My impression is that accurately dispensing powder is around an order of magnitude more complex and more expensive than liquid handling.
What solutions have you seen for accurate powder dispensing (both proprietary and open-source)? Do you have any experience with these? What are the different mechanisms for powder dosing that make the most sense for automated chemistry/mat.sci.?
Some related discussions:
cc @shijing
A short project I did during graduate studies is 3D printed spatula, inspired by Cook et al who dispensed solids into 8x30 mm glass vials with purchased plastic spatula.
Instead of purchasing spatula of various sizes and selecting the one closest to the desired mass, I modeled and printed 3 sets of spatula with a known volume, and used them to draw a calibration curve (blue dots). This allowed me to calculate the spatula size for dispensing the volume I need.
As shown by the figure, the final printed spatula manages to have decent accuracy in isolation, but in practice I found that lighter powders did not dispense as well into those 8x30 mm vials due to static in the glovebox.
The benefit to this method is the low cost, and drawback is that a new spatula needs to be printed for every solid. A new calibration curve may also need to be drawn every so often if higher accuracy is desired (a large batch of solid may take on moisture over time, etc), and a scooping motion is required (vs just dispensing downwards).
For this particular project, I just manually tapped the spatula until each scoop is level. If transferred to automation, it doesnāt seem too difficult to physically level it by passing something against the top of the scoop part to push off excess powder.
J. Alsenz developed another straightforward method for small scale solid dispensing with a positive displacement pipette.
The pipette is adjusted to the desired volume, and the empty space is packed by manually pressing the tip into the solid multiple times. Across multiple pipette tip sizes, this method can dispense from around 0.6 mg to 20 mg, with average CV around 10%.
What I like about this method, is that the motion is closest to liquid handlers like Opentrons, so itās potentially a tool change away from being integrated into an existing robot. The tips might even be 3D printable, so it does not rely on a particular vendorās supply (which are fairly expensive).
The drawback is the limitations on volume and solid types it can handle. For every solid there will be a maximum tip length and width that it will either not stay in the tip, or be stuck and unable to be pushed out. I quickly tried packing table salt into a 1 mL syringe, for example, and it would not stay there as expected.
I am currently designing and building a low-budget, automated dispensing module for different salts.
I have settled on using an auger system for the dispensing. Now my accuracy boils down to:
Pros for using an auger:
Cons:
Curiously, it seems like the firearm enthusiast community has made some attempts at tools along these lines. I wasnāt expecting these things to require or achieve the level of precision they seem to be getting.
However, when I think about my workflow, I think that just dispensing/metering is only one component ā when Iām thinking about partial automation these days, what Iād really like is a tool that can not just dispense but recover materials e.g. after weighing or ball-milling.
Does anyone have any insight into the specs required to make something like this work?
For a few months, Iāve had an Autotrickler v4 set up to use. Nice design, very fast, accurate (+/- 1 mg), the design is rather modular, and they even sent me instructions for how to control it via bluetooth programmatically (rather than via the app that comes with it). This means Iāll be able to control it via MicroPython with a bluetooth-enabled microcontroller (e.g., Pico W), just need to get around to setting it up. Currently, Iāve been demoāing it with jello powder dispensing. Another cool thing to note is the scale (A&D FX-120i - 122g x 0.001g ā CE Products Inc - Reloading Scales + AutoTrickler Products) - it was chosen in large part because it has a fast response mode, which is important for dispensing where the rate of mass change is relatively small.
However, it seems to be better suited for larger volumes, the materials would likely need to be swapped out with something more chemical-friendly, and the designs arenāt open-source. There is also GitHub - eamars/OpenTrickler: The open source powder trickler solution with DIY hardware and software., but it wonāt be as easy as the plug-and-play solution that Autotrickler v4 is. Of course, ātricklersā arenāt the only approach. Itās great to see the variety of solutions proposed here. We also have the Metter-Toledo āsalt shakerā and the Chemspeed āoverhead gravimetric dispenserā approaches as well.
Hey Sterling,
Thanks for this note! One thing that I am thinking about & dealing with in my experiments is that typically itās not enough to just dispense powder, but I also need to āpick it back upā after a certain operation (e.g. ball milling or acoustic mixing).
Do you have any ideas or tools that accomplish this? Iāve been looking into powder pipettes but the accuracy is quite low and also Iām worried about vacuuming small particles into a hopper that can then react together if we use a variety of different precursor compounds.
Do you have any ideas or tools that accomplish this?
This has been on my mind since you brought it up a while back - I agree itās important (especially for solid-state materials science applications), but it seems like a tough challenge. Do you have a specific workflow in mind?
Perhaps worth looking at how Cooper group with PowderBot and Ceder group with ALab (probably best to check SI and videos) dealt with these issues. I think both required milling of some kind are were somewhat of workarounds specific to their setups. Hard for me to think of something general that doesnāt involve mimicking a human with a spatula.
I am attempting solid dispense using a glass/plastic auger used for chemistry.
I used a 3d printer adapter and a stepper motor with control board
I am currently debating what scale to buy to use for feedback:
Cheapest Option $200
Sterlingās Suggestion, More Expensive $580
It needs to have digitized I/O for control. I am not sure of where I will use this but I wanted to try prototyping it.
Looks like a cool project!
An additional option that may be worth considering is the A&D HR-100A (0.1 mg resolution) CE Products scale, which also seems well-geared towards powder dispensing applications in terms of fast response:
Sending RS232 commands using MicroPython and a Pico W would probably look something like whatās summarized in this transcript.
The A&D FX-120i Manual - English linked to on the A&D FX-120i website shows the commands on page 53 of the PDF (section 15). Iām not an expert with RS communication, but it seems like it would work, just need the appropriate RS232 to TTL module, preferably one with screw terminals.
Not a direct answer, but maybe the powder handling pieces from here can spark some ideas: https://www.emeraldcloudlab.com/instrumentation/. I think the ideal solution will depend on the lower and upper limits on the mass you want to pick up.
I like what they classify as the analytic balance, with the 0.1 mg resolution and maximum 220g
Also looks really simple to work with Pico W. I just learned you can ask chat GPT about how to DIY electronic communications. Adapters are not expensive
Something like dispensing precursors ā mixing ā high-energy ball milling ā sintering? And Iād need to pick up the homogenized mixture after the ball milling. This is doubly complex because the ball milling will reduce the particle size, meaning that things like a powder pipette are less viable as the vacuum will ingest the particles.
Iāve also been disappointed with the repeatability of powder pipettes.
