Alper Kaan Soydan | Technische Universität Ilmenau: Why does scaling down solder bumps to the microscale drastically reduce assembly yield, and how does time play a role?

How does fluidic self-assembly compare with mechanical and electrophoretic transfer on flexibility and substrate independence?

Mechanical placement forces a trade-off: change the chiplet size and you must change the process parameters and the whole accuracy budget. Fluidic methods remove that constraint, so the substrate is independent of the process mechanism. Pitch size or design can change and the approach still works, which is where the flexibility benefit of fluidic placement comes from.

Electrophoretic methods are very good with miniaturization today, with the field now talking about nanorods and the microLED-to-nanoLED space. The catch is active voltage routing, which you may not want in your backplane inside a liquid environment. Orientation control of the LED is also challenging in this method. Both constraints are intrinsic to how the capture is driven electrically.

ELUX is the commercial example of fluidic surface assembly. Like electrophoretic approaches it relies on physical capture: wells are defined in the substrate, and chiplets moving in the fluidic environment fit into them and are trapped. ELUX adds pillars on top of the microLED, so the parts tumble across the flat surface instead of getting stuck, which delivers orientation control.

In this short video, you can learn:
* Mechanical placement ties process parameters and accuracy to chiplet size, while fluidic methods keep working when pitch or design changes.
* Electrophoretic assembly miniaturizes well toward nanorods and nanoLEDs, but needs active voltage routing in a liquid backplane and struggles with LED orientation.
* ELUX uses substrate-defined wells for physical capture and adds pillars on the microLED so chiplets tumble rather than stick, giving orientation control.

📋 **Clip Abstract** Fluidic self-assembly is compared against mechanical placement and electrophoretic methods on flexibility and substrate independence. The clip shows why fluidic approaches tolerate pitch and design changes, what limits electrophoresis, and how ELUX's well-and-pillar capture delivers orientation control.

About the speaker:
* Speaker: Alper Kaan Soydan
* Company: Technische Universität Ilmenau
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

How do 10-micron fluidic self-assembled microLEDs achieve half-micron alignment, >12,000 yield/throughput, and post-assembly contacts?

Two-micron-thick microLEDs were assembled into 10-micron square receptors, reaching more than 12,000 yield or throughput. The demonstrated microdisplay area was 0.85 centimeter by 2 centimeter, containing 12,600 assembled devices with 29 missing receptors in that example. The array used a relaxed pitch. Achieved alignment accuracy was half a micron, with around two-micron angular alignment. The process used a period PSS.

After full self-assembly, top contacts were integrated by PCVD silicon dioxide passivation, followed by opening windows to access the indium tin oxide top contact. Conductive polymers were then applied by spin coating. This is a relatively easy process, but the conductive polymer brings reliability challenges. The team produced an operational microdisplay unit with some level of addressability. These steps define the post-assembly contact integration for the fluidically assembled array.

The demonstration array was 0.85 centimeter by 2 centimeter, with 12,600 devices and 29 missing receptors, indicating a relaxed pitch. Alignment was half a micron, angular alignment around two microns. Electrical and optical performance were verified, including a green peak. The assembly relied on a period PSS, called unlikely for industry use. The result is an operational microdisplay with addressability, but conductive-polymer reliability remains a key integration concern.

In this short video, you can learn:
* Two-micron-thick microLEDs were fluidically self-assembled into 10-micron square receptors with more than 12,000 yield/throughput.
* The demonstration achieved half-micron alignment accuracy and around two-micron angular alignment across a 0.85 centimeter by 2 centimeter array with 12,600 devices.
* Top contacts were integrated by PCVD silicon dioxide passivation, indium tin oxide window opening, and spin-coated conductive polymers, yielding an operational microdisplay with addressability but reliability challenges.

📋 **Clip Abstract** Fluidic self-assembly placed two-micron-thick microLEDs into 10-micron square receptors, reaching more than 12,000 yield/throughput and half-micron alignment accuracy with around two-micron angular alignment. The 0.85 centimeter by 2 centimeter demonstration contained 12,600 devices with 29 missing receptors, and post-assembly top contacts used PCVD silicon dioxide passivation, indium tin oxide access, and spin-coated conductive polymers, giving an operational microdisplay with addressability and reliability concerns.

About the speaker:
* Speaker: Alper Kaan Soydan
* Company: Technische Universität Ilmenau
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

How did researchers overcome the oxidation barrier of 10-micron solder bumps to achieve a 99.77% fluidic self-assembly yield?

The process medium is the first control point in solder-driven fluidic self-assembly. Atmospheric oxygen and liquid dissolved oxygen must both be checked. Atomic absorption spectroscopy measures element concentration in liquid taken from the surface-assembly procedure. Indium is the most active element in low-melting-point solders, which are mostly lead-free. Removing atmospheric oxygen changes the detected concentration at the critical 10-micron solder-bump scale.

With atmospheric oxygen removed, 10-micron solder bumps show a fivefold reduction in detected element concentration. In the fully deoxygenated case, the reduction reaches 20 times. That matches the 20-times reduction in surface-to-volume ratio relative to 200-micron solder bumps, where 100% yield was achieved. The 20 number is treated as a key idea: smaller bumps expose far more surface, so oxygen control becomes decisive.

Full self-assembly requires all conditions to be correct, otherwise it will not work. Two further criteria matter: reduction from the PA side and chiplet mobility, meaning chiplets must be dispersed in the environment. Accounting for these factors in the fully deoxygenated medium produced 99.77% yield. This historic barrier was overcome in the array.

In this short video, you can learn:
* Atomic absorption spectroscopy measured element concentration in liquid from the surface-assembly procedure to track oxygen effects.
* Removing atmospheric oxygen reduced detected element concentration fivefold at 10-micron solder bumps and twentyfold in fully deoxygenated medium.
* A fully deoxygenated medium, combined with PA-side reduction and chiplet mobility, yielded 99.77% self-assembly yield.

📋 **Clip Abstract** Solder-driven fluidic self-assembly at the 10-micron scale is limited by atmospheric and dissolved oxygen, which attack active indium in low-melting-point, mostly lead-free solders. Atomic absorption spectroscopy quantified the effect: fivefold lower detected concentration with atmospheric oxygen removed and twentyfold lower in fully deoxygenated medium, matching the twentyfold surface-to-volume ratio change versus 200-micron bumps and enabling 99.77% yield when PA-side reduction and chiplet mobility are also controlled.

About the speaker:
* Speaker: Alper Kaan Soydan
* Company: Technische Universität Ilmenau
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven