Sander is an Applied-physics graduate from TU Delft. Worked at a semiconductor start-up during his studies, decided building a company would be the most interesting way to shape his career. In 2010, a TU Delft professor suggested combining light and electron microscopy. Sander and co-founder Andries Effting took the patent out of the university and founded Delmic.
Core conviction: almost every process that matters plays out at the nanoscale, yet the instruments that look at that scale are still operated by hand. “A Ferrari permanently stuck in traffic.” Delmic’s purpose is to remove that traffic.
Tell us about your company.
Delmic is a Delft-based deep-tech company building automated workflows for electron microscopy. We were founded in 2011 as a spin-out of TU Delft and AMOLF, and today more than 220 research institutes and companies in over 30 countries use our technology. We are headquartered in Delft, with presence in Japan and the United States.
For the semiconductor world, our relevance is cathodoluminescence (CL). Our SPARC platform is the leading tool for characterising compound semiconductors and related materials, including GaN, SiC, µLED and quantum emitters, at the nanoscale, with 100+ systems installed globally. Our newer FLASH platform extends that into industrial R&D. The common thread across everything we build is simple. Sample in, data out.
What problems are you solving?
Our view is that the microscope itself is no longer the limit, the manual work around it is. For many applications, the problem of resolution is solved, but those applications tend to be the ones where human-intensive operation is the bottleneck. Sample loading, acquisition and choosing where to measure still depend on expert operators.
In compound semiconductors specifically, yield-limiting defects are often invisible to optical inspection, either because they are too deep or too small. As a result, today’s feedback loop is either incomplete (when using optical inspection) or very slow (when using electron microscopy). In a field where iteration speed directly impacts production yield, that lack of information is expensive. We automate the measurement so that data comes back in hours, not weeks, and so researchers can base conclusions on robust data volumes rather than a handful of manual measurements.
What application areas are your strongest?
Our deepest strength is cathodoluminescence for compound semiconductor and optoelectronic materials. More specifically, this includes power electronics such as GaN-on-SiC epiwafers, yield qualification for µLEDs and displays, and the characterisation of quantum emitters and advanced photonic materials such as diamond and hexagonal boron nitride. In these applications, techniques such as carrier-lifetime mapping and emission-uniformity measurements at sub-10 nm resolution go beyond what conventional optical techniques can achieve. This expertise is built on a broad academic foundation. SPARC is also widely used in nanophotonics, quantum optics, geology and mineralogy. Our installed base of more than 100 SPARC systems has given us extensive application knowledge across these fields.
Alongside the materials business, we run an equally strong life-science portfolio in cryo-electron tomography and large-scale (volume) electron microscopy. Useful context because it’s the same automation platform and the same software core applied to a second market.
What keeps your customers up at night?
For our semiconductor R&D customers, it is yield, and specifically the defects they can’t see. Nanoscale defects that optical inspection misses translate directly into yield loss and higher
costs, while the feedback loop needed to identify and correct these issues is often too slow. When it takes days or weeks to learn whether a new epitaxial recipe worked, every iteration becomes expensive. The pressure increases further when competitors can iterate faster, especially as GaN and SiC wafers scale to 200 mm and 300 mm and new compound-semiconductor fabs come online.
Underneath that is a broader data-confidence problem that we hear across every segment. Researchers often draw conclusions from very limited measurements simply because measuring takes too long. They lie awake wondering whether what they’re seeing is the rule or just a coincidence.
What does the competitive landscape look like, and how do you differentiate?
We are unique in the level of integration we do on commercial SEM platforms, and the resulting control and automation we offer in this field.
In CL there are two kinds of competitors. Providers that sell capable but hardware-agnostic add-on detectors, but where the workflow still depends entirely on the operator’s expertise with both the SEM and the detector, a little to no workflow automation is offered. Other providers pursue the same integrated CL-SEM vision we do, but offer their own electron column that is widely regarded as underperforming on electron-optical quality. We deliberately work with best-in-class electron columns while offering the level of control and automation of a fully integrated instrument.
In short, we differentiate in four areas. We offer broad modality coverage, with six CL imaging modes including time-resolved measurements and polarimetry. SPARC works with high-quality commercial SEMs rather than locking users into a proprietary electron column. We have one of the largest installed bases and applications maps in the field. And finally, we are increasingly automating the complete measurement workflow.
What new features / technology are you working on?
The headline at IMC21 is the latest generation of SPARC, called SPARC 3.0 which is a platform solution with improvements across the board. It also comes with automation out of the box, such as whole wafer mapping and automatically selecting the right measurement locations. Combined with our ODEMIS control software, the instrument now drives the beam, stage and detector autonomously and delivers analysis-ready data, turning CL from a slow, expert-gated add-on into something that can run largely by itself.
Beyond that, we are working on the next generations of FLASH, our CL platform purpose-built for compound-semiconductor R&D, and we’re working on an AI-assisted analysis that acts as a validated gateway to the best community and commercial analysis models rather than reinventing them.
How do customers normally engage with your company?
Most relationships start with a technical conversation rather than a sales pitch. The first point of contact is often one of our application specialists, who works with the customer to
understand the specific measurement challenge and, where it makes sense, runs a demonstration or sample trial.
Other important entry points include major conferences such as IMC, Microscopy & Microanalysis, MRS, SEMICON Europe and dedicated compound-semiconductor events. We also receive referrals from electron-microscope OEMs for CL capabilities they do not provide themselves, as well as enquiries through our website and published application notes.
Entry-level detectors typically have a shorter sales cycle and are often sold through OEM or distributor partners. Flagship platforms such as FLASH involve a longer, more consultative process and are sold directly in Europe and North America, with distributor support across Asia-Pacific.
Come find us at booth 381 at IMC21 in Liverpool!
Also Read:
CEO Interview with Phillip Stanley-Marbell of Signaloid
CEO Interview with Stephen Strausser of VON ARDENNE North America, Inc.
CEO Interview with Ajit Prabhu of Quest Global
Share this post via:
Five Billion Pulses Later: What DUV Optics Testing Reveals About Semiconductor Tool Uptime