In our inaugural episode, we speak to Parth Shah, PhD, a software engineer and former process engineer at Intel about the incredible processes behind microprocessor manufacturing, his love of all things aerospace, research in plasma technology and how his seemingly disconnected experiences weaved into a crucial breakthrough during the design of a new fab line for the next-gen microprocessors for Intel.
Parth Shah:It's probably one that always like, comes to my mind and always, in a way haunts me and also fascinates me at the same time. [background music]
Siddhit: Pashi presents the Means of Production, a podcast about what it really takes to build, maintain, and scale the processes that produce the physical products that power our world. Every episode, we ask a manufacturing expert to walk us through the nuts and bolts of how they do their job. We explore how and why they got into manufacturing, dive deep into the hardest problems they've solved on production lines, and discuss their thoughts on what's broken in manufacturing today and how those things can be fixed.
SiddhitThis podcast is hosted by Siddhit Sanghavi, Pashi's US Manufacturing Operations Lead, and former assembly engineer at Ford Motor Company.
Siddhit: And now season one, episode one. Our guest is a software engineer and former process engineer at Intel, and also a PhD holder in plasma technology. Please welcome Parth Shah., but before we start a very quick disclaimer from our guest.
Parth: I work at Intel, but this is my own opinion, and it's not the opinion of Intel Corp. I am not the spokesperson for Intel official or otherwise.
Siddhit: I'm very happy that you are our first guest in this debut episode of the Means Of Production, welcome. So we've known each other for a very long time, right? Since University Of Mumbai and we've both came here, we both basically gone into technology fields and then after your PhD work, you've been at Intel for a while where you've done a lot of process engineering for Intel microchips. And since then, we haven't really discussed like what it is that you exactly do and all the work that you did over there. So firstly, just tell us what you do right now and what you did so far at Intel and we go from there.
Parth: Yeah, sure first of all, thanks for inviting me to this podcast and thanks for the introduction as well. So let me elaborate on that introduction that you gave me, so let me give you a bit of my background before I start going on about my work at Intel. So my background in undergrad was production engineering, which is like manufacturing engineering then I didn't really like that much. So I wanted to focus on the thermal and the fluid sciences part of the study that we did in production engineering. So that inspired me to go to Masters in aerospace engineering because my background was so different. I had to take a lot of prerequisites in introduction to flight and introduction to fluid dynamics, and even some material science courses before I can be proficient enough for aerospace engineering. And when I was in my Masters, I realized that I really liked how rockets work and the propulsion part of it. So I thought of doing a PhD and in jet engines or rockets and stuff, so that kind of made me apply to a lot of schools where the propulsion and or combustion where their main forte. So my PhD thesis was in combustion, but it was not just combustion and I had to wear all sorts of hats.
So my, my thesis was actually plasma assisted combustion where basically, we looked at how ions and electrons interact with hydrocarbon flames and how they interact with nanoparticles that are formed in flames. And once I was done with my PhD, I wanted to find a job in aerospace industry but due to visa restriction, I couldn't find one. So I found a post-op, which was kind of close to my work, basically it was close to the principles that govern the plasma assisted combustion with applications and semiconductor industry. So I found a post-op position and Minnesota where I looked at how nanoparticles formed in plasma environment can help with depositing Silicon nanoparticles on a subscript. I got a job as a process engineer at Intel, and as a process engineer, my job was to work in a team where we either process development on plasma edge patterning layers for Intel's next generation microprocessors. So I was working on a plasma edge tool, so again the plasma edge tool is nothing but a tool which has a chamber which runs at like a very low pressure and you ignite a plasma using helium or nitrogen gas. And when you mix that with some other carrier gas, like CF4 and the plasma that is generated helps in the removal of material from the Silicon wafer to make patterns on the wafer.
So basically our job was to pretty much find ways to manufacture these chips and focus on these plasma edge layers in the metal zero, metal one side. And also the other part of my job was to sustain the plasma edge tools, so now these tools that we work on are like millions of dollars. The tool I was working on was around $50 million, so every engineer was given a task to be responsible for one tool 24 seven. So that helped me understand how a plasma edge tool works and how to interact and collaborate with the vendors that provide us with the tools in order to help us troubleshooting them. So whenever, even if it's like 3:00 AM on a Saturday night or Sunday morning, something happens to the tool, like we have to somehow find ways to troubleshoot it and have it fixed. So basically, there were two parts of my job, one was processed development part where we worked on the wafers and the second one was maintaining the tools. I think it was a good experience and I did that for about three, three and a half years and after that I wanted to reconnect with my computational background. So I wanted to transition to something on the computation side. So that's why I found a job, which is more software focused where we create design rules, very cool design rules for layout designers. I can elaborate, well that's pretty much a short or a long like summary of my journey from my undergrad to wherever I am today.
Siddhit So before we move on Parth, can you give us an overview of how microprocessors are manufactured and the different steps involved?
Parth: Yeah, of course. So let me actually start with the broad overview of how microprocessors are made in a place like Intel or Global Boundaries or Samsung or any big company that makes microprocessors on a very large scale. So what goes into these microprocessors? You need basically two major components, one is like an extremely clean environment which is also known as clean room to avoid dust particles from the outside environment and then the other major component is called, specialized equipment for three basic functions. The three basic functions are central deposition, patterning, and edging. So lithography comes under patterning, so to illustrate how microprocessors are made let's talk a little bit more about how integrated circuits are also called, as ICs are made on what these are. So these ICS integrated circuits are made up of structures called transistors and these transistors need to be connected, right? To make a complete circuit and you need like metal lines called interconnects, which connect these transistors to complete these circuits. Now these transistors are extremely small, with features sizes that are like tens of nanometers could be microns, but nowadays, like it's in 5, 10, 20, 30, 40, 50 nanometers structures. So basically when you have transistors of such scales the way it helps us is that it allows us to integrate a large number of transistors into a very tiny space to create a very powerful computer. Okay, so in order to make sure that these transistors even work, which are so small, even a small piece of dust must not land on a chip. So a major development in these processes in these labs are to avoid these dust particles to land on the chip. So these dust particles can be from the outside environment, but most of these dust particles actually come from within the chamber. So I talk a little bit more about that later on, so to give you a perspective of how tiny these transistors are and how widespread these are in 2009, 2010 there were 10 quintillion transistors that were manufactured, okay.
Parth: Yeah, which is 250 times the number of rice grains eaten worldwide in that same year.
Siddhit: Oh, my God!
Parth: Yeah, and the interesting part is like the price to produce a single for that same price we can produce about like 125,000 transistors. So you can try to estimate or try to guess like how small these transistors are and how many transistors are being manufactured like every second in these labs.
Siddhit: And also, how efficient and standardized the process has become to be able to beat the production cost of rice, right? So that's great.
Parth: I think this is more like an anecdotal kind of example, I found this somewhere online so I thought it's like a cool thing to talk about and to just give a good perspective of how tiny these transistors are and how the entire effort goes towards making them even tinier every day and every year just to follow the Moore's law. So these IC materials are basically of two types, metals that are conductive and dielectrics that are not conducted. So the metals and the dielectric that we use to fabricate transistors and integrated circuits are very small and thin, like the thickness is a hundred nanometers or less and think about human hair, which is about a hundred microns thick. So the thickness of these fins or these devices are at least like thousand times smaller than the thickness of a human hair. And to give you another perspective, that's like comparing the height of a statue of Liberty to the height of your standard smartphone or any mobile device that you have.
Siddhit: Oh, wow! That is blowing my mind, I thought you were going to say statue of Liberty with a human but this is a more stark comparison, wow, go on please.
Parth: Yeah, so now let's come down to the three equipment that perform the three basic functions which I would repeat, first thin film deposition, second would be patterning and third would be edging, let me start with thin film deposition. So thin film deposition is usually the first step after they get Intel or any semiconductor fab company gets into the process of chip making. So they get the first like circular wafer in the form of Silicon and it's the stickiness of that Silicon wafer is about a couple of millimeters thick, and this Silicon is also known as substrate. So the first step is usually thin film deposition, and that involves governing a substrate, which is Silicon in this case with another thin material and substrate is in this case, is the Silicon wafer or something equivalent that you can hold with your tweezers. So the aspect ratio of that substrate is very high, like the diameter would be let's say 12 inches which is around 300 millimeters and the thickness would be like three millimeters. So the aspect ratio is like one is 200 and you can make transistors in the substrate, but now you need to make electrical connections between the transistors. So what you'll do on the substrate, is first you'll deposit a thin uniform metal layer on the substrate and then pattern it using something like photolithography . I talked more about photolithography a little bit later.
Siddhit: So what metal is this, is that common knowledge? What metal is it or is that secret or proprietary or something?
Parth: I mean, nowadays it's like all fancy materials, like which are usually secretive. But it could be something as simple as like copper or yeah, usually copper. I'm mostly see a lot of copper so I don't want to, it's usually copper because it's like highly conductive, that's why they use copper. So to a wild particulate matter dust particles the devices are not only manufactured in clean rooms, but since they are extremely sensitive to the chemicals or gases or even the chamber that they resided, they are processed under extreme vacuum conditions because air itself can be reactive to these processes and air itself can be a problem. So to give you another example a lot of these processes are conducted at like, let's say 10 or 20 millitorr and atmospheric pressure is 760 torrs . So this is like less than a thousand of an atmospheric pressure. So it's an extreme vacuum conditions, but because it's under extreme vacuum conditions like these tools are extremely sophisticated, so they need to go from very high pressure to very low pressure. So they have a lot of equipment within the tools in order to supplement this transition.
Siddhit: And Parth, sorry to interrupt, but how many units of torr did you say this was? You said very low.
Parth: Yeah, 20 millitorr.
Siddhit:Millitorr, okay so just for the perspective of the audience, one torr is 0.0013 standard atmosphere, and this is like one thousand of a torr, 1 millitorr, so that's extremely low.
Parth: Yeah, so this is a thin film deposition, so now let's move on to the fun part, which I like which is called lithography. So lithography, there are many types but the most widespread technique of lithography is called photolithography. Okay, so it's exactly how you develop a photograph where you expose a certain part of the wafer to light and the remaining part of the wafer will be covered. You'll cover the wafer with a stencil resist called photoresist. So the part that is exposed to the light can be washed over later on, and what remains on the wafer is the stuff that was covered under the stencil, which is also called photoresist. Basically, lithography is like a particular wavelength of light, which interacts with that special material that's been coated on the substrate. So you have the appropriate light that reacts with that material and the second important part is the kind of the stencil, the photoresist you need. So the qualities you need with this photoresist, is first of all it should not react with the light because you just want the material away from that stencil to react with the light and it has to be also very precise because that's the shape that you want after all on the substrate. So the issues that you run in lithography is like, as you go to smaller and smaller feature sizes the wavelength of the light kind of starts mattering. Because the wavelength of the light, like if you remember from your basic physics, the standard wavelength is like as small as like 10 nanometers, 20 nanometers, a hundred nanometers, and the feature sizes that we're trying to nowadays create have the same wavelength.
So if you have wavelength that is bigger than the feature size, then you basically cannot create that pattern. So in order to work around that problem, like they are nowadays trying to develop a similar patterning process using EUV, which is lithography, but using extreme ultraviolet lithography. I will talk more about it later on because I'm not really familiar with it right now. But and up until now, when UV wasn't available what they would use to do is rely a lot on edging. So you would create like big patterns because traditional lithography would have size limitations and those patterns then would translate down through the film using plasma edging also called dry edging and specialty about plasma edging is like, we use plasmas which are like charged gases in order to remove material from the substrate and then you remove like material through plasma. Most of the times you get structures that are in the form of tapered holes. So from large sizes on the top, you can taper it down to smaller sizes using plasma edging. So as you get smaller and smaller without UV, you'll have to rely a lot on plasma edging in order to create like smaller feature scales as compared to UV which gets rid of that problem. And again, edging which means removal of material, it relies on lot on charged gases and charged gases means like they react with the metals in the chamber, which means like it creates a lot of dust particles and other defects on the wafer. So even though plasma edging kind of resolves that issue, but it's not the best way to work around it because like we have to develop processes in order to mitigate the dust particular issue. So EUV is like the Holy grail towards developing the next generation microprocessors because it solves a lot of issues when it comes to processing and even though EUV tools, like they cost like hundreds of millions of dollars they would eventually prove cost effective when it comes to patterning wafers. So that was kind of like a brief overview of how chips are made. But of course, like I just gave you like an overview but in reality, there are like thousands of different processes that take place in order to make a chip and another fun fact is like, it takes at least 45 days to make one chip, which is longer than to make an aircraft, for example, like 747. So since chip making relies a lot on chemical processing, the fabs run at a 24 seven pacemakers because like if you leave some lead time between two processes it can cost Intel money, because wafers can get oxidized for example, or again, have some undesirable characteristics because you left sometime between two processes. So a lot of effort goes into reducing the time between two processes and that is done through automation and through robots.
Siddhit: hat's fantastic Parth and very deeply technical also. So let me try and attempt to kind of summarize a little bit of what you said, and I'll still probably get something to wrong. So when the chips, the microprocessors that have to be created, get so small, the wavelength of regular light can be at the same size and at that point you cannot create these microprocessors. Are we using ultraviolet light because their wavelength is smaller than the rest of the spectrum? Is that the reason?
Parth: Yeah, that's a good question, I think that is one of the reasons, but again, like, I'm not that familiar with it in order to give you like the exact picture of why exactly we are using it.
Siddhit: Okay, I was just wondering, because there's UV and then there's x-ray so if that explanation is correct then in the future, we might see this being done with x-rays because they're even smaller than, than you realize.
Parth: Yeah, but I think like x-rays don't react with a lot of materials, that's why you use x-rays to scan your bones, right? Because they just pass through your skin, so a lot of photochemical reaction. Even though they are wavelengths are much smaller, there's still other issues that come with it.
Siddhit: I see
Parth:The other problem would be the obvious one is to deal with the safety because x-rays can be, can create a lot of safety issues.
Siddhit: Yeah, agreed, I think what you're saying points to the combination of its ability to react with certain materials, plus it's wavelength not just as it's variant, so you're right about that. Now with the pattern, what I'm understanding from you is that the pattern is a lot like how you would want to design, some people in India would call it like a 'rangoli' like you would drop this colored powder onto the floor and where it does not fall through the holes, it's not coming on the substrate and then the light is reacting with what is not covered in this template and that is what gets edged, if I'm using the correct term.
Parth: No, etching is the following process.
Parth:They use those chemicals to wash it away.
Siddhit: I see and a lot of the difficulties that were being faced by this process in terms of the size, in terms of the photo sensitivity, in terms of the dust or whatever it is we'll hopefully solved or are being solved by EUV .
Parth: Yes, correct.
Siddhit: And can you again, explain why how plasma comes into this picture?
Parth: So of course, so stencil, which is like photolithography, it takes care of the patterns across the wafer, right? But plasma edging is used to create like holes that goes laterally into the wafer. Like, do you see the difference between the stencils and between going across the wafer and through the wafer?
Parth: So that's the major difference between lithography and plasma etching and the way plasma is solving the issue is like, let's say lithography, like left of pattern of 50 nanometer hole and there is no way for lithography to go lower than that. So what plasma does is it removes materials, but as it's removing material into the substrate or into the thin-film, that's been deposited it kind of tapers down like it's not an exact vertical hole, so it kind of gets the taper into the hole. So if the top dimension is like, let's say 15 nanometers, you can end up with a dimension which is like 25 nanometers when you reach to the end of the hole, like a tunnel into the thin film. So that's how plasma edging is solving lithography problem is by creating a taper so that you get holes that are smaller inside. So let's think about plasma doing a job of tapering down from 50 to 20 nanometers then it's followed by something called as weightage or something called as clean. So they, again like wash the wafer with some X chemical and that chemical washes away the top part of the hole and all you remain with is a smaller hole down in the bottom.
Siddhit : Yeah, I do see, so it's kind of like making a hole in a two layered cake and then the top layer of the cake is like eaten. So the hole, which was 50 nanometers earlier might be like 35 nanometers or something and then go onto 25 at the bottom. So yeah, it kind of lets you go to the third dimension of depth of that material, whereas photo lithography is only surface.
Siddhit: Right, so yeah, I kind of understand that more and I hope the audiences also understood, but I will leave articles and I'll leave links to articles and some PowerPoints that I find explaining this process as best as I can. Yeah, but thank you Parth, I think that was a really great explanation.
Parth: Oh, thank you.
Siddhit: I think you did answer part, or quite a bit of the first question, which is what made you get into the field in the first place. But is there anything you'd like to add about why it is that you were interested in aerospace and what made you get into that in the first place?
Parth: Of course, Oh I'd love to do that, that's the best part. So since I was a kid just like any other kid who gets influenced by NASA. I always love like space and planes and in India at that point of time, like when I was applying for my undergrad, there weren't too many schools offering engineering in aerospace or aeronautical sciences and the other thing was, I was pretty much like an average student at that point. So I wasn't like good enough for IIT or I wasn't even like knowledgeable enough about like what exists outside of what people around me talk about.
Siddhit: Right, you don't know what you don't know.
Parth: Exactly, yeah so, I don't know what exactly. So I didn't know the dots enough in order for me to connect basically, I only used to keep hearing about mechanical, electronics or computer engineering. So I had one of these choices and I realized after my physics class, in my high school, that most likely I would like the mechanical engineering field. So I picked something like production engineering because it was closer to mechanical engineering and the other reason was like, the school was pretty much at a walking distance from my house. So there were like multiple factors, which made me choose production engineering when I was an undergrad. But what made me choose aerospace engineering is again, like my love for planes and rockets and also, as I was doing my undergrad, like I really liked the terminal engineering and fluid mechanics courses. While I was doing aerospace engineering, I really loved the propulsion part, the propulsion classes which where classes like rocket propulsion and also combustion and some other classes like incompressible, fluid mechanics. So that's what really got me more and more interested towards getting a better understanding of how rockets work and with that side like, I felt like I wasn't yet good enough just with a Masters, I need to really get deeper and deeper with my study and hence I decided to do a PhD and I went for a PhD at NC State. So that's how my journey walled from like production engineering to propulsion, to plasmas.
Siddhit: It's very interesting and very, I guess very circuitous also, but I guess one thing that is a peculiar is that you started with production engineering and I was your batchmate, but you also ended up in a production environment at Intel. So you can say that you were able to combine your production base with your plasma tech, deep dive to give you a good combination of skills, which probably, that is a rare combination to get right. Knowing how to manufacture plus knowing plasma tech so that would be very niche.
Siddhit:o that's fantastic, so in this time when you were designing the process at Intel for the microprocessors, what was the hardest problem you faced and how did you solve it?
Parth: So there were a couple of them, but it's probably one that always comes to my mind and I'm always like in a way haunts me and also fascinates me at the same time. So as you know, like the nodes in semiconductor industry are getting smaller and smaller and smaller, like every two years as per the Moore's law. So I was working on the 10 nanometer node.
Siddhit: Sorry to interrupt you, but can you just tell like our listeners, what a node is?
Parth: Oh yeah, definitely. Actually, node is nothing but a name of a product, they call it three nanometers, it's in the form of three nanometers, five nanometers, 10 nanometers. Like I would say years ago, the nodes actually meant a physical dimension on a chip, like 24 nanometer would mean that there is a certain pitch length of the transistors on the chip. But nowadays the nodes don't really mean anything, it just means the next process node that a company is manufacturing. So right now 10 nanometer doesn't mean like there is an actual 10 nanometer physical dimension on the node, it just means it's the next product that Intel, or maybe DSMC is coming out with once they're done with the previous node, which was 14 nanometers. So just to summarize like nodes don't really mean anything, it probably has to do with little bit with the density transistor density on a chip, but right now it pretty much doesn't mean anything, it just means the next generation product that a company is coming out with.
Siddhit: Okay, got you. Go on.
Parth: So back to the problem so in these nodes like for example, I was working on the 10 nanometer node, which means there are densities and structures on the chip that are like smaller than the previous nodes. So as we go smaller and smaller the transistors kind of start interacting with light, they start interacting with dust particles. So imagine if you have a transistor that's like, let's say a hundred nanometers and a dust particle is 10 nanometers. So it doesn't really affect the hundred nanometer transistor because the transistor size is too big for the dust particles. But now imagine like the transistors and the densities are shrunk down by 10 folds. So now the transistors on the chip are comparable to the size of the dust particles. So now these dust particles scan heavily influenced the performance of the chip, like one desk particle can cause like major defects on a chip. So as we go to smaller and smaller dimensions controlling dust inside a tool where the chips are manufactured and process are really important. And there's not just like one or few processes when it comes to manufacturing chips, there are like thousands of manufacturing processes when it comes to chip. Like every chip takes anywhere from like 45 days to 90 days to manufacture from the first day it's a Silicon to the chip that we get on our computer, it takes anywhere from 45 to 90 days.
So every process is really important, like when it comes to controlling dust particles and these dust particles are not just the particles that we find in the environment, there are dust particles that are formed inside the tool as well. Because you're working at like really low pressures and very high pressure gradients and you're also striking a plasma, which is basically high energy electrons, like hitting like all parts of the plasma chamber and they interact with the chamber and that itself can create a lot of dust particles. So basically, one of our most important jobs is to reduce this defect on the chip and when I was working on a chip there was this issue that just wouldn't go away for like almost a year and I was assigned with that problem along with a couple of team members of how to mitigate that issue. And so the way the semiconductor chip manufacturing companies work is like they get a tool from the vendor and we have to design a process on that tool. Like there is little to no room of modifying the vendors hardware per se in order for us to make the process work. So we have to get whatever tool we are given and do whatever process modifications we can do in terms of like process knobs, like temperature, pressure, high voltage, and there are like tons these knobs and we have to find like the best condition so that it doesn't give us that defect.
So now we kept working on this problem and it was resolved, it was going nowhere and we were just banging our heads. And what I suggested is like, okay why don't we just have a meeting with the vendor, I know it's like not traditional and it's unconventional to ask for them to make modifications to their hardware, but let's find out if it's actually possible. So we arranged a meeting with the vendor and we sat with them and they stole us the same thing, we cannot change their hardware, this and that, blah, blah, blah. Like you have to deal with the process yourself and we can help you with the process development, but there is no way to change the tooling or any hardware on the tool. Because that would mean like they have to change the whole supply chain or whatever on their manufacturing side. But I kept bugging them and I did my own research as to like what these parts that go into the chamber are made of like what material they're made of. I didn't have like access to any simulation software so I had to like brought groups that could help us, like with simulating all of these conditions with that material. So I had to sit with them, I had to sit with a vendor and show the vendor a lot of these simulations and prove to them, like what's the source of these particles that are appearing in our chamber but there's no way for us to mitigate them unless they change the hardware. So after a lot of back and forth, after a lot of data and simulations the vendor agreed to test out some new hardware on the tool and in terms of like, as little as screws.
So we change from one material to some another alloy and we tried and tested a lot of these and eventually after like three months or so, we see some mitigation in dust particles when they make a certain type of modification. So I would say that was the most challenging part to someone who doesn't work in the fab, like it may appear that it's just a dust particle, but it's because it's so small, it's like such a headache to us. And like after kind of solving that problem, like we were able create this new tradition sort of were under extreme or extenuating circumstances we can ask the vendor to help us with modifying the parts. So this one problem, like helped us to change the company culture in a certain way, because earlier where her hands were tied because of the one policy, but now if you cannot get the process running, we can resort to this last option. So it's something that I feel proud about, of course it was not just me it was like a lot of other people who are involved but that's like one of the biggest problems that were solved in the plasma etch group.
Siddhit: Yeah, so the solution to this problem not only needed, in-depth technical know-how about the interference of dust particles and their effect on the process, but also a great deal of project management and coordination, right? So it really sounds like a unique and at the same time, very common manufacturing problem because you have design groups, you have vendor groups, you have production groups and they can't always talk to each other because they are talking in different languages. And I think here you and your teammates kind of brought them together it looks like, and kind of changed the way they communicate with each other. So kudos on that, I think it was a great incident you described. So it is something to be proud of, especially if it changes a little bit of how the company conducts its business.
So, well the last question I guess, is if you had a magic wand to change one thing about how your works and this could be the job that you described or a previous job you had or anything like that, what would it be and obviously, this would be within reason but, go ahead, what would that be?
Parth: I would say, one thing I would change with a magic wand is the company culture when it comes to manufacturing. Like Intel being a very old company still I would say resorts to some old school techniques, like having passed down in the morning at 7:30 AM, which means like process engineers and technicians have to be in the fab gowned up by 7:40 AM, I think. Basically, which means that you have to get up at like 6:00 AM if you live like 30 minutes away and get ready within a very short period of time and go to work get there by 7:40 AM in the fab. So I feel that's a pretty old school way of thinking, because now with newer technology, we can have those things like anywhere as long as you have a computer and obviously, all of us have computers. So I would suggest that, and like most of these semiconductor companies, they need to start resorting to new age techniques and they should stop or kind of like transition from old school thinking to do the new era thinking where you have technology and science to help you with your work-life balance. There's no need to still stick and adhere to these old school routines. I think COVID has kind of indicated that it is possible to work from home even as a process engineer, work very effectively and I think Intel has seen a great boost in productivity in people, especially process engineers because of COVID. So I think like Intel might eventually get rid of that requirement of being in a fab, like early in the morning, which just doesn't make sense to me. Other than that, I would say a couple of things like again, Intel has this very secretive like work culture, like even within groups, even between groups, there is not a lot of collaboration and there's a lot of IP restrictions within Intel groups. Forget about like collaborating with other companies or with universities or schools when it comes to developing chips. So I feel like that probably kind of proves to be harmful when it comes to the long-term success of a company, like having too many IP restrictions. So that's something that I feel if I had that magic wand, I would change that.
Siddhit: Awesome, yeah thank you so much.
Siddhit: Yeah, I think what you touched upon is something many industries face and the auto industry is no different. And you're certainly right, COVID also impacted some of those and what you are referring to, I guess, is what many people call like the industrial era calendar, right? Like you have to get inside a big warehouse at a certain time and you can get out only at a certain time. But there's so many new technologies like AR and VR and what you just mentioned, just remote connection to the hardware that might make this much easier if not eliminate it, or at least make the lives of process engineers easier.
Siddhit: And I guess this is my plug for Pashi that Pashi is browser-based so you design the whole process, you add machines on this visual drag and drop editor, and you just say, okay, this is the first machine of this process, this is the second machine, third machine. You write the logic and you can operate it from that browser, so once the hardware is connected after that, it is possible to run Pashi from anywhere. And also, look at production analytics from anywhere, which is a lot of what you might also be doing, looking at the results and looking at the numbers that are generated by the machine. So yeah, I think that's how things are going and it should open up thank you for the answers. I think you touched upon all the critical problems that you would face as a process engineer and also your magic wand answer was quite on topic.
I guess before we close, I want to ask you like a fun question which I didn't prepare you for because I wanted you to just think about it on the spot, which is in 2051, if your grandchild were to step into a factory what would they see or you as an older person step into a factory, what would you see?
Parth: Of course, I would expect a lot of things to be automated. I would hope to see a lot of robots doing the work of a process engineer. So I know about a lot of companies like Glam research, applied materials, they are working on a lot of projects where basically they're trying to use machine learning AI to indirectly like replace a process engineer. So as engineers, like we have had to collect a lot of data manually, in the sense that we had to go to the fab, like remove wafer, cut a chip out of a wafer, and then look at it under a microscope and then manually measure like the different dimensions using SEM or TEM , which is like electron microscopy. So that entails a lot of like manual work and whenever you involve a human, of course there's a lot of human errors of it. So the accuracy and precision goes down if you rely too much on engineers, right? So basically, all these companies are working on ways to automate all of this and use deep learning techniques for image recognition and pattern recognition.
So I feel that's what I'm going to expect, like if I walk into a fab, I'm going to see a lot less engineers and I'm going to see a lot of stuff that has traditionally been run by engineers be completely replaced by machines. And not only that, for example every few weeks technicians have to perform preventative maintenance on our tools and that involves a lot of like manual labor and it may sometimes create a lot of injury or also sometimes leads to defects within the wafers. Because when it comes to cleaning, also there is like a very comprehensive aspect that every technician has to follow, one step you're in there and bring the temperature down after a couple of minutes. So you have to be careful at every step, so I feel a lot of these things will also be replaced by robots and according to my understanding, companies like Intel are already working on it to replace potential hazardous work by an actual robot. So I feel in 2051, I'm going to see a lot of these jobs being replaced by AI not just AI, robots, or some machine.
Siddhit: Yeah, that is the hope many of us have that we can reduce errors, reduce fatigue, and have these robots and yeah, that is a trend going on. And hopefully that feasibility increases more than what it is right now of very sensitive processes, like the one you described.
Parth: Right. Right.
Siddhit: So, yeah that's a great answer, thank you Parth so much for coming on the podcast and sharing your experiences and your answers about your career. I I'm sure it will not only be interesting to hear in general, but also like young aeronautics engineers or young semiconductor manufacturing engineers would find it very, very useful, and interesting. So yeah, I hope to maybe see you again on this podcast and you will see this on our website very soon. So thank you so much and have a great day.
Parth: You too, thanks for having me.
Outro : [background music] If you enjoyed this conversation, please subscribe to the Means Of Production Podcast, for more stories from people behind all the manufactured goods we use, love, and depend on. This episode was made possible by Pashi the operating system for manufacturing. Pashi unifies the entire production process for any product, encompassing operator instruction and data interfaces, stage logic, and parameter thresholding, machine interfacing and configuration support programming and coordination and stage to stage production flow control into a single Pashi program. Check us out at pashi.com and until we meet again, have a fantastic day, and take care.
Music from Uppbeat: "Falling" by Zayner. License code: SYFMAQN9WWBAPCSU