Wednesday, September 4, 2013

Mass flow controllers and finding one's path

I went to school for eleventy billion years and ended up with a bachelors, masters and Ph.D in electrical engineering (EE). I often wondered along the way if I should have switched majors at some point.

It probably happens to everyone, but as an undergrad I took a lot of required classes I wasn't really into. I discovered I was much more interested in learning how the individual semiconductor components worked and were fabricated than designing circuits that used them, or analyzing signals that controlled them.

NMOS and PMOS logic circuits were kind of interesting... but how they were made and worked at the fundamental level was WAY more interesting.


The nice thing about graduate school is you can delve into whatever little niche that interests you. For my PhD work, I specialized in metal-organic chemical vapor deposition (MOCVD) in the fabrication of laser diodes on nonpolar crystal orientation of GaN. This incorporated a little EE, and a lot of materials, chemistry, and solid state physics. It was a great way for me to steer away from my circuits-heavy background towards the fundamental physics and fabrication of electronic devices.

And yes, telling people I made lasers was so cool.

MOCVD, also called metal-organic vapour phase epitaxy (MOVPE), is a deposition technique for thin films and semiconductors. It's a big machine that involves metal-containing molecules called precursors which are routed through various gas lines to a reaction chamber where the sample, called a substrate, sits at high temperature. The precursors combine with other process gasses in the chamber and, over time, atomic layers of semiconductor material get deposited on the surface. This layer-by-layer deposition, shown below, is also referred crystal growth or crystal epitaxy.

In short, MOCVD is both the gigantic tool and the process I used to "grow" lasers.

Horizontal gas-flow MOCVD system showing growth of Gallium Arsenide.  The precursor chemicals containing Ga and As flow into the chamber and react on the surface of a hot substrate, leaving behind Ga and As atoms to incorporate into the perfectly crystalline film, while the rest of the precursor leaves through the exhaust.  Growth entails a continuous flow of precursors for a specified amount of time, depending on the rate of deposition and how thick you wanted the layer to be.

To "grow" an electronic device you first define all the layers in the structure you want, then you write recipes on a computer that controls the the MOCVD system specifying which precursor chemicals to use, when to turn them on, and how much of them should flow into the chamber at a given time. Each layer gets a little of this and a little of that and the amounts of each are finely calibrated.


Schematic diagram of a MOCVD system with two process gasses (hydrogen and nitrogen) and six precursors, with individual valves and mass flow controllers that control what and when to flow into the reactor. 


When you work in deposition equipment like I do, you get really familiar with mass flow controllers or MFCs. These small but necessary instruments are used to control how much process gas or precursor flows through the lines to the reactor. As a recipe writer and tool operator, one of the parameters you specify is the flow rate of each precursor through their respective MFC and when to turn them on or off. They direct traffic so to speak.


Relationship between velocity (v_bar) and flow rate, Q which is measured in units standard cubic centimeters per minute, sccm (pronounced "skim") or standard liters per minute, slm (pronounced "slim"). 

Just like I loved learning about the inner-workings of electronic devices as an undergrad, I love learning about the inner workings of the pieces of equipment used to make them. MFCs are a tiny part of a giant deposition system like MOCVD, but alone they are pretty interesting.

Here's a cool video from Sierra Instruments that discusses how MFCs actually do their thing.



Isn't that neat? They even use MFCs to make beer! Yay, beer!

In my new job, I'm basically a recipe writer and tool operator for a similar CVD-type system. My group specializes in the deposition of thin film dielectrics, which is used as a single layer in semiconductor devices. MFCs are still really important.

I'm still learning a lot and getting used to the fast-pace of industry work. Most of my coworkers come from backgrounds in chemistry, physics, and materials science. My EE background sets me apart in some ways and sometimes I worry I lack enough relevant background and have some catching up to do in that regard. Whenever I feel especially stupid, I remind myself: I have a Ph.D. in engineering. I'm not stupid. I should be able to figure it out. 

Hopefully, one day my knowledge of undergraduate level circuits will be an advantage... Perhaps in meetings with customers when they mention gates or FETs and I know exactly what they're talking about?  Honestly I have no idea. I may have a Ph.D. in engineering and a long term full time job that I like, but I still don't know what I want to be when I grow up.

So far, though, every path I take seems to lead me in the right direction.

The many paths of electrical charge and Kirchoff's current law.