Silicon Carbide High Voltage Schottky!!!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi guys,

I was looking for something else and stumbled upon Silicon Carbide High Voltage Schottky diodes up to 600V from Infineon

http://www.infineon.com/cgi/ecrm.dll/ecrm/scripts/prod_ov.jsp?oid=21102&cat_oid=-8681

Let me first say a few words about Schottky diodes and why we don't all use them today: First of all, they have only been available in relatively low voltage. A few years ago this voltage was 60V. Then it went to about 100V. It is very easy to get spikes with more than that so they are not really the most stable product. Also, considerable reverse current -- I mean these guys get HOT. Now that is enough about history -- let us move on to what I wanted to say.

I think I am onto somthing here. While I have only briefly checked out the datasheets for the 300V devices, I am impressed. Check out the leakage current -- it is like zero. They also claim zero reverse recovery as well as positive recovery. Forward voltage appears high for a Schottky, but the fact that it is linearly increasing with current is not a problem -- this is also the characteristic of traditional soft recovery units.

It is worthwhile to consider that Infineon out of the blue displaced IRF with vertical high voltage MOSFET's where the tradition up to that time (a couple of years ago) was that on-resistance for such devices increased logarithmically with voltage rating. That trend was broken to linear ... and IRF followed suit.

Silicon Carbide is a device material I have never heard of until this point. While I do not know intimate details of this material for semiconductor use, I am open minded and feel that Infineon may well be onto a winner here.

My guess is that Infineon has some very smart people on board.

I am quite excited, so bear with me -- I wanted to share immediately! I have marked the thread with a questionmark given that I have not tested this new product.

Petter
 
http://www.grc.nasa.gov/WWW/SiC/SiCReview.html

theoretical appraisals have indicated that SiC power MOSFET's and diode rectifiers would operate over higher voltage and temperature ranges, have superior switching characteristics, and yet have die sizes nearly 20 times smaller than correspondingly rated silicon-based devices

To be achieved of course

I rest my case and await for FET's ... (I understand JFET's will be next)

Petter
 
Infineon SiC Schottky diodes

Hello,

I'm the Applications Engineer for Infineon in North America for Power Semiconductors. So take my opinion with a grain of salt or two... ;)


SiC has some excellent material properties - breakdown field strength over ten times higher than silicon, high thermal stability (doesn't go intrinisic until above 500°C), and thermal conductivity comparable to copper.

However, it's very properties also make it much harder to work with and fabricate semiconductors. Current production wafers are 2", though we expect to introduce 3" late this year or early next. Defect density is an issue in the raw wafer material; fortunately, given the thermal conductivity and breakdown field strength, it's possible to fabricate fairly small chips that can handle ten amps, and with the thermal properties of true majority carrier devices (like MOSFETs), they parallel fairly well.

We've built PN diodes as well as schottky diodes, and as noted above, have built high voltage JFETs- and are developing composite transistors using low voltage MOS as a driver with a cascode JFET SiC switch.

The primary application is in current switching at high frequencies, SMPS power supplies in the 200 - 500 KHz region. They cut losses dramatically compared with the best ultrafast silicon diodes.

And yes, I think they work pretty nicely for "high end" audio; very minimal RC snubber to get a very "quiet" rectifier; but like silicon schottky's, their weakness is a lack of high surge current capability compared with PN diodes; because their is no conductivity modulation at high injected current levels. Forward drop is accurately modelled as a junction potential in series with a differential resistance that has a positive temperature coefficient. So, you can get thermal runaway on gross conduction overloads. But there are ways around that, too.


-Jon
 
I spent a lot of time getting a 3300 watt 435 vdc boost regulator to behave itself using a Harris RHRG3060 diode. The main problem was the overshoot voltage on the switching FET because the turn-off snubber for the diode only allowed a fairly spongy path to the main output cap so it was difficult to properly limit the peak drain voltage. If these schottkys were available 18 months ago, and particularly in bigger sizes then life would have been a lot easier.

I thing SiC schottkys for continuous made boost regulators as used in PFC front ends for SMPS's will be the best thing since sliced bread. What is the forward turn-on time like for these diodes?

GP.
 
The primary dopants for n and p material haven't changed- except that different ones are used depending on the degree of carrier concentration desired- so, for example, boron is used in a p-layer backside emitter (in the collector) for IGBTs, where low hole emission is desired in proportion to the electron current.
While Infineon does have a line of small signal silicon discretes, new development (aside from RF) is minimal; I'm not involved in those products.

The biggest changes have come about because of new equipment (much finer feature sizes) and the development of more "3D" vertical structures, as opposed to the more conventional planar structures. This is where the ability to create the vertical p-wells for our charge compensatio MOSFETs comes from (introduced in '98; commercial name "CoolMOS"), and also for our new low voltage deep trench MOSFETs, (p-channels announced, n-channel being sampled, announed in July) which have greatly reduced gate to drain charge/capacitance, and very low body diode Qrr (under 10 nC). Some of the processing technology and experience obviously comes from our memory chip work...

All these devices, unfortunately, are being developed for switching applications, not linear... the last power amp I designed with MOSFETs used Magnatec complementary lateral devices from England- similar concept to Hitachi's, but much better execution-- they're used in the Ayre V1. Thermal Q point around 250 mA, available in multichip parts, and fairly acceptable matching between P & N channel- though I ended up using a compound output stage design anyway, so device polarity matching wasn't much of an issue. Lately, (last two years) I've been working with the extended beta bipolars, also in non-loop feedback designs. Those are found in the recent Ayre amps, the V-5 and V-6.

Best regards,

Jon
 
Wow, you never know who is listening!

JonMarsh,

So when can we expect SiC high power JFET's? Current JFET's are low current, ultra-low power and low voltage.

What more can you tell us about low voltage deep trench devices? My selection criteria for such devices usually starts with picking a modern unit -- and selecting the one with the highest on-resistance :) while still having reasonable thermal resistance, and sometimes moving up in rated voltage beyond my needs. My guess is that a small footprint device in a large carrier would be very interesting to try for Audio. At what time would a small number of such devices be available for testing, and are there any that fit the bill? I know they are intended for switching -- but my brain struggles with accepting that new technology cannot be applied effectively to high end audio :) I particularly feel such units would be useful in creating cost savings due to ability to run at elevated temperatures, but the salient point here is optimal sound quality for now.

Which brings me to my final point: I searched Infineon for the SMPS eval board. I have long been wanting to use SMPS as a power supply. The eval board looked pretty radical but was set up for the wrong voltages, and probably designed more for efficiency than low noise. Is it relatively easy to create 200W units that have a single 30V output using Infineon technology? Ideally, the PSU should be switchable or infinintely variable (down in voltage) but I would be happy to settle for 30V single voltage or variants thereof. I sense that Power Electronics is still some form of black art, so would very much welcome your thoughts on this issue!

Thanks!

Petter
 
Hi All,

Wanted to add just some links to this old thread, as this seems to be main SiC thread so far.

Petter said:
[...]So when can we expect SiC high power JFET's? Current JFET's are low current, ultra-low power and low voltage.[...]

Cree some months ago put out their second generation power MESFETs in SiC:
http://www.cree.com/ftp/pub/crf24010.pdf

The parameters don't exactly imply an audio application, but if anybody is in the most exotic amp contest...

Judging from this link, it's also good for the most expensive amp contest (especially considering how many of these you must parallel for 8 Ohm load):
http://www.theta-j.com/creesic.htm

There is also more usefull stuff at Cree, this SiC diode looks fine:
http://www.cree.com/ftp/pub/CSD20030.PDF

Regards,
Peter Jacobi
 
SiC material

These silicon carbide (SiC) Schottky devices appear to be everything I ever had on my wish list. What I find amazing is the extremely low loss due to leakage current, particularly at elevated temperatures. Up until now, that has been the Achille's heel of Schottky devices, as well as the limited voltage capability. I'll definitely be using these soon. My first reaction was that I wish I had these available to me 3 years ago, for an offline SMPS design. But, then again, my competitors didn't have them either, so that's fine.

I first remember SiC being used for blue LED lamps 10 to 15 years ago. They were quite dim. At 20 mA forward current, they were barely visible outdoors in sunlight. Then Gallium Nitride (GaN) came out around 1995, and blue LED lamps made with this material were much brighter. As a result, GaN completely displaced SiC in the blue LED market, and GaN is also used for ultra bright green LED lamps as well. Recently, a combination of GaN and SiC has been used for LEDs in order to increase lifetime.

This is a significant breakthrough. Best regards.
 
We have been developing SiC JFETs for some years (since 1999), and another generation will be undergoing testing around May-June. They are intended more for HV switching applications, though, being combined with an 55V Si MOSFET to make a cascode switch. A single switch has an Ron of about 350-400 milliohms, with a breakdown voltage of 1500V. No, that's not a typo. The chip isn't all that large (4.5 mm2), so it's not sized for linear applications, but for switching, and minimum conduction loss.

New generations of 600V diode will be released later this year, and probably 300V by the end of the year.

Best regards,

Jon Hancock

Infineon Technologies, NA.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.