Innovation in the Lab

FROM THE OCTOBER ISSUE: Ethanol laboratories use HPLCs to find issues early in fermentation, but increasingly have been using the data for process optimization.
By Susanne Retka Schill | September 19, 2017

High-performance liquid chromatography (HPLC) is the workhorse of the lab, each day analyzing multiple samples used to monitor fermentation and identify key compounds that could indicate contamination. In the early years, the analyses were primarily used to spot trouble early on, when corrective steps could be taken easily.

Today, operators also are using the data generated to fine-tune the process. Plus, as the ethanol industry has matured, lab managers are being asked to do more; the new yeasts and enzymes being introduced require accurate baseline data for comparisons during product trials; different compounds are being tracked; and interest in monitoring fusels in fermentation is heightened. Thus, consistency and accuracy is front and center in the life of the ethanol lab. That can be a challenge in an industry where many lab workers have learned their skills on the job.

Kristi Plack, chief operating officer at Bion Analytical, works with lab managers to calibrate equipment, keeping it precise and optimized. Practices vary greatly, she says, from calibrating daily to only calibrating when a column or a mobile phase is changed. She admits that her lab’s practice of calibrating for every run isn’t a feasible approach for an ethanol plant, but she recommends at least a couple times a week. “The big thing is, you want to be consistent.”

Bion Analytical, based in Sioux Falls, South Dakota, makes standards — known solutions that are used for calibration and validation. Calibration is done with one standard and the validation should be done with another with a different lot number, she says. “You know the known value of your validation standard; you should see the validation standard randomly above or below the known value.” If not, it’s time to begin checking into possible biases. Having a notebook for each HPLC helps with troubleshooting. Was a mobile phase changed? Was the column cleaned on schedule? “You think you’re going to remember things like that, but you don’t.”

An Apt Analogy
High-performance liquid chromatography has a language of its own. Jim Mott, field technical support supervisor for HPLC maker Shimadzu, has an apt description for how HPLCs work. “It’s so silly it actually works,” he says with a chuckle. “There’s a fire at a very small zoo that only has elephants, deer and rabbits. To protect the animals, the zookeepers open the gates and let them run through the forest. The forest has a fence so the animals must pass through the forest. A zookeeper is stationed on the other side to count the animals as they exit the forest.” The trees represent the stationary phase —  the column. The sample is the animals and the fire is the mobile phase forcing the animals through the forest. “The rabbits are going to be able to squeeze through small spaces, the deer need a bit larger space and the elephants need the biggest space. The result is that the rabbits come out first, followed by the deer, then the elephants. The interaction between the animals (sample) and the trees (stationary phase), as they are being forced through the forest, causes the different species to be separated. The zookeeper (detector) quantifies the results of the separation.”

For an ethanol plant, an HPLC is a relatively cost-effective and quick way to monitor the primary molecules of interest — the starches being converted to sugars, the sugars being converted to ethanol and other compounds that indicate how well things are proceeding.

There may be instruments that are more precise and accurate, Mott says. “There are better ways to separate and measure carbohydrates, but then you might not see the ethanol. Other technologies may be more expensive or more well-suited for a research lab.” The HPLC, as commonly configured for the bioethanol production lab, is a good compromise for a lab that must process multiple samples a day. “The analysis time is typically 25 to 30 minutes per sample,” he says. “And they are usually pulling between four to 10 samples per fermentation, and many plants have four to eight fermenters in process. There are samples coming in all the time.”

New Applications
“Most people use HPLC to monitor fermentation,” says Mike Smith, group leader in biofuel technical service at Novozymes. “They use it to see if someone missed an addition, for example. Say an operator forgot to add urea or GA (glucoamylase); you pretty much have 24 hours to take corrective action.”  Forgetting to add yeast or stalled fermentations are increasingly rare, he says.

Now, plants are taking it to the next level. “Innovators are using HPLC data to optimize their performance. For example, they might start measuring five hours before drop to see if they are maximizing their fermentation time.” Another would be looking for peak glucose at around 10 hours and then bringing that peak down. “You can time your GA,” he explains. Some producers are looking at fusels — compounds produced by stressed yeast that become inhibiting if levels get too high — with their HPLC, he adds, but they have found the gas chromatograph to be better for that job.

The next step for those wanting to get more out of their HPLC, Smith says, is to learn more about the science behind the data. “I would like people to look at actual chromatography. What those peaks are and what they aren’t. I think there’s a lot of misunderstanding on what the DP4 peak is, for example.”

Mott explains DP stands for degree of polymerization, but it can be thought of as dextrose polymers. “If you think of glucose being the primary fermentable sugar, maltose is DP2 — two glucoses linked together. DP4 is four.” That peak on the chromatogram is referred to as DP4+, since it actually can be a mix of DP4 to DP40, an aggregate of many carbohydrates. As the starches get broken down to smaller pieces, they become fermentable sugars and the starch peak gets smaller. What becomes confusing, however, is that the DP4+ peak doesn’t disappear at the end of fermentation. Some think it means lost yield, but it could be other interfering molecules.

Testing Parameters
Mott has been experimenting with different parameters to see if making tweaks to HPLC methods might help give better results. He presented his work at the International Fuel Ethanol Workshop in June in Minneapolis. He has run experiments comparing the typical operating conditions recommended by manufacturers for column temperature, length of the column, different acidic levels in the mobile phase and the flow rate.

“When looking at fermentation samples where other materials are interfering, changing temperatures can give you a cleaner view of what you’re interested in,” he says. When using standard solutions for the experiments, it wasn’t clear which temperature value was best. “It looked like carbohydrates early in the chromatogram preferred lower temperatures and the organic acids and alcohols liked higher temperatures, but you have to find the best single temperature for the analysis. With ferm samples, there was a definite improvement around maltose as the column temperature increased. Looking at 70 Celsius column temperature, you seem to get a cleaner separation of the carbohydrates from some of the interfering secondary metabolites.” Temperature adjustments might also be helpful in analyzing other compounds in the carbohydrate area, such as fructose or trehalose.

With greater interest in tracking fusels, Mott also experimented with a shorter column. Fusels can be detected with the same HPLC operating conditions used for fermentation monitoring, but the analysis time can be long, up to three hours. “Since fusels elute after ethanol, a long column may not be needed for adequate analysis,” he says. “Using the same hardware with a 150-millimeter column (where a 300-mm column is mostly used for fermentation monitoring) will cut the analysis time in half. Increasing the flow rate by up to 50 percent will further shorten the analysis time without detriment.” He’s been able to get all the detectable fusels within 50 minutes, he reports, a big improvement over the potential three hours with standard parameters.

Fusel analysis is a more frequent question these days, Mott says, along with concern over residual carbohydrates at the end of fermentation. “I’ve seen instances where the carbohydrates are pretty well consumed, and [the plant labs] may identify the wrong peak glucose or maltose.” As long as the HPLC is in good condition, the retention times for the various peaks don’t change.

Education has been a big part of Mott’s mission since Shimadzu began supplying HPLCs to the ethanol industry back in the early 2000s. “We used to say they build ethanol plants where the corn grows, and not where there is an immediate supply of trained chromatographers.” Bringing in people with university-level training in fermentation isn’t always economically feasible, he admits, but he’s seen plants take people with minimal science background and turn them into effective lab technicians and managers.

“It’s been interesting to see how people progress,” he says. “I’ve been around some of these plants since the day they were first started up and have seen the same lab manager be on site ever since. They go from being very green — saying ‘I’m not quite sure why I’m standing here and doing this’ — to asking very intelligent questions, based off 10 years of experience.” They can look at a chromatogram and decide whether it’s quality and reproducible data. If it isn’t, they are asking how to test and find the problem. “I’m proud of them. It’s taken a great deal of effort on everyone’s part. Now that they’re asking good questions, that’s cool.”

Author: Susanne Retka Schill
Freelance Journalist