Isfet

An ISFET is an ordinary MOSFET with its gate metal stripped away and replaced by an ion-sensitive membrane sitting directly in a fluid, so the transistor's threshold voltage shifts with the chemistry of whatever liquid touches that membrane instead of with a voltage applied by a wire. Where a standard MOSFET's gate electrode sets the channel's conduction state through a fixed metal contact, an ISFET floats that gate electrically and lets a reference electrode in the sample fluid do the job instead, so protons or other ions binding to the membrane's surface change the effective gate charge and, with it, the current flowing between source and drain — a direct, label-free translation of chemistry into an electrical signal that a standard CMOS process can fabricate. Piet Bergveld described the structure in 1970, but it took three more decades of CMOS process maturity before a version of it became something a foundry could actually mass-produce reliably at a competitive cost.

Ion Torrent turned that threshold-voltage shift into the basis of an entire DNA sequencing platform, and that is the single clearest commercial proof that ISFETs can be fabricated at the scale a real chip business needs. Jonathan Rothberg's team built sequencing chips that place millions of individual ISFET wells over a CMOS die, each well holding a single DNA strand; as a DNA polymerase incorporates a nucleotide, it releases a hydrogen ion, that ion shifts the local pH by a fraction of a unit, and the ISFET directly under that well reports the shift as a tiny but readable voltage change. Thermo Fisher's Ion Proton and Ion S5 instruments run chips with up to several million of these sensor wells, each one acting as an independent ISFET, fabricated with the same lithography and ion implantation steps a logic fab runs for any other CMOS product, just with a post-processing step that etches away the metal and deposits a sensing membrane in its place. The sequencing-by-synthesis result is a chemical reaction read out electronically rather than optically, which is the reason Ion Torrent sequencers skip the fluorescence imaging and optics that competing sequencing platforms depend on.

Drift and trapped charge are the two problems that keep ISFETs from behaving like textbook MOSFETs, and both trace back to the same root cause: an ion-sensitive membrane is a much less stable interface than a clean metal gate. A fresh ISFET's threshold voltage creeps for minutes to hours after first contact with fluid as the membrane hydrates and ions slowly redistribute across it, an effect largely absent in a conventional MOSFET whose gate dielectric never has to equilibrate with a liquid; most practical ISFET designs wait out an initial settling period or apply a calibration curve in firmware rather than trying to eliminate the drift at the device level. Trapped charge builds up in the membrane and gate oxide under sustained bias and ionic exposure, nudging the threshold voltage further over a sensor's working life, which is why foundries building ISFET lines for diagnostics customers specify tighter oxide quality and lower trap-density targets than a standard logic process needs, and why most commercial parts are rated for a fixed number of measurement cycles rather than being treated as a permanently reusable sensor.

Treat an ISFET as a CMOS logic transistor wearing a chemistry-reading membrane rather than as some separate class of exotic sensor, because that framing is exactly why Ion Torrent, blood-gas analyzers, and a growing set of point-of-care pH and glucose chips could all be built on standard foundry lines instead of specialized MEMS or discrete-sensor processes.