Overview Lab 1: Field Oxide Brandon Baxter, Robert Buckley, Tara Mina, Quentin Vingerhoets Lab instructor: Liang Zhang Course Instructor: Dr. Gary Tuttle EE 432-532 January 23, 2017 In this lab we created our first layer of oxide on the bare silicon wafers using the wet oxidation process. This is the first step for processing our CyMOS device and it is important, because it will form the base oxide layer for the entire device, into which we will later cut holes for our adding n-type and p-type dopants and then re-cover with an oxide layer, afterwards. Because we were going to expose our wafers to high temperatures for this oxidation process we first had to perform the standard clean at the beginning of the lab. This was important because we had to remove organic and inorganic contaminants on our wafers before exposing them to intense heat from the furnace. Brandon was selected to read each step of the SOPs for the standard clean process out-loud, and Tara was selected to actually perform each step of the SOPs at the Wet-Process Benches. To clean the wafers, we first submerged them in a heated, basic solution of ammonium hydroxide at 75 degrees Celsius. Then, after briefly putting it in a tub of dilute hydrofluoric acid and then rinsing with deionized water, we immersed it in an acidic solution of hydrochloric acid at the same heated temperature. Afterwards we had to dry the wafers in the spin rinser and dryer machine before putting it in the furnace. Once we completed all of the steps in the SOPs for the standard clean we proceeded to the wet oxidation procedure to grow the silicon dioxide layer on the surface of our wafers. This layer was supposed to be about 250 nanometers thick. Using our knowledge of the wet oxidation procedure we calculated how much time we needed to expose our (100)-oriented silicon wafers in the furnace, which was set to be at 1100 degrees Celsius. However, we suspect that the furnace was at a higher temperature than we thought, so we actually grew more oxide than we were expecting, resulting in a layer that was about 370 nanometers in thickness. To do this process for wet oxidation, we first had to load the wafers in the furnace and then push them to the middle of the furnace extremely slowly, about 5 inches per minute. Then, after ramping up the temperature to 1100 degrees Celsius, we turned the bubbler on to heat up deionized water for the wet oxidation process, which will enter the furnace as steam mixed with nitrogen gas. Then, after waiting for the amount of time we calculated for our desired oxide thickness, we had to turn off the bubbler and ramp down the temperature of the furnace and very slowly remove the wafers from the furnace. Starting Wafers Our team has 10 wafers in total, which at the beginning stage of the process are all bare (100)-oriented silicon wafers that are 3 inches in diameter and about 380 micrometers in thickness. The lot number for our silicon wafers was written to be 1511-4023, though we were
not sure what this code represented exactly. Our wafers are all n-type, they are doped with phosphorus ions, and at the beginning of the CyMOS process they all start out as essentially identical bare silicon wafers. Of these 10 silicon wafers, 4 of them will become our CyMOS devices as a final product by the end of the semester. Another 4 of these wafers are test wafers which will be maintained with the same thickness of oxide as our silicon wafers and in the future, for each time we need to etch the oxide layers on our CyMOS devices, we can use one of our test wafers, which should have the same thickness of oxide, to determine how long we need to etch the wafers by etching one of our test wafers clean and timing how long it takes. The 2 remaining wafers are backup wafers, in case any of our other 4 production wafers are damaged. For this lab we are using wafers that are n-doped with Phosphorous. Before processing our Silicon wafers we did some preliminary testing on them to determine more precisely what the initial doping concentration of our wafers are. To do this we had to measure the average resistivity of each wafer using the four-probe technique. Using this method we determined the current and voltages measured to determine the average resistance to calculate the resistivity of our wafers, we measured it to have the following resistivity on average: 3.948 Ω cm From this resistivity value we measured we were able to calculate their approximate doping concentration, determining this concentration to be the following, on average: 15 3 1.1 * 10 cm This doping concentration value is important to know the approximate value for in our wafers, because of future calculations we may need to do for this. See the recorded values for the four-probe technique on the photo of the scanned worksheet titled Manual four-point probe measurements in the Appendix section at the end of our report. Also in the Appendix, see details of the characteristics of our starting wafers on the photo of our scanned process traveler sheet titled Starting Material. Standard Clean Before we were able to perform the actual wet oxidation process we had went through the standard clean process which is used to clean the wafer of organic and inorganic particles to remove contaminates. This was particularly important to do before the wet oxidation processing because we need to put the silicons in the furnace. In fact, in any high temperature process for the silicon wafers, it is critical that the standard cleaning procedure be performed beforehand. This is because, if there are any contaminants on the wafer when putting it in the furnace, these contaminants get more deeply incorporated into the wafer and can ruin its functionality.. Overall there are two main chemical solutions we use for specific purposes in the cleaning process. The first solution is a basic solution that is created in the tub labeled SC-1 by mixing ammonium hydroxide and hydrogen peroxide. The main purpose of this first solution is
to remove organic materials like fingerprints, dead skin cells, and other organic particles that have imprinted or fallen on the wafers from people working in the lab. The second solution, which is mixed in the tub labeled SC-2, is an acidic solution which is made by mixing hydrochloric acid and hydrogen peroxide. This solution is used to perform an ionic clean and remove metallic particles from the wafer, which come from using metallic tweezers to handle them. Both solutions are diluted in deionized water and are supposed to be heated to about 75 degrees Celsius before the wafers are submerged in them. After preparing the two solutions the wafers are first placed in the SC-1 tub with the ammonium hydroxide. Then, after waiting 15 minutes for the wafers to be fully cleaned by the basic solution, the wafers are briefly placed in a tub of dilute hydrofluoric acid and then rinsed in the cascade rinse tub. Then the wafers are placed in the SC-2 tub with the hydrochloric acid and once again rinsed in the cascade rinse tub. Finally, once the wafers have been exposed to both solutions for the cleaning process the wafers still need to be dried. They are placed in the spin rinser and dryer machine so that all water droplets are removed for the next processing stage in a way that does not damage the wafers. Wet Oxidation After measuring the doping concentration of our n-doped silicon wafers, and performing the standard clean process, we were finally able to start doing the wet oxidation process. The first step we underwent was to perform the calculations to determine the amount of time we needed to place the wafers in the 1100 degrees Celsius furnace for in order to create an oxide layer of a certain desired thickness. This calculation work is done and shown on a scanned copy of our handwritten work, which can be found in the Appendix section of this document, which gave us the following time value for the wet oxidation process: 12.25 min We used the equations to actually calculate this amount of time, but looking at the graph for the wet oxidation process for 1100 degrees Celsius and a desired thickness of 0.25 micrometers, we can see that this amount of time can also be verified, approximately, by considering the curve drawn on this graph, giving us somewhere around 0.2 hours:
To grow the oxide we placed the wafers in long, thin ovens that had just enough room for our 3-inch wafers in their quartz boat carriers. After we set the temperature for these furnaces to their standby states of 800 degrees Celsius with 1 standard liter per minute of dry Nitrogen gas we gently loaded our wafers into the furnace. This loading and pushing process took a relatively long amount of time because we had to enter the wafers into the center of the furnace slowly using the long push rod and sliding them in at a rate of about 12 inches per minute. So, because the furnace was about 5 feet long, this process took about 5 minutes. The reason we had to enter our wafers so slowly into the furnace was because of the large temperature difference we were subjecting them to from room temperature to the 800 degrees Celsius in the center of the furnace. If the wafers increase in temperature too quickly they become more likely to shatter, the same way that glass can crack and shatter if its temperature changes too quickly. After the wafers were loaded we then had to ramp up the temperature again to be our desired temperature for the wet oxidation process. Thus, we set the furnace to 1100 o C and waited for the oven to reach the proper temperature. While the furnace was heating to this temperature, we also turned on the bubbler, which heats the water to almost boiling, reaching a 98 degrees Celsius. The bubbler was set to allow 200 ccm of nitrogen gas to flow through it. This water in the bubbler is the water used for the wet oxidation process, and before starting this process, we wanted to wait for the bubbler to increase its temperature to almost 96 degrees Celsius at least, since there was a plus-or-minus 2 degrees from the desired temperature for the acceptable range. Because our bubbler was taking a surprisingly long time to get to this minimum 96 degrees Celsius, once it had reached a temperature of 95.7 degrees Celsius, our lab TA said that this was close enough and that we could proceed with the wet oxidation process after all, so we started the process even though we had not quite met the 96 degrees Celsius minimum. Once the water was hot enough, to start the wet oxidation process, we had to flip a switch that would send the nitrogen gas and steam mixture into the furnace tube where our wafers were
to start the oxidation process. Waiting for the calculated amount of time, which we did at the beginning of the lab, after 12.25 minutes (or 12 minutes and 15 seconds), we stopped the oxidation process by flipping the switch back so that only dry nitrogen gas was allowed to flow into the furnace tube. After turning off the bubbler and shutting off the flow of nitrogen, we ramped the furnace temperature back down to 600 o C, and left our wafers in the furnace overnight to be removed the next day by the next lab group. These details of our wet oxidation process are also presented in scanned page the process traveler sheet entitled Field Oxidation. Results Overall the first lab seemed to go relatively smoothly. We calculated the amount time we needed to expose the wafers in the 1100-degree Celsius furnace, double-checked these calculations before proceeding, and in general there were no notable problems we ran into during the lab. However, we obtained interesting results for our oxide layer, which we also attempt to explain the cause of in this section. During the next lab period we were to check the oxide thicknesses for the silicon wafers, and see the results of the work we had done from the field oxidation process the week before. To check the oxide thickness, we had to use a device in the back of the lab room which was created by a company called Filmetrics. This device uses known optical properties, like the refractive index of a specific wavelength of light for specific materials, and measures the amount of reflectance of this shined light in order to determine the thickness of our oxide layer. Thus, when using the Filmetrics machine, we set the machine to determine the thickness for the following setup: Top material - Air Middle material - Silicon dioxide Bottom material - Silicon ( about 380 micrometers in thickness ) So, knowing the optical properties of a specific wavelength of light through air, silicon dioxide, and silicon, the Filmetrics device, using a complicated equation, can determine the thickness of
the silicon dioxide by looking at the amount of reflectance from the light. Indeed, based on the reflectance value that the Filmetrices devices sees, we can get many different possible values for the thickness of the oxide, since the reflectance, which is a function of the thickness of the oxide, tends to repeat itself as the thickness of the oxide layer increases. Giving the Filmetrics device an approximate thickness, which is our goal of 250 nanometers for our oxide, the device is able to determine the most likely oxide thickness based on the reflectance value it measures. You can observe this repetition in the function for one of the graphs we obtained from the Filmetrics machine for one of our measurements: When doing this process, we measured one of our test wafers using a 4-by-4 wafer mapping of the oxide thickness for the first test wafer. Then for the other three test wafers, we only measured 1 point in the center of the wafer. For the measurements we got of the wafer mapping, we got the following values for our oxide thicknesses, in nanometers: 1 370.0 nm 5 370.2 nm 9 371.6 nm 13 370.0 nm 2 370.1 nm 6 371.3 nm 10 371.9 nm 14 370.4 nm 3 370.3 nm 7 371.6 nm 11 372.1 nm 15 370.2 nm 4 369.9 nm 8 371.3 nm 12 371.8 nm 16 369.8 nm
For the other test wafers, we did not perform a mapping, like did we the first one, and we simply measured the thickness of the oxide in the center of the wafer. Doing this, we obtained the following results: Test Wafer 2: Test Wafer 3: Test Wafer 4: 370.2 nm 371.1 nm 370.7 nm Looking at these results from our measurements, we found we had grown far more oxide than we had intended. We had meant to grow 250 nm, but we had grown between 369.8 and 372.1 nm, averaging to 370.8 nm of oxide. Asking our lab TA, Liang, about why this might have occurred he said that he thought the furnaces may have been hotter than what we were setting them to. So, even though we set our furnace to be at 1100 degrees Celsius and we used this temperature to calculate the amount of time we had to put the wafers in the furnace, it is possible that the furnace was at a higher temperature than what we were expecting. However, even though our oxide is significantly thicker than what we were expecting our lab TA said this difference should not be a critical concern, for it will not prevent our CyMOS devices from being functional ones. Appendix Here is the hand-written results we obtained from our calculators for calculating the amount of time we needed to perform the wet oxidation process for a desired thickness of 250 nanometers in a furnace at 1100 degrees Celsius:
Here were the measurements we obtained from doing the four-point probe technique in order to determine the resistivity of our wafers and, from that measurement, the doping concentration of the phosphorus n-type dopant:
Here is the filled-out lab sheet of our process traveler, describing the characteristics of our silicon wafers before we began any processing of them:
Here is the sheet of our process traveler describing the details of the steps we went through for the field oxidation processing of our wafers, which was done in this lab session: