Caffeine Project

Project Overview

For this project I worked with my advisor, Brad, and a fellow student, Will Fox, to determine the caffeine levels in different samples of coffee. Our original goal was to determine whether a lighter or darker roast of coffee would have a higher concentration of caffeine. We then branched out to a few other questions involving the caffeine levels in the green coffee beans, the caffeine levels in decaf coffee, and how much caffeine the coffee maker we have extracted through each run of the coffee grounds. This was accomplished by using the HPLC to collect data and using Igor and Excel to analyze the data. This page includes procedures for making the standards and samples, as well as procedure for running the HPLC. This project allowed us to gain a more comprehensive understanding of how the HPLC works and how to effectively analyze and portray data we gained from the HPLC.

Roasts

The difference between L0 and L5 roasts is the highest temperature the beans will reach in the roaster. The “L” stands for the level of roast. As the levels go up, the beans are roasted at higher peak temperatures. So the L0 roast is considered to be a "light roast" and the L5 roast is considered to be a "dark roast."

Figure 1. Chemical structure of caffeine

Making the Samples

Caffeine Standards

For this project we had to make caffeine standards to compare the samples to.

The following link describes how we made these standards:

Caffeine Standards

L0 and L5 Samples - Trial 1

To make the coffee for the samples, 2.6 oz of beans were weighed out. These beans were then grinded, and 38 g of the beans were placed into a pourover filter. Then, 640 g of boiling hot water were poured over the grounds, using the chemex, allowing the grounds to bloom as needed. This process was done with both L0 and L5 beans.

L0 and L5 Samples (and Green Beans) - Trial 2

After this first round of coffee, we discovered that the coffee samples were too concentrated for our standards, so we cut the amount of beans to 30 g of L0 and L5. This round we also made a sample of 15 g of the green coffee beans with 320 g of boiling water. We used a 0.45 micron filter and syringe to put this solution into the HPLC vials to ensure there was no particulate matter present.

Final Results for Preparation of L0 and L5 Samples (and Green Beans)

After looking at the data from the second round of coffee, we discovered that the concentrations of caffeine in our samples were still too high for our standards. We tested L0 and L5 again, diluting 50% (400 microliters of the sample and 400 microliters of RO water). We also tested L0 samples through the traditional coffee maker, running the grounds through three separate times to analyze how the concentrations of caffeine were affected by extraction. This test was also done with green coffee beans as well.

Trigonelline and Nicotinic Acid

Once the data for this trial was analyzed, we decided to run L0 samples three times through the coffee maker again to validate our previous results. In this trial, we also tested standards for 300 ppm trigonelline and 10 ppm nicotinic acid, which are both also found in coffee. The trigonelline standard was made by mixing 0.03 g of trigonelline and 100 mL of water. The nicotinic acid sample was made by mixing 0.001 g of nicotinic acid and 100 mL of water.

Decaf Samples

After all of these trials, we decided to test the amount of caffeine in decaf coffee. 2.6 oz of decaf coffee beans were weighed out and grinded. We then ran these grounds through the coffee maker three times. These samples were not diluted because we already expected the concentrations of caffeine to be fairly low.

Instrumentation

After making the standards, we tested the standards using the HPLC to see where the peaks were for the different concentrations of caffeine.

The following link goes into more depth about working with the HPLC:

HPLC

Once the standards were tested, another HPLC trial was done including the L0 and L5 coffee samples.

An additional HPLC trial was done with coffee that had been diluted. This set included L0, L5, and the green coffee bean samples.

The HPLC trials with decaf coffee were not diluted due to the small amounts of caffeine in the samples.

What the HPLC Tells Us

We ran these samples through the HPLC at the wavelength of 273 nm, which is the lambda max for caffeine. Running the samples at this wavelength tells us how much the samples absorb at this wavelength.

Data Analysis

To analyze the HPLC data, the data was exported as an arw file. This file was then loaded into Igor, which we used to make chromatograms. Chromatograms allowed us to gain a better understanding of the how each concentration of caffeine absorbs at 273 nm for our standards, and how each sample absorbs at 273 nm. We used the information from the chromatograms to make standard curves in excel. The slopes of the standard curves gave us an equation we could use to find the concentration of a sample using the absorbance.

Example standard curve data can be found here.

Equation

The following equation was used to find the concentration of caffeine in our samples. The equation is solved for x, which is concentration, then the absorbance is plugged in for y and solved.

y = 0.0041x + 0.0071

R² = 1

Data Summary

From the data we collected, we found that the green beans had more caffeine than the L0. We also found that if L0 is run through our coffee maker three times, the second run through will have the most caffeine. However, this was not true for the decaf and green bean runs through the coffee maker. These followed the more expected result of the first run having the most caffeine, followed by the second run and the third run having the least amount of caffeine. That being said, small amounts of caffeine were found in the decaf coffee.

As for results of L0 and L5, we need to run further tests to get a more reliable answer as to which roast has more caffeine.

Caffeine in a Standard Cup of Coffee

To see how our results compare to large companies' caffeine levels per cup of coffee, we decided to configure the concentrations of caffeine in the Short, Tall, Grande and Venti sizes of the dark and medium roast coffees from Starbucks. The data sheet and our calculations showed that the dark roasted coffee ranged from 545-575ppm in caffeine levels. Their medium roast coffee ranged from 655-690ppm caffeine. These ranges show that their medium roast coffee has more caffeine than the darker roast coffee. The concentration should be the same for each of the sizes of coffee, which is why the nutritional sheet for Starbucks is interesting. The nutritional sheet had different sizes of the same roast listed at different concentrations of caffeine, which should not be the case. We found that with the same grounds, the caffeine concentration levels are different, which could be why the concentration levels for their coffee are skewed.

The USDA states that the average cup of coffee has a caffeine concentration of about 400 ppm. If you add the caffeine concentrations of our trials of the Pot 1-Pot 3 samples, it adds up to about 400 ppm. This shows that our coffee maker is not very efficient in extracting all of the caffeine from the coffee grounds.

Starbucks Nutritional Value PDF

USDA Cup of Coffee Caffeine Levels

References

Monmouth Coffee Project

Journal of Chemical Education Article about the Chemistry of Coffee

Starbucks Nutritional Value PDF

USDA cup of coffee caffeine levels

HPLC for Beginnners Video