# EPIC Online Technology Meeting on PIC Packaging and Testing

https://www.youtube.com/watch?v=A36K3HaBAKo

[00:05] foreign.
[00:22] from epic and this is the series of online technology meetings that we run on a photonics this year and today we have a very exciting topic there is a big packaging and testing.
[00:38] but I will hear from Seven speakers representing all the value chain of photonics Industry working with big and we'll start with the presentation of Peter Wright from Kindle Institute and then during Louis from fix what I also has a look here from the ocean from nanascribe Luke from icon photonics and niclaus from Mario Optics Joshua from python Tech and Richard from.
[01:08] Yellow.
[01:11] Uh, apart from this amazing setup lineup of speakers, we also have over 80 people in the room tonight with us.
[01:19] And those are representatives of the companies from every chain.
[01:26] Um, in the valley chain, so we start with uh people working on RNG for picks, semiconductor manufacturers and equipment manufacturers, as lasers as well as companies that provide lasers and optics, and all the way to system integrators and the end users of slash technologies.
[01:49] And uh, also very important today, I have my colleague Antonio Castella with me.
[01:56] Hello Antonio.
[01:59] Hello everybody.
[02:01] And yeah, welcome everybody, and it's good to be today with this nice topic and also this grade five exposures that are.
[02:09] supporting the event to date.
[02:11] Um, as you can see, I will sponsor today.
[02:12] Are the companies Fix, uh, Content, and
[02:15] Tech, Accept, Results, Micro Optics, and we
[02:19] want to give some time to introduce the
[02:21] companies and to speak a little bit of
[02:23] what they are doing.
[02:25] Um, so first one.
[02:27] D from FX, the first doors, you can tell
[02:30] us what is Fix doing? Thank you, Antonio.
[02:33] Yeah, so we are Fix and we are
[02:35] established in 2017 and we do the
[02:38] packaging of photonic integrated circuits,
[02:40] MM-based components in different volumes,
[02:43] so really helping from prototype to high
[02:45] volume production, and we're up, we're
[02:48] leading expertise in hybrid pick
[02:51] assembly, fiber attachment, and ultra high
[02:53] frequency electronics interfacing.
[02:56] So currently we do packaging up to like 120
[02:58] gigahertz.
[03:01] So thank you very much, uh, and now we
[03:04] have next, uh,
[03:05] speak on Tech, and we have today Josue
[03:08] para, uh, that you can say a few words.
[03:11] About your company, yes, thank you very much for the kind introduction.
[03:15] And my name is Research and Development Engineer at 5 Contact Ireland.
[03:21] If I contact has already more than 20 years in the market and we specialize in providing automatic photonic solutions for the alignment, assembly, and testing of a photonic integrated circuits.
[03:36] Um, I mean all what involves um, let's say from micro optics, from the pigs, from the different electronic components and electronic components as well.
[03:47] If we have to fix them, cure them using maybe some epoxy, some like a soldering and of course, we are focusing as well in the testing of these components.
[03:59] The community attach not only at a high level, but also at wafer level.
[04:04] Thank you, thank you very much.
[04:06] We can move to the next one, the company I am Tech, we have Andreas Freytack.
[04:12] floor is yours for telling us a little bit about your company.
[04:15] good afternoon everybody.
[04:17] thank you very much for the kind introduction.
[04:20] um I'm from amtec Key account manager since three years now.
[04:24] um mtech is located in Berlin with around 230 people and we are specialized in the advanced electronic packaging services for our customers starting from the process development due to work on production.
[04:40] if you can just switch to the next slide.
[04:42] um yeah as you can see we offer services from the prototyping Supply Chain management and industrialization but we can also integrate test systems into the production line and we are working in a broad range of different markets.
[05:00] starting from medical industry automation semiconductor and also data telecommunication where we see a growing demand for these kind of applications into into.
[05:13] semicon fiber and recently also started.
[05:17] Aerospace products.
[05:19] thank you.
[05:21] so thank you Andreas for this.
[05:23] interesting introduction we moved to.
[05:25] music Kim he's going to talk a little.
[05:28] bit about accessories so um.
[05:32] hello everyone Antonio can I.
[05:36] share with my screen.
[05:38] because maybe then I need to ask you.
[05:41] clicking a lot of buttons so maybe I.
[05:43] will share.
[05:44] okay you can share yes it will be the.
[05:47] same slide but.
[05:51] just wait a second.
[06:04] do you see the screen.
[06:07] yeah thank you okay thank you thank you.
[06:11] Antonio for the introduction.
[06:13] so accessories is leading manufacturer.
[06:15] micro Optics including micro lenses lens arrays and also cylindrical lens.
[06:21] cylindric lens arrays and also diffractive care elements.
[06:25] our micropics products can enable many kinds of these the optical transceivers.
[06:31] some examples are shown here.
[06:33] if you look at inside there are transmitting and receiving sub-assemblies so-called Cosa and Rosa.
[06:40] let's see what it does inside like a laser diode coupling the light from with the micro optic to the optical circuits and coupling to the fiber and receiving back.
[06:50] what I'm talking about this is these days a lot of optical setups look like this complex highly populated.
[06:58] for instance applications are FMC stabilizer or Quantum Computing or Quantum sensing but those applications you cannot make industrial mass production.
[07:09] so what will be the solution which is a peak alternate interior circuit this one even we can.
[07:16] produce as small as like a few millimeters of dimension.
[07:20] these are why we are here today talk about our micro Optics really cope well to nautical circuits and Optical transceivers.
[07:27] if you just place this Optical circuit or P to this position this can be neutralized as good as Optical transceivers in communication Market.
[07:40] so today I think the peak assembly and testing and packaging I think you could appreciate our micro Optics very well.
[07:46] not only this thing I want to talk about but for the highly integration for example this on the right side the peak has a lot of channels in order to hire integration we need a lot more features like a mechanical feature for instance close tabs spacers precise dicing or metallizations for soldering.
[08:06] all these additional features we can make in way forever Way Forward means we can scale up more than million micro or lens chips per month delivery to you this is what we can do.
[08:19] today for you so that's all thank you.
[08:23] thank you music.
[08:25] perhaps you can move back to our presentation.
[08:31] so I would like to sponsor this Asus micro Optics I don't see anybody from shoes in the room so um well I want to introduce you this company they are a supplier of white quality micro Optics components and they are working in different markets like that on Telecom also sensing of course Automotive markets and semicolon medical.
[08:55] and I just want to finish saying thank you to all our sponsors for supporting these events we see that this picky packaging and then it is very very interesting topic for everybody so we are glad that we do are here today with us and just a final word before going back to even.
[09:14] um for everybody just join as a panelist uh you will receive an invitation so be.
[09:19] Sure that you join to pass a panelists.
[09:21] So you can make questions directly.
[09:22] Participate in the chat and and.
[09:25] Participated later discussion with us.
[09:29] With the first Speaker I think.
[09:32] Yeah thank Antonio for introducing our sponsors today and uh without further delay uh I would I would like to invite Geraldine Lewis the chief commercial officer from uh fix Orthodontics packaging to share us share with us what you have on your mind.
[09:53] Yeah especially since Peter was having a fire alarm right so I'm going first.
[09:59] That's right purchasing his firefighting skills now.
[10:10] Let's see so you can see my screen yeah.
[10:13] You're good to go okay.
[10:15] So thank you very much I would like to give a a small update on like big.
[10:20] packaging motivation Trends and challenges that we are seeing so I already mentioned uh briefly in the introduction uh that we are a spin-off actually from leonex International so a silica nitrite Foundry and they noticed that there's a need to do the volume packaging and ramp it up so we started our operations in 2018 but since then we have been growing as a completely independent back packaging Foundry so if you have an idea then like where we are located in the in the packaging fields what we notice is that not everybody knows what big packaging is about so I have a small animation here to illustrate what are the processing steps that we are required to execute uh before you have a photonic integrated that shortcut which is packaged in a device um and some of the steps that you're seeing here are like the die bonding the the polishing steps the active alignment and six degrees of freedom we need to do accurate epoxy spends on like these small chips especially when you look to
[11:21] Indian phosphide chips that typically have an area of like a few hundred microns by a few hundred microns.
[11:28] You start to appreciate the scale of it.
[11:30] Since here everything looks very big and easy to access but in reality the chip can become very small depending on the application that they're serving.
[11:40] You can actually see like find one of the final steps where we're actually doing the hermetic sealing of a module uh and closing it off doing the wire bonding and then hermetic shielding to make a hermetic hermetic package.
[11:55] If you're looking then like the motivation behind packaging well there are certain areas that we need to improve and certain challenges that we need to overcome and to just provide already like where we what we need to improve as a roadmap is the signal and power distribution is to protection from the devices when you're looking mechanically electrically and hermetically we have the optical performance so the losses that might be like a a very critical target and we
[12:23] need to do thermal management and depending on like which Market you're in.
[12:27] you might be focusing on one product or one area harder than the other and of course what we need to overcome is when people are looking for a better Optical performance.
[12:38] it's like how can we maintain the cost or even lower the cost in the end.
[12:41] the same is when you're adding multiple components in a package.
[12:45] one of the things that you're going to increase as well is or reduce its reliability.
[12:52] so you would like to increase the reliability even though the devices get more and more complex and of course with the higher frequencies the electrical crosstalk and the losses are getting more and more critical throughout the complete design.
[13:08] so to provide some of the market drivers and to liberate with some images of things where we have been doing let's see where we can play a role and what what impact they have on each other.
[13:19] so let's first take the lower cost.
[13:22] so as we are a contract manufacturer it
[13:25] means we like to assemble modules that are designed by our customers.
[13:27] however we realize that a lot of our customers start with the atomic integrate circuits and don't think immediately out of the module.
[13:34] so off the shelf prototype packages are available but for volume a lot of customers they would like to have a customized package that fulfills all of their requirements.
[13:45] and it becomes clear that someone needs to take the responsibility of the product design as well as the the ownership of the products.
[13:53] and of course we can support our customers along the journey but in the end it needs to be clear that the product responsibility is owned by the customers and not by companies that do the contract manufacturing using standardized options to start with.
[14:06] they make sure that the investment in volume scale up tools devices can be pushed forward until a specific product and Market is getting more mature.
[14:19] and then the electronics Industry especially in the consumer Market it's clear that when you need to invest in expensive tooling like malls in order to
[14:27] reduce the part cost in photonics we are typically working with metallic hermetic housings and the cost scaling of that is completely different and it it may lower the cost initially when you're when you're starting but when you're looking through the scale up at the scale down in terms of cost is not similar as it is in the electronics and what we also see is when you do tape house in two different Fabs for multi-project way for prototyping we typically get two different chip formats and therefore also when it comes to the packaging we also require two different printed circuit boards and two different products to get them in so lack of standardization between the foundries and and the chip formats are already limiting us to reuse the designs from one Foundry to another um to eliminate elaborate maybe unlike uh design these devices that are optimized for packaging in early design phases well if we are seeing to to
[15:28] volume production uh we have some customers that are looking to Consumer parts and using 20 Channel spot size conversions including polarization maintaining fiber array we're not seeing that scaling towards consumer applications.
[15:43] so I think it needs to be clear like when you're looking at it have a look at it when you want to reduce the cost what kind of interfaces do you really require if you require a single chain operation maintaining interface don't design a full fiber array with all channels to be polarization maintaining and of course having some margin in the power budget in the end will help you to allow for some manufacturing tolerances and especially more tolerant designs and volume markets is are highly benefiting from this.
[16:11] when you look to markets like Quantum of course this is a no-go area and customers are hoping for hero Parts one time after the other so improved manufacturing tolerances and new designs would overcome this Challenge and boost the loss limits.
[16:28] stuff
[16:33] so
[16:35] for Prototype packaging we have our ROM characterization package standards
[16:40] so these are flexible easy to adapt packages which are configured as a Lego style package and we have a full Suite of Chip dimensions and electrical pad configurations already available
[16:51] and for the RF interfacing the modules can be easily in upgraded with a custom radio frequent interfacing boards and what you see here are two different designs
[17:01] so one is the design that we started scraping off ourselves from scratch
[17:06] and the right one with the blue base that's one that we developed together with any picture Consortium
[17:13] foreign
[17:19] so customers that are using standardized housings for volume packaging uh the direction of their choice is highly dependent on the needs of what they have for the the package and in the end if
[17:30] you want to qualify a product for certain markets or if you're looking to validate your initial chips to take them off the optical table and show them to investors of your potential customers.
[17:40] the advantage of these packages that are already told we keep them on stock so that means there's no minimum order quantity or upfront housing design costs to get started which makes it really an easy way to to do customized prototyping.
[17:57] in order to scale up production of our customers we need to be able to reduce the operator impact on the product cost and in order to perform this we focus on batch level automation on our tool set providing flexible scale up and once volumes are high for a certain Sonic integrated circuits a dedicated production line can be established as it's already done from the Datacom device Manufacturing.
[18:23] so here you see two movies of examples where we do a hybrid integration of a laser diet and bottom a flip chip integration.
[18:33] foreign.
[18:40] Next on the list, the integration.
[18:43] So we are trying to stretch the limits.
[18:45] Of hybrid integration ranging from code.
[18:47] Package optics chiplets hybrid.
[18:49] Integration, heterogeneous integration.
[18:51] And in the example here, you see a demonstrator we built on with multiple chip technologies like Indian phosphide, silicon nitrite, and poly boards.
[19:00] Taking the advantages of each individual chip platform.
[19:02] And due to the high-end assembly processes, the light can travel from one ship to another without degrading the optical signal too much.
[19:11] So this has been done within the EU Tera way projects.
[19:17] And here you see another example showing the hybrid integration of silicon nitrite in combination with silicon stomachs with various DC and RF electrical interfaces.
[19:25] Again, to overcome the challenges on one platform and leverage another platform on its own.
[19:33] strengths and as you can see here this is again using the the flexible prototyping technology that we have but then with a custom printed circuit boards applying the RF signals into the chips.
[19:49] miniaturization it's also good to know that it's not always about miniaturization so on one side yes we're definitely focusing on that uh but what we are seeing is where Electronics is scaling with each node providing smaller and more energy efficient devices.
[20:04] photonics isn't always about making it smaller as an example I have here a giant Quantum processor that we have been built within the Q mode project together with quex quantum here you see a wafer size photonic integrator circuit in the middle of the printed circuit board and it has more than three thousand Thai whiny by bones interface with multiple printed circuit boards to allow for a dense interfacing and underneath this module there's a three
[20:33] kilowatt chip cooler attached to take
[20:36] out the heat that is being generated
[20:37] within the Chip And to provide a thermal
[20:41] stable platform for the Optics
[20:45] yes
[20:48] similar as the example before it's in
[20:51] photonics it may require more complex
[20:53] wire bonding schemes than the electronic
[20:55] counterparts to allow for standard
[20:57] printed circuit board technology so the
[21:00] lack of serializers on the chip and a
[21:01] large number of iOS that are on eye
[21:03] level need to be interfaced to the
[21:05] boards staggered wire bones are an
[21:08] approach to overcome this challenge but
[21:10] are not widely used yet in the industry
[21:18] as we mentioned photomics and electronic
[21:20] scale differently but more than Moore's
[21:23] Law is also applicable here as well so
[21:25] in photonics for example we can go
[21:27] integrate these lenses in this case it
[21:29] was 3D printed I think as a mung Sikh
[21:33] was just mentioning from accentrex you
[21:35] can also make them wafer scale and then
[21:37] you can flip chip them up I think the
[21:39] big Advantage here in this case when we
[21:41] do the printing of the lenses we can
[21:43] reach accuracies with placements with
[21:45] respect to the grating coupler of about
[21:47] 100 nanometer and this is a tolerance
[21:50] that cannot be obtained yet by a flip
[21:53] chip assembly processes today
[21:55] Next Generation next the integration on
[21:58] the thin film ceramic board make sure
[22:00] that the wire Bond density on the
[22:01] phonics can be natively wire bonded to
[22:04] the to the dense thermally conductive
[22:06] interposure and this therefore this
[22:09] design is helping to get more
[22:10] functionality per cubic millimeter
[22:13] instead of focusing on the Square
[22:14] millimeters of the chip design itself
[22:21] modules go up to higher frequencies and
[22:23] where on one side we are discussing with
[22:25] customers that declare 100 megahertz RF
[22:28] on the other side we are stretching the
[22:30] requirements to a staggering 120
[22:32] gigahertz and we have simulation tools
[22:34] like ads and River wire bonding
[22:36] stretching these requirements on the
[22:38] design and Manufacturing sites
[22:43] foreign
[22:46] ERS of course before a product is being
[22:48] accepted for a certain Market thorough
[22:51] tests might be executed to reveal the
[22:53] weak points within a design so test
[22:55] plans for integrated stomachs they're
[22:57] still pretty limited and quite often
[23:00] it's telecommunications and telecordia
[23:02] requirements used in the initially
[23:04] reliability start however one needs to
[23:07] check the applicability for Telecom and
[23:09] Laser since in Telecom a laser needs to
[23:12] function 24 7 for a period of 25 years
[23:16] whereas in automotive it only needs to
[23:18] last for ten thousand hours in operation
[23:21] over a period of 15 years and most of
[23:24] the time your device is actually in an
[23:25] off State
[23:29] foreign
[23:30] so therefore I to conclude the
[23:31] presentation I have some key takeaways
[23:33] so we help our customers in an early
[23:35] stage with packaging roadmap for their
[23:39] products to help them out from a
[23:40] prototype to a qualified product so
[23:42] don't look only to your device you want
[23:45] to have next but more like where you
[23:46] want to be in the end and draft down the
[23:49] whole roadmap from there in packaging
[23:51] and that may help you as well to secure
[23:53] enough funding and and not be surprised
[23:57] in between of the expenses that you're
[23:59] running into of course we try to reuse
[24:02] as much as possible from existing
[24:03] designs to limit scale-up cost and to
[24:06] mitigate design risks and we work on new
[24:09] processes to further stretch the
[24:10] integration possibilities well it's not
[24:13] yet available today may become available
[24:15] as an option for you in the near future
[24:17] but when you require product tomorrow
[24:19] please then start by using a known
[24:22] process so the number of risks that
[24:24] you're stacking into a product when you
[24:26] launch it you would like to limit that
[24:28] as much as possible
[24:30] think of how your product scales
[24:32] what do you require and what components
[24:35] do you require to scale up with a lower
[24:37] cost and polarization maintaining fibers
[24:39] as you can see today they remain rather
[24:42] expensive even though the volume
[24:43] requirements will go up quite a bit in
[24:46] the future
[24:47] photonics does have similarities and
[24:49] difference with respect to electronics
[24:52] and I think we just need to be aware of
[24:54] them since they made advantages or
[24:56] disadvantages for your products
[25:00] with this I would like to conclude my
[25:02] presentation I'm open for any questions
[25:03] thank you
[25:06] thank you very much to Ryan that was a
[25:08] great opening presentation
[25:09] and yes I would like to remind you that
[25:11] you can we have a possibility to ask
[25:14] questions to all of our speakers just
[25:16] use the raise hand button at the bottom
[25:18] of your screen
[25:20] and do we have any questions in the
[25:22] audience
[25:26] in fact during there was a you showed us
[25:30] a large variety of packages from from
[25:33] the ones we're used to all the way up
[25:36] the vehicle scale packages I'm wondering
[25:38] uh do you what are the trends that you
[25:41] see for photonic packaging is the
[25:43] variety of different packages keeps
[25:46] growing or are we going towards the
[25:48] standardization
[25:49] I think what we are seeing clearly is
[25:51] that the variety for certain packages is
[25:54] uh decreasing so I think Telecom and
[25:57] data comb and since they're using phonic
[25:59] integrated circuits for a longer time
[26:01] you see a stronger converge I think when
[26:04] I look to other markets like lidar
[26:06] Quantum medical I see them diverging
[26:09] even more and more since they have
[26:10] completely different needs so for
[26:12] Quantum people are very happy to pay a
[26:15] few Euros more as long as you're
[26:16] focusing on the Optimal Performance that
[26:19] you can get out of a device whereas for
[26:21] medical you're looking to consumable
[26:23] parts and it needs to be as low cost as
[26:25] possible and it doesn't need it it now
[26:28] does need to be reproducible but it
[26:30] doesn't need to last typically for like
[26:32] tens of years we are working on quite
[26:35] some of these applications that require
[26:36] consumable parts so they need to work
[26:39] out of the package you use them once and
[26:41] then you discard them and so they have
[26:43] completely different yeah requirements
[26:45] when it comes to packaging and scale up
[26:48] right right it strongly depends on
[26:50] application yeah all right
[26:53] um we have a question from the audience
[26:55] from Carol from Portland
[26:57] nice overview uh I like that
[27:02] um on your last slide you showed an
[27:03] intriguing picture of an epoxy package
[27:07] yeah so far has been as far as I know at
[27:09] least fairly Limited in in this industry
[27:12] is that something that you see
[27:14] progressing or is it still rather
[27:16] experimental well what we are seeing is
[27:19] a
[27:21] hematic packages are the de facto
[27:24] standards within the Telecom industry
[27:26] because of the long lifetime that they
[27:28] can withstand but other markets may be
[27:31] open for new approaches so if you're
[27:33] looking to automotive and also depending
[27:35] on which chip platform you're using
[27:37] using LCP or polymer packages might be
[27:40] like a much better way to scale up to
[27:43] reduce the cost if you're sticking to
[27:45] the materials we had so far so we are
[27:47] not changing the bill of materials then
[27:48] the cost will not scale uh in in the
[27:51] volume so having injection molding
[27:53] solutions that will definitely help and
[27:56] we need to work with our customers to
[27:57] figure out can we use semi-hematic
[27:59] packages for a lot of applications and
[28:02] is that good enough
[28:04] thanks
[28:07] all right and Marco from agns
[28:11] please go ahead hi
[28:13] hi Marco uh
[28:16] I have a question regarding the the
[28:18] supply chain so I agree with you this
[28:21] processes are so special that these are
[28:24] perfect uh way to start and as a niche
[28:28] into the market when this is combined
[28:31] with the uh when when Co package optic
[28:34] comes there are also conventional flip
[28:38] chip assembly so flips upon substrate
[28:42] smt mounting and so on yeah how these
[28:46] two will somehow play together yeah
[28:48] we'll fix support both
[28:52] so flip chip smt and the final module
[28:56] basically
[28:57] or do you see in the future also that
[29:00] the photonics might Remain the niche
[29:03] where is to play and then the rest will
[29:06] be done by all sets like a CM core or
[29:09] jset wherever it is
[29:11] yeah I think there's multiple ways in
[29:14] there so we play a role so we have flip
[29:16] chip equipment and we do a lot of first
[29:18] level packaging so for example a chip to
[29:21] a interposer so that you have a
[29:23] redistribution layer that can go to a
[29:25] standard BGA approach and customers like
[29:28] that we can do like a fiber attached so
[29:30] basically then you have a BGA with fiber
[29:33] pre-attached and then a office or
[29:35] existing oset can take this as a module
[29:38] and solve it or integrate it into the
[29:40] the complete system so I think people
[29:42] are trying to figure out like where are
[29:44] we going to do what's to build a
[29:46] complete device when it comes to chiplet
[29:49] stacking so we also do a lot of stacking
[29:51] of like Electronics directly on top of
[29:54] photonics and then like on the
[29:55] redistribution layer VGA package so it's
[29:58] really like getting the different
[29:59] functionalities in the first level
[30:01] package
[30:03] I understand now
[30:05] thank you yeah welcome
[30:08] micro Optics also have a question for
[30:11] the room so Wilfred you can proceed
[30:22] do you need to understand
[30:29] can you hear me now now it's okay okay
[30:33] um I uh thank you German for that very
[30:36] broad presentation so you showed this 3D
[30:39] printed lenses earlier yeah
[30:42] um what is your experience with 3D
[30:44] printed lenses because
[30:47] we know that they're not yet the
[30:49] equivalent of of micromanufactured micro
[30:52] etched uh micro fabricated just by
[30:56] etching or by imprint so this is our
[30:58] experience from 3D printed lenses do
[31:00] they give you sufficient good
[31:01] performance that you can have a
[31:03] representative
[31:04] setup for for your clients then or so I
[31:07] think what we are seeing with the the uh
[31:10] the 3D printed lenses the biggest
[31:12] advantages you can play with different
[31:14] Optical designs throughout the gift your
[31:17] journey so what we are seeing is like a
[31:20] multi-project way for run depending on
[31:21] where you're getting it from may have
[31:23] some differences even across the chips
[31:25] that you're getting so having 3D
[31:27] printing as a option is then really
[31:30] helpful for the prototyping at least to
[31:32] find your optimal lens configuration
[31:34] when it goes to volume production we are
[31:37] not seeing that this 3D printing will
[31:39] play like a a big role the only thing
[31:42] where it will be very important for is
[31:44] making the Masters that can be then be
[31:46] replicated actually for wafer level in
[31:49] printing so that you're using basically
[31:50] a nano imprint lithography setup to
[31:53] replicate it over the complete wafer
[31:56] okay and and the quality you've seen so
[31:58] far is that sufficient so coupling or
[32:01] optically the quality that we are seeing
[32:03] is really staggering so uh we have
[32:05] lenses but that we are delivering
[32:07] printed on fiber rash so either for
[32:10] collimation or for a parallel Optics to
[32:13] do testing of atomic integrated circuits
[32:15] and that works out very nice also for
[32:18] wave level testing and on the other side
[32:20] we do printing for example on top of
[32:22] silicon stomachs to get a collimated
[32:24] bundle out of a silicon photonics chip
[32:26] which is also working very well if
[32:29] you're looking to the the quality of the
[32:31] lenses but especially also to the high
[32:33] through positioning you're getting with
[32:35] respect to the grading
[32:36] okay yeah thank you
[32:40] all right thank you thank you
[32:43] um just a very last question from my
[32:45] music accessories
[32:49] thank you for mentioning about the way
[32:52] forever micro Optics from US public but
[32:56] uh I have an additional question
[32:58] regarding to this 3D printed Optics how
[33:01] about reliability you mentioned the
[33:03] Azure decoder test usually polymorphics
[33:06] is not recommended tacodia testing
[33:08] schemes so how is your experience with
[33:11] that so I think for a multi-project way
[33:14] for packaging it's really about
[33:15] prototyping first validating the
[33:18] performance of your chip and and testing
[33:20] that so that's where we see that these
[33:22] lenses are being utilized a lot right
[33:24] now second step is when you integrate it
[33:27] into a product and I think for a product
[33:29] when it's fully hermetic these lenses
[33:32] can be utilized as well I think the
[33:34] biggest barrier is typically the
[33:36] moisture or the energy density and those
[33:38] are kind of tests that we are actually
[33:40] executing also today to figure out like
[33:42] well what is the reliability level that
[33:45] these lenses have and they protect it
[33:46] but that's all in a unprotected
[33:48] environment
[33:50] thank you I agree about that yes thank
[33:54] you
[33:55] all right start with a very intense
[33:57] discussion today thank you again Jerome
[34:00] and let's uh for the second speaker who
[34:04] actually was supposed to be the first
[34:05] speaker to the today so Peter O'Brien
[34:08] the head of research group in the Jindal
[34:10] Institute and the director of the
[34:12] photonic packaging pilot line picks up
[34:15] Peter is floor is yours
[34:23] so can you see my screen I'm going to
[34:26] share it yes perfect okay thanks
[34:29] everybody uh nice to uh see you all
[34:31] again today apologies about the uh
[34:34] confusion we had a fire alarm so we had
[34:36] to evacuate the building but fortunately
[34:38] it's all okay so um very interesting
[34:41] talk from fix and
[34:43] um fixer a partner in our pilot line and
[34:45] I'm going to talk a little bit about our
[34:47] pilot line but more around the area of
[34:48] Technology sustainability so one of the
[34:51] key kind of uh requirements are
[34:54] activities in in the pilot line funded
[34:57] by the European commission a number of
[34:58] years ago is that we're a sustainable
[35:00] financially but also technologically and
[35:03] I think this is really important because
[35:05] um there was a question about the future
[35:06] so where we see packaging going is it's
[35:10] it's still very much in change and you
[35:13] know uh looking at a lot of the
[35:16] packaging it's done part by part what we
[35:18] call package level
[35:20] um and really where we want to go to is
[35:22] Wafer level much like What's Done in the
[35:24] the microelectronics world so developing
[35:27] a sustainable technological approach in
[35:30] the future is very very important
[35:32] um it's not fully there yet but we are
[35:34] working on it and I just want to talk
[35:36] about what we're doing and we're in our
[35:39] pilot line and related projects and to
[35:42] to enable that so
[35:44] um just very very quickly and we you may
[35:47] have seen these slides before so pixap
[35:49] is a distributed pilot line we have
[35:51] multiple partners across Europe I'm
[35:53] continuously growing and one of those
[35:55] important industrial scale of Partners
[35:57] is fix who have kind of been with us
[35:59] from the start and
[36:01] um grown and increased our capacity and
[36:04] technological capabilities and Jerome
[36:06] mentioned about some of the pilots are
[36:08] the Prototype packages we worked
[36:10] together to develop those so when you
[36:13] want to look at evaluating devices at an
[36:15] early stage rather than investing in
[36:18] Tooling in very expensive packages you
[36:20] can use the picks up fairly standard
[36:22] prototype packages with a range of
[36:25] Technologies very very straightforward
[36:27] and as I say the way it's organized is
[36:30] we have a development kind of service
[36:33] activity primarily located here at
[36:36] Tyndall where we look at engagement with
[36:38] users we give training very important
[36:40] and we also do the high risk development
[36:43] and early stage prototyping and then
[36:45] we're industrial Partners like fix and
[36:47] others like Argo Tech as well and with a
[36:49] few companies in that space we transfer
[36:52] and projects or scale processes with
[36:55] them we also provide those companies
[36:58] with access to equipment that we might
[37:00] have that they may not have available
[37:02] for example ficon Tech are now located
[37:04] with us here at Tyndall so they have
[37:07] pretty Advanced machines and we can work
[37:09] to optimize machines or give access to
[37:11] our industrial Partners you know test
[37:13] before they invest so they can make a
[37:15] good choice of equipment and that's very
[37:17] very important
[37:19] a very important part and I think the
[37:21] key part of what we've done over the
[37:23] years in Pixar has developed up
[37:24] standards they're not formal standards
[37:26] but really around design rules and
[37:29] that's so important because quite often
[37:30] and we still see people coming with
[37:32] chips that we've had no M hand apart and
[37:35] designing at least giving advice so how
[37:39] you lay out your Optical electrical and
[37:41] very important in RF so I was mentioned
[37:43] the previous speaker around ORF
[37:45] packaging so it's very very important to
[37:48] be able to configure devices so this is
[37:49] a 3D stacking of IC driver IC on a
[37:52] silicon photonic chip with DC on RF
[37:55] interfaces and we have design rules
[37:58] around that and we're expanding and
[38:01] bringing those to not just photonics but
[38:03] also co-packaging electronics and even
[38:05] mems integration as well for things like
[38:07] lidar time of flight and so forth so and
[38:10] the design rules are a key part of what
[38:12] we've what we're doing
[38:13] but moving forward
[38:15] um as I say much of packaging is done
[38:17] part by part so even with gold boxes so
[38:20] if you want to scale up your process you
[38:22] basically either pass more devices to
[38:25] your machine and you try and kind of
[38:26] shave off time to reduce the processing
[38:28] time or else you just buy more machines
[38:30] and uh you know with the volumes that
[38:33] we're beginning to see hundreds
[38:35] thousands tens of thousands even
[38:36] millions of Parts scaling rapidly that's
[38:39] a real challenge it's actually a big
[38:41] problem so we need to move more to wafer
[38:43] level so I feel like we can learn a lot
[38:47] from the mems industry so what you see
[38:49] on the right hand side is pretty much a
[38:50] mems type package but
[38:53] um kind of adopting a similar approach
[38:55] where much of the packaging is done at
[38:56] the wafer level even hermetic ceiling is
[38:58] done at the wafer level and then you get
[39:01] Optical access you know how well by by
[39:03] certain means and I think somewhat
[39:05] bonding fibers to packages re to chips
[39:08] is not viable because it's really based
[39:10] on a package level approach so how do we
[39:13] get around how do we move forward to
[39:15] this more wafer level kind of assembly
[39:18] so um
[39:20] what you see on the left hand side is a
[39:21] pretty standard approach and we've been
[39:24] talking about this for some time on the
[39:26] right where we're integrating micro
[39:27] lenses onto the chip now in this case
[39:29] you see a micro lens onto the edge of
[39:31] the Chip And then collimate and
[39:33] expanding the beam and refocusing it
[39:36] into a fiber array
[39:37] and actually that fiber array doesn't
[39:39] have to be an expensive glass fiber
[39:41] block it can actually be molded so you
[39:42] can make a molded micro lens array with
[39:45] receptacles for ribbon fiber and and um
[39:49] you know essentially make a very low
[39:50] cost type solution now this is still at
[39:53] chip level so because the device has to
[39:55] be singulated that in a second where
[39:58] where we're moving towards and what
[39:59] we're looking at doing in one or two of
[40:01] our kind of ancillary projects related
[40:03] to picks up so we're bringing in these
[40:05] new technologies hopefully to scale them
[40:07] up
[40:08] so historically we've been working
[40:10] before where we're using grazing
[40:12] couplers and grazing couplers are
[40:13] fantastic although there's limits to
[40:15] what they can be used for so you can see
[40:17] we made these homemade connectors and we
[40:20] connected uh you know a plugable chip uh
[40:23] with this fiber array now it's not a
[40:26] molded part it's a it's a glass fiber
[40:28] block with micro lenses but again you
[40:30] can see the principle we're kind of
[40:31] separating the uh the fiber from the
[40:34] actual package so we're kind of moving
[40:36] more to kind of chip or even wafer level
[40:39] and just to show you where that kind of
[40:41] Technology does have an immediate fit so
[40:43] this is a bio biosensor so using
[40:46] evanescent sensing and this is work
[40:47] we've done with some Partners iMac and
[40:49] University again robots group where we
[40:52] um looked at biosensing on a silicon
[40:54] silicon nitride platform and again you
[40:57] can see the principle of the expanded
[40:59] beam into the micro lens array and air
[41:02] detection so
[41:04] um basically evanescent sensing
[41:06] um and using photo detectors on chip and
[41:09] and you can see the electrical Pogo pins
[41:11] which basically uh sends the electrical
[41:14] signal but the idea here is that this is
[41:16] a disposable chip so you can pick it up
[41:18] put it in and because the beam is
[41:21] expanded you've got fairly relaxed
[41:22] alignment tolerances so again you're
[41:25] using the expanded beam so it shows you
[41:27] a nice application immediate application
[41:29] for that expanded beam approach where
[41:30] you're not bonding the fibers are
[41:32] putting a laser on the biosensor which
[41:34] is disposable ultimately it's also
[41:36] important to mention we do have a design
[41:38] kit around that so the there is a design
[41:40] rule so everything we do is around
[41:41] design rules just to show you again the
[41:44] principles here you can see the micro
[41:46] lens with the chip and the fluid X I
[41:48] won't dwell on this too much but that
[41:49] actually works quite well and we have
[41:51] with some of our industrial Partners
[41:53] like our argotek developed a scalable
[41:55] process so it reduces the barrier for
[41:57] companies to be able to
[41:59] um you know invest in an early stage
[42:01] kind of prototypes to look at see if
[42:03] microfluidics and integrated photonics
[42:05] can be used to replace other types of
[42:07] sensing Act
[42:09] um kind of modes
[42:11] but um I showed earlier around using
[42:13] grading copers so moving on then to
[42:16] something a bit more kind of relevant a
[42:18] bit more up-to-date let's call it Edge
[42:20] coupling
[42:21] um with large mode sizes what we've done
[42:24] in our planet line because one of the
[42:26] big challenges we see is getting access
[42:28] to sufficient numbers of chips so
[42:31] um you may have heard me mentioned
[42:33] before around reference chips and these
[42:35] are basically very simple Optical or
[42:38] electrical kind of air chips with
[42:40] different types of loopbacks Etc a whole
[42:43] array of packaging structures in this
[42:46] case it's just simple Optical iOS and
[42:49] but we make lots of them we made a large
[42:51] number of wipers and made thousands of
[42:53] chips
[42:54] um and essentially just use them for
[42:56] packaging tests so for example companies
[42:58] like FICA Tech can use them then for
[43:00] machine installs we don't need to use
[43:01] valuable product chips so these approve
[43:05] very very successful and in this case
[43:07] we've used these chips these are Linex
[43:09] based chips with a 10 Micron to try
[43:11] complex platform 10 Micron mode adapter
[43:13] and I'm using chips from Zeus off the
[43:17] shelf and we wanted to demonstrate this
[43:19] plugable type connector so you can see
[43:21] some component details here
[43:24] um and what we've done in this case is
[43:25] uh put the micro lens at the edge of the
[43:27] chip expand the beam and again use this
[43:30] uh homemade connector part so you can
[43:32] see the fiber and this is actually where
[43:35] we're aligning the chip so we're
[43:36] aligning the actual micro lens to the
[43:38] edge of the chip with the loop back so
[43:40] light in measure the light out and
[43:42] maximize the coupling efficiency and
[43:43] we've demonstrated that and you can see
[43:46] here a working demonstrator you can see
[43:47] the coupling efficiencies so when we
[43:49] apply the epoxy and these by the way
[43:52] these chips are diced they're not
[43:53] polished
[43:54] um and because of the index when we
[43:56] apply the epoxy we get pretty good
[43:58] coupling efficiencies
[43:59] so you can start to see now we're moving
[44:02] away from bonding the chip to the
[44:04] faucets or sorry bonding to the fiber
[44:06] array directly to the faucet
[44:08] um and uh making a plugable connector
[44:11] what's also interesting here is that the
[44:13] connector here is not the actual in this
[44:16] demonstrator here it's not the glass
[44:17] block it's actually a molded part it's a
[44:19] molded micro lens array in a um in a in
[44:23] a connector so you can start to see that
[44:25] the price or the cost of this type of
[44:27] module can be driven down
[44:30] so I mentioned earlier about design
[44:31] rules so one of the one of the
[44:33] challenges in this particular case is
[44:35] aligning the micro lens so you need a
[44:37] fiber input the light and connect the
[44:39] light back out and align the align the
[44:41] actual micro lens onto the chip what
[44:43] we've done here now we've introduced a
[44:45] new design rule so these alignment wave
[44:47] guides by
[44:48] um Illuminating from the packaging
[44:50] machine itself you can align the fiber
[44:53] and using these active wave guides here
[44:55] so we can just illuminate and see where
[44:58] the beam is collimated and then all the
[45:00] active wave guides in between are
[45:02] aligned basically it's just
[45:03] self-aligning the active wave guides and
[45:05] we've developed some chips around that
[45:07] so you can see we haven't put the light
[45:09] source on the machine itself but we're
[45:11] using these are just red to pilot light
[45:13] pilot illumination
[45:15] um lights but essentially all you need
[45:17] to do is spot illuminate the grading
[45:19] couplers so you can see the grading
[45:20] couplers here we can quickly find those
[45:23] to Edge and then align the actual
[45:25] intermediate
[45:26] um blue black waveguides for testing as
[45:28] you can see this kind of simple design
[45:31] rule enables the actual micro lens to be
[45:34] quite simply packaged to the edge of the
[45:36] chip
[45:36] but as I said earlier we're trying to
[45:38] move to wafer level so what we're
[45:40] looking at now on in a project called
[45:42] photonically so one of our partners
[45:44] Source micro Optics we're looking at out
[45:46] of playing coupling so using micro
[45:48] prisms and to couple autoplane and then
[45:51] use the micro lens to collimate the beam
[45:54] so we're just starting on this process
[45:55] but you can see these components for
[45:57] example have been developed by Wilfred
[45:59] and the team at sus so using integrating
[46:02] the actual micro prisms outer plane
[46:04] coupling with a with the micro lens on
[46:07] the opposite side of the actual chip you
[46:09] can collimate the beam the beautiful
[46:11] thing here is that it enables 2D
[46:13] coupling and you don't need epoxies
[46:15] anymore you can do solder attached so uh
[46:18] replace solders also it's on the surface
[46:21] so it's much easier from an alignment
[46:22] and the Machine Vision point of view so
[46:24] that's kind of where we're working
[46:25] towards
[46:26] on top of that photonic leap is also
[46:29] looking at the electrical package and it
[46:31] was interesting to see the low-cost type
[46:33] of molded lead frame because we're
[46:35] thinking along the same lines so moving
[46:37] away from the gold box looking at
[46:39] surface mount type packages which are
[46:40] very low cost and standard in the
[46:42] electronics World in this case we're
[46:44] using glasses the the interposer base
[46:46] and also using um true true areas so
[46:50] filled copper veers both for electrical
[46:52] contacting but also thermovies so you
[46:56] can see the concept here where we're
[46:58] using the BGA platform for electrical
[47:00] powering at the bottom and Optical
[47:02] outputs so again using that type of
[47:04] outer plane coupling approach so we're
[47:07] just we're about over a year into the
[47:09] project and we're starting to develop
[47:10] these with some of our partners in glass
[47:12] you can see the true Vias the RDL the
[47:15] redistribution layers the micro Optics
[47:17] integrating lasers Etc this will be done
[47:20] at wafer level it's the big Advantage
[47:22] here then is one of the things we're
[47:24] really trying to get towards is be able
[47:26] to do multi project packaging runs so
[47:28] enabling companies to cost share at an
[47:31] early stage and then to increase the
[47:32] volume as they need to go forward it
[47:34] also enables wafer level testing as well
[47:37] so this is important because it enables
[47:39] us to look at new technologies moving
[47:41] away from the package level approach to
[47:43] more wafer scalable approach and that's
[47:45] very important for future pilot scale
[47:47] packaging so I'll leave it at that and
[47:49] open it to questions thanks
[47:53] thank you thank you very much Peter for
[47:55] a very inspiring talk
[47:57] sure there will be many questions now
[48:00] this is the audience
[48:02] this is your chance
[48:06] Android go ahead
[48:09] Peter thanks for this excellent overview
[48:11] of where we can go in the future with
[48:13] wafer level packaging I was actually
[48:16] wondering when you're looking to these
[48:18] interposers what kind of resistances are
[48:20] you getting with these Vias because I
[48:22] think like when you're looking to
[48:24] printed circuit boards or like Elite
[48:26] frames the the resistances are pretty
[48:28] low but I think for Vios it can be quite
[48:30] different is it
[48:31] well that's a good point because
[48:33] actually what we're trying to do here is
[48:35] two things so we're looking at the
[48:36] thermal as well as the electrical many
[48:38] Vias are normally just um how can I put
[48:41] it they're not filled veers they're
[48:43] coated around the kind of perimeter of
[48:45] are they kind of the inside or the
[48:47] outside of the vehicle so just the wall
[48:48] of the vehicle what we're doing here is
[48:51] filling the vehicle so it'll drop the
[48:53] resistance significantly and also drop
[48:55] the thermal resistance so generally what
[48:57] we're looking at is B is around say 50
[48:59] Micron diameter and we're working around
[49:02] changing the pitch so
[49:04] um yeah it's a very valid point so
[49:06] filling the videos with copper is where
[49:08] we're going to reduce the resistance
[49:10] both electrically and thermally
[49:12] thank you
[49:15] now Ahmed from CMC
[49:18] yeah thanks Peter for the presentation
[49:21] um I want to ask about you know you
[49:23] talked about you know the way for a
[49:24] level packaging for the you know the
[49:26] micropism and then the lens
[49:29] um how is that done in a way for level
[49:31] like um because I'm guessing like you
[49:33] you need to align first the prism and
[49:35] then place the there's two approaches
[49:38] there's two approaches where you
[49:39] actually have an independent or separate
[49:41] micropism so you you basically drop the
[49:44] micropism into a shallow it's not a
[49:46] fully etched so for example many
[49:48] foundries now etch the facet they
[49:50] actually maybe to 100 microns and then
[49:52] they they have this ledge they dice so
[49:54] the process is already in existence you
[49:56] can etch the facet which you're not
[49:59] etching a full-fast addressing is
[50:00] essentially a small pocket around the
[50:02] waveguide the prisms typically are in
[50:04] the order of say 50 microns that's
[50:06] typically so that could be done at wafer
[50:08] level you you basically for example you
[50:11] can transfer print the prism you can
[50:13] pick and place these in and then
[50:15] separately placing micro lenses on top
[50:17] of that well we've been working with
[50:19] Zeus on is integrating all of those the
[50:21] prism and the micro lens in a single
[50:23] unit so these can be pick in place and
[50:26] dropped into these these smaller air
[50:28] pockets okay interesting okay thanks
[50:30] yeah and as I say
[50:32] um because it's at a surface level it's
[50:34] possible to use a um a solder gold tin
[50:38] so you're bonding the the but the base
[50:40] of the the micro lens prism assembly to
[50:43] the top of the Silicon for example and
[50:45] you can use laser assisted Reflow okay
[50:48] so generally approach eliminating epoxy
[50:51] okay so what's the tolerance like the
[50:53] placement tolerance of those
[50:54] requirements around the similar
[51:01] enough or even better
[51:03] that could be done pick and place at
[51:04] that um level is not really a challenge
[51:07] and it's easier because it's on the
[51:09] surface not on the edge I agree yeah
[51:12] okay thanks Peter thanks
[51:14] yeah that does sound like an interesting
[51:16] approach the next question next question
[51:18] is from Amir from CSM and Peter this is
[51:22] uh was usually a very nice uh uh you
[51:24] know talk and I always enjoy it actually
[51:27] sort of uh for this you know vision of
[51:28] way for a scale uh so do you think that
[51:30] you know sort of for Edge coupling H
[51:32] tracets are enough or how to get away
[51:35] from like you know polishing at that
[51:37] racer scale well you know the edge is
[51:39] usually it's not doesn't need to be
[51:41] polished
[51:42] um there are other approaches like
[51:44] membrane couplers and Global foundries
[51:47] of a membrane coupler I think AMF have a
[51:49] similar type of membrane coupler there's
[51:51] different approaches actually that's one
[51:52] of the challenges we face because uh I'm
[51:56] sure fixed see the same Challenge and
[51:58] there's lots of different couplers
[51:59] there's lots of different
[52:01] um Dimensions there's different mode
[52:03] sizes uh so there's lots of choices and
[52:06] I think we need to kind of maybe start
[52:09] to reduce those and consolidate the
[52:11] number of our designs so as I say um you
[52:15] don't need to polish you don't always
[52:16] need to polish so you think the edge
[52:18] facet would be enough to give a sort of
[52:22] a reliable and repeatable okay
[52:29] all right next question Tobias from
[52:31] okay all right next question Tobias from
[52:31] genotics please
[52:33] um yeah thank you um my question would
[52:36] be so Peter you mentioned uh that
[52:39] um with these um yeah
[52:43] um prison and a lens approach you could
[52:45] enable wafer level testing of uh Edge
[52:48] coupling devices so uh but that would
[52:52] require that you
[52:53] um would come you know yeah would need
[52:56] to assemble the complete wafer uh before
[52:58] wavel level testing or would you require
[53:00] a wafer level test before uh to save
[53:02] some parts or or the thing here is that
[53:06] you could do you do all the essentially
[53:08] the key packaging before the testing
[53:19] because if you put the Optics on at
[53:21] wafer level you can now have Optical
[53:23] access
[53:26] you basically just need a collimated
[53:27] light source to basically uh collimate
[53:30] the beam much like the connector plug so
[53:33] you scan across the surface
[53:37] okay thanks okay I will just take uh one
[53:41] last question from Frank Frank focus
[53:44] light Technologies please
[53:45] uh thanks Peter I wonder is it possible
[53:48] to assembly using the cross the
[53:51] electrical fac accessory
[53:55] the witch
[53:56] the you know if you have fac and one
[54:00] side another side of its ICC array so
[54:03] it's a cross structure is it possible to
[54:06] achieve the assembling by this kind of
[54:08] structure
[54:11] um sorry I I'm finding hard to visualize
[54:13] exactly what you're what you're asking
[54:15] it's a also kind of a micro less array
[54:20] but on one side is the FSA it is a
[54:24] cylindrical structure another side is
[54:27] just like a I say say array so it's a
[54:30] crossed
[54:32] I I don't know if you get it okay so
[54:35] you're talking about correcting for
[54:36] astigmatism is it with two cylindrical
[54:38] lenses crossed yes yes yes but it's a
[54:41] monolithic
[54:42] um to one structure yeah possibly but
[54:46] normally
[54:48] um you well you possibly could one of
[54:49] the challenges um and Wilfred knows all
[54:51] about this you need an epoxy on
[54:53] potentially on one side so if you need
[54:56] epoxy to bond the cylindrical lens and
[54:59] Main approach may not work but you can
[55:01] put epoxy dots for example
[55:04] um on outside the active Optical area
[55:07] then you could use cylindrical lenses
[55:09] yes
[55:09] okay thanks
[55:13] all right uh thanks uh to everyone uh I
[55:16] think we can continue this conversation
[55:17] for much longer but uh please stay with
[55:20] us after the end of the event we have a
[55:22] good time to discuss longer uh thanks
[55:24] again Peter and uh with that I would
[55:27] like to pass to the next speaker of
[55:29] today uh Joshua Zimmer sales manager
[55:32] from nanascript
[55:37] okay thank you
[55:39] uh can you see my screen
[55:42] very good thank you so
[55:45] um yeah let's dive right into it uh I'm
[55:48] talking a little bit let's say lower
[55:50] Heights of a flight we are a machine
[55:53] manufacturer and we build 3D
[55:56] microfabrication systems such as the
[55:59] quantum X align system that you can see
[56:01] here on the left our systems are 3D
[56:04] printers for polymer micro lenses that
[56:06] have been discussed already a couple of
[56:08] times today and so we use two Photon
[56:11] polymerization as opposed to
[56:13] Conventional UV polymerization or
[56:16] lithography we use a femtosecond near
[56:18] infrared laser and this gives us a high
[56:21] degree of control over the position and
[56:24] the size of the polymerization that
[56:27] takes place our systems uh use the Laser
[56:30] Source and then a couple of galvo
[56:32] mirrors for a beam positioning and then
[56:36] we go through a microscope objective
[56:38] that focuses the laser light right into
[56:40] the sample volume where the 3D printing
[56:44] takes place in case of the quantum xl9
[56:46] we have two sensor systems that allow us
[56:50] accurate sensor so accurate detection of
[56:54] features like markers and automatic
[56:57] alignment to that so this allows the
[56:59] printing of polymer micro optics for for
[57:02] example among other structures and here
[57:05] we have years of process development
[57:08] experience which allows us to print
[57:11] lenses with a reliable repeatable high
[57:15] quality Optical surfaces as well as
[57:18] finding a compromise in terms of
[57:21] printing speed so some surfaces that are
[57:23] not optically active may have a reduced
[57:26] quality
[57:28] with this you can print directly onto
[57:32] photonic chips at the chip or wafer
[57:34] level in principle and in in such a
[57:38] situation we use the confocal module so
[57:41] one of the detection mechanisms here
[57:43] which looks at the back reflected light
[57:45] through the beam path and this allows us
[57:47] to get a 3D position detection of
[57:51] features on the photonic chip for
[57:52] example alignment markers from previous
[57:55] process steps or chip edges so waveguide
[57:59] Corners things like that and now with
[58:02] the exact knowledge of the position and
[58:04] the angle of the chip you can print
[58:08] coupling structures micro lenses for
[58:10] example aligned in position and angle to
[58:14] the chip so just one quick example here
[58:17] would be these lenses that were printed
[58:20] onto an edge
[58:22] emitting set of of a light here this
[58:26] could these are wave guides this could
[58:28] be
[58:29] um lasers or light sources or something
[58:31] like that and as you can see here down
[58:34] in this comparison between the
[58:36] simulation and the measurement so here's
[58:39] the the lens and we look at the light
[58:42] intensity and you can see there's an
[58:44] excellent match between the simulation
[58:46] and uh the the measurement meaning that
[58:48] we have understood and can predict and
[58:51] can create the right beam profile as
[58:55] requested
[58:57] you can also print on fiber uh so single
[59:01] fibers or Fiber arrays and
[59:04] um so in in that case the setup would be
[59:07] looking something like this for example
[59:09] with a fiber and a v-groove holder in an
[59:12] array and here we don't have alignment
[59:15] markers so instead of the confocal
[59:17] module we use a camera system and fiber
[59:20] core illumination from the other side of
[59:22] the fiber to allow automatic detection
[59:26] of the fiber core and we can also look
[59:29] at the emission direction to also align
[59:32] to the to the emission direction of the
[59:34] fiber and in such a situation again just
[59:37] one example you would automatically
[59:40] align to and print on all of the cores
[59:43] in such a fiber array and again we show
[59:46] that we have a good match between the
[59:48] the simulated beam path and and the
[59:52] measured beam profile a lens fiber
[59:55] arrays like these are available through
[59:57] our partner fix who also spoke earlier
[01:00:01] in the session so if you just contact
[01:00:02] them or go to the web shop you can get a
[01:00:05] set of fiber arrays
[01:00:08] printed with lenses with our technology
[01:00:10] on that
[01:00:12] so just briefly we characterized these
[01:00:16] and there are two typical situations
[01:00:19] that get asked from us by our customers
[01:00:23] so um either you have either a fiber or
[01:00:26] a wave guide with a printed lens where
[01:00:30] you wanna
[01:00:31] um where you want to collimate the beam
[01:00:33] to a relatively wide beam diameter or
[01:00:35] you want to focus to a pretty small spot
[01:00:38] this has already been discussed
[01:00:40] previously or yeah also that with a
[01:00:45] wider beam and the second set of micro
[01:00:47] lenses you can get relaxed alignment
[01:00:50] tolerances or if you just want to have a
[01:00:52] single lens and a no second lens on the
[01:00:54] other side you could use the fiber to to
[01:00:58] directly match the mode field of the
[01:01:00] target waveguide we characterized both
[01:01:03] these setups so up here we have two
[01:01:06] fiber arrays with lenses printed on them
[01:01:09] and we are testing coupling through one
[01:01:12] 0.2 millimeters of free space in this
[01:01:15] case we designed for and tested with a
[01:01:18] green light 533 Nano is a 32 nanometers
[01:01:21] sorry and here we see indeed a relaxed
[01:01:25] alignment tolerance so we have plus
[01:01:26] minus 5 Micron alignment tolerance for 1
[01:01:29] DB of excess loss and we found 0.7 DB
[01:01:34] loss per lens so two lenses in the beam
[01:01:38] path 0.3 of which per lens are just from
[01:01:41] fresnel losses which could in principle
[01:01:43] be mitigated down here we tested this
[01:01:46] other scenario with a lens fiber array
[01:01:49] directly coupling to a silicon and
[01:01:51] insulator
[01:01:52] a situation and here we designed for and
[01:01:55] tested with 1550 nanometers and we found
[01:01:58] losses of 1.7 DB per coupling interface
[01:02:02] for for this test
[01:02:04] uh one last slide we talked about wafer
[01:02:07] level testing and of course if you can
[01:02:09] build your setup that you already have
[01:02:12] um out of plane coupling then then
[01:02:14] you're good to go for the wafer level if
[01:02:16] you have Edge emitting or a general Edge
[01:02:19] coupling uh chips then then you need to
[01:02:22] somehow be able to look around the
[01:02:24] corner if you want to do it on wafer
[01:02:25] level and for this it would be possible
[01:02:28] to get such a fiber array with
[01:02:31] periscopic lenses printed on them and
[01:02:34] then to to use that for wafer level
[01:02:36] testing like our customers here have
[01:02:39] done and published two years ago Optics
[01:02:41] Express where you have here basically
[01:02:44] your uh fiber array and then these
[01:02:48] periscopes can be dipped into these
[01:02:50] etched grooves on the on the wafer to
[01:02:54] look into these wave guides from the
[01:02:58] side
[01:02:59] that's all I hope there's time for a few
[01:03:02] questions thank you very much
[01:03:04] thank you Josh
[01:03:07] and time for questions
[01:03:14] thanks for the overview I want to ask
[01:03:18] about you know the optical return loss
[01:03:20] you know of these lenses another thing
[01:03:22] is the high power capability like and
[01:03:24] also the scalability right like do you
[01:03:28] align and then you do this process quick
[01:03:30] how fast like for example if you do like
[01:03:32] a fiber array like how how quickly can
[01:03:35] yeah so with the whole process the
[01:03:37] automatic alignment per uh per fiber and
[01:03:41] then the printing process we're looking
[01:03:43] at a couple of minutes per lens so
[01:03:46] typical uh lenses that look in the same
[01:03:49] direction as the fiber would be like two
[01:03:51] or three minutes per lens and for the
[01:03:54] periscopes because they have more
[01:03:56] Optical surfaces and they need to take a
[01:03:58] little longer there it's more like
[01:04:01] um six to nine minutes so this is not
[01:04:04] currently a large scale process but it
[01:04:07] can be used for for small scale
[01:04:09] production and especially for uh for for
[01:04:12] prototyping
[01:04:15] oh okay thanks how about how about the
[01:04:17] return loss do you have any number about
[01:04:19] that you tested for I mean the the
[01:04:21] losses
[01:04:23] um we we mentioned here basically so we
[01:04:25] have here 0.7 DB uh total loss of the
[01:04:29] line going through the lens
[01:04:30] and um some of it is reflection some of
[01:04:33] it is just in transmission so the the
[01:04:36] total loss is 0.7 DB
[01:04:39] um but uh so we
[01:04:42] I don't I don't know the number I had we
[01:04:44] have a couple of tests for a few
[01:04:46] wavelengths of the power please contact
[01:04:49] me afterwards so I can give you an exact
[01:04:51] number okay great thanks
[01:04:55] all right thanks uh if there is no other
[01:04:58] question
[01:05:01] I'd like to pass to the next speaker
[01:05:04] thank you Joshua thank you
[01:05:06] um the next speaker of the session is uh
[01:05:08] Jean Luke
[01:05:09] the CTO and the co-founder of Icon
[01:05:13] photonics Paris
[01:05:15] whenever you're ready
[01:05:21] please unmute
[01:05:30] doesn't look you need to unmute
[01:05:42] yes thank you can you hear me now
[01:05:48] okay thank you and so thanks for uh
[01:05:52] letting me uh giving this presentation
[01:05:55] um so I'm going to uh to tell you about
[01:05:59] um uh icon photonics and our technology
[01:06:01] which is made in polymer to enable the
[01:06:05] packaging of integrated floating
[01:06:07] circuits and for the next generation of
[01:06:10] optical connectivity
[01:06:11] so while actually we the contacts
[01:06:14] actually we have already approached uh
[01:06:17] in this in this webinar uh but I would
[01:06:21] like to insist on on the fact that there
[01:06:24] is a great uh rejection of Dimensions uh
[01:06:26] nowadays that we are looking for and
[01:06:29] actually if you are looking for a
[01:06:31] pluggable Optics you go to co-package
[01:06:34] objects and then soon to 3D cheap plate
[01:06:37] and as we say it actually there is also
[01:06:40] another Trend which is to go to lower
[01:06:43] Dimension to get higher speed and higher
[01:06:46] density but also to to go to the wafer
[01:06:50] level in order to to enable a mass
[01:06:53] production capabilities
[01:06:56] uh that's what we are working on and
[01:07:00] actually I can give also two examples on
[01:07:03] on the subject subject that we are uh
[01:07:05] considering first the data.com where
[01:07:09] your uh targeting very high tax rates uh
[01:07:12] with today one out of 12 224 GBS per
[01:07:16] second and and you are running toward
[01:07:18] higher data rates so sure you can do it
[01:07:21] in parallel uh but also per Channel you
[01:07:24] want to increase your debt rate and in
[01:07:26] in this in this progress the packaging
[01:07:30] is getting more complex and you need to
[01:07:33] simplify it uh and to register cost but
[01:07:36] also to enable higher performance
[01:07:39] when you look at the second example
[01:07:41] which is looking for the quantum
[01:07:43] Supremacy there is another challenge
[01:07:45] there you're trying to have a higher
[01:07:50] number of qubits which is a fundamental
[01:07:52] element of this quantum computer and
[01:07:54] today you have up to 433 qubits what is
[01:07:59] the challenge there is that each qubit
[01:08:01] for example is photo detector which is a
[01:08:05] Supra conductor for the detector cooled
[01:08:07] down at cryogenic temperature and the
[01:08:11] packaging of each individual detectors
[01:08:13] is almost macroscopic Packaging
[01:08:17] um so if you really want to go to larger
[01:08:19] number of qubits and if you want to have
[01:08:22] a useful computer you need that you need
[01:08:25] to Target more than 1000 qubit so you
[01:08:30] really need to shrink down your
[01:08:32] packaging for that and to be also uh
[01:08:35] compatible with a very low temperature
[01:08:37] down to the mili k
[01:08:40] so that that's what we are tackling with
[01:08:44] our solution which is a all-in-one
[01:08:47] package actually driving the photonics
[01:08:49] the electronics and the mechanic
[01:08:51] mechanical integration at wafer level
[01:08:54] so we start actually from the device of
[01:08:57] the client either as a wiffer or as a
[01:09:02] chip and we assemble them with these
[01:09:06] structures that we are providing either
[01:09:09] with surface alignment and there you
[01:09:12] will see in detail a polymer structure
[01:09:16] which is shown there which is providing
[01:09:18] the low loss coupling with tapers taper
[01:09:21] shape which is showing a large area at
[01:09:25] the input and a small area at the output
[01:09:28] where you have your detector so it's
[01:09:31] concentrating the beam
[01:09:33] the larger Ray helps to relax the
[01:09:36] tolerance of alignment of the fiber
[01:09:38] and then with mechanical structure which
[01:09:41] is fixing the the fiber on top of it you
[01:09:44] can really have a first mechanical
[01:09:47] alignment then you finish work with this
[01:09:50] Optical focusing of the light on on the
[01:09:53] taper
[01:09:54] this solution which is shown here in
[01:09:56] surface coupling is also used for the
[01:10:00] edge coupling of the rib of a ribbon
[01:10:02] fiber to to pick to the edge of the peak
[01:10:06] if we go in further detail for one
[01:10:09] example of this structure you can see
[01:10:11] here the different parts which are
[01:10:14] separated first you have a silicon
[01:10:17] interposer on which in which you can put
[01:10:19] the the chip
[01:10:21] and then you can have electrical
[01:10:22] interconnections up to millimeter wave
[01:10:25] then you have the the polymer taper and
[01:10:28] the five or older which is on top of it
[01:10:30] and then you have the fiber here is it
[01:10:32] when it is assembled so you can see for
[01:10:35] example that the fiber door is very
[01:10:37] small and for in that example you have
[01:10:40] Trail fiber which are coupled into this
[01:10:43] fiber order this is a kind of advanced V
[01:10:46] groove so you really align precisely the
[01:10:50] different fibers and at the same time
[01:10:52] you fix it firmly so that when you go
[01:10:55] down to cryogeny you don't have any
[01:10:58] movement in any directions
[01:11:00] the polymer tapers you can see here
[01:11:03] different geometries which are more
[01:11:06] compact or taller depending on the
[01:11:10] geometry that you are targeting for
[01:11:12] example here you see the fiber holder
[01:11:14] the taper and the chip which are bonded
[01:11:17] all together and that is a wafer level
[01:11:20] process except the fiber insertion
[01:11:22] obviously which still is individual but
[01:11:25] can be automatized
[01:11:28] so finally our Solutions we have surface
[01:11:31] coupling Edge coupling and surface to
[01:11:34] Edge also which is upcoming soon and for
[01:11:37] the surface coupling these are the kind
[01:11:39] of geometry you can have from an optical
[01:11:42] aperture which is 70 Micron down to 10
[01:11:45] Micron with MMF fiber multimod fiber and
[01:11:48] then we can enable 0.1 DB coupling
[01:11:51] losses and the same for a single mode
[01:11:55] fiber or multi-core Fiber because by by
[01:11:58] sense this technology is Wafer level so
[01:12:02] you can make an array which is a
[01:12:04] two-dimensional array
[01:12:06] in the edge coupling the coupling loss
[01:12:09] or low again below 0.5 DB and that is a
[01:12:14] very good progress compared to state of
[01:12:17] the art and we provide two solutions the
[01:12:19] Standalone option where you align
[01:12:21] actively the chip to depict the the
[01:12:24] fiber holder to the peak or even a
[01:12:26] passive alignment where we have a
[01:12:28] mechanical clipping options to to
[01:12:31] provide
[01:12:32] foreign
[01:12:34] the polymer that we are using
[01:12:36] I was also listening to to one of the
[01:12:40] questions which is about reliability is
[01:12:42] quite robust polymer and then we can
[01:12:44] also have some process on top of it like
[01:12:46] uh zero air gap option that can provide
[01:12:50] and to reflective coating and and the
[01:12:53] polymer is really reliable so it goes
[01:12:56] from millikay to very high temperature
[01:12:59] in excess of 350 degrees Celsius and we
[01:13:03] also have obviously uh this uh dampede
[01:13:06] test cycling test and that makes the
[01:13:09] lifetime of the device uh
[01:13:12] very uh very acceptable
[01:13:16] so our collaboration model actually
[01:13:18] we're providing design and test for for
[01:13:21] the client fast prototyping as well so
[01:13:25] that we can develop the proper Solution
[01:13:27] on top of the chip of the client and
[01:13:29] provides more serious sampling but we
[01:13:31] can also obviously go to fabrication and
[01:13:33] projection from low to medium volume and
[01:13:36] even high volume and that is really the
[01:13:39] idea of the wafer level process that we
[01:13:42] are using
[01:13:45] so this is the end just to keep in time
[01:13:48] with the uh six minutes given and the
[01:13:51] final question that you are asking what
[01:13:53] we can do for you well we can bring to
[01:13:56] packagers Standalone options to ease the
[01:13:58] coupling so in that way we are very uh
[01:14:01] compatible and and complementary uh to
[01:14:05] their approaches so we can provide
[01:14:07] better performance for for you and ease
[01:14:10] in in difficult conditions or easing
[01:14:13] simply the standard one Solutions but we
[01:14:15] can also bring to uh module makers
[01:14:18] higher performance and yields for for
[01:14:22] their production and from cryogenic room
[01:14:24] temperature to high temperature
[01:14:26] and finally we make it smaller but
[01:14:29] denser and better that's what we can do
[01:14:31] for you
[01:14:32] what we are looking for obviously new
[01:14:34] challenges uh
[01:14:36] just like the quantum computers
[01:14:38] trying to tackle the more than 1K qubits
[01:14:43] or even 10K qubits cycling higher debt
[01:14:48] rates when you need photo detectors to
[01:14:51] be smaller for example and then we can
[01:14:53] really help you to shrink down the
[01:14:55] dimensions
[01:14:57] today we already have early stage
[01:14:59] adopters so we want to to expand
[01:15:02] obviously and to be uh and we are ready
[01:15:05] to cooperate with with a with new client
[01:15:08] and exploiting also on mass volume
[01:15:10] capabilities
[01:15:13] all right
[01:15:15] um thank you John Luke and thanks for
[01:15:18] answering clearly to the Epic question
[01:15:20] right away
[01:15:22] um it is now open for more questions
[01:15:25] from the audience
[01:15:29] in fact I was wondering myself
[01:15:31] um what is the current cycle time of
[01:15:34] your process on a chip and what is the
[01:15:37] forecasted cycle time when you go to
[01:15:40] high volume Manufacturing
[01:15:42] right actually
[01:15:44] um going to the typing to a volume
[01:15:48] production is quite easy actually the ID
[01:15:50] is really because we're a welfare level
[01:15:52] adapting the process on top of your
[01:15:54] wafer and Depends obviously of your
[01:15:57] surface state of the different materials
[01:16:00] you have and so on so uh the the longest
[01:16:04] part is clearly as a prototyping phase
[01:16:07] because we need to to have a common work
[01:16:09] together and it's question of weeks and
[01:16:12] months I would say a few months to to
[01:16:15] make it which is a typical clean on uh
[01:16:18] cycle not so much more but a typical
[01:16:21] clean on process development
[01:16:24] all right all right
[01:16:27] um another question from Ahmed from
[01:16:30] cmc's Microsystems yeah thanks uh just
[01:16:33] to confirm like you showed in your slide
[01:16:35] that you have you know a better coupling
[01:16:37] with the vertical uh coupling scheme
[01:16:41] than the edge coupler right it was I
[01:16:42] think point one and the edge coupler was
[01:16:45] 0.5
[01:16:46] repeats right
[01:16:48] okay because I was guessing like uh if
[01:16:52] you have a grating coupler right there
[01:16:53] will be some mod conversion loss right
[01:16:56] right in the case of grading coupler
[01:17:00] actually uh uh you can still uh
[01:17:03] challenge this this uh 0.5 DB on the
[01:17:07] edge actually here you don't have a
[01:17:09] grading coupler this value for the top
[01:17:12] side uh 0.1 DB is without the creating
[01:17:15] copper is clearly directly to a photo
[01:17:17] detector and the creating coupler you
[01:17:19] will be similar to 0.5 DB
[01:17:22] um there is also an advantage with for
[01:17:24] technology with creating coupler is that
[01:17:27] the creating coupler you need to give an
[01:17:29] angle uh to to to to to to to to the
[01:17:33] beam to to the fiber to couple to it and
[01:17:36] actually we will help to make it
[01:17:38] vertical okay
[01:17:40] so you don't have to bother on the fiber
[01:17:44] placement with with an angle to to put
[01:17:47] it on it and the the taper will will
[01:17:50] manage it for you and that has an
[01:17:52] advantage because when you change the
[01:17:54] temperature for example and you have
[01:17:55] your laser on board of your pick the
[01:17:58] wavelength is changing and when the
[01:17:59] wavelengths changes the angle of mission
[01:18:02] out of the Brack couplers is also
[01:18:04] changing so when you go to Vertical uh
[01:18:09] emission again then you are enlarging
[01:18:12] the bandwidth range of your of your
[01:18:15] grading copper
[01:18:18] right so this is not the grating coupler
[01:18:21] it's uh directly to the photo detector
[01:18:24] right
[01:18:25] the point one is directly to a photo
[01:18:27] detector and creating copper is a 0.5 DB
[01:18:30] okay great okay okay thanks
[01:18:35] um if there is any other question
[01:18:39] and if not uh I think again John Luke
[01:18:43] and uh I'd like to pass the word to well
[01:18:46] okay let's take the last question from
[01:18:48] Amir CSM yeah Yannick this is a great
[01:18:52] talk I mean so I see a lot of
[01:18:53] Investments since the last time we met
[01:18:55] uh this you know so value is point five
[01:18:57] point I mean these are inferior actually
[01:18:59] measure them because uh you know 0.5 to
[01:19:02] single mode it's really really really
[01:19:05] really difficult actually
[01:19:07] it is indeed but that is a real values
[01:19:11] so you measured this it's not like in
[01:19:13] like a Target or simulations no yeah
[01:19:16] okay what's the tolerance
[01:19:19] well actually the tolerance is really uh
[01:19:23] in in that case uh it
[01:19:26] um
[01:19:27] on the fiber placement uh was uh
[01:19:34] good question I can't remember uh top of
[01:19:37] my head like that but I can find it back
[01:19:39] and I will I will send it to you I mean
[01:19:42] I I it can be done repeated like you
[01:19:44] know at the sort of in in with the
[01:19:46] statistic I mean this is the probably
[01:19:48] the best of the best I've seen
[01:19:50] uh really in any method of
[01:19:53] packaging yeah but the Tolerance on the
[01:19:56] uh fiber to the peak alignment is is
[01:20:00] really in the range of 0.1 Micron you
[01:20:02] need to have okay that is really a uh an
[01:20:06] active alignment that can provide uh and
[01:20:09] except you have some mechanical options
[01:20:12] to to make it very specifically on the
[01:20:15] fiber tolerance that is a more relaxed
[01:20:17] obviously uh in case in that case it's a
[01:20:21] few microns that you have in terms of
[01:20:23] Tolerance
[01:20:24] okay okay
[01:20:26] super thank you you'll come
[01:20:30] all right uh thank you John Luke again
[01:20:32] passing to the next speaker and I invite
[01:20:35] Nicolaus Fleury business development
[01:20:38] manager from Warrior Optics to share his
[01:20:40] presentation
[01:20:46] yes hello everybody can you see my
[01:20:48] screen
[01:20:49] yeah you're good to go perfect
[01:20:52] so hello everybody
[01:20:54] um yeah it's been really interesting
[01:20:55] following the first couple of
[01:20:57] presentations they were already actually
[01:20:59] a couple of points which really
[01:21:02] um resonates with me and uh which which
[01:21:04] I can only confirm
[01:21:07] um one of which is that you know in this
[01:21:09] little packaging and assembly schemes
[01:21:11] it's really important to consider the
[01:21:14] different aspects at a very early stage
[01:21:17] of the development and to already take
[01:21:19] into account the packaging
[01:21:21] when you actually design a product or
[01:21:23] designer pick so this is also what I
[01:21:26] want to talk about as you see here in
[01:21:28] the title
[01:21:29] um on Electro Optical co-engineering
[01:21:32] um and topic can give you some
[01:21:34] additional insights here
[01:21:35] so first of all let me uh quickly tell
[01:21:38] you who we are very Optics
[01:21:42] um so we are a spin-up from a PCB
[01:21:43] Manufacturing Company actually
[01:21:45] um called veraprint and we're located
[01:21:47] here uh in the Alps in the in eastern
[01:21:51] Switzerland
[01:21:52] um so what we do is we design and
[01:21:54] manufacture photonic boards so we have
[01:21:57] our own two clean rooms to see one of
[01:21:59] them here in the background where we
[01:22:01] actually produce a manufacturer
[01:22:03] but as you might know in the photonics
[01:22:04] field many things are custom steel so
[01:22:08] that's why we're also active you know
[01:22:10] design and development as well
[01:22:13] um you see here a couple of markets here
[01:22:14] we're really active throughout various
[01:22:17] segments photonic sensing Telecom and
[01:22:20] cheap packaging
[01:22:21] um
[01:22:22] again many photonic parts are really
[01:22:26] agnostic to the application such as pigs
[01:22:29] or the packaging aspects for packaging
[01:22:31] challenges are common to many Industries
[01:22:34] that's why we are also very broadly
[01:22:36] working
[01:22:37] so uh what are these photonic ports that
[01:22:40] I'm talking about
[01:22:42] well I'd actually like to make an
[01:22:44] analogy to the electronics World
[01:22:46] um there I mean everybody's aware for
[01:22:49] the semiconductor chip level
[01:22:50] developments there's a lot of us around
[01:22:53] this at the moment as well but then when
[01:22:56] you merge or when you when you put some
[01:22:58] of these chips into an actual electronic
[01:23:01] system and application we typically go
[01:23:03] the route of having a board a PCB
[01:23:05] printed circuit board which is really
[01:23:08] meant to hold or to to package
[01:23:12] several ICS
[01:23:14] because of course one I see even though
[01:23:16] there is no chiplets and everything
[01:23:19] um even one IC or one chiplet cannot do
[01:23:21] the the entire job of a product
[01:23:24] um so this is really what what the board
[01:23:25] level or the pcbs are meant for
[01:23:27] now in the photonics world we have all
[01:23:30] these cheap level developments as well
[01:23:32] and it's been really interesting and
[01:23:34] exciting because we see a lot of new
[01:23:36] platforms coming up
[01:23:38] um costs are going down functionalities
[01:23:40] are going up so it is really great also
[01:23:43] for the applications and systems
[01:23:45] but still um I would argue that
[01:23:48] um and I took this little picture here
[01:23:49] that a lot of
[01:23:51] um applications are sort of overtaking
[01:23:53] this board level approach not
[01:23:55] necessarily uh this is not necessarily A
[01:23:57] Bad Thing
[01:23:59] actually this is a great package that
[01:24:00] you see here it's a great assembly but
[01:24:02] many applications till now they only
[01:24:04] required as in this picture here the
[01:24:07] required um one for example one fiber
[01:24:10] coupling so the applicant part so to say
[01:24:12] if you look at this package was smaller
[01:24:15] than Electronics right because there's
[01:24:16] still of course an electronic board used
[01:24:20] so what we are working on at the career
[01:24:22] manufacturing uh are the so-called
[01:24:24] photonic boards are meant to sort of
[01:24:28] um view this this missing Gap
[01:24:31] um so you can imagine it's it's it
[01:24:33] serves the same purpose it's a tonic
[01:24:34] board it's it's based on a printed
[01:24:36] circuit board but also has an optical
[01:24:39] layer which can be used to couple and
[01:24:42] package several Optical picks or Optical
[01:24:45] components that's the that's a short
[01:24:48] summary of a photonic report
[01:24:49] now of course as mentioned so we can we
[01:24:53] base on on a standard PCB
[01:24:55] technology which means we can also make
[01:24:58] use of
[01:24:59] all the functionalities which has been
[01:25:01] developed there so we can have firmary
[01:25:04] wires as Peter O'Brien mentioned it
[01:25:06] before you can have a high speed RF
[01:25:09] functionality
[01:25:10] to more than 100 gigahertz is actually
[01:25:13] uh is starting to become a bottleneck in
[01:25:16] driving Peaks as well that you get the
[01:25:19] fast Electronics running
[01:25:21] at the same time and as mentioned and is
[01:25:24] what we really produce is an optical
[01:25:27] layer which is on top of such a board
[01:25:30] um so this is actually a sort of a peak
[01:25:33] platform on its own
[01:25:36] um so a planar waveguide platform fully
[01:25:38] passive which in our case is mostly
[01:25:41] based on polymer material
[01:25:43] and here I'm really happy again
[01:25:45] um we've had a couple of talks already
[01:25:47] about polymer uh in in this webinar
[01:25:50] which which I really like because I
[01:25:52] think it's kind of a trend to see
[01:25:54] polymer Rising
[01:25:56] they might have had a little bit of a
[01:25:58] bet
[01:25:59] a reputation in the past but it turns
[01:26:01] out that there are many materials which
[01:26:04] are actually optically very interesting
[01:26:06] many polymer materials and not only
[01:26:09] optically but also
[01:26:10] environmentally so for example the
[01:26:13] polymer material we are using is the
[01:26:15] Cordia proved so it can sustain high
[01:26:18] humidity high temperatures and speaking
[01:26:21] of temperature we've heard it before in
[01:26:23] other
[01:26:24] presentations other materials as well
[01:26:27] can sustain about 300 degrees which is
[01:26:30] similar for our material
[01:26:32] I stated 270 degrees here because that's
[01:26:35] a standard Reflow soldering temperature
[01:26:38] which our material can undergo if it is
[01:26:41] on top of a PCB
[01:26:45] um now we are actually manufacturing
[01:26:46] both multi mode and single mode
[01:26:47] waveguards with different standard
[01:26:49] processes
[01:26:51] um similar to to any other photonic
[01:26:53] platform uh in the middle of course we
[01:26:56] have also a couple of building blocks of
[01:26:59] passive devices filters couplers where
[01:27:02] things combiners
[01:27:04] Etc so
[01:27:05] um in one in some sense you can also
[01:27:07] think of this platform
[01:27:10] of the opportunity to actually put all
[01:27:13] the let's say boring passive
[01:27:16] functionalities of a pick to the to the
[01:27:18] packaging to the board level because the
[01:27:21] main advantage of course using a polymer
[01:27:23] material is that it's less costly than
[01:27:26] say a silicon platform so you're not
[01:27:28] paying for a square millimeter but more
[01:27:31] per square centimeter you can have
[01:27:33] larger boards and so you can actually
[01:27:35] could also reduce the size of your
[01:27:38] actual
[01:27:39] um High complex pick just limited to the
[01:27:42] let's say active functionalities
[01:27:44] and put all the splitting and um and
[01:27:46] glide routing onto the polymer platform
[01:27:50] but actually what's most important on
[01:27:52] this slide is the right side so the
[01:27:53] integration and assembly possibilities
[01:27:55] that we have
[01:27:57] um as mentioned we typically based on a
[01:28:00] PCP board but we're also working using
[01:28:04] glass or silicon as an interposer as a
[01:28:06] substrate
[01:28:07] which is especially interesting if
[01:28:09] mechanical Precision is required as well
[01:28:15] um here a few pictures so this is how
[01:28:17] such photonic Parts can look like and in
[01:28:20] fact the word photonic boards is maybe
[01:28:22] not so true after all because it turns
[01:28:25] out that in all of these pictures and
[01:28:27] projects that you see here
[01:28:30] um the optic or the photonics part is
[01:28:32] only part of the story so the electric
[01:28:36] mechanical and thermal requirements are
[01:28:39] often quite harder to meet because most
[01:28:42] people already focused on the photonics
[01:28:44] so this is sort of well established but
[01:28:47] it as mentioned it turns out the
[01:28:48] electrical characteristics can be very
[01:28:50] important and can be very hard to meet
[01:28:52] as well and especially the combination
[01:28:54] of all of these four domains
[01:28:57] and this is why again
[01:29:00] I put together a very short list of say
[01:29:03] requirements which we believe are
[01:29:07] necessary for a good or Advanced Peak
[01:29:10] Packaging
[01:29:11] which is first of all what we've heard
[01:29:13] before
[01:29:15] that we that we should really focus on
[01:29:18] co-engineering which on the one hand
[01:29:20] means we should consider the packaging
[01:29:23] right before the peak fabrication
[01:29:26] there are still ways of course to
[01:29:28] package a pick for a copy to pick once
[01:29:31] it's coming out of The Foundry but you
[01:29:33] might not be able to get the best
[01:29:36] um best solution after all
[01:29:38] and the second aspect of the first part
[01:29:41] is actually the second point which
[01:29:42] really means that um you have to
[01:29:44] consider all this different disciplines
[01:29:46] so photonics Electronics thermal and
[01:29:50] mechanics all of this at once which is
[01:29:52] really challenging as well a lot of
[01:29:55] different disciplines are necessary and
[01:29:58] but it turns out that this is also
[01:29:59] really relevant if you want a very well
[01:30:02] performing packaging or SMB process
[01:30:06] and the third part is really what we as
[01:30:08] very Optics believe in uh and that is
[01:30:11] that
[01:30:11] um yeah the the packaging approach
[01:30:14] should be scalable in some way because
[01:30:16] in the end we all want these great
[01:30:18] products and chips to end up in in high
[01:30:21] volume applications and um if you're
[01:30:24] using uh fibers fibers are great and you
[01:30:27] we need it at some point but still fiber
[01:30:30] alignments typically are linear
[01:30:31] processes sort of higher the number of
[01:30:33] ports to hire the number of effort
[01:30:37] whereas for example using our platform
[01:30:39] you can
[01:30:40] structure in one go
[01:30:43] hundreds or you could even have
[01:30:45] thousands of parallel channels
[01:30:47] so it's very nicely scalable in that way
[01:30:50] and I think that's really important in
[01:30:53] the entire assembly process in the
[01:30:55] entire supply chain to keep this in mind
[01:31:00] um now one last example how this can
[01:31:03] look this is actually a sort of special
[01:31:05] case
[01:31:06] out of a new project so what you see
[01:31:10] here in the bottom left is a photonic
[01:31:12] interposer that we designed and
[01:31:14] manufactured which is used for the
[01:31:17] indium phosphide based 100 gigahertz
[01:31:19] wavelength selected switch
[01:31:22] um it nicely combines all these features
[01:31:24] so we have a planar waveguide layer
[01:31:27] which is used to couple uh at the same
[01:31:30] time to 16 Optical channels around the
[01:31:33] light has been out inside
[01:31:35] Etc
[01:31:37] um you could actually also do entreat
[01:31:39] mode conversion so you can optimize the
[01:31:41] coupling to the pitch as well as to a
[01:31:44] fiber for example
[01:31:46] then of course Optical interfaces are
[01:31:48] very important we use typically are Edge
[01:31:51] coupling or adiabatic coupling
[01:31:55] um to the peak side as well as Edge
[01:31:56] coupling to the fiber side
[01:31:59] um the electric connections
[01:32:03] um are done here on the Silicon
[01:32:05] Enterprise itself and later on by upon
[01:32:07] it to the PCB as you see on the right
[01:32:09] picture
[01:32:10] and actually so these big wires in this
[01:32:12] prototyped are red and black one are for
[01:32:15] our text controller which are used to
[01:32:17] thermally stabilize this device
[01:32:21] so this is what I want to show uh to you
[01:32:24] um overall we're always looking for
[01:32:26] partners and collaborations which can
[01:32:29] uh either help us expand our technology
[01:32:31] add some building blocks but of course
[01:32:34] also um
[01:32:36] for partners which can make use of such
[01:32:38] a platform so I'm thinking of
[01:32:40] applications here that have a high Port
[01:32:43] count uh multiple inputs or outputs
[01:32:47] Optical both applications which need
[01:32:50] um high speed electrical interfaces
[01:32:54] and overall applications which which
[01:32:57] want to scale to to higher volumes
[01:33:00] so what we can provide is is what I've
[01:33:02] shown here uh these photonic parts of
[01:33:05] photonic interports which I think are
[01:33:07] one way really to consider if you want
[01:33:10] to end up with a high performance and
[01:33:13] also scalable product and application
[01:33:16] I'm happy to answer questions or feel
[01:33:18] free to reach out if this somehow
[01:33:21] resonates with you thanks a lot
[01:33:25] that's indeed a very interesting
[01:33:27] platform and a solution for photonic
[01:33:30] integration Nicholas it is now open for
[01:33:34] questions from the audience
[01:33:36] just use the racing button
[01:33:40] all right and the first question from
[01:33:42] Amir yeah it's very good uh project uh
[01:33:45] Nicolette actually sort of uh he
[01:33:47] mentioned about this you know putting
[01:33:48] these passive components on the board I
[01:33:50] mean uh why like you know sort of it's
[01:33:53] not like how long is the for example the
[01:33:56] direction coplayer we are not saving any
[01:33:57] area
[01:33:58] maybe after chip but the whole product
[01:34:01] would be a lot bigger
[01:34:06] what is the bending radio
[01:34:09] um so the bending radius is about
[01:34:13] um minimum brightness radius between two
[01:34:16] and five millimeter depending on
[01:34:17] wavelengths and process this is a node
[01:34:20] um you're right you're actually not
[01:34:22] decreasing the size but you're reducing
[01:34:24] the cost right and this is actually a
[01:34:26] main driver in for many applications and
[01:34:29] you're reducing the cost because you're
[01:34:30] increasing the area but on a low cost
[01:34:32] platform right on the polymer platform
[01:34:34] as compared to having a larger silicon
[01:34:37] Protonix chip or an Impulse fight or
[01:34:39] whatever
[01:34:40] that's what I mean
[01:34:47] thank you anyone else
[01:34:52] oh I think you mentioned any calls but I
[01:34:54] was just wondering again so what are the
[01:34:57] the working temperatures of this polymer
[01:35:00] but uh so you can sustain uh more than
[01:35:02] 300 for short times
[01:35:04] um more than 100 degrees for for
[01:35:07] continuous operation
[01:35:10] um and the answer is standard decoderia
[01:35:12] uh specifications so 85 85
[01:35:17] for long long hours
[01:35:19] okay so the most important range it can
[01:35:22] withstand all right
[01:35:25] um thanks again for sharing uh this and
[01:35:28] um I'd like to pass to the next speaker
[01:35:31] um it will be Joshua para the research
[01:35:34] and development engineer from ficon deck
[01:35:37] Joshua
[01:35:39] yes so good afternoon everyone and thank
[01:35:42] you very much for having me here
[01:35:45] um so my name is Josue para I work for
[01:35:48] ficon tech service
[01:35:50] although I'm based in Ireland I'm in the
[01:35:53] branch of fight which is a falcon Tech
[01:35:55] Ireland we are based in Cork inside the
[01:35:59] Tinder National Institute
[01:36:05] so when we do our five contact but
[01:36:06] already this slide has been shown and at
[01:36:09] the very beginning of this meeting
[01:36:11] but it just to
[01:36:13] um explain a little bit more is that at
[01:36:16] five contact we already have more than
[01:36:18] 20 years of experience working or
[01:36:20] developing automated solutions for the
[01:36:24] alignment
[01:36:26] assembly and also testing and of course
[01:36:28] even qualifying some of the of these
[01:36:32] assembles and um so we provide a
[01:36:39] as far as or as much as I know the very
[01:36:43] good Edge in terms of
[01:36:45] um
[01:36:46] assembly different components which can
[01:36:48] be used in let's say
[01:36:50] um flip chip and we can assemble
[01:36:54] um different platforms of
[01:36:56] integrated circuits like in this case
[01:36:58] can be from the normal semiconductor
[01:37:00] electronic
[01:37:02] integral circuits but our main
[01:37:06] um
[01:37:07] let's say area of expertise is now in
[01:37:10] the area of photonics so we are dealing
[01:37:12] directly with the photonic interior
[01:37:13] circuits which are like based on the
[01:37:16] platform of silicone silicon nitride and
[01:37:18] in your phosphide so we can assemble the
[01:37:20] lasers we can assemble different Peaks
[01:37:22] and then we can even
[01:37:24] do the alignment of different Optical
[01:37:27] components like in this case can be the
[01:37:29] let's say normal single fiber can be a
[01:37:32] fiber array can be now the even a fire
[01:37:35] with a with the with the lenses can be
[01:37:39] my
[01:37:40] and Michael Lansing rice as well and
[01:37:43] then once that we we are able to align
[01:37:45] all of these very with very high
[01:37:47] Precision like at the stock market range
[01:37:50] we can also attach them so we can use
[01:37:53] different techniques like using a UV
[01:37:55] curing also some thermal compression
[01:37:58] like in the case of slip chip and as
[01:38:01] well as laser assisted bonding
[01:38:04] and as part of today's meeting apart
[01:38:08] from talking about the assembly we also
[01:38:11] are going to be addressing what is
[01:38:13] testing which in this case not only we
[01:38:15] can provide solutions for testing at a
[01:38:18] single or even package level which is
[01:38:20] pretty much the same but also at the
[01:38:23] wafer level so we can also provide some
[01:38:25] of these same Solutions
[01:38:27] as well as we have some other systems
[01:38:30] like laser welding and on inspection
[01:38:32] some sorting of components and some bar
[01:38:36] let's say about stacking
[01:38:39] so at five contact and when I say we
[01:38:41] provide different solutions also it
[01:38:43] depends on the needs of the different
[01:38:45] customers as it can be as a research
[01:38:48] institute also can be like maybe
[01:38:51] companies who have like r d areas or
[01:38:55] departments where they want to
[01:38:57] um investigate what is the best way to
[01:39:00] package their prototypes so we can
[01:39:04] provide Solutions like in this case on
[01:39:07] the left hand side you can see in this
[01:39:08] in on this slide we have like we can
[01:39:11] provide lab systems that are more like
[01:39:12] for r d very simple very accessible from
[01:39:15] the different size sides of the of the
[01:39:17] of the machine but also we can scale up
[01:39:19] depending on the needs of the customers
[01:39:21] to something that is more like
[01:39:23] production-like system and even if it
[01:39:26] requires a major production we can jump
[01:39:29] towards what is the production automated
[01:39:32] system and the very
[01:39:35] um last picture is on the right hand
[01:39:37] side is showing what I want to talk a
[01:39:39] little bit later on that is an inline
[01:39:42] assembly of different machines in order
[01:39:44] to assemble even more complex different
[01:39:48] components like more like systems and
[01:39:50] what is important here
[01:39:52] and it's something also what if I can
[01:39:54] take a
[01:39:55] is the best way we can actually provide
[01:39:58] a good service and solution for for
[01:40:01] assembly and testing to our customers
[01:40:03] like it's very important that we
[01:40:05] understand the requirements of our
[01:40:07] customers
[01:40:08] so understand the process not only the
[01:40:12] process let's say even step by step of
[01:40:14] what has to be done but also it's
[01:40:16] important to understand what are the
[01:40:18] difficulties and sometimes our our
[01:40:21] customers are facing when they are
[01:40:22] trying to do that even at a let's say a
[01:40:25] research and development
[01:40:27] um stage
[01:40:29] so in this case we we had we designed
[01:40:33] our machines for
[01:40:34] a manufacturing a step like
[01:40:39] and in terms of this um meeting which is
[01:40:42] for packaging and testing something that
[01:40:46] um
[01:40:47] one of the tools that we have developed
[01:40:49] recently at five contact that um we
[01:40:52] already are using
[01:40:53] um cameras in order to visualize the
[01:40:55] different components that we have on our
[01:40:57] machines
[01:40:58] and so we can develop the tooling that
[01:41:01] is necessary to handle them but also in
[01:41:03] order to precisely
[01:41:06] um assemble these components
[01:41:10] um I mean we have already systems that
[01:41:11] they have like a the solution of a
[01:41:14] accuracy as 70 between 50 to 70
[01:41:17] nanometers but also we can use that and
[01:41:21] let's say even for attaching like in
[01:41:23] this case I'm showing this slide and
[01:41:25] when we have a a component that is a
[01:41:28] silicon and then we want to for instance
[01:41:32] place a three five material like in this
[01:41:34] case could be like a laser
[01:41:37] and then using fiduation we can use our
[01:41:40] cameras like in gas cameras
[01:41:43] and and then um see to the silicon and
[01:41:48] see the feed which is unprecisely I'll
[01:41:50] align this
[01:41:51] this two components for instance
[01:41:54] I mean this is one of the tools that
[01:41:56] I've recently we've been adding to our
[01:41:57] machines
[01:41:58] apart from the orders that we have from
[01:42:00] many many years of experience
[01:42:03] and even combining these IR Vision we
[01:42:08] can even include
[01:42:09] and some of the
[01:42:12] latest ordering assisted as well so we
[01:42:14] can not only visualize from the bottom
[01:42:18] side but also we can
[01:42:20] and provide some Laser Source you know
[01:42:22] lasers a beam in order in order to bond
[01:42:25] the components
[01:42:28] so I'll go for my next slide
[01:42:30] in terms of testing and we also are
[01:42:33] providing some solutions and for not
[01:42:37] only at the Die Level but also
[01:42:40] at a wafer level so as you can see in
[01:42:43] the picture
[01:42:45] um that is on the left hand side let me
[01:42:47] see if I can yes here obviously this
[01:42:49] picture I mean this picture for sure you
[01:42:51] have seen in many other presentations
[01:42:53] for 500 and in this case we come from
[01:42:57] the top so that would be needed some
[01:42:59] creating couplers but also
[01:43:02] another thing that has been already
[01:43:05] mentioned in this meeting is the
[01:43:06] possibility of using this trip the uh
[01:43:10] printed structures so Periscope like
[01:43:13] structures and where we can access
[01:43:17] um the components from the from the edge
[01:43:20] but in this case
[01:43:23] um it's needed this on wafer trenches so
[01:43:27] we can then access the side of the
[01:43:29] components so there are different
[01:43:30] possibilities that can be achieved with
[01:43:32] already let's say with the with our
[01:43:34] machines and even for that as you can
[01:43:37] see we were working already with Banker
[01:43:40] photonics even Banger photonics they
[01:43:43] offer the machines for doing the
[01:43:45] printing of the of the periscopes lenses
[01:43:48] and this photonic wire bonds and they
[01:43:51] are even using let's say our platforms
[01:43:53] of the of our machines
[01:43:55] so it can be either that somebody gets a
[01:43:59] banger photonics machine which already
[01:44:01] even has this server that we have for
[01:44:03] fighting deck or even
[01:44:04] and let's say get the the polymeric
[01:44:07] components like in this case the lenses
[01:44:09] and the and the
[01:44:11] and the periscopes and then maybe attach
[01:44:14] these components to our machines
[01:44:17] and then we can I mean the customers can
[01:44:19] access and let's say their devices at
[01:44:23] the wafer levels as well using our
[01:44:25] systems
[01:44:27] and something as well that um it's
[01:44:30] important to mention is like right now
[01:44:31] we are talking about maybe
[01:44:34] at some point like it sounds more like
[01:44:36] an r d research study of course it's
[01:44:39] still ongoing because still we are
[01:44:41] lacking of different standards for let's
[01:44:44] say not only for assembly and testing
[01:44:47] but also for the for the machines that
[01:44:49] they have to perform all of these
[01:44:51] activities but what is important as well
[01:44:54] to consider is like
[01:44:55] the production itself so there are
[01:44:58] companies they they already need to
[01:45:01] fulfill certain number of deliveries of
[01:45:02] different with the current Technologies
[01:45:04] so at the moment and we offer also the
[01:45:08] possibility of
[01:45:11] let's say assembling different machines
[01:45:13] so they in order to really do different
[01:45:17] assembly processes like we can hear
[01:45:20] fitting the components then they go
[01:45:22] along the different machines that are
[01:45:25] interconnected
[01:45:26] and with different
[01:45:28] um assembly processes and also as you
[01:45:31] can see here not only assembly it can be
[01:45:33] like maybe to capture read this time
[01:45:36] analyze data
[01:45:37] and then even put laser soldering
[01:45:40] whatever is needed
[01:45:41] and sorting anything and then at the end
[01:45:45] we just get the product
[01:45:47] and in addition to that what is
[01:45:49] important as well for companies who are
[01:45:51] interested in
[01:45:53] a let's say ramp up their their
[01:45:55] manufacturability is that
[01:45:58] also we offer apart from the normal
[01:46:01] software that we have at the moment and
[01:46:03] with machines
[01:46:05] complementary machines that they can
[01:46:07] sorry software that that can even at
[01:46:09] real-time monitor what is happening in
[01:46:12] the process already of the assembly
[01:46:13] process I mean that's the testing
[01:46:15] process so we use some machine learning
[01:46:18] and the AI so we can then complement
[01:46:21] this
[01:46:22] and
[01:46:23] as the assembly process go along even
[01:46:26] the customer gets information of how to
[01:46:28] improve the process how to make it
[01:46:30] faster because even it is possible to
[01:46:32] correct and rectify if there's a problem
[01:46:35] with assembly the acid is already in
[01:46:38] line going with the with the first
[01:46:40] production so it can be a rate of
[01:46:43] feedback and then improve the process as
[01:46:46] it goes along
[01:46:49] um so for instance this already has been
[01:46:51] used for the pollution line of big
[01:46:53] companies uh producing Transit
[01:46:55] transceivers cameras and online
[01:46:57] applications
[01:47:01] also well if I can take apart that we
[01:47:03] are always interested in improving in
[01:47:05] terms of technology for assembly and
[01:47:07] testing we also
[01:47:10] they are aware that we have to provide
[01:47:12] very good service to aware customers
[01:47:15] and in order to do that we are now
[01:47:18] opening more different branches like in
[01:47:21] this case the branch in Ireland in order
[01:47:24] to take to our customers as close as
[01:47:26] possible where they are around the globe
[01:47:29] somewhere some some Labs some some point
[01:47:32] of contact
[01:47:33] where maybe they can approach it easier
[01:47:37] rather than being concentrated in one
[01:47:39] place like in this case the headquarters
[01:47:41] are in in Germany
[01:47:43] and then we can help them to find the
[01:47:46] best solutions for their assembly and
[01:47:48] testing
[01:47:49] so um
[01:47:51] thank you very much and if you have any
[01:47:53] questions feel free to ask now or even
[01:47:55] later feel free to reach out thank you
[01:47:57] very much
[01:47:59] thank you Joshua great presentation I'm
[01:48:01] sure there are many people in the
[01:48:03] audience who already use your machines
[01:48:04] thank you so uh we have a time for just
[01:48:07] uh one or two questions
[01:48:09] please go ahead raise your hands
[01:48:14] let's start with uh Marco from hns
[01:48:18] yeah hi Marco here uh are you thinking
[01:48:21] uh with this machine provide a full
[01:48:25] Electro Optical testing capability or
[01:48:29] some other equipment must be seen in
[01:48:31] order to fully test the device
[01:48:34] that's a very good question and thank
[01:48:36] you very much for asking that because
[01:48:38] let's say
[01:48:39] these machines that we already providing
[01:48:42] they have capabilities for doing the
[01:48:44] optical and electrical testing however
[01:48:46] as let's say the complexity of the
[01:48:49] different systems increase
[01:48:51] also we have to find out or investigate
[01:48:53] what a new ways of doing things like in
[01:48:55] this case something important that Peter
[01:48:56] Ibrahim was mentioning is about this the
[01:48:58] assembly like placing these micro lenses
[01:49:00] and then it is possible to reach out on
[01:49:03] the wafer and with an optical probe but
[01:49:06] also it's important as well to be able
[01:49:08] to
[01:49:09] um access also from the bottom and to
[01:49:11] power up all the different components on
[01:49:13] a wafer for instance so in this case
[01:49:15] even Peter o'bri mentioned about this uh
[01:49:18] project photonic leap and where if I can
[01:49:20] take this to one of the partners in fact
[01:49:22] the Falcon that we are working along
[01:49:24] with the Technical University find
[01:49:25] einhoven with silver ladkowski and shigo
[01:49:28] Congo and we are actually working
[01:49:30] towards the development of better
[01:49:33] probing let's say hybrid probing
[01:49:36] um at the wafer level
[01:49:37] um Solutions so in this case exactly
[01:49:39] what you're mentioning like being able
[01:49:41] even to
[01:49:42] test optically analytically and let's
[01:49:45] say components on a wafer
[01:49:50] okay thank you
[01:49:53] I hope that answers your question Marco
[01:49:55] um anyone else
[01:49:59] because if not I would suggest with that
[01:50:03] we'll move to the next talk thanks again
[01:50:05] Joshua
[01:50:07] um thank you last speaker of the session
[01:50:09] as we moved from packaging all the way
[01:50:12] to testing is a Richard Fury the
[01:50:15] director at yellow
[01:50:17] and then we're very glad to have you
[01:50:19] Richard okay
[01:50:21] can you see my uh presentation there
[01:50:25] yeah you're good to go okay thanks
[01:50:28] um so thank you uh holding everybody
[01:50:31] back for the end of their afternoon
[01:50:33] um so I work for a company called yellow
[01:50:35] uh we manufacture burning and life test
[01:50:39] equipment
[01:50:40] um
[01:50:41] the
[01:50:46] uh primary business is the designer
[01:50:49] manufacturer of systems
[01:50:51] um typically I mean we've got three main
[01:50:55] ranges
[01:50:57] um a low power less than one amp burn-in
[01:51:00] system uh covering about a thousand
[01:51:03] devices
[01:51:05] um
[01:51:06] low power summary less than one number
[01:51:09] covering 5000 devices
[01:51:12] um the main difference between these two
[01:51:14] is that the y1000 offers air cooling
[01:51:18] only and the y4000 also offers water and
[01:51:22] Tech Cooling
[01:51:24] and then we also do a high power system
[01:51:26] which goes up to 300 amps and needless
[01:51:30] to say that is water cooled
[01:51:33] so those assistants we manufacture
[01:51:36] but looking at what we actually do I
[01:51:40] mean the first question is what is life
[01:51:42] test
[01:51:44] um so life test is what you would do
[01:51:46] normally at the early stage of device
[01:51:50] development where you're trying to prove
[01:51:53] the long-term reliability of your your
[01:51:56] devices
[01:51:57] it's sometimes also used as parallel
[01:52:00] wafer qualification so you normally do
[01:52:03] live tests when you're developing
[01:52:05] products but you might do an ongoing
[01:52:07] sampling process
[01:52:10] from Wafers that ties in a bit with what
[01:52:13] Peter was saying where you might put
[01:52:15] some test devices around the edge of the
[01:52:17] wafer
[01:52:18] just to to qualify the way for
[01:52:22] it's also used to find a sweet spot when
[01:52:26] you're going to do production burning
[01:52:27] what is the optimum current and
[01:52:31] temperature to age a device that you
[01:52:33] want to age it quickly so you have a
[01:52:35] short burn in but not so quickly as to
[01:52:39] damage through other mechanisms than the
[01:52:42] normal failure mechanisms
[01:52:45] um life tests typically will run
[01:52:47] anything from two thousand dollars to 8
[01:52:50] 000 depending a bit little bit upon upon
[01:52:52] the product
[01:52:54] foreign
[01:52:55] on the other hand was it might use the
[01:52:58] same equipment is used as a production
[01:53:02] technique to identify
[01:53:06] faulty or early life failures
[01:53:11] so
[01:53:12] in Burnham what you do is you stress the
[01:53:15] device
[01:53:16] electrically and thermally
[01:53:19] and again based upon what your earlier
[01:53:22] studies are you will have chosen a
[01:53:24] temperature
[01:53:26] very rarely above 150 degrees C
[01:53:29] um and we'll line anything from about 85
[01:53:33] to 150 you may also over stress in terms
[01:53:37] of the current Drive which can also
[01:53:39] speed up the aging process
[01:53:43] what you're trying to do is to make the
[01:53:47] failures
[01:53:49] uh due to the the manufacturing process
[01:53:54] of the laser diode in the first place
[01:53:57] um uh occur much earlier in the life
[01:54:00] cycle and particularly before you've
[01:54:02] spent money Packaging
[01:54:05] so that is a principle of burning
[01:54:09] um burning times anything from two hours
[01:54:14] to 168 hours depending upon what the
[01:54:17] product is
[01:54:19] um how sensitive it is I mean if it can
[01:54:21] stand a lot of abuse you can get a
[01:54:24] shorter burning
[01:54:27] the problem about burning is when should
[01:54:30] you do it
[01:54:31] um you start at a wafer level and you
[01:54:34] land up with products in the field
[01:54:36] uh
[01:54:37] um
[01:54:39] if you do your bun in at two earlier
[01:54:43] stage let's say you do it at wafer level
[01:54:46] you may not capture some of the failure
[01:54:50] modes if you do it too late when you
[01:54:52] spend all your money hermetically
[01:54:54] packaging it you've thrown away a lot of
[01:54:56] money in that process
[01:54:59] um
[01:55:00] typically we find that at the chip on
[01:55:03] Carrier stage so sometime after it's
[01:55:06] been singulated but before it's been
[01:55:09] stuck into something more expensive
[01:55:13] um is the optimum time for burning
[01:55:17] um we are looking at way for level
[01:55:19] probing for burning um it's slightly
[01:55:21] different from what phyton tech do
[01:55:23] because you have to
[01:55:26] because you're aging the devices you
[01:55:29] need to be driving many of them at the
[01:55:30] same time you can't do single devices at
[01:55:33] the same time you need to be doing at
[01:55:35] least 5 000 in in a group to make it
[01:55:38] worthwhile
[01:55:39] um and even then you can have to step
[01:55:41] and repeat across a wafer
[01:55:46] we have a big project doing that at the
[01:55:48] moment but there's a say that the
[01:55:51] challenge I keep on going around is
[01:55:55] um
[01:55:56] obviously some of the
[01:55:58] um failure modes are related to the
[01:56:01] original laser architecture but some are
[01:56:04] also related to the subsequent processes
[01:56:07] the singulation
[01:56:09] um the bonding down onto a carrier the
[01:56:11] while bonding can all potentially
[01:56:14] introduce failures
[01:56:17] um and I have to be I have to say I
[01:56:20] don't know many people that know exactly
[01:56:22] what on their failure modes are
[01:56:25] so that's a Whistle Stop tour I was
[01:56:29] conscious I was at the end of the line I
[01:56:31] didn't want to keep everybody hanging on
[01:56:32] too long
[01:56:34] um we're always interested in
[01:56:36] collaboration
[01:56:37] um I think we may be certainly talking
[01:56:38] to Icon photonics and Nana scribe I
[01:56:41] think who have things that they can
[01:56:42] offer us
[01:56:43] um uh and indeed FICO Tech as well
[01:56:47] um but we're also interested I think
[01:56:49] again as Peter said at the beginning
[01:56:51] the earlier you look at how you're going
[01:56:54] to test things the better
[01:56:57] um somebody comes along to us and
[01:56:59] they've already decided what their what
[01:57:02] that break card is going to look like
[01:57:04] and what their pads are going to be and
[01:57:05] you say well that's going to double the
[01:57:07] cost of of doing your burning
[01:57:10] um so early talking about your process
[01:57:12] is important and that's me
[01:57:17] representation thank you very much
[01:57:19] Richard I'm sure this will raise a few
[01:57:22] questions so it is now open for the
[01:57:26] audience please use the resend button
[01:57:32] I was uh I was actually Antonio
[01:57:37] yeah I I have a curiosity because some
[01:57:40] some of these diets are using
[01:57:42] applications like medical or space Are
[01:57:44] there specific regulations to make these
[01:57:47] kind of tests uh for the components that
[01:57:49] are using this uh let's say quite strict
[01:57:52] applications
[01:57:54] as far as I'm a nurse at the moment I
[01:57:57] mean tell codia and have obviously in
[01:58:00] the Telecom side
[01:58:02] um well established you know processes
[01:58:06] and protocols
[01:58:07] it doesn't seem to be as well
[01:58:10] established in the other fields like
[01:58:12] medical Etc although it should be but
[01:58:14] but it isn't at the moment
[01:58:17] okay thank you
[01:58:21] okay anyone else
[01:58:26] Richard I was wondering uh what is the
[01:58:28] typical uh cycle time for a burning test
[01:58:32] um
[01:58:33] typicals probably around 12 hours would
[01:58:37] be a typical cycle would be to
[01:58:40] do an lav of the device at 25 degrees C
[01:58:47] um raise it to 85 to 100 for about 10 to
[01:58:53] 12 hours then bring it back down to 25
[01:58:57] and do a final Liv and then you
[01:59:00] typically look well obviously get dead
[01:59:02] devices and that's easy but um most
[01:59:05] people look at the shift in ith as being
[01:59:08] a good measure of of device performance
[01:59:12] all right all right that could be
[01:59:14] possible to do it both at the at the
[01:59:16] wafer level and then later at the cheap
[01:59:18] level right yes I mean you you could do
[01:59:22] it you you could do a burning at wafer
[01:59:25] level which might exclude a lot of the
[01:59:28] um semiconductor issues
[01:59:31] um but would exclude some of the process
[01:59:34] manufacturing issues
[01:59:36] um but you might then get away with a
[01:59:37] much shorter burn-in at the subsequent
[01:59:40] stage
[01:59:41] um
[01:59:44] okay okay so that's very clear well um
[01:59:47] we're unfortunately already the running
[01:59:50] out of time as it's already 5 PM so okay
[01:59:54] I would like to thank you Richard again
[01:59:56] for the presentation as well as all the
[01:59:57] other speakers of today's uh seminar
[02:00:03] um just a few final words before I close
[02:00:08] this session
[02:00:17] so uh we're very glad that you joined us
[02:00:20] for this epic online technology meeting
[02:00:22] on big packaging and then testing we had
[02:00:25] the pleasure to hear uh seven
[02:00:27] presentations today from my Kindle
[02:00:29] Institute from six packaging from Nana's
[02:00:32] crib icon subtonics as well as vario
[02:00:35] Optics bike and Tech and yellow
[02:00:40] I'd like to remind you that this is a
[02:00:42] very important one but not the last
[02:00:45] otm uh that we're planning online uh
[02:00:49] here you can see a list of upcoming
[02:00:51] events organized by epic and just in a
[02:00:54] week from now we'll have all the similar
[02:00:57] seminar on the earth observation and as
[02:01:01] well in December there is an online
[02:01:04] technology meeting dedicated to
[02:01:06] photonics for food and beverages
[02:01:09] we'll start also next year 2023 with the
[02:01:12] online technology meeting on meta
[02:01:14] materials and meta lenses and followed
[02:01:16] up by laser-based manufacturing for
[02:01:19] implants and later in February there is
[02:01:21] also bio-imaging topics and Quantum
[02:01:24] Metrology and Quantum sensing that kick
[02:01:26] in
[02:01:28] um I'll take also the opportunity to
[02:01:31] thank the team of Epic and remind you
[02:01:35] that epic is always happy to support you
[02:01:37] with technology consultations market
[02:01:40] reports opportunities for networking and
[02:01:44] accessing new markets mentorship and
[02:01:47] photonics HR Support and investment
[02:01:51] uh with this um well just would like to
[02:01:54] just uh thank our our sponsors again for
[02:01:58] uh supporting this event
[02:01:59] and uh we're 800 members already at Epic
[02:02:03] so we're always welcome to join us
[02:02:05] thank you very much
[02:02:08] and goodbye
[02:02:11] thank you even goodbye thank you bye
[02:02:13] everybody thank you