 Does it make sense to produce chemicals using microbes? Well, the chemicals you choose matter a lot. Hi, I'm Tony or Wenja Wu. I'm a PhD candidate from the University of Wisconsin-Madison, and I'm going to introduce to you our work on identifying technoeconomically promising bio-based chemicals. This work is in the general area of microbial conversion. We know that biotechnologies today are quite advanced. We can use metabolic engineering tools to genetically engineer microbes like E. coli and E. yeast to produce a lot of different chemicals that are currently produced mainly from fossil fuel feedstocks like petroleum. And we call these products produced through such biological routes, biobased chemicals. Researchers in this community have been hoping to see such bio-production processes replace a portion of the traditional processes for environmental benefits and also potentially economic advantages. But there's a fundamental question here, which is which of these chemicals should we target? What are the promising ones to be produced biologically? This is the question we answer in this work. Now, there have been a few studies in the past that tried to address this problem, but most of them focused on identifying platform chemicals that could be converted to final products later through chemical conversions. But in fact, a lot of those chemicals can be produced directly by microbes without having to go through the platform chemicals. In other works, the focus was on studying a few interesting chemicals that currently attract a lot of attention. But we want to consider a much bigger pool of candidates. And also in most of the past studies, discussions on the economics have been quite limited, especially the downstream separation cost, which was rarely considered. But the separation process that converts a very dilute product stream into high purity final product is actually very expensive. In many cases, that accounts for 60-80% of the total production cost. Therefore, it needs to be considered. So these are the additional challenges we address in this work. The way we do it on a very high level is we start from database chemicals, and then we find out which ones can be produced through microbial conversion. And finally, we develop a few screening criteria to identify the promising chemicals. More specifically, we start from 691 database chemicals, which is the intersection between HPV, or high production volume chemicals, which include all commodity chemicals and a portion of fine chemicals. We intersect this with the biochemical databases like keg and metasike, where all chemicals that exist in any biological systems are collected. And the intersection here is 691. Then we develop a set of metabolic models based on flux balance analysis, which essentially uses mathematical equations to represent all the reactions that could happen inside a microbe. So these models can be used to identify from the database chemicals which ones can be produced by microbes like E. coli and yeast. There turn out to be 209 of them. Another use of the metabolic models is to estimate two very important parameters. One is yield, which is how much product we can get per unit of sugar we feed to the microbes. And the other parameter is residence time, which represents how fast the reactions happen. These two parameters are then used to estimate the bioconversion cost based on this industrial scale fermentation process we developed. And this cost is then used to develop the first of our three screening criteria. Specifically for each chemical product, its current market selling price has to be greater than this bioconversion cost to be economically feasible. And this has to be true because at this point we haven't even included downstream separation cost yet. The product stream we get here is actually very dilute. It could have only say 1% product, but in the end we want a product that is 99% pure, for example. The other two criteria are developed based on market data. This second criterion enforces a minimum market volume or demand because we do not want to build a tiny fermenter to serve a very small market. So in this example, we set the minimum threshold to be 20% of the biorefinery size proposed by NREL, National Renewable Energy Lab. The third criterion is based on market size, which is the price times the market volume. And this has to be large enough to meet revenue expectations to, for example, recover capital investment. In this example, we choose a minimum of $5 million annual revenue. And also we assume that we can replace the product currently sold on the market because our bioprocess has cost or environmental advantages, but we can replace no more than 20% of the current market. That's our assumption. So this translates to $25 million market size. This way, we identify 45 candidate chemicals that satisfy all the three criteria. Next, all we need to do is include the downstream separation cost. Once we have that, we can modify criterion one a little bit. And now we say the price has to be greater than the bioconversion plus separation cost. Essentially, we're enforcing a positive profit margin here because it doesn't make sense to adopt a bioprocess if it's not profitable. Now, how do we estimate the separation cost for all the 45 candidates? This is actually done by using a method we've been developing in the past few years. Essentially, we considered all the 17 most common separation technologies like centrifugation, filtration and distillation, as well as all the possible ways that these technologies could be used together to achieve a separation task. So all the possibilities are embedded in this so-called superstructure. And then with mathematical optimization methods, we can identify the cost minimal process for any product. So with this method, we further generated a set of curves that express the separation cost as a function of the product tighter or the product concentration after bioconversion, but before separation. And we have a curve for each of the five classes of products. For example, this one is for a product that is extracellular, which means it is excreted to the outside of the microbial cells, and it's also insoluble in water and lighter than water in terms of density. Now, how do we use the curves? From the metabolic models in the previous slide, the tighter for each product can be estimated. So we can simply read the corresponding cost from this curve. And then we incorporate that into the modified criterion one to finally identify the promising chemicals. And here's the results. And we're only showing five examples here. You can see for each product under different production conditions, like maximum yield or maximum productivity, is it promising, which means does it satisfy all the screening criteria? And what is the profit margin? We also estimated the brick even tighter that would render zero profit margin for each product. So metabolic engineers and microbiologists could look at these values and think, okay, I'll have to engineer my microbes to achieve these numbers in order to meet the minimum economic requirements. We'd also like to explain what promising really means here. It does not mean that industrial scale production today would be profitable. Instead, it means that it can be profitable if reasonable advances in metabolic engineering and separation technologies are achieved in the future. And this is actually why we have deliberately made optimistic assumptions such as maximum yield or productivity and minimum separation cost, because we do not want to cut off any products. But if you're interested in a more accurate technoeconomic evaluation of a specific product, then you could consider this list as a starting point. And then you could try to find experimental data, for example, to determine yield and residence time for cost estimation, and then perform a detailed process synthesis. This work is more intended for early stage analysis. Finally, I'd like to mention the flexibility of the framework. There are a few things that could be modified to cater to your specific scope. For example, the market price and demand data can be updated. The estimated yield and residence time can be changed. You could modify the threshold values for the screening criteria or even include additional criteria. And also the bulk conversion and the separation costs can be estimated in different ways. So all of these are customizable. So does it make sense to produce chemicals using microbes? From this work, now you see the chemicals you choose really matter a lot. In the end, we'd like to thank our sponsors, the U.S. Department of Energy through the GLBRC Center and National Science Foundation through the EFRI program. Thanks for watching.