Hello, this is Chris Mack, and today I'm going to give a talk that's a little different than most other talks that I have given over the years. I joined the semiconductor industry in 1983, so that's over 40 years ago. So as 2023 comes to a close, I decided to give a little retrospective of that career and maybe even impart a few lessons that I may have learned along the way. So what does a 40-year career in the semiconductor industry look like?
Well, this is what it looks like in terms of my headshots over those years, and you can see that I've undergone a little bit of a change. Notice the last headshot is 2011. About 10 years ago I decided to stop aging, so in fact that is the last headshot I'll probably ever use. But the more traditional way of telling your story in a career is with the career bio. Now don't read this.
They're all the same. They say basically the same thing. I went to college here.
I graduated there. I got this job. I worked on this. I got these accomplishments, maybe these awards. And all of these things led in a very linear way to the most important thing that I've ever done.
which is what I'm doing right now. That's the way all these bios read. And the nature of how we write them implies this linear progression from the beginning to the end.
That's really not what most people's career looks like. And that's not what my career has looked like. It's more like a random walk. So what I'd like to do is talk about the randomness. that has happened throughout my career and what it has meant to me.
Let's begin with some advice. Well, I'm going to end with some advice, but what I'd like to do is start with the kind of advice that you've probably all heard. It may be the most famous, the most common career advice is the kind of thing you get at the commencement speech, follow your passion. So, Steve Jobs.
Famous 2005 commencement speech says, you know, if you're doing what you love, then it's never really work. But how does this idea of following your passion jibe with my experience of a career which is more of a random walk? How are you following anything if most of the things that have happened, the inflection points in your career are all somewhat random?
So I'd like to try to. See how we can understand these two things together. So let's step through that bio beginning with my Education I went to a small school in Indiana called Rose-Hulman Institute of Technology.
I love that school. It was great for me It's a beautiful campus. This is what it looks like today.
It wasn't quite as beautiful 40 years ago, but still really nice and I got four degrees from there. And if you look at these degrees, you would think they are the perfect set of technologies in order to understand lithography, which is what I spent my career working on. You would think there was a plan, but no, there was no such thing. I went to college and I didn't really know what I wanted to do. I liked physics, I liked chemistry, and I couldn't decide which major to make, so I kept putting off the decision.
So one semester would go by and I'd say, well, next semester I'll decide. And then the next semester would come and I'd say, well, I won't decide yet. I'll decide later.
And so I just kept doing both, pursuing both areas and never did decide, which turned out to work out really well for me. Well, then if we look a little further, you see I get a master's degree in 1989. But hold on a second. That's a seven-year gap.
You know, there's a lot of stuff that you can read in a bio based on the things that are not said. So what is this missing piece, this time frame that's not accounted for? Well, when I graduated, I immediately went to graduate school.
I went to Caltech. My goal was to get a Ph.D., to become a professor, to do research, and to teach. Well, that did not work out because after one semester, I dropped out. That's right, I'm a Caltech dropout.
There was a girl involved. But I'd I really just wasn't ready to go to graduate school at that time. I needed a break. I needed to do something different. So all of my plans, which I had carefully thought about, to get my PhD and become a professor, those plans didn't work out.
Well, turns out the girl didn't work out either. But it did result in my first job at the National Security Agency. That might seem like an odd choice of a career.
Well, I had worked there over the summer just before going off to graduate school. So it was kind of an obvious place for me to go back to because I needed a job. So I got a job at the National Security Agency. Now, this is an important piece. To get a job at the National Security Agency, you need a top-secret special intelligence clearance.
It is the highest clearance of any agency. We laughed at those piddly clearances they gave away at the CIA. We had real clearances at NSA.
And it takes about a year, nine months to a year, to get your clearance. So when I interviewed for a job and was offered a job in a new, brand new microelectronics research organization, My boss thought I would be showing up in nine months to a year, but I already had my clearance because I worked there over the summer and I showed up the next week. She had no idea what to do with me. I didn't have a desk.
I didn't have a chair. I didn't have any responsibilities. Nobody knew what I was going to be doing.
They just weren't prepared. Well, another engineer in the office felt sorry for me and he said, go out on the loading dock. There's a box, a piece of equipment that he had just ordered, and he wanted me to take it apart, put it together, and read the manual.
What was in that box? A piece of lithography equipment. The only reason I spent my career doing lithography was because this box happened to be on the loading dock when I arrived in need of something to do.
It could have been anything, and I'd be doing something else. There was no plan involved. But this was just the beginning because we were building out a new lab, and this new lab required...
lithography expertise. We bought a stepper, a real manufacturing piece of lithography equipment, but that lab actually took many years to build and be ready to install that stepper. So, I had learned about lithography. And this is when I discovered a series of papers written by Rick Dill and coworkers at IBM called the Dill Papers from 1975. And these papers changed my life.
When I read them, I knew this is what I wanted to do. This was the set of science works that transitioned lithography from an art to a science by applying mathematical equations to describe exactly how lithography works from the optical imaging to the exposure and development of the photoresist. And that leads then to the next piece of this equation after this bio.
After reading those papers, I decided the best way for me me to learn about lithography was to write my own lithography simulator called ProLith. The positive resist optical lithography model. From 1985 to 1989, while I worked at the NSA, I developed this software and I decided to give it away.
I gave it away for free. I gave my first paper on the topic in 1985. I told anybody in the audience that wanted this software, I'd give them a floppy disk and... They could go and install it and run it.
And hundreds of people took me up on the offer. Well, by 1989, the lab had been finished, the stepper had been installed, and my boss told me, time to stop working on this lithography simulation stuff and go to work in the lab. Problem is, I really like this lithography simulation stuff.
So I decided to do something else, to quit this job and to form my own company. And from 1990 to 2000, I ran Findlay Technologies. We commercialized the software. And it turns out that by giving away the software for free, it meant it was in the public domain. I didn't think about it at the time, but it was the perfect thing to do because now anybody, myself included, could take that source code and turn it into a product.
So that's... What we did for 10 years, our company grew and was quite successful before being acquired in the year 2000. But that's not quite it, because also on this bio, you'll see that I got my Ph.D. in chemical engineering from the University of Texas in 1998. Now, you work through those numbers and you realize that I got my Ph.D. while... I was running my company. Well, that was not an easy thing to do, especially at a prestigious university like the University of Texas. And I couldn't have done it except for one man, Grant Wilson.
Grant Wilson is probably the... most important, smartest resist chemist in the world. Here he is with his wife, Debbie, and the emperor of Japan.
When he was awarded the Japan prize, he's been awarded pretty much every prize there is. And he welcomed me into that department at the University of Texas and said, he would work around my schedule, my commitments running Finley Technologies and pursue... the research that I needed to do to get my PhD.
My PhD was on modeling solvent effects in optical lithography, and in case you were tempted to ever look up that PhD thesis and read it, well, I'll save you a lot of trouble. Here's the summary rendered as a haiku. When heat is applied, solvent flows through polymer.
Your model is wrong. I wrote that at a bar the night after I defended my thesis. Now in 2000, I sold my company to Finley Technologies. I'm sorry, I sold Finley Technologies to KLA Tinko, now KLA.
And as a part of KLA for five years as Vice President of Lithography Technology. my focus began to shift. KLA is a metrology and inspection company.
They're deeply involved in the measurement and interpretation of data. And data has always been a part of my world in modeling because we had to calibrate these models. But the focus of what I did started to center around data, and data has become an important thread through all aspects of my career. I think it's an important thread through most people's careers in the world of science and technology, but I especially spent a lot of time thinking about data. Well, in 2005, my first child was born and I decided to quit my job at KLA.
I needed more time and so I started focusing my career intentions on a website, lithoguru.com. It's a little bit strange to think that my career was in developing my website, but it was all about putting all of my interests, my knowledge, the things I had been working on, and the things I continued to work on up for public consumption. on this website. I called myself a gentleman scientist because I no longer needed a wage-paying job to pursue my interests in science.
So I worked on the gentleman side, writing essays. working on lots of things, but much of what I did was in the writing area. And the first thing I did in 2005 was to work on my textbook, Fundamental Principles of Optical Lithography.
which was published at the end of 2007. I had been working on this book for a long time, but in 2005 I realized that every year I was getting further away from finishing it. That's because the technology of lithography was changing faster than I was actually keeping up with my writing. That's a really bad situation. So I basically worked full-time for two years and did nothing but write this book, got it finished, and it's one of the highlights of my career.
I've also written a few other things. My most recent book, I'm also very proud of, How to Write a Good Scientific Paper, was born of my experiences as the editor-in-chief of a scientific peer-reviewed journal called Jam 3. And this book, by the way, is amazing. available for free thanks to the generosity of the publisher SPIE. Now I didn't really give up on my desire and my dream of being a teacher and in fact I taught part-time at the University of Texas for over 25 years. I ended up Teaching a wide range of courses in a large number of departments.
I taught as a visiting scholar for a semester at the University of Notre Dame and another semester at the University of Canterbury in New Zealand. But most of my time was spent teaching at the University of Texas. And it was great.
I had a wonderful time over those 25 years. And in the last few years, I became really interested in the concept of the flipped classroom. And so the last two classes that I taught, chemical processes for micro and nanofabrication, and my data analysis class from data to decisions, I flipped those classrooms. What does that mean? It means I recorded all my lectures, I required the students to watch the lectures before they came to class, and then we spent our class time discussing, working problems, going through all of those things to cement that understanding in the students.
It's called the flipped classroom and I really enjoyed doing that. But I needed a way of delivering these lectures to my students and the obvious easy and free way to do that was YouTube. So I created a YouTube channel and posted my lectures and at the end of that first semester in 2013, exactly 10 years ago, I had 23 subscribers because I had 23 students in my class.
10 years later, I now have 23,000 subscribers. I still don't know how that happened. People just started discovering these lectures.
I thought maybe a couple of other people might find them useful, but I did them just for my students in my classes. But they've become quite popular. I've got three different classes, two full semester classes with all the online material, plus a review on statistics and on my YouTube channel, maybe a couple of music videos that I've made as well.
Now, all of this is going on. And you see there was no plan to any of it, but it worked out quite well. Another thing that happened when I was writing my textbook back in that 2005-2007 time frame, I discovered a topic.
that fascinated me called stochastics. Stochastics is the fundamental randomness that occurs at the molecular or atomic level. At that level, every event, every process is probably Probabilistic, not deterministic. Well, I had spent my whole career looking at the world from a deterministic perspective. That is Proleth, the modeling software I wrote, encoded deterministic equations.
If you have certain outputs, you get certain inputs. Excuse me. If you have certain inputs, you get certain outputs every time.
But in real life, the output of every experiment has some randomness associated with it. That's stochastics. And in lithography, that can result in roughness of the edges of the features or variation in the dimensions of the features to the point where sometimes features can even be completely missing.
And this is due just to stochastic variations. So I became fascinated with understanding this phenomenon. And as I studied it, I realized that this was going to be the most important thing. important thing going on in lithography and in patterning in the semiconductor industry for this reason.
We are a victim of our own success in this industry. Moore's Law. We have been shrinking the feature size approximately 12% every year for over 50 years. Now that shrinking has slowed a bit, quite a bit, lately.
But nonetheless, you take this incremental improvement and you do it every year. for 50 years, it's really quite dramatic. We've had multiple three orders of magnitude reduction in the feature sizes we use to make our chips from the beginning of the integrated circuit era. Well, every one of those features has had stochastic variability.
Every one of the edges we print is rough. Every one of the dimensions we print has variability due to the fundamental randomness that occurs on that molecular level. It's just that normally, over most of the history, not normally, but over most of the history of the semiconductor industry, that variation has been insignificant compared to other things going on, other sources of errors, and insignificant compared to the feature sizes we were printing.
That is no longer the true. We've been shrinking. The feature size so much that now stochastic variations are the dominant source of variation in the patterns that we print. And as of today, they are the biggest headache in semiconductor manufacturing when it comes to patterning. So, turns out, understanding stochastics, which I started working on 15 years ago in earnest, is one of the most important things we could be doing to move forward with Moore's Law.
One of the reasons that this is true is stochastic scaling is quite different than other scaling. What do I mean by scaling? Well, errors need to scale.
If we shrink a feature, we need to shrink the errors in the package. patterning of that feature in proportion, or maybe a little faster because of how things get harder when features get smaller. But stochastic variability is far more than a linear scaling. it is a volumetric scaling.
If I shrink the volume of material within which stochastic events are occurring, the variability goes as one over the volume. And the volume in our case, in patterning, is the critical dimension, the feature size that we're trying to print. So the variability is scaling as one over the feature size cubed rather than the more conventional one over the feature size or a little bit more than that.
And as a result, Stochastic variations are getting worse much faster than the other errors that we had been used to controlling over the last 50 years. So the control of stochastics is becoming harder and harder every time we shrink our feature sizes. So this really became clear to me it would be a control problem.
And control starts with data. The fundamental... rule of process control is you can't control what you can't measure.
So when I started looking at data to understand the problem of stochastics, I began to doubt the data very much. Most of the data is based on the use of a scanning electron microscope to measure the features. And so here's an example image with some features on it, and we need to measure the roughness.
But the SEM images are very noisy, and the noise in the image looks a lot like stochastic. variations and it's hard to separate those two out. The conventional way of dealing with noisy data is to average, take more data points, and the the mean will have an uncertainty that goes down as 1 over the square root of the number of measurements that you make. But that's not true when what you're trying to measure is the variability. You can't just average away this kind of noise because it's contaminating the actual roughness.
So a new approach had to be used. taken and I realized that the existing approaches simply did not work so I invented a new one where I could use modeling of the scanning electron microscope instrument to figure out where the edges were in order to avoid the problem of noise in the sim. So you see the convergence here my experience in modeling immediately made me think about modeling the instrument of the sim.
understanding of the data and stochastics made me understand what would be required to get a good answer. Well, a little over seven years ago I realized that this invention was going to be worth something, so I started a company. I teamed up with my old business partner Ed Cherrier and the two of us founded Fractilia, the computational metrology company.
The combination of all these things that I had been doing in my career became the perfect background for what I am now, which is a metrologist, someone who focuses on the science of measurement. Wow, so this is what I'm doing right now. It's a lot of fun. It's very exciting. It's a lot of work.
And if you think about all of the things that happen to me in this career, I am perfectly positioned. to be doing what I am now. But none of it was a plan. My career did not follow a plan. However, here's a couple of my favorite quotes that will explain what happened to me and what I think happens to many people.
Eisenhower, when talking about life in the battlefield, said, Plans are useless, but planning is indispensable. Basically, as soon as you hit the battlefield, all your plans go out the window. But without that planning, you wouldn't be prepared for what happens to you. And likewise, Louis Pasteur said, When describing observations in science, chance favors the prepared mind.
These two things are a core of my philosophy of being prepared to take advantage of the randomness that happens in your life. So back to that advice. Should you follow your passion? Well, that implies a couple of things.
One, you only have one passion. And two, that you can follow it, that you can chart your own course, go in this straight line along the path that is your passion. But that, in my experience, is unlikely to work. Maybe somebody will be lucky, and that will exactly be what happens to them. But for me, the twists and turns in my career...
were often a result of circumstances outside of my control. So my advice is a little bit different. Whatever path you find yourself on, find something to be passionate about.
Do the right thing. Do something important. And be passionate about it because there is something on your path to be passionate about. And if there's not, that's when you have to change paths. But I think there's plenty.
In almost any career, any field, certainly in the science and technology world, there's plenty to be passionate about. And this has turned out to be the way it's worked for me, and I hope the way it can work for pretty much anyone. Well, I appreciate you indulging me on my reminiscing over the last 40 years of my career, and thank you.