Gari Arutinov | Holst Centre: Why do contact pick-and-place tools fail the four pillars of advanced packaging, and what does a laser pulse replace them with?
01:13 - 03:04
Other snippets from this talk
Summary of the clip:
Why do contact pick-and-place tools fail the four pillars of advanced packaging, and what does a laser pulse replace them with?
The mid-end layer sits between classical front-end wafer fabrication and packaging, where compute, power, photonics and memory building blocks come together. Its performance depends on how well and precisely those blocks are assembled, at what throughput and yield. The stated problem: how to efficiently assemble these Lego building blocks. Four pillars define the requirement: ultra-high precision, handling increasingly larger, thin and fragile components, maximizing package real estate, and pristine interfaces for efficient electrical and optical connection.
Achieving those four pillars at production throughput with conventional contact-based pick-and-place systems becomes increasingly difficult, if not impossible. The alternative eliminates contact altogether: a laser pulse triggers release of the component from proprietary release stacks, transferring it in a non-contact manner toward the target substrate. Because the process is non-contact, stress is minimised and contamination is reduced for both the components and the receiver, and eventually the product. The release happens within microsecond range.
The microsecond release is what enables support for high-throughput processing, and, as stated, it is intrinsic to the approach. Together, non-contact transfer addresses the pillars where contact tools struggle: precision, fragility, package real estate and interface quality. Photonics, memory, compute and power blocks can be placed without the mechanical touch that limits conventional systems. That is the claim: replace contact with a laser pulse, and keep production throughput.
In this short video, you can learn:
* The mid-end sits between front-end wafer fabrication and packaging, and its performance depends on precision, throughput and yield when assembling compute, power, photonics and memory blocks.
* Four pillars are named: ultra-high precision, handling larger thin fragile components, maximising package real estate, and pristine interfaces for electrical and optical connection.
* A laser pulse releases the component from proprietary release stacks within microseconds, transferring it non-contact to the target substrate.
📋 **Clip Abstract** Conventional contact-based pick-and-place struggles to deliver the four pillars at production throughput, so the approach removes contact entirely. A laser pulse triggers release from proprietary release stacks in the microsecond range, reducing stress and contamination while supporting high-throughput transfer of components toward the target substrate.
About the speaker:
* Speaker: Gari Arutinov
* Company: Holst Centre
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#NonContactTransfer, #ProprietaryReleaseStacks, #MicrosecondRelease, #MidEndAssembly, #AdvancedPackaging, #HeterogeneousIntegration
This is a highlight of the presentation:
High Precision at the Speed of Light: Laser-Induced Forward Transfer for Advanced Chip Integration
More Highlights from the same talk.
04:58 - 06:42
What transfer speed and yield has laser die assembly actually demonstrated for microLEDs at panel scale?
What transfer speed and yield has laser die assembly actually demonstrated for microLEDs at panel scale?
For microLEDs, throughput and yield remain dominant challenges: these plates require millions of tiny dies to be assembled. Here, 10,000 microLEDs were transferred within two seconds on a panel, with transfer yield above 99.9%. Speed has been showcased, but overall assembly throughput depends on motion control and matching donor-to-receiver designs, where pitches are matched. The open question is whether distribution and micro corrections are needed; the transfer yield potential shows it is not a bottleneck within throughput.
Speed overall is not enough; deploying to real use cases is key. The work showcased transfer of 250,000 tiny dies onto a substrate that mimics a smart eyewear application: concave eye lenses. Despite the tiny size of the dies, dimensions are on the slide, transfer yield was maintained above 98%. This moves the assembly challenge from a lab demonstration toward application-specific geometries, where die dimensions and receiver curvature matter.
The next step is industrial deployment. Together with industrial partner ITECH, a showcase pre-industrial system has been built for this application domain. It is capable of achieving micron-level alignment accuracies at very high throughputs. This combination targets real use cases rather than only demonstration speed, linking transfer speed and yield to application-specific eyewear geometries and panel-scale integration needs.
In this short video, you can learn:
* 10,000 microLEDs were transferred in two seconds on a panel while transfer yield stayed above 99.9%.
* 250,000 tiny dies were transferred onto a substrate mimicking smart eyewear concave eye lenses with transfer yield above 98%.
* A pre-industrial system built with ITECH reaches micron-level alignment accuracies at very high throughputs, while overall throughput depends on motion control and matched donor-receiver pitches.
📋 **Clip Abstract** MicroLED panel assembly faces throughput and yield limits, but the demonstrated transfer of 10,000 microLEDs in two seconds at above 99.9% yield shows transfer yield is not the bottleneck. Scaling to smart eyewear-like concave eye lenses, 250,000 dies were transferred above 98% yield, and an ITECH pre-industrial system now targets micron-level alignment at very high throughputs.
About the speaker:
* Speaker: Gari Arutinov
* Company: Holst Centre
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#MicroLEDTransfer, #TransferYield, #ConcaveEyeLenses, #MicronAlignment, #MicroLEDDisplays, #ARDisplayManufacturing
09:41 - 11:54
Can a non-contact process handle fragile 20-µm power dies and enable 3D edge optical interconnects?
Can a non-contact process handle fragile 20-µm power dies and enable 3D edge optical interconnects?
Power electronics components can be millimeters across yet only tenths of a micron thick, making them extremely sensitive to cracks and stresses. Conventional pick-and-place systems fail to handle these increasingly thinner and fragile components without damage. That mechanical fragility became the third application domain and a perfect test bench for a non-contact approach. It demonstrated collective pickup of 20-micron-thick, millimeter-size, three-by-four-millimeter dummy power devices from commodity tapes as fabricated onto release tape carriers.
After collective pickup, dies undergo selective transfer without cracks or deformation, with successful release onto lead frames and mid-frames. Across three application domains, the work addressed key bottlenecks: throughput, precision, and mechanical fragility, topped up by overall cleanness and selectivity. The non-contact method is not limited to transferring dies onto two-dimensional substrates. It opens a path from 2D layouts to 3D architectures by mastering how dies are released and oriented.
Control of release and orientation enables 3D stacking of devices, whether thick or thin, small or large. Components can be brought onto the edge of a chip or substrate. That edge placement is extremely relevant for optical connection between chips, because it is the closest the chips need to go instead of routing through the RDLs. The non-contact approach therefore extends beyond 2D die transfer toward 3D edge integration and optical interconnect.
In this short video, you can learn:
* Conventional pick-and-place cannot handle millimeter-size power devices that are only 20 microns thick without damage.
* Collective pickup from commodity tapes and selective release onto lead frames removed cracks and deformation in the demonstrated dummy power devices.
* Controlling die release and orientation opens 3D stacking and edge placement for optical chip-to-chip connection instead of RDL routing.
📋 **Clip Abstract** Mechanical fragility in power electronics exposes the limits of pick-and-place for millimeter-size, 20-micron-thick dies. A non-contact approach achieves collective pickup, selective transfer without cracks or deformation, and release onto lead frames, then extends to 3D stacking and chip-edge optical interconnects.
About the speaker:
* Speaker: Gari Arutinov
* Company: Holst Centre
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#NonContactPickAndPlace, #CollectiveDieTransfer, #FragilePowerDies, #LeadFrameRelease, #AdvancedPackaging, #HeterogeneousIntegration
06:57 - 09:21
How close is die assembly to the sub-200-nanometer three-sigma accuracy that photonic coupling demands?
How close is die assembly to the sub-200-nanometer three-sigma accuracy that photonic coupling demands?
Integrated photonics performance is extremely sensitive to alignment accuracy. A tiny misalignment of the die directly translates to optical losses, and the interface must be exceptionally clean for efficient optical connection. Release stack carriers can also act as processing carriers. Before transfer, substrates can be cleaned and activated efficiently on a carrier level. Besides transferring dies precisely, the approach prepares interfaces for bonding, so optical connection quality is optimized.
Indium phosphide photonic components have been assembled onto various receiving platforms, supported by an adhesiveless bonding strategy and compatibility with various photonic architectures. A roadmap target was set at sub-200 nanometer three-sigma accuracy, critical for highly efficient optical coupling of devices. Across the last one and a half years, progress moved from one micron in 2025 to 500 nanometer one sigma early this year, then to 350 nanometer three sigma in latest results.
Photonic photo detectors were successfully integrated onto the SOI platform and packaged into a fully functional module. This module demonstrates 200 nanometer precision accuracy supported by the adhesive-less approach. The latest 350 nanometer three-sigma result brings the effort firmly onto the sub-200-nanometer target. The stress bench shows how carrier-level cleaning and activation prepare bonding interfaces while precise die transfer addresses optical loss sensitivity.
In this short video, you can learn:
* Integrated photonics is an ideal stress bench because tiny die misalignment directly becomes optical loss and interfaces must be exceptionally clean.
* Indium phosphide photonic components have been assembled onto various receiving photonic platforms with an adhesiveless bonding strategy.
* Precision progressed from one micron in 2025 to 500 nanometer one sigma early this year, then 350 nanometer three sigma, with photonic photo detectors on SOI packaged into a fully functional module at 200 nanometer accuracy.
📋 **Clip Abstract** Integrated photonics demands exceptionally clean interfaces and alignment so precise that tiny die misalignment does not become optical loss. Using release stack carriers for cleaning and activation before precise die transfer, the work assembled indium phosphide photonic components with adhesiveless bonding onto various photonic platforms, advancing from one micron in 2025 to 350 nanometer three sigma and demonstrating photonic photo detectors on SOI packaged into a fully functional module at 200 nanometer precision.
About the speaker:
* Speaker: Gari Arutinov
* Company: Holst Centre
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#IndiumPhosphide, #AdhesivelessBonding, #Sub200NanometerAccuracy, #SOIPhotodetectors, #IntegratedPhotonics, #OpticalPackaging