Mettler-Toledo International Inc. (MTD) Earnings Call Transcript & Summary

Hi, hello again. Thank you for joining this live webinar today. On the topic of an innovation and resistivity measurement, or manufacturers of semiconductors. Before we start, I just want to let you know that the webinar is being recorded for possible future years. Any Q&A session at the end won't be included in any future use. [Operator Instructions] My name is Philip Barnes, I'm value content manager for Mettler-Toledo Process Analytics. Our presenter today is Joel Kenyon. Joel has spent the last 12 years in product management for industrial analytical equipment, and he is the Global Product Manager for conductivity and transmitters at Mettler-Toledo Thornton and I will now hand over to Joel.

All right. Thank you, everybody. As Philip mentioned, my name is Joel Kenyon. I am the Product Manager for conductivity and transmitters at Mettler-Toledo Thornton. And I'm excited to talk with you today about our latest innovation in resistivity measurement. So today, we'll go over a few things. To start, I'll go over just a brief overview of Mettler-Toledo Thornton, who we are and what we do. We'll review the measurement principle for measuring resistivity in a water or aqueous sample. And we'll go over some of the challenges that face users who are trying to measure resistivity accurately within an industrial environment, particularly within the semiconductor industry. And then we'll take a quick look at some market trends that are driving this sort of innovation in the resistivity measurement from the Mettler-Toledo perspective and then we'll get into our solution to those challenges and trends, which is the UPW UniCond. And then we'll take a quick look at where in the process the UPW UniCond would have typically been installed. [indiscernible] start. Thornton is a business group within the Process Analytics division of Mettler-Toledo. And at Thornton, we are particularly focused on pure and ultrapure water analytics. Thornton has been a leader in the production of high-quality resistivity measurement equipment specifically for the semiconductor industry since the 1960s. So we've been doing this for over 6 decades at this point. We are very heavily involved in industry organizations like USP and the various other pharma companies, for the pharmaceutical industry as well as Semi and the ITRS road map committee for the semiconductor and microelectronics fabrication industries. I mean we have very good working relationships with almost all global players within the water system fabrication business. So just about any company that builds water systems for use in pharmaceutical, semiconductor or power generation industries, we have a very strong relationship with. Next, I'll just quickly go over the measurement principle for a 2-electrode resistivity measurement, and as I'm sure most people here are aware, resistivity is generally used as a sort of generalized indicator of water quality that is used to measure and control the quantity of ions in water or aqueous sample. It measures total ionic concentration. This is why I say it's sort of a generalized indicator of water quality. It is not ion specific. It relays the total amount of contamination in the sample. It is based on a theoretical -- the measurement itself is based on a theoretical 2-electrode plate configuration, whereby you have 2 electrodes of unknown surface area spaced a known distance apart from each other. We're using the relationship between the surface area of the 2 electrodes and their spacing to determine a mechanical constant, which is called a cell constant. And we use that relationship in combination with the current carried by ions in solution between the 2 plates to determine what the conductivity or the ability of the solution to conduct [indiscernible]. Higher conductivity indicates more contamination in the sample. Resistivity is the inverse of conductivity. So when conductivity is high, resistivity will be low and vice versa. In practice, what we do is we take 2 electrodes, we connect them. We plant an AC voltage across the 2 electrodes, and again, ions in the solution will act as carriers of current. And the flow of current between the 2 electrodes is directly proportional to the amount of the ions dissolved in the solution. The 2-electrode plate that the measurement is based on -- the 2-plate electrode that we talked about that the measurement is based on is not feasible or realistic for installation into process applications. As a result, Thornton pioneered the design of the concentric electrode design. This is the typical design that you'll see in the marketplace today where you have 1 outer electrode ring with an inner electrode inside of it, so cylindrical inner electrode inside it with an insulator separating the two. This is done so that we get a homogeneous electric field, which increases the signal quality in addition to allowing better sample flow through the conductivity measurement cell. The sort of relationship between cell constant and conductivity solution -- or the ideal cell constant for a specific conductivity solution has to do mostly with when talking about this concentric ring design as we do mostly with the size of the inner electrode and the spacing of the electrodes. You'll see a 0.01 centimeter constant sensor here on the right and the [ 0.01 ] constant sensor here on the left. And as you can see, the sensor with the smaller cell constant has a much larger diameter inner electrode and thus much less spacing between the inner electrode and the outer electrode. These sorts of electrodes are ideal for measuring ultra-high resistivity samples in sort of non-aqueous solutions. This would be used for measuring in media like alcohol or glycols. The 0.1 centimeter consent sensor is very well suited for measuring in aqueous samples, particularly those of high resistivity like UPW and Mettle-Toledo was standardized on this 0.1 centimeter constant sensor for the vast majority of applications simply because, as I mentioned earlier, it allows for much better flow -- sample flow through the conductivity cell, it also eliminates or helps reduce the risk of accumulating particulates from things like corrosion products or resin fines. It also allows faster rinsing, especially in high conductivity applications where the sensor may be exposed to a high conductivity sample, where contaminants may stick to the electrode, you find much faster rinsing when using a 0.1 constant sensor. In addition, as you can see, the thermal nets of the inner electrode are the 0.01 constant sensor, makes it a little bit less responsive to changes in temperature. This becomes very important, especially when making compensated resistivity measurements. Okay. Now I'll just go over some of the primary challenges to measuring resistivity accurately within an industrial environment. And within the semiconductor industry, the very first is the actual accuracy of the sensors that are available. Plus or minus 1% accuracy has sort of been the standard for many years. But there are also some manufacturers out there that make sensors that have a 2% accuracy specification and some as high as 3%. The SEMI F63 guideline specifies that water for use in wafer fabrication at the point of distribution should be 18.18 megaohm centimeters and resistivity plus or minus 1%. And you can see that if the sensor being used to measure resistivity as an accuracy of 1%, this means that as soon as you deviate from the ideal 18.18 megaohm centimeter water quality that it's very possible that the sensor can either read out of spec, while the water is, in fact, still in specification or the sensor can read in spec, while this water is actually out of specification. And this is all due to the measurement accuracy of the sensor and not necessarily any actual variation in water quality. This number in terms of sensor accuracy also, in many cases, customers tend to assume that this number assumes temperature compensation is being accounted for. And it's very important that they know for sure because many manufacturers will, in fact, specify uncompensated resistivity accuracy or being big use as to whether that accuracy specification is compensated or uncompensated. And what this means is that if that measurement accuracy is uncompensated, then there is an additional variation that needs to be accounted for when temperature fluctuates. So when temperature fluctuates, the resistivity will fluctuate if it's not compensated. And even if it is compensated, adding that compensation will add some additional amount of error to the accuracy specification. It's very important to consider this while understanding how accurate the sensor is actually measuring in the user's process. As we discussed, the next piece is the effect of sample temperature on the resistivity measurement itself. Resistivity measurement is dependent upon temperature. The mobility of ions increases at higher temperatures. As ion mobility increases, conductivity increases. And as we know, resistivity is the inverse of conductivity. So when conductivity increases, resistivity decreases. So, it's a whole way of saying that when temperature over solution increases, resistivity will decrease. In a typical salt solution, the amount of change per degree C in the resistivity of the solution is about 2% per degree C. In UPW, that coefficient can be anywhere between 4% and 7% per degree C depending on the quality of the UPW. That means that for a single -- that for a single degree C change in process temperature, the resistivity reading of that water could change by up to 7% if the measurement is not temperature compensated. This is why we compensate measurements. Resistivity is most often in most industries other than the pharmaceutical industry made as a compensated measurement. This way we can reduce the variability due to simple temperature fluctuations and the resistivity measurement, so that we can focus on changes in resistivity that are actually due to water quality, not just small changes in temperature. In addition to temperature of the process or the sample, ambient temperature can also have an effect. When we get differences between process temperature and ambient temperature, we can see temperature gradients. And those temperature gradients if the temperature sensing element of the sensor is not constructed properly, can cause heat to bleed into or out of the sensor. And as a result, this causes artificially high or artificially low temperature readings, which have a direct result on the compensation of a temperature-compensated measurement, and this introduces error into the measurement. As you can see in the example on the right, baseline resistivity is around 18.12, but when the delta -- the temperature delta between the process temperature and the ambient temperature reaches about 20 C, we see a difference between the starting point and the resistivity measurement had elevated ambient temperature of about 0.25 megaohm centimeters caused by a 20-degree C difference between the ambient temperature and the process temperature. This can cause fluctuations as well. So the fourth primary challenge is noise and signal disruption. As most of you are aware, the industrial environment that these sensors are measuring in is replete with things like pumps, variable frequency drives. There are challenges associated with static buildup in plastic piping systems, especially if the various components and analytical equipment are not properly grounded. In a sort of common situation, it's very easy to see something like we do on the right -- in this chart on the right, where we see on the order of 0.1 megaohm peak-to-peak standard variation as sort of standard across the measurement. And when we look back at the SEMI-F63 guidelines, 18.18 plus or minus 1% 0.1 megaohm peak-to-peak variation is more than 50% of that error budget wrapped up in just the noise being introduced to the measurement circuit without looking at any actual variation in water quality. In addition to this sort of broad noise that can be -- that can affect the measurement, there are also -- there's also the opportunity for much larger signal disruptions that can cause communications drop out entirely. This can cause 0 readings that artificially flag alarms or just loss in communication or loss of visibility of water quality to the control system. Now some of the industry trends, there are some very prevalent industry trends that are driving sort of increased scrutiny on water quality at the fab level. And what we see are trends like artificial intelligence, automotive applications, as well as the increasing commoditization of consumer electronics and miniaturization of -- simultaneous miniaturization and increased complexity of the components that make up these consumer devices -- sort of driving a need for higher quality UPW to help optimize yield of these processes while simultaneously increase scrutiny on water usage and water shortages globally are driving a need to increase the amount of water that is reclaimed or reused in the process. So because of these industry-level trends and we are starting to see increased focus at the plant level, with engineers being required to meet stricter UPW requirements in order to accommodate acceptable yield levels while implementing smaller line widths on wafers. In addition, those tighter budgets associated with commoditization and increased scrutiny on water use mean that while they are simultaneously -- simultaneously while they are under pressure to deliver higher quality UPW, they are also under pressure to meet increasing demand for reclaim, reuse and recycling that water as well. So at Mettler-Toledo Thornton, we've seen these trends towards smaller line widths causing increased scrutiny on UPW quality as well as this drive towards reclaim, reuse -- higher rates of reclaim, reuse and recycle and realize that these evolving needs required a resistivity sensor that was sensitive enough to meet the increasing requirements of UPW quality as well as helping customers make faster decisions. And what we did is we worked with industry partners to develop new firmware, electrical and mechanical enhancements in order to deliver a more accurate, more stable and robust resistivity sensors that could help these customers to meet these evolving needs. And the result of that collaboration is the UPW UniCond. As I mentioned, we have worked with numerous semiconductor manufacturers to help develop the sensor. And that collaborative process resulted in a sensor that is more stable, more isolated from ambient effects like temperature and noise, and provides an accuracy that is unparalleled in the industry. So just to give you a quick overview, when we talk about accuracy, the example I gave earlier, is the SEMI-F63 guideline where ultra-pure water at the point of distribution should be 18.18 megaohm centimeters plus or minus 1%. Most fabs or foundries will set action levels at around 18, maybe as low as 17.9. And as we discussed earlier, most resistivity sensors on the market currently have an accuracy specification of plus or minus 1%. The UPW UniCond provides an accuracy specification of plus or minus 0.5% compensated resistivity on a compensated resistivity measurement. So as an example, let's say that a fab is producing water that has an actual quality of 18.15 megaohm centimeters, and their alarm limit is 18. If their actual water quality is 18.15 megaohm centimeters, that should be well within their allowable limits with a -- when using a resistivity sensor that has a 1% accuracy specification, it's very possible that even if this water is actually 18.15 megaohm centimeters that the resistivity sensor will report that water as out of specification. Even if -- and it can report that water is out of specification while still being within that sensors out of specification. What this means is that the customer is now in a situation where they could be taking unnecessary action to solve a problem that doesn't really exist in their water system. Because that they could be shutting down a process line and performing some sort of preventative maintenance like regenerating resin or replacing resin when they don't actually need to ahead of schedule incurring downtime and costs that they shouldn't have had to incur. With the UPW UniCond if the actual water quality is 18.15 centimeter -- megaohm centimeters, the 0.5% compensated resistivity accuracy means that the customer does not have to worry about unnecessarily triggering an alarm. There is no situation where if water quality is actually 18.15 megaohm centimeters that the UPW UniCond would report an out-of-spec reading. This ensures the customer that they are not taking the necessary action and prevents them from unnecessarily having to perform maintenance or shut down a process stream in order to perform that maintenance and lose potential production resources down -- further downstream. Perhaps the more concerning example, sorry -- perhaps the more concerning example is what if the water quality is actually out of spec? So again, water quality is intended to be 18.18, plus or minus 1%. If the alarm limit is set at 18, then a reading at 17.9 should trigger an alarm. If the sensor has a 1% accuracy specification, it is entirely possible that if the water quality is actually 17.9, that, that sensor could report that water as in spec. This means that in a situation where the sensor's accuracy is plus or minus 1%, the customer could very conceivably be sending water to their downstream processes that is well out of specification. It could contribute to decreased yields in their fab. With the UPW UniCond and it's 0.5% compensated resistivity accuracy, if the actual water quality is 17.9 megaohm centimeters, it will report that water as out of specification. There is no situation where if the actual water quality is 17.9 And the alarm limit is 18 that the UPW UniCond will report that water is in spec. This helps the customer ensure that the water that is reaching its points of use or is leading its point of distribution going to the various tools and points of use is of sufficient water quality to help them meet their yield targets. They don't have to worry about sending water to their points of use that is going to bring contamination along with it and thus harm their wafer yield. Sort of the second piece is what we -- what we've done to mitigate the effects of the electromagnetic interference that I've spoken about earlier. As the industry pushes closer to sub 3-nanometer technology nodes, sensitivity to contaminants becomes of increasing importance. So you need to really be able to distinguish noise from actual changes in water quality. And what you see here is in the example on the right-hand side, we've put the UPW UniCond in process along with a competitive sensor. In the gray trace -- in the gray jagged trace is the competitive sensor, reading in a relatively constant stream of UPW. The blue trace on the top, this is the UPW UniCond and you'll see that there is an order of magnitude improvement between the competitive sensor and the UPW UniCond. The UPW UniCond sensor is stable. It is not noisy. And it makes it possible to actually see real changes in water quality through the noise. Now really test the capability of the UPW UniCond sensor, we wanted to simulate sort of a high noise event, something like a pump turning on right next to it or a drive turning on or build up of static electricity, something along those lines. So we injected 3-volt peak-to-peak noise directly into the UPW UniCond's measurement circuit. And the trace here in orange is what happened when we did that. So even with noise directly -- injected directly into the UPW UniCond's measurement circuit, it is still 10x more stable than a competitive sensor. What this means is less time spent troubleshooting, less time spent trying to tie a change in resistivity to a pump turning on somewhere else in the process or to a drive activating right next to the transmitter. Anything along those lines, we can eliminate those factors from resistivity fluctuations and determine that any changes in resistivity are actually due to water quality and not any other environmental factors. Like we discussed earlier, ambient temperature effects can also have an impact on resistivity readings. And with UPW UniCond, we have built in very good thermal isolation to prevent anything like that from happening. Temperature control, both process and environmental, is very energy and cost intensive. And with the UPW UniCond, it becomes much less critical to control those things very tightly in order to get an accurate resistivity measurement. What we see here on the right is again a competitive sensor mounted in line with the UPW UniCond. And these sensors were measuring UPW well housed in an environmental chamber. The gray trace is the resistivity value from the competitive sensor. The green trace is ambient air temperature. The blue trace is the resistivity reading from the UPW UniCond. You can see that the competitive -- the resistivity reading from the competitive sensor very closely correlates to changes in ambient air quality as to say that ambient air quality is directly affecting or ambient air temperature is directly affecting the resistivity reading from the competitive sensor. Meanwhile, the UPW UniCond is flat, there is no effect from ambient temperature on its resistivity reading. Again, this means that we eliminate 1 more environmental factor from influencing the resistivity measurement. The user is truly certain that things like HVAC systems turning off and on at different times a day over seasonal changes in ambient air temperature are not causing artificial changes in the resistivity measurement, and any changes that they are seeing are truly due to water quality, providing them greater insight into their process. Sort of the last piece is the process temperature or temperature compensation. As part of this UPW UniCond project, we both redesigned our temperature measurement circuit as well as reapproached temperature compensation or refined our temperature compensation algorithms. Thornton developed the first temperature compensation algorithms, the Thornton [indiscernible], decades ago. And we haven't rested on our models. We've continued to reevaluate temperature compensation and have done whatever we can to implement it in a more accurate fashion. And the latest result of that is what we've done with the UPW UniCond. What you'll see here -- and because of these efforts, the UPW UniCond is the only sensor available that is able to specify plus or minus 0.5% accuracy on a compensated resistivity measurement. And what you'll see here on the bottom is an example of what that means. On the left, you have a competitive sensor. And you can see here the orange trace is process temperature, the blue trace is resistivity. You can see that when prices temperature is high, resistivity is low. And again, when process temperature is low, resistivity is high. Total change in temperature on this chart is about 1 degree C and the change in resistivity at each peak is about 0.1 megaohm centimeters. Now as we've discussed earlier, when temperature increases -- when process temperature increases, resistivity will decrease. So you would -- on an uncompensated veteran, you would expect that when temperature goes high, resistivity goes low. Now the scale of this change -- of this change in resistivity tells us that this measurement is, in fact, compensated. They are doing some level of temperature compensation. However, they are still getting the 0.1 megaohm centimeter change for every degree C that the process temperature changes. That means this again, consider the SEMI-F63 guidelines, 18.18 centimeters, plus or minus 1%. The error budget for the process is 1%, that's 0.18 megaohm centimeters. Here, more than half of that error budget is being consumed because of a 1 degree C change in process temperature. That means that there is now only 0.08 megaohm centimeters allowable variation in the process that can actually be due to real changes in water quality. You now have a sensor that is inaccurately compensated that is causing fluctuations in resistivity readings because of very small changes in process temperature. And on the right-hand side, you have the UPW UniCond. Again, similarly, 1 degree C changes in process temperature, and they have -- because of the temperature compensation, they're having no effect on the compensated resistivity reading. Again, this means that the customer can be certain that any changes in resistivity are due to real changes in water quality, not some other component. It minimizes the importance of energy and cost-intensive processes like managing process temperature and ambient temperature. It allows you to actually see the real water quality independent of environmental effects like process temperature, ambient temperature or electromagnetic interference. And this is applicable at most measurement points within the semiconductor water cycle, particularly at that point of distribution. Water quality at that point of distribution is critical to downstream yield in the wafer fabrication process. With the UPW UniCond, the user can be certain that water both entering and leaving that point of distribution is pure enough to meet their yield targets in those wafer fabrication processes. They can avoid taking unnecessary action when water is still in spec, and they can be absolutely certain that the water that they are sending further downstream is of the required quality. And in that way, the advancements of the UPW UniCond are helping meet these tightening tolerances also upstream. Further upstream, you can increase or decrease tolerances of upstream treatment processes like RO or mixed deionization or ultrafiltration steps. The UPW UniCond can help tighten tolerances on these upstream processes as well. At the point of use are the various tools that are used in wafer fabrication, water or UPW is heavily involved. So water from the point of distribution is used at these various points of use as a substrate for chemical mixing for various processes like epitaxy, etch and photolithography. It's also used in cleaning between each step. We use a process called rinsed resistivity to make sure that water leaving the tool is suitably clean when compared to water that came in to make sure that the water -- that the wafer is clean enough to move on to the next step. And these processes, the UPW UniCond can help make sure that incoming water is of sufficient quality to maintain yield targets. We're also reducing rinse time by making sure that by providing a more accurate -- a faster, more accurate indication of when the rinsing process is complete. And they can also help measure more accurately while accommodating higher process temperatures required for some of these tools. Some of these tools will require hot UPW, UPW that has been heated to 50 degrees C, 60 degrees C, even sometimes as high as 80 degrees C in order to make the process more efficient. With the improved temperature measurement circuit, temperature compensation and thermal isolation of the UPW UniCond, these higher temperatures can be accommodated with minimal effect to the measurement quality. The third place where these sensors come and provide value is in the reclaim reuse decision free that we discussed earlier. It can help provide a more accurate picture of whether water should be sent to wastewater treatment or diverted to reclaim processes. And if they are diverted to reclaim, they can provide a much clearer picture on when it's suitable to divert water to reuse in applications like cooling towers, boilers, heat exchangers, that sort of thing or if it's clean enough to be diverted back through the treatment process or recycled and reused in the wafer fabrication process. Using the UPW UniCond at these measurement points will give the user a much more accurate picture of what the water quality coming into the decision point is and help them reduce the amount of water sent to waste as well as making sure that they only send water of sufficient quality to various reuse and recycle streams. So to summarize, industry trends like artificial intelligence and various automotive trends as well as commoditization -- further commoditization of consumer electronics have led to increasing needs for complex chip design that are driving scrutiny or increased scrutiny on UPW water quality at the fabs. This increased complexity and increased capability in electronics means that the water quality required to meet yield targets is ever increasing. And in order to meet the needs of those or to meet those evolving needs at Mettler-Toledo Thornton, we worked very closely with leaders in the semiconductor manufacturing industry to design [indiscernible] that specifically met those increasing -- those increasing needs, those evolving needs within semiconductor UPW. And the result is the UPW UniCond, which provides best-in-class stability to help cut through noise to see actual water quality and remove the effects of various environmental factors like electromagnetic interference or ambient temperature fluctuations or process temperature fluctuations in order to provide the most accurate possible measurement of resistivity in a manufacturers or semiconductor manufacturers water system. In addition to providing leading analytics with the UPW UniCond, the UPW UniCond is part of a broader portfolio that is very well focused on the semiconductor industry. So with additional measurements like sodium, silica, low PPB, TOC, optical dissolved oxygen, pH, online microbial detection, the Mettler-Toledo Thornton portfolio is very well suited to help semiconductor manufacturers meet all of their needs for water quality monitoring at point of distribution, point of use, reclaim reuse and even further upstream and some applications for things like RO and makeup water treatment. So again, the UPW UniCond, with all of its advances is just the latest piece of that broad Mettler-Toledo portfolio to meet those industry needs. With that, I've gone about 45 minutes. So we have about 15 minutes left for questions. I'll be happy to answer any questions from the group.

Thanks, Joel. That was very informative presentation. If anybody's got any questions for Joel can enter them into the chat or feel free to unmute yourself and ask Joel directly. And while we may be waiting for some questions, Joel, can I ask, are there any considerations that you need to make as to how a resistivity sensor should be installed?

So this is intended to be an online measurement. The ideal situation for us is to install the sensor directly into a pipe. However, it is also very possible to install the sensor on a side stream using a flow housing. Again, we just want to make sure that for installed out of flow housing, that the flow to the flow housing is sufficient to provide at least 1 foot per second flow velocity through the sensor. And that's just to make sure that flow doesn't stop and water doesn't become standing on the sensor. That's the only requirement really in terms of installation.

Any questions from anybody for Joel, before we wrap up? There's a question.

Hello, Joel. Yes, I have a question because most customers chose the NPT installation in UPW, but we have the Tri-Clamp installation, so what's the difference between these 2 installations?

So Tri-Clamp installations are typically -- again, as you mentioned, they're relatively atypical within the semiconductor industry. They're used just for an increased level of hygiene really. The Tri-Clamp process fitting is a hygienic process fitting that's typically used very extensively within the pharmaceutical industry. But in addition to those properties, Tri-Clamp fittings are generally more well suited to high-temperature and high-pressure applications. In terms of the effects on accuracy or the effectiveness of removing noise or ambient temperature effects or process temperature effects from the measurement, there's really no difference in sensor performance between the Tri-Clamp and the NPT sensor. It's really just about the process conditions and customer preference for the process fitting. Like I mentioned, the Tri-Clamp fitting is a hygienic fitting. It's also more well suited to high-pressure and high-temperature applications.

Thank you. Anybody else with a question for Joel? It looks like that's a no. So I just like to thank everybody for attending today. Feel free to contact us if you have any questions about our portfolio on the new UPW UniCond. Thank you to Joel for the presentation and have a good day from me and bye for now.

Thank you, everybody. Thanks for coming.