# SABIC GRAND TOUR AND THE CHALLENGE OF REFLOW SOLDERING

https://www.youtube.com/watch?v=wTGr8Io2VVE

[00:08] I am in Bergen op Zoom, in the Netherlands, at SABIC Photonics Centre of Excellence.
[00:13] And I'm here because I want to understand how thermoplastic material can drive the data at the AI demand rate.
[00:21] Harrie Hoekstra, you are the chief scientist at SABIC in the optics sector.
[00:25] Could you tell us a bit what is this laboratory for?
[00:29] Sure, Harrie.
[00:32] Well, as you know, any micro optics like micro lens array are key components in the modern photonics.
[00:37] Uh they're used to manipulate light uh for sensing, uh for coupling of the fiber to the chip.
[00:45] And uh and we want to uh yeah, show you today that thermoplastic resins are a good material to select for these micro lens arrays.
[00:54] When your customers come to this laboratory and they see this, by the way, this is fascinating.
[00:57] When they see this, why should they care?
[01:01] Yeah, well, this actually we're going to
[01:03] Show you today all the way from the resin chemistry to the compounding, to the molding, the testing of the molded articles all the way up to the design.
[01:12] Uh and how customers can actually use our design advice to to pick our material in their end application.
[01:18] But first we start with the basics.
[01:21] So, what we do here is called compounding.
[01:23] We add different polymers, different resins.
[01:25] We add them with a loss-in-weight feeder into a compounding machine where we have a transportation screw with kneading elements to form a homogeneous blend of the two resins.
[01:37] Here, at the die sides, we have a molten strand, which is cooled in a water bath, and then being transported into a cutter.
[01:48] As you can see here, here we have the cutter.
[01:52] And in the end, we end up with these pellets.
[01:54] These pellets we sell as SABIC Specialties to our customers, which are the micro molders.
[02:00] But at SABIC Specialties, we want to be more than
[02:04] only a resin seller.
[02:07] So, we want to go up in the value chain.
[02:09] We help our micro molders with selecting the right machinery for molding.
[02:13] So, now that we have the compound ready, what are the next steps?
[02:16] The next step actually is micro molding.
[02:18] So, with micro molding, we turn these pellets into an optical element.
[02:23] And we do that actually by having a steel mold and do a diamond turning very high precision.
[02:31] We generate a cavity.
[02:31] In that cavity, we inject our plastic at very high temperature, very high pressure.
[02:36] And then we can injection mold very small optical parts, 10 mg or even 100 mg.
[02:43] And we do that by very high replication accuracy.
[02:47] Gabri, I guess there is not a chance that you can show me the micro molding station.
[02:51] All right, Jose. Let's go to the micro molder.
[02:52] Let's go.
[02:55] So, Gabri, where [music] are we now?
[02:57] This is the processing lab.
[02:57] Here we do the injection molding.
[03:00] We have very [music] big machines, obviously.
[03:02] But today, we talk about micro molding.
[03:02] So,
[03:05] We're going to our smallest machine we have here, our micro molding machine.
[03:09] And my colleague Young Joong Choi, lead scientist within SABIC, will explain how the micro molder works with X Ultem.
[03:17] Young Joong, thank you so much for being with us today.
[03:19] Young Joong, this place is is massive.
[03:20] So, for me it's it's fascinating that we're talking about micro molding in such a huge place.
[03:28] Yep. Actually, this is our processing lab.
[03:30] Here we have all the injection molding machines of all different sizes.
[03:33] So, we can mold the larger parts like a bumper or a fender.
[03:38] But we will go to the micro molding machine that can mold the very small optical element at high precision.
[03:44] So, let's go to the our micro injection molding.
[03:45] From the point of view of the customer, why should they care about this?
[03:50] Because here we test our materials that our materials can be injection molded or different shapes of the part from small size to bigger size also there you can see some of the examples.
[04:01] So here we test our materials that can be suitable
[04:07] for applications that customers do care.
[04:10] Yeah.
[04:11] Gabry just showed us the the compounding station.
[04:13] We saw the material and now you can make here micro lenses free for micro optics.
[04:18] You can make the micro optical elements.
[04:22] Yeah, exactly.
[04:22] So let's see here we have our micro molding machine.
[04:26] Let's start here.
[04:28] So you saw how our granulates are made.
[04:31] So here we feed our granulate into this hopper.
[04:34] So this is screw will melt it down those granulates into the melted form and those molten polymers are injected into a mold cavity which is high precision machine by a diamond turning.
[04:49] We call high precision what micrometers?
[04:52] It's a sub micron accuracy.
[04:55] So we can mold we can mold a very small part for example 20 mg part with surface then surface roughness over 5 nanometer range.
[05:03] Yeah.
[05:05] You can do less than 5 nanometer surface
[05:07] Roughness? Yeah.
[05:10] Can I actually see some of these molds?
[05:10] Yeah, sure.
[05:10] We have some samples.
[05:15] So there we can make millions of part at exactly same shape.
[05:22] So here we have the micro lens arrays molded.
[05:26] This part has around 25 mg of the part size.
[05:31] So it's very tiny and that's why we have this and 20 times bigger micro lens array model.
[05:37] So you can see we have the eight lenses in a single row and the lens to lens pitch distance is 250 micron and the surface roughness is 5 nanometer length scale.
[05:51] So, we can produce the identical part at a million per year.
[05:55] So, this injection molding machine is highly cost-effective uh high volume mass production technique using our thermoplastic resins.
[06:05] For example here, you can see eight
[06:08] identical lenses in a single shot.
[06:11] So, we can further improve that kind of the the throughput either 16 or 32 lenses per single shot.
[06:19] Sheng-Chung, I have done my homework.
[06:21] I know that the darker color are X TEM and the lighter color ULTEM lenses.
[06:23] So, X TEM has been for many years the market leader for reflow soldering.
[06:31] Yes, exactly.
[06:32] of applications are you looking at for these microlenses?
[06:35] This can be used for co-packaged optics in a data center because all the electronic component should go through the reflow process for electronic assembly.
[06:46] And the other one, the lighter color, the ULTEM, what are the main applications you target?
[06:51] ULTEM is mainly used for pluggable optics, also for active optical cables.
[06:57] It is kind of the material choice for that kind of the data com applications.
[07:01] So, that is the micro molding station.
[07:04] What is this?
[07:06] So, this is the biggest molding machine
[07:08] we have in Bergen op Zoom.
[07:10] This is a completely the opposite application.
[07:12] This is intended to mold the big part, the huge part.
[07:17] For example, we have one application or one example here.
[07:20] So, it has multiple gating points.
[07:23] So, we call it sequential gate.
[07:26] So, we have inject material at the multiple different locations and we can mold also huge big part, for example, bumpers, fenders, that kind of applications.
[07:36] So, we do beyond this kind of micro molding or optics.
[07:39] It can be automotive or health care.
[07:44] Also water management is a solid provide material in a diverse applications.
[07:49] It is so nice to have here the micro molding for the micro optics and then the macro molding for the large thermoplastic structures.
[07:56] It is it gives a perspective of what the kind of things that we can do with this material.
[08:00] Really great.
[08:01] You cannot do anything if you cannot measure it.
[08:05] Can we go now to the metrology side?
[08:08] Optical lab.
[08:09] Let's go.
[08:13] Gabi, we have a rule at Optica.
[08:15] If you cannot measure, you cannot sell it.
[08:18] You have to measure all your structure, properly characterize them.
[08:21] And you told me that there is an optical lab here in which you do this metrology side.
[08:24] Can we see it?
[08:25] Sure.
[08:25] Let's go to the optical lab right now and I'll explain you, Jose.
[08:30] Jianzhong already gave us a teaser about X 10 and Ultem differences.
[08:35] How can customers decide whether they should use one material or the other?
[08:38] Yeah, well, Jose, there are actually three or maybe four reasons to pick our material.
[08:43] First of all, free-form optics, right?
[08:44] With injection molding, you can make any kind of shape.
[08:46] MLAs or for instance this particular part,
[08:52] which is a model 10 times the size of the real X 10 part.
[08:56] You can see there is an lens array here coupled with a fiber.
[09:01] And you have a same fiber array on the bottom at the big side.
[09:03] So, the light is bended 90 degrees via internal reflection.
[09:09] You can also add alignment pins.
[09:11] And these alignment pins actually take care that the alignment with the laser and detector is not lost if you go to higher temperatures for instance in the data center.
[09:20] So, these are the options you have with thermoplastic resins.
[09:22] And as you can imagine, not so easy to do with glass.
[09:28] So, second reason actually is well, scalability.
[09:30] I think as Jianzhong explained, we can make many many parts in a year, Uh, especially with multi-cavity molding.
[09:38] Uh, and that is important, right?
[09:40] If CPO becomes mainstream, then you have to have mass production capability.
[09:44] Third reason.
[09:46] Well, with Ultem, we have already a heritage of being used in applications like uh, pluggable optical transceivers.
[09:54] Um, and well, customers select Ultem because it has the lowest CTE amongst other thermoplastic resins.
[10:00] So, yeah, of course that is very important.
[10:03] We also pass uh, industry test protocols like Telcordia.
[10:06] And that is also a key to to be able to specify it in this
[10:11] Space.
[10:12] And then finally, we developed a new resin called Extem, which is a thermoplastic polyamides.
[10:16] And that has a glass transition temperature of 280° C.
[10:23] And therefore, you can use Extem as optics if you want to make sure it withstands reflow soldering.
[10:27] So, now with Extem, you can actually do the packaging and alignment before the final reflow reflow step.
[10:37] And that is a very helpful for the outside companies.
[10:38] It's very unique to combine optical material with a flow soldering.
[10:41] What are the main applications, the main market drivers of this?
[10:46] Yeah, well, we hear a lot that for co-packaged optics, you want to really align the optics already on the board before the final reflow step.
[10:54] Not only with CPO, but also with sensors, optical sensors, time of flight.
[10:57] Uh, we see a new trend in automotive that people want to go to optical ethernet connections.
[11:03] And they all want to align the optics before the solder reflow step.
[11:07] And if you want to use a thermoplastic resin, obviously there's only one resin that you can pick, and that is Extem.
[11:12] How do we have the optical lab at the testing lab?
[11:14] What are you measuring here?
[11:17] Yeah, well, uh, what we measure here is not what we have normally have on standard data sheets, right?
[11:22] The standard data sheets are more for the injection molders, our direct customers.
[11:26] But for an optical designer or optical engineer, what he or she needs are the optical constants.
[11:31] So, optical constants are actually the refractive index as function of temperature, function of wavelength.
[11:38] And that's what we measure with this refractometer.
[11:42] So, here we uh measure the optical constants, we add them to our data sheets,
[11:48] uh and more importantly, we add them into a regular optical solver packages like Ansys and Zemax.
[11:52] So, that means that the optical designer can use his standard solver to design the part with with our material.
[12:02] But what an optical designer also needs is the transmission.
[12:04] So, let's go to the next machine.
[12:07] This is a spectrophotometer um where we show the optical transmission of an uh Extem part, 1 mm
[12:15] color plaque.
[12:17] And you can see that the transmission is very high at the typical O-band or C-band for single mode.
[12:24] It's also uh transparent at the multi-mode wavelengths, 850 or 960 nm.
[12:29] Transmission is roughly 87, 88%.
[12:34] Some customers want to increase uh to close to 100%.
[12:36] So, what we have done, we contacted a couple of anti-reflective coating companies,
[12:43] and they developed an recipe for the anti-reflective coating uh that adheres to Ultem and Extem.
[12:48] It stays intact, so there's no cracking upon reflow soldering.
[12:54] So, if you really want to go to close to 100% transmission, then we can contact the customers with these companies.
[13:01] By the way, we also work with uh optical adhesive companies.
[13:05] So, a lot of companies have developed optical adhesives that work very well with Extem and Ultem.
[13:10] Sometimes even adhesives adhesive works as a kind of CTE
[13:16] implication risk between the substrate and uh and Extem.
[13:20] Refractive index and transmission are very important parameters for optical designers, but when you want to make a proper optical design, tolerances are actually also very important.
[13:29] How you characterize the different tolerances in both the parameters?
[13:32] We go to the next machine.
[13:34] Uh where we have our metrology uh set up and Youngjun's going to explain what kind of things he can measure with the Maxcope.
[13:41] So here is the microscope.
[13:43] What are you going to be doing with this?
[13:44] So this is optical microscope.
[13:47] So also Abby mentioned the design freedom of injection molded part.
[13:51] As you can see here, this is very complex lens shape.
[13:54] So we have the eight lenses in the RX and TX side.
[13:58] So that also has this alignment pin to be connected with MPO connector.
[14:03] Also there's another matching lens arrays to be connected to photonic integrated circuit.
[14:11] Also we have the glue patches here.
[14:13] So we can measure for example the pin to
[14:16] pin distance, which is very critical for alignment of the lens to fiber array.
[14:23] So we can also measure all these kind of uh lens dimensions accurately.
[14:29] Also this is multi-mode.
[14:32] So the tolerance is a bit more relaxed compared to the single mode fiber optics.
[14:37] But also as Abby has developed the single mode co-packaged optics lens arrays.
[14:39] So you are also under micro molding shop.
[14:45] I showcased the uh say injection molding of such micro lens arrays.
[14:48] But here is a single mode fiber optics.
[14:50] The alignment tolerance is much more stringent.
[14:55] So also here we have the laser confocal microscope.
[14:57] For example.
[15:00] So the lens to lens distance pitch is 250 micron currently.
[15:04] And these lens arrays should be accurately aligned with the photonic integrated circuit and fiber arrays.
[15:11] So using this laser confocal microscope, we have
[15:17] developed metrology methodology which can measure lens to distance well below one micron accuracy.
[15:25] So we achieved 0.1 micron gauge R&R.
[15:28] So we can measure the lens distance very accurately.
[15:34] Also, we can measure the 3D profile of the lenses.
[15:37] So you uh downstairs you told us that you can manufacture millions of units.
[15:42] Are you measuring each of them?
[15:44] No, this is not a production site.
[15:46] So we select some of those samples.
[15:48] We test the batch process.
[15:50] But uh that kind of testing can be I would say automatized in a real production environment.
[15:57] However, for me the main message that I want to give the audience is that if you want the right material to have it compatible with a flow soldering, you have to count the Savic, especially for the co-packaged optics.
[16:07] What is this?
[16:09] Yeah, this is a reflow machine.
[16:11] And why is that important?
[16:13] I mean, in uh in in a modern photonics world, the electronics and the optics are
[16:17] packaged extremely close to each other,
[16:19] right?
[16:19] So that means that your optics need to withstand typical electronic assembly techniques like surface mounting technology or reflow soldering.
[16:29] Well, reflow is done uh at 260° C.
[16:32] So most thermoplastic resins will deform at that temperature.
[16:36] So at Sabic Specialties, we worked very hard the last couple of years to develop new monomers, new resins with a glass transition temperature of 280° C.
[16:47] So you still can do the injection molding, uh but you end up with a micro-optic part that withstands reflow soldering.
[16:53] And why is that important?
[16:55] Well, now all of a sudden, you have the option of using thermoplastic resins.
[17:00] You can align your optics on the board before the final reflow step.
[17:05] And that uh saves a lot of cost for the OSA companies.
[17:08] What I can tell you is that from the photonic design perspective, very few people take this into account.
[17:14] And then they go to the packaging house and the packaging house told them about reflow soldering and they have to come back in
[17:18] Design.
[17:21] At what step do you want to talk to optical designers?
[17:22] That's a good question.
[17:23] Yeah, well, actually they will not invest directly in in the in micro molding.
[17:28] They will for instance start with a prototype.
[17:30] This is for instance a very nice prototype and before they go commercial, they want to make sure that the lens bit size, so the distance between the different lenses from the side and from the bottom, but also the distance between the two alignment pins, they will not change upon reflow soldering, right?
[17:48] Otherwise, you lose the alignment.
[17:50] So, what we then do, we invite those customers to come with their prototypes and what we do, we do the reflow test here.
[17:58] Here you see the actual part for made from Axtem.
[18:01] We just put them in the reflow machine, put them on the conveyor belt and here is the protocol, a typical Gidec protocol where we go a little bit higher than 260 degrees three times.
[18:15] That's what the Gidec protocol dictates.
[18:17] And then at the end, after a few
[18:19] minutes.
[18:24] we take the parts and then we're going to measure critical dimensions for this part in our metrology lab where we just have been.
[18:32] And if the
[18:34] And if the distance is acceptable, and if the delta dimension is acceptable, and of course it's a little bit more critical with single mode versus multi mode, then our customer can decide to go from prototype to real commercial parts.
[18:51] How do you guarantee that the part will be working?
[18:53] There are certain tolerances for each application.
[18:56] Yes.
[18:56] Well, of course there is a CTE challenge, right?
[18:59] The Axtem has a higher CTE as the substrate.
[19:00] So, what we then do, we add some fixation points like alignment pins that there is a thermal expansion when the temperature goes up, but due to these fixation points, the alignment is not lost in the lateral direction.
[19:15] Video thing is the final customer, the one that comes to you.
[19:16] But, I guess there are different OSATs in the world.
[19:19] Like I can mention a few like Amkor.
[19:21] Jabil.
[19:24] They When they want to enter new markets like CPO, how do they interact with you?
[19:27] Yeah, these companies really want to stick to 260° reflow soldering.
[19:33] And, they also want to use thermoplastic resins for the reasons I just explained earlier.
[19:36] But, then you need a resin that survives that 260 reflow.
[19:41] Because all the machines at the OSAT companies are already at 260° and they don't want to lower the temperature for us.
[19:47] So, therefore we developed this resin called Axtom.
[19:50] This has been fantastic.
[19:52] I would like to sit down, have a cup of coffee with you and talk about applications.
[19:55] Let's do that.
[19:55] Let's go to the demonstrators.
[20:05] I'm so excited.
[20:06] We saw the compounding, we saw the micro molding, we saw the metrology.
[20:09] Let's talk about applications.
[20:11] What is making you excited these days?
[20:12] Yeah, we built some demonstrators to convince customers to use our material.
[20:16] So, we've been to OFC the last few years where we had a booth and where we showed
[20:19] to the customers Axtom demonstrators.
[20:22] At OFC 25,
[20:25] we had a service coupler, a grating
[20:27] coupler. And, that was exciting. It was
[20:30] the first thermoplastic resin surviving
[20:31] reflow.
[20:33] But, the feedback we got from customers
[20:34] were twofold. They want to see
[20:36] detachability of the fiber array unit.
[20:39] And, they want to also see an edge
[20:40] coupler. So, we spent quite some time
[20:43] together with Tyndall to develop this uh
[20:46] particular demonstrator where we used
[20:48] two Axtom MLAs between the fiber array
[20:51] unit and the PIC with a kind of expanded
[20:53] beam optics. And, my colleague Yang Yun
[20:56] will explain in more detail.
[20:58] >> Detachable fiber connectors are really
[21:00] the future. They're the only way that we
[21:02] can manufacture co-packaged optics in
[21:04] volume production. You should know Yang
[21:05] Yun that they are at the Optica Global
[21:07] Advanced Manufacturing Alliance meeting
[21:09] on the 2nd and 3rd of June. We have a
[21:10] whole panel to talk about detachable
[21:12] connectors. Why should they use your
[21:14] material?
[21:15] >> Yeah, sure. Here you can see the one
[21:17] lens array is attached to fiber array
[21:20] and another lens array is edge coupled
[21:23] to the peak side. So here we have the
[21:25] SLD light source. So this light is laser
[21:28] source comes from one of the fiber. This
[21:31] lens array will expand the beam and
[21:33] collimate. And another lens array will
[21:36] focus the expanded beam to peak side.
[21:40] Then this peak is a simple wave guide
[21:42] loop. Then it returns back and passing
[21:45] through another mating lens arrays and
[21:48] goes to this optical power meter. So
[21:51] here you can see the signal loss in a
[21:54] loop back wave channel. So this beam is
[21:57] expanded. Also the tolerance is a lot
[21:59] more relaxed. So here we can have this
[22:02] kind of detachable operation. Now you
[22:06] can see minus 100 dB so signal is
[22:08] completely gone. If we reattach
[22:15] Now you can see the signal is back
[22:17] because the currently is built on the
[22:19] Lego block. It shows a little higher
[22:21] signal loss than before but that is a
[22:23] before of a bit of this
[22:25] deflection of the base plate. But we we
[22:29] did this kind of the make and break
[22:31] cycles over 40
[22:34] cycles then we observed average 2.2 dB
[22:37] signal loss. But also it should be
[22:41] noted that the signal loss comes from
[22:44] the peak design not by the our lens
[22:46] array.
[22:47] >> At the Optica Global Photonics Economic
[22:49] Forum this year, Jin Hasegawa one of our
[22:51] keynote speakers is coming to talk about
[22:53] detachable connectors, the road map,
[22:55] Senko's road map on detachable
[22:56] connectors. Why should they use your
[22:58] material?
[22:59] >> Yeah, so This detachable operation
[23:01] provide the value to customer. For
[23:03] example, higher manufacturing yield also
[23:06] serviceability at the field. Also, it
[23:09] offers kind of a multi-generation
[23:11] hardware upgrade at the data center
[23:13] operation. So, this are the micro lens
[23:17] array fabricated by XT lens provide this
[23:20] kind of detachable operation and it is a
[23:23] huge value to customers.
[23:25] >> Detachable connectors are the only way
[23:27] to make co-packaged optics at volume.
[23:30] This is the right material for that.
[23:31] >> Exactly. So, this material provided that
[23:33] high scalable mass production with a
[23:36] design freedom. You can make the high
[23:39] freeform lenses with all different
[23:42] aspects in a single molded part.
[23:44] >> Gabry, thank you so much again for
[23:46] constantly supporting OFC. What did you
[23:48] learn there?
[23:50] >> Yeah, we also got feedback from
[23:51] customers that they don't like epoxy for
[23:54] optical adhesives that much. It works
[23:56] for the packaging, but they have
[23:57] concerns on the long-term stability in
[24:00] terms of yellowing and embrittlement.
[24:02] So, what we did in this demonstrator, we
[24:05] actually use a key advantage of ultimate
[24:07] XT that it's very easy to metallize.
[24:10] So, what we did, we metallized the pads
[24:13] on the side of this MLA.
[24:15] We just put them on the printed circuit
[24:17] board and underneath we put some solder
[24:20] paste. And then during the reflow, it
[24:23] will automatically align with the flexel
[24:25] and detector. It's so-called capillary
[24:27] self-alignment.
[24:29] And and yeah, therefore you don't need
[24:31] to use epoxies. You see here on the
[24:33] demonstrator that we coupled two MLAs
[24:37] uh sticked on the on the PCB by the
[24:39] solder paste and the signal strength is
[24:42] very strong.
[24:42] >> Gabry, I can tell you one thing, you're
[24:44] very politically correct. The photonics
[24:45] industry doesn't like or saving hate a
[24:48] little bit epoxy. Is there really a
[24:50] possibility that we can get rid of epoxy
[24:52] for most of the applications?
[24:53] >> Oh, yeah, sure. I mean, if you can stick
[24:55] XT
[24:56] XT and ultimate on the board without an
[24:58] epoxy, then and it is reliable over the
[25:00] time, then I think it's a great
[25:01] solution.
[25:02] >> We are in the verge of making a
[25:04] revolution. Gabi, I could stay here for
[25:06] days and having so much fun, but I need
[25:08] to go to another event. Could you tell
[25:09] me in short words why people should care
[25:11] about extem and ultem?
[25:13] >> Yeah, what well what we like customers
[25:15] to do is like you, come to our
[25:18] optoelectronics center of excellence and
[25:20] we will learn customers that material
[25:24] and design go hand in hand to develop
[25:26] the optical interconnect they are
[25:27] looking for. We will give them the same
[25:29] lab tour as what we gave to you, Jose.
[25:32] And and and that's what we want, right?
[25:34] We don't want only to be a resin
[25:36] supplier, we want to be a total solution
[25:38] provider and that's actually what we at
[25:41] Sabic Specialties care for.
[25:43] >> The reflow soldering compatibility can
[25:45] actually revolutionize CPO
[25:46] manufacturing. We are working very hard
[25:48] at Optica to make sure we have a
[25:50] resilient supply chain for the future of
[25:52] CPO and definitely Sabic is part of
[25:55] this.
[25:56] >> [music]