A Schottky Diode Can Cook Itself in Reverse
The reason to reach for a Schottky is the low forward drop, a couple of tenths of a volt where a silicon PN junction asks for the better part of one. The reason it comes back to bite is the other end of the curve. Reverse leakage in a Schottky starts out large and climbs steeply with both temperature and reverse voltage.
A diode leaking in reverse is dissipating power, the reverse voltage times the leakage current, and that power lands in the same junction whose temperature sets the leakage. Heat lifts leakage, higher leakage lifts dissipation, dissipation lifts the junction temperature again.
When the thermal path cannot carry that heat away fast enough, the loop does not settle. It runs until the part is gone, and the trigger can be nothing more than a hot ambient and a reverse voltage sitting near the rating.
The trade you cannot escape
A Schottky's forward drop and its reverse leakage come from the same knob, the metal-to-semiconductor barrier height. Lower the barrier to buy a smaller forward voltage and the same change hands you more leakage and worse high-temperature behavior. No silicon Schottky is both ultra-low-drop and low-leakage, because the physics that grants one removes the other.
Where the runaway happens
The datasheet leakage figure is the first trap, because it is quoted at 25 °C while the number you live with is the one at your real junction temperature. Schottky reverse leakage tends to rise by something like a factor of two for every ten degrees or so of junction heating, and it rises again as the reverse voltage approaches the rated value. A part reading a few microamps on the bench can be pulling a hundred times that once it sits hot and reverse-biased near its limit.
Follow the loop with numbers and the danger gets concrete. Say a small surface-mount Schottky shows single-digit microamps of reverse leakage at room temperature, the figure the schematic was sized around. Drop it into an enclosure that runs the junction to 110 or 120 °C, hold it near its rated reverse voltage, and the leakage no longer reads in microamps; it can climb toward the milliamp range, and a milliamp or two across thirty or forty volts is tens of milliwatts to over a hundred sitting in a package whose junction-to-ambient resistance can run into the hundreds of degrees per watt. That dissipation lifts the junction, the hotter junction leaks more, the extra leakage adds more dissipation, and whether the diode finds a stable point or walks off a cliff comes down to a race between the leakage curve and the copper. Stiffen the heat path and the system settles at some elevated but survivable temperature. Starve it, seal the box, add a summer afternoon, and the same part on the same rail runs away and fails short. Nothing in the room-temperature bench data hinted at it, because the whole mechanism only wakes up once the junction is hot.
The applications that walk into this are the predictable ones. A Schottky rectifying the output of a boost converter sees the full output in reverse every cycle and usually lives in a warm enclosure. An ORing diode between two supplies blocks reverse voltage around the clock. A reverse-polarity-protection diode bakes whenever someone connects the input backward and the board keeps drawing current.
In each case the part was chosen for its forward drop, sized off the 25 °C reverse spec, and never plotted at the temperature it would reach. It passes every bench check at room temperature and dies in a thermal chamber, or in the field the first hot week.
Silicon carbide buys an exit, at a price
Silicon carbide changes the math. A SiC Schottky leaks far less and keeps that low leakage as it heats, so the runaway loop loses the fuel it needs to start. The wide bandgap is doing the job the silicon barrier could not.
The bill comes in forward voltage and in price. SiC parts carry a higher drop and cost more, and they earn it where the reverse stress, the temperature, or the blocking voltage would push a silicon Schottky into the danger zone to begin with.
Choosing one that survives the summer
Derate the reverse voltage with room to spare instead of buying right up to the rating, since both the leakage and the chance of avalanche grow as you approach it.
Read leakage off the curve at your worst-case junction temperature, not the headline 25 °C value, and multiply it by the reverse voltage to get the heat the thermal design has to carry.
Hot diodes leak.
The clean switching is the part you keep. With almost no stored charge, a Schottky stops conducting the moment voltage reverses, which is why it belongs in fast converters where a PN diode's recovery loss would hurt; that virtue is real and stands apart from the leakage debt.
Then confirm the package and board can shed the dissipation with margin, because this failure is a thermal one wearing an electrical mask. When an alternate is needed and you are buying through distribution, qualify it on the reverse-leakage curve and the thermal resistance, not the forward drop; an independent distributor like In Fortune Electronics can match the exact part and bring the full curve set with it. A Schottky picked only for low forward voltage was picked on the one number that says nothing about whether it survives reverse bias in the heat.