millibeep

As part of a project I am working on, I needed to make some microfluidic devices. For those that aren’t familiar, these are usually chips that are made with channels, some few microns tall and wide that can handle a few microliters of fluid (hence the name: micro- fluidics). Typically, and indeed in my case, these channels are made of some soft material and bonded to glass, so that the channel exists as an impressed feature in the material and is bounded by the glass, creating a pipe that is easy to image with some microscope. Anyway, I made some earlier today and thought it would be neat to document the process here.

The first thing to do, is make or buy a substrate that has whatever channel design you want printed on the surface as a positive image (i.e. it sticks up off the surface) because this is what will stick into your soft material and create the channel; a process known as ‘soft lithography’. In my case, I have a silicon wafer with a simple design on it made with a UV-curable chemical called ‘photoresist’, which is also an interesting process, but is really best as a subject for another blog post.

You can just barely see the raised up channels, which are about 25 microns tall and 500 microns wide. I needed to be careful handling this, as the photoresist could easily be scratched off or distorted, altering my final channel shape. Here’s a close-up image of the channels.

Now that I have my wafer and channel design, I need the material into which I will actually cast the shape of the channels – the “soft” part of soft lithography. In this instance, I’ll use a substance called ‘PDMS’ which stands for polydimethylsiloxane, which is long and hard to pronounce, but the main takeaway is that it is a highly elastic silcone rubber with some neat properties (go check out the linked wikipedia page for more info).

The PDMS that I am using comes as a two component binary mixture. Once you mix component ‘A’ with component ‘B’, a chemical reaction starts and the rubber starts to cure. The neat thing about having a two component system is that some properties of the final product (in this case, namely stiffness/elasticity) can be tuned by the initial ratio of ‘A’ to ‘B’. With the specific PDMS that I am using, a 10:1 ratio of A:B seems to work best.

Luckily, the aforementioned 10:1 ratio is of mass rather than of volume (although I think they would be pretty close). In this case, it makes it pretty straightforward to measure out the correct amounts.

Oops. Close enough.

Now that I have my two components added together, I need to mix them together really well. For this I just used a plastic stirring rod and elbow grease. However, this opened the door to another problem – my PDMS now has a LOT of bubbles in it.

Normally, if I was mixing two component epoxy or something similar, I wouldn’t care about bubbles in the mixture. However, because I will be using this mixture to make channels with really small feature size, any bubble that sits on the channel imprint could completely disrupt the design. Therefore, I need to get rid of these bubbles.

De-bubbling the PDMS is not as complicated as it might seem. A common feature in laboratories and suspiciously-well-equipped garages is a vacuum chamber. High enough vacuum will increase the differential air pressure between the inside and the outside of the bubbles, causing them to expand and eventually burst. All I have to do is stick my cup of silicone under vacuum for a little bit and all the bubbles will be sucked right out!

It can happen that the bubbles foam way up like a shaken soda – in this case it is best to release the vacuum slightly, or just gently slam the whole box down on the bench to disrupt the bubbles into bursting. After about 20 minutes or so, the last of the bubbles had burst and I had well-mixed, de-gassed PDMS ready to pour over my wafer.

Before I should do anything more with the polymer, I need to prepare my wafer to hold the PDMS over it long enough for the silicone to cure. Remember the tinfoil from the earlier picture?

Ta-da!

Now I am ready to cast the PDMS over the wafer. In order to not undo all my hard de-gassing work, I need to be exceedingly gentle when pour the liquid polymer, so as to not introduce any more bubbles that could disrupt the fluid channel features. I needed two hands for this, so luckily a fellow lab denizen was nearby who could take a photo for me.

It is usually best to pour from one corner and let the highly viscous polymer gently creep across your wafer like so much molasses in a molasses flood or similar natural disaster.

Once the PDMS was poured, I set it on a level surface to cure for 48 hours. Alternatively, I could have stuck it into a 80 degree centigrade oven for an hour or two to cure it the same amount. This wafer won’t be done today, but luckily I had another wafer I had poured a couple days ago that was ready to go.

Once the PDMS is set, I now just have to cut away the foil and excess polymer, and cut out the individual devices.

At this point, caution is essential, as the crystalline silicon is very brittle and putting too much force on it will snap it (which is bad because they can be fairly expensive).

Now I have pieces of PDMS that have a pattern negatively imprinted into one side, which will be bonded to a piece of glass to create a sealed channel. However, in order to access the channels from outside the device, I need to punch holes in the material that I can connect to tubing afterwards. For this, I use a small biopsy punch, but a polished, flat-end injection needle will work too (an angled needle is no good, as we need to punch out the ‘core’ of polymer, leaving a hole).

It’s important to punch on the end of the channel feature so that the hole can access the imprinted channel.

Now I have to clean the feature side of each device to prepare them for bonding to the glass – luckily this is made very easy by one of mankind’s best, most useful inventions:

Scotch tape!

The tape will both protect the features on the PDMS from debris while I prepare the glass for bonding, and when I peel it off, the tape will bring with it any debris that is already there. Wonderful!

The next part is where I clean the microscope slides and prepare them for bonding, but since this step just involves cleaning the slides with isopropanol, we can skip the pictures. The critical thing is to make sure they are clean and dry.

Next, I take my cleaned glass slides and taped up PDMS chips over to the plasma cleaner. I won’t get into the details here (feel free to check out the linked wikipedia page) but the long and short of it is that the plasma both cleans the surfaces of the PDMS and glass, as well as activates OH groups in the PDMS and glass (both are silicon based) so that they will form bonds with each other without the need for any adhesives or clamping.

Into the science box!

Once the chip and glass are situated, the chamber is evacuated with a pump, and the ionizer is turned on, creating a bright pink plasma that bathes the devices.

Once the plasma is active, I usually let the machine run for two minutes to make sure the surfaces are nice and activated. Once the time is up, I have to work quickly to pull the devices out, and flip the PDMS chip onto the glass slide (features down). Once the two surfaces make contact, they will immediately bond and pull together (no need to press on the PDMS). Then, I like to let the bonded devices sit on a hot plate at 80 degrees (centigrade) for a few hours to ensure good bonding.

While the bonded devices are toasting to completion on the hot plate, I can prepare the fluid pin connectors that I will use to connect tubing to the channels on the device. You can buy specialty connectors for this, but I like to use blunt, flat end needles. If you gently bend the needle 90 degrees, you can twist it out of the syringe base, leaving you with a handy, L-shaped pin connector.

At this point it is lunchtime and so I leave the pin connectors and devices on the hot plate alone and go have lunch.

By the time I get back, the devices have bonded sufficiently, and I can hook up some tubing. To do this, I gently push the L-shaped connectors I just made into the holes I punched earlier, like so:

From this point it is pretty straightforward to hook up some tubing to the pin connectors and create a functional microfluidic device, which is exactly what I did for this example chip.

The application for which I need these devices is a little more complicated, and actually involves a second device that has an on-board, PDMS valve system, but that’s a more complicated post for another time. In the meantime, enjoy this video of the chip in action!

/millibeep