Piezo drop-on-demand inkjet head
Updated: Feb 5, 2020
A surprising amount of chemistry comes down to moving liquids around in a controlled way, this sounds simple in theory but actually becomes incredibly difficult to do in automated systems. A recent project required a liquid deposited on a porous material in a highly uniform and controlled manor. After trying many different approaches I turned to inkjet printing.
Inkjets are devices which spit out droplets of liquid on command from a reservoir. The most common type used in printers today do this by rapidly heating a fluid chamber, however a second type uses piezo electric materials driven in pulses to kick out tiny droplets. Since I plan on playing with many different liquids and have experience working with piezo crystals I went with the latter option. Combining one of these heads with a moving stage leads to an inkjet printer that should solve my problem.
First looking at the commercially available options, household inkjet printers already come with a moving stage and print head with sometimes over 300 individually controlled jetting heads:
These can cost as low as £30 however the requirements for printable fluids are quite strict and would need some heavy reverse engineering. Inkjet valves for chemistry do exist, they can usually handle pretty much any liquid and dispense a controlled amount with incredible precision:
BUT the cost of these things is around £10,000 :(
So lets see what we can do with some 3d printing and a positive attitude
For a prototype I just gathered parts which I already had laying around and designed a chamber around them. The design can be seen at the top, its a straight channel through the whole valve. The reservoir is at the top, with a round chamber capped by the piezo crystal and the nozzle is a 0.2 mm 3d-printer kind.
To drive the piezo crystal I just used mains voltage through an isolating transformer, this is not ideal as the pulse timings cant be changed but a good start for testing the concept.
So after filling up the reservoir with some soapy water, the 3d print was not as water tight as I had hoped, but I decided to give it a shot anyway:
If you ignore the large droplets leaking through the side you can clearly see 2 jets fire out of the nozzle when the piezo crystal is actuated.
Stepping up the game
For the next prototype I got the biggest piezo crystal I could find and some pneumatic hoses to tidy up the plumbing. The design is quite similar however there is a key difference, the reservoir is external and connected by a flexible silicone tube. The jetting head itself is mounted on a height adjustable stage, so lifting or lowering the nozzle with respect to the liquid level in the reservoir adjusts the hydrostatic pressure in the chamber.
Also Instead of using an o-ring and custom thread to secure the piezo crystal to the chamber I simply added another component which screws onto the chamber and a thin layer of silicone provides a good seal.
The height difference between the main chamber and the reservoir should be based on the radius of the nozzle used (r), surface tension of the liquid (sigma), density of the liquid (rho) and the strength of gravity on the planet you intent to use this on (g).
In practice this is set by flipping the head upside down, pulsing the crystal to dislodge any air bubbles, then attaching to the height adjustable stage and lifting until fluid is no longer dripping out of the nozzle.
To drive the crystal I experimented with several approaches:
First one was to drive the piezo parallel to an inductor. If the mosfet is pulsed then the collapsing magnetic field across the inductor can build up incredibly high voltage spikes. Using inductors in the mH range I managed to drive my crystal at around 400V while starting with less than 20V. This did create some intense audible kicks in the crystal but did not work to eject droplets. It turns out that pulse width plays a bigger role than pulse intensity.
So the next idea was to use a H-bridge to deferentially drive each side of the crystal, however I did not have any high side gate drivers so I came up with this:
Each side of the crystal is either pulled high to around 30V or pulled low to 0V. The gates are driven by a logic inverter IC so the device requires a single pulse and the width sets the period that the crystal is flexed for. This means that the crystal is always held in deflection towards the chamber, and flexed in the opposite direction during a pulse, creating a 60V drive signal. This is also a good idea as the piezo is driven in complete cycles which eliminates historesis problems which would reduce the reliability.
This prototype was printed on my significantly improved 3d printer and turned out and after a while of fine tuning I was able to get the jetting valve to reliably spit out droplets:
This version uses a 1 mm nozzle, has no leaks and is extremely reliable. Each droplet is around 1 uL. I attempted to measure the repeatability by mass of 20 droplets but after 10 rounds the deviation was less than 0.2 mg so far less than i can accurately measure using an analytical balance.
By using a smaller nozzle the droplets can be made microscopic, even to the point where they are almost impossible to see and evaporate within seconds.
Each of those graduations is 1 mm and that is still not even close to the limit of this device.
The droplet generator can even be used in 'rapid fire' mode where the droplet stream appears like a continuous jet:
So this device should be more than sufficient for my purposes and even opens up a whole new array of projects. I'm currently designing several more, and building them out of materials much more resistant to organic solvents. So there will surely be follow-on projects looking at printing exotic materials and automated droplet scale chemistry.
So no need to spend 10k on a commercial model, some scraps and a 3d printer perform just as well!