We all know that process control is everything. And this is true more than ever before at each node. That is, if there is a notion of “more than everything.”
Of course, one of the critical parameters in pretty much any process is temperature. Monitoring it to make sure nothing is going awry is, to put it mildly, a good thing. How to do that then?
Most wafer-processing equipment involves one or more chucks to which a wafer is secured (usually by a vacuum). And those chucks often have temperature sensors in them, so that’s a start. But the chuck temperature isn’t the same as the wafer temperature – heat has to get from the wafer to the chuck, and if you really want a direct picture of what’s happening on the wafer, you may not be satisfied with knowing the chuck temperature.
So to do this, why not measure an actual wafer? This is, of course, doable – just attach a bunch of temperature probes to various positions on a wafer and then run it through the process. The main problem here is that you’ve got this pigtail of wires to deal with from the probes, meaning you can’t seal down the process chamber, meaning you’re still not getting a real in-situ picture of the temperature as the wafers experience it when being processed – like with a plasma going.
KLA-Tencor’s SensArray division takes things one step further with a wireless approach. They do this by building a silicon sandwich. First they have a wafer with temperature probes plus a battery and some mild CPU capability and memory. Then they cover that with a blank wafer so that, at the end, the top looks like a fresh wafer.
They can then put this in the transport system and monitor conditions before, during, and after processing, getting a full picture not only of the chamber, but also of the environment in the immediate vicinity. They can store temperature as a function of position on the wafer as well as time.
Once the wafer comes out, it can download the data wirelessly. KLA-Tencor has a souped-up FOUP (Front-Opening Universal Pod, a super-tricked out wafer cassette for those of you/us that haven’t walked through a fab in a while) they call the Basestation 300 (or 300Z). This lets you run a sensor wafer through an entire line (almost), download the data from the wafer, and charge the battery on the sensor wafer. Depending on the level of integration with the wafer equipment, the downloaded data could be used to tweak equipment settings immediately – which is pretty close to real time given that all the measurements can be made in about 5 minutes.
There’s only one gotcha with the wireless approach: it can’t be used in any process step involving a temperature over 140 °C (the wired version can go up to 1100 °C). That’s because of the battery and the circuitry that needs to operate while in the chamber. So diffusion, for example, can’t be measured wirelessly.
Now… you’d think that a single wafer sandwich would suffice for all of the various process steps, but it’s not quite so simple. So far they had an EtchTemp version available for measuring the SiO2 etch step, but that can’t be used for measuring Si etch – because it would etch the top wafer. So they’ve announced one that has an extra Y2O3 coating to protect the wafer from being etched; they call it EtchTemp-SE.
Then they announced one called ScannerTemp for measuring the uniformity and stability in the scanner. But here they can’t tolerate the extra thickness of the EtchScanner wafers, which are 40% thicker than a standard wafer. So this is thinned back down to the thickness of a standard wafer and is calibrated tightly within the 20-24 °C range for immersion measurements.
Finally, they announced a WetTemp-LP (low profile) version for wet process temperature measurement. This also has standard wafer thickness, but the temperature range and accuracy are broader than that of the ScannerTemp edition.
You can find more info on their release…