Printable dielectrics for RF, Bonding in heterogeneous integration
Welcome to this week edition of our technology newsletter with a focus on additive electronics.
One housekeeping note: the masterclass and tour spaces taking place before our inaugural conference and exhibition in Eindhoven (Future of Electronics RESHAPED | 12-13 OCT 2022) are almost full. We are now in tocuh with venue to see if additional space can be released. We recommend that your reserve your spot ASAP.
Topics for this week: EHDjet printed QDs on microLEDs | R2R single-step gravure printed perovskite photovoltaics | Gravure printed microbumps for microLEDs | 3600ppi full color "silicon" displays | Lasers in microLEDs | Printable dielectrics for RF and MW devices | Pitch scaling and bonding in heterogeneous integration | Microfluidics and Electrohydrodynamic printing | Graphene to market: breaking down regulatory barriers.
High-PPI RGB microLEDs, printed electronics, and quantum dots?
The three themes are closely linked since QDs can be digitally printed as color conversation materials atop blue microLEDs to enable wide color gamut RGB uLED displays without requiring a separate transfer step for each color. Join TechBlick's event on microLEDs to learn more www.TechBlick.com/microLED
Inkjet is the common technology investigated for such a purpose. As shown below by Prof.Armin Wedel, however, its 4pL droplet is too large, allowing at best a 40um pixel and not able to reach even 850 dpi.
Electrohydrodynamic printing (EHD) can however address this issue. In EHD, the droplets are pulled out by an electric field from a nozzle which sits close (50um or so) to the surface and thus requires a good printing facility.
As shown below, the droplet volume is only 0.5pL, enabling 1-10um pixels in the lab and 15um reproducibly. This will enable one to achieve 850ppi and 1000ppi!
Slide 2 shows an example of a QD color filter (QD-CF) for a microLED display deposited using EHDJet. Here, 15um pitch is reported, achieving 1000ppi. The roadmap will be to evolve the technology towards even 2000ppi!
These are excellent advacements of the art and technology, paving the way for the development of high-PPI microLED technology.
Of course, EHDJet is a relatively new technology. It is mainly single head and slow, although multi-head print heads are emerging. Nonetheless, it is an elegant solution for depositing color filters on high-PPI microLED displays.
To learn the latest about these technologies joint TechBlick's specialist event on microLEDs and Quantum Dots where Prof. Wedel will also present: www.TechBlick.com/microLED
You can hear from the likes of Samsung, Sharp, Yole, ASMP, Coherent, Nanosys, CEA, AUO, Allos Semiconductor, KIMM, Luxnour, Omdia, Playnitride, micromac, and many many more
R2R gravure print perovskite photovoltaics in a single step without antisolvents?
This would be a major step towards industrialization. Here, we discuss the transition from 2-step printing to one step printing with antisolvent to one-step printing with no antisolvents. Riikka Suhonen et al discussed the latest developments at TechBlick's event in Dec 2022. Here is a summary
2-step approach: In general, most approaches are based on a 2-step printing in which the lead iodide (from PbI2-DMSO ink) is first gravure printed on a printed SnO2 NPs layer and then dried. The DSMO is then washed away in a water and isopropanol path and the remaining porous layer is dipped into a chemical second path to form MAPbl3. The 'pilot' R2R runs yield PCE of 9.7%. This approach requires two chemical steps, slowing the process. Furthermore, handling the porous Pbl3 layer is difficult in R2R environment and oncersion to FA- or FACs-perovskite challenging
1 step printing + antisolvet: the standard antisolvent is ether but this can not be printed due to high volatility. Therefore, there has been huge effort in developing an antisolvent which could be printed industrially and was environmentally friendly. VTT et al developed the tBuOH:EA system. This way they achieved fully R2R gravure printed perovskites with efficiency of 13.8%. This is an elegant solution. Nonetheless, There is a desire to eliminate the antisolvent step as it will require a spray or bath step together with solvent fumes.
1 step printing: Here they used starch as a rheology modifier with perovskite precursor, forming a viscous ink which can be used to print well-defined patterns. In this R2R 'lab' printing starch-based MAPb3I inks they used IR annealing and hot air annealing. The first lab runs showcased a champion result of PCE 9.9%. This is still an early stage development but shows that reasonble efficiencies can be achieved using R2R gravure printing with only a single step!
Of course - these results are early stage. Lifetime remains an issue and development area. Nonetheless, this is an important field to watch further.
We will soon announce the agenda for our 2022 edition of TechBlick's event on organic, perovskite, and tandem photovoltaics.
Gravure printed microbumps for microLEDs
As microLEDs inevitably shrink in size, the micro-bumping requirements for the microLED dies becomes more challenging. Direct wafer-based printing based on gravure offset techniques offers a promising solution in this regard. Indeed, this is another field where printed electronics can play a role.
Komori has recently achieved excellent results, which will be unveiled at TechBlick's upcoming microLED event on 30 Nov-1 Dec 2022: www.TechBlick.com/microLED
As seen in the slides below, gravure printing can print microbumps printed using flux paste, achieving a printing precision of 5 µm within a range of 300 mm. The first slides show the precision of the printing position on a wafer. In particular, it compares it with screen printing, showing how gravure printing advances the fine feature printing capability w.r.t screen printing (+/-10 um although screen printing too can and will also advance)
As shown in slide two, the minimum diameter that can be printed with SAC (Sn, Ag, Cu) solder paste is 6 μm and the distance between the centers of the bumps is 30 μm. Reflow has been successful with a minimum diameter of 10 µm. This way for example, a microLED die in the size of 30um by 50 or 80um can be supported.
Furthermore, as shown in slide three, this technique also offers the possibility to control the thickness by printing several diameters. The smaller the bump diameter, the higher the aspect ratio.
This are very nice results, showing the viability of gravure printing technique for microbumps. This technology can support current and near-term generations of microLEDs but will it evolve as microLED dies further shrink in the longer term?
Join Komori and other members of the community to learn about all aspects of microLEDs from GaN microLED technology to transfer and tiling technology to bumping and color conversion technology and beyond. You can hear from the likes of Samsung, Sharp, Yole, ASMP, Coherent, Nanosys, CEA, AUO, Allos Semiconductor, KIMM, Luxnour, Omdia, Playnitride, micromac, and many many more
“Silicon” Displays with an incredible 3600ppi full color using microLED and QD technology?
Sharp (HIRANO Yasuakie et al) will join us from Japan to explain this technology at the upcoming TechBlick event on microLEDs and quantum dots (www.TechBlick.com.microLEDs).
As shown in the slide below, first blue-only uLEDs are formed on a sapphire substrate. Here, one LED array contains 352 x 198 micro LED dies of 24 um x 8 um in size. In parallel, an LSI chip containing the driving circuitry is formed on a silicon wafer. Here, the cathode (N-type electrode) and anode (P-type electrode) are fabricated for each micro LED die to apply driving voltage independently to each die. The Au bump electrodes are fabricated in accordance with the pitch of the LED dies. The two substrates are flip-chip bonded using Au-Au bonding. Here one can already see the parallel to the silicon and optoelectronic industry (vs. the traditional thin film display industry!). Next, the sapphire layer is removed via laser lift-off. Finally, Cd-free quantum dots (green and red) are deposited atop the microLED dies to enable R G color conversion. This way one achieves RGB colors.
The device architecture is shown in slide 2- here one can see the location of GaN uLED dies, Au bumps, as well as light shielding walls and quantum dots (QDs). This way, a full color 1,053 ppi display is formed.
However, given the small size of the emissive area of uLEDs, the brightness is low. An innovative solution here is to switch from individual driving cathode electrodes to a common one, thus freeing up more spaces for uLEDs. As shown in slide three, the light emission in one pixel was improved from 23% to 38%. As a result, brightness of 11 knits was achieved. This is an excellent progress. Of course, it is not the final game as even at 11 knits the brightness is not yet not sufficient for outdoor AR applications.
Join us and your industry peers on 30 NOV – 1 DEC 2022 at our first-ever specialist microLED and QD event to hear more about this technology from Yasuakie-san et al: www.TechBlick.com/microLED
How lasers help in MicroLED display production?
See slides below to learn. One of the biggest manufacturing challenges in uLED display production is the transfer step given the speed and yield requirements. As shown in the slides below by Oliver Haupt from Coherent Inc., lasers can play an important role in this step, both when all three colors (R G B) microLEDs and also when only blue microLEDs need to be transferred.
To learn more join TechBlick's first ever specialist event on microLEDs on 30 NOV - 1 Dec where Oliver will present this technology www.TechBlick.com/microLED
The process flow for both cases is shown below. In case of RGB MicroLEDs, first a temporary carrier is attached to the sapphire substrate on which GaN uLEDs are grown. Laser Lift Off (LLO) is deployed to de-bond the sapphire substrate, releasing the carrier wafer with the detached GaN microLEDs. Next, controlled UV spots are used to release the individual microLEDs onto the final substrate holding the TFT active backplane layers. These process can be repeated three times, each time for a different uLED color. In all steps, of course, excellent and optimized control of the laser profile/parameters in harmony with the right adhesive material properties are required.
In the case of blue-only microLEDs, the final backplane substrate is brought into contact with the GaN sapphire substrate. The GaN uLEDs transfer to the final substrate via the LLO process. Three color capability is then achieved by color conversation, e.g., QDs or small-sized phosphors.
The results show the example of microLED RGVB transfer. The parameters are shown in the slide including microLED size, pitch, laser energy density, donor-receiver distance, etc. It can be shown that a different color is transferred with each shot. Thus, in three shots all R G B microLEDs are placed at the right spot! As the subset in slide 2 shows, the laser can in each step/shot process an area of roughly 2.83cm2.
To learn more join our world-class event on microLEDs and QDs. More info on www.techblick.com/microLEDs
Heterogenous integration is the key to the future of computing
Here, the limiting factor is often the interconnect density (pitch) as well as bandwidth and energy consumption of the I/O.
Indeed, as we move towards platforms where multiple dies, potentially from different foundries, are all integrated into the same package, the issue becomes extremely important because die-to-die communication becomes the bottleneck.
The first slide below are from Intel, presented by Dr Sabi at a conference in Sept 2021 online. Here, you can see the intended evolution of the technology. First, EMIB was launched. Here, a silicon bridge with <55um pitches serves as a small (2x2mm to 8x8mm) communication link between two separate dies in a package. This approach is an alternative to the standard silicon interposer technology.
Next was the development of Foveros platform, allowing face-to-face integration of dies from different foundries onto a single package all connected via silicon base logic die. As you can see, this technology will evolve with aggressive pitch scaling and a potential transition to direct Cu-to-Cu bonding from microbumps.
The second slide shows this trend further. It shows the evolution of interconnect pitch as heterogeneous integration advances. The common technology is flip chip BGA (FCBGA). The pitch here is limited to around >120um. Next, EMIB was launched. Here, the pitch was reduced to 55um thanks to the silicon bridge technology. Now there is Fovereos which is based on die on wafer technology. The next generation will based on HBI or hybrid bonding.
These heterogenous integration platforms can enable the integration of dies from different foundries. The challenge is though that each foundry has its own I/O designs, making easy compatibility difficult. It is helpful for the industry to develop common standards to enable a plug-and-play solution.
The third slide shows the need to transition towards Cu-Cu bonding. As the left chart shows, solder based microbumps can support the technology until around 15-20um. Beyond this pitch level, a transition to Cu-Cu bonding becomes necessary. With this transition comes the possibility to increase bump density to over 10000/mm2. This is vital so that the I/O size and bandwidth do not limit overall system performance in complex multi-die packages.
But will all suffice? The fourth slide shows need to transition from Cu bonding to optical I/O technology. As the table shows, optical I/O (OIO) can increase shoreline density by a factor of 4, reaching 1.6 Tb/s/mm. It will also improve power efficiency by some 35%.
Finally, as shown in the fifth and final slide, this technology will need to evolve. The current (Sept 2021) demonstration was for an on-package OIO able to achieve >1Tpbs/mm @ 6 pJ/bit. The target is a fully integrated OIO able to achieve 10 Tbps/mm @ just 1pJ/bit conversation.
Heterogeneous integration is THE technology space to watch babak sabi
Challenge: printable dielectric materials for RF and MW devices?
The dielectric material is often the bottleneck against fully printed high-performance RF and MW devices. This is often a neglected challenge as the emphasis is mainly on the conductive layer. Indeed, the development of a suitable low-loss digitally-printable dielectric material with high and controlld resolution is a technical challenge. In this 3-min video, Yuri Piro from University of Massachusetts Lowell explains why this is challenging:
"So coming up with a non-polar material that you can form on the spot with low processing conditions and low polarity is difficult and you really can't use these conventional approaches."
Microfluidics and Electrohydrodynamic printing (EHD)?
EHD is a promising digital printing technology for going beyond the resolution limits of inkjet. Most examples showcase electronic or display related applications.
However, in a recent TechBlick talk, as shown in slide 1, Dr Aart-Jan Hoeven showed an example in microfluidics where EHD could delvier value. Here, this technology could enable the electrode widths or pitches to be narrowed from 30-40um (possible with industrial inkjet) to perhaps 1-5um using EHD, thus saving space. This will support the miniaturization trend of microfluidics, making possible to even integrate them into the human body.
In slide 2 DoMicro BV 's laboratory-scale nano printer can be seen in more detail. It is able to deposit ultrafine features digitally! This DM50-ENP printer is generating significant interest and was developed as part of E-Nanoprint-Pro project