The Makers Guide to Microfluidics

Imagine a highway without any lane divides. Cars travel down the road generally all going the same direction, but without any lane lines it can be hard to stay straight. Cars get jostled all around and bump into each other, the flow of traffic is chaotic, and overall things on the road are hard to predict. A microfluidic device is like taking that road and adding lane lines in. Particles and fluids pretty much stay in their lanes while flowing through microfluidic channels, forming parallel areas of flow that are predictable and controllable. This is called laminar flow, and with it we can do some really cool things. Chips can separate microparticles based on their size or shape, analyze incredibly small amounts of fluid (helpful in things like RNA research), or generate microdroplets that contain a single cell.  

Microfluidic chips are usually made in laboratory clean room conditions using relatively complex multi-step fabrication processes that, while easy enough to do in the lab, do not lend themselves well to the home hobbyist. In this guide I will show the process that I found to be best in terms of fabrication of microfluidic chips using tools that could be found in the average makerspace, or above-average home workshop. These steps do not need to be completed in a clean room, and hopefully are within grasp of anyone interested in making their own chips. Many amazing papers and github projects by even more amazing people went into the honing of this method, and I will try to list them all at the end under “Further Reading.”

Tools

• 3D Printer: Preferably resin, though I've successfully used an Ender 3 in a pinch so anything goes
• Milling Machine or Laser Cutter: You will need one or the other, or preferably both. They don't need to be particularly nice machines to still get decent results. a K40 or 3018 CNC would be alight, if a little finicky.
• Hobby Knife: Particularly one with a square end
• Lab Stands: I just use these because we had them already. Anything that can hold the syringe will work fine/better. (Could be a problem to solve with a 3D printer)

Supplies

• 3M 468MP: This is the adhesive sheet that we'll use to bond our chips and seal the ports. It's cheap.
• PVC Aquarium Tubing
• Aquarium tubing on/off valves and 3-way connectors
• Electrical Tape
• Good quality latex balloons
• A bike pump, air pump, perisaultic pump, or good lung function

Prototype microfluidic devices are most commonly fabricated using a process called soft-lithography. In short, chrome or mylar masks are used to etch SU-8 photoresist on silicon wafers in order to create master molds. A castable silicon called PDMS is poured over the mold and then bonded to glass with oxygen plasma. While this process is fast, easy, and effective for properly funded bio labs, it is a little out of reach for the average home tinkerer or makerspace inhabitant. SU-8 and PDMS are pretty expensive substances, and the chemicals needed to process the master molds are pretty nasty. 

Instead of soft lithography we are going to be directly milling our microchannels into polymethyl methacrylate (PMMA, also known as acrylic) using either a desktop milling machine or a CO2 laser. The piece of acrylic with the channels milled into it (the flow layer) is bonded to another piece of acrylic using widely available 3M adhesive sheets or, optionally, PDMS.

In order to interface with the microchannels we will need some ports. I use 3D printed luer-style surface mounts. These have a lot of flexibility depending on your fluid handling system. You can either use high-end medical tubing, luer fittings, and pumps ($$$) or aquarium tubing, balloons, and disposable syringes ($). More on that in step 8.

Microfluidic chip designs are widely varied and range from the simple and understandable to the mind-boggling and trauma inducing. If you have a specific need I recommend searching around some repo’s like metafludics for existing designs and starting there. If you want to play around and start to learn by doing, there is an excellent browser-based tool called 3duf that you can use to quickly design chips. Of course any CAD or vector design software like fusion 360, inkscape, or illustrator can be used as well. 

For the purposes of this tutorial I will be making a simple Y-channel. These are great for testing out fabrication methods, fluid handling systems, and just getting used to microfluidics in general. Y-channels have some actual helpful uses too, like fluid mixing, droplet generation, etc. I’ll link some cool papers on the subject in the “Further reading” section.

I am using my trusty Nomad 883 Pro with 1/64” and 1/32” flat endmills from Carbide3D to do my milling. You do not need to use a machine nearly this nice however. Really impressive results can still be had from using a <$300 “3018” style CNC from aliexpress or ebay, you’ll just have to spend a bit more time on setup for each run. 

1. First I import the design file into carbide create and set the channels to be on one layer, and the outline of the chip and the ports to be on the other.
2. A 1mm thick piece of acrylic will be used for both the flow and cover layers. It's a good idea to get an actual measurement of the thickness of the acrylic with a micrometer or calipers.
3. I start by surfacing a piece of MDF to act as a wasteboard, then apply a layer of blue painters tape to both the acrylic and the MDF. Cyanoacrylate (super glue) is deposited onto the tape-covered wasteboard and the acrylic sheet is pressed gently onto it (tape side down). This superglue + painters tape trick is really handy for workholding, and I use it for pretty much every CNC project I do involving large flat things. 
4. Once the acrylic is on the machine I place a 1/8" pin in the collet and zero it out using a rolling paper (#blazeit). Then swap for the 1/64” endmill and find the new Z-zero with my tool setter. A 1/64" endmill will result in roughly 400um wide channels. 
5. I want the total depth of the channels to be .4mm, so I do two passes at .2mm depth, with .2mm of stepover. For the feed rate I pretty much max out the machine and do 635 mm/Min at 10k RPM.
6. Once the two parts of the chip are milled out, remove them from the wasteboard and set them aside.

I've attached the .SVG file of the VERY basic chip that I am milling here. This can be imported to pretty much any piece of cad/cam software and used to generate the GCODE needed for your own machine (mill OR laser)

If you don’t have a CNC machine or the urge to learn to use one, a CNC CO2 laser can also be used. The goal here will be to set the power and speed of the laser in such a way that it cuts only partially through the acrylic. In my experience each laser is different enough that you should take the time to experiment and find the settings that work best for your specific machine. I usually do this by cutting 5-8 lines of varying power/speed, then cutting out a box and using an inspection microscope to inspect the resulting channels. Ideally you want the most consistent and semi-circular channels possible with as little kerf as possible. Using individual pieces of paper to offset the focus of the laser can result in some really good results. 

Above you'll see a picture of a test chip with 5 channels, ranging from a power of 20% at 20mm/s, gradually increasing to 40mm/s. Note how the box is meant to slice through the channels, this allows for a better look under the microscope.

For reference, in the pictures above I am using a a 80w CO2 laser at 20% power and 40 mm/s. This results in a somewhat triangular channel with some kerf. While this will work okay for simple flow cells, I'd probably want to experiment more with focus offset to try and get a better, less melty channel.

Once you have figured out your “feeds and speeds” you can simply place your acrylic sheet on the laser bed, load your .DXF file into the cutting software and you’re off to the races. Figuring out the right settings for your machine and acrylic is tricky, but once you have them it’s super easy to laser-mill new designs with very little effort on your part. 

Next we will need some way to interface with the chip. There seems to be 1000 methods to do this, and every way sucks in its own special way. I have found that 3D-printing some surface mount luer connectors sucks the least. I use a cheap resin printer (Photon S, $99 at the time of writing) but pretty much any 3D printer will do the job. Even a stock Ender-3 can produce workable parts, you’ll just have to test them all and toss out the ~40% that leak. Print at the highest resolution that you can manage.

I’ve included an STL of the mount I made. There are definitely improvements that could be made here, and ports will have to be tested for leakage, but all in all this is a pretty cost-effective method of interfacing.

I’m using 3M 468MP sheets to both seal the chips and attach the ports. 468MP is just a layer of adhesive between two removable paper layers. These are used in the 3D printing world a lot and are really cheap for the quality. I recommend picking up a big pack of them as you’ll find a ton of uses for this sort of adhesive once you have it in your life. Other sorts of adhesive can be used if sourcing becomes an issue. There are some really high-end medical adhesives that are super sticky and super thin that are available, but you’ll definitely pay out the nose for them.

For actually cutting the adhesive I use a CO2 laser. I simply place the sheet onto the laser bed, load up the outline of my chip and the “port_donut” file and let the laser do it’s thing. Always cut out more than you’ll think you need as adhesives can be a little finicky to work with and you may want spares. I always end up totally messing up at least once :P 

If you don’t have a laser cutter you can also use a knife (aka Xurography, which sounds really cool). You can use a hobby knife and a steady hand for this, but a silhouette or cricut cutter really comes in clutch here, especially for the little circles.

1. Collect the 2 pieces of acrylic that you milled earlier and clear them of any dust or debris. Compressed nitrogen is very very good at this if you have it, compressed air also works well. 
2. Grab 3 printed ports and set aside a handful of port_donuts. Carefully remove the white paper from one donut and place it onto the flat of the port, making sure not to block the hole. Grab the flow layer of your chip and hold it at eye level with the channels facing you. Carefully Remove the remaining 3M side of paper from the adhesive and guide the port onto one of the port-holes. If you mess up and block the port, just gently pull it off, remove the adhesive, and try again. Repeat this for all the ports needed for the chip. 
3. Once the flow layer is ported, grab the cover layer and the chip-sized piece of adhesive you cut earlier. Remove the white paper from the adhesive and carefully place it onto the cover layer. If placed well, remove the 3M side and place the flow layer onto it. Press gently all around the chip and try to work out as much air as possible.
4. Congratulations, you just put together your first microfluidic chip!

For general fluid handling, proper syringe pumps are the best. However, there is another option for fluid handling that does not require mechanical assistance at all: pressure driven flow using balloons and syringes. 

This is a sort of silly but super effective and really cheap method to drive fluids, and the method I will be using for this write-up. Effectively you blow up some balloons and connect them to your chip and you are in business. I like to use PVC aquarium tubing and the various valves and connectors that are sold alongside them. These can be purchased online or at almost any pet store/walmart. The bang-for-buck ratio here is incredible. 

1. Start by disassembling your 5ml syringe. On the plunger there will be a black rubber stopper; pull this off.
2. Using a hobby knife cut an X all the way through the rubber stopper.
3. Cut a 6-8” portion of PVC tubing and shove one end through the rubber stopper so that it is just peeking out the other side. Place the rubber stopper back into the syringe with the tubing still attached.
4. Cut more sections of PVC tubing and connect them to the valves as shown in the picture.
5. Take a latex balloon and place it around the end of the tubing, then using electrical tape secure it tightly. This takes some practice. 
6. Play around with the setup to see what valves you need to open/close to inflate the balloon, push air out of the syringe, etc. Once you are comfortable with the setup, you can pull some liquid into the syringe, inflate the balloon, connect some tubing to the inlet ports on the chip, and slowly release the air from your balloons.

In my experience, fluid handling is the toughest part of making microfluidic devices. If you have fluid handling systems (peristaltic, pressure driven, or syringe pumps) then you are a very lucky person indeed and you can pretty much skip this section. For everyone else, let’s get into it:

Syringe pumps are expensive, finicky machines that are simple in concept and deviously complex in execution. There are a bunch of DIY syringe pump plans out there, and they are all.. Okay-ish. I’ve personally built a half dozen and they are all pretty bad for the cost. Even bargain-tier 3D printed and arduino based pumps are around $75-$100 each in parts which put them solidly in the “not worth it” category for me. Luckily there are some options for the casual tinkerer that will not break the bank.

DIY syringe pumps, Best of the Worst:

1. Someone much smarter than me named Vittorio Saggiomo designed a set of syringe pumps using an Ender 3. First you print the parts needed, then you cannibalize the machine and use those parts to finish the pumps. This is a really cool solution since you end up with 3 programmable pumps for around $160, without having to source a whole bunch of small individual parts. If you, like me, have an Ender 3 that’s fallen into disuse this would be a fun weekend project to try. I haven’t personally built these, but the low barrier of entry makes these enticing.

If everything works and there are no leaks (there will probably be leaks) you will start to see your fluids flow. You can change the flow rates by inflating/deflating the balloons to varying diameters or manually squeezing the balloons to increase flow rate on one side or the other. Place your chip under a microscope and see how the change in flow rates affects the fluids at the Y interchange and what shapes you can make. Can you block off one arm of the Y entirely? Try getting the rate juuuust right to make some “drip” droplets. This is just one of an infinite amount of designs you can try, so hop on metafluidics and start trying out whatever you like! 

You'll note in the photo above that I have binder clips around the outside of the chip. This can help keep the chip sealed during high pressure situations like droplet formation

The chip made in these photos ended up failing. But what happened? If you look closely in the first picture you'll see a small channel that formed from a print defect in the luer port. Under the microscope this defect is much easier to see. The 3M sheet adhesive is pretty good at plugging up small defects, but one of this size is no match for the pressure levels that the chip sees. In this situation I would just toss that port and go with another one.

You can also accidentally punch through the acrylic if you are using a laser cutter and have intersecting channels. Basically, the laser cutter will "mill" out one channel, then when it is milling out the second it will go over that first channel again and ablate more material, sometimes leading to a puncture. This can be prevented by designing the channels such that they are either milled in one go, or have gaps the width of the dot size on either side of a channel so that the laser does not go over the same spot twice. This takes a little bit of trial and error to figure out on.

• Minimum feature size using this method is relatively large at ~250um (using a 1/100” end mill). This can mean certain types of chips cannot be made with these fabrication methods (pinch flow droplet formation uses 80-100um orifices to generate droplets, for example.)
• Multi-layer and controllable designs (i.e. on-chip valves) are possible, but require the use of PDMS as a substrate instead of 3M 468MP. PDMS is awesome and you should get some, but it is kind of expensive and you’ll need a vacuum chamber and a lab oven to work with it. Certainly not a deal breaker by any means (also honestly a toaster oven is fine) but it does up the fabrication time a little. Stay tuned for another guide on making PDMS membranes for these chips. 
• At-home photolithography using silicon wafers, photomasks, and PDMS would be rad as heck and is theoretically possible. I started working on this specifically a week ago and it is promising, but probably a year away from being viable (for me, anyway).

By and large most of the method outlines by this guide has been formed over the past few years from numerous papers and zoom calls. I'll try to list all of the major inspirations.

• http://2017.igem.org/Team:BostonU_HW/IntrouF - This is a phenomenal reference for making PDMS + PMMA based chips by the fine folks at BU. Probably one of the best resources for milling microfluidic chips in the home shop
• https://pubs.rsc.org/en/content/articlelanding/2018/LC/C8LC00471D - The paper that informed the balloon based fluid handling system. I highly recommend reading the paper as it gets into the nitty gritty of flow rates etc. Excellent read.

Here are some great links to learn more about microfluidic design theory, fabrication, etc.:

• https://www.pnas.org/content/104/48/18892 - Great paper on the basics of particle manipulation
• https://link.springer.com/article/10.1007/s10404-018-2048-2 - Goes much deeper into the how and tao of micrmilling on desktop-class CNC machines
• https://www.bu.edu/moss/files/2015/08/Sapienza2015_Microfluidics.pdf - Slide show with some introductory information to microfluidics
• https://pubs.acs.org/doi/full/10.1021/ac800492v - Fine control liquid handling method that I've been meaning to try
• https://blogs.rsc.org/chipsandtips/2012/04/18/easy-and-inexpensive-fabrication-of-pdms-films-of-different-thicknesses/?doing_wp_cron=1610216320.4139440059661865234375 - Step by step guide to make PDMS membranes
• https://openwetware.org/wiki/Pneumatic_Valves_-_Alex_Brosseau,_Anthony_Brouillard - This site is a goldmine of info in general. This specific page is about micro valves.