up the
[31:03] contact by yourself well thank you
[31:05] Thomas
[31:07] anyway yeah Wilfred would you have a
[31:09] question Wilfred no from F thank you uh
[31:13] uh uh thank you Ian Alex uh very good
[31:18] talk great nice new slides from light
[31:21] counting I haven't seen those yet very
[31:22] good um the um yes it's getting quicker
[31:26] than what we saw so I mean
[31:29] um everybody's pushing for quantum dot
[31:32] lasers you said it Andreas especially
[31:34] from Andy beheer is is making a big he
[31:37] thinks it's the next big thing next to
[31:39] lium n so you you mentioned a little bit
[31:42] the the heterogeneous integration that
[31:44] will take probably some
[31:46] time do you see when but most people
[31:50] still use free space assembly you showed
[31:52] the HG genuine and the opol link so how
[31:56] how do they integrate the system then do
[31:58] you know and I think you you mentioned
[32:00] there's no isolator this makes things
[32:02] much easier and like they if you talk to
[32:06] those big trans
[32:08] manufacturers they say no problem to do
[32:12] active alignment we have 100 machines
[32:15] yes you know in a row that's what I'm
[32:18] getting at yes keep doing keep doing it
[32:20] keep doing it there are so massive CAPIC
[32:23] investment uh which is already armatized
[32:26] and this compan is now like major trans
[32:29] manufacturers grow this year they will
[32:30] grow like 1.5 times in their revenue so
[32:35] and uh that's why today I would say you
[32:39] know it's not the major challenge for
[32:42] transceiver assembly houses to do active
[32:45] alignment and to integrate laser you
[32:47] know you put laser you put silicon
[32:49] photonic chip you put one lens you put
[32:52] Optical isolate you put another
[32:54] lens okay so you know you don't need an
[32:57] isolate dat from what I understood that
[32:59] no no no no this Advantage it's cost
[33:02] saving it's also eel saving and less
[33:05] it's less troubles and so on yeah yes
[33:08] but it's not I would say it's not a he
[33:10] app for active alignment when it comes
[33:13] not worry about active alignment so
[33:14] because I will show that later our
[33:16] Optics are all actively aligned nowadays
[33:18] with the laser but when we do designs
[33:20] now we always have to take the the the
[33:24] the big chunk of isolator into account
[33:26] and this will be gone with if they would
[33:28] use your laser is that correct yes no
[33:30] that's correct no they all like it you
[33:32] know and there is a shortage envisioned
[33:34] now forw dfb and also the funny thing
[33:38] also for optical isolator there is a
[33:39] shortage you know because only two
[33:42] companies in the world build build this
[33:43] Garett garet you know yes ah I see thank
[33:49] you all right and next question from
[33:52] Sylvie menzo from ctil photonics syv
[33:55] please yes thank you even thank you
[33:58] Alexa I'm sorry I cannot put my camera
[34:00] I'm I'm sick today I have a a very bad
[34:03] call but I'm I'm happy to to have seen
[34:06] your your very good talk alexe I have a
[34:09] question you mentioned you you have um
[34:12] it
[34:12] seems the industry is ready to to make
[34:15] the um Quantum do G Marite on 18 in
[34:19] Wafers how about the patterning of of
[34:22] the of at this wafer scale is it ready
[34:27] is there 6 in you know 8 in it's
[34:29] like 6 in yes 6 in is uh is it's a
[34:34] commercial standard now you know like
[34:37] there are here you know FaceTime
[34:40] recognition based on gum aide pixels
[34:43] like pretty big arrays and there are
[34:45] several billions of those arrays already
[34:47] produced in the
[34:49] world do do you see any when do you
[34:52] foresee the 8 inch uh supply chain
[34:55] readiness
[34:58] it's not my headache for now you know my
[35:01] headache how to ramp to six inch okay
[35:09] thanks all right so that's uh all for
[35:13] you so no more
[35:15] questions in this case I suggest we
[35:18] thank Alexi again and looking forward
[35:22] again to set up set up your connection
[35:24] with the Thomas who spoke to us earlier
[35:27] and also whoever else is interested in
[35:30] uh scaling up the uh amazing uh CW quum
[35:34] do lasers uh please feel free to reach
[35:37] out to me or alexe directly uh with this
[35:40] um I'd like to pass the word to the
[35:42] first Speaker from our counterpart
[35:45] organization TOA and uh the first
[35:48] representative of Taiwan in our meeting
[35:50] will be Ted KU the director of apak of
[35:53] electronics and he will talk about the
[35:55] apac's Silicon photonic application
[35:58] solution then uh yes uh can you see the
[36:03] my screen not yet please share the
[36:05] slides and you'll be good to go can you
[36:09] see it can you can you see it no no okay
[36:19] sure okay now it's seem to be
[36:22] starting yeah okay uh today they are to
[36:26] introduce the AP photon is the our
[36:30] technolog uh for the related
[36:33] application so all quick to to to give
[36:37] you the idea about our company then talk
[36:40] more about our the cor technology then I
[36:44] trying to to give an idea for the the
[36:47] CPU the fiber cering solution then we
[36:51] looking for the the partner to do the
[36:53] development together then give you the
[36:55] summary today
[36:58] okay so the AP after we fund it from the
[37:01] year 1998 so more than the 20 years the
[37:05] AR call uh transceiver shiing to the
[37:08] market and we are the IPO company in
[37:11] Taiwan so with the H the financial
[37:14] performance to support the longterm the
[37:16] business because the ult communication
[37:19] the generation to generation investment
[37:21] isqu quite huge and the third one is the
[37:24] quick talk about our the product
[37:26] portfolio we support a very
[37:28] comprehensive transer from the very low
[37:31] data rate to the 800 and right now we
[37:34] are developing the 1 point6 Terra for
[37:37] all kinds of the application no matter
[37:39] the the data center the switch or the
[37:42] the mobile front even though the F to
[37:44] the
[37:45] home then today because I will talk more
[37:48] about the SEC photon is related then we
[37:50] have the techage platform to support the
[37:54] this
[37:56] application so this I uh give you the
[37:59] introduction about our the technology of
[38:02] from the design to the manufacturing as
[38:04] the package and all automation related
[38:08] of course for the appal trans design we
[38:11] need to have cability from all kinds of
[38:13] the
[38:14] simulation uh like the highs speeed
[38:17] secetary Optical system the Thal and the
[38:20] the stress the simulation and of course
[38:22] the vitation
[38:24] platform then for the materal wise
[38:28] there so many OSI be used in the in the
[38:32] in ulture receiver so we have the no how
[38:36] how to pick up the roosy how to do the
[38:38] validation and of course the uptic
[38:40] related no the the active or the pive
[38:44] components and high high speed the sub
[38:47] TR and of course um the O
[38:51] composition then there some software
[38:55] funware and shim is compared the the
[38:58] development and of course the for the
[39:00] reability validation platform then for
[39:03] the package wise uh we have the internal
[39:06] the T and uity bonding capability and
[39:10] high preced the tri the package te
[39:15] knowledge then for for the automation of
[39:17] course I got the D bang y bang Aigner
[39:21] the burning and of course the
[39:23] testing then in the left side I give the
[39:26] more idea about our the second Photon te
[39:29] platform because we we have experience
[39:33] to development 800 dr8 the we gr uh
[39:38] based CH photonis the transceiver then
[39:41] for the 400 G dr4 we pick up the H CPO
[39:45] in the the pick so basing on such kind
[39:48] of the transceiver PL transceiver design
[39:52] experience so our technolog platform to
[39:55] support Second phonus will including
[39:58] like again for the artical system design
[40:01] then the precedure the DI all kinds of
[40:04] dice package then the paive the fa
[40:06] attachment uh to the V gr pick and ative
[40:10] or uh T transmitter side and the reer
[40:14] side cine and of course the index MOS
[40:18] system and of course for the CER is the
[40:22] higher power consumption with the the
[40:24] heat so the good theral management
[40:27] capability for sh out laser and of
[40:30] course we also can handle for the remote
[40:32] laser D with the pm5 measurement then we
[40:35] also have the do REM the C article
[40:39] module to support the the CPU
[40:45] application okay then this page uh there
[40:48] some high level idea about our the CPU
[40:52] or fi cou solution because CPU is very
[40:55] hot in Taiwan right now so our we come
[40:57] out our idea and looking for the some
[41:01] partnership at uh the feature of course
[41:05] for CPU the most people we we talking we
[41:08] talking with are looking for a detach of
[41:10] alol
[41:11] interconnection and how to have a such
[41:15] such kind of the mechanic design the
[41:17] most important uh is the for the very
[41:21] the the C window then we design the
[41:24] whole artical system then to p uh basing
[41:29] on if we using the the MCH the the Mori
[41:33] diameter is the the track here uh our do
[41:38] some simulation then we can handle the
[41:42] the window around the 20 micrometer to
[41:45] this the to DV the C Windows to support
[41:49] for detachable the mchan design then for
[41:53] the peck uh we can support the vertical
[41:56] cing so you can see it's the the pattern
[41:59] something the structure like a TS TMC is
[42:02] a vertical Caroline and in the right
[42:04] button sign we can also support H
[42:07] Caroline with the V group such kind of
[42:09] the the peck then for the Caroline
[42:13] process our the system our idea is to do
[42:17] the P alignment and for the Torrance we
[42:22] we we need the Bing accur to P the 2
[42:25] micrometer so we can support the daugh
[42:27] the the cine uh the better Carin do uh
[42:30] the better car efficient the da Carin do
[42:34] then for the con wise because the right
[42:37] now our idea is to come out something
[42:40] like the mini Mt or Mt like the me
[42:43] mechanical structure but of course for
[42:46] if needing more channels we can discuss
[42:49] with our partner to come out the the the
[42:52] potential the architecture yeah so the
[42:55] purpose today first is to looking for
[42:58] the pner sh for the JDM so uh our
[43:02] partner wishing to to to pick up the the
[43:04] second phon is the the design company
[43:07] and the also the package
[43:10] company okay so I do a summary here uh
[43:14] as our introduction we have ready to go
[43:16] the C photon is the the package and the
[43:20] testing capability and we have the the
[43:23] sprin recer for the second phon is the
[43:25] pl board the transceiver
[43:28] then the second scene is the the the the
[43:31] the major the the Target today to to
[43:34] looking for the partnership from the pck
[43:37] or the allet then the potential uh need
[43:40] to discussion uh the subject for import
[43:43] for the pick we need to discuss what MD
[43:46] from you the way gu design and because
[43:49] our Target is to do this some page
[43:51] alignment for our dance uh then uh we
[43:55] need to discuss the the Ping P design uh
[43:58] on the you the peck then for the allet
[44:01] is the the high level for the example
[44:03] the simulation we're looking for the
[44:06] below the two micrometer the post shift
[44:08] acute sponded I think it should be the
[44:10] quite mature in in the market but anyway
[44:13] if we have a partner we can to discuss
[44:15] how to work together then for our our
[44:19] our our portion of course we do the
[44:21] design the L fa solution design and also
[44:26] the detect the mechanical the solution
[44:29] design yeah then if you have the any
[44:32] question you can ask me or you can
[44:33] contact I show the my email in the
[44:36] St yeah it's the my presentation
[44:41] today all right this was Ted K the
[44:44] director of apak of tectronics thank you
[44:46] very much Ted for a very clear
[44:49] presentation and uh also for summarizing
[44:52] your needs I mean for example exactly
[44:55] thank you very much wilred I was already
[44:57] about to PO you about the optical design
[44:59] system so please go ahead with your
[45:00] question so what what you were showing
[45:02] in your request for CPO I think we might
[45:05] have a solution for you that's exactly
[45:07] what we're working on now and I will
[45:09] show it in my presentation so uh so do
[45:12] you have more specifics on on
[45:15] on uh on exactly what you need there uh
[45:19] of course I can I can wait in for your
[45:21] plation but basically we already contact
[45:23] uh you the the China uh the China team
[45:26] and your
[45:27] yeah right okay thank you
[45:30] yeah thank you okay okay so the miracle
[45:34] has already happened all right and on
[45:36] the packaging and test side I expect you
[45:39] can you wait for the fen Tech
[45:40] presentation as well so y Chen will
[45:43] present their capabilities I'm sure you
[45:46] will find some interesting overlaps
[45:48] there as well yeah
[45:50] right yeah yeah yeah okay good all right
[45:55] any more questions from the room
[45:59] okay I have a short questions uh just
[46:00] like you mentioned uh this is very
[46:02] interesting presentation from um uh AEK
[46:05] actually I guess I have visited apek uh
[46:07] in China before uh like more than 10
[46:10] years before and uh we uh were really
[46:13] interested uh to be uh to see the
[46:16] potential coroporation with uh um Apec
[46:20] uh and later I will present our
[46:21] Solutions and so that we can maybe have
[46:24] more discussion uh offline and by the
[46:26] way I'll be also in Taiwan uh in uh
[46:29] November so maybe we can also uh and
[46:32] also in Milan China so maybe we can find
[46:33] a time to also organize a visit yeah
[46:37] sure welcome you to visit
[46:39] us okay that's it so that we found all
[46:43] the solutions for you right here all
[46:45] right thanks again for your talk and uh
[46:47] I suggest we pass to the next
[46:49] presentation as already mentioned by
[46:51] Wilfred noil the chief scientist of
[46:53] focus light and Wilfred will talk about
[46:55] the micro Optics as enabling technology
[46:58] for
[46:59] Datacom well FR whenever you're ready
[47:02] thank
[47:04] you
[47:11] um
[47:15] so yeah sorry thank you yeah thank you
[47:19] Ian and and epic for inviting me to this
[47:23] event and uh and I'm very happy to
[47:26] present what what do here so uh
[47:29] you you might know focus light already
[47:32] as a laser company and recently focus
[47:34] light has uh has extended um um our
[47:39] activities so we actually have now two
[47:41] locations in Switzerland one in Zurich
[47:43] and one in in in Nel still of course we
[47:46] have several facilities uh and and and
[47:51] um manufacturing sites in China with our
[47:54] headquarters in Chia and uh at and and
[47:57] dongan is mostly the the um the backend
[48:01] processing and
[48:03] um and recently we also uh acquired
[48:07] heptagon which is also now in Singapore
[48:10] so um having said that so we are having
[48:14] now these four pillars basically in our
[48:17] company so the one is what we call
[48:19] Photon generation this is assembly and
[48:21] stacking and building of sub modules for
[48:24] lasers so we're making a lot of laser
[48:27] uh prepackaged lasers that can as
[48:30] modules as subm modules that then
[48:32] become uh uh crucial Parts in bigger
[48:35] systems then what we call Photon control
[48:37] is everything that has to do with micro
[48:40] Optics and Optics in general so we make
[48:42] uh we design our Optics ourself that can
[48:45] be then combined with the lasers but we
[48:47] also sell that of course separately so
[48:50] each of these column can be can be uh is
[48:53] an individual business unit or or or
[48:57] business for us then we do complete
[49:00] Solutions mostly for the semiconductor
[49:02] industry so we make alignment lasers for
[49:05] for the um how do you say for the uh for
[49:08] the kneeling in the display industry and
[49:11] uh and last but not least is our new
[49:13] column it's the um what we call the
[49:17] global uh uh the global protronics
[49:21] Foundry where we offer Foundry Services
[49:26] based B on imprint technology and and
[49:29] and vacuum injection molding so here we
[49:31] built complete micro Optics so for
[49:34] example we make little little stacked
[49:37] microl lenses so all wafer level Optics
[49:39] that are very interesting and competive
[49:41] to to classical injection uh plastic
[49:45] molding lenses so we we we this goes
[49:47] into ey tracking and systems like this
[49:50] so so this is a commercial Foundry with
[49:54] super high volume for uh consumer
[49:57] applications also what we do there is
[49:59] our Automotive lighting uh so we are
[50:02] also in automotive
[50:04] lighting uh
[50:06] headlamps and and also in the so-called
[50:09] welcome or light carpet that you see
[50:11] sometimes for
[50:13] projectors um what do we do for this um
[50:17] industry that we targeting today or
[50:20] looking at today so silicon photonics so
[50:22] similar to the previous slides um or
[50:25] previous presentation we saw so we also
[50:28] offer a complete design work and we're
[50:30] very happy to hear now that Quantum do
[50:32] lasers don't need an isolat anymore that
[50:34] makes everything much easier so so we do
[50:37] microlens designs so we design the micr
[50:39] lens so what is blue here that's our
[50:41] part what's gray that we typically take
[50:44] from the client and then we find the
[50:46] best possible solution we also do
[50:48] tolerance simulation and then we go into
[50:49] manufacturing so we also like the
[50:51] previous two speakers we are also wafer
[50:53] level we do wafer level Optics and so we
[50:56] work on a wafer and and make little uh
[51:01] polymer uh halfes on there that are then
[51:04] by by matching technology transferred
[51:06] into the substrate so on the end you
[51:08] have a
[51:08] monolithic uh microoptical element like
[51:11] here this is a monolith monolithic
[51:13] silicon lens so and then we dies to the
[51:16] size that the client
[51:18] needs um now we saw that already before
[51:22] uh for in the keynote I just want to
[51:24] mention so this is the big thing now the
[51:26] the AI clusters but there's a lot still
[51:30] uh going on in the classical Data Center
[51:33] Technology and the Telecom industry as a
[51:35] matter of fact so the the there was
[51:38] recently a presentation from Sienna uh
[51:41] where they showed how they actually
[51:44] going into the edge now to 2 to 10 kilm
[51:46] that's very new that means now you can
[51:49] even get coherent Optics Cent pluggable
[51:52] transceivers at the edge of the um at
[51:56] the edge of the data center they're not
[51:58] going inside the data centers but this
[52:00] is being discussed at well but they
[52:02] already getting very close which is a
[52:04] completely new thing uh especially for
[52:06] 800 gy LR and
[52:09] ZR okay um similar as before in the
[52:13] keynote the new players now in Datacom
[52:17] so if you would have looked at this
[52:19] configuration for data Telecom and
[52:21] consumer a few years ago it was
[52:23] completely different so it's now very
[52:25] dominated by by the Asian market in
[52:27] Datacom where Telecom is still
[52:29] classically spread across the globe uh
[52:32] now inom coherent and Source photonics
[52:35] jab they're all e opol link they're all
[52:37] produce heavily in Asia so this is where
[52:41] uh where we are focusing obviously now
[52:42] our Market
[52:44] on and uh the volumes are are
[52:47] mindblowing huh we all heard that
[52:49] several times already so what do we do
[52:51] in this domain again so we do for fibers
[52:53] we do the coupling lasers uh the
[52:56] coupling lenses uh and also for the
[52:59] lasers we do the coupling optics for the
[53:00] cation here so we do either fuse silica
[53:04] lenses that are then index matched to
[53:05] the typical glass fiber or single mode
[53:08] fiber and the Silicon lens is better for
[53:10] high numerical apertures pick or laser
[53:13] diets so when you have a mode field here
[53:15] of one between one and six microns which
[53:17] is typically so lasers are typically
[53:19] with one and three microns uh and pics
[53:21] are typically between three and eight
[53:23] microns so we can have silicon lenses
[53:26] that can handle these High um numerical
[53:30] Divergence numerical aperture and
[53:32] Divergence so what we can do here is
[53:34] also making arrays so that are then uh
[53:38] CTE match to the pick so if you have a
[53:40] pick with up to 16 channels we see now
[53:42] 16 to 20 channels coming more and more
[53:45] often for for the 800 gig 1.6 Tera we
[53:48] see 20 channels now coming and that's
[53:51] where you need um silicon microlens
[53:54] arrays uh that are CTE match to the pick
[53:58] um yeah I might just mention that uh we
[54:02] are going from free space now into a
[54:04] silicon photonic so we still have a lot
[54:06] of requests here for the for the 400 200
[54:10] 400g it's still going so it's not
[54:12] because 800g and 1.6 Tera is coming that
[54:14] the 400g and 200g are not there anymore
[54:17] there there's still a lot of demand for
[54:18] this and we make typically these
[54:21] microlens arrays for the cimation so
[54:24] these can either be singlets or or
[54:25] arrays but the the trend is very
[54:28] important going now to make the
[54:30] multiplexes and the modulators and
[54:31] silicon photonics was also nicely
[54:34] explained in the keynote um so that
[54:36] means you need Optics here to either
[54:39] couple the light to a to a detector but
[54:42] typically what's more often the case
[54:43] it's laser to laser to pick coupling and
[54:47] and um and uh or pick to fiber coupling
[54:52] so this is how our micro Optics are used
[54:55] then so you have some type of pick with
[54:57] a lot of functionality here and then we
[55:00] make different types of microl lens
[55:01] designs and and and shapes and form
[55:06] factors that are then adapted to what
[55:08] the clients need so we are again here on
[55:10] the laser diet side we help the clients
[55:12] with the design we get all the input
[55:15] information and then we do the the
[55:18] refocusing Optics here for the laser
[55:21] diet uh sometimes these are very high
[55:23] Ina systems you have very very large
[55:26] angles here 60° or so and then on the
[55:29] other side of the pick typically there
[55:31] is a pluggable connector so you want to
[55:33] have a cimation lens so we do that as
[55:35] well and then we also do the lenses that
[55:37] go in front of the single mode fiber so
[55:39] we offer uh these three
[55:42] lenses and um and we are uh able to do
[55:48] um to help you here to find the best
[55:51] solution uh so what the new development
[55:54] is is when you have an edge emitter so
[55:56] people kind of step away from the
[55:58] grading coupler now these days for
[56:00] different reasons um so there's a lot of
[56:03] silicon nitr right now spot size
[56:05] converters available and also uh you
[56:08] want to go broadband and for this we
[56:10] offer this this solution now with a
[56:13] microprism so we have this little
[56:15] integrated lens with a prism and this
[56:18] vertical sidewall then that you can
[56:20] basically attach to the end of your pick
[56:22] and then uh you use some off the shelves
[56:25] single mode fiber connector so we I can
[56:27] either cimate here or generate a certain
[56:30] beam diameter according to what you want
[56:32] but what is more coming now we saw it
[56:34] also earlier today is you have when you
[56:36] make a pick you typically have some type
[56:38] of trench here already and then we what
[56:41] we're working on now actually already
[56:43] with with one Foundry we're working on
[56:45] the um uh what we call the drop him
[56:48] microprism so you can push this micro uh
[56:52] prism inside this Tren so you
[56:54] co-designed that with a pig manufacturer
[56:56] and where we where we then can directly
[56:59] cimate the beam that comes out here and
[57:03] um what we see more and more often
[57:06] nicely that we saw this pattern already
[57:08] before so we're also quoting this
[57:10] pattern from tsmc what tsmc is offering
[57:12] now uh this co- package optics for 2
[57:15] energy per line um where you have a pick
[57:19] and then you have some type of silicon
[57:21] carrier the light goes through the back
[57:22] of the Silicon carrier there's some type
[57:24] of integrated lens there and then what
[57:27] happens actually here the light comes
[57:28] out as a as a very well defined angle
[57:31] and we see that now more and more often
[57:32] from all types of clients where tsmc is
[57:36] basically defining um the optical
[57:39] interface that comes above so we most of
[57:41] our Optics that we design these days are
[57:44] are to match what tsmc is doing here so
[57:48] uh so tsmc offers some type of output
[57:51] here and and we do an equivalent output
[57:54] for other silicon photonics
[57:56] uh uh systems or or we just design the
[57:59] optical coletor that's based on the tsmc
[58:01] design so we are very flexible
[58:04] so and uh so what we do there exactly is
[58:09] either we have these uh you have some
[58:11] type of TI mirror inside or grading
[58:14] coupler inside the pick and we do some
[58:16] matching Optics here so you have a cated
[58:18] beam that comes out either verticle at
[58:20] at the uh angle or we offer um uh uh
[58:26] microoptics that you can attach the
[58:28] backside to your um silicon photonics
[58:31] device and then we also do the fiber
[58:34] coupling here so and
[58:37] um so we are very flexible with the with
[58:40] what type of microl lenses we offer so
[58:42] we have these backside cavities that are
[58:44] actually used as an beam expansion so
[58:47] that means we have a mix between air and
[58:49] silicon here to get a certain cated beam
[58:52] size here so this is so you can have a
[58:54] very thin silicon piece so if a space
[58:57] becomes rare these days in these
[58:58] transceivers they want to get smaller
[59:00] and thinner especially when you have to
[59:02] pack or stack two uh uh uh uh uh um two
[59:08] wdm in there so that means you need to
[59:11] you have very little space so by
[59:13] combining the expansion in air and
[59:15] silicon we can make these nice trenches
[59:17] on the back
[59:18] side so this is how it looks like then
[59:21] in reality so you have some type of pick
[59:23] or APD here then we do the Silicon M
[59:26] lens that is in recessed and people then
[59:28] put on microprism sometimes with
[59:31] a uh integrated more or less assembled
[59:35] preassembled isolator and then coup into
[59:37] single mode fiber so we see that quite
[59:39] often here this type of
[59:42] configuration and uh what is a little
[59:44] bit newer what we're working now on are
[59:46] these defract of optical lenses that are
[59:49] um how do you say they are
[59:54] fully uh off AIS typically they are
[59:57] fully flat they are high fi factor and
[01:00:01] they're fully integrated into this uh
[01:00:04] into the Silicon slab and then you can
[01:00:07] use it for an optical phas array but so
[01:00:10] this is typically for some type of free
[01:00:12] space Optics um for let's say for
[01:00:15] lighter applications or free space
[01:00:17] Communications so we see that now more
[01:00:19] and more often and we're already working
[01:00:21] on this with different clients so this
[01:00:23] is so they come with a grading CER
[01:00:25] output and then we match our lenses and
[01:00:28] the nice thing with this deflective
[01:00:29] Optical lenses each lens can be
[01:00:31] different so you can really generate
[01:00:33] some output field here in each lens um
[01:00:37] uh goes all the way to the edge so
[01:00:39] there's no gaps between the lenses so
[01:00:41] it's um very versatile so this is a
[01:00:44] summary again of what we can do so we
[01:00:46] can offer these single lenses this
[01:00:48] recess lenses for the assembling with a
[01:00:50] prism or we do these all these different
[01:00:52] features here that help you to assemble
[01:00:55] and for your vision systems and the
[01:00:57] prism lens and also we offer uh gold tin
[01:01:00] coating very thin actually Less Than 3
[01:01:03] micrometer thin so these are not
[01:01:04] electroplated these are very thin layers
[01:01:06] that keep the alignment very well uh in
[01:01:09] the
[01:01:10] z-axis and that's all from my side thank
[01:01:15] you thank you Wilfred great summary as
[01:01:19] always um I'm sure we will have some
[01:01:23] interesting questions about the micr
[01:01:25] lenses
[01:01:26] um yeah maybe we start already with ctil
[01:01:29] photonics syv
[01:01:31] please yes thank you very much wifed you
[01:01:35] mentioned you so you mentioned you have
[01:01:37] multi Channel Solutions can you comment
[01:01:40] um on the performance um so typically
[01:01:44] these are our clients who do this uh
[01:01:46] sometimes they share the data sometimes
[01:01:47] they don't so typically they go with
[01:01:50] what we've seen now is 20 channels this
[01:01:52] is quite common now so that we we have
[01:01:55] done now several
[01:01:56] chip delivers for 20 channels we don't
[01:01:59] usually get a feedback but when the
[01:02:01] client orders again then I think it's
[01:02:03] okay so it's a very good question I like
[01:02:05] to have the answer as well typically the
[01:02:07] the it's so 1x4 is standard 1 by8 has
[01:02:11] been standard we did 1 by 12 for many
[01:02:13] years already so and now the new
[01:02:16] channels are are 1 by 20 so there we we
[01:02:19] are still waiting for the feedback but 1
[01:02:21] by8 and 1 by 12 we have been doing for
[01:02:23] several years already now thank you
[01:02:27] all right and now Cliff Wong from wave
[01:02:30] splitter Technologies
[01:02:32] please I just want to ask if you think
[01:02:36] the grand cine will become the
[01:02:38] mainstream or H cine a b group we had a
[01:02:43] many many alignment system but I think a
[01:02:46] great covering is tsmc we all want to
[01:02:49] promote a solution so how do we think if
[01:02:53] the grand compy will become the main
[01:02:55] stream
[01:02:56] so that's a very good question excellent
[01:02:58] I like to know the answer as well so
[01:02:59] tsmc is very big and of course that's
[01:03:01] why I said tsmc effectively sets the
[01:03:03] standards now with their grading
[01:03:05] couplers uh but we see more and more
[01:03:08] often
[01:03:09] um uh Broadband solution which are not
[01:03:13] based on grading coupler so these are
[01:03:15] Edge emitters so most of the lenses we
[01:03:18] design now are for Edge emitters um um
[01:03:22] it's not clear yet I think there will be
[01:03:23] both going so the The Edge emitters and
[01:03:27] the grading couplers so grading couplers
[01:03:28] have typical loss of 2 to 4 5 DB uh
[01:03:32] maybe syby can comment on this you might
[01:03:34] know better than me but um but typically
[01:03:36] the grading coupler has a certain loss
[01:03:38] and with the edimer we expect less loss
[01:03:41] and everything is about low power now um
[01:03:45] of course the technology a little bit
[01:03:46] more complex when you do edits so it's
[01:03:49] an excellent question so we do a lot of
[01:03:51] optics for et so we have been doing
[01:03:52] those for many years so
[01:03:56] okay so maybe our end customer they said
[01:04:00] that they don't like to take a fiber so
[01:04:04] maybe that's why they want to use the
[01:04:07] GRE caring yeah yes so it's I I think
[01:04:10] it's Case by case so also the vrps have
[01:04:12] pros and cons so we saw that Intel
[01:04:15] recently in their recent presentation
[01:04:18] they said they last year they showed a
[01:04:19] lot of vgro Technologies this say they
[01:04:22] this year they said they're going away
[01:04:24] from the vgrs so we have to see what's
[01:04:26] coming H
[01:04:29] yeah all right thank you C yunen from
[01:04:33] bch
[01:04:36] please Al thanks uh to your presentation
[01:04:39] very fre very insightful and especially
[01:04:42] about the micro Optics uh which is also
[01:04:45] uh one of our Core Business actually
[01:04:48] competent I just wondering um so uh
[01:04:52] because internally we also discuss a lot
[01:04:54] about the passive and active assembly uh
[01:04:57] what will be the trend uh for the future
[01:04:59] and uh do you have a more specific
[01:05:01] number about the tolerance which is
[01:05:03] allowed for the uh coupling uh between
[01:05:07] the uh so once the light go through the
[01:05:10] micro lens which is cated and need to be
[01:05:13] aligned into the fiber array right and
[01:05:15] this step what will be the the tolerance
[01:05:18] uh so far you can achieve with your
[01:05:20] product that's a very good question as
[01:05:22] well so what we see now more and more
[01:05:24] that the mode field that coming out of
[01:05:26] the picks are between three and six
[01:05:28] microns so sometimes they get a little
[01:05:30] bit bigger but the spot size converter
[01:05:31] needs space so people try to keep it as
[01:05:34] small as possible so we see it between
[01:05:36] three and six microns now the typical
[01:05:39] mode fields from the pick um so then you
[01:05:41] can estimate the alignment accuracy
[01:05:43] needs to be half a micron or so or
[01:05:45] better um the um and on the other side
[01:05:51] we see a strong Trend towards a beam
[01:05:53] diameter between 40 and 60 microns
[01:05:56] and more and more now we see a lot of
[01:05:58] pitch of 127 microns pitch so and
[01:06:01] there's the beam diameter of about 40 45
[01:06:04] microns this is becoming almost a
[01:06:06] standard now we see that a lot now um so
[01:06:09] it's an excellent question so I think
[01:06:10] the sweet spot for the for the um 127
[01:06:15] micrometer pitch will be around 40 to 50
[01:06:18] microns and for the 250 Micron pitch it
[01:06:21] will be between 50 and 80 micr let's say
[01:06:23] so the mo the beam diameter outside and
[01:06:25] then you can estimate your tolerance so
[01:06:28] a very good question so and and for the
[01:06:31] we also have qu we we we our recent
[01:06:34] project where we do the microprism so
[01:06:36] there's the idea also to look into
[01:06:38] passive alignment but this will be
[01:06:40] tricky but this is something when we say
[01:06:43] passive that means you don't need light
[01:06:45] you do everything with vision um so uh
[01:06:48] and so that's still like semi-active
[01:06:51] let's say so that puts also a very high
[01:06:54] requirement on the accy of on our on our
[01:06:56] lens design so and we working so we we
[01:07:00] we are very good at this this is our
[01:07:02] core business doing very precise front
[01:07:04] and backside features um yeah I guess
[01:07:08] this is also something related to the uh
[01:07:10] mechanical design uh not only the the
[01:07:12] lens the optical design but also the
[01:07:14] mechanical um uh how say it um
[01:07:18] connection reputability right which will
[01:07:20] also be part of the the tolerance yeah I
[01:07:23] think this is a area maybe we have also
[01:07:26] a potential Corporation in the future so
[01:07:29] we as a packaging company and us as
[01:07:31] Optical uh companies we can maybe work
[01:07:34] together to to find a uh solution yes we
[01:07:37] have very close contacts to the team in
[01:07:39] Breman so we yeah and uh yes let's inim
[01:07:43] actually
[01:07:45] yeah good thank you yes okay thank you
[01:07:50] but we have to go more more in depth
[01:07:52] there huh so that's right yes yeah okay
[01:07:55] and in particular also expectations from
[01:07:57] our collaboration with the Taiwanese op
[01:07:59] electronic manufacturers willfred so yes
[01:08:03] so for Taiwan um we saw it earlier so
[01:08:06] that we like to work closer with the
[01:08:08] with the pck manufacturers there and
[01:08:09] also the assembly companies and and uh
[01:08:13] what we're actually looking into now is
[01:08:15] getting into pdk so process development
[01:08:18] kits from the Silicon photonics
[01:08:20] companies so we we're working very
[01:08:23] towards this so we have already good
[01:08:24] contacts with the software companies but
[01:08:26] at the same time we're also looking into
[01:08:28] the ad case so the assembly design kits
[01:08:31] and uh I think this is uh more dominant
[01:08:34] in Asia southeast Asia especially and uh
[01:08:38] yes and they would like to find Partners
[01:08:40] especially in the in Taiwan and then
[01:08:42] other Asian countries to to to get into
[01:08:46] the um uh to make micro Optics and
[01:08:49] microlenses that are on site available
[01:08:51] at the assembly companies this would be
[01:08:53] something very interesting for us so
[01:08:56] yeah okay okay all right so if that
[01:08:59] rings the bell to any of you please
[01:09:01] contact W or me and yeah we thank you
[01:09:04] again for the talk willfred and I pass
[01:09:06] the word to our next speaker uh Cliff
[01:09:08] Wong the director of wave wave splitter
[01:09:12] Technologies and he will talk about the
[01:09:14] wave splitter laser chips and Optical
[01:09:16] engine process
[01:09:20] flow Cliff the floor is yours
[01:09:32] they're good to
[01:09:34] go
[01:09:36] okay yes all
[01:09:41] good you can
[01:09:44] start
[01:09:46] okay I first of all I will introduce
[01:09:50] West Brit so we faor she in 1996 so
[01:09:55] so you can see our company name is a web
[01:10:00] spr so it's a passive
[01:10:04] component so before that we do the focus
[01:10:09] on wdm dwdm and the AWG product but uh
[01:10:15] after 2003 we want to become a IPO
[01:10:21] company in Taiwan so we establish a
[01:10:26] Rd team for the chip side and the
[01:10:29] transceiver side in Taiwan so you can
[01:10:31] see the 2020 and the 2021 we establish
[01:10:35] transer module and Manufacturing in Ty
[01:10:40] and establish n chip manufacturer in T
[01:10:44] and we also develop 10 key and 25 key
[01:10:48] it's high high speed T ler and
[01:10:52] uh and uh also in 2021 and 2022 we also
[01:10:57] develop 3010 high power L is the 70 minw
[01:11:03] first at 70° C and uh in 2022 and
[01:11:10] 2024 we also develop 100 Mil for
[01:11:17] C it's our goal office the headquarter
[01:11:21] in Taiwan and uh as I mentioned the side
[01:11:26] in Tai and the laser chip side in T and
[01:11:30] we also have the sales office in fremar
[01:11:35] and uh we also had a Japan team Japan is
[01:11:38] the sales office in Japan and we right
[01:11:41] now we have the it's the our
[01:11:44] manufacturer side we also build up the
[01:11:47] paa paa and theer side Ina and the CH is
[01:11:52] our supply chain manager team in
[01:11:57] China okay we had our core technology we
[01:12:02] had stre main application why is C laser
[01:12:07] including the chip laser chip and the
[01:12:09] laser chip array it's the high power
[01:12:12] laser and you can see the right side
[01:12:15] it's a 10 g 25 G 50 g DM laser chip is
[01:12:20] high speed so we also focus on fiber to
[01:12:24] the hall fiber to ruin and the left side
[01:12:27] is the sensor Optical sensing so we also
[01:12:31] develop 50/50 Naro Nano L with DB laser
[01:12:35] and the
[01:12:36] 653 g sensor DB laser so we had the high
[01:12:40] power and the high speed and the sensor
[01:12:43] chip is our core
[01:12:46] technology it's very simple de a chip
[01:12:51] process FL so we buy the AP five hourss
[01:12:56] prior and we also find the o to do the
[01:12:59] waer process after L we inh house to do
[01:13:04] the SC the wafer SC the wafer to the bar
[01:13:10] and we do the face coating and to do the
[01:13:13] a HR coating and after that we separate
[01:13:18] the BR bar to chip then we do the D C
[01:13:22] and the CC to do the relability
[01:13:25] and the bur
[01:13:28] testing so I will I want to show our CH
[01:13:33] Cur Rel hyp laser we also the 17 m w is
[01:13:39] in mass production we mention that 70 m
[01:13:43] is in 70° C okay and the 100 m right now
[01:13:48] we the sample evaluation so and we
[01:13:53] because we had a lesser chip so we also
[01:13:55] use our laser chip to develop our
[01:13:58] Optical engine and the transer so we
[01:14:01] also had a four 100 G dr4 Optical engine
[01:14:05] and uh by the next quarter next quarter
[01:14:10] next year we will ready the 800 key 2x
[01:14:15] dr4 LP receiver this
[01:14:18] year and uh also we had the highspeed
[01:14:23] uh ler chip so we had a 10 ke 25 G 50 g
[01:14:28] so we use our chip to develop our P
[01:14:32] product so like B
[01:14:35] orer and the and the
[01:14:38] 50/50 n chip is we also developing right
[01:14:44] now and I will show the it's a very
[01:14:47] simple uh ler chip process dra it's the
[01:14:52] it's the we process okay you can see the
[01:14:56] first draft is the AP W currently we use
[01:15:00] the 2 in or 3 in do the to so we buy the
[01:15:05] AP and the then we had the first mask is
[01:15:09] to do the reach so we use a dry and the
[01:15:13] W Hing to do the reach because our laser
[01:15:18] chip is a reach web guide okay so we are
[01:15:22] we we don't use the pH structure so we
[01:15:25] use the r guide structure to develop our
[01:15:28] laser chip so so you can see a reach so
[01:15:32] we use the dry and the wave aging to do
[01:15:35] a reach and after reach we use the
[01:15:38] preservation F to protect it's a silicon
[01:15:42] nitrate and after that we can you can
[01:15:46] see we we use the second mask we open
[01:15:49] the content and after that we start to
[01:15:52] do the P metal it's a Ty PT go to do the
[01:15:56] p and then we do the
[01:15:59] aing and then after that we do the go
[01:16:04] prating you can see the high power L the
[01:16:07] most important is the hit spraing so we
[01:16:12] will increase the go cin sness to help
[01:16:18] the his spreading so the typically they
[01:16:21] use the one micro sign but right now we
[01:16:25] use the six to add micro sness to help
[01:16:32] our hit
[01:16:34] braing and after that we do the laping
[01:16:38] then to do the and metal okay so it's a
[01:16:41] very very simple W
[01:16:46] process and it's our C
[01:16:50] laser you can see is our 100 Mil W sh
[01:16:54] laser is typically the IV curve okay so
[01:16:59] it's the the different shape is the
[01:17:01] different
[01:17:02] temperature so it's very very so it's
[01:17:06] just a show we can reach the 100 m stab
[01:17:10] at the 70 degre
[01:17:12] c okay so as I mentioned before our core
[01:17:16] technology is our stabilizer so we use
[01:17:19] our DP chip to develop our Optical
[01:17:22] engine and the also L2 receiver because
[01:17:28] we are we are not a T1 company so we
[01:17:31] cannot get the DS DSP source so it's not
[01:17:36] easy to get the DSP right now so we all
[01:17:39] only can develop
[01:17:42] P
[01:17:44] yeah okay and we we had 400 G and 800 G
[01:17:50] Optical engine you can see our left side
[01:17:53] is 400 G Optical engine it's a four
[01:17:56] Channel plan 100 G and you can see the
[01:18:01] left side is 800 key Optical engine so
[01:18:05] you can see the two stabul but apply to
[01:18:09] the
[01:18:10] one one pick one pick H
[01:18:16] chel so later I will show the optical
[01:18:20] engin the step by
[01:18:23] step Optical engine so we we first step
[01:18:27] is we we dunting our stabul laser into
[01:18:30] the S so it's a tic B into the it's the
[01:18:36] first
[01:18:38] step second step we we di
[01:18:41] another another component such as our we
[01:18:45] di on the pick we di on the lens holder
[01:18:49] with di
[01:18:51] Co into the the transmitter side the
[01:19:00] base and after lat we put the isolator
[01:19:04] and the semister and the MPD into the
[01:19:08] transmitter side
[01:19:12] base and the second we will put the TI
[01:19:18] and
[01:19:19] the PD R into a receiver side
[01:19:26] you can see it's the receiver side fber
[01:19:31] array it's active
[01:19:37] alignment and after L because when we
[01:19:40] finish the receiver side we will start
[01:19:43] to do the transmitter side and uh right
[01:19:47] now we select the the pick up that use
[01:19:50] the V gr solution so right now we use
[01:19:55] the big GL to do the V GL is a paf
[01:19:58] alignment so it's well easy to do the F
[01:20:03] attach and the high earray for us so we
[01:20:08] select the V gr to use the theu pick
[01:20:14] company then when we finish the TS fber
[01:20:18] with the empty connector so right now
[01:20:21] you can see after that the the final
[01:20:25] step is we use the alignment to do the
[01:20:29] test lens alignment yeah the the final
[01:20:34] process and the before that we had
[01:20:38] finish the transmitter fiber array fiber
[01:20:42] array passive alignment so you can
[01:20:44] through here to monitor the power
[01:20:48] level so so you can see we when we do
[01:20:52] the test t t side the next alignment you
[01:20:57] don't know the power label so but before
[01:20:59] that we had to do the transmitter fiber
[01:21:02] so we can use this here to monitor the
[01:21:05] power
[01:21:07] level
[01:21:08] okay so that's all from my side thank
[01:21:14] you all right thank you very much Cliff
[01:21:18] uh interesting
[01:21:19] presentation um so if you have any
[01:21:22] questions for our speaker please raise
[01:21:24] the hand
[01:21:25] in the meantime I wonder c um so at what
[01:21:29] scale can you produce this
[01:21:32] sub um assemblies and uh everything you
[01:21:36] show to us and also what is your
[01:21:39] expectation for bridging with the
[01:21:41] European photonic industry Consortium is
[01:21:43] is there any particular need you would
[01:21:46] like to cover yeah I think uh the most
[01:21:49] important for us is I think the bottom
[01:21:52] neck will be the alignment
[01:21:56] yeah so you can see the we can do the
[01:21:58] lens alignment it's it's very low
[01:22:03] through I think when we use the maybe
[01:22:06] China Company equipment or Japan company
[01:22:10] equipment the cut will be around the 5
[01:22:15] pie per hour so you can calculate if you
[01:22:19] want to reach the 100K per month output
[01:22:24] maybe you you need to buy
[01:22:27] 30 30 set alignment machine it's only
[01:22:31] alignment machine but we you need to do
[01:22:35] a Tang W bang and the L lens alignment
[01:22:39] maybe fiber alignment is if you don't
[01:22:41] use the vehicle maybe you use the h c
[01:22:46] yeah or Grand C so I think the most
[01:22:50] important the botle neck will be
[01:22:52] alignment machine and asign know uh if
[01:22:57] you press P press order to the alignment
[01:23:01] machine Supply the lead time will at
[01:23:05] least six months maybe
[01:23:09] longer yeah so that's why we cannot
[01:23:13] compete with the China
[01:23:16] spr because they had many many equipment
[01:23:20] such as maybe in online maybe y link
[01:23:23] it's a
[01:23:25] company they had more than 100 set
[01:23:30] equipment in their side yeah so I think
[01:23:35] maybe I you can see many
[01:23:37] many active activity will be used the
[01:23:42] alignment so I think for me it's very
[01:23:45] very important to get this alignment
[01:23:49] equipment prior support right now okay
[01:23:53] okay we got it well we have have a f
[01:23:54] contct right right there with us uh y CH
[01:23:57] have any quick comment on this maybe you
[01:24:00] can see the time will be wrong yeah this
[01:24:03] is exactly uh What uh we see that a lot
[01:24:07] of uh customers are uh the challenge for
[01:24:09] a lot of customers of us uh and this is
[01:24:11] also reason why we uh uh this is the F
[01:24:15] cont Tech preposition to provide the uh
[01:24:19] cap uh packaging Technologies uh in high
[01:24:22] through Pro and also e and also I will
[01:24:26] introduce this uh because I will be the
[01:24:28] next speaker I will introduce more about
[01:24:31] our technology in uh my presentation and
[01:24:34] offline we can also uh have more
[01:24:36] conversation so like I said I I will be
[01:24:38] in Taiwan in November and maybe we can
[01:24:41] also organize a uh offline meeting to
[01:24:43] present our Technologies um in a better
[01:24:47] way okay perfect yes so please pay
[01:24:50] attention to the next presentation uh
[01:24:52] yeah very quickly Wilfred what's your
[01:24:54] question
[01:24:56] thank you uh so my question is you
[01:24:58] showed a lot of pwn um so how do you so
[01:25:01] there's a lot of hyp for AI uh all the
[01:25:04] transceivers that go into the AI
[01:25:05] business do you make lasers for both
[01:25:07] markets or you concentrating on the pond
[01:25:10] and I would be interested in how these
[01:25:12] two markets compare in terms of cost
[01:25:14] volume reliability so are there big
[01:25:17] differences
[01:25:20] or is very big diverse yeah you can see
[01:25:26] the the
[01:25:28] uh uh for the P application is uh maybe
[01:25:33] 10 g 25 G 50 g p application it's a high
[01:25:38] variant and uh for the AI data center
[01:25:43] you can see
[01:25:46] currently we can see the many Channel
[01:25:49] supp such as inol or yink currently l
[01:25:54] still using the
[01:25:56] EML so they don't use the Silicon
[01:26:00] phon because currently they can get
[01:26:03] the chip yeah and uh yeah I think but
[01:26:09] they
[01:26:11] don't I think maybe for the maybe for
[01:26:15] the 800 G or 1.6 ter they will change
[01:26:21] but the currently I heard they they
[01:26:23] don't change because they Ste can get
[01:26:26] the chip from Japan
[01:26:29] supplier yeah so you can see the such as
[01:26:33] their customer Nvidia or Google they
[01:26:36] will consign the
[01:26:39] chip to Inola or yink so they don't
[01:26:43] worry about their chip
[01:26:46] Supply so they don't so so so right now
[01:26:50] you can see
[01:26:52] it's they are still using the EM so you
[01:26:56] can see how to convince them to use the
[01:27:01] silic maybe the cost maybe maybe if they
[01:27:05] can they can get EML
[01:27:09] chip because as you know em chip they
[01:27:12] need to recross two two times so the
[01:27:16] capacity is
[01:27:19] limited but for high power L maybe we
[01:27:22] can increase the quantity very
[01:27:25] so I think in the future high power L
[01:27:29] will be
[01:27:32] the be future yeah okay thank you okay
[01:27:37] yeah thank you C once again for your
[01:27:39] presentation and um let me already uh
[01:27:42] switch to the next speaker um so Ying
[01:27:46] Chen already heard from him uh pres
[01:27:49] presenting Pon Tech and Y Chan will talk
[01:27:52] about ecosystem for photonics packaging
[01:27:53] and
[01:27:56] testing is your okay thank you everyone
[01:28:00] uh let me share my
[01:28:06] screen can you see my screen
[01:28:09] already Yes just remove the thing on the
[01:28:12] top yeah okay it's not okay yes better
[01:28:18] okay yeah one more okay first of all
[01:28:20] yeah good first of all um thanks to Epic
[01:28:23] and and also uh tosia to give us the
[01:28:26] this opportunity to present uh our
[01:28:29] Technologies so my name is inan I'm
[01:28:31] sales manager and product manager for
[01:28:33] Asian Pacific region and overseeing the
[01:28:36] indirect uh sales uh such as in
[01:28:38] Singapore Korea is and following up um
[01:28:42] cityon phonics almost uh more than 10
[01:28:45] years so fire cont is one of the German
[01:28:48] hidden uh Champions uh we are focusing
[01:28:51] on photonics packaging and testing
[01:28:53] Solutions
[01:28:54] we provide the solution uh
[01:28:56] worldwide and since um photonic now
[01:28:59] integrated into semiconductors uh the
[01:29:02] challenge of semiconductor companies uh
[01:29:05] is the increased um packaging accuracy
[01:29:08] requirement and also need to maintain
[01:29:11] still the high uh throughput and also
[01:29:14] yield and also the quickly involving uh
[01:29:17] product development uh as previous the
[01:29:19] slides uh the previous speaker mentioned
[01:29:22] already
[01:29:24] and but this is what exactly the uh
[01:29:26] solution from Fon Tech can provide as a
[01:29:29] ecosystem so we have uh different
[01:29:31] platforms uh which is suitable for the
[01:29:33] MPI phase new product development uh to
[01:29:36] the high volume manufacturing which are
[01:29:38] proven by the top uh semi uh conductor
[01:29:41] companies uh the comprehensive
[01:29:43] capabilities we have is uh such as uh
[01:29:46] High Precision passive uh active
[01:29:48] alignment and also we can do the testing
[01:29:51] and packaging on with level and also on
[01:29:53] D level
[01:29:55] um all platforms of Fon Tech is using a
[01:29:58] unified API which can ensure the
[01:30:00] stimulus transition from um uh MPI to
[01:30:03] high manufacturing can also minimize the
[01:30:06] disruption and also accelerate the time
[01:30:08] to
[01:30:09] Market modular machine design is another
[01:30:12] Advantage we we have which enhance the
[01:30:15] spare parts management reduce the
[01:30:17] maintenance cost and also at the end
[01:30:19] lower the total cost of ownership I
[01:30:21] think this is essential for all the
[01:30:23] suppliers uh for the all the uh
[01:30:25] manufacturing companies the high yield
[01:30:28] and through proo is another Advantage uh
[01:30:31] which maximize the overall equipment
[01:30:33] efficient uh Fitness and also ensure the
[01:30:36] operational uh efficiency in the
[01:30:39] following minutes I will demonstrate how
[01:30:41] we can achieve
[01:30:44] that so let's dive into the total
[01:30:46] solution we provide from waiver to die
[01:30:49] so we provide the optical electrical
[01:30:51] testing Solution on wer level with this
[01:30:54] 3D printed uh prow scope uh solution we
[01:30:57] are able to do the H coupling already on
[01:31:00] W
[01:31:01] level um for the double side Optical
[01:31:03] engine um um we are developing the W
[01:31:07] level testing solution together with our
[01:31:09] customers and also Partners which are
[01:31:12] able to hardock the ETA uh directly on
[01:31:15] the uh
[01:31:17] testing system and for the highspeed
[01:31:20] testing so we also provide the uh
[01:31:23] packaging solutions for hetrogeneous
[01:31:26] integration uh of 35 device uh onto
[01:31:29] silicon fing chip on W level but also on
[01:31:33] Die Level with very high accur accuracy
[01:31:35] uh we are talking about 0.2 micrometer
[01:31:38] in uh three
[01:31:40] sigma so for the uh Die Level testing we
[01:31:44] also provide the high-speed OE testing
[01:31:46] Solutions with parallel testing process
[01:31:49] which can uh increase the throughput
[01:31:51] significantly
[01:31:56] for the customer product um actually
[01:31:58] previous speaker um uh has already uh
[01:32:01] talking about the transceiver modules
[01:32:03] quite a lot so we have the uh different
[01:32:06] packaging solutions for the traditional
[01:32:08] plaque B transceivers and also for the
[01:32:10] cor packaging uh Technologies and also
[01:32:12] including uh the external Laser Source
[01:32:16] we have already delivered uh over 150
[01:32:19] machines to the largest city conf foric
[01:32:21] uh company uh to their cm in uh South
[01:32:25] Asia for the lens attachment and also
[01:32:28] FAU attachment uh of the transceiver
[01:32:31] modules
[01:32:33] um um almost 80% of the companies who is
[01:32:36] developing the CPO uh packaging
[01:32:38] Technologies are cooperating with us uh
[01:32:41] therefore we our vesatile platform can
[01:32:44] provide uh Solutions such as uh fiber
[01:32:47] into VG grp attachment lens array
[01:32:50] attachment and also fiber array
[01:32:52] attachments directly on the
[01:32:54] uh uh CPO
[01:32:58] module beside this Optical uh already
[01:33:01] mentioned Optical alignment and testing
[01:33:03] Solutions we all utilize our
[01:33:05] understanding and experience in Optics
[01:33:07] and also together with our flexible uh
[01:33:10] platforms we also provide the solution
[01:33:13] the line solution for FAU assembly
[01:33:15] Technologies uh working with the largest
[01:33:17] the semiconduct company in Taiwan so uh
[01:33:20] due to time limitation I won't be able
[01:33:22] to uh uh explain too much about that
[01:33:25] basically we uh can start with a ribbon
[01:33:28] or uh single fiber and do the laser
[01:33:30] stripping cleaving and then end up with
[01:33:33] the end of line testing uh such as um uh
[01:33:37] pitch size measurement uh return loss
[01:33:39] measurement insertion loss measurement
[01:33:41] and so
[01:33:43] on if you have need uh any more
[01:33:46] information please uh feel free to
[01:33:48] contact me
[01:33:51] offline in terms of high uh accuracy
[01:33:54] hetrogeneous integration we provide also
[01:33:56] a special flip chip uh bonding
[01:33:58] technology with uh assistance of ir
[01:34:01] image system uh we can achieve submicron
[01:34:05] uh alignment accuracy so the princip of
[01:34:07] this technology is to uh use the as the
[01:34:11] picture shows we uh um I make a
[01:34:16] pointer so we use the IR image system
[01:34:19] and also IR IM illumination we can look
[01:34:22] through the P chip
[01:34:24] and we can align the fiducial on the top
[01:34:26] side of the P chip and bottom side of
[01:34:28] the uh 35 device die um so uh and
[01:34:32] Achieve uh submic accuracy uh um and we
[01:34:36] use the Advanced Vision process and
[01:34:38] together with the very accurate motion
[01:34:40] system we can already achieve uh 0.2
[01:34:43] micrometer uh in three sigma and uh
[01:34:47] beside alignment with fiducia we can
[01:34:49] also um uh use the wave guide as image
[01:34:52] to do the uh alignment
[01:34:59] okay let me move to the next slide okay
[01:35:02] in terms of high volume manufacturing uh
[01:35:05] we provide the Standalone machine uh
[01:35:07] with fully automatic uh loading system
[01:35:10] we are able we are also able to uh Lin
[01:35:12] up multiple machines um uh with the
[01:35:15] conveyor system and provide the inline
[01:35:17] solution so this is one example we
[01:35:20] provide the whole line solution for the
[01:35:23] uh for this is for automobile and uh the
[01:35:25] whole assembly line uh has like more
[01:35:28] than 20 machines for the camera
[01:35:30] assembly our solution is um semi S2
[01:35:34] certificated and also can fulfill the
[01:35:37] cisen uh requirement like I mentioned
[01:35:39] before we have already delivered a
[01:35:41] couple hundreds of machine for customers
[01:35:43] and now in this year we are delivering
[01:35:45] uh over 50 machines for One customer uh
[01:35:48] for the 1.6 terabyte transceiver uh
[01:35:51] modules
[01:35:56] we build up our software architectures
[01:35:58] around our uh packaging and testing
[01:36:01] Solutions and we use machine learning
[01:36:04] Technologies uh nowadays more uh we can
[01:36:06] call it AI Technologies to proceed uh to
[01:36:09] process or data we can gather inside the
[01:36:12] machine and provide the following uh
[01:36:14] functions such as um custom
[01:36:17] visualization for process parameters uh
[01:36:20] we can put all of uh manufacturing data
[01:36:22] on one dash sport uh inline key
[01:36:25] parameter monitoring uh which we call
[01:36:28] sometimes uh preventative maintenance
[01:36:30] the machine will uh monitoring the uh
[01:36:33] the data in production and if there's
[01:36:36] any uh unnormal uh Trend then the
[01:36:39] machine will automatic send uh alarm to
[01:36:42] the operator or
[01:36:44] engineers and we have also the
[01:36:46] capability to do the inlined uh motion
[01:36:49] parameter optimization active or passive
[01:36:51] alignment Improvement we use deep
[01:36:54] learning for inspection function we have
[01:36:57] also viral system uh as a simulation
[01:37:00] Tool uh customer can use it for for the
[01:37:02] product development and also process
[01:37:04] development without touching the real
[01:37:06] machine which is running in production
[01:37:09] um and we also have another very
[01:37:12] important function is so-called process
[01:37:14] uh control uh software uh imagine you
[01:37:18] have over 100 machines uh in the
[01:37:19] production line and if you need to
[01:37:21] update the process uh um you can update
[01:37:25] uh all the process on on one machine for
[01:37:27] one time without changing the parameters
[01:37:30] or process one by one which is very time
[01:37:32] consuming and huge
[01:37:35] effort so as a global company will
[01:37:38] provide a local service support from
[01:37:40] different fonex sites uh we have very
[01:37:42] strong service team uh in the United
[01:37:44] States Europe and uh in China and also
[01:37:48] in Thailand and we are uh establishing a
[01:37:51] new service team in Malaysia and also uh
[01:37:54] one new subsidiary in Taiwan with the
[01:37:57] application uh Labs uh in Asia uh in
[01:38:00] Asia in Europe and also in the US we
[01:38:03] also uh support customer uh product
[01:38:05] development and also proof of
[01:38:07] concept uh this is a very short
[01:38:09] presentation uh about our company and
[01:38:12] solutions um now we have couple minutes
[01:38:14] for
[01:38:17] questions okay okay thank you y Chan uh
[01:38:20] very nice presentation of f Tech
[01:38:24] um very briefly my question is yeah what
[01:38:26] are your expectations for this
[01:38:28] collaboration with Taiwanese of
[01:38:30] electronic
[01:38:31] manufacturers yeah so uh we see the uh
[01:38:35] the the the whole activities in Taiwan
[01:38:38] because now Taiwan is a very important
[01:38:41] market for us and it's also one of my
[01:38:44] region I'm supporting so uh since uh the
[01:38:47] biggest semiconduct company in Taiwan
[01:38:49] and also the whole osat and syst
[01:38:52] integrators is working together for
[01:38:54] cityon foric and we need to uh work very
[01:38:58] closely uh together with them so that we
[01:39:00] can uh get the uh Market Trend in the
[01:39:03] first moment and also work uh together
[01:39:06] with the packaging and testing solution
[01:39:08] uh with them and we believe that this
[01:39:10] will change the whole supply chain in
[01:39:12] the uh coming 5 to 10 years for foron
[01:39:16] X yeah yeah that's that's that's a good
[01:39:20] summary of it uh yeah hope well you
[01:39:22] already got
[01:39:24] in conversation with several of our
[01:39:26] previous speakers so I'm sure it's
[01:39:28] already becoming useful for you so yeah
[01:39:30] please feel free to uh poke in if any of
[01:39:35] this interests you as well um if there's
[01:39:38] no further questions I suggest we move
[01:39:39] to the last speaker as we're already
[01:39:41] short on time uh thank you y again and
[01:39:44] the last speaker uh is Justin Bry who
[01:39:48] represents ACH and he will talk about
[01:39:51] how motion control enables modern data
[01:39:53] comp Technologies so for logistic
[01:39:55] reasons um Justin recorded his
[01:39:58] presentation and uh later on his
[01:40:00] colleague Simon Chen uh will be able to
[01:40:03] take your
[01:40:05] questions
[01:40:10] so coming
[01:40:16] now hello and thank you for attending
[01:40:18] this presentation my name is Justin Bry
[01:40:20] and I'm a business development manager
[01:40:22] with Aerotech a global Precision
[01:40:23] automation company that is headquartered
[01:40:25] in Pittsburgh Pennsylvania for those of
[01:40:27] you who don't know Aerotech we've been
[01:40:28] in business for over 50 years
[01:40:30] specializing in producing automation
[01:40:31] technologies that enable extremely high
[01:40:33] precision and high throughput
[01:40:35] applications in a variety of Industries
[01:40:37] including the optical Communications
[01:40:38] industry our long Heritage in this
[01:40:40] industry has positioned us well to
[01:40:42] assist our customers in solving the
[01:40:44] challenges of manufacturing and testing
[01:40:45] the devices that enable modern data
[01:40:47] centers should you have any follow-up
[01:40:49] questions related to this presentation
[01:40:51] or if you're looking for General
[01:40:52] automation guidance please reach out to
[01:40:54] me via email or LinkedIn or feel free to
[01:40:57] ask questions at the conclusion of this
[01:41:00] webinar industry shaping trends like Ai
[01:41:03] and high performance Computing have been
[01:41:05] evolving at a Breakneck Pace continuing
[01:41:07] to push the limits of how we interact
[01:41:08] with the digital world as these
[01:41:10] Technologies grow more powerful and more
[01:41:12] ubiquitous in our daily lives they're
[01:41:14] driving an insatable demand for
[01:41:15] additional computational power at
[01:41:17] hyperscale data centers demands that are
[01:41:19] currently outpacing More's Law And while
[01:41:21] headlines tend to focus on this growth
[01:41:23] in compute demand the bandwidth needs
[01:41:26] for communicate communicating
[01:41:27] information from one system to another
[01:41:29] within data centers is following a
[01:41:31] similar trajectory as established
[01:41:33] technologies have begun to see
[01:41:35] diminishing returns and in some cases
[01:41:37] begin to reach the limitations of
[01:41:38] fundamental physics many people have
[01:41:40] claimed that the traditional
[01:41:41] understanding of Moors law is quote
[01:41:42] unquote dead or at the very least
[01:41:44] meaningfully slowing down because of
[01:41:47] this large scale deployment of Paradigm
[01:41:49] shifting Technologies such as silicon
[01:41:51] photonics and advanced packaging will be
[01:41:52] necessary NE to meet the demands of
[01:41:54] applications like Ai and high
[01:41:55] performance Computing in today's webinar
[01:41:58] we'll delve deeper into these
[01:41:59] Technologies and discuss some of the key
[01:42:00] process challenges in manufacturing and
[01:42:02] testing
[01:42:03] them speaking of silicon photonics this
[01:42:06] technology offers several distinct
[01:42:07] advantages when compared to traditional
[01:42:09] semiconductor-based electronic devices
[01:42:11] first bandwidth and latency Optical
[01:42:14] signals have been a foundational
[01:42:15] communication technology for decades
[01:42:17] enabling data transmission at
[01:42:19] significantly higher speeds and at
[01:42:20] greater bandwidth than Electronics alone
[01:42:22] additionally
[01:42:23] Optical signals are more resistant to
[01:42:25] external interference reducing signal
[01:42:27] attenuation particularly over long
[01:42:29] distances high-speed Optical
[01:42:30] interconnects within data centers take
[01:42:32] advantage of these properties the next
[01:42:34] benefit is Energy Efficiency Optical
[01:42:36] technologies that more tightly integrate
[01:42:38] photonic components shorten the lengths
[01:42:40] of conductive copper on the device
[01:42:42] reducing power draw and heat generation
[01:42:44] this is particularly critical as the
[01:42:45] power demands of data centers are
[01:42:47] growing
[01:42:48] exponentially finally device footprint
[01:42:51] adoption of silicon photonics
[01:42:52] Technologies has allowed transceiver
[01:42:54] manufacturers to pack more functionality
[01:42:56] into a smaller space increasing the
[01:42:58] density of optical interconnects
[01:42:59] additionally while still in the early
[01:43:01] stages commercially emerging
[01:43:03] Technologies like co- package Optics
[01:43:05] offer the promise of replacing the
[01:43:06] industry standard plug pluggable
[01:43:08] transceivers with more integrated
[01:43:10] devices that are far smaller in both
[01:43:11] size and power demands because of these
[01:43:13] advantages Optical transceivers
[01:43:15] utilizing silicon photonics based
[01:43:18] photonic integrated circuits already
[01:43:20] make up a substantial portion of the
[01:43:21] optical transceiver Market particularly
[01:43:23] for the highest speed devices and this
[01:43:25] share is consistently
[01:43:27] growing another important and related
[01:43:29] trend of the fundamental Technologies of
[01:43:31] modern transceivers is the shift towards
[01:43:33] tighter integration at the package
[01:43:34] device level traditionally the optical
[01:43:36] engine of the transceiver which includes
[01:43:39] photonic integrated circuits and a light
[01:43:40] source is housed within the plugable
[01:43:42] transceiver itself as shown in the top
[01:43:45] diagram however enabl by Advanced
[01:43:47] packaging techniques such as
[01:43:48] heterogeneous integration where
[01:43:50] components of different materials are
[01:43:51] packaged together new architectures
[01:43:53] including onboard Optics and co-packaged
[01:43:56] Optics are being developed to continue
[01:43:58] reducing both the physical size and
[01:44:00] power demands of these devices by
[01:44:02] shortening the electrical paths on the
[01:44:04] device signal robustness is improved and
[01:44:06] electrical resistance is reduced
[01:44:08] lowering the PS consume per bit of data
[01:44:12] transferred tighter integration is
[01:44:14] driving significant advancements in
[01:44:15] Optical transceivers but the downside of
[01:44:17] these increasingly complex architectures
[01:44:19] is that this complexity also introduces
[01:44:21] new challenges in manufacturing test in
[01:44:23] and packaging perhaps the most demanding
[01:44:25] aspect of both manufacturing and testing
[01:44:27] Optical transceivers is the need for
[01:44:28] extreme Precision often the micrometer
[01:44:31] and even nanometer of scale to meet
[01:44:33] these stringent requirements
[01:44:35] manufacturers often must invest in
[01:44:36] advanced equipments and equipment and
[01:44:38] processes including High Precision
[01:44:40] positioning
[01:44:42] stages so what specific processes
[01:44:44] require these micrometer and nanometer
[01:44:46] levels of precision while not an
[01:44:48] exhaustive list I plan to cover a few
[01:44:50] today including wafer and die Level test
[01:44:53] T in of photonic integrated circuits
[01:44:54] which is often performed to ensure
[01:44:56] quality picks before device packaging
[01:44:58] takes place Precision Alignment for
[01:45:00] advanced packaging techniques including
[01:45:02] heterogeneous integration and 2.5d and
[01:45:04] 3D chip stacking and Optical alignment
[01:45:07] and assembly where components within a
[01:45:09] transceiver are aligned to ensure
[01:45:10] adequate signal strength as you'll see
[01:45:12] on the subsequent slides all these steps
[01:45:14] of the manufacturing utilize Precision
[01:45:16] motion control systems in different and
[01:45:17] sometimes related
[01:45:19] ways as I mentioned previously silicon
[01:45:22] photonics Wafers can be manufactured
[01:45:24] using traditional semiconductor methods
[01:45:26] making silicon a very compelling option
[01:45:28] for photonic integrated circuits similar
[01:45:30] to to traditional integrated circuits
[01:45:32] both wafer and die L testing methods are
[01:45:34] used for quality control but testing
[01:45:36] pick is a much different and more
[01:45:38] complex process than testing traditional
[01:45:40] electronic ic's this is because testing
[01:45:42] photonic integrated circuits requires
[01:45:45] testing of both the electrical and
[01:45:46] Optical paths on the die for electrical
[01:45:49] paths a simple physical connection is
[01:45:50] usually all that is needed to conduct a
[01:45:52] test but for optical pads devices are
[01:45:54] typically sensitive to the positioning
[01:45:56] of the probes rather than just contact
[01:45:58] because of this precise probe
[01:46:00] manipulation is required including in
[01:46:02] several degrees of freedom and in
[01:46:03] complex kinematic
[01:46:05] arrangements as an example shown here is
[01:46:07] a diagram representing Edge coupling on
[01:46:09] a silicon photonic wafer Precision
[01:46:11] linear and angular alignment between the
[01:46:13] probe and photonic device under test is
[01:46:15] required as misalignments measured in
[01:46:17] micrometers and even nanometers can
[01:46:19] affect the signal loss during testing
[01:46:21] additionally the adjustment made during
[01:46:23] these tests typically need to occur at
[01:46:25] the tip of the probe so a system capable
[01:46:27] of of pivoting about this point of
[01:46:29] interest is often needed so how are
[01:46:31] these Precision alignments executed High
[01:46:34] Precision Positioning Systems such as
[01:46:35] the Sixaxis direct drive Aerotech
[01:46:38] solution shown here enable highly
[01:46:40] accurate placement of the optical probes
[01:46:42] to ensure consistent and repeatable
[01:46:43] device testing combined with aerotex
[01:46:46] automation one motion Control software
[01:46:48] and drive Electronics this system is
[01:46:50] able to Pivot SE several degrees of
[01:46:51] freedom about the assign workpoint and
[01:46:53] Achieve o off-axis error motions
[01:46:56] measured in singled digigit micrometers
[01:46:58] despite the workpoint being hundreds of
[01:47:00] millimeters from thees mounting Point
[01:47:02] automation one simplifies configuration
[01:47:04] of these workpoint offsets eliminating
[01:47:05] the need for the user to perform complex
[01:47:07] linear algebra additionally alignment
[01:47:10] applications like this also typically
[01:47:12] take advantage of automation 1's Optical
[01:47:14] alignment algorithms to optimize the
[01:47:17] position of the probe based on an
[01:47:18] optical signal strength reading that is
[01:47:20] sent to the controller the controller
[01:47:21] uses this signal strength to adjust the
[01:47:23] position of the system based on several
[01:47:25] built-in algorithms identifying the
[01:47:27] ideal position for the probe to maximize
[01:47:29] the
[01:47:30] signal meeting stringent test
[01:47:32] requirements like these often require
[01:47:34] careful selection of an appropriate
[01:47:35] motion system whether the system is a
[01:47:37] Serial kinematic system like the one
[01:47:39] shown here or a parallel kinematic
[01:47:41] system such as a hexapod there are
[01:47:43] several advantages and disadvantages
[01:47:45] associated with both Technologies but
[01:47:47] one key advantage of Serial kinematics
[01:47:49] is that the off axis error motions
[01:47:50] introduced throughout the range of
[01:47:52] travel of each aess are much easier to
[01:47:54] characterize and therefore calibrate
[01:47:56] this improves the repeatability
[01:47:57] performance of the system and reduces
[01:47:59] the risks of unexpected 3D err motions
[01:48:01] to get a bit more specific the six- axis
[01:48:03] serial kinematic positioner shown here
[01:48:05] includes three degrees of freedom in
[01:48:06] each linear direction as well as three
[01:48:08] degrees of freedom in each rotational
[01:48:10] Direction roll pitch and yaw this system
[01:48:13] utilizes non-c contct Motors and high
[01:48:15] life cycle cross roller bearings to
[01:48:17] ensure the industrial robustness and
[01:48:19] high stiffness while also minimizing 3D
[01:48:21] geometric errors capable of achieving 10
[01:48:23] nmet steps at the workpoint offset as
[01:48:25] shown in the plot which was captured
[01:48:27] using a capacitive probe this system
[01:48:30] singled digigit nanometer level in inos
[01:48:33] stability and minimum incremental motion
[01:48:35] are highly suitable for wafer probing
[01:48:38] applications next I wanted to touch on
[01:48:40] Advanced packaging a key set of
[01:48:42] technologies that are enabling not only
[01:48:43] the next generation of semiconductor
[01:48:45] devices but photonics enabled devices as
[01:48:47] well Advanced packaging encompasses a
[01:48:49] variety of manufacturing methods being
[01:48:51] developed and deployed primarily by the
[01:48:53] semiconductor industry some of which
[01:48:55] like tsmc's Coos have been fully
[01:48:58] commercialized for years many others
[01:49:00] like 3D stacking and hybrid bonding are
[01:49:02] in various stages of development and
[01:49:04] commercialization with expectations of
[01:49:06] broader utilization in the coming years
[01:49:08] Advanced packaging methods like
[01:49:09] heterogeneous integration which combine
[01:49:11] multiple chips of different materials
[01:49:13] into one package are critical to the
[01:49:15] pursuit of more tightly integrated
[01:49:16] optical devices mentioned previously
[01:49:19] this is because the material properties
[01:49:20] of silicon prevent it from lazing so for
[01:49:22] an evolving technology like co-packaged
[01:49:24] Optics to truly integrate the Laser
[01:49:26] Source into the package device
[01:49:28] non-silicon materials will be needed to
[01:49:30] be incorporated regardless of the
[01:49:32] materials being integrated as the
[01:49:34] feature sizes of these devices push into
[01:49:36] the low single- digigit micrometer
[01:49:38] ranges such as in the case of hybrid
[01:49:40] bonding the alignment accuracies
[01:49:42] required during the integration process
[01:49:43] continue to tighten well into the
[01:49:45] nanometer
[01:49:48] levels various 2.5d and 3D stacking
[01:49:51] methods are being used for header genous
[01:49:53] integration while the specifics of each
[01:49:54] process vary and different manufacturers
[01:49:57] use different and often proprietary
[01:49:58] methods these processes almost always
[01:50:00] require Precision Alignment at some
[01:50:02] point given the size of the components
[01:50:03] being integrated for example when
[01:50:05] aligning chips to an interposer for 2.5d
[01:50:08] stacking individual chips are stacked
[01:50:09] onto an interposer and the mic microbump
[01:50:12] contacts and through silicon Vias of the
[01:50:13] chip must align with the redistribution
[01:50:16] layer within the interposer to enable
[01:50:18] connections to other chips in the
[01:50:19] package these types of alignments
[01:50:21] require precise linear positioning of
[01:50:23] each ship in the X and Y directions
[01:50:25] followed by Motion in the Z direction
[01:50:26] that is closely controlled for both
[01:50:28] position and force for wafer to wafer
[01:50:30] processes additional challenges are
[01:50:32] introduced as the angular positioning of
[01:50:34] the wafer becomes even more important
[01:50:36] than during a chip scale process for
[01:50:38] example for a 300 mimet wafer if the
[01:50:40] wafer is rotated about the vertical axis
[01:50:42] by even as much as 100th of a degree
[01:50:45] this results in a linear misalignment at
[01:50:47] the edge of the wafer that's larger than
[01:50:48] 25 microns and misalignments of this
[01:50:50] magnitude will result in significant def
[01:50:52] effects and therefore reduce yield as
[01:50:55] would be expected for any high volume
[01:50:56] manufacturing operation and especially
[01:50:58] so in the semiconductor industry both
[01:51:00] high yield and high throughput are
[01:51:01] critical so high performance Positioning
[01:51:03] Systems that can combine the necessary
[01:51:05] Precision with high Dynamics and with
[01:51:07] high Dynamics are foundational to an
[01:51:09] optimized process
[01:51:11] flow as mentioned previously
[01:51:13] heterogeneous integration is required
[01:51:15] for co-packaging of Optics onto a single
[01:51:16] substrate just recently a major
[01:51:18] manufacturer of optical Communications
[01:51:20] equipment debuted one of the first
[01:51:22] commercial available products that takes
[01:51:24] advantage of this technology an optical
[01:51:26] switch that utilizes fan out wer level
[01:51:28] packaging to co- package photonic
[01:51:30] integrated circuits with traditional
[01:51:32] ic's however scalability of these
[01:51:35] processes remains a key challenge for
[01:51:36] the industry as many of the current
[01:51:38] approaches require aligning and stacking
[01:51:40] chips individually which is highly timec
[01:51:42] consuming and therefore throughput
[01:51:44] limiting compared to most traditional
[01:51:45] semiconductor processes because of these
[01:51:47] limitations new parallel processing
[01:51:49] methods are constantly being explored to
[01:51:51] reduce cost and increase output one
[01:51:54] parallel processing method that is being
[01:51:56] explored at the lab level for
[01:51:57] heterogeneous integration is
[01:51:58] microtransfer printing microtransfer
[01:52:00] printing is a process that has been
[01:52:02] developed primarily for the production
[01:52:03] of another emerging technology microed
[01:52:06] displays microtransfer printing involves
[01:52:08] the transfer and attachment of an array
[01:52:10] of components simultaneously in a wafer
[01:52:12] to wafer format typically since the
[01:52:14] wafer substrates can be different
[01:52:17] materials microtransfer printing is a
[01:52:18] technology of interest for heterogeneous
[01:52:20] integration at scale that being said
[01:52:22] said this wafer to wafer process is
[01:52:24] subject to the Precision Alignment
[01:52:25] challenges mentioned previously and
[01:52:27] therefore require highly precise
[01:52:29] positioning equipment like the example
[01:52:31] shown on the bottom right for execution
[01:52:33] this system utilizes an Aerotech planer
[01:52:35] air bearing XY
[01:52:37] stage and includes a total of 10 axes of
[01:52:39] motion for conducting planarity
[01:52:41] sensitive alignments between the various
[01:52:43] components the exceptional geometric
[01:52:45] performance and dynamic capabilities of
[01:52:47] the air bearing stage enable inspect
[01:52:49] alignments to be achieved at high speeds
[01:52:51] satisfying the throughput require of
[01:52:52] this specific
[01:52:54] process after the chip scale packaging
[01:52:56] processes are completed the final
[01:52:58] assembly of optical communication
[01:52:59] devices often requires High Precision
[01:53:01] alignments as well even though the
[01:53:03] feature sizes are often larger than at
[01:53:05] the chip packaging stage Precision
[01:53:07] positioning at several degrees of freom
[01:53:08] is critical to reducing insertion losses
[01:53:10] very similar to the visioning challenges
[01:53:12] discussed during the waiver and die
[01:53:14] Level testing section misalignments
[01:53:16] during the final assembly can result can
[01:53:18] result in devices that do not meet
[01:53:20] quality and Signal specifications as an
[01:53:22] example the plot on the bottom right
[01:53:23] shows Optical signal strength during an
[01:53:25] alignment across a 20 micrometer by 20
[01:53:27] micrometer range as you can see the
[01:53:29] signal drops off considerably with just
[01:53:30] a few microns of Mis alignment
[01:53:32] fortunately alignment algorithms like
[01:53:34] the one in aerotex automation One
[01:53:36] controller are purpose built to solve
[01:53:37] these challenges additionally Clos Loop
[01:53:40] position and force control is often
[01:53:41] necessary during these bonding processes
[01:53:43] as Parts can shift due to epoxy carrying
[01:53:45] temperature fluctuation and other
[01:53:47] factors in the case of some types of
[01:53:50] optical devices aligning multiple
[01:53:51] components simultaneously can also be
[01:53:53] necessary shown here is a solution from
[01:53:55] Aerotech that was for a specific Optical
[01:53:57] device manufacturing application in the
[01:53:59] case of this system two separate Optical
[01:54:01] components needed to be aligned at once
[01:54:03] each with six degrees of freedom
[01:54:05] positioning flexibility to accomplish
[01:54:07] this Aerotech mounted two separate six
[01:54:09] off positioners in an inverted
[01:54:11] Arrangement onto one of our linear motor
[01:54:13] based Gantry systems all being
[01:54:15] synchronously controlled within a single
[01:54:17] instance of automation one the linear
[01:54:19] motor technology enables the system to
[01:54:20] achieve High Dynamics while moving
[01:54:22] between different devices shortening the
[01:54:24] device-to-device travel time while
[01:54:25] maintaining Micron level positioning
[01:54:27] repeatability once in position the six
[01:54:29] off positioners are able to align the
[01:54:30] components as needed using the tool
[01:54:32] center point programming tool
[01:54:35] highlighted in the animation once
[01:54:37] aligned high bandwidth Clos Loop Force
[01:54:39] control across multiple degrees of
[01:54:40] freedom ensures the appropriate forces
[01:54:42] are applied during final assembly
[01:54:44] process as you can see automation
[01:54:46] Solutions like this one require the
[01:54:47] deployment of complex High Precision
[01:54:49] mechanics combined with Advanced control
[01:54:51] features which which are among our
[01:54:52] favorite kinds of projects at Aerotech
[01:54:55] to summarize I hope it's now clear that
[01:54:57] Precision automation plays a critical
[01:54:58] role in the manufacturing and testing of
[01:55:00] today's modern data in infrastructure
[01:55:03] demand for bandwidth within data centers
[01:55:04] is increasing exponentially due to
[01:55:06] Technologies like AI a trend that shows
[01:55:08] no sign of slowing down with traditional
[01:55:10] Technologies starting to encroach on the
[01:55:11] boundaries of fundamental physics
[01:55:13] Innovations in Paradigm shifting
[01:55:15] Technologies like Silicon photonics and
[01:55:16] co- package Optics will be needed to
[01:55:18] address these demands Precision motion
[01:55:20] control already plays an important role
[01:55:22] in the manufacturing and testing of
[01:55:23] these Technologies a role that is set to
[01:55:25] become even more critical as optical
[01:55:26] devices become more integrated thank you
[01:55:29] for your time and attention today please
[01:55:31] feel free to contact me on LinkedIn or
[01:55:32] via email at J bry.com if you'd like to
[01:55:34] arrange a follow-up discussion my
[01:55:36] colleagues William yay and Simon Chen
[01:55:38] from our Aerotech Taiwan team will now
[01:55:40] take some time to address questions from
[01:55:41] the
[01:55:43] audience thank
[01:55:47] you all right uh so this was the
[01:55:50] presentation from otch and
[01:55:52] I believe we have Simon with us to
[01:55:55] answer the questions hello
[01:55:59] Simon um yeah so yeah please so this is
[01:56:02] the last Talk of the today's session so
[01:56:04] very quickly uh please raise your hands
[01:56:08] if you have a question for our speaker
[01:56:11] today um in the meantime yeah I'll ask
[01:56:14] yeah Simon so what is um I mean it's
[01:56:16] very clear the proposition of ACH but
[01:56:18] what would you expect from this
[01:56:20] collaboration with the European photonic
[01:56:22] industry Consortium what what would be
[01:56:24] your needs that uh we as a European
[01:56:27] industry can
[01:56:30] cover yeah let me try to answer that Sam
[01:56:33] is my colleague Al William joining here
[01:56:35] uh yeah just uh I think this is the
[01:56:37] first time we actually join this event
[01:56:40] together and Aerotech we are one uh the
[01:56:43] company that is quite different from all
[01:56:45] other speakers today so we are kind of
[01:56:47] the motion motion control automation
[01:56:49] company so we can up upper straing
[01:56:51] compared to like f cont Tech they are
[01:56:53] kind of machine Builder uh like a focus
[01:56:56] light and and they all others they
[01:56:59] provide uh kind of uh the final product
[01:57:02] but we kind of like component of the
[01:57:04] entire machine so we would like to
[01:57:06] explore the possibility to finding the
[01:57:09] new trends of more uh difficult motion
[01:57:13] challenges that base on these new
[01:57:15] photonic devices so because uh as all
[01:57:18] the speakers U mentioned today about the
[01:57:20] alignment accuracy requirements and
[01:57:22] about the throughput requirements uh
[01:57:25] it's very clear that uh we see like for
[01:57:28] example the previous speaker Cliff he
[01:57:30] mentioned about the throughput is very a
[01:57:33] big bottleneck for him so that's kind of
[01:57:35] thing that we want to identify will that
[01:57:37] problem that kind of bottom neck is
[01:57:39] something related to motion that we can
[01:57:40] address so and not limiting to that and
[01:57:43] just anything related to motion control
[01:57:45] what can we do for kind of the photonic
[01:57:48] industry sort of like alignment
[01:57:51] attachment testing processes oh these
[01:57:53] are something we are very
[01:57:55] interested all right all right okay this
[01:57:57] is very clear so I sure this might ring
[01:58:01] the bell for some of us in the audience
[01:58:05] and uh if if anyone please raise the
[01:58:08] questions now and if not we're really
[01:58:11] hitting the U the time limit right now
[01:58:14] so uh anyways thank you very much for
[01:58:17] your presentation Simon and also your
[01:58:19] colleagues who actually recorded it
[01:58:22] um I believe we can be just uh in touch
[01:58:26] after the meeting for follow-ups and uh
[01:58:29] either way so we thank also our partners
[01:58:33] from toshia and all of the speakers who
[01:58:36] presented today for making this uh
[01:58:38] meeting possible let me just uh share
[01:58:41] very quickly some information with
[01:58:43] you um so this was the Epic online
[01:58:47] technology meeting on integrated fonics
[01:58:50] manufacturing with you together with our
[01:58:52] Partners from uh Taiwanese op electronic
[01:58:55] semiconductor
[01:58:58] Association um so yeah once again we
[01:59:03] heard very exciting talks from alalum
[01:59:06] aacop Electronics focus light with
[01:59:08] spitter Technologies p tech and ACH and
[01:59:13] uh before you forget uh let me let me
[01:59:17] introduce you the next occasion when we
[01:59:19] can meet and discuss um this uh the
[01:59:22] related topics topics of photonics
[01:59:25] integration and semiconductors and this
[01:59:27] will be very soon already in 11 and 12th
[01:59:30] of November uh in Munich during the
[01:59:33] famous semicon Europe Expo that takes
[01:59:36] place in November 12 to 15 so we'll have
[01:59:40] the offline meeting of Epic members and
[01:59:43] also all the associated Partners uh
[01:59:46] there in Munich right next to the Expo
[01:59:48] and then we'll go to visit the Expo
[01:59:50] together so uh yeah don't miss the
[01:59:52] chance there is only two weeks left so
[01:59:54] please register now on our epic -
[01:59:57] phonic.com website and this will be the
[02:00:01] our next occasion to meet in person so
[02:00:04] with this I'm just there to remind you
[02:00:06] that epic uh ponic industry Consortium
[02:00:09] is there to support you with technology
[02:00:11] uh expertise Market studies networking
[02:00:14] mentorship recruiting and investment
[02:00:16] opportunities and I'd like to thank my
[02:00:19] whole team of Epic that made this
[02:00:20] meeting possible and all of you today
[02:00:23] for your attention thank you very much
[02:00:25] and I see you on the next
[02:00:29] occasion
[02:00:32] great all right thank you thank you very
[02:00:34] much indeed