Editor’s note: This article was first published by Roast Magazine‘s January/February 2024 issue, and is reprinted here with permission.
Freezing Green Coffee
An Evaluation of the Impact on Sensory Quality Over Time
By Chris Kornman and Isabella Vitaliano
Between the months of December 2022 and June 2023, The Crown: Royal Coffee Lab & Tasting Room designed and carried out a research project to investigate what sensory changes might be observed in green coffees that have been frozen in commonly available packaging types: Ziploc and vacuum sealed. We stored, roasted and cupped three coffees over six months and compared them with a control sample stored in GrainPro packaging in a climate-controlled room set to between 20 and 22.2 degrees C (68 and 72 degrees F).
By the end of the project, our cuppers generally scored frozen coffees in either of the frozen packages higher than the control samples—particularly coffees with higher initial quality ratings—but failed to reliably differentiate control versus test samples in blind triangulation sets.
We believe that, even in home or small-scale commercial environments with limited packaging options, there may be some benefits to freezing the highest-quality green coffee for long-term storage. We believe that this practice may be less beneficial for moderate- to low-scoring specialty arabica coffees. We also note that freezing versus conventional storage impacted the coffees’ behavior in the roaster. We acknowledge that professional tasters were overall inconsistent in their ability to identify differences across the entire project.
Literature Review and Context
Data-backed claims on the impact of storing green coffee in freezing temperatures are few and far between, and peer-reviewed articles are even scarcer. Some leaders in the industry believe that freezing green coffee allows them to halt the aging process of coffee to maintain its flavor and integrity. Conversely, unsupported information exists claiming the moisture loss from freezing brings out flavors similar to past-crop coffee. These sources insist that freezing green coffee during any period of transit or warehouse storage would diminish the quality.
Tests run by our team at The Crown resulted in some insights into how a brief freeze impacts long-term moisture in green coffee (first published by Daily Coffee News in March 2018). Over the course of five months, control samples in this research had a substantial change in moisture reading, while the frozen samples were more stable.
One peer-reviewed article on freezing green coffee—titled “Comparison of chemical compounds and their influence on the taste of coffee depending on green beans storage conditions,” published in Scientific Reports in February 2022—covered different temperatures over a 12-month period. It compared washed- versus dried-processed coffees and found that the sensory quality of washed coffees diminished in the higher temperature storage range. The researchers compared jute to GrainPro and found that of the five temperatures (-10, 5, 10, 18 and 20 degrees C [14, 41, 50, 64.4 and 68 degrees F]) the 18- and 20-degree C conditions yielded lower cup scores. Interestingly, storage bag type did not impact quality loss in a significant way in this study.
Research Project Methodology
The aim of this evaluation was to determine the impacts of freezing the green coffee on the resulting cup quality during a six-month time frame. By assembling a sensory panel of three judges with more than 40 years of accumulated experience, including one Q Grader, we felt our small group was well-qualified to analyze these coffees.
The project design focused primarily on sensory analysis and Specialty Coffee Association (SCA)/Coffee Quality Institute (CQI) scoring over this time frame. Three washed coffees of different origins—a Guatemalan, an Ethiopian and a Kenyan—were selected and split into sample sets: control, Ziploc, and vacuum sealed. The control was stored in a GrainPro bag inside a lidded plastic Cambro container at room temperature; the Ziploc bag was sealed by hand; and the vacuum-packed sample was sealed shut using a commonly available FoodSaver brand system. Both the frozen samples were held in a commercial kitchen freezer in a 0 to -4 degree C (32 to 24.8 degree F) temperature range throughout the duration of the project.
Green coffees were allowed to defrost at room temperature for one day before roasting. Each origin was assigned an individual roast recipe (designed to replicate a production-style environment) on an Aillio Bullet using a “burner profile” style automation—where the digitally controlled roaster adjusts the heat setting at pre-programmed time intervals.
Figure 1 – Guatemala Roast Profile
Figure 2 – Kenya Roast Profile
Figure 3 – Ethiopia Roast Profile
On the day following roasting, coffees were cupped and triangulated by our sensory panel. Roasted samples were cupped and scored (traditional SCA/CQI style) in an initial sensory analysis in December of 2022, and thereafter at three regular intervals in January, March, and June of 2023.
Sensory analysis included a set of six sample groups to triangulate per coffee type, along with a supplemental cupping set to obtain scores and descriptors. Every storage sample was compared against each other set twice in each triangulation round. The cuppings were staggered at one-week intervals to avoid palate fatigue.
Limitations
Our study acknowledges several limitations. Our moisture and water activity data were gathered at irregular intervals and relied heavily on previously studied performance of different coffees. The coffee sample set is small and not representative of many regions and processing methods. We made a small adjustment to our roasting protocols in the final round of testing. Our triangulations did not conform to international standard methodology for triangle testing. Some cuppers were absent from certain sessions, which may have skewed data on certain triangulation and cup score performance (though we believe that overall trends remain relatively accurate and repeatable). It is probable that results could be altered or improved by initiating the study closer to the time of the coffee’s arrival, and continuing it for a longer duration. Additionally, we noted that some roaster controls were not optimal for consistency and repeatability, and that as coffees changed during storage so too did their behavior during roasting, which likely had some impact on both cup quality evaluations and triangulations. As with most studies of this type, more work must be done to verify our conclusions.
Discussion and Analysis
Changes and Observations in Physical Quality
Freezing coffee resulted in a few noticeable differences in the physical qualities of the coffee when compared to the control coffee stored at room temperature. In our testing, both the Ethiopian and the Kenyan coffees remained insubstantially altered during the six-month investigation, either at room temperature or in frozen conditions. We surmise this was likely due to their overall lower initial moisture content (9.5 to 10 percent). The Guatemalan coffee, however—which entered the study at 11.8 percent moisture and a relatively high 0.62 water activity level—was affected by room temperature storage, and it lost 1 percent moisture and 0.07 water activity over the course of six months. However, both the Ziploc and vacuum-sealed frozen Guatemalan coffees retained their original readings within a very narrow margin (+/- 0.1 percent moisture and 0.01 water activity).
During roasting, we observed changes in the coffees’ behavior under established recipes. Frozen (and subsequently thawed) coffees from all origins began to take longer to reach first crack (+30 seconds by the end of one month in storage). Additionally, likely because of this change in heat absorption (and the adherence to our original and now insufficiently powerful burner profile settings), cuppers began to note more roast-related flavors in the control samples across all three origins during testing after one month’s duration, and especially after three months of storage. This observation was reinforced by higher cupper accuracy in month three across all coffees, when cuppers correctly identified the control roast versus either of the frozen samples with an impressive 75 percent accuracy.
This created an unexpected problem in our testing. Our initial goal was to try and taste flavors in the frozen coffees related to storage conditions, not roasting conditions. We felt that the experiment’s original design had failed in this purpose.
With strong confirmation that the storage environment was affecting the way the coffee roasted, we chose to alter our roasting protocols for the final round of sampling and homogenize final roast color (rather than following a strict temperature and time profile) to best test the sensory impact of freezing. The result was that cupper accuracy returned to lower levels, reflecting the groups’ overall average just under 50 percent.
Observations on Affective Cup Quality Evaluations, Total Score and Attribute Ratings
Cuppers evaluated each coffee using the SCA/CQI 10-attribute scoring sheet at five intervals: (1) initial evaluation sample roast (on a Proaster 2-barrel machine, between August and September 2022 for all three coffees); (2) Dec. 9, 2022, Bullet roast to establish roast profiles and evaluate final control quality prior to freezing; and during the months of (3) January 2023, (4) March 2023, and (5) June 2023, representing one, three, and six months of storage, respectively.
All coffees in all storage packaging declined in overall cup score between December 2022 and June 2023, with the control samples showing the greatest declines. The Ethiopia lost the most total points from 86.8 in December to a low of 83.5 (control) in June (a decline of 3.3 points). The Kenya lost 2.3 points from 87.2 to 84.9 (control), while the Guatemala fell 1.3 points from 83.9 to 82.6 (control). This trend mostly holds when comparing the sample roast scores taken from the coffee’s U.S. arrival date. It could be construed that coffees with higher total scores and better qualities are those with the most to lose.
Overall, cuppers generally agreed in ranking, total scores and individual attributes when comparing the same coffee under different storage conditions. However, there was little consistency in descriptive terminology, and cuppers generally did not align well with each other on the numerical value assigned to attributes and overall cup score. Cupper disagreement is represented by the shaded areas on each coffee’s cup score evaluation infographics.
Figures 4 5 6 – Total Cup Quality Ratings Over Time
For all three coffee origins, we observed higher overall scores on the frozen coffee samples than the control storage coffee, with few outliers. In both the Kenyan and the Guatemalan samples, the vacuum-sealed coffee consistently outperformed the Ziploc-stored coffee. Curiously, the Ethiopian coffee had an inverse performance, where the Ziploc-stored coffee quality consistently rated higher across all three cuppings compared to the vacuum-sealed beans. Comparing December 2022 to June 2023, in the best storage condition for each coffee, the frozen coffee cupped within -1.1 to -0.4 of the original score, a marked improvement over the performance of the control sample (-3.3 to -1.3 points).
Figure 7 – Difference in Cup Score by Packaging Type—December 2022 vs. June 2023
More granularly, between December 2022 and June 2023, across the seven scaled attributes on the SCA/CQI form (aroma, flavor, aftertaste, acidity, body, balance, and overall), cuppers noted near-universal declines in aroma, flavor, acidity, and aftertaste in all coffees, with one or two exceptions based on storage types.
Figure 8 – Kenya : Difference in SCA/CQI Attribute Score By Packaging Type December 2022 vs. June 2023
Figure 9 – Ethiopia : Difference in SCA/CQI Attribute Score By Packaging Type December 2022 vs. June 2023
Figure 10 – Guatemala : Difference in SCA/CQI Attribute Score By Packaging Type December 2022 vs. June 2023
When describing the coffees, some cuppers mentioned flavors associated with green coffee staling in certain samples.(Cuppers may experience this as a muting of the coffee’s intensity, or as drying, woody, or paper-like aftertaste. Sometimes simple terms like “fade” or “aged” are used in the industry to summarize coffee tastes associated with staling green coffee.)
One cupper, tasting blindly, indicated the presence of “age” flavor for all three control samples in June 2023, though that cupper was unable to accurately identify the off-cup during triangulation at a higher rate than other cuppers.
Observations on Discriminative Analysis
An important component of our sensory testing involved three professional coffee tasters, including one licensed Q Grader. Cuppers were tasked with triangulating each coffee origin by storage type (i.e., choosing one “off cup” position randomly in line with two “like cups”). Single-origin tables were set with replicate pairings, so each cupper tasted each coffee’s storage type compared with each other storage type twice. If a cupper could correctly identify the odd cup at a rate of 33 percent, they would equal the expected rate of random chance.
Overall, as a group the tasters correctly identified 68 off-cups out of 138 total sets, for a combined accuracy of 49 percent. This could be characterized as better than chance, but still unreliable.
Cuppers were better at identifying the control sample when paired with either frozen sample than they were at comparing Ziploc versus vacuum-sealed. This trend holds true across all three cuppers, all three coffees, and all three evaluations.
Figure 11 – Cupper Identification Accuracy Percentage (Organized by Coffee Origin and Packaging Type)
Cupper accuracy was highest in March 2023, after three months in the freezer—the mid-point of the six-month study—at an overall rate of 64 percent accuracy across all cuppers, coffees and sessions that month. This was likely due to differences in roasting color/duration, as noted in a previous section related to observations of physical quality. The final cupping in June 2023 returned to the overall average
for cuppers of just under 50 percent, after adjusting the roast levels to improve homogeneity in color and development.
Cuppers were overall less accurate with the Guatemalan sample than they were with the two other origins (38 percent for Guatemala compared to 50 percent for Kenya and 58 percent for Ethiopia). Cuppers were especially accurate with Ethiopian control samples when compared to frozen storage types, posting 69 percent accuracy in this category averaged across all evaluations.
The Q Grader logged the best individual cupper performance, and correctly identified the Kenyan coffee’s control versus vacuum-sealed samples in all six sets across the three triangulations. This boosted their overall average on the Kenyan coffee evaluations to 67 percent, and their total accuracy rating across all triangulations was the highest of the group at 56 percent (compared to 50 percent and 44 percent for the other two cuppers). As with the other two cuppers, high accuracy in the Kenyan and Ethiopian coffee identification was buffered by poor performance across the three Guatemalan triangulation sessions.
At large, the cuppers’ relatively unreliable accuracy at blind identification of differently stored coffees does not support the practice of freezing green. However, when comparing the highest-scoring coffees, cupper accuracy approaches 66 percent, a relatively good indication of freezer performance versus unprotected green coffee storage for the best-quality coffees. Lastly, under our conditions (six months in storage, freezer temperatures stable between 0 to -4 degrees C/32 to 24.8 degrees F), the cuppers inability to distinguish between vacuum-sealed and Ziploc-stored packaging types does not indicate substantial differences in storage media.
To Freeze or Not to Freeze
During this study, we observed that coffees stored over time at room temperature, even with protection such as GrainPro liners, tend to lose cup quality in both total cup score values and individual attribute ratings. From our evaluation, even given the small sample sizes and limitations related to methodology, we believe this decline in quality may be slowed—but not completely halted—by freezing the green coffee.
Much discussion has been made about “deep freezing” and/or vacuum sealing green coffee to preserve quality during freezer storage, but our study suggests that even a standard commercial freezer operating between 0 to -4 degrees C (32 to 24.8 degrees F) and simple storage packaging like a commonly available Ziploc bag are better than GrainPro packaging at room temperature for preserving quality, when rated by total cup score and individual attributes on the SCA/CQI scale. This was especially true for higher-quality coffees observed in our study. Coffees of lower starting quality may benefit less from frozen storage.
