How Is A Flowcell Made? How Much Does It Cost?
Last time I looked at a HiSeq X flowcell and tried to document its construction. This approach seems to be used on most of Illumina’s recent sequencers (HiSeq, MiniSeq and NovaSeq).
One question I have is what plastic is used to form the channel here. So I tried doing some tests… the material seems to have roughly the same density as water (doesn’t float or sink). It self-extinguishes, and burns with an orange(ish) flame. It seems to get tacky around 300 to 350 degrees C.
Various Illumina patents talk about Black Kapton (Polyimide). But this doesn’t quite add up… I decided to take some of the film from a NovaSeq flowcell as see if I could create a channel with it.
I sandwiched it between two pieces of cover glass and stuck it on a hot plate. With some poking this formed a reasonably stable channel:
Elsewhere in the patents Illumina mention Kapton KJ in its regular and “black” (carbon infused) formulation. This makes slightly more sense. Kapton KJ is a heat sealable and can be bonded to “nearly any type of material”:
Regular Polyimide would probably cost <<$1 per flowcell (HiSeq). Kapton KJ appears to be discontinued the current heat sealable Kapton film being Kapton FN, this would also be significantly less than <$1 per flowcell.
In order to determine the type of glass used I threw the 300um glass in a UV/Vis spectrometer. Below the spectra for cover glass, the HiSeq X patterned substrate, and the NovaSeq patterned (300um) substrate:
For the NovaSeq it was possible to scratch off the patterning (indicating that it is build on top of the substrate). On the HiSeq X it was not possible to remove this. This perhaps explains the odd looking spectra.
Based on the above it looks to me like they use fused silica on these instruments. An 8 inch fused silica wafer costs ~$100 and can provide ~10 HiSeq X sized substrates. Assuming this doesn’t vary too much based on thickness this is ~$20 per flowcell.
Overall based on the COGS it seems unlikely that flowcell materials cost more than a few $10s.
I may try and slap some flowcells together. So subscribe if more information on flowcell construction and costs might be of interest!