# EPIC Online Technology Meeting on VCSEL Manufacturing and Applications

https://www.youtube.com/watch?v=P-0McKKp4lA

[00:00] are we ready enough her finger is up her
[00:02] thumb is up
[00:03] my face is matching 650 nanometers and
[00:07] the color of my tie thanks to the
[00:08] beautiful sun
[00:09] we are having here in the netherlands
[00:11] and almost all over europe today we are
[00:13] global
[00:14] and we are talking about my favorite my
[00:16] favorite topic in photonics
[00:18] we are talking about pixels
[00:21] you probably know me by now you know how
[00:24] crazy i get
[00:25] when anyone speaks about pixel's
[00:27] vertical cavity surface emitting lasers
[00:30] because this is one of the biggest
[00:32] things to ever happen in photonics so
[00:35] far
[00:35] and you can quote me on that but what's
[00:38] next
[00:38] i believe it is time to rethink the goal
[00:41] of the pixel roadmap
[00:42] and here's why suppose we add
[00:45] functionality
[00:46] to today's pixels like the ability to
[00:49] tune its wavelength
[00:50] or direct the optical beam remember the
[00:54] pixel itself
[00:55] as a light source is very low cost but
[00:58] things
[00:58] around the light source are still
[01:01] relatively expensive
[01:02] and bulky so to progress we need further
[01:05] integration of components
[01:07] around the pixels at the size ratio
[01:10] price of the pixel itself
[01:12] and i am talking about mems and micro
[01:14] optics
[01:15] the plan of being able to have dim
[01:17] steering and being scanning pixels
[01:20] now that's fantastic because it is a way
[01:23] of adding
[01:24] another legend of functionality to serve
[01:26] the customer
[01:27] who is always price and size conscious
[01:30] these ideas extend to other businesses
[01:33] that could take further advantages
[01:35] of pixels because this is an economic
[01:38] way
[01:38] to produce light sources with a very
[01:41] precise
[01:42] tunable wavelength take personalized
[01:45] healthcare
[01:46] where smarter pixels could mean the
[01:48] ability
[01:49] to improve the clinical quality of
[01:51] sensors
[01:52] in a wearable or a health patch or even
[01:56] in the next generation mobile phones
[01:58] because
[01:59] pixels only cost a few cents put them
[02:02] in an array and you can make very low
[02:05] cost
[02:05] but highly effective spectrometers
[02:09] when we think of wearables at this
[02:10] moment we think of
[02:12] smart clothing with sensors but using
[02:15] pixels
[02:15] we take things to the next level
[02:18] allowing the wearable
[02:19] not only to sense its surroundings but
[02:22] also see them
[02:23] in 3d companies like prophecy
[02:27] are in the lab exploring the
[02:29] possibilities
[02:31] this opens so many doors to help those
[02:33] with less and ideal vision
[02:36] and imagine the impact on personal
[02:39] mobility
[02:40] when this kind of technology is coupled
[02:42] to a
[02:43] self-driving vehicle but to move faster
[02:46] we need to switch our current thinking
[02:48] around instead of regarding a pixel as a
[02:51] light source to another system
[02:54] suppose the pixel became the platform on
[02:56] which to integrate other things
[02:59] at low price and high volume so how do
[03:02] we persuade pixel manufacturers
[03:04] lumentom 26 osram
[03:08] ams bertilas bank with 10
[03:11] vi systems then what we need is a much
[03:14] more efficient
[03:15] smart optical platform because a pixel
[03:19] is so much more than just a cheap light
[03:22] source
[03:22] that's my vision on monday june 7th
[03:26] i like to invite those companies with a
[03:28] vision on the future of pixels
[03:31] to contribute to a focus discussion on
[03:34] what's next
[03:35] sign up now to access the epic sunroom
[03:38] because thousands of brilliant minds in
[03:40] the epic network
[03:42] can accelerate the path to what's
[03:44] possible
[03:46] welcome everyone and good afternoon the
[03:49] pixels being
[03:50] a platform for photonics is my personal
[03:53] vision
[03:54] not necessarily the one of epic members
[03:56] but in the next two hours
[03:58] and with the speakers that were selected
[04:00] by dr anna gonzalez
[04:02] i think we're gonna have a very clear
[04:03] picture on how to work together
[04:05] my name is jose poto and the cto of epic
[04:08] chief technology observer
[04:10] epic has now a lot of members we are a
[04:12] family of 700 plus members all the
[04:14] companies here in this
[04:15] in this slide are very well known by us
[04:18] what makes a difference to the epic
[04:20] staff is that we know
[04:21] every company we know every member of
[04:23] epic truthfully
[04:25] to make sure that we're connecting with
[04:26] the potential suppliers partners and
[04:27] customers organize events we provide so
[04:29] the network
[04:30] will help you raise capital we have the
[04:32] biggest website to find a job in
[04:33] photonics in the world jobs in
[04:34] photonics.com
[04:35] and every epic member has access to a
[04:38] long list of market reports
[04:40] all of them for you yes part of the epic
[04:42] membership benefits
[04:43] we are almost at the end of the season 4
[04:46] online technology meetings season 4 all
[04:48] the seasons were through the pandemic
[04:50] pay a special attention to the meeting
[04:52] that we have in a couple of weeks
[04:54] on microscopy we are actually collecting
[04:56] some of the key companies to actually
[04:58] find
[04:59] potential collaboration in a business
[05:01] that is growing at the moment but today
[05:03] today
[05:03] is big seller first of all i would like
[05:05] to acknowledge the support of our media
[05:07] partner today electro optics
[05:09] thank you very much for being such a
[05:10] fantastic partner of epic but today
[05:13] this meeting wouldn't be possible
[05:14] without the support of four sponsors and
[05:16] we start with ficon tech
[05:18] all the way from hacking bremen the
[05:20] germany is important to mention them for
[05:23] all they do in the testing of mixes
[05:25] thanks to the multi-channel probe lead
[05:27] technology
[05:28] for batch parallel mix signal test and
[05:31] quality operation but also fiber reality
[05:34] the living technology consultancy
[05:36] company you are developing business
[05:38] in victims so in photonics in general
[05:40] you should speak to fiber reality also
[05:42] pay
[05:42] special attention on the fantastic block
[05:45] vego system
[05:46] all the way from poland from warsaw the
[05:48] 35
[05:49] ap taxi division of ecosystem provides
[05:52] 80 wafers
[05:53] for custom and niche pixel applications
[05:56] but bank with 10 if you're looking for a
[05:57] big size between 720 nanometers to 1.7
[06:00] microns
[06:01] these people provide the fantastic
[06:03] pixels for time of flight or for any
[06:04] kind of 3d sensing technologies
[06:06] all the way from singapore we have focus
[06:10] focus is a very well-known contract
[06:11] manufacturing for
[06:13] electronics providing high volume and
[06:14] low cost as well as high quality
[06:16] services
[06:17] fix in the netherlands the european osat
[06:20] optical semiconductor assembly
[06:22] technologies
[06:23] in the netherlands providing packaging
[06:25] services and testing services photonic
[06:27] integration
[06:29] i'm tech you're looking for a partner to
[06:30] do the packaging but also do the back
[06:32] end
[06:33] to think packaging and backhand at the
[06:35] same time because sometimes you need to
[06:36] wait for thinning and sometimes you need
[06:38] weight bonding at the same time imtech
[06:40] is your key partner for packaging when
[06:41] you also need
[06:42] back-end services and finally last but
[06:45] definitely not list
[06:46] a company everybody is talking about nil
[06:48] technology a new
[06:50] micro optic manufacturer based on evil
[06:53] lithography breaking the malt
[06:54] and then nano imp lithography making
[06:56] volume production
[06:57] or high-end nano optics no micro-optics
[07:01] nano-optics and a person who knows
[07:03] nanotechnology and who knows photonics
[07:05] who's made a difference
[07:06] in the growth of epic is my very good
[07:08] friend doctor
[07:10] anna gonzalez buenas tardes
[07:15] thank you very much for this nice
[07:17] introduction and yes we have a very nice
[07:19] meeting in front of us today
[07:20] here you can see the agenda we we
[07:23] managed to bring
[07:24] one of the top market leaders in
[07:27] manufacturing big cells
[07:29] and they are going to talk about the
[07:30] last developments also
[07:32] about what are the challenges for the
[07:34] supply chain that will be
[07:36] today here and if you can go to them yes
[07:39] here
[07:39] we have an amazing supply chain so you
[07:41] will have the companies
[07:43] offering a big wafer companies offering
[07:46] egg coatings
[07:46] filters micro updates laser technology
[07:50] photonic interrupted circuits all the
[07:52] manufacturing and packaging
[07:54] testing services well everything that
[07:56] they will need
[07:57] and also last but not least i would like
[08:00] to mention the european projects the
[08:01] european initiatives
[08:03] with which we are collaborating first of
[08:05] all passion passion is developing an
[08:07] entire infrastructure
[08:09] for a metro network so yes they
[08:12] integrate big cells with
[08:14] silicon photonics they do a very
[08:16] interesting transceivers
[08:17] also we have pixel pixels for packaging
[08:20] and assembly of photonic integrated
[08:22] circuits
[08:22] also we have a jpeg say if you are
[08:25] interested in going to pilot production
[08:28] a for indian phosphite photonic
[08:30] integrated sequence jpegs has now
[08:32] an open call available so contact us for
[08:35] more information
[08:36] and finally photohub so if you are a
[08:38] company trying to develop a new product
[08:41] using photonic
[08:42] technologies photograph can help you
[08:44] with the expertise
[08:46] but also with financial support and with
[08:48] training
[08:49] and that's it jose what an amazing job
[08:52] you have done these are all the
[08:53] companies who register for the meeting
[08:56] today on pixels if you want to get in
[08:58] touch with any of them
[08:59] they are here looking for partners this
[09:01] is a unique opportunity
[09:04] and this meeting is also live streaming
[09:06] youtube so hello youtubers thank you for
[09:08] joining from youtube you can remain
[09:10] anonymous because people in the zoom
[09:11] room has to be very active
[09:12] you can remain quiet but you want to get
[09:14] in touch with any of the participants
[09:15] today and you have a good business case
[09:16] for them
[09:18] hos.podepic.assac.com i would love to
[09:20] make
[09:21] that connection and let's get the
[09:23] meeting started we had a really strong
[09:25] challenge today
[09:26] because we had some of the top
[09:28] manufacturers worldwide in pixels
[09:30] so who should open the meeting actually
[09:33] i would tell you that it was a coin toss
[09:34] but it wasn't we really wanted trump
[09:37] to start the meeting today for many
[09:39] reasons but first because it's a
[09:41] european
[09:42] success story a family-oriented business
[09:45] that is driving
[09:46] innovation worldwide in laces for
[09:49] manufacturing
[09:50] but also in vertical cavity surface
[09:52] emitting lasers
[09:53] joseph panker thank you very much for
[09:56] taking the role
[09:56] of opening or vixel event of the year
[09:59] the floor and the attention of everyone
[10:02] goes to trump
[10:06] joseph good afternoon
[10:10] joseph pankard from trump we make this
[10:12] meeting live
[10:14] so we always have prepared for this
[10:16] joseph i don't think we can hear you
[10:18] because maybe you are muted and i do
[10:20] have a mac to celebrate such
[10:22] accomplishment
[10:24] so joseph we still can't hear you are
[10:26] you muted can you hear me now
[10:28] loud and clear all the way from akin
[10:31] right in germany the floor is yours
[10:34] so and can you see my skype now yes if
[10:37] you go to slideshow mode we will see
[10:39] with the quality it deserves perfect
[10:42] okay
[10:43] thank you very much jose my name is
[10:45] joseph brancat and i it's a it's a great
[10:48] pleasure to
[10:49] open this meeting for those who have met
[10:51] me before
[10:52] and a few of the the participants i'm
[10:56] sure
[10:56] are friends and old colleagues i have
[10:59] been managing this
[11:00] uh this company for quite some time when
[11:02] it was phillips photonics and
[11:04] in 2019 we were luckily really luckily
[11:07] acquired by a very renowned
[11:09] laser company which is trump and trump
[11:11] is is switching gears now on this
[11:14] and i happily went back to my to my
[11:17] passion
[11:18] which is technology and handed over the
[11:20] uh general management ship to
[11:22] bartol schmidt which many of you know so
[11:24] today is really a privilege to talk
[11:26] about
[11:27] manufacturing and applications about
[11:29] vixels which actually is a topic
[11:31] as broad as to cover a full conference
[11:33] so what i try to do is to
[11:35] to to give you just three slides one is
[11:38] on the challenges that we are facing
[11:40] today
[11:40] as we speak the second one is on the
[11:44] tomorrow and tomorrow has begun already
[11:46] today
[11:47] and the third one is well the challenges
[11:50] that we see coming along
[11:51] if we really want to to take the full
[11:54] potential of the vixel opportunities and
[11:57] it's it's going to be a pleasure to
[11:59] invite all of you to be part of that
[12:02] that story so the challenge we are
[12:04] facing today it's in one single word is
[12:07] process control we
[12:08] as trump and previously phillips we have
[12:10] shipped close to 2 billion pixels
[12:13] and largely without any field returns so
[12:15] we have done single in multimode we have
[12:17] done all kind of wavelengths
[12:19] when it comes to areas we have done
[12:20] small large regular irregular
[12:22] addressable non-addressable high-speed
[12:25] data command you can see a couple of
[12:26] examples on the left side
[12:28] so the top left is a densely packed
[12:30] single mode with 16 micrometer pitch
[12:33] and a random pattern and the 56g datacom
[12:36] chip that we have released recently and
[12:38] adjustable chips
[12:39] now if you look at where is the
[12:41] challenge and where is the money today
[12:43] actually we're spending 20 percent of
[12:45] of the the money to make a vixel we
[12:47] spent on the ep
[12:48] and that's adding value we're spending
[12:51] 40 percent of the money in front in
[12:53] processing so
[12:53] litho etching and so on and so on that's
[12:57] but then another 14 has to be
[13:00] spent on testing and qualifying and this
[13:03] is just adding cost
[13:05] so here's the invitation so if anyone
[13:08] can help us
[13:09] to reduce what we consider as adding
[13:12] cost
[13:13] so compositional control in the epi
[13:17] dimensional control in the in the full
[13:19] inch
[13:20] six inch wafer or four inch wafer
[13:22] processing so that's
[13:23] decisive for the yield and then finally
[13:26] we have to do testing we can't get rid
[13:28] of testing but then
[13:29] if you can help us to reduce and to
[13:31] speed up the testing
[13:33] it's going to bring us major steps
[13:35] forward
[13:37] that brings me to the next slide and the
[13:40] next slide is
[13:40] the challenges we are facing tomorrow as
[13:42] a community and i would call this
[13:44] functional integration
[13:45] a concept that is of course well known
[13:47] in the silicon industry
[13:49] called moore's law in that context
[13:50] because you're putting more and more
[13:52] function on it on the same chip
[13:53] the same is happening actually in the
[13:55] vixel the gallium arsenide
[13:57] platform that we are using is is good
[13:59] for making lasers these are the pixels
[14:01] it's good for making leds it's good for
[14:03] making photo diets it's good for making
[14:04] transistors or photo transistors
[14:07] but also a lot of passive optical
[14:08] elements like lenses polarization
[14:10] gratings diffusers wave guides
[14:12] and finally also skip skip chill the
[14:15] scale packaging
[14:16] devices again on the left side you can
[14:18] see a flip chip
[14:19] data comp device where you can see that
[14:22] on the top left
[14:22] we have even integrated collimating
[14:24] lenses on the back side just for fiber
[14:26] coupling
[14:27] polarization grating which is
[14:30] implemented on the facet of a laser and
[14:32] on the on the lower side you can see
[14:34] some backside optics in a flip chip
[14:37] device where you can do a kind of
[14:39] scanning so it's a fan
[14:40] out structure and and finally also the
[14:42] bumping and
[14:43] mounting devices now the challenges
[14:47] where you can contribute to is really
[14:49] that
[14:50] well if you if you integrate additional
[14:53] functions then it's
[14:54] again about process control which is
[14:56] very important and remains
[14:58] important but it's also the increased
[15:00] complexity
[15:01] we are doubling easily doubling or
[15:02] tripling the number
[15:04] of masks that we are employing and
[15:07] if you have so many functions on the
[15:09] same chip then there is just no way you
[15:11] can do this
[15:12] just by hand waving design rules but
[15:14] what we need is design rules for the
[15:16] manufacturing but also for the device
[15:18] properties
[15:19] and a whole bunch of modeling if we
[15:21] compare ourselves to where the silicon
[15:23] industry is
[15:24] then of course we are three decades
[15:26] running behind and we have to close that
[15:28] gap in order to
[15:29] to make full use of the potential of
[15:31] these additional functions that we put
[15:33] onto onto the same chip
[15:35] as was by the way also referred to in
[15:37] the introduction slides of jose
[15:40] and that brings me to the final one and
[15:42] the final one is the next
[15:43] challenge that that keeps us busy for
[15:46] the next
[15:46] couple of years i would say so extend
[15:50] and create new applications and that's
[15:52] again an open invitation i'll give you
[15:54] one example of
[15:55] something which we hadn't on our roadmap
[15:58] a couple of years ago but then all of a
[15:59] sudden we discovered that
[16:00] if you do with a pixel face recognition
[16:03] or proximity sensing
[16:04] then of course you can also detect pm.5
[16:07] particles so very
[16:08] fine dust particles which is very
[16:09] important for environmental health and
[16:11] together with our partner bosch
[16:13] sensortech we have developed the sensor
[16:14] which you can see on the the upper left
[16:16] side
[16:17] three millimeters in size so very very
[16:19] complex it consists of three vehicles
[16:22] and the detection principle is such that
[16:24] the vixel sends out
[16:25] light that is back reflected then from
[16:27] the particles
[16:29] that back reflected light is coherently
[16:31] interfering with the laser itself that's
[16:33] what we call
[16:34] self-mixing interference then in the
[16:37] lasers we have integrated the photodiet
[16:39] so that's a piece of functional
[16:41] integration already and then we detect
[16:43] the change of the laser light and
[16:44] thereby detect also the presence of a
[16:47] particle and even the speed of a
[16:48] particle and then we have to do some
[16:50] clever
[16:52] signal processing that generates a pm.5
[16:55] value that sensor is very very compact
[16:58] and it does not require any venting
[17:00] holes and it just
[17:01] for for the purpose of this meeting is
[17:03] an example of where from a
[17:05] very unexpected angle we got a new
[17:08] application which
[17:08] hopefully will eventually be a
[17:10] broad-scale applied
[17:12] sensor that you you can find everywhere
[17:14] where you take care about the
[17:16] environmental
[17:17] quality of your air so that brings me to
[17:20] the last sentence that so the challenge
[17:22] for the epic community is really invent
[17:24] invent invent come up with new
[17:27] inventions
[17:27] come up with new applications because i
[17:30] believe that we have kept only a
[17:31] fraction of what pixels can
[17:33] can do for all of us and with having
[17:35] said this
[17:36] i thank you for listening and i'm happy
[17:39] to engage with
[17:40] all of you to excel in manufacturing on
[17:43] the wanted side and the second topic was
[17:45] the application so come up with new
[17:47] ideas and we are happy to support this
[17:49] with
[17:49] new vixen technology thank you very much
[17:53] thank you very much joseph super great
[17:55] presentation we liked
[17:56] very much and maybe i would like to ask
[18:00] you to summarize maybe in
[18:02] in maybe to elaborate a little bit more
[18:03] about the challenges that you
[18:05] presented at the beginning you know
[18:07] about the
[18:08] how to improve the testing and the other
[18:11] challenges maybe you could summarize for
[18:13] our
[18:14] in companies here in thumb
[18:17] well the testing of course is is very
[18:20] broad we have optical testing for for
[18:22] the integrity of the chips for defects
[18:23] that we have to test we have the full
[18:25] electro optical testing which is usually
[18:27] 100
[18:28] testing but then we have also
[18:29] qualification of the wafers where we
[18:31] take samples and that sample testing is
[18:33] is a
[18:33] destructive testing and especially the
[18:36] the the reliability testing in in the in
[18:38] the field of data com that's very
[18:40] challenging very demanding
[18:42] and it takes an awful lot of time
[18:43] because we need to build up a database
[18:45] we need to to demonstrate that
[18:47] the devices are lasting for very very
[18:49] long and anything
[18:51] that can help us to shorten that testing
[18:54] to have more clever ways of building up
[18:56] that statistical base
[18:58] uh to do on wafer testing as much as we
[19:01] can
[19:02] will greatly reduce the time to market
[19:04] of new ones
[19:05] but it's also reducing the cost that we
[19:07] have to incur
[19:09] so it's it's it's really a full bunch of
[19:11] testing that that starts with in-line
[19:13] testing during production and it starts
[19:14] with the epi
[19:15] once once you you don't do the epi the
[19:18] correct way
[19:19] you have hardly any way to repair this
[19:21] first downstream
[19:22] so anything which which which which
[19:24] helps uh
[19:25] making you good start into processing is
[19:28] is welcome
[19:29] all the way up until testing of the well
[19:33] optical properties for instance okay
[19:36] thank you we have a few companies in
[19:38] these meetings such as spike on tech i
[19:40] am sure they can comment later about
[19:42] this
[19:42] right now and we have already won one
[19:45] question in the chat well
[19:47] more than one let's go to theater from
[19:50] what's up tech photonics a little would
[19:53] you like to make this a question by
[19:55] yourself
[19:56] um hi thanks thanks for calling me um
[20:00] i'm i i come from the raman spectroscopy
[20:02] side and
[20:03] pixels would be just perfect light
[20:04] sources for portable raman
[20:07] if and only if their bandwidth could be
[20:09] shrunk a little bit
[20:10] i think they're on the order of a
[20:12] nanometer and if they could go to 0.3
[20:14] or nanometers or below at any wavelength
[20:17] around 800 nanometer give or take
[20:19] 30 nanometers or something that would be
[20:22] just
[20:22] that which has a perfect match for for
[20:25] portable raman spectroscopy
[20:27] and i was just wondering is that is that
[20:28] possible
[20:30] i i do believe it's possible so if you
[20:32] if you just
[20:33] send send your your wish specification
[20:36] i'll
[20:37] or certainly connect to the experts and
[20:38] as a matter of fact we have been
[20:40] discussing that that application a
[20:41] couple of years ago
[20:42] but it was probably too early in the in
[20:44] the uh in the
[20:46] in the development but it's it's
[20:48] definitely possible
[20:51] okay thank you very much um yes i am
[20:55] looking forward to receiving your email
[20:57] so
[20:58] yes i'm sure you can talk later okay
[21:00] further about this
[21:01] and now we have dominique from photon
[21:03] design
[21:04] i see you raise your hand yes
[21:09] yes my question uh
[21:12] joseph you mentioned that uh yeah so
[21:16] we we uh we model pixels and uh so i was
[21:19] very curious
[21:20] um that you mentioned that you you need
[21:23] uh much better modeling to manage the
[21:26] complexity that
[21:27] your pixels now have and can you
[21:30] elaborate on
[21:31] what do you see the challenges still for
[21:34] on the modelling side
[21:35] what what what um what are you
[21:38] looking for well it's it's it's it's the
[21:41] really the full spectrum yeah so i
[21:43] i don't expect we can do all in one
[21:45] model but it's it's it's the the
[21:46] spectrum of
[21:47] starting with the with the model of the
[21:49] the ap of course uh for for
[21:51] b for high speed for short pulse speed
[21:53] for high efficiency
[21:55] so there are all kind of design freedom
[21:58] that we have
[22:00] but then if you have a look at the the
[22:02] pictures that i show here then of course
[22:04] there's a lot of
[22:04] optical modeling that is also very
[22:07] necessary so the the amount of modes
[22:09] that we have in the laser the
[22:10] behavior of the modes along a pulse
[22:14] train that is
[22:14] sent to the pixel um polarization
[22:17] control scc
[22:18] mode control so there's a lot of
[22:20] publications also on controlling the
[22:22] modes
[22:22] i'll give you an example so we if you
[22:25] want to have single mode lasers
[22:27] and of course the the way we do it is to
[22:29] to have a very
[22:31] narrow oxidation aperture
[22:35] and then it's becoming single mode
[22:36] automatically with with number of
[22:38] challenges that come along and all of
[22:40] that
[22:41] is is partly covered by the models
[22:45] many of them are self-made but i think
[22:48] time
[22:48] is there to have more professional
[22:50] modeling platforms on which we can
[22:53] predict the properties of of the the
[22:55] optical properties but also the electric
[22:57] properties of the lasers so it's really
[22:59] the full spectrum of all the properties
[23:01] that we
[23:02] we need to model and we we have
[23:05] quite some own modeling activities
[23:07] mostly
[23:08] in cooperation with institutes but we
[23:11] are not at
[23:12] at the position as the silicon industry
[23:15] would be where you have professional
[23:16] companies just
[23:17] focusing on modeling
[23:21] and everyone is using that
[23:25] well it'd be interesting to talk to your
[23:26] your colleagues later
[23:29] be very happy to um i mean we
[23:32] we throw it on design we can't do
[23:35] everything but um
[23:36] a number of things that you raised i
[23:38] think
[23:40] are already problems that we've solved
[23:43] okay just another example the lower left
[23:45] one is is the bumping and mounting there
[23:47] of course the thermal
[23:48] mechanical model they are they're really
[23:49] very important but those ones
[23:51] are commercially available pretty easy
[23:53] so so that's
[23:54] that's that's one of the the areas where
[23:57] we are pretty well covered but
[23:59] nevertheless an important topic okay i'm
[24:02] sure that you yes the button design
[24:04] software can help you to all these
[24:06] challenges so i i'm sure you can have a
[24:08] talk
[24:09] later okay thank you then maybe mark
[24:12] you also have your hand right uh yes sir
[24:15] thank you uh
[24:16] just a quick question for joseph uh can
[24:18] you talk at all about uh
[24:20] how many many of your shipments are a
[24:22] thousand nanometers and higher
[24:24] and do they tend to stay more towards a
[24:27] thousand
[24:27] than the 1300 given the difficulties
[24:31] that go to 1300
[24:32] right now well well i think if you look
[24:35] at the
[24:35] the worldwide shipments so not not the
[24:37] trump shipments only then the the
[24:39] the bulk of all the shipments is 940
[24:42] that goes into into smartphones and
[24:44] that applies also for us and the second
[24:46] uh
[24:48] biggest volume is 859 meters that goes
[24:51] into
[24:51] data com applications these are the two
[24:53] driving applications but what we see is
[24:55] that
[24:56] the field is opening up widely there's
[24:59] recent of course
[25:00] recently of course a lot of questions
[25:03] and rumors about even longer wavelength
[25:05] but
[25:06] we have seen in specialty applications
[25:07] also shorter ones like in
[25:09] oxygen sensing and i do believe that
[25:12] we'll see
[25:13] along with additional application we'll
[25:15] see the full spectrum of all wavelengths
[25:17] coming
[25:18] also in datacom where people are talking
[25:20] about different wavelengths that you're
[25:22] using for cwdm applications
[25:25] but the historically the bulk is 940 and
[25:28] 850.
[25:32] okay thank you very much we have another
[25:34] question so will fred from swiss micro
[25:36] optics would you like to make the
[25:37] question by yourself
[25:39] yes thank you very much anna for letting
[25:41] me uh ask this question right here
[25:43] so very impressive again this vixen
[25:46] business i i'm like jose it's really
[25:48] fascinating to see what's going on there
[25:50] the the you showed in your last slide
[25:53] the
[25:54] pm 2.5 applications so this is clearly a
[25:58] new application compared to the other
[25:59] bulk of application that you mentioned
[26:01] earlier in data
[26:03] telecom and or datacore mostly and and
[26:05] 3d sensing
[26:07] so from what i i think there are more
[26:10] commercial challenges than than
[26:12] technological challenges
[26:14] for for bringing pm 2.5 with fixes to
[26:18] the market is that a
[26:20] correct uh assessment or or what is your
[26:23] experience there
[26:25] yes yes i mean it's it's it's always a
[26:28] um a question on whether the
[26:30] let's say the the the take rate or the
[26:33] acceptance of a new sensor principle is
[26:35] is
[26:35] is widely accepted by the market and
[26:38] what is what is of course
[26:39] very clear is that sensing functions
[26:42] will largely increase because sensors
[26:44] are
[26:44] in in general the the interface between
[26:47] the analog and the digital world
[26:49] and environmental sensing is also
[26:51] something which which
[26:53] will definitely increase the question is
[26:56] on on whether
[26:57] that specific form factor which is very
[26:59] very compact
[27:01] uh whether this is necessary for most of
[27:04] of the applications or where you whether
[27:05] you can do with a more traditional and
[27:07] conventional one
[27:08] at the end of the day my
[27:12] personal experience once you can do
[27:14] things very compact and you integrate
[27:17] that eventually it will be taken and
[27:19] that's what we expect for for this
[27:21] sensor as well
[27:22] it's from a performance point of view
[27:24] it's it's just matching
[27:26] uh bulky um standard pm
[27:29] uh 2.5 sensors so there's
[27:33] performance point of view is this is
[27:35] perfect
[27:36] but of course it's not it's it's a
[27:38] little bit more expensive here so that's
[27:40] that's uh
[27:40] yeah that that's my experience okay
[27:43] thank you very much
[27:45] i thank you for this a now i see iwan
[27:47] from iqe
[27:49] iwan hello how are you doing
[27:52] hi hi anna i'm doing very very well
[27:54] thank you
[27:55] uh nice to be part of this excellent
[27:57] meeting
[27:58] i have a one question for yourself
[28:00] joseph
[28:02] um when you mention the
[28:05] role that the epi manufacturer has to
[28:07] play in
[28:08] in pixel manufacturing what are the sort
[28:11] of things that you like
[28:12] i understand you said about the improved
[28:15] uniformities
[28:17] and compositional control and so on
[28:20] would
[28:20] what else would you like to see from the
[28:22] epi manufacturers then if you
[28:24] are looking for improved yield on the
[28:27] device manufacturer's end
[28:28] is it some aspect of the performance of
[28:32] the wafer that is more related to the
[28:34] device performance as
[28:35] other than say the materials
[28:37] characterization
[28:39] provided by the epi provider well you
[28:42] you know where we're coming from it's
[28:44] it's that we we order epi
[28:46] to uh and i'm talking about uh dark dark
[28:49] history of course we order
[28:51] we we order app or we make appy and and
[28:53] and we specify the
[28:56] the the basic parameters of the ap
[28:58] compositional
[28:59] um uh properties and and then the
[29:02] dimensional properties and then at the
[29:04] end of the pixel processing you discover
[29:05] whether the api has been has been okay
[29:07] or not
[29:08] and of course we have come a long long
[29:10] way so we can we can much much better
[29:12] characterize whether the epi
[29:14] is is really what it should be and that
[29:17] is
[29:18] first of all finding the right metrology
[29:21] and we have
[29:22] even cooperated many years in in
[29:24] defining a better and better metrology
[29:26] but but then also controlling your
[29:27] equipment because what would i say
[29:30] is if there's any inhomogeneity or any
[29:33] any
[29:34] mistake or whatever in the epi
[29:37] it's impossible to repair afterwards and
[29:40] sometimes
[29:41] we only discover when we do the final
[29:43] qualification and that's of course
[29:45] something where we
[29:46] clearly would expect that that we we
[29:48] improve and we can
[29:50] continuously improve the quality of the
[29:52] epi and also the the
[29:54] metrology just to to decide after the
[29:57] epi growth on whether it's good happy or
[29:59] not
[30:00] just to avoid that we are adding value
[30:02] and adding cost
[30:03] afterwards and have to uh for for no no
[30:05] no good reason that that's one the the
[30:07] second one is
[30:09] yes well as you see that there's
[30:12] a request for widening the the
[30:15] wavelength
[30:16] that that we we have to supply for
[30:18] pixels and
[30:20] this is very very much of course an epic
[30:22] question
[30:23] so we have to under 1300 nanometers for
[30:26] many years and we know what
[30:27] challenges are around that but but there
[30:29] is certainly an area where epic can
[30:30] contribute
[30:31] big times
[30:34] i would like to congratulate you on the
[30:36] fantastic presentation that you gave but
[30:38] i'm not ready to let you go yet i'm
[30:40] going to come back to this slide
[30:42] and the reason why i want to come back
[30:43] to this slide is because you say that
[30:44] the challenge
[30:45] is in process control and you talk here
[30:48] about the
[30:49] impossible to repair mistakes which i
[30:50] understand but defining the yield and
[30:53] the target is to
[30:54] speed up the testing i know that anna
[30:57] already asked you to comment on the
[30:59] testing
[31:00] parts but i would like to to retake that
[31:02] we are talking about
[31:04] colorimetry and we are talking about
[31:06] wavelength tracking
[31:07] what kind of testing do you need to
[31:09] perform during process control do you do
[31:11] a pl measurement why could you tell us a
[31:14] little bit about this because i would
[31:15] like
[31:16] some epic members
[31:21] for the device testing the process
[31:23] control from the ap and also afterwards
[31:25] for the device testing so both well it's
[31:26] the full spectrum i mean
[31:28] we all know it's sims measurement it's
[31:30] tm measurement it's it's uh
[31:32] it's uh uh depending on what you do is
[31:35] uh
[31:35] wrenching the fracture mate geometry
[31:37] it's pl measurement
[31:38] uh it's uh and then i it's probably not
[31:42] exhaustive but
[31:43] there's a whole a full spectrum of
[31:45] measurements and i think that that's
[31:46] pretty well known in the community
[31:47] what's what needs to be done
[31:48] and what can be done we have seen
[31:51] companies developing spectrometers
[31:53] like advantages that they are providing
[31:56] this kind of
[31:56] spectrum analysis and i was wondering if
[32:00] that's enough
[32:01] because we are targeting now many
[32:02] different markets in which modulation
[32:04] is a very important parameter and to
[32:07] what extent
[32:08] is enough to monitor the wavelength of
[32:10] pixels
[32:12] uh that in gives enough
[32:15] information for the yield is something
[32:17] that they still have in mind and it's
[32:18] not clear to me
[32:19] actually i am not sure that just having
[32:21] a
[32:22] wavelength measurement is enough there
[32:24] are many companies in the in the room
[32:26] that provides a testing of
[32:30] when they are a device level but joseph
[32:32] could you comment maybe on this before i
[32:33] go to them is there a
[32:35] enough information from from the
[32:37] spectrum
[32:38] that we can actually guarantee that the
[32:40] vixen will work at device level
[32:44] well guarantee is always relative terms
[32:47] i mean to to make a pixel work
[32:49] in itself just sending out light is is
[32:52] probably not not
[32:53] a big challenge but to to make it work
[32:54] for 10 years
[32:56] to to to take the example of the data
[32:58] com under on the uh
[33:00] on the uh um 24 7 operation
[33:04] so that that's that's a very big
[33:06] challenge and and
[33:07] you you cannot decide on whether it's
[33:08] going to work just by measuring the
[33:10] wavelength
[33:11] that's that's not enough so there's a
[33:13] lot more that needs to be measured
[33:15] and again the the amount of knowledge
[33:18] that we are building up by correlating
[33:20] the device performance to the
[33:21] measurement that we do in the epi
[33:23] is building up over the last years and
[33:26] and
[33:26] it's not that we we we can decide only
[33:28] at the very end where the device is
[33:30] working according to the specification
[33:32] we we know much much earlier in the
[33:33] chain
[33:34] however it doesn't stop
[33:37] i think still even today you can have
[33:41] a characterization of the epi with a
[33:44] number of hidden parameters which you
[33:46] don't characterize which for instance
[33:48] come by different growing conditions
[33:52] and which at the end lead to different
[33:55] properties of the pixel and that level
[33:57] of control we would like to achieve that
[34:00] by growing the epi and the same by the
[34:03] way
[34:04] applies also to other processing steps
[34:06] like oxidation or metallization or so on
[34:08] but but
[34:09] to to grow the epi to an extent that we
[34:13] can
[34:13] safely predict the properties uh
[34:16] at the very beginning already on the
[34:18] next two on the on this meeting of two
[34:20] hours by the way in this meeting one
[34:21] hour and 25 minutes left
[34:23] if we can find some information about
[34:25] the testing challenges for pixels and
[34:27] also for the integration with micro
[34:29] optics and pixels
[34:30] we're gonna have a successful meeting so
[34:31] let's see how it goes we're gonna ask
[34:33] many of the vixen manufacturers in the
[34:35] room later we have vertilas we have
[34:36] bandwidth 10.
[34:37] i want to understand a bit more what
[34:39] kind of room for cooperation is here
[34:41] but in interaction with micro optics
[34:42] joseph there is a question from you from
[34:44] youtube
[34:45] and that question is from the ceo of
[34:48] icon photonics carlos viana this company
[34:50] is actually integrating micro optics at
[34:52] wafer level on pixels
[34:53] and carlos vienna is wondering i could
[34:56] see some micro optics on pixels in your
[34:58] presentation
[34:59] could you tell us a bit more about the
[35:01] beam control
[35:02] challenges beam control challenges that
[35:04] you have
[35:08] well first of all there's there's this
[35:10] is a specific
[35:11] optics the all the optical element that
[35:14] you can see on these
[35:14] pictures are optical elements that are
[35:17] etched into the gallium arsenide itself
[35:19] so it's it's gallium arsenide optics
[35:21] it's monolithically
[35:22] part of the the wafer of course you can
[35:24] you can also use
[35:25] hybrid optics by by bonding optics at
[35:28] the
[35:29] surfaces and we have done that as well
[35:31] and we know how to do that as well
[35:33] and at the end is of course to
[35:36] characterize the optical surfaces
[35:38] with respect to their geometrical
[35:42] properties that that's that's of course
[35:44] a big challenge that we have
[35:46] and then of course in the during the
[35:49] process of making these optics we have a
[35:51] lot of challenges in making sure
[35:52] that this process is stable and
[35:54] controlled
[35:56] the challenge from trump was on process
[35:58] control monitoring and he also asked the
[36:00] epic members to think about applications
[36:02] doomfist here looking for partners
[36:05] and joseph is here giving one of the
[36:06] best presentations so far at an online
[36:08] technology meeting joseph
[36:10] thank you very much for taking the
[36:12] opening slot
[36:13] from trump what a lineup we have today
[36:16] from trump we go to
[36:17] broadcom counter stainless system
[36:20] engineer and program manager at broadcom
[36:22] concert thank you very much for being
[36:24] with us today tell us how the epic
[36:26] members can have broken
[36:28] be even greater than what it is the
[36:30] floor is yours
[36:32] ah thanks josie thanks for the
[36:34] invitation
[36:36] i would like to show you a little bit
[36:39] about
[36:40] the latest progress of our data com
[36:43] solutions
[36:44] uh reliable mixes as well as the
[36:46] industrial
[36:47] applications um from i am i'm
[36:50] responsible for the system development
[36:52] of
[36:53] some of the industrial products and
[36:56] especially sensor products
[36:58] spectrometers and
[37:01] free space transceivers so
[37:04] you will see it in my presentation but
[37:06] in general
[37:08] we are always interested in
[37:09] collaborating
[37:11] around our module solutions so we are
[37:14] constantly searching for the
[37:16] better lensing systems better deflective
[37:18] elements
[37:19] lower cost testing solutions of course
[37:22] um
[37:24] even are we supposed to be seeing some
[37:26] slides
[37:27] yeah we are not yet
[37:30] so window share a screen
[37:33] there you go okay so let's
[37:37] uh start let's say with our pixel
[37:39] technology
[37:41] high speed pixel technology the base
[37:44] baseline technology
[37:47] the the latest and greatest we have now
[37:50] is
[37:50] the afcd v84 lp so it's a
[37:54] four times 100 gig solution so this is
[37:57] for boosting the 100 gigabit per channel
[38:00] so going to 50 gigabytes and
[38:04] using the the well established pum4
[38:07] modulation
[38:08] having four discrete
[38:11] amplitude levels per bit that means
[38:15] basically two bits per bout very
[38:18] basic modulation scheme but we with
[38:20] quite some overhead required
[38:23] but i have to admit the bandwidth
[38:27] improvement of the vix's is not
[38:30] not fully not fully taking uh up
[38:33] uh with the with the bit rate
[38:36] development so
[38:37] pushing the bandwidths for the vixes is
[38:41] let's say getting harder and harder at
[38:43] the moment
[38:44] we are approaching the 30 giga
[38:47] hertz however of course we would like to
[38:50] have more
[38:50] we need more than 30 gigahertz so the
[38:53] the pressure
[38:54] to the electric uh equalization
[38:57] systems is getting higher and higher
[39:00] luckily we also
[39:01] we are also focusing on the digital
[39:03] signal processing and the equalization
[39:05] part so
[39:06] therefore later i will show you a new
[39:09] set of
[39:11] samples uh where the complete the
[39:14] complete transceiver
[39:15] uh chain is uh discovered
[39:19] for four times wire kick in general
[39:21] broadcom we
[39:22] are more or less focusing on data
[39:24] centers
[39:26] all kinds of solutions for the data
[39:28] centers from the storage
[39:30] with the ethernet communication uh
[39:33] down to the optical physical layer and
[39:36] this is let's say
[39:37] kind of a core product from our company
[39:40] the older ones those are already
[39:43] productive
[39:43] they are the four times 50 gigabits per
[39:46] second
[39:47] pump four weeks array and the extended
[39:50] temperature array so the v64 xt
[39:53] is a special series which i am also
[39:55] using for industrial products
[39:57] because it's qualified for an extended
[40:00] temperature range
[40:01] so the typically we have 5 to 75 degrees
[40:05] in this case we have minus 20
[40:07] to 90 degrees substrate temperature for
[40:10] the 25 gigabit
[40:13] non-return to zero classical modulation
[40:17] um on the right hand side you see
[40:20] i mean it's already two years old but
[40:23] this is from the beginning when we
[40:25] when we started shipment the
[40:27] qualification
[40:29] or of the lifetime this is maybe uh
[40:32] i also have to admit this takes a long
[40:34] time you need to sample for very long
[40:36] time data
[40:37] and this is quite the same on our site
[40:41] so that
[40:42] yeah i would say even at least 40
[40:45] percent of the
[40:46] of the cost of the whole pixel
[40:49] development and so on so
[40:50] that's the testing and qualification is
[40:53] a big
[40:54] big issue yeah um
[40:57] and yeah i mean the we always guarantee
[41:00] for minimum 10 years lifetime with some
[41:03] let's say safety margin this is the same
[41:06] for our data com customers as well as
[41:08] for industrial customers
[41:10] i think i should move to the next slide
[41:15] so this is now showing the complete
[41:18] 100 gigabit per channel component
[41:21] solution
[41:22] lineup we are sampling this at the
[41:25] moment we expect the full
[41:27] productive release by end of this year
[41:30] and
[41:30] it consists of the vixel array itself i
[41:33] mean it's
[41:34] slightly modified from the from the
[41:36] first page but it's more or less the
[41:37] same physical
[41:38] physical device just a different test uh
[41:42] decay here and of course then around the
[41:45] pixel the vix elastic core component but
[41:47] around the pixel
[41:48] the drivers very important also the
[41:51] files
[41:52] which are which are let's say
[41:57] handling all the signal processing from
[41:59] the receiver side as well as from the
[42:01] transmit side
[42:03] some pin photo detector reference arrays
[42:07] and the transit potential amplifiers
[42:10] so the whole the whole core
[42:13] technology for the transceivers
[42:16] and just on the upper right side you
[42:19] just see
[42:20] um yeah the picture from the 50 gigabyte
[42:23] measurement
[42:25] the eyes are open however they could
[42:28] look
[42:28] nicer so this is something we need to
[42:32] further further improve also from from
[42:35] equalization point of view
[42:39] in general the whole lineup is optimized
[42:41] as usual for multi-mode fibers
[42:45] in this case the the om4 it's
[42:48] standardized fiber
[42:49] and we are we are more or less always
[42:53] focusing on the 850 nanometers
[42:55] wavelengths
[42:57] because this is let's say the core of
[42:59] our of our applications
[43:01] is always the fiber application
[43:03] therefore the mixers
[43:04] are developed around the fiber
[43:06] applications and therefore 850 is
[43:09] is more or less always the same the
[43:11] unchanged target math links so
[43:13] if we have and this is now let's say my
[43:15] next slide
[43:16] if you have different applications where
[43:18] we for instance are not using fibers
[43:21] here we have two examples from the
[43:23] industrial uh
[43:24] space where we use air very weird
[43:28] medium then we just use air
[43:31] then of course we are not forced to use
[43:33] 850 nanometer it's just convenient
[43:36] because the technology is available but
[43:37] we could go or make even we should go to
[43:40] different wavelengths
[43:41] sometimes other wavelengths are better
[43:44] suited for certain
[43:45] free space applications in terms of buy
[43:48] safety and so on so 1300 nanometer would
[43:51] have a big advantages here
[43:53] of course and 1500 as well
[43:56] for the optical free space communication
[43:59] uh we have a product
[44:00] launched now a couple of weeks ago
[44:04] even it's just a 1.25 gigabit full
[44:07] duplex transceiver the special
[44:10] feature of this transceiver is that it
[44:13] is not using two different wavelengths
[44:15] it's using the same wavelengths for
[44:17] receiver and transmitter
[44:18] it's using the same lens for receiving
[44:20] and transmitting it's using the same
[44:22] device
[44:23] so it's a full duplex transceiver
[44:27] which is using a spatial spatial
[44:30] decoupling of the
[44:32] receiver and the transmit path the
[44:35] spatial decoupling is used uh
[44:37] by the use of a fresnel structure for
[44:39] the transmitter
[44:41] and the reflective lens below for the
[44:43] receiver
[44:44] so transmit that receiver are separated
[44:47] spatially
[44:48] and also the data streams are spatially
[44:50] separated
[44:52] so the unique thing of this transceiver
[44:54] is that it's really 330 degrees
[44:56] symmetrically
[44:57] symmetrically so you can use it in
[45:01] rotating in your rotating application so
[45:03] we have one
[45:04] liner scanner approach using this but
[45:08] let's say our main application is
[45:12] this medical so going through glass
[45:14] windows
[45:16] where wi-fi or other wireless
[45:19] solutions are prohibited
[45:23] and what we are developing now is
[45:26] an extension to higher towards higher
[45:30] data rates five gigabits plus you think
[45:32] that's uh
[45:33] not a big deal however even even using
[45:36] our let's say
[45:37] latest uh almost latest type and with
[45:40] fixes i mean
[45:42] 25 gig plus we are coming to some
[45:45] limitations here because
[45:47] uh from eye safety and other conditions
[45:50] we are forced to
[45:51] to keep a very low uh modulation current
[45:55] also in our applications a very special
[45:57] thing is ios power consumption
[45:59] so typically when when we offer some
[46:02] transceivers we
[46:03] are not often using higher data rates
[46:06] but maybe
[46:07] lower the the power consumptions for our
[46:10] for our applications
[46:11] for the industrial business units and
[46:15] in this case it's the same we have to
[46:17] use very small modulation currents
[46:19] and so this even five gigabit is very
[46:22] challenging then
[46:23] for this free space transceiver and last
[46:26] but not least our time of flights in the
[46:28] series
[46:29] they got productive last year and we are
[46:32] adding more and more module types now to
[46:34] the family
[46:37] one the series now is up to 30
[46:40] meters we are developing now an extended
[46:43] range for up to 50 meters
[46:46] distance outdoor so also the core
[46:48] element for this
[46:49] is um also a new pixel
[46:53] where we where we improved the peak
[46:56] current and the brilliance
[46:57] basically of the vixel reduced the
[47:01] impedance
[47:02] but kept let's say the rise in four
[47:04] times or the bandwidth
[47:05] more or less the same so even the high
[47:09] bandwidth fixes are important for this
[47:11] time of flight
[47:12] solutions because the time of flight we
[47:14] are talking about tens of picoseconds
[47:16] which we have to differentiate so
[47:18] whenever the bandwidth is changing or
[47:21] rise times are
[47:22] changing and so on this will directly
[47:24] relate to distance errors
[47:26] and they have to be calibrated in a very
[47:29] complicated way normally so that more
[47:31] stable the
[47:32] the pixel the more stable the detector
[47:34] the the easier and the lower cost
[47:36] the system so
[47:41] okay so one example here i mean
[47:45] getting let's say three times 30 devices
[47:48] measured over 9000 hours this is
[47:50] an actual measurement of the latest
[47:52] pixel uh which we are using the latest
[47:54] generation now
[47:55] um at 100 degrees substrate and 40
[47:59] milliamps cw
[48:01] typically for the for the modulus we are
[48:03] using about only one percent duty cycle
[48:07] or with 40 milliamps a little bit more
[48:09] 1.5
[48:11] in order to to be eyesafe this means
[48:14] this already translates into a very very
[48:17] long lifetime but however you can
[48:18] imagine
[48:19] once you are leaving laser class one
[48:21] once you are boosting the power once you
[48:23] are boosting the
[48:24] the peak uh powers and so on uh then uh
[48:28] of course you
[48:29] you have new challenges ahead but let's
[48:32] say our
[48:33] our main focus here is to get the
[48:36] compact modules
[48:37] uh ideally they should be reflow
[48:40] solderable
[48:41] so the the free space transceiver is
[48:42] really flows orderable
[48:44] the time of flight sensor is not yet
[48:46] reflow soil durable so it uses a more a
[48:49] traditional lens system
[48:50] but we are going to we are going to
[48:53] develop a reflow
[48:54] capable device so
[48:58] so that's more or less uh that was my
[49:01] brief overview
[49:03] so that bandwidth is always important
[49:06] even for applications where you don't
[49:08] think uh
[49:09] the speed of the pixel is of any
[49:11] relevance or you think you have much
[49:13] room but
[49:14] often in in reality you face some
[49:16] conditions
[49:18] such as a very large temperature span
[49:21] or a very small current or
[49:25] extreme reliability conditions or
[49:27] whatever where you
[49:28] where you use some of the margins of the
[49:32] anyhow
[49:33] available technology
[49:36] so our track record so far was doubling
[49:39] the bit rate every three years
[49:41] and let's keep fingers crossed that we
[49:44] that we
[49:45] uh maintain at this place but what i see
[49:48] from let's say from system point of view
[49:52] more and more ip will be will be uh
[49:55] generated on the electronic
[49:57] side and on the signal processing side
[49:59] to keep at the space
[50:02] thank you very much gunter for a great
[50:04] presentation gunter you are an epic vip
[50:06] so you know how this works by now epic
[50:09] is always looking for companies who can
[50:10] help you be even greater
[50:12] so give us a couple of challenges the
[50:14] epic question
[50:15] how can you help them help you
[50:18] so in general i have to say again
[50:21] besides from the
[50:22] buyer seller let's say relationship we
[50:24] always are
[50:25] searching for new elements so new
[50:28] deflective elements
[50:30] new diffuser technologies to boost isafd
[50:34] limits
[50:34] mems is a very important topic so
[50:38] once we once for our sensors we need a
[50:41] good mirrors um for let's say
[50:45] uh for deflecting the light so i have
[50:48] shown you a very simple time of light
[50:50] sensor which is basically flush
[50:51] flashlighter but what
[50:53] what is also used are lighter lighter
[50:56] solutions with
[50:57] certain scanners mechanical scanners
[51:00] and also for our spectrometer family we
[51:04] are searching for
[51:05] for mems solutions now to be more
[51:08] flexible to use
[51:09] different detector technologies or
[51:12] combine
[51:12] let's say a low low grade array
[51:15] pixel together with many many receiver
[51:19] pixels
[51:20] so one challenge is to get large arrays
[51:22] of the mixers so i have shown you
[51:24] an array of four that's a very small
[51:27] array
[51:28] um if you want to make an array of 20
[51:31] 000
[51:32] i mean we are probably not the company
[51:34] to do that there are others
[51:35] having more experience but what we see
[51:38] on our testing so far the yield is a
[51:41] dramatically it's a dramatic issue and
[51:44] when you get to larger arrays so ether
[51:46] you you improve the let's say
[51:50] the defect control on ep taxi and the
[51:52] process
[51:53] you build in some some um
[51:57] let's say redundant channels some nice
[52:00] ideas
[52:01] or you have good uh yeah a good
[52:05] a good mirror system a good optical
[52:07] system which which somehow adapts and i
[52:09] think
[52:10] uh adapting the optics also the passive
[52:13] optics
[52:13] lens elements to the active and
[52:17] alignment and so on so similar as the
[52:19] cameras has
[52:20] started 10 years ago with autofocus this
[52:23] will be a core technology in the future
[52:25] but this is a wave level technology
[52:27] correct this is a wafer level technology
[52:29] otherwise it becomes too expensive which
[52:31] is something that you cannot deal with
[52:32] absolutely so let's try to find your
[52:34] partners for wafer level packaging of
[52:36] optics do we have anybody from
[52:38] neil technologies in the room for neil
[52:41] technology i think have somebody all the
[52:43] way from
[52:43] denmark right nicholas hanson neil
[52:46] technologies we are looking for partners
[52:48] for broadcom to actually
[52:49] integrate high-level micro-optics
[52:53] with very high precision can you help
[52:56] yes let me let me give you
[53:00] first a short introduction to to nilt
[53:05] so uh nil is an optics company uh we
[53:08] have three different product categories
[53:11] we have the future displays where we
[53:13] make
[53:14] masters for wave guides and then we have
[53:17] the meta optics and the
[53:19] diffractive optics and with the mid
[53:22] optics and the diffractor optics we make
[53:24] both receiver optics so focusing lenses
[53:27] and we also make emitter optics
[53:30] such as diffusers fan outs collimators
[53:36] and below here you can see some some
[53:39] typical examples
[53:41] what is interesting about the diffusers
[53:44] that we're making
[53:44] is that you can go from a really high
[53:47] intensity to a very low intensity
[53:50] very quickly so from ninety percent down
[53:54] to ten percent in just
[53:55] below one degree then we're also using
[53:59] diffractive optics
[54:01] [Music]
[54:02] and we managed to control the
[54:04] manufacturing to keep the zero order
[54:07] so the light that does not interact with
[54:10] a structure
[54:11] we managed to keep that to below 0.1
[54:14] percent
[54:15] then there's also a very high degree of
[54:18] freedom when it comes to controlling the
[54:19] intensity
[54:21] and this is a direct consequence of
[54:23] using a refractive optic
[54:26] to generate the light pattern
[54:29] then we also do fan outs like dot
[54:32] patterns
[54:34] and here there's also a huge flexibility
[54:37] in
[54:38] a number of dots field of view and all
[54:41] of that
[54:41] or field of elimination and one
[54:44] challenge here is to to get a low
[54:46] uniformity
[54:49] error and we are able to reach below 10
[54:54] variation this particular example here
[54:57] is
[54:57] polarization independent but i'll also
[54:59] mention a little bit later on that it's
[55:00] possible to
[55:01] deliberately use uh the polarization in
[55:05] the device and then also collimators
[55:08] um based on on does or met optics
[55:12] there were able to to get a very low
[55:14] beam divergence
[55:16] and you can tailor it to the specific
[55:18] pixel array
[55:20] and you can have tight tolerances on the
[55:23] focal distance below 5 microns
[55:27] so all of these different type of optics
[55:31] diffusers and fan outs
[55:33] and collimators they can be done in
[55:36] different ways
[55:37] that they can be done with diffractive
[55:39] optics can also be done with meta optics
[55:42] or with micro lens arrays
[55:45] and so which one is the best
[55:49] it really is the the key challenge uh
[55:51] necklace is in the integration
[55:53] so i think we need to find a way that we
[55:55] can have
[55:56] a pixel wafer like the one that was
[55:58] presented by true by joseph before
[56:00] and a micro optics wafer level-based
[56:02] manufacturing we like the one that niels
[56:04] does
[56:04] is there any any demonstrator or
[56:07] anything that we could do together i pay
[56:09] for the shipping expenses of micro
[56:10] optics wafer to broadcom from
[56:13] wrong and nil is there anything that we
[56:14] can do today so if you look on the
[56:16] integration side
[56:18] uh what we bring to the table you can
[56:20] see the modules
[56:21] uh here down to the right so you have a
[56:26] conventional module where you have a
[56:28] pixel a cod major
[56:30] user um what we're doing now is we're
[56:33] trying to make this uh
[56:35] cheaper and smaller so simplifying the
[56:38] module
[56:39] and also simplifying the whole assembly
[56:41] chain
[56:42] you need to have an offline discussion
[56:45] this is one of the ones to watch in
[56:46] europe
[56:47] and this is something that you can do
[56:48] but allow me to continue because they
[56:50] have so many people coming asking for
[56:52] helping broadcast so i would like to
[56:54] continue to have a question from one of
[56:56] your
[56:57] potential or maybe current or future
[56:59] customers we're going to go to
[57:01] leo space photonics
[57:05] what's on your mind thanks for saying
[57:08] thanks
[57:09] for the very nice presentation um
[57:12] here what we do we develop transceivers
[57:16] for space applications
[57:17] and specifically for optical inter
[57:20] satellite links
[57:21] optical interests of the light links so
[57:23] basically fiber optic links within the
[57:24] satellite
[57:26] so we approached the vehicle from the
[57:27] harsh environment
[57:29] point of view now the requirements for
[57:31] space
[57:33] although some years back nobody cared
[57:35] about
[57:36] speed and perhaps temperature was not
[57:39] a prime issue and so forth during the
[57:41] past years things have changed a lot
[57:43] so today we are asked to design for 25
[57:46] gigs for a space application
[57:48] and the end users are already asking for
[57:50] 50 gigabit per second
[57:52] and the interesting part is that the
[57:54] space industry although
[57:55] it is considered to be you know a
[57:57] conservative one it is one of the first
[58:00] industries that have adopted the
[58:01] midboard optics approach in order to
[58:04] improve the bandwidth
[58:05] to remove any sort of bandwidth
[58:07] limitations so for space
[58:09] although it sounds peculiar it sounds
[58:11] strange
[58:12] the vehicle is starting to become one of
[58:15] the major bandwidth
[58:16] bottlenecks so the question is the
[58:19] following
[58:21] um the challenge that we have here is
[58:23] that we need
[58:25] now we need high speed and we need high
[58:28] temperature so the temperature of the
[58:30] vehicles will experience will be in the
[58:31] range of 100 degrees c
[58:33] and the transceiver now nobody wants her
[58:37] metric so then
[58:38] so the parts that are inside the
[58:39] transceiver they need to show some sort
[58:40] of a self
[58:42] hermeticity to be able to protect
[58:44] themselves
[58:45] um so my question is the first question
[58:47] is how do you see the visibility of a
[58:49] high
[58:50] bandwidth and high temperature vection
[58:53] from the product
[58:55] parts that you have from the product
[58:56] codes i would say
[58:58] how do you see the feasibility of a v64
[59:01] xt
[59:02] temperature with a v84 lp performance
[59:06] is this something that you can see okay
[59:08] this is something
[59:10] you're gonna get two answers in one
[59:12] because i think broadcom and trump both
[59:14] got to comment on this
[59:15] counter first and then joseph
[59:20] how do you see the scalability of the
[59:22] data rates
[59:24] trying to squeeze the power consumption
[59:26] so okay i understand
[59:27] we can go higher in terms of speed but
[59:30] is there any
[59:31] possibility of keeping the power
[59:32] consumption down of keeping the bias
[59:34] of the direction in uh low does it yeah
[59:38] i think that's possible i mean we can
[59:40] sorry
[59:41] we can we can engineer the uh the
[59:44] buyer's current and keep it still low
[59:46] uh the the issue is when you have a
[59:48] large temperature span i mean when you
[59:50] when you only i mean require a high high
[59:53] high bandwidth at 100 degrees that
[59:56] should be possible
[59:57] the efficiency is less of course but
[01:00:00] other effects are somehow somehow
[01:00:03] better but when you when you want to go
[01:00:06] from minus 40 or even minus 60 to 100 we
[01:00:09] have an issue
[01:00:10] i have to say and so whenever we can
[01:00:13] limit somehow the temperature spent
[01:00:15] then we are in the game and on the other
[01:00:17] side of course
[01:00:19] i think for space we probably need to
[01:00:21] discuss also about redundancy or so
[01:00:24] we have not we have not done space
[01:00:28] we have no experience with let's say the
[01:00:30] the hardening of the technology for
[01:00:33] you need to join our meetings on space
[01:00:35] technology especially on free space
[01:00:38] communications you will be amazed
[01:00:39] the volume of this market this is a high
[01:00:42] volume market now
[01:00:43] but joseph you want to comment as well
[01:00:45] yeah yes i i think because it's one of
[01:00:47] my favorites
[01:00:48] it's it's it's truly one of my favorites
[01:00:50] there's a solution to high temperature
[01:00:52] and that solution is going away from 850
[01:00:54] nanometers to 940 980 or 1000
[01:00:58] because just for physical principles you
[01:01:00] can operate
[01:01:01] a 980 under the same stress conditions
[01:01:05] 40 degrees 30 40 degrees higher than you
[01:01:07] can operate in 850.
[01:01:08] the reason that datacom is with 850
[01:01:12] is just a historical standardization
[01:01:15] reason
[01:01:16] but there's no physical good physical
[01:01:19] argument
[01:01:20] for that choice it's a historical choice
[01:01:22] and
[01:01:24] and if if if there's a compelling reason
[01:01:26] to go to much
[01:01:27] higher temperatures which i expect is
[01:01:29] coming to coming
[01:01:31] and we see that many times then i think
[01:01:34] we have to discuss changing wavelengths
[01:01:36] yeah i think that that every expert in
[01:01:39] the community would confirm this even
[01:01:40] broadcast
[01:01:41] i want to address after the
[01:01:43] presentations by element126 i want to
[01:01:45] have a round on how we can make tunable
[01:01:46] big cells together on wavelengths
[01:01:48] before that we have a question
[01:01:57] thank you very much for joining showing
[01:01:58] that this meeting is european oriented
[01:02:00] but not european exclusive
[01:02:02] what's on your mind thank you jose
[01:02:05] and we're have a lot of business in
[01:02:07] europe so friend of europe as well and
[01:02:09] our headquarters in europe
[01:02:11] so i just want to follow up a question
[01:02:13] first congratulations to ganthea
[01:02:15] broadcom the 100g vixel is very
[01:02:17] impressive
[01:02:18] that's another good news because
[01:02:20] yesterday i saw a news uh
[01:02:21] lumento ocularia had a 100g direct
[01:02:24] modulator lasers
[01:02:26] so that's great um so just want to
[01:02:29] let you know we actually had done some
[01:02:32] passive lens attachments array attached
[01:02:34] onto vixels
[01:02:35] and wafer level packaging and there's
[01:02:39] all different kind of process
[01:02:40] and so i'd like to follow up with you to
[01:02:43] see
[01:02:44] what's your needs if our product can
[01:02:46] help and we have process experience
[01:02:48] to study okay very good so that's that's
[01:02:51] great
[01:02:51] so i um um there's a german saying i
[01:02:55] don't know that that
[01:02:55] lies in a detail or something it's
[01:02:57] always there's always
[01:02:59] at the end a stupid reason why you have
[01:03:02] to build up a discrete module
[01:03:03] whatever but we always want to avoid it
[01:03:06] so having a wafer level technology is
[01:03:08] always the desire
[01:03:09] and maybe for some reasons for this
[01:03:11] transceiver we had to go away again but
[01:03:13] it's already very small
[01:03:15] but sooner or later with some
[01:03:18] let's say flip chip technology or
[01:03:22] some some adding the pixel to the
[01:03:23] silicon and so on
[01:03:25] in a defined way more or less
[01:03:28] we should be we should have another
[01:03:30] scheme yes
[01:03:31] whatever level that's the trend we're
[01:03:33] seeing and particularly for
[01:03:34] smartphone market there's a lot of uh
[01:03:38] the i will say stop a short that
[01:03:42] the lens directly attached to a waiver
[01:03:45] level pixel
[01:03:46] array and that has to be very high
[01:03:48] accurate
[01:03:49] sub micron wave level packaging has to
[01:03:52] be passive
[01:03:54] because you need extremely volume
[01:03:57] kind of a throughput um so uh we're
[01:04:00] seeing a lot
[01:04:01] and something we can contribute a little
[01:04:04] bit and
[01:04:05] we have to work together with device
[01:04:07] guys
[01:04:08] thank you very much yik yang i want to
[01:04:10] continue because they you know what
[01:04:12] people come to this meeting looking for
[01:04:14] pixels they want to find a vixen they go
[01:04:17] with challenges so after the first two
[01:04:18] presentation i think it is time to go to
[01:04:20] another one
[01:04:21] few side is a company making quantum
[01:04:23] random number generators and they are
[01:04:24] looking for a pixel
[01:04:26] they are looking for a pixel domenico
[01:04:27] what are you looking for
[01:04:30] thank you i mean for the introduction so
[01:04:32] yeah i mean uh
[01:04:34] we are a startup based in barcelona just
[01:04:36] a quick introduction about keysight
[01:04:39] we founded the company three years ago
[01:04:41] and our first
[01:04:43] product we launched first product last
[01:04:45] summer
[01:04:46] which is actually uh a quantum random
[01:04:48] number generator
[01:04:50] we are very interested in uh in big cell
[01:04:52] since in the last uh
[01:04:54] two years and a half you've been
[01:04:57] studying uh
[01:04:58] quantum effect in pixel and and we want
[01:05:00] to use it
[01:05:01] for um cost effective
[01:05:04] uh very highly integrated
[01:05:08] component for uh especially for the new
[01:05:10] market such as
[01:05:11] iot and 5g so i mean we are
[01:05:14] looking uh for companies which would
[01:05:18] like to engage with us
[01:05:19] and to develop this highly integrated
[01:05:22] solution
[01:05:23] and we also in the last two years
[01:05:27] uh explored this effect and filed
[01:05:30] dominica are we supposed to see what is
[01:05:31] language
[01:05:32] with the specifications yeah i mean i
[01:05:34] can i can yes because i want the specs
[01:05:36] these people everybody all the pixel
[01:05:38] manufacturers are to be here
[01:05:40] waiting for this and they're they want
[01:05:41] to write them down the numbers
[01:05:43] oh i mean i don't have the numbers here
[01:05:45] but let me get you someone i i like to
[01:05:48] share
[01:05:48] those numbers privately actually if
[01:05:51] that's
[01:05:51] not a problem i just have i mean a
[01:05:54] general
[01:05:54] school show it's choice
[01:06:01] windows share screen and then you select
[01:06:02] the screen that you want to share with
[01:06:03] the universe
[01:06:04] yes i guess you can see it right yes
[01:06:08] okay so i mean as i said we are
[01:06:11] based in barcelona 23 people are now
[01:06:14] 35 of us all the phd in
[01:06:18] related physics or engineering
[01:06:22] and as i said i mean we have
[01:06:26] uh basically developed solutions for
[01:06:29] safer connectivity and advanced
[01:06:31] computation mostly
[01:06:33] quantum random numbers for those two
[01:06:35] main applications at the moment
[01:06:37] and as i said we are looking well i mean
[01:06:40] our
[01:06:41] main other value are basically speed and
[01:06:44] high quality of their numbers we were
[01:06:47] able to generate
[01:06:48] and we developed
[01:06:50] [Music]
[01:06:52] those products and tested in several
[01:06:54] applications like
[01:06:55] data center and performance computing
[01:07:00] those are some use cases you said
[01:07:02] enhanced security
[01:07:04] qkd and and monte carlo simulation
[01:07:07] mostly
[01:07:10] so i mean what i was saying we would
[01:07:13] like to engage with some
[01:07:14] mixed manufacturer to to
[01:07:18] bring those devices to the next level
[01:07:20] especially for consumer electronics
[01:07:22] and and yeah i mean i don't have the
[01:07:25] numbers here but
[01:07:26] um i have all the numbers so if you can
[01:07:30] get in contact with
[01:07:31] uh trump or other another momentum is
[01:07:34] australia i mean
[01:07:37] this is the company that put one of the
[01:07:39] first quantum products in the market
[01:07:41] this is the company that has attracted
[01:07:43] interest from many many companies
[01:07:45] worldwide and amazed
[01:07:46] by the introductions they have made that
[01:07:48] they have follow-up and this is a
[01:07:50] winning horse in quantum technologies so
[01:07:52] please
[01:07:53] i'm going to introduce you to all the
[01:07:54] pixel manufacturers in the room that you
[01:07:56] have been in touch with
[01:07:57] i want them to help you because it's
[01:07:59] very important for us to help you
[01:08:01] after the next two presentations we're
[01:08:02] gonna address the the topic of tunable
[01:08:05] pixels but the next two presentations
[01:08:07] are quite
[01:08:08] interesting we got the two market
[01:08:10] leaders
[01:08:11] on pixel manufacturing lumenum and 26
[01:08:15] back to back who gives more than this
[01:08:18] let's start with lumentum
[01:08:19] my very good friend and vixen lover
[01:08:22] david
[01:08:23] chesky thank you very much for being
[01:08:24] with us this
[01:08:26] afternoon here in the netherlands tell
[01:08:28] us how we can help lumentum being
[01:08:30] greater than already
[01:08:31] the floor and the attention of everyone
[01:08:33] goes to david
[01:08:39] oh sorry one second here
[01:08:44] let's see the background with david and
[01:08:46] we can see him as well
[01:08:48] baby hi there hello good day everybody
[01:08:51] yeah thanks for your time today it's uh
[01:08:53] it's good to be here thanks jose
[01:08:54] um as you know uh you got me really
[01:08:57] excited about your video so i'm
[01:08:59] gonna share with uh everybody here uh
[01:09:01] some great stuff
[01:09:02] in uh manufacturing for vixels and uh
[01:09:05] see where we can get to a lot of great
[01:09:06] challenges there
[01:09:07] a lot of great discussions already and
[01:09:09] um
[01:09:10] i'm going to talk a little bit more
[01:09:11] about where we're at today so let me uh
[01:09:13] without any further ado
[01:09:16] as joseph mentioned before we're
[01:09:17] shipping a lot of devices so we've got
[01:09:19] over a billion devices with zero field
[01:09:23] fuel failures in uh an extraordinarily
[01:09:26] uh high volume
[01:09:26] large scale devices for consumer
[01:09:28] electronics and what that's really done
[01:09:31] is enabled us to build on our
[01:09:33] manufacturing platform and that's the
[01:09:35] key to
[01:09:36] to the next generation of technologies
[01:09:38] and applications really so let me go
[01:09:39] through a little bit with you
[01:09:41] on where we're at with that and we'll
[01:09:43] talk a little about some of the
[01:09:44] challenges so
[01:09:45] um you know where we started this all
[01:09:47] goes back
[01:09:48] you know five or six years ago when we
[01:09:50] really introduced that six-inch
[01:09:52] pixel manufacturing platform and uh you
[01:09:54] know you go back to 2017 there with some
[01:09:56] of the first
[01:09:58] high-volume consumer electronics large
[01:10:00] arrays
[01:10:01] that were introduced there and that
[01:10:02] really changed the manufacturing
[01:10:04] paradigm and the infrastructure and it
[01:10:06] really enabled us to start building on
[01:10:08] top of that
[01:10:09] and i consider that uh gen one of vixel
[01:10:12] manufacturing technologies we've
[01:10:13] improved it over the years
[01:10:15] multi-junction
[01:10:16] epi for vixels was a big improvement
[01:10:18] which started to improve the the power
[01:10:20] conversion wall plug efficiencies
[01:10:22] up to 48 and we're well beyond that
[01:10:24] today
[01:10:26] we've also increased our manufacturing
[01:10:29] capabilities with smaller and tighter
[01:10:30] pitches
[01:10:31] um and today you know we're starting to
[01:10:33] get into that second generation of pixel
[01:10:35] technology and that's really
[01:10:37] what we're talking about today is how do
[01:10:39] we do more integration
[01:10:40] how do we make more functionality and
[01:10:42] things like multi-layers of metal
[01:10:45] making systems matrix addressable we can
[01:10:48] do all of those things today
[01:10:50] and that's really what we're building on
[01:10:51] with new applications
[01:10:53] and proliferating where pixels are today
[01:10:57] in addition you know we continue to
[01:10:58] implement new epi with uh
[01:11:00] now five and six jvic uh epi for vixels
[01:11:04] with 400 watts peak
[01:11:06] a lot of these are automotive
[01:11:07] applications but a lot of that is also
[01:11:09] applicable to
[01:11:10] uh many of the other consumer and
[01:11:13] industrial applications that we're
[01:11:14] talking about here as well
[01:11:16] um and so where are we going in the
[01:11:18] future now we're looking at
[01:11:19] more integration all in one with
[01:11:22] backside emitting
[01:11:23] integrated micro lenses addressability
[01:11:26] uh
[01:11:26] longer wavelengths that we've as we've
[01:11:28] been talking about today at 300
[01:11:30] nanometers 1400 nanometers and beyond
[01:11:32] and then polarizing these light sources
[01:11:34] for advanced applications
[01:11:36] so let me give a brief introduction as
[01:11:37] to what that looks like for
[01:11:39] our second generation of technology in
[01:11:41] the platform here and
[01:11:42] again you know we talked a lot today
[01:11:44] about the challenges and challenges are
[01:11:46] integrating more functionality and what
[01:11:48] are we doing to address that it's really
[01:11:50] again
[01:11:50] four of these things here backside
[01:11:52] emitting integrated micro lenses
[01:11:54] multi junction epi and addressable
[01:11:56] emitters and what that really allows you
[01:11:58] to do is to do
[01:11:59] a lot of the things and complement
[01:12:01] what's being done with micro lenses
[01:12:03] with these uh things that the team that
[01:12:06] nil technologies is doing uh with the
[01:12:08] mems teams
[01:12:09] make those systems more efficient and
[01:12:10] also cost produce them provide more
[01:12:12] integration
[01:12:13] and compact uh systems at the same time
[01:12:16] it gives you a lot more flexibility
[01:12:18] with what you can do with the vixel
[01:12:19] today so i'm going to give you an
[01:12:21] example
[01:12:22] a little bit here of what you can do
[01:12:24] with an addressable array so
[01:12:26] i i show here on the left an image of a
[01:12:28] vixel array
[01:12:29] that's broken into multiple sections and
[01:12:32] you can light up different segments
[01:12:33] at different times and again be able to
[01:12:36] do this
[01:12:37] in a way that allows you to optimize
[01:12:39] your system design
[01:12:40] customize new things and really do
[01:12:42] different things with it
[01:12:44] another example of what that addressable
[01:12:46] array would look like is now we can do
[01:12:48] interleaving of these
[01:12:49] looks a lot like an led where you can
[01:12:51] turn them on and off
[01:12:52] in any configuration whatever you want
[01:12:54] to do and again that's
[01:12:56] all driven by the fact that we can do
[01:12:58] these second generation of technologies
[01:13:00] with
[01:13:00] multi layers of metal with
[01:13:02] addressability um it's
[01:13:04] it's extremely complex it looks very
[01:13:06] simple but it's extremely complex
[01:13:08] primarily because of the the power
[01:13:10] levels and the current that you're
[01:13:11] trying to get to with
[01:13:12] these size of vixel arrays um and then
[01:13:15] another great example is what you can do
[01:13:17] here with kind of like
[01:13:18] two different arrays interleaved within
[01:13:20] each other
[01:13:21] uh you can see we're turning one off and
[01:13:23] the other on at the same time
[01:13:24] uh with different electrodes there and
[01:13:27] you can imagine that that could be in
[01:13:28] any kind of pattern or configuration
[01:13:31] again
[01:13:32] because we are doing the second
[01:13:33] generation of technologies and
[01:13:35] this one on the right here is an example
[01:13:36] of you can modulate the intensity
[01:13:40] of this pixel array and use the same
[01:13:42] collinear optics
[01:13:43] but maintain a constant current source
[01:13:46] so
[01:13:47] you maintain a lot of the wavelength
[01:13:48] stability a lot of the other parameters
[01:13:50] that would change if you try to modulate
[01:13:52] the intensity for
[01:13:53] a single emitter um again this is
[01:13:56] enabled by
[01:13:56] the ability to do addressability at
[01:13:58] multiple layers of metal
[01:14:00] uh building on top of that that first
[01:14:02] generation here in our second generation
[01:14:04] platform
[01:14:05] let me give you another quick example of
[01:14:07] what this looks like with integrated
[01:14:08] optics now we talked a little bit about
[01:14:11] wanting to do complex optics with a doe
[01:14:15] or a diffuser
[01:14:16] and you see what that example would look
[01:14:18] like here on the left with a traditional
[01:14:21] package
[01:14:22] package optics with a driver i see in a
[01:14:23] pixel
[01:14:25] we're working to integrate all of that
[01:14:26] into one chip and put that all on top of
[01:14:29] uh one driver i see and in different
[01:14:33] implementations there you can kind of
[01:14:35] see on the left and the right
[01:14:36] it's the same approach it's the same
[01:14:38] technology using backside
[01:14:40] emission uh it's the same technology
[01:14:42] using micro lenses
[01:14:43] uh but we were able to do different
[01:14:45] things with it you can kind of see
[01:14:46] obviously
[01:14:47] uh that gallium arsenide is is written
[01:14:49] upside down because we're using a flip
[01:14:51] chip approach
[01:14:52] and we're emitting out of the back side
[01:14:53] so the top of that vapor you can see is
[01:14:55] between the gallium arsenide chip and
[01:14:57] the sub mount then on the left
[01:14:58] we've got two contacts three contacts
[01:15:00] there on the right we've got
[01:15:02] multiple contacts so more addressability
[01:15:04] there you can see different lensing
[01:15:06] on the right versus the left and you can
[01:15:08] envision that you could do that
[01:15:10] in different ways within one ship um you
[01:15:12] can do different
[01:15:13] you know light up different areas within
[01:15:15] one ship do
[01:15:16] diffusers collimation um beam steering
[01:15:20] all of these kinds of things can occur
[01:15:22] once you're able to implement this and
[01:15:24] that's really where we're at today
[01:15:25] and i'll show you an example real quick
[01:15:27] of what that looks like um you see what
[01:15:29] the
[01:15:29] chip looks like on the left hand side
[01:15:31] there at the bottom a
[01:15:32] photograph of the backside of a chip a
[01:15:35] whole bunch of emitters there in the
[01:15:36] middle you see
[01:15:37] uh the the um unlens uh near field
[01:15:40] pattern
[01:15:41] how they light up and then in the middle
[01:15:43] there you see the the lens emission uh
[01:15:45] in the near field uh images
[01:15:47] um and you can see the effect of the
[01:15:48] lens is how that spreads out how that
[01:15:50] changes the image the uh the intensity
[01:15:52] pattern
[01:15:53] in the near field and then on the right
[01:15:54] in the far field and so you know the
[01:15:56] challenges here really
[01:15:58] are twofold one is you know what we
[01:16:00] talked about before is this
[01:16:01] addressability
[01:16:02] and how do we do that what do we do with
[01:16:03] it what can we do with it what do you
[01:16:05] want to do with what is
[01:16:06] the limits of the applications and the
[01:16:08] technologies and where can we go with
[01:16:10] that
[01:16:11] um another challenge here is what can
[01:16:13] you do with integration what are the
[01:16:15] what are the kinds of of lenses uh
[01:16:17] integrated optics that that are useful
[01:16:19] for these different systems
[01:16:21] what are the different systems we can
[01:16:22] design with these kinds of capabilities
[01:16:24] built in
[01:16:25] another is how much more stacking can we
[01:16:28] do on the top right there
[01:16:29] and you've got all of this built in with
[01:16:31] things that we know how to do integrated
[01:16:33] micro lenses
[01:16:34] backside emission flip chip how do we
[01:16:36] mount that onto a driver ic make this
[01:16:38] compatible with
[01:16:39] with silicon technology and really be
[01:16:41] able to to go to the next level of
[01:16:43] integration here
[01:16:44] um and again you know one of the one of
[01:16:46] the comments made before was the silicon
[01:16:48] platform and that's what we're really
[01:16:49] looking at right now is how this
[01:16:52] vixel manufacturing platform looks so
[01:16:54] much like a
[01:16:55] an evol and evolution like the uh like
[01:16:58] the silicon manufacturing platform and
[01:17:00] that's what we're able to do today
[01:17:02] and where does this all get done again
[01:17:03] it's your traditional markets today and
[01:17:05] the smartphones
[01:17:06] you see a lot of efforts going into
[01:17:07] automotive both in cabin as well as the
[01:17:10] long range
[01:17:11] lidar applications um but what's really
[01:17:13] exciting today is a lot of this
[01:17:15] integration
[01:17:16] and customization of the optics and the
[01:17:18] electronics
[01:17:20] really enables these iot markets where
[01:17:22] we're seeing things like
[01:17:23] access control and industrial
[01:17:25] applications
[01:17:26] and even wearable applications and
[01:17:28] that's what's really exciting because
[01:17:30] those things are going to drive the
[01:17:32] demands of billions and billions
[01:17:34] more devices and the manufacturing
[01:17:36] platform today
[01:17:37] is robust enough but what we need to do
[01:17:39] next is really
[01:17:41] add on top of that all these extra
[01:17:43] technologies and capabilities
[01:17:44] to be able to supply these markets in a
[01:17:46] cost-effective manner
[01:17:48] and really enable them to get out there
[01:17:49] and that's what's exciting to me about
[01:17:50] pixels today is that
[01:17:52] um you know we've done a great job of
[01:17:54] putting them into devices into billions
[01:17:56] and billions of devices
[01:17:57] all these companies that you're
[01:17:58] listening to today um
[01:18:00] and they're going to go into billions
[01:18:02] and millions more because we've got
[01:18:03] these manufacturing platforms that as
[01:18:05] jose said in his intro you know it makes
[01:18:07] it a cost effective platform that we can
[01:18:09] now do
[01:18:10] a whole bunch of other things that we
[01:18:11] never thought about doing before
[01:18:13] so just real briefly yeah kudos to you
[01:18:16] jose
[01:18:16] great great visions and again things
[01:18:18] i've been talking about for years and
[01:18:19] years and years you know gen one
[01:18:21] we've got these anything from from low
[01:18:23] power to mid power and even today into
[01:18:25] high power and
[01:18:26] you know gen 2 is really a combination
[01:18:28] of these new technologies built on top
[01:18:30] of there
[01:18:30] and then the future direction is really
[01:18:32] where we're going to take this some of
[01:18:33] the discussions we've had with long
[01:18:34] wavelengths and polarized
[01:18:36] systems for advanced technologies i
[01:18:38] won't go through the summary but we're
[01:18:40] really evolving here today
[01:18:41] you know we're focused on manufacturing
[01:18:43] platform new products
[01:18:45] and what we really need as i said those
[01:18:47] challenges before what do we do with
[01:18:48] these things
[01:18:49] and also at the bottom their ecosystem
[01:18:51] partners
[01:18:52] are really important for establishing as
[01:18:55] module integrators detector solutions
[01:18:56] and electronic integration to make
[01:18:58] all of this happen so thank you very
[01:19:01] much for your time look forward to some
[01:19:03] questions thank you very much david
[01:19:06] super nice presentation
[01:19:08] as always so okay i heard a lot of your
[01:19:11] challenges they are mainly related to
[01:19:13] micro optics and packaging right
[01:19:15] so i am sure we are going to have a we
[01:19:18] have companies that are going to comment
[01:19:21] now i'm sure that i actually see a
[01:19:23] question from phi contacting the chat
[01:19:24] but what about testing do you have any
[01:19:27] challenge uh on testing
[01:19:29] did you mention something testing has
[01:19:31] been uh
[01:19:33] an extraordinary challenge from day one
[01:19:35] i mean going back to 2015-16
[01:19:37] testing is what enables all of this to
[01:19:39] happen um
[01:19:41] pixels especially at these wavelengths
[01:19:43] are complex waivers
[01:19:44] and you need extraordinary not just
[01:19:46] electrical and optical testing but also
[01:19:48] mechanical handling systems and
[01:19:49] we've been you know uh uh innovating
[01:19:52] across that
[01:19:53] for all of these years and and uh you
[01:19:55] know it's working with partners that
[01:19:56] allows you to continue to make progress
[01:19:58] there and
[01:19:59] um some of the interesting things we're
[01:20:00] doing today in testing are for
[01:20:02] backside emission which is a completely
[01:20:04] new way to look at a way for
[01:20:06] uh testing on wafer and then also uh
[01:20:09] with integrated optics like diffusers
[01:20:11] collecting that light and analyzing that
[01:20:13] and then also the the longer wavelength
[01:20:15] so
[01:20:15] all of these things are those are the
[01:20:17] challenges today with those those new
[01:20:18] technologies in the test world
[01:20:21] okay because now i see a question in a
[01:20:23] chat in the chat
[01:20:24] from instrument systems a karthik
[01:20:28] a karthik maybe would you like to make
[01:20:32] the question by yourself but maybe you
[01:20:33] have one slide to show
[01:20:35] what the instrument system is doing
[01:20:39] yeah absolutely uh i hope everybody can
[01:20:43] see and hear me
[01:20:45] yes we can see and tell you very well if
[01:20:46] you want to share your screen
[01:20:49] yeah so i will quickly uh give everybody
[01:20:53] a quick uh overview on what we do first
[01:20:56] of all thanks for the invite and it's
[01:20:58] been
[01:20:59] it's a really pleasure it's been a
[01:21:00] really great discussion
[01:21:02] all around on pixel manufacturing
[01:21:05] technologies so we at instrument systems
[01:21:08] we provide uh solutions for
[01:21:12] all sorry yeah we provide solutions for
[01:21:15] all kind of uh pixel testing
[01:21:17] comprehensive along the entire uh value
[01:21:20] chain
[01:21:21] uh so to speak so here i can
[01:21:26] so here you have from the wafer level uh
[01:21:28] to the chip packaging to the driver
[01:21:30] module and
[01:21:32] towards also at the at the end uh device
[01:21:35] level
[01:21:36] so here basically i would like to
[01:21:38] quickly uh show you the two devices that
[01:21:40] we have the
[01:21:41] uh that enable near field and far field
[01:21:44] characterization so the on the left you
[01:21:46] see the near fields
[01:21:47] uh characterization device that is the
[01:21:49] vtc 4000
[01:21:51] states for uh pixel testing camera so
[01:21:53] it's a comprehensive near field
[01:21:55] characterization device it allows you to
[01:21:57] make
[01:21:58] a single emitter analysis including the
[01:22:00] radiant flux the polarization
[01:22:02] angle and degree of each single emitter
[01:22:05] you can of course combine it with a
[01:22:06] spectrometer
[01:22:07] and measure the wavelength of single
[01:22:10] each
[01:22:10] single emitter and apart from that of
[01:22:12] course you can also measure the
[01:22:15] uh the beam quality parameters like the
[01:22:17] waste of the emitters
[01:22:18] uh the numerical aperture the m squared
[01:22:20] value so on and so forth
[01:22:22] uh the good really uh uh great thing
[01:22:24] about is this the camera
[01:22:26] is completely and absolutely calibrated
[01:22:28] and traceable to
[01:22:29] our national institute which gives them
[01:22:32] uh
[01:22:32] really uh unprecedented uh uncertainty
[01:22:35] budgets
[01:22:36] so especially when for example for the
[01:22:40] far
[01:22:40] field which i'm coming to soon
[01:22:43] when it comes to laser eye safety
[01:22:45] parameters or determining the laser
[01:22:47] class of the dod
[01:22:49] such traceable data is always is always
[01:22:52] an add-on advantage
[01:22:55] on the right side you can see the
[01:22:56] far-field characterization device that
[01:22:58] we
[01:22:59] supply it's the vtc 2400 so
[01:23:02] it's basically a comprehensive far-field
[01:23:05] characterization device it
[01:23:06] consists of a monochromatic camera with
[01:23:09] a light permeable screen
[01:23:10] and a mount that you can see here
[01:23:13] uh for the for for the vixel or for the
[01:23:17] duty
[01:23:18] the vtc 2400 also uh is absolutely
[01:23:22] calibrated
[01:23:23] you can measure the absolute radiant
[01:23:25] flux the angular and spatial
[01:23:27] distribution
[01:23:28] of the uh spatial distribution uh of the
[01:23:31] emitted light the opening angle the
[01:23:33] cross section profile
[01:23:34] the emission pattern the the middle and
[01:23:37] the
[01:23:37] outer peaks you can identify the hotspot
[01:23:41] on the screen
[01:23:43] which basically is one of the also an
[01:23:45] important factor in determining uh the
[01:23:48] laser class of the dot
[01:23:49] we've spoken about the eyes eye safety
[01:23:52] considerations
[01:23:53] here and here also we as i mentioned we
[01:23:57] provide a complete traceable uh
[01:24:00] calibration with uncertainty budgets and
[01:24:02] a certificate so
[01:24:03] it's uh it's an important factor for the
[01:24:06] for the eye safety considerations
[01:24:11] okay thank you very much yeah just one
[01:24:14] last slide and i'm done
[01:24:16] sorry um yeah so just wanted to quickly
[01:24:19] summarize so we have the entire
[01:24:20] portfolio for
[01:24:22] test pixel testing solutions for the
[01:24:24] temporal the electrical the optical
[01:24:26] spatial
[01:24:27] uh solutions or the spatial measurements
[01:24:30] uh on top of that we also have a time
[01:24:34] resolution so measurement so
[01:24:35] if we want to we also offer solutions
[01:24:39] at a nanosecond pulsing uh
[01:24:41] characterization for
[01:24:42] for devices that supply uh for devices
[01:24:46] uh that need to be characterized at a
[01:24:49] nanosecond
[01:24:50] policy okay thank you very much for
[01:24:53] this explanation very interesting
[01:24:55] technology to
[01:24:56] test it to test the pixel arrays right
[01:25:00] so okay we have plenty of questions in
[01:25:03] the chat so maybe we can go now to phi
[01:25:05] contact
[01:25:06] greg do you want to make the question by
[01:25:10] yourself
[01:25:23] i'm sorry right um my camera was not
[01:25:26] working
[01:25:26] um for most of this uh session so that's
[01:25:29] when my picture keeps disappearing
[01:25:31] um some very interesting talks i think
[01:25:34] we're hearing a lot
[01:25:35] um about the increasing uh
[01:25:47] greg i think we are experiencing some
[01:25:49] technical problems here
[01:25:51] eh maybe we can come back to you later
[01:25:54] okay
[01:25:55] if you like i'll answer this question
[01:25:57] yes sure
[01:25:58] do you like it yeah this question is uh
[01:26:00] general question uh regarding the arrays
[01:26:02] is vixel redundancy built into array
[01:26:04] stacks
[01:26:05] so um the pixel redundancy is built into
[01:26:08] many products yes
[01:26:09] um you know it really is product
[01:26:12] dependent and system dependent so
[01:26:14] um you know we've got a lot of different
[01:26:16] products that address
[01:26:17] the systems in different ways um within
[01:26:19] the the
[01:26:20] individual epi stack uh it's there is no
[01:26:23] redundancy but we do a lot of testing to
[01:26:25] ensure
[01:26:26] um a lot of the uh pixel
[01:26:29] as you can see on the previous system
[01:26:30] you know we're measuring every single
[01:26:32] vixel in an array we measure
[01:26:33] not just whether they light up but how
[01:26:36] much power they have
[01:26:38] in relative to each other and there's a
[01:26:40] lot of criteria that we need to meet
[01:26:42] depending on the system applications so
[01:26:43] there's some some versions of
[01:26:45] vixel arrays have redundancy uh others
[01:26:48] uh not as critical
[01:26:50] thank you thanks for jumping in can you
[01:26:52] hear me
[01:26:54] yes okay the question i have is are you
[01:26:57] still testing sequentially
[01:27:00] um well i mean no look we light up an
[01:27:02] array at the same time
[01:27:04] so we've got one array and depending on
[01:27:05] whether if there's matrix adjustability
[01:27:07] there may be different kinds of uh
[01:27:09] waveforms that we need to use or
[01:27:10] different functions that we need to test
[01:27:12] them
[01:27:12] in different manners so it really is
[01:27:13] system and application dependent
[01:27:16] thank you david okay thank you very much
[01:27:19] and now
[01:27:20] wilfred from seuss micro optics do you
[01:27:22] have a comment related to
[01:27:24] micro optics yeah thank you um very
[01:27:27] impressive
[01:27:28] what to see that that you can integrate
[01:27:30] it now
[01:27:31] the micro optics in in indirectly in the
[01:27:34] wafer
[01:27:35] in the gallium arsenide how do you see
[01:27:38] um
[01:27:38] the combination of this integrated
[01:27:41] optics with external micro optics so
[01:27:43] we're very strong in imprinted micro
[01:27:45] optics so we do
[01:27:46] we do this for automotive applications
[01:27:48] and really high volume now
[01:27:50] we implant polymer and glass optics so
[01:27:53] the question is
[01:27:54] can you combine your optics with these
[01:27:56] types of optics would there be a chance
[01:27:58] to do this or
[01:28:00] or does it have to be a still
[01:28:04] assembled chip level assembly or even
[01:28:07] even on a package level okay great
[01:28:10] question and
[01:28:11] uh look i think the goal is integrate as
[01:28:14] much as possible on wafer on ship
[01:28:16] right there's some things in optics that
[01:28:18] you can't do at short focal lengths or
[01:28:19] short flow distances so
[01:28:21] um i think what what you can do is
[01:28:24] utilize the on-chip integration to
[01:28:26] optimize your offshore
[01:28:29] optics and whether that's making less
[01:28:31] complex making your packaging less
[01:28:33] complex
[01:28:34] the more you can do on chip and again
[01:28:35] this is this is a a lesson that the
[01:28:37] silicon industry has taught us
[01:28:39] over the last 60 70 years how many years
[01:28:42] is it yeah
[01:28:43] 60 years or so right i can't even can't
[01:28:45] even remember but um
[01:28:46] look it's integrate as much as you can
[01:28:49] more functionality
[01:28:50] brings down the cost brings down the
[01:28:51] system complexity and what that does is
[01:28:54] enables
[01:28:54] new things to be done more ways to do it
[01:28:56] and you know more efficiencies to be
[01:28:58] gained
[01:28:59] and then you guys will advance things
[01:29:00] and then that'll make everything you
[01:29:02] know it's a
[01:29:03] it's a self-fulfilling prophecy so it's
[01:29:05] uh i i would say it's a combination of
[01:29:07] both
[01:29:08] okay and and you know can i have a
[01:29:10] follow-up point
[01:29:13] do you want to do that on wafer level or
[01:29:16] or do you want to integrate that or do
[01:29:17] you leave that up to your clients
[01:29:19] basically this integration no no we we
[01:29:22] do all of the integration on wafer we
[01:29:23] basically provide ships
[01:29:25] to to our uh our customers right
[01:29:28] and so however if we could add more
[01:29:30] functionality now with
[01:29:31] extra packaging correct right
[01:29:35] they asked us to do more functionality
[01:29:37] they also integration because
[01:29:38] they want to to cost reduce you know
[01:29:42] size reduce and improve the performance
[01:29:44] and do different things with their
[01:29:45] systems and that's the way to do it is
[01:29:47] integration
[01:29:47] okay well then we should talk
[01:29:52] okay very good a next question that we
[01:29:54] have in the chat uh
[01:29:56] vadodayal from photonic components a dfm
[01:29:59] do you want to make the question by
[01:30:00] yourself
[01:30:03] a question regarding the size yeah yeah
[01:30:05] yeah actually david
[01:30:06] babu from hong kong yeah yeah yeah
[01:30:09] actually you know
[01:30:11] all the speakers you know have excellent
[01:30:13] presentation telling
[01:30:14] very high volumes so my fundamental
[01:30:16] question do you think
[01:30:18] the excel volumes reaches eight inch fab
[01:30:23] i i don't see it just yet i think
[01:30:25] there's
[01:30:26] a lot of capacity in the in the gallium
[01:30:28] arsenides uh
[01:30:29] pixel space today at six inch um and
[01:30:32] it's scalable
[01:30:33] so um and and i think there's a lot of
[01:30:36] challenges with eight inch but it's it's
[01:30:37] it's not
[01:30:38] it's a conversation you and i've been
[01:30:39] talking about for many years right but
[01:30:41] it's
[01:30:41] yeah yeah yeah challenges with eight
[01:30:43] inch are obviously you can you can get
[01:30:45] more chips on a wafer but it's a
[01:30:47] completely different manufacturing
[01:30:49] platform right and a lot of it is
[01:30:50] silicon based so
[01:30:51] now you're going to start to do things
[01:30:52] in a silicon fab rather than a
[01:30:54] three-five fab and
[01:30:55] um i i think that there's still a lot of
[01:30:58] volume
[01:30:59] uh scalable in in six inch three five
[01:31:02] facts
[01:31:03] okay that's great yeah thank you thank
[01:31:05] you bye
[01:31:08] okay david so i think you mentioned
[01:31:10] before that you were doing the epic
[01:31:12] wafer process by yourself
[01:31:14] in house i mean uh the way for that we
[01:31:17] do
[01:31:18] we do everything uh for fabulous or for
[01:31:21] families okay yeah so then we do some
[01:31:23] in-house uh
[01:31:24] epi and processing for uh other uh
[01:31:28] pixel products and laser products as
[01:31:30] well very good so maybe now is the right
[01:31:33] moment to introduce you one of the
[01:31:34] one of our companies that is launching a
[01:31:37] new ap preferred process for developing
[01:31:39] pixels
[01:31:40] hey jasek from ecosystems are you here
[01:31:43] yes
[01:31:44] yes thank you anna thank you all the
[01:31:47] great presentations here
[01:31:48] i was very interesting to hear all the
[01:31:50] speakers
[01:31:52] and i would be happy to tell a few more
[01:31:53] words about bigger system
[01:31:56] so um uh vico uh
[01:31:59] is a well salvaged company within the
[01:32:00] market for over 30 years
[01:32:02] in photonics and during the time he
[01:32:05] worked with many permanent companies
[01:32:07] who also are on the stock exchange so
[01:32:10] bigo is a public company
[01:32:12] and in 2019 we established a new uh
[01:32:16] three five epi foundry and invited top
[01:32:18] uh experience in neptucci to
[01:32:20] delete this um epi foundry we have the
[01:32:23] g4 reactor
[01:32:24] multiple wave phase configurations very
[01:32:26] easy access to sims
[01:32:28] and many other characterization methods
[01:32:31] and
[01:32:31] we are in the pixels area we focus on
[01:32:34] short and both short and long
[01:32:37] wavelengths
[01:32:39] in our manufacturing capabilities uh we
[01:32:42] showed that with the characterization
[01:32:44] methods we show that we have
[01:32:46] quite great growth control and we meet
[01:32:49] all the tolerances
[01:32:50] in uniformity doping levels
[01:32:54] appearances levels and so on and also
[01:32:56] one
[01:32:57] interesting thing is that in-house we
[01:33:00] have the capability to do the prototypes
[01:33:02] to confirm the quality of our materials
[01:33:05] so yeah thank you very much and we are
[01:33:07] very interested in a collaboration on
[01:33:09] specifically custom niche and
[01:33:12] top performing products thank you
[01:33:18] anna i think you're muted sorry i was
[01:33:21] mute yes
[01:33:22] so thank you very much for this i'm sure
[01:33:24] it's vertical uh
[01:33:25] an extra call with you and david later
[01:33:28] okay so then a let's move ahead in the
[01:33:31] program and let's go for the next
[01:33:33] presentation so yes i am very happy to
[01:33:36] introduce to you a julian from
[01:33:38] 26. so julian if you want the floor is
[01:33:41] yours
[01:33:42] thank you and first of all let me
[01:33:44] apologize my webcam is not working so
[01:33:46] you know not see my face which is maybe
[01:33:48] not special but
[01:33:54] okay so um thanks a lot for
[01:33:57] inviting me to this presentation today
[01:34:01] we've heard a lot of very interesting
[01:34:03] topics
[01:34:04] around pixels around manufacturing
[01:34:06] challenges
[01:34:07] around new technologies what i wanted to
[01:34:11] uh to center my talk on
[01:34:12] is the trends in the pixel illumination
[01:34:15] and trends can be driven either by the
[01:34:17] technology
[01:34:18] or they can be driven by the application
[01:34:20] so i want to spend a little bit of time
[01:34:22] on both those aspects
[01:34:25] um in terms of market
[01:34:29] demand clearly there is there was a
[01:34:32] market demand there is still a market
[01:34:34] demand
[01:34:34] by space and that really fueled
[01:34:37] um do you see my screen yes yes we
[01:34:41] can see it very well no problem
[01:34:45] so it really fueled the demand on the
[01:34:48] very large manufacturing
[01:34:49] infrastructure and david talked about
[01:34:51] the six inch and how the six
[01:34:53] inch is probably sufficient for today's
[01:34:55] demand on pixels
[01:34:57] um so this infrastructure is here and
[01:34:59] it's here to stay
[01:35:00] and now it can be used for new demand
[01:35:04] and the demands for example will come
[01:35:07] from
[01:35:08] currently the face unlock for example
[01:35:09] and the photography and the world facing
[01:35:12] cameras and mobile phone and will deploy
[01:35:15] in the future
[01:35:16] into augmented reality and when i say
[01:35:18] augmented reality that means automatic
[01:35:19] reality mixed reality virtual reality
[01:35:24] another field where more and more 3d
[01:35:27] cameras
[01:35:28] are being implemented is in white goods
[01:35:31] and the the first simpler example is
[01:35:33] probably a vacuum cleaner
[01:35:34] where you want a more efficient mapping
[01:35:37] of the surrounding
[01:35:38] you also want to make sure that you are
[01:35:42] spanning the entire space that you want
[01:35:44] to vacuum but
[01:35:45] other white goods um will also
[01:35:49] get their 3d camera for user
[01:35:51] authentication for example for
[01:35:53] customization
[01:35:54] and even if you generalize white goods
[01:35:56] into uh
[01:35:57] cameras and what used to be called
[01:36:00] webcam or surveillance cameras
[01:36:03] then 3d cameras will become very
[01:36:06] prevalent
[01:36:08] then uh it was mentioned briefly i think
[01:36:12] david mentioned it before
[01:36:13] in the automotive space we have both the
[01:36:16] in-cabin applications
[01:36:17] where you want to monitor drivers or
[01:36:20] driver monitoring systems
[01:36:22] or you want to monitor the occupants
[01:36:24] occupancy monitoring systems
[01:36:26] and for example you can measure with a
[01:36:28] 3d camera you can measure the body
[01:36:30] position
[01:36:31] so that the airbag can be triggered
[01:36:33] without any
[01:36:34] risks of injuries you can also customize
[01:36:38] the cabin based on who is driving
[01:36:40] for example the ac preference or the
[01:36:42] seat
[01:36:44] position there is also gesture control
[01:36:47] for infotainment
[01:36:48] so there is a really a lot of
[01:36:50] applications that are enabled
[01:36:52] by a 3d camera in in cabin application
[01:36:55] and then exterior automotive that's
[01:36:57] typically what we talk
[01:36:59] when we say lidar or adas autonomous
[01:37:02] driving
[01:37:03] and that again can be several type of
[01:37:06] application can be collision avoidance
[01:37:08] pedestrian detection
[01:37:09] uh can be the auto
[01:37:13] automatic braking systems so there are
[01:37:15] really a lot of systems that would rely
[01:37:17] on the depth information both in
[01:37:18] cabin exterior
[01:37:22] and finally what we are seeing as well
[01:37:25] is that more and more in the industrial
[01:37:28] space
[01:37:28] maybe what's going with the term factory
[01:37:30] 4.0
[01:37:32] also here you want to map your
[01:37:33] surrounding and you want to
[01:37:35] have the most efficient um
[01:37:39] understanding of the environment that
[01:37:42] you're
[01:37:43] navigating so for example for automatic
[01:37:46] forklifts
[01:37:47] you want to make sure that you recognize
[01:37:49] the shape of the packages that you
[01:37:50] are handling so that you don't damage
[01:37:52] them and that you can carry them safely
[01:37:54] to the place where you want to draw
[01:37:58] so i'm not going to go into the
[01:38:00] specifics of each of these applications
[01:38:02] but each of these applications will have
[01:38:04] specifics
[01:38:05] on the illumination that means on the
[01:38:07] pixel that's also on the pixel system so
[01:38:10] what it needs to be integrated with
[01:38:12] for example driver optics um
[01:38:18] and that's why it's important to
[01:38:20] understand these trends
[01:38:22] so that you can steer the technology of
[01:38:24] your illumination systems
[01:38:26] to address them
[01:38:30] so one of the trends uh and it was
[01:38:32] mentioned
[01:38:33] earlier as well by other speakers is
[01:38:36] multi-junction pixels
[01:38:38] um in a multi-junction pixel maybe for
[01:38:41] those that are not familiar with it
[01:38:43] um you so in a standard pixel with a
[01:38:46] single junction you have a
[01:38:47] gain region which is sandwiched between
[01:38:49] two mirrors
[01:38:50] and the light oscillates between the two
[01:38:52] mirrors gets amplified
[01:38:53] starts lazing that's why the light comes
[01:38:56] out from the top
[01:38:57] that's why it's a vehicle in a
[01:38:58] multi-junction pixel
[01:39:00] you stack multiple gain region between
[01:39:02] the two mirrors
[01:39:04] so what it means is that for the same
[01:39:05] surface area of the mission you get
[01:39:07] higher peak power
[01:39:09] so you increase the power density of
[01:39:11] your pixel chip
[01:39:12] which has many advantages one of them
[01:39:15] obviously is that because the power
[01:39:17] density is increased
[01:39:18] for a customer that looks for a certain
[01:39:20] power level the chip size is reduced and
[01:39:22] the chip size usually scales with it
[01:39:24] with the price so the cost is then we
[01:39:27] lose
[01:39:28] but on top of that um you operate by
[01:39:31] cascading
[01:39:32] several gain regions that means you
[01:39:33] increase the voltage
[01:39:35] um the operating voltage but you reduce
[01:39:37] the current
[01:39:38] and when you want to go to very high
[01:39:40] peak power which is typically done with
[01:39:42] high current
[01:39:43] for a single junction it's very hard to
[01:39:45] design a driver
[01:39:46] that can drive fast enough to steal a
[01:39:48] high level of peak
[01:39:50] current so when you increase the voltage
[01:39:52] you simplify the driver design and you
[01:39:54] enable a faster ryzen prototype
[01:39:58] so for example in the applications such
[01:40:02] as
[01:40:02] direct time of flight such as for
[01:40:04] example lidar
[01:40:06] this is very advantageous this is an
[01:40:09] example of a five junction
[01:40:10] so five junctions that are stacked on
[01:40:12] top of each other five junction chip
[01:40:15] uh about a one meter square chip a
[01:40:17] smaller knitting area
[01:40:19] um where we get about a bit more than
[01:40:21] 500 watts of
[01:40:22] peak power which amounts to 1.2 kilowatt
[01:40:25] per
[01:40:26] square and clearly that's a regime
[01:40:28] that's very interesting
[01:40:30] for direct time of flight and especially
[01:40:33] for flash
[01:40:34] systems direct time of light where it
[01:40:36] might start to compete with
[01:40:37] other automotive systems based on
[01:40:40] edumating lasers
[01:40:43] um when you use this type of chips as
[01:40:45] well in a
[01:40:46] mobile application or in mobile
[01:40:49] applications
[01:40:50] the advantage is you get smaller chip
[01:40:52] you get lower cost
[01:40:54] but also because you drive them at your
[01:40:56] current
[01:40:57] um the impact of electromagnetic
[01:41:00] interference is reduced
[01:41:02] and when you have a very compact system
[01:41:04] um
[01:41:05] where the let's say the driver the chip
[01:41:07] and maybe an antenna
[01:41:08] are in close proximity this is important
[01:41:12] you don't want your 3d camera
[01:41:14] elimination system to interfere with
[01:41:16] your reception of your phone for example
[01:41:18] mobile phone
[01:41:23] in terms of the system trends
[01:41:28] the typical illumination for a time of
[01:41:30] flight is what's called a flood
[01:41:31] illumination
[01:41:32] where you have a pixel uh and let's say
[01:41:36] a diffuser and the
[01:41:37] role of the diffuser is to shape the
[01:41:39] beam that is emitted from the ignition
[01:41:40] system
[01:41:41] to shape it so that it matches the
[01:41:44] camera field of view
[01:41:46] um as we talked about
[01:41:50] this can be integrated so typically
[01:41:52] traditionally it has been done
[01:41:54] at the heterogeneous level so there is a
[01:41:56] pixel and there is a diffuser in the
[01:41:57] package
[01:41:58] but now more and more these two
[01:42:00] functions get integrated
[01:42:02] and we even talked about with a scale
[01:42:04] optics there that
[01:42:05] that is definitely a trend um
[01:42:09] then we can go so that the advantage of
[01:42:11] these flood eliminations is
[01:42:12] it's rather simple you illuminate the
[01:42:15] entire scene at once
[01:42:17] and you detect the depth information of
[01:42:20] the entire scene
[01:42:21] also at once so you have no lagging
[01:42:22] effect no ghosting effects
[01:42:24] but the problem is that now you dilute
[01:42:26] your emitted power
[01:42:28] to the entire scene that means that you
[01:42:30] need to drive
[01:42:31] the illumination to pretty high peak
[01:42:33] power to reach a certain threshold on
[01:42:35] each pixel on the camera
[01:42:37] now the alternative is that instead of
[01:42:39] illuminating the entire scene you
[01:42:40] eliminate one single pixel at a time
[01:42:42] scanning system so here you have a laser
[01:42:45] you have a scanning system that deflects
[01:42:47] the beam to the right position on the
[01:42:49] scene
[01:42:50] and then gets the image on the detector
[01:42:53] now the advantage of that is that you
[01:42:54] have very
[01:42:55] very high peak power density at the
[01:42:57] level of the pixel
[01:42:58] so you can get to very long reach the
[01:43:01] inconvenience is that you have a
[01:43:02] mechanical system to scan
[01:43:04] which is not desirable especially in
[01:43:06] automotive systems
[01:43:08] and and you also have the lag that you
[01:43:11] need to raster
[01:43:11] the beam across the entire ceiling so
[01:43:13] that will lead to coasting artifacts
[01:43:16] another alternative which is maybe also
[01:43:19] something that we see as a trend
[01:43:22] is uh it's kind of an intermediate
[01:43:24] between the two
[01:43:25] instead of flooding the entire scene um
[01:43:29] with power you actually project dots
[01:43:32] and those dots then concentrate the
[01:43:35] amount of power
[01:43:36] on a position that gets them and then
[01:43:39] gets image on
[01:43:40] the sensor so it's a way to increase the
[01:43:44] reach that the
[01:43:45] typical flood illumination system will
[01:43:47] be able to reach
[01:43:48] um by again concentrating the power on
[01:43:51] slots
[01:43:54] and finally we also talked about it
[01:43:57] david
[01:43:58] should show some data there is
[01:44:01] addressability addressability can do
[01:44:04] several things
[01:44:05] so here is an example of why it's useful
[01:44:07] i talked about the scanning system that
[01:44:09] requires a moving part
[01:44:11] to deflect the beam in the right
[01:44:12] direction and lidar
[01:44:14] but now if you have an addressable array
[01:44:15] and you imagine here that you have an
[01:44:17] array with
[01:44:18] each line being addressable individually
[01:44:22] if you turn on the laser one by one with
[01:44:23] an optical system in front
[01:44:26] then you can imagine that your beam will
[01:44:28] then scan
[01:44:30] the scene at different time and here
[01:44:33] again you get
[01:44:34] a concentration of the power on the line
[01:44:36] that you're illuminating
[01:44:38] now the drawback well the advantage
[01:44:40] let's say first
[01:44:41] is that you have no moving part which
[01:44:43] for a lidar is very
[01:44:44] desirable the drawback is that you trade
[01:44:47] off
[01:44:48] the moving parts for a larger chip where
[01:44:51] you need to have this addressability but
[01:44:54] on the other hand because you benefit
[01:44:56] from the manufacturing infrastructure of
[01:44:57] the pixel
[01:44:58] this trade-off tends to be beneficial
[01:45:04] and maybe the last slide to give you an
[01:45:07] overview of
[01:45:08] who 26 is and what we do so we are
[01:45:11] one of the biggest suppliers of pixels
[01:45:14] out there we've heard numbers of billion
[01:45:17] of lasers
[01:45:18] billion uh which is a really outstanding
[01:45:22] number
[01:45:22] thinking back like five or six years ago
[01:45:25] so we have shipped also in this range
[01:45:27] probably more than 30 billion emitters
[01:45:30] probably even in one year so those
[01:45:33] numbers are
[01:45:33] real um pixel manufacturing is in full
[01:45:36] swing
[01:45:37] and we benefit from the six inch
[01:45:40] manufacturing that we have
[01:45:41] and in our case it's all vertically
[01:45:43] integrated
[01:45:45] so we can do single emitters which you
[01:45:46] see here this is used for example as a
[01:45:49] proximity sensor
[01:45:50] we can do arrays where they're all in
[01:45:52] parallel
[01:45:54] we also package those pixels into smt
[01:45:56] packages with a diffuser for example
[01:45:59] we can do packages with more integrated
[01:46:01] functions
[01:46:02] for example with the driver um
[01:46:05] and there will be more and more
[01:46:08] development
[01:46:09] with more and more functionality both at
[01:46:10] the package level and the cheap level
[01:46:12] moving forward thank you
[01:46:17] very much julian super nice
[01:46:19] thank you very much julian super nice
[01:46:19] presentation so
[01:46:21] okay you have been showing and you have
[01:46:23] been explaining different challenges
[01:46:26] and maybe you can propose to the
[01:46:28] companies that are in this meeting
[01:46:30] a few challenges how how they can help
[01:46:33] you
[01:46:35] um yeah i mean one topic that came back
[01:46:38] several times is testing
[01:46:40] and uh testing in a way is a challenge
[01:46:43] but i also agree with
[01:46:45] the first uh presentation from joseph
[01:46:46] that we all need tests
[01:46:48] because we have to if we could get ric
[01:46:51] if we could get away without testing uh
[01:46:53] that would be very beneficial so that
[01:46:55] goes back to
[01:46:56] the comment about the epi and app
[01:46:58] uniformity and
[01:46:59] control and inline controls those are
[01:47:02] very
[01:47:03] uh important but to come back to the
[01:47:06] testing
[01:47:06] where we have the challenges around the
[01:47:09] automotive space and short force
[01:47:10] applications when we drive now arrays to
[01:47:13] the range of what i showed the 500 watt
[01:47:15] with let's say currents of 150 amps in
[01:47:18] nanosecond
[01:47:20] this is not something that you can
[01:47:21] easily do and this is something that
[01:47:23] you cannot do at this this moment on
[01:47:26] wafer level
[01:47:27] so that means that if we need to test
[01:47:29] statistics
[01:47:31] uh whole way through it is still a
[01:47:33] challenge and at the same level
[01:47:35] for reliability when we're used to
[01:47:37] testing thousands of chips
[01:47:38] for reliability models uh also here when
[01:47:41] we want to do these reliability models
[01:47:43] in short falls
[01:47:44] high peak power this remains still a
[01:47:48] challenge
[01:47:51] okay thank you very much julian yes i
[01:47:52] see testing is something that
[01:47:54] it's a challenge that all the
[01:47:56] presentations had in common during this
[01:47:58] afternoon
[01:47:59] okay so maybe now i would like to give
[01:48:02] the floor to
[01:48:05] do you have a comment right i sure do um
[01:48:09] i i definitely agree on the uh the eight
[01:48:12] inches
[01:48:12] it's not it's not hardly around the
[01:48:14] corner the gallium arsenide
[01:48:15] we're going to stick with the six inch
[01:48:17] uh i just wanted to make a few points uh
[01:48:19] given that we just uh had the two
[01:48:21] dominant vendors speak and i i think
[01:48:23] it's important
[01:48:24] uh you know that we point out that
[01:48:25] momentum did pull off with winston me a
[01:48:28] minor miracle in 2017
[01:48:30] with a six inch and uh uh there's no
[01:48:33] doubt about that
[01:48:34] but we're going all all this time
[01:48:36] without mentioning
[01:48:37] the big elephant in the room and that's
[01:48:39] apple apple's the big customer here
[01:48:41] and the problem with apple is uh they uh
[01:48:44] don't exactly give good direction and uh
[01:48:47] in that context i would like to ask uh
[01:48:50] if i can
[01:48:51] both david and julian uh
[01:48:54] you know doesn't that become problematic
[01:48:57] in that
[01:48:58] don't you need especially in lumeniums
[01:49:00] case because uh
[01:49:02] you're using you know you have that
[01:49:04] particular business model
[01:49:05] uh that's heavily dependent on outside
[01:49:08] vendors uh
[01:49:09] how do you implore these vendors to be
[01:49:11] able to change on a dime uh
[01:49:14] and with uh with the uh 2-6
[01:49:18] 2-6 ought to be congratulated because
[01:49:20] you have uh
[01:49:21] you really come back on the vertical
[01:49:23] integration uh the last days
[01:49:25] of uh you know fittessar you know was it
[01:49:28] was problematic
[01:49:29] uh uh i know you lost at least one uh
[01:49:33] good designer that was
[01:49:34] very much in touch with with apple uh
[01:49:37] so my question uh to 26 in particular to
[01:49:41] juuling would be
[01:49:43] um is there a need uh for outside
[01:49:45] designers that might help you going
[01:49:47] forward
[01:49:50] okay so i i see there are two questions
[01:49:53] here one is for julian for
[01:49:55] two six and one is for david momentum
[01:49:57] who wants to start
[01:49:58] it's julian's turn you can go first sure
[01:50:02] julian go ahead thanks such a gentleman
[01:50:06] so i believe the question was about
[01:50:08] whether we see the need for
[01:50:10] exterior designer chip design
[01:50:13] in general we tend to do we tend to have
[01:50:16] the philosophy of being practically
[01:50:17] integrated that
[01:50:18] also increase the design however
[01:50:21] if there are some skill sets
[01:50:24] or there are certain reasons of
[01:50:26] bandwidth for example
[01:50:28] we are not adverse also to contacting
[01:50:31] consultants for experts but let's say
[01:50:34] that's not the standard practice
[01:50:37] and um and to comment to your statement
[01:50:40] about finisar and the achieve design so
[01:50:44] we still have we have a lot of cheap
[01:50:46] designers and they don't
[01:50:48] all come from the finisher side we also
[01:50:50] had a lot of chip designers from the 36
[01:50:51] side
[01:50:52] so the combination of the two is really
[01:50:54] uh world-class
[01:50:56] chip design capabilities
[01:51:00] okay thank you and david now would you
[01:51:02] like to comment
[01:51:04] yeah sure no i i think you got a lot of
[01:51:06] great points there mark and julian you
[01:51:08] know
[01:51:09] uh a lot of successes there at uh at 2 6
[01:51:11] and and i think your point mark
[01:51:13] is really well made is that five six
[01:51:15] years ago this was a monumental effort
[01:51:18] by
[01:51:18] uh several groups of uh talented
[01:51:21] companies
[01:51:22] are really working hard it was kind of a
[01:51:24] a moonshot mission
[01:51:26] and uh it worked it's successful um uh
[01:51:29] you know before that
[01:51:30] momentum was also an integrated device
[01:51:32] manufacturer but
[01:51:33] we found that uh you know with the
[01:51:35] economics of the
[01:51:36] uh the consumer electronics market um it
[01:51:39] really makes sense to move to the same
[01:51:41] model that has been proven to be
[01:51:42] effective for
[01:51:44] across all of the markets and
[01:51:45] technologies for the most part
[01:51:47] especially in consumer electronics so
[01:51:49] question how do you get these companies
[01:51:51] to scale with you and change
[01:51:52] on a dime that's what they've been doing
[01:51:54] for years and years and years and this
[01:51:55] is
[01:51:56] just the first time that they've been
[01:51:57] doing it for pixels and for optics and
[01:51:59] that's really the difference is that
[01:52:01] this is not it's not
[01:52:03] how do you continue to change it's what
[01:52:05] do you change to next
[01:52:06] because now we can do all of these
[01:52:08] things we can scale we've got so many
[01:52:10] great talented people and organizations
[01:52:12] that can do these things and
[01:52:13] we're aggressively going after these new
[01:52:15] business opportunities
[01:52:18] okay thank you very much for this
[01:52:19] discussion now uh julian i would like to
[01:52:22] introduce you one of the
[01:52:23] companies that we have here in this
[01:52:25] meeting and
[01:52:26] i'm sure a company has a 2-6 you are
[01:52:29] always looking for new materials and
[01:52:32] it gets more resistant material so we
[01:52:33] have here is ben from delos
[01:52:36] espen maybe would you like now to take
[01:52:38] the floor and explain a little bit
[01:52:40] what is the law doing yeah thank you
[01:52:43] anna
[01:52:45] let me share my screen
[01:52:51] so now it should work
[01:52:55] so i'm from the company dalo we are a
[01:52:58] manufacturer of
[01:52:59] industrial adhesives and high-tech
[01:53:02] materials
[01:53:03] and what we can offer is or our
[01:53:06] materials for the lenses for example
[01:53:09] or for the whole pixel packaging to
[01:53:13] bond and to fix it
[01:53:16] what uh what the advantages of
[01:53:19] our materials are that we are
[01:53:22] have a teas adhesion on various
[01:53:26] substrates
[01:53:27] on the typical mixture substrates have a
[01:53:30] low out gassing behavior
[01:53:32] no bleeding to on the substrates can
[01:53:35] provide small bond lines high
[01:53:38] temperature stabilities and
[01:53:40] many other behaviors here
[01:53:44] we are also deliver ultra black
[01:53:48] materials to block some light or
[01:53:52] yeah to defuse the materials
[01:53:56] lens bonding in general active alignment
[01:54:00] so many things if you have any questions
[01:54:03] contact us
[01:54:04] on the photonics experts at wda
[01:54:07] or visit us at the photonics plus
[01:54:10] event which is coming up in the next
[01:54:13] weeks
[01:54:14] that we also attend and have our own
[01:54:16] booth so if you have
[01:54:17] any challenges in the packaging please
[01:54:20] come back to us we are
[01:54:23] looking forward to your challenges here
[01:54:27] thank you very much has been a very nice
[01:54:30] a
[01:54:30] very nice slide and now maybe is the
[01:54:32] moment
[01:54:33] to introduce another company that we
[01:54:35] have in the room another
[01:54:37] pixel manufacturer a christian from
[01:54:39] bertilas
[01:54:42] would you like to comment a little bit
[01:54:45] what is bertila
[01:54:46] doing um i think most of it has already
[01:54:50] been said let me switch on my camera
[01:54:52] just a moment
[01:54:56] hi hi everybody hi hello christian how
[01:54:59] are you doing
[01:55:01] hanging in all right yeah
[01:55:04] i think most of those things have
[01:55:05] already been touched on
[01:55:07] i mean we as a long wavelength pixel
[01:55:09] manufacturer always
[01:55:10] of course interested in uh the
[01:55:12] applications going long wavelength
[01:55:15] i think i heard some comments uh in
[01:55:17] earlier presentations like
[01:55:20] from broadcom mentioned that he sees
[01:55:22] some demand
[01:55:23] in higher wavelength other than that
[01:55:27] of course we face as virtual as the same
[01:55:30] challenges as
[01:55:31] all the other manufacturers we have all
[01:55:33] these challenges on design side
[01:55:36] um i think broadcom and the others
[01:55:38] showed uh
[01:55:39] high bandwidth pixels on gallium
[01:55:41] arsenide basis going up to 30 gigahertz
[01:55:44] if you look at the in the indian
[01:55:45] phosphite space uh what we have is the
[01:55:48] base technology
[01:55:49] we are i want to say i'm more or less
[01:55:52] stuck around
[01:55:53] 20 gigahertz right now so we are trying
[01:55:55] to find the path
[01:55:56] up to 30 gigahertz uh having said this
[01:55:59] our lasers are can be modulated up to 50
[01:56:02] gigabit per second
[01:56:04] uh and anna you mentioned this project
[01:56:06] earlier the u project the passion
[01:56:09] where we uh realized uh a two terabit
[01:56:13] per second integrated transmitter based
[01:56:16] on pixel technology
[01:56:18] and silicon photonics so that's a big
[01:56:20] thing for us it really is really looking
[01:56:22] at the integration assembly technology
[01:56:25] of taking our vehicles and combining
[01:56:27] them with the silicon photonics
[01:56:29] technology
[01:56:30] and in the first step of course as a
[01:56:32] hybrid technology
[01:56:34] in the future maybe we are seeing also
[01:56:37] possibilities to really directly
[01:56:38] integrate
[01:56:39] uh three five materials together with
[01:56:43] silicon photonic platforms
[01:56:45] um yeah what what else um i've seen
[01:56:48] a lot of talk about that assembly and
[01:56:51] test
[01:56:51] test is a big thing for us we do a lot
[01:56:55] of testing as anybody else in the pixel
[01:56:57] space
[01:56:58] uh we test during the course of the
[01:57:01] production of the wafers of course
[01:57:03] uh there was talk about how do you check
[01:57:06] the
[01:57:07] epi structures if they are in spec or
[01:57:09] not you don't want to find out once you
[01:57:11] finish
[01:57:13] the wafer that's not not the right
[01:57:16] time to do so we have a lot of testing
[01:57:19] actually implemented and
[01:57:20] also latest steps later in the
[01:57:23] production
[01:57:24] phase where we can compensate for
[01:57:27] inaccuracies that may happen
[01:57:29] in the epitaxy manufacturing or in the
[01:57:31] processing
[01:57:32] so that's really key
[01:57:36] but what's also very important for us is
[01:57:38] we are mostly selling into the single
[01:57:40] mode laser space today but we are
[01:57:41] looking into the
[01:57:43] 2d array space now so we have prototypes
[01:57:45] we are
[01:57:46] engaging with potential customers right
[01:57:49] now
[01:57:50] so the packaging of you know high power
[01:57:53] pixel arrays is a key to us so
[01:57:55] we're looking out for uh partners
[01:57:58] in that area and then anything around
[01:58:01] burning and testing today we
[01:58:03] we sell single pixels so every every
[01:58:06] pixel we
[01:58:06] we ship today sees the burn in is being
[01:58:09] tested numerous times
[01:58:11] uh we're also looking to scale that
[01:58:13] everything uh
[01:58:14] we we're using today we have developed
[01:58:16] in-house because of a lack of
[01:58:18] really standardized equipment being
[01:58:20] available
[01:58:21] and also at the high price tag of those
[01:58:25] types of equipment so those are the key
[01:58:27] topics we are looking at
[01:58:29] it's been a really wonderful discussion
[01:58:31] so far and thanks to everybody
[01:58:33] contributing to this it's really a great
[01:58:34] community thanks
[01:58:36] thank you very much christian and now
[01:58:38] let's move to the last day
[01:58:40] talk of the of the agenda so last but
[01:58:42] not least
[01:58:43] i want to introduce you to phil from
[01:58:45] bandwidth 10. so
[01:58:47] feel if you want a the floor is yours
[01:58:50] thank you jose so bam with tim has
[01:58:53] started in 2011.
[01:58:55] can you go to the next slide
[01:58:58] and and basically the goal of the
[01:59:02] company is to
[01:59:03] commercialize tunable pixel technology
[01:59:05] by using hcg
[01:59:07] technology that was developed at
[01:59:08] berkeley uh today we're 55 people
[01:59:11] worldwide with a core team
[01:59:15] in u.s and taiwan and
[01:59:18] uh 11 phds in the organization so we
[01:59:22] have a very strong design
[01:59:23] development team and as many people
[01:59:26] indicate
[01:59:27] we are doing a lot of modeling and
[01:59:30] focusing on the development of the
[01:59:32] technology
[01:59:33] today we're doing a wafer scale
[01:59:36] integration of the mems technology in
[01:59:38] the device
[01:59:40] you want to go to the next slide
[01:59:44] as you can see on the right hand side
[01:59:46] over here we have a bim
[01:59:48] structure that actually uh moves
[01:59:50] changing the cavity length of the device
[01:59:52] which allows you to tune single mode
[01:59:54] devices
[01:59:56] long wavelength uh we started out at
[01:59:58] 1550 we have also
[02:00:00] integrated devices at 1060 and other
[02:00:02] wavelengths
[02:00:04] shortly we'll have wavelengths as it's
[02:00:05] 1310 and
[02:00:08] there's certainly no reason why we can't
[02:00:11] move to
[02:00:12] 720 and so on we've demonstrated we've
[02:00:14] delivered 1653
[02:00:16] devices as well for different
[02:00:19] applications
[02:00:20] but having the mems integrated device
[02:00:23] allows you to have a very low cost
[02:00:26] swept source or you can actually lock it
[02:00:29] onto a wavelength to modulate it and
[02:00:30] stay within 100 gig
[02:00:34] spacing for optical uh
[02:00:40] communication applications and
[02:00:45] do 10 gig modulation as
[02:00:49] uh vertilos indicated we're also looking
[02:00:52] at going up to 25
[02:00:54] gig and higher speeds next slide
[02:01:01] one of the issues that i've heard many
[02:01:03] many times is you can't make a high
[02:01:05] reliable
[02:01:05] pixel at long wavelengths
[02:01:09] one of the challenges as everybody has
[02:01:12] talked about is testing these devices
[02:01:16] to get significant amount of test data
[02:01:19] it takes a long time and as you can see
[02:01:21] we've got
[02:01:22] 18 000 which is a couple years of
[02:01:24] operation
[02:01:26] uh on devices of 155
[02:01:29] devices on one group here plus we've got
[02:01:32] another
[02:01:34] group that's actually a smaller
[02:01:37] cell and you can see we've got over
[02:01:40] three and a half million device hours on
[02:01:42] the device
[02:01:43] we're very good fit rates the end
[02:01:46] very good confidence level however the
[02:01:48] biggest challenge of that
[02:01:50] to get there is to be able to test the
[02:01:52] device early on
[02:01:54] and as everybody indicated that testing
[02:01:57] the wafer
[02:01:59] today is a very critical part of it
[02:02:02] however
[02:02:03] in most applications testing the wafer
[02:02:05] can't tell you if the device is going to
[02:02:07] work
[02:02:08] to your specifications so you have to
[02:02:11] fully build the device and test it
[02:02:13] before
[02:02:14] you can actually have good
[02:02:15] characterization of the device and know
[02:02:17] if it's going to meet all the
[02:02:18] specifications
[02:02:20] requirement next slide
[02:02:26] we're focusing on a wide variety of
[02:02:28] applications and now
[02:02:30] you can see fiber bright grading gas
[02:02:32] sensing swept sources for
[02:02:34] such as oct as well as lidar and other
[02:02:37] applications
[02:02:38] medical imaging is a very important part
[02:02:40] of the area where you're doing oct
[02:02:43] as well as some other applications that
[02:02:46] customer came up with where they're
[02:02:48] looking at
[02:02:49] doing uh special imaging
[02:02:52] to look at cancer cells and so on
[02:02:55] uh lidar applications proximity uh
[02:02:58] people are using the tunable source to
[02:03:01] be able to get higher resolution regards
[02:03:03] to lidar applications
[02:03:04] optical communications so obviously use
[02:03:07] vmcc
[02:03:09] allows you to have a tunable link that
[02:03:11] is a very low cost
[02:03:13] solution to the overall application otdr
[02:03:18] so it's another application where
[02:03:20] basically people are looking at
[02:03:22] time domain reflectometry be able to
[02:03:25] modulate the device as well as tune it
[02:03:28] allows you to get very uh good
[02:03:30] resolution
[02:03:32] mode hop free be able to hold the device
[02:03:35] stable
[02:03:37] and tune it to a particular wavelength
[02:03:39] is going to enable
[02:03:40] technology that is not available today
[02:03:44] i mean trying to manufacture a device to
[02:03:47] a very precise wavelength
[02:03:49] is very difficult however you can easily
[02:03:51] tune it to a
[02:03:52] very precise wavelength and maintain it
[02:03:55] there
[02:03:57] uh today we offer solutions and from a
[02:04:00] basically chip all the way down to the
[02:04:02] driver's uh device so we can do teal
[02:04:04] cans toes
[02:04:06] application fiber pigtails and tunable
[02:04:08] transceivers
[02:04:10] uh the key in the near future is be able
[02:04:13] to broaden
[02:04:14] the tuning range so the tuning range is
[02:04:17] uh much broader for instance full c-band
[02:04:20] full
[02:04:21] l-band and so on but you will also see
[02:04:24] [Music]
[02:04:27] higher power as well as a
[02:04:30] faster modulation bandwidth next slide
[02:04:37] our value statement basically is we can
[02:04:40] supply
[02:04:41] a swept source uh for different
[02:04:44] applications we have very narrow line
[02:04:46] with
[02:04:48] we certainly can't get as general as
[02:04:49] some edge emitters but
[02:04:51] the line width is much better than most
[02:04:54] people
[02:04:54] expect and we have multiple wavelengths
[02:04:58] the 1654 wavelengths all the way down to
[02:05:01] 1060 wavelengths in the very near future
[02:05:03] we'll have 1310 devices
[02:05:07] we do design modeling and
[02:05:10] we basically purchase epi outside we do
[02:05:13] have a vertically integrated fab where
[02:05:15] we're
[02:05:16] doing everything from doing wafer
[02:05:18] processing all the way up to
[02:05:21] package assembled devices and we're
[02:05:24] working with cms to do
[02:05:26] some of our external packaging and so on
[02:05:28] outside so we can do volume
[02:05:30] wrap up quickly next slide
[02:05:41] okay uh what can bandwidth 10 do
[02:05:45] we can deliver cost effective tunable
[02:05:47] swept sources
[02:05:49] for a wide variety of different
[02:05:50] applications
[02:05:52] we need to understand what your needs
[02:05:53] are and what your applications are
[02:05:56] what can you do for us as we can work
[02:05:59] with you to help
[02:06:00] integrate the solution as well as
[02:06:02] understand
[02:06:03] the pros and cons of pixel tunable pixel
[02:06:08] and how to integrate the device into
[02:06:10] your system
[02:06:11] with minimal amount of difficulty
[02:06:15] so you need to understand what your
[02:06:17] application is and
[02:06:18] how how to optimize the device for your
[02:06:22] application
[02:06:24] thank you very much i appreciate all
[02:06:25] your time and efforts
[02:06:28] and i appreciate your time i already
[02:06:29] apologize you're going to finish a few
[02:06:31] minutes late but i think it's totally
[02:06:32] worth it the topic of tunable pixels is
[02:06:34] very very important
[02:06:36] thank you for answering the epic
[02:06:38] question philip but i want to talk
[02:06:40] to your potential customers and see what
[02:06:43] are their needs we have one of the key
[02:06:45] companies
[02:06:45] in the world of lidar in the room the
[02:06:48] company is
[02:06:49] stick thank you very much for being with
[02:06:51] us eckhart gerster
[02:06:53] if you came to a meeting full of pixel
[02:06:55] manufacturers
[02:06:56] can you tell me your needs and also can
[02:06:59] you
[02:07:00] can you see there is some room for
[02:07:01] cooperation with bank with 10 of the
[02:07:02] tinable pixel
[02:07:07] you know we make this meeting live and
[02:07:09] sometimes we hire robots to make the
[02:07:11] voice over
[02:07:12] and this kind of thing we hire a very
[02:07:13] good one eckhart can we try again
[02:07:17] i think we're having some sound
[02:07:19] crawlings with eckhart
[02:07:22] herker let me come back to you and we'll
[02:07:24] hire you a better robot for the for the
[02:07:26] voiceover
[02:07:27] before that i think there is some clear
[02:07:29] questions in the room
[02:07:30] greg from ficontech what's on your mind
[02:07:36] well i hope you can hear me this time
[02:07:38] loud and clear and
[02:07:40] good the um what i'm picking up out of
[02:07:43] this meeting is i almost have the
[02:07:44] impression that
[02:07:45] the testing was if you like the the
[02:07:49] plate second cousin to the idea of
[02:07:51] manufacturing
[02:07:52] and functional developments i almost
[02:07:53] have the impression that out of this
[02:07:55] meeting
[02:07:56] you could have potentially three in the
[02:07:57] future right one on
[02:07:59] one on applications one on functional
[02:08:02] development and another one on testing
[02:08:04] and what i'm going yes i'm going
[02:08:06] integration on different
[02:08:08] on different application fields yeah we
[02:08:10] need to do this and the idea is to go
[02:08:12] into that but phil
[02:08:13] i want to reshape the question from greg
[02:08:15] on to the testing
[02:08:16] when it comes to the tunable pixels
[02:08:18] testing is complex
[02:08:20] testing is difficult and testing is to
[02:08:22] be fast how do you address it do you
[02:08:24] test every pixel
[02:08:25] individually and you tune over the whole
[02:08:27] wavelength range
[02:08:28] or do you have like an average of in in
[02:08:31] vixel arrays
[02:08:32] no every device is tested individually
[02:08:35] on a wafer scale testing to start with
[02:08:39] so we tune the device completely across
[02:08:41] the tuning range and test it
[02:08:43] and then on the discrete depart after
[02:08:46] it's assembled
[02:08:47] we also test the part individually yes
[02:08:50] it does take a significant amount of
[02:08:52] time and that
[02:08:53] today is one of the biggest bottlenecks
[02:08:55] that we have regards to
[02:08:57] uh testing the device on a wafer scale
[02:09:00] as well as
[02:09:00] testing the device on discrete
[02:09:03] components such as a to package or a
[02:09:05] fiber pigtail package
[02:09:07] the problem with integration with the
[02:09:09] integration vertical integration of
[02:09:10] pixels is we normally typically complex
[02:09:13] the testing processes
[02:09:15] on that we do have many companies in
[02:09:16] epic that can help with that for example
[02:09:18] road microtech
[02:09:19] reinhardt is with us in the room if
[02:09:21] anybody needs some some testing
[02:09:23] challenges they need to they need to
[02:09:24] contact reinhardt
[02:09:25] good to see you rode microtech here in
[02:09:27] the room let's go back to seek
[02:09:29] let's see if the robot is stick or
[02:09:31] healthy right now eckhart
[02:09:33] can you hear me now loud and clear thank
[02:09:35] you for okay
[02:09:36] sorry about that this was a technical
[02:09:39] problem
[02:09:40] the requirements we have is mainly
[02:09:43] brightness
[02:09:44] and reliability so multi-junction is a
[02:09:47] good good path
[02:09:48] definitely and yeah reliability
[02:09:51] is is also very very important for us as
[02:09:54] well
[02:09:54] so that's why i have a question to phil
[02:09:57] regarding the
[02:09:58] the activation energy he was mentioning
[02:10:01] so did you determine and measure these
[02:10:04] values you you mentioned 0.5
[02:10:06] sorry 0.7 electron volt and 0.35 or is
[02:10:11] it
[02:10:11] was this just taken from literature no
[02:10:14] we've actually done
[02:10:15] devices at multiple temperatures as well
[02:10:18] as multiple
[02:10:19] drive currents and actually done the
[02:10:21] measurements
[02:10:23] so it's a fairly
[02:10:27] significant amount of work to actually
[02:10:30] come up with those numbers
[02:10:32] definitely yes that's why i'm asking
[02:10:35] okay thank you
[02:10:36] when it comes to the pixels in the lidar
[02:10:38] segment eckhart we've been talking about
[02:10:40] this
[02:10:41] for already a couple of years what is
[02:10:43] the current state of the art
[02:10:45] are there big cells today integrated or
[02:10:48] how long does it take
[02:10:49] or most important what do you need from
[02:10:51] all these guys i brought they did they
[02:10:53] creme de la creme
[02:10:54] what do you need from this to to make a
[02:10:56] pixel
[02:10:57] in the lighter system so there
[02:11:01] as of today there are vixes already
[02:11:03] integrated in
[02:11:04] in lidar systems that's that's yeah
[02:11:07] state of the art but
[02:11:09] compared to edge emitting lasers the
[02:11:12] vixels are still
[02:11:13] still lower in brightness and that's why
[02:11:16] we
[02:11:16] we strive to to get
[02:11:20] more higher brightness mixes
[02:11:23] higher brightness pixels not not more
[02:11:26] pixels
[02:11:27] in a race no no no not more vixens
[02:11:30] vixx's array does not really make too
[02:11:32] much sense then we have more power but
[02:11:34] we have also more
[02:11:35] area so we need high power on a small
[02:11:38] in a small area source area source of
[02:11:41] area
[02:11:44] number in your head in because i measure
[02:11:46] power in vats do you have like a number
[02:11:48] like
[02:11:49] like a magic number that i could i have
[02:11:50] everyone here i even have fleer here as
[02:11:52] well so
[02:11:53] is there a number in your head
[02:11:57] in terms of jewels or if we could if if
[02:12:00] we could
[02:12:01] could just
[02:12:05] increase the the area of a single mode
[02:12:07] pixel then it would be fine but but
[02:12:09] typically the the the uh
[02:12:13] light power per area is decreasing for a
[02:12:16] higher
[02:12:16] higher um area pixels always
[02:12:20] always unfortunately but this is not the
[02:12:22] case for for multi junction
[02:12:24] it is it is not but it is the case for
[02:12:27] for pixels
[02:12:28] with that request you left it so open i
[02:12:30] expect to make a lot of introductions
[02:12:32] for you i would like to remind if you're
[02:12:33] going to get
[02:12:34] in touch with any of the participants
[02:12:35] today do send me an email consider
[02:12:37] postatepictestasup.com
[02:12:38] and i will make the introduction philip
[02:12:40] thank you so much for having the first
[02:12:42] morning coffee with us
[02:12:44] i would like to finish this meeting by
[02:12:47] going back to where we started i would
[02:12:49] like to go back to joseph
[02:12:50] from trump so you could do my job today
[02:12:53] i would like to delegate giving the
[02:12:55] final remark
[02:12:56] to trump because i'm so happy that you
[02:12:58] were here
[02:12:59] what do you think we should do what do
[02:13:00] you think of the pixel community that we
[02:13:02] have put together at epic
[02:13:04] and what do you think we should do
[02:13:05] together now
[02:13:07] well well i think as i said in my last
[02:13:10] slide thinking about the next 10 years i
[02:13:12] think
[02:13:12] the challenge is really to fight the
[02:13:14] applications that are driving the volume
[02:13:16] that
[02:13:16] that are also driving the revenues for
[02:13:18] all of us such that we can scale and
[02:13:20] bring all
[02:13:21] all the the the nice concept the nice
[02:13:24] ideas
[02:13:24] nice invention on the road i think
[02:13:27] that's that's uh
[02:13:28] that's that's what we're all looking for
[02:13:30] we have seen that
[02:13:32] clear mega trend and you started and
[02:13:34] everybody really agreed with you it's a
[02:13:35] mega trend adding pixel functionality by
[02:13:38] vertical
[02:13:39] integration and that was that was
[02:13:40] fascinating that was all the idea when i
[02:13:42] came to the meeting and all of you
[02:13:44] individually told me that this is the
[02:13:45] way to go and we're going to do a lot of
[02:13:47] things to go in the direction
[02:13:49] the integration with micro optics we're
[02:13:50] going to push this like crazy
[02:13:52] next meeting integration with mems we
[02:13:54] didn't speak speak to sound microsisters
[02:13:56] but they were in the room
[02:13:57] interaction with them something that
[02:13:58] we're going to push a lot we
[02:13:59] underestimate the difficulty of the air
[02:14:02] gap
[02:14:03] in high reliability systems we need to
[02:14:05] address this incoming meeting
[02:14:07] the process monitoring the epic wafer
[02:14:09] and the pixel post processing
[02:14:11] and accelerating testing and the guild
[02:14:13] calculation those are challenges that we
[02:14:15] need to address
[02:14:16] with this community there is a of course
[02:14:18] a drawback like everything
[02:14:19] every time you add functionality you
[02:14:21] increase the cost and complexity which
[02:14:22] is something that we are going to
[02:14:23] address as well
[02:14:24] for me this was really a truly fantastic
[02:14:27] meeting but also
[02:14:28] was the beginning of an adventure epic
[02:14:30] is here to make more and more big self
[02:14:32] events and as greg free profile context
[02:14:35] said we need to organize more
[02:14:36] pixel meetings on this topic so we
[02:14:39] prepare we are going to bring this
[02:14:41] community together and bring more and
[02:14:42] more end users coming here
[02:14:44] looking for vertical cavity surface and
[02:14:46] universes
[02:14:48] and soon we are going to start traveling
[02:14:50] again so all i want to ask you is it
[02:14:52] doesn't matter if it's spicer
[02:14:53] if he's trasenica if he's moderna if his
[02:14:56] giants go liguantosis get vaccinated as
[02:14:58] soon as possible because i need to start
[02:15:01] traveling again i would like to organize
[02:15:03] a vixen meeting
[02:15:04] at a key location bring all of you
[02:15:06] together and then go for dinner
[02:15:08] and have a very nice beer afterwards all
[02:15:10] right happen
[02:15:12] stay healthy wash your hands and wear a
[02:15:14] mask see you soon bye