The U.S. Department of Energy’s Bioenergy Technologies Office (BETO) hosted a two-day webinar series, Biocarbon Incorporation into Transportation Fuels via Co-processing in Refineries, highlighting key takeaways from the Bio-oil Co-processing with Refinery Streams project. Below is a transcription of, “Webinar Day 2: Biogenic Carbon Tracking and Measurement in Co-processing of Biogenic Feeds in Petroleum Refineries,” held on September 27, 2023.
.....
Simone Hill-Lee, BETO
Good afternoon. This Zoom call is being recorded, and may be posted on DOE's website, shared or used internally. If you do not wish to have your voice recorded, please do not speak during the call or disconnect. Now, if you do not wish to have your image recorded, please turn off your camera and participate only by phone. If you speak during the call or use a video connection, you are presumed to consent to recording and to the use of your voice or image. We'll be getting started momentarily. We have just a few housekeeping reminders before our program gets underway. During this presentation, those of you in the audience will not have the ability to unmute or to turn on your camera. And since this is the case, we encourage you to submit your questions, using the Q&A function. You should see an icon either at the bottom of your screen, or perhaps at the top. As the chat functionality is disabled for this program, we would encourage you, even if you're having any technical issues to submit those issues to the Q&A and we'll do our best to answer those and help you. Please feel free to submit questions at any point during the presentation today, and every effort will be made to address all relevant questions. So, thank you for your attention at this point, I'll turn the program over to my colleague, Dr. Robert Natelson.
Robert Natelson, BETO
Thanks Simone. Today's webinar will be on Biogenic Carbon Tracking and Measurement and Co-processing of Biogenic Feeds in Petroleum Refineries.
I'll now provide some brief overview of BETO, the funding agency, and then we will dive into individual talks on biogenic carbon tracking strategies from the DOE national laboratories.
The work presented today is funded by the U.S. Department of Energy Bioenergy Technologies Office. BETO is an applied research and development office that partners with the national laboratories, academia, nonprofits, and industry to support pre-commercial research.
BETO’s mission focuses on reducing greenhouse gas emissions in the transportation and chemicals sector by using the nation's biomass and waste resources. BETO’s goal is efficient and scalable technologies, for low cost and low-carbon intensity, sustainable aviation fuel and other strategic fuels and chemicals. The strategy has been published in multiple documents this year and they are listed here and widely available online, including from the BETO website. BETO also funds the Billion Ton Study, and I'm sure many in this audience know there's been rapid growth in the use of vegetable oils and biogenic fats, oils and greases on the order of millions of tons of feedstock to produce billions of gallons of biofuels, often in a co-processing configuration.
But conventional wisdom says these feedstocks will not be enough in the coming years. New feedstocks and conversion technologies need to be unlocked to meet the demand for low-carbon density, fuels and products, and one example of these new feedstocks and new conversion technologies we heard about last week is the co-processing of bio-oils and bio-crudes from woody biomass and sewage sludge, which we heard about last week from Reinhard Seiser from the National Renewable Energy laboratory, and Huamin Wang from Pacific Northwest National Laboratory.
Reinhard mentioned that part of this multi-laboratory team project is to explore R&D related to the regulatory risk in co-processing and that means measuring the biogenic carbon in co-processed fuel. Methods are available. But there are other methods under research and development, with possible advantages or disadvantages. So in today's talk, we'll hear about research and development, and some of the alternative methods.
BETO is organized into four program areas, renewable carbon resources, conversion technologies, systems development and integration, and data modeling and analysis. Today’s work is supported by the Systems Development and Integration program. Together these programs work towards supporting the BETO goals, including the production of three billion gallons of SAF or sustainable aviation fuel by 2030, supporting 35 billion gallons of SAF by 2050, and to support methods that do this cost effectively, and with greenhouse gas reductions of at least 70% compared to the conventional alternative. Before we dive in, some additional background, the audience today is mostly biofuel R & D and petroleum refining engineers and scientists. The speakers are coming from training in geology, and they have applied expertise in physics and chemistry to research methods in biogenic carbon tracking, related to biofuels, and also this application requires understanding of the carbon cycle of different carbon isotopes, including the plant biology, knowledge of biological carbon and fixation. So all to say that this is truly very multidisciplinary work. Everyone is coming from their own point of view and part of the purpose for me today is to figure out a frame of reference. Nevertheless, I'm very excited to hear our speakers today, because it always makes me feel smarter when I hear them talk.
The first speaker will be Dr. Zhenghua Li. Dr. Li is a senior scientist from the Los Alamos National Laboratory, or LANL. He serves as the team leader for the isotope and analytics team. His research interests includes isotope, geochemistry, analytical chemistry and research in renewable biofuels and clean energy, climate change and carbon sequestration. He has accumulated over twenty five years of experience in geochemistry and wet chemistry. Over the past six years he has actively contributed to the advancement of renewable carbon training tracking techniques,for the co-processing of bio-oils and bio-crudes for petroleum feedstocks in refineries. Doctor Li, the floor is yours.
Zhenghua Li, Los Alamos National Laboratory (LANL)
Thank you, Robert. Hello, everyone. Today I'm going to talk about the biogenic carbon tracking and measurement by LSC 14Carbon and IRMS d13Carbon in co-processing. The LSC stands for Liquid Scintillation Counting. IRMS stands for Isotope Ratio Mass Spectrometry.
This diagram is showing you how bio-oil is co-processed with VGO through FCC or hydro-processing to produce four products. There are gas phases, aqueous, liquid hydrocarbon, and the coked catalyst.
So the goal of co-processing is to maximize the biogenic carbon incorporation into a useful product. Here is a liquid hydrocarbon. We don't want the biogenic carbon to get into a gas phase, aqueous or coked catalyst. So to track the biogenic carbon distribution among the products, currently, the gold standard is to use the AMS 14Carbon method. The AMS stands for Accelerator Mass Spectrometry, which is an ASTM method 6866.
So our task in the co-processing project is to develop a simple, fast, less expensive method with a capability for online biogenic carbon measurement.
So in the next few slides, I'm going to talk about the LSC 14Carbon method of development. I want to mention, this is team work, today I’m just a presenter.
This slide showing is a background slide, showing you how the radio14Carbon can be used to quantify biogenic carbon contained by LSC. The diagram on the left, showing you that the radiocarbon is formed by cosmic rays through neutron reaction with 14Nitrogen isotope in the atmosphere.
After 14Carbon is formed it is immediately reacted with oxygen to form CO2 gas, and the CO2 gas is fixed by trees or by vegetation through photosynthesis. As soon as the CO2 is fixed by trees, the 14Carbon starts to decay with a half-life of 5,730 years because the 14Carbon is a radioactive atom.
So people use the 14Carbon contained in modern plants as a reference is defined as 100 pMC, is a percent Modern Carbon. It's a proxy of 100% biogenic carbon.
So the older plant, like the dead tree, or dead vegetation contains less 14Carbon due to the radioactive decay. Fossil fuel contains zero 14Carbon because it is a million years old.
So when we mix the bio-oil with the crude oil by measuring the radiocarbon decay activity, (a) we can tell how much bio-oil is mixed with crude oil. So the diagram on the right is showing you how the biogenic carbon is calculated. Let's say, if we, if the biofuel is made from modern plant, the 14Carbon container should be 100 pMC, which is 100% biogenic carbon. The decay activity is 13.56. So for an unknown sample, if we obtain decay activity rate for 8.0 minutes per gram carbon, so we can calculate biogenic carbon percentage using these equation. So this equation calculate, if we get an 8.0, then we can get the 60% biogenic carbon. So how this works, you can see the X-axis is a biogenic carbon fraction, Y-axis is made of radiocarbon activity.
So, for fossil for petroleum, the biogenic activity is zero percent. But there's a some activity over here which is a background activity. And then for modern plants is 100% biogenic carbon, so the activity is 13.56.
If we got the number over here, let's say 8.0. So use this line, go this direction than go down so we can calculate how much biogenic carbon in these unknown samples is how things work. So, the radioactive decay activity A can be measured by using a liquid scintillation counting LSC. So this LSC, we have three types of LSC in our lab, so they are pretty small in terms of footprint, since this is a liquid scintillation counting. That means all the samples need to be converted into liquid like a solid sample, biomass or gas, need to be a liquid phase. So for liquid fuels, we mix the fuels with a scintillate, also called the cocktail. The simulator produces photons, when 14Carbon decay occurs. The number of photons immediately are related to energy of decay products. The detector, also called the photomultiplier tube, detect lights.
But there are some complications, the scintillater itself can produce the photons. So this is called a background, and also the photon can be blocked by color or biochemical compound. But if by color, we call it the quenching effect, if biochemical, by color itself, its called the counting efficiency problem. So these two problems we have to deal with if we use the LSC Method.
So, we have developed three LSC methods for three types of samples. If the fields samples are clear or near clear, we can use the direct measurement method. We have an optimized data acquisition range of interest window counting times and other optimized parameters.
Our method can achieve 0.4% uncertainty compared to ASTM 6866, which is the 14Carbon AMS method in the range of 1 to 10 percent biocarbon in the samples. I will talk about this in more details in a minute.
If the field samples are lightly colored, we have developed the decolorization method, but I want to mention decolorization, removes some biogenic carbon in some samples. But I will talk little bit more about this in a minute.
So then the next thing is, if the samples are dark or a solid sample, there's no way we can use direct measurement or use a decolorization method. So we have developed a CO2 conversion method. We convert the sample into CO2 through combustion, and then use the amine to capture CO2 in a liquid phase, and then mix the amine with the simulator for LSC 14Carbon counting.
So here's more about the Direct Measurement. These are the results of from Direct LSC Measurement. We tested the different fuel samples, they are gasoline, jet fuels, diesel, and B100. This diagram is showing you a good correlation between LSC and the AMS result.
The arrows are very small. You can see it's almost, it's just a plus or minus 4. You know, this is showing the deviation from the AMS result from LSC that is pretty small. The Direct Measurement is a no-prep approach, which means that we don't need to prepare a sample for LSC measurement. We just directly mix the fuel sample with the scintillate directly and put it on the machine to measure the biogenic carbon content. So this is the faster way to measure the measurement. So this method has been published in the field.
This time, I'm showing you the attribution of uncertainty. The Y-axis is the RMSE, it stands for Root Mean Square Error. The X-axis is the counting time. So what does this diagram mean? Basically it shows you the longer counting time, the smaller uncertainty. Also, the uncertainty depends on the fuel properties. Dark samples show less counting efficiency, and then larger uncertainty. The high counting efficiency results in low uncertainties, so pretty complicated. You know, the clear sample is the best. Of course, the last example caused a lot of issues in terms of the counting efficiency and the quenching factor. So here is a more about the decolorization method. This slide is showing you a diesel decolorization. These are commercial diesel that we purchased from a gas station that label that D as the petroleum diesel B20, B100, and you can see they are pretty colorful. We investigated different techniques, we included adsorbents, and used the catalyst that break it down, some larger compound oxidation method that you use on ozone oxidation. We found that the adsorbents with the clay and silica gel, including the aluminium oxide, was the most effective. Showing you the diagram on the right that the decolorization reduced uncertainty without affecting biogenic carbon interpretation for B100, but they are decoloring preferentially removed biogenic carbon in B20. So, this is related to some compound like an aromatic molecular in the B20, which contains more biogenic carbon in their compounds that will cause the problem. But, we don't expect the blended, co-processed samples to be affected by this artifact because co-processing homogenizes biogenic carbon among compounds. So, for the dark samples or solid samples, there's no way we can use the above two methods, as I mentioned, to measure biogenic 14Carbon using LSC. So we have developed the CO2 conversion method. To quantify biogenic carbon in a sample, our approach is to develop biogenic carbon calibration line. We use two tanks of CO2, which is for different, very different biogenic carbon content. Tank 1 CO2 has 2.7% of biogenic carbon and Tank 2 CO2 has 96.2% of biogenic carbon.
So the diagram on the upper left showing you an excellent correlation between absorption of the carbon mass and measurement of 14Carbon counts per minute for Tank 1 CO2. This means that the CO2 can be quantitatively absorbed by amine at the different CO2 saturation rate.
The data point within the dashed line represents the CO2 saturation rates are greater than 80%. So the diagram on the bottom left, showing you a good correlation between absorbed carbon mass and the measured 14carbon counts per minute for Tank 2 CO2. Again, the data point that is within the dashed line represents the saturation rate larger, greater than 80%. So, the diagram on the upper right is LSC 14Carbon spectrum of Tank 2 CO2, which contains 96.2% biogenic carbon. These four spectrums are corresponding to four vials containing various amounts of CO2.
The black line is a background 14Carbon signal. The blue line is the CO2 14Carbon signal, because Tank 2 contains 96.2% biogenic carbon. So the 14Carbon signals today are very strong. So this sample is very easy to measure if the biogenic Carbon contained is high.
Our method indicates a very low 14Carbon signal like 0.1 gram carbon with a 2.7% biogenic carbon can be meaningfully detected through CO2 amine captures.
So we have to come back to a biomass combustion experiment to verify the calibration line using 14Carbon standard IAEA-C3 cellulose, which is an international standard.
IAEA-C3 cellulose, which is a biomass content 129.41 pMC. So the diagram on the left is LSC 14Carbon spectrum. The diagram on the upper left is the background spectrum. This sample content zero CO2 is just to simulate. We didn't put any CO2 in this bias just used as a background.
The diagram on the bottom left is the IAEA-3 CO2 14Carbon spectrum. That's 0.87 grams CO2 absorbed in this vial, so you can see the signal is very strong even if it is a small amount of carbon CO2 absorbed in the blue line versus the black line. So using our calibration curve we obtain a value of 130.87 pMC which is very close to the published value, as you see in the table on the right. So, we also have to carry out two CO2 gas mixing experiment. We mix Tank 1 and Tank 2 CO2 at a different ratio. We got a very good result and that result was very close to the expected values. I want to mention here that the CO2 conversion method works for all type of samples, fuel samples, a solid sample, clear, darker, whatever, as long as it can be combustible, converted into CO2.
So our future work will be included in the test fuel sample with low biogenic carbon content, and reproducibility. Reduce the instrument background noise, because at Los Alamos, the background, it is almost 3 cpm, and there are some labs at very low elevations that are at 1 cpm, and that is very good. We also like to develop a high capacity CO2 absorption medium. I think this is all about the LSC Method development.
My next topic is about biogenic carbon tracking through IRMS d13Carbon analysis. This is a background slide about how the d13Carbon can be used to track biogenic carbon in co-processing. Using 13Carbon to track a biogenic carbon is based on the difference of a d13Carbon value between the feedstocks, here is bio-oil and VGO. So the diagram on the left is showing you the d13Carbon range of natural materials. As you can see, the d13Carbon values of C3 plants are very close to VGO, but the C4 plants are very different from VGO, far away from VGO. This means the C4 plants derived bio-oil has a better traceability for biogenic carbon tracking.
The center diagram is showing you the carbon atmospheric fractionation during C3 plants for photosynthesis. The atmospheric CO2 d13Carbon is minus -8o/oo. During the photosynthesis, there are two major carbon values as fractionation occurs. This results in the d13Carbon value in the final product biomass to be around -26o/oo.
The picture on the right is showing you an IRMS instrument for 13Carbon analysis. As you can see, this environment has a very small footprint, and it's relatively easy to use.
So what is the sensitivity for using d13Carbon to track biogenic carbon distribution in co-processing with VGO? The diagram on the left is showing you a model, the sensitivity for blending the C3 plants and the C4 plants bio-oil with VGO, which has d13Carbon equal to -30o/oo. If you, if we blend, let's say, if we blend a 4% bio-oil with 96% of VGO, the change of d13Carbon value is, in blend, is very, very small, you know, you can see over here.
But for the C4 plant, its big, you know, is that the blue bar is much bigger. So this means C4 plant with VGO, it's easy to be tracked. So, in order to precisely determine the 13Carbon value for C3 plant derived bio-oil co-processing, we have derived a cold C method, and that's showing on the diagram on the right, because bio-oil is a volatile viscous oil sample, they are very difficult to measure due to their fractionation. So this is the method that is working very well and has already been published.
Another question we have addressed is what is the limit of a detection for using d13Carbon to determine biogenic carbon. So the limit of detection depends on the analytical uncertainty for C3 plants derived by the analytical method that we can detect .50% blending level, if the analytical data is -0.26 o/oo. For C4 plants, we can detect a .12% blending level, (-.13 o/oo), it's very, very sensitive. So the diagram on the right shows you a controlled experiment, where we mixed a very small amount of commercial diesel with a large amount of petroleum diesel, so we are able to detect a very well, very low blending levels over here.
So, we carried out multiple real co-processing experiments on blending levels to test the 13Carbon method. The diagram on the left is showing you the bio-crude and bio-oil that were co-processed with VGO, X-axis is a biogenic carbon content calculated based on the blending level. The Y-axis is a d13Carbon value of co-processing the product. This is co-processing, it's not just a simple blending. So this study reviews a significant correlation between biogenic carbon and the d13Carbon value in co-processing.
So, we have verified our d13Carbon result by comparing them with AMS 14Carbon analysis, and as you can see on the right, the result is very, very good. The diagram on the top right is showing you the correlation between AMS 14Carbon and d13Carbon, for the bio-crude sample showing a very good correlation with a very small error bias. The diagram on the bottom right, showing you the correlation between AMS 14Carbon and d13Carbon, for CFP forest residual samples. Again, very good correlation. But since there is just these three samples, the uncertainty is pretty big, you know. The error bar is pretty big compared to the bio-crude. So I want you to pay attention to the X-axis, for the bottom diagram, the difference of d13Carbon value between samples is so small, but we must still be able to obtain the meaningful correlation. This is very good. So another thing I want to show you is that here's a very good example to demonstrate that the d13Carbon analysis can be used to track biogenic carbon, a distribution co-processing. We have to co-process the biogenic carbon bio-oil with the VGO at same blending level, which is 9.7% but different co-processing conditions. Four co-processing conditions were tested. The d13Carbon analysis indicates that the reaction condition to produce more biogenic carbon incorporation. The [first] reaction condition one [1] produced the least biogenic carbon incorporation. So this was confirmed by AMS 14Carbon management as well, which indicates the 8% biogenic carbon in condition two 6% biogenic carbon in condition one [1]. This demonstrate that the 13Carbon is very useful and works well to track biogenic carbon incorporation in the co-processing. So all I was talking about was the qualitative correlation between the d13Carbon products and content. This slide is showing that we can actually quantify biogenic carbon by using 13Carbon method, after the fractionation factor is determined. These three diagrams are showing you the quantified biogenic carbon contained in three sets of a co-processing experiment. The green column is d13Carbon by biogenic carbon content. The red column is the AMS 14Carbon derived at the value, and the blue column is the yield base value. They are in a pretty good agreement. But the uncertainty for d13Carbon is big because we are dealing with a C3 plant. But you know for the bio-crude co-processing, unfortunately we need more. Yeah, this is pretty good if we can determine the fractionation. So the key takeaway for this IAMS method is that the the d13Carbon analysis can be a viable way to track biogenic carbon, but the feedstock must be available for a 13Carbon analysis. I think this is my last slide. So, with that, I'm going to turn it back over to Robert.
Robert Natelson
Thanks, Zhenghua. Our next talk will be on laser spectroscopy and stable carbon isotopes for tracking biocarbon. Next slide please.
Our next speaker today will be Dr. Sophie B. Lehmann. Dr. Lehmann is an Earth Scientist and leads the Stable Isotope Lab at the Pacific Northwest National Laboratory (PNNL). While pursuing her Ph.D. at The John Hopkins University, and as a postdoc at the University of Pittsburgh, she focused on climate and sedimentology and stable isotope geochemistry. Since joining PNNL, she has focused on application and method development related to light stable isotope analysis to track biogenic carbon to blended and co-processed fuels and feedstocks. Her work includes techniques in spectrometry and spectroscopy. Other research interests include the application of light stable isotopes to environmental science, biomineralization, carbon sequestration, and forensics. Dr. Lehman the floor is yours.
Sophie Lehmann, Pacific Northwest National Laboratory (PNNL)
Thank you. So, for this part of the webinar, we're considering optical spectroscopy for tracking biocarbon. Let me just get my laser pointer ready. So, like with IRMS, Optical Spectroscopy can also provide d13C values.
So why Optical Spectroscopy? It's been shown in other scientific fields that it produces precise and reliable d13C data. This instrument is field deployable, and it's even been taken up into airplanes and driven around in vans as a mobile system to collect data.
Through discussions, we've learned that there's interest from the Industry Advisory Board to have fast, reliable and inexpensive biocarbon trackers, and that biocarbon tracking instrumentation would ideally be near site or on-site and provide near real-time data.
Also, that an instrument would be able to track low ratios of biocarbon in co-processed fuels. As Zhenghua discussed, we're interested in stable carbon isotopes, 12C and 13C for tracking biocarbon during coprocessing. This is because we can rely on these isotopes to reach accurate results that we can get faster than counting radio carbon 14C.
Zhenghua’s stable isotope data, or the d13C, like he had mentioned, were produced using IRMS. What we want to do is to see if Optical Spectroscopy can monitor d13C as well as IRMS does for transportation fuels and for co-processed and blended fuels with low biocarbon ratios. In addition, we're assessing if this analytical capability can be adapted for refinery applications. Can Optical Spectroscopy provide near real time measurements from the refinery floor, for example.
The stable carbon isotopes that we're looking at are 13C and 12C. They're commonly discussed as d13C. d13C is a calculation that allows us to look at very small differences in the ratio of 13C to 12C for a sample and then compare that against that ratio for a standard. Essentially, what this is doing is providing a more user-friendly value to work with than these very small values related to the very small differences in isotope ratios. Zhenghua’s work identifies that d13C correlates with percent renewable carbon for our blended and co-processed fuel. Because d13C can provide accurate estimates of percent biocarbon d13C data could help us monitor and optimize incorporation of biocarbon during co-processing.
The d13C method is based on the difference in d13C between fossil and biocarbon feed stocks. Let's see here in the schematic. The d13C trends with percent biocarbon in a linear and predictable way between those two feedstocks. With increasing biocarbon, the d13C of a co-processed or blended sample, is going to move closer to the d13C of the biocarbon feed stock, and that will be along this mixing line. So, d13C measurements are made using Isotope Ratio Mass Spectrometry, or IRMS and Optical Spectrometry. In this case, we're specifically looking at TILDAS, which is Tunable Infrared Laser Direct Absorption Spectroscopy. So IRMS works by passing CO2 through a large magnet to separate 12C from 13C CO2 and then the d13C is calculated using those data. I'll talk more about how we use TILDAS to get d13C data in the next slide.
So both IRMS and TILDAS produce reliable high-resolution d13C data. They both operate in the gas phase. This means that a fuel sample would need to be combusted to CO2. This requires a combustion device. IRMS tends to be more expensive, and it can take up a larger space than TILDAS, it requires a lab environment with low vibrations and a controlled climate, and there needs to be space to facilitate heat distribution. While samples can take a short time to analyze using IRMS, samples from a refinery will need to be sent out to a lab. So you can’t necessarily be able to get near real-time information. In comparison. TILDAS is a smaller instrument. It's less expensive. This instrument is field deployable. I had mentioned how there are these mobile labs that bebop around and get data. The instruments has a rugged design, and this means, TILDAS has the potential to be applied on site at a refinery, and because a sample doesn't necessarily need to be sent out to a lab for analysis, it has a potential to produce near real-time measurements.
Okay, so how does TILDAS work? As in IRMS, when analyzing samples for d13C, samples are converted to CO2 via combustion. We worked with liquid samples in this in this project, but samples can also be solids or gases, though we'll still need to convert that to CO2. For TILDAS, CO2 is introduced into a chamber and a laser bounces back and forth and back and forth through this chamber. d13C is calculated from light transmission measurements at specific wavelength absorption bands. These correlate to our 13CO2 and our 12CO2. These molecules have absorption bands at other wavelengths as well. But these bands were specifically selected to maximize the instrument response, and to minimize interference with other molecules.
Due to the interest of using d13C as a fast and accurate alternative to counting radio carbon 14C for monitoring and optimizing the biocarbon during co-processing, we developed a method to prepare fuels and feedstocks and to assess TILDAS measurement performance. So a large amount of the method development has gone into preparing samples and tuning TILDAS for this application. Here's just a little overview of what we did. Samples were injected into a combustion oven and the CO2 was captured and purified. We took a sample and had two aliquots from each sample. These were them prepared and analyzed using IRMS and using TILDAS so that we could compare these methods.
Our goal here is to significantly reduce the hands on component of the current sample preparation and analysis, and then to analyze samples as a near continuous process, using only TILDAS. This whole procedure would be done onsite at a refinery and in a largely automated way. So let's take a look at our data. This figure shows data from IRMS vs TILDAS. The d13C of samples are plotted versus percent biocarbon. The data here are blended fossil fuels and biofuels. These blends have biocarbon between zero and 10%. There are two different sets of blended samples and these correspond to blendlines for renewable fuels made of C4 plants and C3 plant. The dashed lines here are our 95% confidence intervals for the combined IRMS and TILDAS data and the solid line is a model showing where we would expect the data to ideally plot. What we find is that there's a strong correlation between IRMS and TILDAS data and there's an r-squared close to one for both of these trend lines and a tight 95 percent confidence interval and for both of the tread lines. We think our results have improved over time as we've continued tuning our system, and we think that, moving forward, we will be able to get even better data by streamlining the process and going from a more manual process to an automated system. Something I want to point out here is that the d13C values for the C3 line, were slightly offset for TILDAS and IRMS measurements, but in both cases they have the same slope. We think this offset is related to sample processing. Because they have the same slope across the space, we have removed that offset for this plot, but we're trying to understand why there is that offset. However, it doesn't change the fact that our results suggest that TILDAS can be used for tracking biocarbon at low ratio blend ratios. IRMS provides precise and reproducible data that are comparable to IRMS.
Okay, so just taking a step back, we know that there is interest in a fast, inexpensive and automated analytical system that produces near real-time data, and that can be used onsite for monitoring and optimizing the incorporation of biocarbon in co-processed fuels. So this is a stable method, it has a potential to provide near real-time data on-site. However, this has yet to be tested onsite, and we would also need to do some fine tuning in the industrial setting. We have a method for preparing samples and analyzing d13C and all connected together for TILDAS. This method provides consistent, high precision measurements. So this is a less expensive and faster system for data collections than IRMS. There is more development that is needed to automate our system and combine the sample preparation with the TILDAS and that we need to do that before we can deploy our system. But we are currently integrating the sample preparation in TILDAS to simulate our near real-time and automated measurements. Moving forward we really want to partner for insights on how this method of biocarbon tracking can best be employed at refineries to incorporate more biocarbon in co-processed fuels. So this project was part of a three-lab collaboration, and we want to acknowledge everyone. In bolded text are the names of our team at the Pacific Northwest National Laboratory and our partners at Aerodyne Research. And with that, I'd like to thank you for your time.
Robert Natelson
Thank-you. We'll now open up for some Q&A, with their speakers, Sophie and Zhenghua, with the remaining time. Perhaps go to a a first question, perhaps for Zhenghua, and these questions came in throughout the talk, so some of them may have been answered or partially answered during the talk. Zhenghua, how much sample is needed for the CO2 conversion method and what is the measurement error?
Zhenghua Li
Yeah, this is a good question. You know a 100% saturation rate is 2.5 grams, CO2. So the optimal saturation rate is 80%, so if we can get there like 2.0 grams or around 2.2 grams, still 2 grams, that would be great that we enhance the signal to noise levels. This is all related to how much biogenic carbon is contained in a sample. If carbon content samples are higher, then we are using less amount of CO2, if this makes sense, you know, if your sample contains a very small amount of biogenic carbon, then we have to use a lot of CO2 like 2.5 or 2.2 grams CO2 will be good, we generate the useful signals. But sometimes, I don't know how much it like biogenic carbon, so this will we always maximize the CO2 content in in the absorbant, which is about 2.5 CO2 grams in the in a vial.
Robert Natelson
Next question is FTIR spectroscopy a viable optimal method in addition to Tilda?
Sophie Lehmann
I'm not totally sure that if FTIR looks at the isotopic ratios, that’s something to look at more closely.
Zhenghua Li
I use the FTIR but you know, it's not very quantitative. FTIR can be very sensitive, and also can be used for a carbon astro bioanalysis. But FTIR as a standard, we don't have a good standard to quantify, you know FTIR signals. Mostly, FTIR Okay, the relative intensity. How much is that? So that's why [AMS/IRMS] is gold standard. But in the future, you know, it could be a way to develop a method that if we can develop a standard using FTIR, then we can read an FTIR spectrum, and then we can get some, useful, meaningful ratio of carbon isotopes.
Sophie Lehmann
And if we label the feedstocks, then we can understand that label carbon is going to…
Robert Natelson
Next question, I see you will need d13Carbon for the fossil feed, for the vacuum gasoil, for example, how do we get that when refineries are in a dynamic not a process, these are the good questions. I'm not sure much we can discuss here but bring that up…
Zhenghua Li
Yeah, this is kind of a critical question, for the IRMS method, because IRMS method that is based on the feedstock composition. If the feedstock is not available, then there was no way we can compare the meaningful values, so that why we also need to know the feedstock like crude oil or bio-oil, the 13Carbon. We can analyze it easily, but that material should be available for 13Carbon analysis. You want to interpret the product that accompanies the team value.
Robert Natelson
Thanks. Another question, again, there is a lot of complex questions, proposing very scenarios. Can any of these methods measuring carbon be done during production, is there some type of sensor within the equipment?
Zhenghua Li
You know I think this would be like an online method… Sophie can talk a little bit more about that. What the purpose of the online method that we're trying to get, PNNL has in development?
Sophie Lehmann
Yeah. The idea, well one idea, is to hook it up to the processing line, and the sample, you know, to see how that d13Carbon is changing as you're actively changing coprocessing parameters. Do you see that d13C in near real-time moves closer to our bio-carbon d13C. Kind of like in that plot that Zhenghua showed towards the end, or some processing parameters produced a product with more biocarbon. So, but yes, we we would like to, you know, conduct it to the processing system. I want it to continue, not necessarily, but it would be, you know it wouldn't be real-time, but it would be pretty close to real-time.
Zhenghua Li
Yeah, I think if we can develop a sensor, that would be great, but sensors are very hard to detect an isotope. You know, here at Los Alamos, that we develop some sensor for some other isotopes, but not the light carbon, hydrogen oxygen, those are the other isotopes. But it would be great if that could be developed or the technology can be developed.
Robert Natelson
Next question. Again the [questions] came throughout the talk… When checking the biogenic carbon, did you adjust the calculations based on whether the original feedstock was C3 or C4 plants? How would this present challenges? You might not know the feedstocks, and of course there's a lot of really complex questions proposing various scenarios in that is outside the scope of this R&D, but I'll let you come to that.
Zhenghua Li
Yes, this is the critical issue for IRMS, for tracking biocarbon and the feedstock must be available. That's why we developed the second method as a 14Carbon. The 14Carbon method, independent of a feedstock, let's say we just go to a gas station, take a sample. Then we measure the biogenic carbon, we can do it. But there's no way we can use the IRMS to tell how much biogenic carbon is in a sample from a gas station. So because this is the limitation for using the IRMS method, you can use the 14Carbon method in a gas station. For example, using the 14Carbon method, you can get the biogenic carbon content. This is the advantage, if we use the 14Carbon, the AMS 14Carbon is the gold standard, and also the ASTM method, 6866, is the method that people use now.
Robert Natelson
Time for maybe one more question. How discrepant should fossil and biogenic feedstock be in d13Carbon signal to obtain the good results for lower biogenic carbon content.
Zhenghua Li
I'm not quite sure I get this question. Can you say it again?
Robert Natelson
How's discrepant should fossil and biogenic feedstock be in d13Carbon?
Zhenghua Li
OK, so yeah, if you look at the sensitivity background, you know this is a pretty, AMS is pretty sensitive, you know, based on the blending level put together. This is clearly described on the slide, also for publication, what's the meaningful difference of 13Carbon value between the feedstocks. But this all depends on the levels, blending levels, co-processing levels.
Robert Natelson
Thank you, Zhenghua and Sophie. We've gone over our scheduled time. So with that, we will thank our speakers again, and thank you all for joining and listening.