Launch HN: Phase Biolabs (YC W22) – Converting CO2 to Carbon-Neutral Chemicals
We’ve built a lab-scale prototype that is a 1.5L bioreactor with a microorganism inside that 'eats' carbon dioxide and hydrogen gas, converting them into chemicals as it grows. Here's a demo video I just made for HN: https://www.youtube.com/watch?v=RUIT3RUeUPE. We are currently making ethanol in the lab but unfortunately CO2-based ethanol cannot be legally sold as a beverage, so industrial solvents it is :)
You can do two things with carbon: you can capture it, or you can use it. Both are hard, but the latter is harder, mainly because carbon dioxide is so small.
Capturing CO2 is usually done by attaching it to something else, usually another molecule, which is how we can extract it from a dilute gas stream or 'pull it' out of the air. But the CO2 molecule is only temporarily transformed.
Using CO2 is a different ball game, usually referred to as CCU (carbon capture and utilization). For this you need to permanently convert the molecular structure itself, and since you are working with extremely tiny pieces of matter, you need extremely precise machinery.
The challenge with converting CO2 is doing it efficiently. It needs to happen with as little energy as possible and to be as precise as possible. If you want to convert CO2 into X, but you also produce Y, and Z, that is a problem which will show up in the cost. Our solution is bio-based CCU, but there are also electrochemical and thermochemical technologies, each with advantages and disadvantages. And there are other bio-based approaches, such as making trees more efficient (e.g. Living Carbon W20). All are valid strategies.
In biology, CCU is known as carbon fixation. During my PhD I was engineering microbes to convert wastes into renewable chemicals and fuels, so I began to study biochemical carbon pathways, which led me to carbon fixation. I began to realize how important carbon fixation is at a macro level (carbon cycle) and how the process works, but also that it is extremely inefficient and can be optimized. For example, the trees in your garden don’t grow very fast. This is due to photosynthesis being 2-4% efficient. I’ve always wanted to start a startup and that has always been in the back of my mind, so I did things that I enjoyed that could also help towards reaching that goal, which led me to this.
Advances in synthetic biology mean we can do things that weren't possible 20-30 years ago. The amount of tinkering that we can do has substantially increased (and costs have dropped), and our understanding has grown due to a rise in data and analytics. We can borrow strategies that have worked in the past in other fields and apply them in new ways.
Since biological carbon fixation is precise, but very inefficient, our approach is to take that precision and enhance it using synthetic biology into a process that is efficient, scalable, and productive enough for industrial application. We're using microorganisms that can naturally fix carbon, and transforming them into mini factories. Our microorganisms are 7x more energy efficient than naturally occurring plants or algae and in theory can produce almost any molecule found in nature directly from CO2.
Carbon fixation is catalysed by a carbon fixation (biochemical) pathway, which is simply a set of enzymes that catalyse a sequence of steps/reactions. The enzymes attach electrons and hydrogen ions onto the CO2 molecule, while removing the oxygen, one step at a time. This process can be called reverse combustion, but whereas combustion is uncontrolled and explosive (litera...
90 comments
[ 3.1 ms ] story [ 181 ms ] threadI have tons of questions, but my first is: How do you measure the products that come out of your bioreactor?
https://airminers.org/
Saw some talks years ago by Daniel Nocera, in which they did water splitting and fed hydrogen to engineered bacteria to make more complex products. Cool stuff - I'd imagine the end products, fuels in their case, would still be way too expensive.
There's interesting stuff going on in modifying microbes to make cannabinoids and other natural products. Would be kinda cool to combine these two :)
I agree, Daniel Nocera's team have done some really cool research and you're right that renewable fuels in general would be too expensive. I would argue that the cost of fossil fuels are artificially low because their current cost doesn't account for the environmental damage they cause. If we had to pay their true cost, the difference might not be so great.
I like to dream and so our long term ambition is to try and make complex molcules like cannabinoids directly from CO2, but we're some ways off.
We need to campaign to make these costs clear and keep companies accountable
Is there a possible future where these critters could eventually turn a clean CO2 source into recreationally drinkable fluid?
And yes they could convert CO2 into recreational drinking fluid, and I learned from someone else in this thread that it would be totally legal!
I imagine you would have to be source pure inexpensive green hydrogen in order for the CO2 balance to be net negative/neutral if that is a goal.
Best of luck - always glad to hear about novel approaches/companies.
The stoichiometric equation is:
2CO2 + 6H2 > C2H5OH (ethanol) + 3 H2O [∆G -107.4 +/- 36.8 kJ/mol]
You're absolutely right, we require low cost green hydrogen (electricity). The carbon balance depends a bit on what kind of carbon you use.
Is the core innovation for your company around the bioreactor structure or the chemistry? I.e. are there other chemical reactions that you are looking at building on?
Core of the innovation is around the engineering of the microorganism. One way to think of it is that it is similar to chip designs. In the 80s (I think) ARM designed chips that had super low power consumption. They patented that design and those chip designs are why we enjoy better battery life on our devices today. We have a similar approach in terms of where our IP resides.
In terms of chemical reactions, we can in theory produce almost any chemical compound found in nature directly, and all chemicals in multiple steps. Our process is anaerobic so we can't do reactions (yet) that require an oxidation step. For context, there are more than 200,000 organic compounds found in the biosphere.
We are going to need low cost green hydrogen. Existing electrolyser technology uses precious metals and so they are expensive/current limiting factor. However there are lots of startups working on bringing to market electrolysers that use non precious metals.
In terms of chemical processes, fermentation is quite simple. In addition to the bioreactor you need a device to control the gas inputs and you need some downstream processing (product extraction). For ethanol distillation is most commonly used but there are newer lower energetically demanding techniques that can be used for extraction. Not all products can be distilled out though, so extraction is somewhat product dependent.
edit: some interesting hydrogen companies
https://www.sunhydrogen.com/technology (like a solar panel but for hydrogen, uses light energy to split water into H2 + O2)
https://www.alchemr.com/technology/ (electrolysers that use non precious metals)
https://www.h2pro.co/technology (membrane free electrolysers)
There are also several companies developing 'turquoise' hydrogen, which is a plasma based technology. I have no connection to this website, but the first few paragraphs it lists a few companies in this space: https://www.h2-view.com/story/four-more-technologies-for-tur...
1. How efficient is this process, i.e. how much ethanol does your lab setup produce per litre of reactor volume and hour?
2. Is the need of hydrogen a problem, i.e. is the required amount of hydrogen for real world usage at scale easily available or would this require significant additional hydrogen production capacity?
EDIT: Number two has mostly been answered in another comment.
Your other comment relates to productivity and can be expressed as space time yield (STY). Like the above we need to improve on this and is part of our next steps.
It seems to me that the energy content of the Ethanol isn't too different from the natural gas going into the plant so I don't see how you get ahead doing this... If you've got to add extra energy or divert some of the input energy to make hydrogen to fuel the reactor how do you end up ahead?
So first, if you need to do 'work', then definitely just use electricity directly to do the job, which is much more efficient.
I agree it wouldnt make sense to split natural gas (into CO2 + H2) and then recombine it back into ethanol, although oil companies would love to do this as they have billions in stranded assets in the form of natural gas.
Ideally you couple some process that generates CO2 (not from burning fossil fuels) with renewable electricity to recycle that carbon back into useful chemicals to displace petroleum derived chemicals. Two examples of this would be cement manufacture and industrial brewing. But yes you need an external energy input, like with most things.
As a side note, the impact of this depends on where you get your energy (renewable of course) and your carbon. Some companies have caught onto this. For example Unilever created a carbon 'rainbow' to separate the types of carbon. Recycling renewable carbon is the goal here.
Pardon me but this doesn’t sound right. If you want to generate heat, burning natural gas is going to be a lot more efficient overall than first burning natural gas at a power plant then transmitting electricity to your facility to convert it to heat. Similarly with rotational energy, etc. Your second point stands: if you power your process by solar, wind, or hydro you could get ahead of CO2.
What are your thoughts on using renewable electricity for heating applications as a way to displace burning of fossil fuels?
Solar can also be used to heat things more directly than first converting to electricity (and taking a big loss on that), but then you really are subject to when the sun shines. But if you put your facility in a desert in Arizona you’ll probably do quite a bit with a set of mirrors used to heat specific objects.
For me that has alwayd been the hard part to understand about CCU, where is there a market large enough to absorb that volume? And where the product does not get burnt or emitted anyway in the end?
Even though the technology is on the shelf it's not being deployed largely because there is no financial incentive to do so... Yet the widespread use of this technology really needs to be happening now if we want any of these carbon capture things to happen.
But yes we really need better systems to incentivize the capture and storage or utilisation of CO2. Carbon taxes a great place to start.
As we find new profitable uses for CO2 the demand for it will increase, which should help create new carbon value chains.
If you do some back of napkin math on petrochemicals, which accounts for roughly 20% of oil usage there is a huge opportunity to displace petroleum using recycled carbon.
Global oil consumption is roughly 100 million barrels / day (today). 1 barrel is 160 kg, so annual petrochemical volumes are roughly 1.1 billion tons of product (20 million * 160 kg / 1000 (to get tons) * 365 days). That is at todays consumption. Chemical usage is expected to grow over the next several decades. Of course this is ignoring recycling carbon into e-fuels. There will be a need for those too.
In terms of actually scaling the technology, heavy industry is widespread and is a source of large scale point source emissions, ranging from as little as 10,000 tons of CO2 emissions / year all the way up to 10 million tons of CO2 / year. It is all about retrofitting these industrial sites with this type of technology to supply local markets the chemicals they need. This ignores the other sources of carbon that will become available via carbon capture (stationary or mobile) as well as direct air capture. It's tough to imagine exponential growth, but things can be very different by 2040.
https://www.chemistryworld.com/news/reprogrammed-bacterium-t...
Because you're here, and partly out of vicarious self-interest (my kid is about to graduate with a biochem bachelors and is looking around at industry work before heading back to grad school):
What's the day-to-day work here look like? I have a good idea of what most software companies, and even a lot of hardware companies, do. I have no notion of what line engineers and tech workers at a company like this are doing. Is it mostly modeling work? Materials engineering stuff? What's the software stack here, if any?
If you are working in analytics, running assays (tests) on things like blood or urine samples (hospitals or clinical trials) or a more recent example would be a covid clinic, the day can be very monotonous. There is a lot of paperwork involved due to the regulations you need to adhere to (GMP, GCP, GLP). This is one of the reasons I didn't like working in pharma. It's better now to things becoming digital, but the point is the work can be very repetitive.
If you are working in an R&D lab things are more dynamic. You might be running similar experiments from one day to the next, but the context is always different. Even though you hit a roadblock and get stuck for a day, a week or a month, as things progress the type of work will change as the project evolves/progresses.
You can work in industry in either of the above environments, both provide valuable experience. Industry is stricter and more rigid than academic labs.
Day to day it's still very hands on. Things are progressing such that you spend less and less time in the lab as things become automated and the workflow becomes digitised, but you still need to go into the lab even if it is to setup the robot. We don't yet have robots to control the robots, although maybe sooner than we think. At high level, most R&D lab employ some sort of design, build, test, learn (DBTL) workflow, even if they don't call it that. Depending on what the focus is, each step in that cycle will be slightly different.
The amount of software is growing every day for all applications. You have everything from basic software like Lab Information Management Systems (LIMS) to help with basic ops to more complex software to help plan workflows and analyse data (Synthace) to much more specific software like protein modelling (Rosetta) or genetic manipulation (Geneious) and the list goes on. I am barely scratching the surface here. I regret not having more training in python.
edit: not a perfect article, but to give you more of a flavor for software in synbio/biotech, check this out: https://www.builtwithbiology.com/read/the-synbio-stack-part-...
Depending on what you want to do you build a different style of plasmid. If its genetic modification (ex. using CRISPR) you use one type, if it's testing a new pathway, you build another. You use software to help with the design of everything and to define and explore the solution space.
To make it high throughput we usually test things using in vitro (cell-free systems) before actually moving into the host. In vitro work has a faster DBTL cycle than in vivo work. We test strains in smaller experiments (20-100 ml) before moving to bioreactors (1-2L).
We would like to automate more and build a more robust R&D pipeline to support faster DBTL cycles, but you can be limited by the epuipment available. Doing highthroughput automated work is great for productivity, but it costs more. So has been challenging to implement everywhere we would like due to resources.
Maersk has commissioned 8 ships to run on methanol. For context, Maersk owns 550 ships. Gives you an idea of the size of the transportation fuel problem.
Business issues aside, micro-organisms always produce a lot more stuff than what you want, and they behave differently depending on the micro-environment, and of course they mutate...
It's kind of hard to go up against DARPA projects and national labs that have been doing this for decades.
This is one of the challenges we face against companies that are spinning out research that has been publicly funded for many years or on a more personal note, going up against people who came from more prestigious institutions. But we think we have identified a niche that is worth pursuing.
I think Zymergen is an interesting case study and serves as an example to companies developing 'new products'. Like most things there is no perfect solution. New products open new markets, new opportunities, and may seem less risky at the beginning, but what happened to Zymergen is an example of what can happen when rolling out new products (in this space of course). Drop in replacements for example don't face those same risks, but they have other challenges of course.
Why not?
Edit: A little searching here seems to say that synthetic ethanol can be used as a food ingredient and must be accurately labeled. The policy doesn't say that it can't be used in beverages. However, I assume that it's use in many beverages are defacto banned as synthetic alcohol is not part of the tradition grain bill. I don't see why it can't be used in the liquors or a new type of beverage.
https://www.fda.gov/regulatory-information/search-fda-guidan...
Another edit: it appears that FDA lists ethanol as GRAS for general food product use, and that TTB approves the use of GRAS ingredients. If you use synthetic flavors, it looks like that affects the labeling. I assume you would just have to label the alcohol consistent with it's source.
This explains how Air Company is selling their ethanol as vodka.
We don't normally think of the containers where wine or sour mash are fermented as bioreactors, but they are.
> You can do two things with carbon: you can capture it, or you can use it. Both are hard, but the latter is harder, mainly because carbon dioxide is so small.
You can either store (CCS: "Carbon Capture and Storage") or/and use (CCU: "Carbon Capture and Utilization") it.
Fixing the typo:
> You can do two things with carbon: you can store it, or you can use it. Both are hard, but the latter is harder, mainly because carbon dioxide is so small.
Also, probably wouldn't say that it's hard because CO2 is small. Instead, probably just say because [it's got a low enthalpy](https://politics.stackexchange.com/questions/30089/why-is-th... ). Otherwise, it'd be like saying that water's hard-to-use because H2O is small, or that gold's hard-to-use because Au is small.
Hah.. yeah.. I like to remain anonymous, so it's difficult to discuss my own background and experience directly, but I fully get what you mean.
I've stood in front of conferences at places like AIChE and in front of the upper-staff at some of the world's largest companies and delivered talks on topics that I consider to be the basics of this field, and there seemed to be surprisingly superficial comprehension. It was weird, and communication seemed like a huge barrier.
A lot of work in this field may be more social in getting folks to accept and support stuff.
I've been interested in if we could use large bodies of water (like lakes or ponds) to both get a lot of interaction with CO2 from the air, and get energy from the sun to capture carbon. Do you think fermentation could work outside, or would we need a more resilient organism like algae or even seaweed?
My idea is to float a dark membrane on top of the water, with a thin layer of water on top. The dark membrane heats the water, sustaining the algae, and then the algae get robotically harvested. Could that work for your organism?
If you use an electrolyser to make hydrogen from water, your energy source is actually the electricity used to drive that reaction. The energy is stored as hydrogen gas.
Large bodies of water are great passive systems to capture CO2. For the outdoors, and if using sunlight is the energy source, algae and seaweed are definitely great candidates for capturing CO2.
Unfortunately your idea wouldn't work for our microorganisms as it is anaerobic, so it would die if exposed to air. But I like your idea for using sunlight to improve the growing conditions of algae + using automation for harvesting!
The more immediate impact for CCU is the emissions reduction achieved via displacement of fossil derived solvents.
Like traditional chemical processes which use metal catalysts, superior catalyst design improves the performance and ultimately the economics of the process.
Edit: I should have mentioned that it's not just the catalyst that has been improved, the design of the process itself has been improved. So upstream (gasification) and the design of the bioreactor also impact how well the process works.