The spreading resistance probe is basically a glorified ohmmeter which does not sense doping type per se. Whether a profile gets plotted with "n" or "p" doping type is solely determined by the operator's input to the data reduction computer. The operators have a number of ways to determine type:

• We can rely on information supplied with the sample. If you tell us you have a boron implant into a p-type substrate, we may not check further unless we suspect something different based on the appearance of the measured resistance data. This has, on a regular occasion, bitten us in the hindquarters. Many times we find that what you tell us to expect and what we actually find are not even in the same neighborhood. Especially when the samples are for R&D – which isn’t to say that quality control hasn’t had its fair share of surprises. So we have determined that information provided by you has a less than desirable level of accuracy. Sorry. Therefore this method is our least favorite. Let’s move on to a few other options shall we?

• Minor flat location, though not absolutely reliable for smaller wafers, and not even available on larger wafers, will tell us substrate type. As we learned in the old school, with reference to the major flat:
• If the minor flat is 180 degrees from the major flat the wafer is N-type <100>
• If the minor flat is 90 degrees to the left or right the wafer is P-type <100>.
• If the minor flat is 45 degrees up on the left or right the wafer is N-type <111>
• If there are no minor flats the wafer is P-type <111>.

Since junctions are relatively easy to spot in the raw data (they cause a cusp in the profile), we can count junctions up from the substrate to get an indication of type for any portion of a profile. Unfortunately the industry is moving towards larger and larger wafers and putting flats on them is thought to consume too much surface area or to cause a severe wobble when spun. This has led to the introduction of the notch. Wafers with notches do not have substrate type and orientation indicators.

• The most commonly used method is the hot-probe. We have equipped our instruments with auxiliary circuitry which allows us to switch to a mode where one of the two probe tips is heated several degrees centigrade above ambient – hence "Hot probe". Then we run a second profile on the sample were we measure the open circuit voltage rather than the resistance between the probes. The polarity of the voltage, determined by the Seebeck Effect, tells us the conductivity type. The method is reliable, except on very thin layers – particularly when they are sandwiched between high concentration regions such as emitters and collectors, The sampling volume is several times larger than that for resistance measurement. When the heated probe touches the silicon heat is transferred to the sample. If there are both N and P regions in close proximity, sometimes the one that has the higher concentration will dominate the signal. Probe operators learn to recognize the shape and amplitude of the measurement and can intuitively determine the correct type. The method can also be uncertain in very high resistivity material. When hot probing samples in the 10⁴ and higher resistivities the N-type mobility being higher than P-type mobility can lead to false N-type signals. Quite often when probing high resistivity we will do 2 hot probe scans – the second being in reverse polarity. If the resistivity is too high the reverse polarity hot probe will not swap to the type opposite that indicated by the normal polarity hot probe.

• Chemical stains are often used, but are notorious for being misleading. They do provide supporting information though, and are used to confirm junction location. This is handy when we are suspicious that we may have fallen out of a pattern. But that’s a topic for another technical note... The Philtec Safe-t-Stain that we use can be handled without chemically resistant gloves and does not have a shelf life. It comes in 5 concentrations for staining lightly to heavily doped silicon. It will normally stain N-type material and leave P-type unstained. However we have seen it stain the P-type and not the N-type. Sometimes it even stains both!

• Experienced probe operators can often study a raw resistance profile and determine doping type. For a given pair of probe tips, n-type data is often noticeably less noisy than p-type. Other methods, such as measuring photovoltages or I-V characteristics, are feasible but we don't currently use them.

In the end, the probe operators sometimes just have to guess, but with all the clues available they nearly always guess right.