Shortcutting Graduates’ Path To Productivity In Manufacturing And Test

Semiconductor companies are working with universities to custom-build engineering curricula so new hires can hit the ground running.

popularity

Manufacturing, test, assembly, and analytics companies are finding unique ways to engage with universities in an effort to shore up the talent pipeline.

The industry is recruiting graduates from universities across the U.S. while partnering with local institutes to serve specific needs. Industry/university co-operation includes:

  • Mapping job descriptions
  • Providing curricula frameworks
  • Co-creating classes
  • Running labs
  • Gamifying analytics
  • Hiring interns, co-ops, and part-time master’s students
Company Annual graduate hires Typical degree level, major hired Example universities Training Methods, Resources, Partnerships
Advantest About 40 since 2021 BS, MS, some PhD; usually EE San Jose State, Arizona State, many In-house Year-1 Onboarding program; co-produces classes with universities, conducts labs for their curriculum
Amkor About 500 for proposed OSAT in Ariz. Primarily BS, MS but PhDs also in the mix; semiconductor, electrical, chemical, mechanical, material science, EE Arizona State, Grand Canyon, UT Dallas or Austin, Purdue, others In-house training, internships, apprenticeships, partnerships with community colleges, universities
Brewer Science Not mentioned Science, mainly chemistry; engineering Multiple, local and national Internships, co-ops
DR Yield A handful BS or MS in CS, informatics, telematics, EE TU Graz Hire after BS and students work part-time while doing MS then hired full-time; local partnership to bring women into tech
Intel Many BS, MS, PhD but a shortage at MS, PhD; about 50% need sponsorship Top science, engineering schools across U.S. and those close to facilities like ASU Internships, apprenticeships, certifications, investments, donating equipment
PDF Solutions 20 to 30 globally; hired 4 from a 24- student class at Carnegie Mellon U. 75% MS/PhD, 25% BS; EE/ECE, CS, data science, AI/ML Carnegie Mellon, UC Berkeley and other UCs, Stanford, MIT, many more Internal training, conference resources, industry-led class on analytics for IC manufacturing
Teradyne 40-50 BS, some MS, some PhD. Engineering and Technical (EE, mechanical, Computer Engineering); some hires in Supply Chain & Operations UMass, Amherst and Lowell, Wentworth (WIT), Worcester (WPI), Cal. Polytechnic, Cal. State, HBCUs including Rutgers,
ASU, Purdue, Texas Universities
Teradyne Learns internal training; co-ops, internships; Global Semiconductor Alliance; Women’s Leadership Initiative; Northeast Microelectronics Coalition; China


[Key terms: Bachelor of Science (BS), Master of Science (MS), Electrical engineering (EE), computer engineering (CE), computer sciences (CS), Master of Business Administration (MBA). Note, this is the second article in a series of three. The first focused on design and AI companies. The third will present the university perspective.]

Greenfield facility needs
When dealing with a greenfield manufacturing site, a company needs people to fill a wide range of jobs from construction, including piping and electrical work, to production, including different types and levels of technicians and engineers.

Amkor is slated to receive CHIPS Act funding for a new site in Arizona and has a global system and team for setting up facilities, which has been used in Portugal and across Asia. The Arizona area has quickly gained a ready supply of construction workers needed by the other semiconductor facilities completed and being built in the Phoenix area. The company is working with local community colleges to develop technicians, which is where the company sees the largest gap. Amkor is also reaching down to the high school level to develop programs to encourage STEM education. As for engineers, it is turning to local universities to help fill the future gap.

“We need a fairly large engineering workforce, and we feel pretty good about securing those positions,” said David McCann, senior vice president and business units chief of staff at Amkor Technology. “We’re working with ASU, Purdue, and Grand Canyon University on mapping our job descriptions into their curricula. ASU has about 30,000 students in their engineering programs, the largest engineering school in the country, so the supply will be strong. The challenge for us is gaining name recognition. Unless you’re in our industry, you likely have not heard of Amkor. You also likely do not know what an Outsourced Semiconductor Assembly and Test company (OSAT) is.”

To build awareness, Amkor hosts industry education events to talk about the supply chain and where it fits in, and invites other companies to join and excite people about manufacturing in the U.S. “We give people something to hold in their hand and say, ‘This can go into an AI server or phone’ to gain their understanding and interest. We want to get them excited about our business segment and manufacturing.”

Amkor is aiming to get in front of ASU, Grand Canyon, and Purdue students on the right timeline for them to be hired when they graduate. “If we are too early, meaning before we are ready to hire, we could disappointment students. If we are too late, we could impact our schedule.”

ASU is ready to meet the challenge. “At this point in time, these fabs are in the construction phase,” said Binil Starly, professor and director of the School of Manufacturing Systems and Networks within ASU’s schools of engineering. “But then they’re also slowly transitioning into people who would be trained to operate these fabs.”

Multiple engineering disciplines are needed:

  • Manufacturing engineers who are going to have to integrate with the machines that are connected in there;
  • Mechanical engineers who have to design and operate these machines and understand it at the mechanics or a physics level;
  • Chemical engineers, because there are gases and chemicals involved, and mixing of them;
  • Environmental engineers, because these are nasty chemicals that need to be managed correctly;
  • Cybersecurity majors. Many students do not realize how much data is being generated by the machines in these fabs and you need a robust, secure framework for collecting this data;
  • IT and software majors, because it’s not like your typical data that is generated in a Google IT-like environment, and the IT folks have to be entrenched with the process, chemical, and manufacturing engineers.
  • Material science engineers and electrical engineers.

Government, companies, and universities all shaping curricula
In light of the CHIPS Act, the government is telling universities, educational institutions, and workforce paths of all types to get ready and focus on the semiconductor industry, said Sam Rosen, who leads the regulatory strategy for Teradyne. “They’re coming to the companies looking for advice on, ‘What are the skills you need? What’s the curriculum that needs to be taught? How do we gain this technical focus?’”

The key is to acknowledge that, yes, curriculum may need to be modernized to meet changes in the technology landscape within industry. But with change landscape keeping the curriculum focused on the product fundamentals with an emphasis on skill-based continuous improvement, this will be a critical point of how we think about shaping the early career workforce to be more successful for when in role, said Ryan Moran, talent acquisition and Center of Excellence lead at Teradyne.

“We’re also preparing for when or if those curriculums are adjusted,” Moran said. “How we can better prepare students for industry? We want to pick that baton up with new hires by first assessing what are the remaining gaps in their knowledge, and then start to equally look into our talent build strategy to continue the development rather than just trying to acquire the perfect talent match. To do this, we’ve had deeper partnerships with educational partners and industry affiliations to grow a workforce and expose all a parties at different levels.”

Teradyne’s training department is building asynchronous training modules that are three- to eight-minutes long. “This is the sweet spot for training,” said Rosen. “And then we couple that with lab time, which has resulted in interest from the U.S. universities.”

Its university curriculum has classes on fundamentals, maintenance, programming, and advanced topics, and each of these is multiplied by its different test platforms.


Fig. 1: A sample university curriculum for test. Source: Teradyne

“As we build our new learning management system, the question is, ‘How do you take sections of training and create labs that are applicable to real-world applications?’” said Rosen. “You need to separate the fundamentals from the application-specific information, and then figure out how to integrate these activities to ensure successful knowledge transfer.”

Running labs and co-producing classes
Companies can help universities by providing input into curricula, and can even take on some of their workload. In Advantest’s case, the company co-produces classes with Arizona State and San Jose State and runs their labs.

“Our model has been to align with a standard course curriculum,” said Mark Abate, senior director U.S. applications and training at Advantest. “We’ll identify a professor teaching a certain class, and we’ll talk to that professor with their department chair and say, ‘You have the curriculum. You teach the lectures, and then you can come to our facility. Or, we’ll put a tester on your campus and our staff will administer all the labs. We’ll take care of designing the labs, the hardware, the software, the lab manuals, and they can offload that burden right to us.”

Through this involvement, Advantest gets exposure to the universities’ students and staff, and it is expanding its reach by identifying regions with workforce needs or by listening to its customers. For example, in the Arizona State program, members from NXP joined the Advantest team as lab assistants to work with the students. And in Texas, NXP and TI requested the company partner with particular universities to meet their workforce needs.

“They said, ‘Listen, with these massive ramps going on, we’re struggling to find capable staff. Please help us get this message into the university,’” said Abate. “We also help train students so when companies get them, they can hit the ground running. They like to know people coming out of universities have not only awareness, but some experience. That’s why we promote a hands-on environment versus, say, a virtual learning environment. We want people to come in, touch the equipment, touch the load boards, work in a real lab environment, so they know what that workplace will look like before they get there. Once they get there, they’re very confident and ready to learn.”

Finding courses on test and measurement is a challenge, but Advantest is in discussion with ASU about a class aimed at RF device testing, and San Jose State is offering a mixed-signal design and test class at the undergraduate level. “We’re intercepting those folks in their junior year, and it’s been a wonderful experience so far for everybody,” said Abate.

Once hired by Advantest, a graduate or new hire goes through a rigorous training process.

“We call it year-one onboarding,” said Abate, though it may take longer. “Technical training is done mostly face-to-face, hands-on, with supplemental online training that’s AI driven, focused on test, technology, products, design, methods, and quality. The second stage we call joining the home team, like their account team. We give them additional focused training, assign mentors, and begin their software skills assessment and training. The third phase we call the bigger picture, and that’s when they learn the roles of the other account team members, such as sales, system development, R&D, field service, etc., and how their role as an application engineer interacts with these account team roles.”

Pizza, cash, and the appeal of AI/ML to lure students into IC analytics
Like test and measurement, there is a lack of classes on IC analytics compared to classes on semiconductor devices and processes, according to Andrzej Strojwas, CTO at PDF Solutions and formerly a professor of ECE at Carnegie Mellon University.

“Where nobody was really filling the gap is how to connect the processing to the data analytics that needs to respond to the huge amounts of data available, instrumentation, and so on,” he said. “Only a small fraction of the data is used. A lot of it is thrown away. Right now, universities are offering classes in cloud computing, AI, machine learning, and so on. But on the example of Carnegie Mellon, 95% of the students will go and work for hyperscalers and have nothing to do with the semiconductor industry.”

With this in mind, Strojwas and his team came up with a class focused on advanced analytics and ML aimed at processing IC manufacturing data. Students form teams and have to solve real problems. The professors used pizza to entice students to the informational seminar and offered financial awards for the winners. The goal is to scale the class across the country with data and tasks relevant to local universities and partner companies.

For example, one group was asked to identify the source of variations for the critical dimension of the gate, Strojwas said in a presentation. Students were given access to PDF’s Exensio platform on a private AWS cloud, but not all of its capabilities. They got data from two sources — one from PDF via a Virtual IDM database based on historical data standardized from previous customers, and the other from Intel, which supplied up to two hundred gigabytes of real data sets sanitized for privacy purposes. Intel also provided guest lecturers.


Fig. 2: A data analytics class at Carnegie Mellon University in partnership with PDF Solutions and with data from Intel. Source: PDF Solutions

The classes have been a big hit, and others agree that prizes, real-life challenges, games, quizzes, AR/VR, and other GenZ-friendly tools could be key when it comes to attracting students.

“Gamification of all types of learning has been increasing over the past four years because we are highly remote now,” said Danielle Ferguson-Macklin, DEI program manager at  Teradyne. “Humans learn better when there’s a game or quiz assigned to it, no matter how you deploy it. There is also the rise in the leveraging of virtual reality and augmented reality to immerse folks in the experience.”

When hiring for new technology such as AI/ML, the key is to attract people with the right soft skills.

“It’s not reasonable to assume that you’re going to find somebody with generative AI experience for multi years, when it’s only been in place less than two,” said Cindi Harper, vice president of talent planning and acquisition at Intel. “What we look for are certain attributes of continuous innovation, like they’re fearless, trying new things, troubleshooting, problem solving. They have enough knowledge of circuit design and semiconductor and can take that to the next level through trial and error, and then we train them from within. We try to stay up with the curriculum so that it continuously evolves.”

In-house training paramount when skills aren’t taught at university
Generally speaking, U.S. universities are more amenable to adapting curricula to meet industry’s workforce needs. Europe takes a view that universities should operate independently with soft cooperation only.

“I know that universities in Germany, for instance, are even more reluctant to do this than here in Austria,” said Dieter Rathei, CEO, DR Yield. “They always feel like business should be separated from their curricula and a student’s academic career should be pure, and engaging with companies might compromise that. Unfortunately, that means that the curricula that you would need at the universities are usually lagging behind in the adoption of the skill sets needed for the current labor market.”

Even if a university was willing to adapt its curricula, knowledge and skills can quickly become obsolete.

“Semiconductor technology is evolving fast and, theoretically, universities could keep up with that base,” said Rathei. “But we are in the field of IT services for the semiconductor industry, and things are even developing faster. So a good approach is where the university should lay the foundations and the framework of thinking and problem solving, and we do the onboarding and teach the required skill sets that’s needed.”

Others agree. Even though the Netherlands’s Eindhoven University of Technology was founded on request of the industry, local companies such as ASML have no direct influence on its curriculum. Most engagement is through guest lectures and, even then, the scope is set by the university.

“Industry is very deep in our DNA and in our thoughts and our way of operating, but always in soft cooperation modes,” said Paul Koenraad, a member of the Photonics and Semiconductor NanoPhysics group in the Department of Applied Physics at TU/e, and dean of the TU/e Graduate School. “I’m absolutely sure that our University Council, and also many people in the Netherlands, are very keen that such a direct connection between industry and universities is not happening. And another reason why I personally, as dean, have a great problem with that is I know that the students have a choice to make after they leave, and if, say, ASML would say, ‘Train all your people for our company,’ but then they are not going to hire all of them, then people are left in a very bad spot.”

Koenraad’s colleague agrees. “ASML said explicitly they don’t want that,” said Martijn Heck, professor in the Photonic Integration Group in the EE department at TU/e. “When I had to go to the advisory board with a new institute, they really didn’t care about all the fancy new things and all the upcoming markets. They said, ‘We just need good people. Keep educating them in the way you want.’”

In this context, streamlined onboarding systems become crucial. “There are no experts in the field of yield software for the semiconductor industry available on the market here, so we have to train them ourselves,” said DR Yield’s Rathei. New hires typically work part-time while gaining their master’s and then are hired full-time. Ideally, they visit a fab test floor sometime in their first year. “It’s very important that people here know they are not working on an abstract concept, so they are seeing the greater good of what they are enabling at the end.”

Internships versus co-ops
Internships and co-ops give candidates hands-on experience and a chance to see if a job is a good fit while building relationships. Meanwhile, companies get a chance to see if they want to hire a person full-time, instead of recruiting through resumes and interviews only.

In semiconductor manufacturing, a key factor is whether a candidate is going to be comfortable working in the very tightly controlled environment and shows enthusiasm, said Rachel Jung, director of strategic partnerships at Brewer Science. The company needs universities to teach the foundation well, and then it is prepared to train interns and co-ops on its very specific chemical processes.

Key differences between the programs:

  • Internships are usually paid and are often about 20 hours a week while a student is still studying or on summer break.
  • Co-ops are always paid and are typically 40 hours when a student has taken one or two semesters off from studying.
  • An internship usually requires course credits and a grade, perhaps even fees.
  • Co-op employment is a paid position and you’re essentially an employee with the company.

Teradyne has a robust internship and co-op program and now aims to meet students even earlier. “If we only start engaging with universities when we have positions coming available, we’re behind the eight ball,” said Moran. “We want to be ahead of that curve. So we are thinking about how to engage the student body through not just career events, but other program avenues to nurture a pathway into the semiconductor industry. Then, as students come into these intern and co-op positions, we’ve already invested in a development process.”

A crucial element is the feedback loop between student, company, and university, which can help ramp time to productivity and provide a better experience.

“We do internal engagements to really understand what our interns and co-ops are experiencing on campus,” said Teradyne’s Ferguson-Macklin. “But how does a university trust us enough to take our feedback? We must cultivate the relationships. If you’re industry and you’re not on campus – participating in their judging events, networking sessions, industry nights – none of the feedback that you give can really take hold because there’s no one there to nurture it.”

Conclusion
The semiconductor supply chain will need to use every tool in the box and include every sector of the community to solve the talent shortage. In this context, hiring is often based on a great attitude rather than just a degree.

“We tell engineering students that a fab environment is a very disciplined environment, in the sense that certain things have to happen or else things mess up,” said ASU’s Starly. “That regimen, and that protocol following, is such an important skill. In our classes and engineering, we do have that stringent discipline, and this is why companies come to hire you. They don’t come because you have a bachelor’s degree in mechanical engineering. They come because you’ve gone through the rigor and through that discipline of following instructions and protocols, and that can’t be taught fast. That’s built up over time, even from the high school level onward.”

Related Reading
Chip Companies Play Bigger Role In Shaping University Curricula
Design and AI companies are using a range of tools to help graduates become productive more quickly. Some are feeding their requirements directly to university.



Leave a Reply


(Note: This name will be displayed publicly)