Who am I?

Hi! My name is Dmitry Janushkevich, and I run a small business aiming to provide semiconductor-related services. If you are in need of:

  • Counterfeit detection,
  • Product intelligence,
  • Input data sets for the purposes of analyzing, evaluating or teaching the concepts, processes, systems or techniques, or
  • Beautiful semiconductor images,

… then I can help you get your project done within budget and a short turnaround time.

What can I do for you?

Today, I am glad to offer a range of services to the general public.

Semiconductor imaging

This constitutes the core service offering of the lab. I can do the following for you:

  • Part decapsulation: Taking the die out of the packaging is the first step to seeing anything at all.
  • Die imaging: Using optical microscopy, I can take panorama images of effectively unlimited sizes down to the 1000x magnification.
  • Die deprocessing: Often times, features of interest are obscured by metal layers above and thus have to be removed.

As a result, you will get (a set of) images of the die(s) inside. Other options can also be discussed.

Previous results

Quite a few of my previous works are available publicly at the SiliconPrawn archive. Here is a selection of them, linking to browsable panoramas:

Capabilities in detail

Below you can find more technical information regarding the lab’s capabilities on offer.

Decapsulation

Decapping eye candy

There are several options available for extracting the die out of the package.

PLEASE BE WARNED: Decapsulation might result in damage/destruction of samples while yielding no usable result despite my best efforts; thus, providing multiple samples is highly desirable.

  • Ceramic lidded packages are universally successful just by desoldering the lid. The chip will likely stay functional.
  • Ceramic (non)windowed packages almost always succeed unless the bottom part fractures from stress and takes the die with it. The chip will not be functioning due to frit fracturing.
  • Epoxy DIP packages likely to succeed unless there is something funky about the die e.g. it’s a multi-chip package, dies are very fragile, etc. The chip will not be functioning.
  • Epoxy QFP, DFN, and simple BGA packages almost always succeed as there is not much material to stress the die. The chip will not be functioning.
  • Other package variants can be discussed individually.

At this time, there is no option on offer which would allow for the chip to stay functional after decapsulation except as noted above.

Optical imaging

Imaging eye candy

The imaging setup in daily use consists of a fully motorized AmScope ME580-PZ-T with the following set of optics:

  • LMplan 5x/0.13
  • LMplan 10x/0.25
  • Nikon M Plan 20x/0.4 ELWD
  • Nikon Plan Apo 20x/0.75
  • Nikon Plan Apo 40x/1.0, oil immersion
  • Nikon Plan Apo 60x/1.40, oil immersion
  • Nikon Plan Apo 100x/1.35 NCG, oil immersion
  • … and a selection of other, more experimental objectives.

This setup typically allows for good results with process nodes down to approximately 350nm. Work is always underway to improve the setup and push the resolution to the physical limits of optics on hand.

Image processing

Imaging eye candy

Imaging itself is done using a Raspberry Pi HQ camera; the output is then processed further using proprietary software developed in-house. The software stack makes it possible to acquire a single-depth image as well as perform focus stacking to capture visible features at any depth. The software then stitches individual tiles to produce a panoramic image of the whole die or a particular region of interest, with output options of a single image e.g. a JPEG file and a web-based deep-zoom image dataset for easy inspection. Other output options may be made available upon request.

Deprocessing

Deprocessing eye candy

Sample deprocessing is performed manually using a polishing rig and selection of abrasives and polishing pads. Some degree of corner and edge rounding is normal and to be expected; in most scenarios, this does not impact the analysis. Deprocessing via lapping typically involves several process-and-image iterations, the amount of which depends heavily on the sample in question. For older process nodes with one metal layer, deprocessing using hydrofluoric acid is a preferred option. Due to considerable amount of work involved, each deprocessing image will be quoted individually.

PLEASE BE WARNED: Deprocessing might result in damage/destruction of samples while yielding no usable result despite my best efforts; thus, providing multiple samples is highly desirable.

Multi-focus panoramas

Using focus-stacked data sets, it is possible to construct multi-focus panoramas: a set of stitched images with individual layers being in focus, instead of all of them at the same time as would normally be the case using the focus stacking approach. This may be of particular interest when handling deprocessed samples. However, there is a considerable workload involved with marking raw data up; tools are being developed to automate the process but we are not there yet.

Something not on the list above?

I am always interested to hear about your particular requirements. Let’s discuss what options are available to get you the best results!

Get in touch

Should you find yourself in need of any of my services, please do not hesitate to get in touch via the form below or directly via email: infosecdj AT airmail DOT net. I am always happy to discuss any ways I can be of service for your project.

Impressum

Dmitry Janushkevich, registered business in Denmark under CVR 12345678, VAT ID: DK 12345678.