I’ve been informed by the punctuation police that I am no longer allowed to employ exclamation marks in my column titles. That’s like asking me to write while hopping up and down on one leg with one arm tied behind my back, but I’ll try to be brave and soldier on. Suffice it to say that for anyone building a semiconductor fab who wants to save time and money while reducing risk, this column is for you!!! (You’re welcome!!!)

As I’ve mentioned before, I’m incredibly lucky. We obviously need not make mention of my outrageously good looks, sparkling wit, and internationally noted sense of style (during our daily FaceTime calls, my dear old mom, who lives in the UK, often remarks: “What on Earth are you wearing? I hope you’re not going out dressed like that!”). What I’m talking about here is the fact that I get to meet and greet many interesting people who introduce me to technological concepts I’d never even considered before.

As a case in point, I was just chatting with Brian Culhane, who is the CEO at Basesite. Brian boggled my brain, and now I hope to boggle yours, so this would be a good time for you to don appropriate clothing. However, before we plunge into the fray with gusto and abandon (and aplomb, of course), let’s first remind ourselves of a few salient points, starting with the fact that our modern civilization is wellnigh totally dependent on semiconductors, so we would be up the proverbial creek without a paddle if we found ourselves unable to obtain these devices.

As part of the COVID-19 pandemic, we all became painfully aware of the potential for supply chain disruption. Things aren’t helped by rogue nations who have already attacked or who may be tempted to strike other countries. According to The Economist, for example, “Taiwan produces over 60% of the world’s semiconductors and over 90% of the most advanced ones.” So, can you imagine the problems that would ensue if Taiwan were to be subjugated by an unfriendly power?

This is one reason why the CHIPS and Science Act—which was signed into law by President Joe Biden on 9 August 2022—is such good news. This act authorizes roughly $280 billion in new funding to boost domestic research and manufacturing of semiconductors in the United States. According to an article on Electronics360, “Throughout the U.S., there are 73 new semiconductor fabrication facilities that are planned for construction or are currently under construction.”

Similarly, the European Chips Act is a legislative package intended to encourage semiconductor production in the European Union. And other countries are cognizant of the need to ramp up their own semiconductor fab capabilities. As noted by the Australian Strategic Policy Institute (ASPI), for example: “ASPI’s 2022 report, Australia’s Semiconductor National Moonshot, laid out the strategic reasons why Australia must embark on a capacity-building initiative to create a homegrown semiconductor manufacturing ecosystem. Every item on the Australian federal government’s List of Critical Technologies in the National Interest is dependent on semiconductors.”

In a crunchy nutshell, all this means that a lot of companies, countries, and organizations are either already building—or are poised to start building—semiconductor fab facilities. I wonder if any of them would be interested in speeding the process while saving vast amounts of money. Unfortunately, when it comes to semiconductor space (where no one can hear you scream), I don’t know anyone who dons the undergarments of authority and strides the corridors of power, so—if you do rub shoulders with anyone like this—please feel free to point them at this column.

What comes into your mind when you hear someone say the word “tools”? Different people will have different responses depending on their engineering discipline. In my case, my knee-jerk reaction is to think of things like my long-nosed pliers, my wire cutters, and my soldering iron. By comparison, in the case of the people tasked with building semiconductor fabs, “tools” may refer to machines like the Endura Clover from Applied Materials (I introduced this machine in my Metrology Maketh the Man (and Woman, and Silicon Chip) column).

Applied Materials’ Endura Clover mainframe (gray) with cassettes (orange) and a collection of
wafer processing chambers (blue) (Source: Applied Materials, Inc.)

A fab may easily have say 2,000 of these “tools.” This is where I must admit to being a bit shortsighted, because I’ve typically considered only fabs that have already been established. As part of this, I’ve written columns like Boosting Semiconductor Fab Productivity by up to 20 Percent, Accelerating and Reducing the Cost of Semiconductor Process Development, and Want to Create Silicon Chips? Want Them Fast? Want them Cheap?

What I hadn’t thought about were tasks like deciding where all these machines should be physically planted in relation to each other to optimize the production flow. At an even “lower level” (conceptually speaking), I had never considered what it took to actually power these machines.  

This is where Brian started to boggle my brain, starting with the fact that fabs can consume hundreds of megawatts of power, thereby requiring them to have their own electrical substation(s). Also, it turns out that each “tool” can demand multiple different supplies, each with its own voltage and current requirements. Some supplies will be DC and others will be AC. Some of the AC supplies will be single-phase while others will be three-phase, and so on and so forth. As a result, we are talking about tens of thousands of connections being handled by thousands of power panels that are distributed around the fab. Just for giggles and grins, things are often in flux, especially in the early stages of the fab’s foundation.

Now, let’s take a brief aside to discuss Brian’s role in all of this. In 2003, he graduated with a degree in Electrical Engineering from University College Cork (UCC), which is the National University of Ireland. Brian’s first job was with a project management company, but he ended up as a gopher, doing whatever he was told to do without any real responsibilities. After five or so years of this, Brian did what any of us would be tempted to do in his situation—he took a year off, including six months cycling (also known as bicycling or biking) from Thailand back to Ireland.

Upon Brian’s return to Ireland, after pottering around in different positions, he moved from Cork to Dublin where a semiconductor fab was being built. In 2013, Brian was given all the responsibility he had previously craved when he became the lead engineer on the electrical integration team (also known as “tool hook-up”) for the latest tool installation project on the site.

I was going to preface this next bit by saying “believe it or not,” followed by telling you that—at that time—at least in this particular domain, the semiconductor industry used Excel spreadsheets to manage everything. However, I fear this would be redundant because we’ve all “been there” and “done that” in our own domains.

Having vast amounts of data being modified and passed back and forth between different groups whilst being represented in Excel spreadsheets is prone to risk, especially when there are multiple copies (revisions) floating around. Thus, Brian wrote the spec for a program to assist with the electrical pre-assignments (tool hook-up). Initially, this was developed by a contract programmer who was assigned to the project. However, due to the need for constant updates coupled with frustration with the turnaround times when using the developer, Brian developed the software skills necessary to support the application himself.

To cut a long story short (which is opposite to the way I usually do things), the use of this tool exploded, first by other members of Brian’s team, and then by other disciplines on the project, including electrical, mechanical, LSS (life safety services), and communications. It wasn’t long before Brian’s code had 500+ on-site users spanning everything from design to construction.

It also wasn’t long before the application started to be used in a second fab, which was being built in Israel. However, this application wasn’t seen by the company Brian worked for as being part of their core business. Happily, the company gave Brian permission to leave, to take his application with him, and to supply these services directly to the semiconductor fabs, all of which happened in 2018.

Now we find ourselves in 2024, and things are racing along. The latest incarnation of Basesite’s platform accepts unstructured project data, cleans it, structures it, and performs advanced AI optimization to automate building services design while providing optimal implementation solutions.  Although I can’t name specific semiconductor vendors here, I can say that—using Basesite—one fab saved $20 million on panels alone. Also, gap analysis, which previously took anywhere from 6 to 12 weeks, is reduced to a matter of hours or days.

Although Basesite’s original focus was on semiconductor fabs, the company’s tools are equally applicable to all sorts of deployments, including automobile manufacturers, data centers, pharmaceutical manufacturers, and… well, pretty much any high-volume manufacturing (HVM) facility. As a result, the company—which is already growing in leaps and bounds—is poised to explode (if only I were a younger man).

So, once again I’ve been exposed to a problem (and solution) of which I previously had no knowledge. I love this stuff! How about you? Do you have any thoughts you’d care to share?