Prototype Design: Creating a Machine to Remove Microplastics from Ocean Water
Advisor: Ms. Abodouma
Internship Location: The Marine Mammal Center: 2000 Bunker Rd, Sausalito, CA 94965
Onsite Mentor: Joelle Vallejo
“With great power comes great responsibility” is the proverb popularized by Spider-Man in 1962. In that same decade, the reign of plastic began. Today, some 358 trillion particles of microplastics are floating upon the surface of the world’s oceans, and that’s only the surface of the problem. We still do not fully understand the long-term consequences of microplastics, but we know that microplastics release toxic chemicals linked to DNA damage, cancer, and reproductive complications into their surroundings as they break down into smaller particles. Although some research has been underway to discover methods to remove microplastics from our environment, much of that research has not found success beyond the lab bench. This project is an attempt to make a small dent in the 358 trillion by bridging the gap between research and its application. I will automate a method developed by Fionn Ferreira to remove microplastics from water with ferrofluids, by creating consecutive prototypes to refine this idea. Interning at The Marine Mammal Center, I will conduct lab testing to determine my prototypes’ efficacy as a way to guide my design process and gain a more complete understanding of the impact of microplastics on the health of our global ecosystem. It’s time we took up our “great responsibility.”
Project Blog
Post 1: Blog 0: "With Great Power Comes Great Responsibility"
“With great power comes great responsibility.” This was the proverb popularized by Spider-Man in 1962. In that same decade, the reign of plastic began.
How have we been bearing our responsibility to the environment?
Not quite well. In our frenzy for a cheap, versatile material, we have neglected to fully understand the life cycle of plastic. Here’s the basic idea:
Extraction of ethane and propane from fossil fuels
Processing and manufacturing with resin to form plastic
Distribution and consumption
???
Well, it turns out that of the 8.3 billion metric tons we’ve since produced, 6.3 billion metric tons are sitting in landfills or hanging about as litter (1). Plastic doesn’t naturally decompose; instead, it breaks into smaller and smaller components called microplastics.
According to a 2023 report from the UN, microplastics are thought to release toxic chemicals as they break down, which have been linked to DNA damage, cancer, and reproductive complications (2). Today, there are about 358 trillion particles of microplastics floating about in our oceans today (3).
I hope to make a dent in this 358 trillion by designing a prototype to remove microplastics from ocean water through this Senior Project.
Now, this sounds pretty far-fetched! However, my background refined throughout high school in biology, chemistry, and physics, can help me answer the tough, theoretical questions behind this interdisciplinary project. On the other hand, my experience building small electronics projects in my free time, such as a RC duck or a pill dispenser, have been the key to my understanding of the product development process. I have been interning at The Marine Mammal Center for my Senior Project to understand the context behind my project. I’ll continue working with Ms. Abodouma, my wonderful internal advisor.
It’s time we took up our responsibility to the environment...
…and walk in the steps of Spider-Man!
Works Cited:
Parker, Laura. “A Whopping 91 Percent of Plastic Isn’t Recycled.” National Geographic, 22 Jan 2024. https://education.nationalgeographic.org/resource/whopping-91-percent-plastic-isnt-recycled/. Accessed 26 Jan 2025.
“Chemicals in Plastics.” UN Environment Programme, 2023, https://wedocs.unep.org/bitstream/handle/20.500.11822/42505/Chemicals-in-plastics_Summary.pdf?sequence=1&isAllowed=y. Accessed 12 Dec 2024.
Asher, Claire. “Microplastics Pose Risk to Ocean Plankton, Climate, Other Key Earth Systems.” Mongabay, 9 Oct. 2023, news.mongabay.com/2023/10/microplastics-pose-risk-to-ocean- plankton-climate-other-key-earth-systems/. Accessed 6 Nov. 2024.
Post 2: Blog 1: There's a bottle on my desk
Hello Orange Juice lover!
There’s this bottle on the corner of my desk. It’s been here since last November, but this week, I finally got around to it.
It’s a translucent container, and its contents are clearly a deep, midnight black. Lifting it up, it’s significantly heavier than expected at first glance. Shaking it, its contents stick stubbornly to the bottom.
Well, what’s in the bottle?
Expired Orange Juice.
Just kidding. There’s no Orange Juice here.
This is a bottle of Iron Oxide. According to the label, it’s Black Iron Oxide, Fe3O4, 99% reagent grade, 44 Micron powder in a 4-ounce bottle. Apart from its scary name, the fun thing about this specific type of rust is that it’s magnetic. In fact, its other name is “magnetite.” Think of this as those iron fillings you might have played with in science class, but in a powdered form that can be dissolved in a solvent, like oil.
Speaking of which, I dissolved a bit of this fancy rust in half a tablespoon of vegetable oil. At the nanoscale level, these magnetite particles became suspended within the oil, forming a basic ferrofluid. At the macroscale level, this oil-magnetite mixture began to exhibit some interesting properties: It’s certainly a fluid at first glance, but when exposed to a magnet, the fluid is immediately attracted to it and hugs the surface of it. (When you do this, try wrapping the magnet in plastic wrap to make the cleaning process easier! Personal experience speaking…)
Higher-grade ferrofluids that you may see online contain a surfactant to reduce clumping within the carrier fluid so that the final ferrofluid contains about 5% magnetic particles, 10% surfactant, and 85% carrier fluid (Magcraft). These will exhibit distinct “spikes” when exposed to a magnet because of the balance of forces between the attraction the nanoparticles within the ferrofluid feel in alignment with the magnetic field and the resistance they face from gravity and surface tension.
Next week, I will be using the basic idea of ferrofluids to create a proof of concept to remove microplastic particles dissolved in water. For now, though, this is pretty fun to play with as I finish reading John Pichtel and Mathew Simpson’s book Microplastics: Behavior, Fate, and Remediation, which will guide my prototyping in the upcoming weeks.
See you then, Iron Oxide connoisseur! No more Orange Juice…
Works Cited for Blog #1:
Magcraft. “What is a ferrofluid?” Magcraft, 31 Jan 2015. https://www.magcraft.com/blog/what-is-a-ferrofluid. Accessed 23 Feb 2025.
Post 3: Blog Post 2: Every Answer Leads to More Questions
Hello Iron Oxide connoisseur!
At this point, you’re probably wondering how this fancy rust will help me clean microplastics from water. Still sounds pretty crazy, huh?
This week we’ll be doing a Proof of Concept.
The following process was developed by Fionn Ferreira, the winner of the 2019 Google Science Fair. Through my Senior Project, we will be adapting his scientific method into a prototype that can be used to automate the process.
Here’s the basic idea:
Given a solution of microplastics and water, add a drop of oil (vegetable oil preferred) and mix thoroughly.
Since both oil and most microplastics are hydrophobic while the water is hydrophilic, the microplastics will clump within the oil (“like mixes with like”).
To collect the oil, mix in an equal volume of magnetite to the solution. The magnetite will become suspended within the oil, creating a weak ferrofluid within the water.
Use a magnet to collect the ferrofluid from the water. Several passes may be needed, depending on the strength of the magnet.
Take a look here for a video demonstration of the process.
Crazy still?
Although the current process visibly reduces the cloudiness of the water by removing the microplastics, I still have so many questions! What is the optimal ratio of magnetite to oil in the ferrofluid? Does the order of steps matter? Can I combine the oil and magnetite beforehand while still preserving the method’s effectiveness at cleaning up microplastics? Can I recycle the spent ferrofluid? How long does the cleaning process take? Can I make it faster? And the biggest question of all: How will I automate this?
I’m still refining my understanding of using magnetite and oil for microplastic removal. This week, I’ve narrowed my reading down from general textbooks to recent research papers and literature reviews. Some of my questions are being answered (use a 1:1 ratio between magnetite and oil in the ferrofluid by volume for a 97% collection rate!); others won’t be until I get my hands on the supplies I’ve ordered for my upcoming prototype.
If you come back next week, we’ll be moving on to planning and prototyping–join me for Hack-A-Blender!
See you then!
Post 4: Blog 3: A Bath and a Roadblock
Hello!
Haven’t you have just been waiting on pins and needles for this update? Maybe?
In any case, you’d be glad to hear that all parts have arrived!
This week’s original goal was to create a basic prototype from a household blender to make sure the basic electronic components work together. However, progress was slower than expected. I was only able to reach an intermediate step testing the components.
My prior experience with electronics has been limited to two or three simple projects beforehand, so this week has been quite the challenge for me – with plenty of moments where I just wanted to bonk my head against the wall. I’ve been figuring things out by trying things, getting terribly stuck, and asking my dear friend Google for help.
Here’s what I’ve got right now:
Plugging the Arduino in, this contraption pumps air from both of the air pumps. One pump is connected to the blue bottle through a tube. As the air in the bottle is compressed, the liquid at the bottom is pushed out of the bottle and into the cup, successfully filling the tub for our puppy.
Filling the bath for this puppy may seem unrelated to our final prototype, but it is an important intermediate step that allows me to test the components as they work together. In fact, it has revealed an important problem that I didn’t realize while designing. When both motors are running, the Arduino’s digital pins don’t provide enough voltage to fully operate the electromagnet. Although there is some magnetic force being generated, the electromagnet can barely hold a small screw in place. I need to connect the electromagnet to the external power source with the motors instead of relying on the Arduino’s pins, unfortunately, for it to be strong enough to pull ferrofluids out of water.
To do so, I’ve set an order for a TIP102 transistor. It will be similar to the H bridge I currently have for my two motors. This transistor will act as an intermediary between our electromagnet and Arduino, allowing them to communicate even while the electromagnet is running at a higher voltage that would fry our Arduino if directly exposed.
Although I’m disappointed that I’ve run into this roadblock, at least the pup is clean. Hopefully she enjoyed her bath! As I continue to wait for parts, I’m learning how to use OnShape to help me with the modeling for 3D printed components in the later parts of this project.
Hopefully, next week, we’ll get beyond providing the plumbing for a dog bath.
Bye!
Post 5: Blog Post 4: “10,000 Ways That Won’t Work”
Hello!
We all make mistakes. When asked about his, the great inventor Thomas Edison replied, “I have not failed 10,000 times—I’ve successfully found 10,000 ways that will not work.”
I thought I’d been on the brink of developing a successful prototype with my Hack-a-Blender scheme this week, but it turns out that I have not. I’ve merely found another way that will not work. In this post, I’ll go over what the prototype looked like and break down its key shortcomings.
This week’s prototype is–as the name suggests–an adaptation of a household blender. It has two air pumps attached to the two sides of the blender and an electromagnet secured to the side. I’ve drilled two holes, one about three centimeters in diameter and the other about the width of a pencil. The larger hole is for the electromagnet while the smaller one is for a polymer clay funnel through which extracted ferrofluid can be removed.
In theory, the first pump would bring in the contaminated water to the blender. The second pump would bring in the ferrofluid. The blender would mix the two together. The electromagnet would pull out the ferrofluid from the water as the blender spins. The water would be drained out through a small hole on the bottom of the blender, leaving behind the ferrofluid residue attached to the electromagnet, which would then flow out through the polymer clay funnel.
See picture:
Why doesn’t it work?
I’ve identified two primary problems with this prototype.
1. The blender mixes at such a high speed that any ferrofluid attracted to the magnet is forcibly washed back into the mixture, causing few particles remain on the electromagnet after the water is drained out. Since the blender is premade, I cannot reduce its speed.
2. Although the TIP102 transistor I ordered last week has made the electromagnet much stronger than before, it has a small active circle only about a centimeter in diameter that is actually magnetic. When separated from the water by a layer of plastic, the electromagnet's pull is weak at best. Even if the water speed were reduced, many consecutive passes would be needed in order to extract any ferrofluid from the water.
These two problems work together to render this prototype largely ineffective.
I’m disappointed by this turn of events, but I have a Plan B in mind. Hopefully next week I won’t find another one of the 10,000 ways that won’t work…
Bye!
Post 6: Blog Post 5: Plan B
Hello!
Welcome back! This week was quite busy for me!
For starters, I was unceremoniously kicked out of the house into the garage to do my prototyping (for good reason, with the racket I was making sawing, nailing, and muttering!). After clearing a table in the garage, I threw out the remnants of Hack-a-blender and began daydreaming and sketching alternative builds.
Here’s the initial idea I landed on, inspired by a robotic arm:
An actuator would convert the rotational motion of a continuous servo into vertical linear motion, lowering a positional servo into a container. The positional servo would slowly sweep through the water in the container with an electromagnet to collect ferrofluid. Working together, the two servos would lift and rotate the electromagnet to a separate container, where the ferrofluid can be dispensed.
Going from pencil sketch to reality was much harder than I thought! There were a ton of electronic, hardware, and software issues that all needed to be resolved. I can’t go over all of them with you, but I’ll give you just a taste with this one:
After building the framework and adding in the electronics of my prototype, I began testing each electronic component with simple lines of code to make sure each part was still working properly after assembly. Unfortunately, this revealed the fact that the positional servo attached to the electromagnet was violently jittering, causing the entire framework of the prototype to vibrate.
According to one online troubleshooter, the problem could be resolved by adding a massive 1,000 μF 25V capacitor to the problematic servo to stabilize the power supply and reduce electrical noise. So I did. And it worked!
Then I went to go get dinner.
Once I got back, I turned on my power supply again. The first servo turned. The second servo turned. And then the whole circuit began to smoke.
With a quick glance at my power supply, I realized that the current had shot up from the usual 0.1-0.4 amps to 4 whole amps. I panicked–pulled out the Arduino’s connection with my laptop and slammed off the power supply. Holding the cord in my hands, I could only watch as a thin trail of smoke wafted from my circuit board and envelope me in that distinctive scent of burnt-out electronics.
I’d short circuited my Senior Project.
Very nice.
I called my dad into the garage to figure out my next steps. Together, we disconnected the Arduino from the circuit and gingerly plugged it back into my laptop. Two red lights blinked back at us. It was alive. Phew!
In a similar way, we tested each component with simple lines of code, waiting on bated breath to see if they’d survived. Miraculously, the servos, transistors, and electromagnet were all functional.
Even now, I’m still not sure what exactly caused the short circuit, but I suspect that it was a combination of that large capacitor and a lack of appropriate resistance in the circuit. However, I took no chances and doubled down on prevention. I removed that overkill capacitor and replaced it with a far more reasonable 100 μF one. I added a 150 R resistor to the electromagnet. I put my servos onto the external power source instead of relying on the Arduino’s 5V pin. I reduced the voltage of my external power supply from 6V to 5V. I switched out my breadboard for a higher quality one. I threw out my flimsy jumper wires and replaced them with solid copper wiring.
Four hours later, when I turned that power supply back on again, I had seemingly come back to the same point I’d been at before dinner. Namely, I still had that jittering servo in my prototype. However, the short had forced me to create a far safer circuit. Disaster had pushed me to reconsider the design aspects I’d taken for granted.
(Eventually, I figured out that the jittering servo was inherently faulty. After changing it out, the prototype was perfectly fine.)
As you can probably imagine, this week was quite the adventure! I’m glad though that it wasn’t all for naught. Here’s a link to a video of what I have so far: https://youtu.be/qly2lhNupBY
Next week, we’ll move on to testing and calibrating this machine! Hopefully I don’t short anything else out…
Bye!
Post 7: Blog Post 6: Across Space and Time
Hello!
This week I’ve been calibrating my prototype to make sure that it is both precise and accurate every time it runs.
Calibration is like synchronizing a complex dance across time and space. The electromagnet must be plunged into the water at the right depth to reach the water without scraping the container bottom, so the continuous servo must be coded to spin for 300 milliseconds. Once the electromagnet is turned on, it is slowly rotated 60 degrees through the water to sweep and collect ferrofluid. In order to dispense the collected ferrofluid, the continuous servo must spin for–not 300 milliseconds–460 milliseconds to raise the electromagnet back into the original position against the force of gravity. Then, the second servo snaps counterclockwise 70 degrees, and the continuous servo drops down to delicately place the electromagnet on top of a sponge. The electromagnet turns off, allowing the collected ferrofluid soak into the sponge. Only then does the whole system reset. Finding these “magical” numbers has been a relentless process of trial and error to progress from vision to dry run to wet run.
Well, does the prototype work?
Yes, but it could clearly be better.
Although the prototype is indeed able to remove ferrofluid from the water sample, there are a number of dead zones where ferrofluid cannot be removed. I added a servo to the prototype with an attached brush to agitate the water and improve circulation so that the electromagnet is able to remove more of the ferrofluid.
However, since both the electromagnet and the brush move in arcs, there is still built up ferrofluid that are beyond their range, no matter how many times the sequence is run. See image:
This robo-arm prototype is a significant step up from Hack-a-blender. I’ve learned a lot from building and calibrating it, but there is still significant room for improvement. First of all, I must think of a way to minimize–or better yet eliminate–dead zones. Furthermore, based on some preliminary testing, I believe that my electromagnet is still too weak, despite being connected to the external power source. A permanent magnet has a much better weight to strength ratio that I should consider incorporating into the next prototype.
Well! Back to the drawing board then…see you next week!
Bye!
Post 8: Blog Post 7: Plan C || Hot Glue and Hope
Hello, hello!
This week, I’ve been working on Plan C, the third prototype! I’m a bit tight on time, having spent too long prototyping, so this post might be a bit short!
As mentioned last week, I wanted to incorporate a permanent magnet into my designs to simultaneously improve energy efficiency and create more suction. Sounds a bit paradoxical for sure, but it might make more sense if you take a look at this planning diagram here:
Actually, did that help? Probably not….it’s still pretty confusing.
Essentially, the magnet is attached to a pole that can slide freely up and down a vertical z-axis and spin in the xy plane, thanks to a bead I sculpted out of polymer clay. A 3D-printed shell can control the movement of the magnet using two servos to power the two degrees of freedom the magnet has, moving it into or out of the water. Finally, a continuous servo attached at a 90-degree angle to the shell can spin with a cut-up toothbrush to thoroughly agitate the water sample, eliminating the issue of dead zones, like those in the previous prototype.
Here’s what it looks like right now:
I’m aware that it looks quite messy, but it is operational. Here’s a video of the prototype in a dry run: https://youtu.be/nblCfA6Swl4.
There are still some hiccups to be straightened out with the position of the water sample and the the coordination of the prototype, but I’m sure I’ll get it figured out by next week!
If you’re around next week, we’ll do a wet run and see if Prototype 3 lives up to expectations!
Bye!
Post 9: Blog Post 8: Does it work?
Hello!
My prototype has undergone a few changes since you last saw it two weeks ago!
Long story short, I replaced the magnet-on-a-stick mechanism that I’d been using to lift up the magnet with a weak spring (made with florist’s wire!). Another slightly stronger but smaller spring sits on top of the bead to dampen unwanted movement. This makes the vertical motion of the magnet a lot more efficient and less damaging to the surrounding components.
Furthermore, I changed out the positional servo on my linear actuator for a continuous one so that I am able to double the vertical range of motion from 7 notches to 14.
Here’s what it looks like currently:
Now–the big question–does it work?
Well, this third prototype looks very promising, since it no longer suffers from the problems plaguing the previous few. The rotational motion of the brush is able to thoroughly agitate the water sample so that there are no more persistent dead zones; the removal of the electromagnet reduces energy consumption so less current needs to be supplied.
Here’s a quick video of magnet/spring and the brush mechanisms in action: https://youtu.be/e-VhGBm_zpc
I’m confident enough with this prototype to move on to the next steps. I’m planning on testing the machine with water polluted by microplastic particles. I’ll be using a light microscope to test water samples before and after purification with my prototype to determine the machine’s efficacy in removing microplastics from water. If I have extra time, I also want to try remaking the prototype while keeping the design to make it look–um–a bit more polished…
Bye! Catch you next week!
Post 10: Blog Post 9: Working with what we've got!
Hello!
The “lab results” are back! This week I’ve been running wet-run tests on my prototype to see if it is truly able to remove microplastics from water.
My “lab” consists of one of those kiddie play microscopes with a mirror (it does magnify though!), a handful of slightly scratched slides, and a phone to take pictures. If the results look a little scruffy then, remember that they’ve been done with a elementary schooler’s kit, not a professional lab.
Anyways…we’ll just have to work with what we’ve got!
I ground some yellow PLA into microplastics with a file. These particles were mixed into 20 mL of water, and a small sample was taken at this stage to act as a control for future comparisons. The rest was combined with the usual: half a gram of magnetite and 1.25 mL of vegetable oil. I ran the mixture in my prototype for a total of 28 runs. Some ferrofluid remained stuck on the bottom of the container, but I don’t think they affected the water quality anymore at that point, since they weren’t floating in solution. Changes happened most quickly in the first 10 or so runs and decreased at an increasing pace over the next few runs, in a way much like a logistic curve approaching a limit. At the end of the runs, another sample was taken from the water.
Well, what are the results?
Here are some pictures of the control slide (pre-treatment) at 300x magnification:
The microplastic particles are clearly visible as yellow masses in the water.
And here are pictures from the post-treatment slide at the same magnification:
Surprisingly, from this preliminary testing, I haven’t been able to pick out any of those yellow particles in these pictures. The original paper from Fionn Ferreira said that this ferrofluid method had an 85% efficacy rate on average. There are a couple of “bubbles,” but other than that, the water has been restored.
Seeing those slides under the microscope brought me so much joy! It was incredibly satisfying to see how the work of the previous two months was coming together. To be honest, there were many, many times over the past few weeks when I was convinced that the stress was too much. I found myself seriously considering dropping Senior Project. However, my internal advisor, Ms. Abodouma, kept me in the grind through the lows, week after week, check-in after check-in. Thank you, Ms. Abodouma!
Oh, wait. It’s not even the end of Senior Project yet… I’ll have to save the rest of my shout-outs for the symposium.
I’ve already begun working on a neater version of this prototype by incorporating some 3D-printed parts. Here’s a sneak peek—
Hopefully I’ll be able to finish this version by next Friday!!
Bye for now! I’ve got to get back to it!
Chloe
Post 11: Blog Post 10: Rube Goldberg and a Reflection
Hello!
I squeezed, squeezed in this final prototype right before the deadline, which is tomorrow! It's quite similar to its predecessor (“Hot Glue and Hope”). I’ve preserved the essential concept while improving the execution: The spring that once enabled the magnet’s vertical motion has been replaced with a 3D print that looks like a piston to allow for smoother motion with less resistance. The polymer clay bead has been replaced with an arc-shaped piece that looks like a Hot Wheels track to allow for more durable use.
These two improvements have not only made the final product appear neater but also have allowed me to stack more magnets in my mechanism, meaning that this prototype should be more effective at removing ferrofluid.
Here’s what it looks like right now:
What do you think?
Well, once again, it was put to the test. While the previous version required 28 runs to remove most of the ferrofluid, this version required just 19. Furthermore, the brush mechanism of this version is more accurate, so even the ferrofluid congealed on the bottom of the container has been scraped off and removed from the solution. Compare these two images:
Prototype 3.0 after 28 runs:
Prototype 3.1 after 19 runs:
I have yet to test this version with microplastics and a lab microscope, but I hope to finish that before the Symposium. I suppose I’ll be working on this until the very end, accelerating to through the finish line.
Speaking of which, is there a “finish line”? I’ve been scraping away at this project for ten weeks now, and I have to say that I’m not completely satisfied. There’s just so much I wish I had more time and access to resources for.
1. I wish I could have had the opportunity to visit a proper lab where I could compare my water samples before, after, and throughout my prototype’s cleaning process under a strong light microscope or with a spectrophotometer. These two tools could have helped me quantize the efficacy of my prototypes with hard data and optimize the number of runs that should be done.
2. I hope that in a future prototype I can streamline the entire microplastic cleaning process using a single machine, from pumping in contaminated water into the machine, to mixing in the ferrofluid, to draining the cleansed water. Right now, my prototype can only do a fraction of this process by itself.
3. Finally, if I want to make a significant difference in the amount of microplastics in our bodies of water today, I must expand my scale. There’s no way anything can happen if this prototype has a capacity of 20mL and must run for an hour each time.
I’ve tried my best with this project given my current skill set and the 10 weeks I was allotted, but I don’t think it’s enough. I will be studying Mechanical Engineering and probably taking on an additional major/minor in Robotics at Carnegie Mellon University next Fall. I still don’t know if that will be enough.
Yet…although hard work doesn’t guarantee success, success can never come before hard work.
So I must try.
On that slightly depressing note I’m sorry to leave you with…this might be my last blog for Senior Project! I hope you can make the upcoming symposium and check out all the projects my peers and I have been working on!
Bye for now.
Chloe Zhang