Kyoto engineers buried the gate to keep a silicon carbide transistor controllable at 600 C

Start with the threshold voltage, because that is where the design work went. A JFET switches when the gate's field opens or pinches the channel, so where the switching point sits depends on exactly how much dopant sits where ^[2]. In silicon carbide, some dopant atoms travel deeper into the crystal than the process intended, and the Kyoto group reports that this can move a conventional JFET's threshold by more than 2 volts once the device is hot ^[12]. Their fix is geometric: put the gate beneath the channel and dope it heavily, so stray atoms arrive in a region already saturated with dopant and the channel-gate profile barely registers them ^[16]. Mitsuaki Kaneko, the study's first author, blames the field's slow progress on "trying to apply silicon-era thinking to a fundamentally different material" ^[15].

Leakage is the second of the two failures they set out to fix ^[11]. Above 350 C the substrate itself becomes less resistive, so current sneaks around the channel while the transistor is nominally off, producing wrong signals and unbudgeted power draw ^[13]. The answer was two doped wells around the device, whose boundaries block that bypass path even as the substrate turns conductive ^[17].

Worth putting the temperatures side by side. 600 C is 100 C above the 500 C at which the best previous JFETs could run long term ^[14][1], and 250 C above where the leakage problem starts ^[3]. Against Venus it leaves 140 C of margin over a surface that reaches 460 C ^[10][2]. But the limit the authors name for past missions is time, not peak heat: silicon electronics held landers to a few hours ^[8], and Venera 13's record on the surface is 2 hours 7 minutes, or 127 minutes ^[9][4]. What was measured here is how cleanly the device switches and how closely its threshold matched the value predicted from channel thickness and doping, across a range beginning at room temperature ^[18]. An operating lifetime at 600 C is not in that set.

On a lander, ambient temperature is not a design variable. In a turbine or a borehole it usually is, and the price is paid in cooling hardware and in the distance between a sensor and the electronics reading it, which is why geothermal drilling and aerospace engine control sit next to deep-space exploration in the authors' own list of targets ^[7]. The device family suits that work for a second reason: a JFET's operation does not depend on an oxide layer, which is where its lower noise comes from, and low noise is what a sensor front end wants ^[5]. The same physics caps its ambition, since JFETs are harder to shrink than the MOSFETs in consumer processors ^[4], so the integrated circuits this leads to will be modest ones by consumer standards. A switching point you can predict at temperature is the precondition for building any of them, and that, more than the 600 C label ^[1], is what the work demonstrates.