Hannah Echt

Project title: Characterization of Naturally-Occurring Alpha Diketone Emissions and Exposures at a Coffee Roasting Facility and Associated Retail Café

Degree: MS | Program: Occupational Hygiene (OH) | Project type: Thesis/Dissertation
Completed in: 2020 | Faculty advisor: Christopher D. Simpson


Background: Alpha diketones such as diacetyl and 2,3-pentanedione have been used as artificial flavorings in a variety of industries and are produced naturally when food products such as coffee beans are roasted. Exposure to these compounds has been associated with bronchiolitis obliterans, a rare and severe respiratory disease characterized by inflammation of the small airways of the lungs. There is evidence from previous research studies and National Institute for Occupational Safety and Health (NIOSH) Health Hazard Evaluations (HHEs) that coffee production workers have been exposed to alpha diketone emissions at levels above recommended health guidelines. However, there is a need for more research on how widespread this exposure risk is and what process elements are associated with the highest levels of exposure in the coffee industry. There are a limited number of studies characterizing occupational exposures to and emissions of alpha diketones at coffee roasting facilities. Even fewer studies have been published on occupational exposures to and emissions of alpha diketones in retail cafés. Only one other published study has quantified differences between light roasted and dark roasted coffee beans. No published studies exist that have evaluated the potential of total VOCs, CO, and CO2 to serve as lower-cost surrogate measurements for alpha-diketone concentrations. The purpose of this thesis was to fill these gaps in knowledge by determining which steps in the coffee production process are associated with the highest alpha-diketone emissions at a small craft coffee roaster and associated café, determining the extent to which direct reading measurements of CO, CO2, and total VOCs can serve as lower-cost surrogate indicators for diacetyl concentrations, and conduct a limited survey to quantify the effect that the process variable of roast type (i.e. light, medium, and dark roast) has on diacetyl emissions from grinding beans. Methods: Data were collected over 4 days of sampling in October 2019 at a single coffee roaster and associated café in Olympia, WA. Integrated personal and area air samples for diacetyl and 2,3-pentanedione were collected from 7:00 AM to 3:00 PM on each sampling day, corresponding to the full-shift roasting, grinding, and packaging operations at the facility’s roastery. Area samples were collected behind the shelves in the café, on the barista counter next to the espresso grinders, next to the roaster, next to the grinders in the roaster’s grinding station, and next to the hopper. Personal air samples for alpha diketones were collected on workers who operated in the coffee roasting, grinding, and packaging area adjacent to the café. A MinieRae 3000 PID (RAE Systems, Inc.) and a Photovac 2020ppb PRO Photoionization Monitor (Photovac, Inc.) were used to measure continuous total VOC concentrations at the roasting, grinding, and packaging stations. The TSI Q-Trak Indoor Air Monitor Model 7575 was used to measure continuous CO and CO2 concentrations at the roasting, grinding, and packaging stations. Probes of direct reading instruments and sorbent tubes were co-located at each sampling location, positioned to be facing towards the work activity being performed, and placed at approximate breathing zone height. For the emissions experiments, integrated area air samples for diacetyl were collected over 30-minute intervals for each roast type and the sorbent tubes were positioned next to the grinder at approximate breathing zone height. A MiniRAE 3000 PID (Rae Systems, Inc.) was used to measure continuous total VOC concentrations during each grinding experiment. Sorbent tube samples from the field and from the emissions experiments were analyzed for alpha diketone mass using GC/MS. 8-hour TWA diacetyl and 2,3-pentanedione personal exposures were calculated for each day of sampling. Medians and 5th and 95th percentiles of diacetyl, 2,3-pentanedione, CO, CO2, and total VOC emissions were determined from each sampling location. Correlations between diacetyl concentrations and each direct reading measurement were assessed. Finally, diacetyl mass emission rates in ng of diacetyl per gram of coffee ground were determined for each roast type. Results: Diacetyl concentrations were elevated in five of the seven personal samples obtained as compared to the NIOSH Recommended Exposure Limit (REL) for diacetyl – 5 ppb as an eight-hour time-weighted average (TWA) and one of seven personal samples exceeded the NIOSH REL for 2,3-pentanedione – 9.3 ppb as an eight-hour TWA. On days where workers were doing more grinding and packaging of ground beans, exposures to diacetyl and 2,3-pentanedione were higher compared to days where less grinding and packaging of ground beans was done. Median diacetyl and 2,3-pentanedione emissions were highest at the hopper (172 ppb and 110 ppb, respectively), followed by the grinder (57.3 ppb and 21.0 ppb, respectively), roaster (7.5 ppb and 3.5 ppb, respectively), barista (3.1 ppb and 1.5 ppb, respectively), and background areas (2.0 ppb and 0.89 ppb, respectively). Correlations between diacetyl and total VOCs, CO, and CO2 showed that the model using air concentration of total VOCs as a predictor for air concentrations of diacetyl had the highest R2-value (R2 = 0.95, p-value = 4.8 * 10-15), followed by the model using CO2 (R2 = 0.58, p-value = 0.001), and the model using CO as an indicator for diacetyl air concentrations (R2 = 0.09, p-value = 0.34). Based on our limited survey, French roast was associated with the highest mass emission rate of diacetyl, followed by medium espresso, espresso, and white coffee. Conclusions: Results from the observational study indicated that coffee production workers at this facility had elevated exposures to diacetyl and 2,3-pentanedione compared to NIOSH recommended guidelines. Area sampling showed that the areas with the highest alpha diketone emissions were the grinder and the hopper, which are both areas associated with tasks involving ground roasted coffee (i.e. grinding and packaging ground beans). Using the model that predicts air concentrations of diacetyl from the air concentration of total VOCs, the average diacetyl concentration when the total VOC concentration is 400 ppb is 7.0 ppb (95% C.I.: -2.1 – 16.2 ppb). Because the 95% confidence interval includes values both above and below the NIOSH REL, this model cannot be used to reliably predict whether an exposure is elevated compared to NIOSH health guidelines at this total VOC concentration. Constraining this model to be based on observed diacetyl concentrations that were less than 100 ppb improved both the accuracy and precision of the original model’s estimates. Using the constrained model, the model predicts that an exposure is elevated compared to the NIOSH REL for diacetyl at total VOC concentrations in excess of 400 ppb. Results from the emissions tests showed that diacetyl mass emissions generated from grinding French roast were significantly higher than those of espresso, medium espresso, and white coffee. White coffee had a very low diacetyl mass emission rate compared to the other roasts, and its rate was significantly lower than that of espresso and medium espresso. While a one-way ANOVA test and post hoc pairwise comparisons showed significant differences in diacetyl mass emission rates between the different roasts, the mean diacetyl emissions from the French roast is one-third higher than that of the medium espresso and espresso roasts. For smaller operations, distributing grinding and bagging tasks more evenly throughout the work week, especially for darker roasts, could reduce worker’s daily TWA exposures to alpha-diketones. While this could improve compliance with health guidelines, the benefit to worker health is dependent on whether or not cumulative exposure to alpha diketones increases risk of respiratory impairment more than peak exposure. Peak exposure would be reduced using this administrative control, but cumulative exposure would remain the same. Future research could focus on designing effective engineering controls, in the form of local exhaust ventilation, with the goal of reducing alpha diketone emissions and exposures, as well as conducting similar studies to this thesis at other small-scale craft coffee roasters and cafés in order to better understand the variability in these emissions and exposures within these types of facilities. URI