Fast and Slow Roasting of Alcohol-and Honey-infused Coffee Blends: A Comparative Study of PAH and N-PAH Emissions

Polycyclic aromatic hydrocarbon (PAH) and nitrated PAH (N-PAH) emissions during coffee roasting were investigated. The fast roasting of green coffee resulted in mass and toxic concentrations approximately 1.6 and 3.2 times higher than in the slow roasting method (5,069 ng Nm –3 and 14.1 ng BaP-TEQ Nm –3 , respectively). However, when considering all the coffee formulations, only the mass concentrations between the two roasting methods were statistically different ( p = 0.05). The initial concentrations of total mass PAHs and PAH4 (BaA, Chr, BbF, and BaP) in green coffee beans (A) were 57.8 ng g –1 and 0.591 ng g –1 , respectively. Following roasting, the PAH concentrations decreased to 34.8 ng g –1 and 40.5 ng g –1 during fast and slow roasting, respectively, with PAH4 concentrations below the detection limit (LOQ = 0.2 ng g –1 ). N-PAHs in the flue gas, green coffee, and roasted coffee beans were also below detection limits, except for the case of fast roasting of B-3 (coffee infused with bourbon whiskey and granulated sugar) where concentrations were 0.503 ng g –1 . The flue gas profile was predominantly composed of naphthalene. On the other hand, the coffee bean profile had significant fractions of three-ringed compounds. Generally, the PAH congener profiles for both the fast and slow roasting methods were similar. For N-PAHs, only 2-nitropyrene was detected in the fast roasting of B-3. The PAH diagnostic ratios indicated that the source of PAHs in green coffee was pyrogenic in nature and more specifically, petroleum combustion. Finally, the emission factors for the fast and slow roasting of green coffee were 1.36 mg kg –1 of coffee (7.49 µ g BaP-TEQ kg –1 ) and 0.483 mg kg –1 (1.35 µ g BaP-TEQ kg –1 ), respectively. These results narrow the research gap on toxic emissions during coffee roasting and would be of particular interest in independent coffee shops where roasting is done in-house.


INTRODUCTION
Coffee roasting transforms the physicochemical properties of green coffee, giving it the characteristic aroma and flavor that has made it one of the most consumed beverages in the world, with a global consumption of over 171 million 60 kg bags in 2023, according to the International Coffee Organization (ICO, 2023).However, harmful compounds are also generated during the process, including acrylamide and polycyclic aromatic hydrocarbons (PAHs) (Duedahl-Olesen et al., 2015;Binello et al., 2021;Huang et al., 2023), which are detrimental to human health (Li et al., 2023;Room et al., 2023).For example, Houessou et al. (2007) reported that phenanthrene (Phe), benzo[a]anthracene (BaA), and anthracene (Ant) form near 220°C, and that higher roasting temperature (250-260°C) generated pyrene (Pyr) and chrysene (Chr).Ciecierska et al. (2019) roasted coffee using an electric heating system for 25-26 min at temperatures ranging from 125 to 135°C and found that the contamination levels of 19 PAHs in the coffee beans were mainly attributed to the low-molecular-weight PAHs, while the more toxic high-molecular-weight PAHs, known for their mutagenic and carcinogenic, contributed only 2.0%-24%.Additionally, the study found that both green and roasted coffees had similar profiles.
Nitrated PAHs (N-PAHs), which are more toxic than their parent compound, have also been detected in roasted coffee (Ko et al., 2018).Ko et al. (2018) detected N-PAHs in coffees sourced from Colombia, Brazil, and Vietnam roasted at various combinations of temperature and time: 150, 180, 210, 230, and 250°C for 5, 10, and 20 min.The results showed that N-PAH concentrations gradually increased with rising temperatures and roasting time, with concentrations reaching up to 5.74 ± 0.81 µg kg -1 .
Nevertheless, it's important to note that all previous studies in this field have exclusively concentrated on PAH and N-PAH concentrations within the coffee bean themselves, without investigating the emissions from the flue gas of coffee roasting machines.This is particularly of concern in independent coffee shops that roast their coffee beans in-house.Furthermore, because coffee roasting machines are operated indoors and have low chimneys, exposure to the emissions could be significant to staff, customers, and people nearby.Lin et al. (2022) found that roasting coffee beans, tobacco smoking, and occupant density were the main activities that affected the indoor air quality in independent coffee shops, contributing to increased particulate PAH concentrations.Additionally, new coffee formulations with alcohol, honey, or maple syrup, which may influence the formation of PAHs and N-PAHs, are available in the market.For example, honey contains several phenolic compounds which could be precursors to PAH formation (Manyi-Loh et al., 2011).
Therefore, this study investigated PAH and N-PAH concentrations in the flue gas during coffee roasting and the roasted coffee bean.The study investigated emissions of these compounds from roasting green arabica coffee and coffee formulations, sold in Taiwan, that consist of green coffee, alcohol (whiskey or wine), granulated sugar, and longan honey.The results of this study will be useful in understanding the contribution of coffee roasting to PAH and N-PAH emissions to the environment and the potential exposure of these compounds through consuming coffee.

Preparation of Coffee Formulations
The green arabica coffee beans were purchased from Golden Malabar Indonesia, while the granulated sugar (Taiwan Sugar Corporation's refined golden sugar), honey (100% Taiwan Longan honey), wine (Louis Jadot Beaujolais Villages red wine, Brown Brothers Cienna red wine, and De Bortoli Deen Vat 5 Botrytis Semillon botrytized white wine), and bourbon whiskey (Basil Haden's Bourbon and Jim Beam Kentucky Straight Bourbon) were purchased locally in Taiwan.The green coffee beans were dried to reduce the moisture content to ~12% by drying at 135°C (Table S1).The dried and cooled green beans were placed in a glass container, and the other ingredients (alcohol, sugar, and/or honey) were added according to the formulas shown in Table 1.These coffee formulas are based on commercial products sold in Taiwan.The glass container was then sealed and placed horizontally on an automated rotating mixer that rotated the container at a rate of 140 rpm for 36 hours.Afterward, the coffee was dried at 40°C and stored in a cool and dry place until roasting.The experiments were carried out in a room in the biotechnology laboratory at Cheng Shiu University.

Coffee Roasting Procedure
An automatic coffee roaster with a capacity of 200 g to 3 kg per batch and equipped with a 6,000 W near-infrared heating source was used for roasting the coffee beans.Each batch was 300 g, and the moisture was measured before roasting using a Kett® PM-450 portable grain moisture meter.An Apex® isokinetic sampling train similar to the one recommended by the U.S. EPA  and S3.
Fast and slow methods, taking 5.62 and 9.65 minutes, were employed to achieve a light roast, respectively.Fig. S1 illustrates the temperature-time profiles for the fast and slow roasting methods.The temperature programs for the methods were preset before roasting.The coffee sample was then introduced into the loading hopper and automatically conveyed into the drum roaster, adhering to the designated temperature conditions.The drum of the roaster was maintained at a steady rotation of 40 rpm throughout both roasting processes.Simultaneously with the commencement of roasting, the flue gas sampling equipment was activated and continued to operate until the conclusion of the roasting cycle.Once the roasting was completed, the furnace door opened automatically, allowing the roasted coffee to be transferred to the cooling tray.After cooling, the coffee's moisture content and roast intensity of both the whole bean and ground coffee were measured.To classify the roast degree, the Lighttells® CM 100 portable coffee roast analyzer was employed.The XAD-2 cartridge was subsequently removed, wrapped in aluminum foil, and stored before analysis.
PAHs and N-PAHs were extracted from the coffee bean samples using microwave extraction.First, the coffee beans were ground and homogenized before 5 g samples were placed in a microwave tube.Volumes of 20 µL of 10 mg L -1 stock solution of PAHs (napthalene-d8, acenapthene-d10, phenanthrene-d10, fluoranthene-d10, chrysene-d12, and perylene-d12) and 20 µL of 1 mg L -1 stock solution of N-PAHs (1-nitrophthalene-d7, 9-nitroanthracene-d9, 1-nitropyrene-d9, and 9-nitrochrysene-d11) internal standard solutions were then added into the tube followed by the extracting solvent: a mixture of hexane, acetone, and dichloromethane (2:2:1 v/v) until the solvents covered the sample.The temperature program for the extraction were as follows: the temperature was increased from room temperature to 120°C in 10 minutes and held for 20 minutes.After the extraction, the contents were transferred to a column with an appropriate amount of glass wool, 0.5 cc × 1 of silica gel, and 2.8 cc × 1 of anhydrous sodium sulfate that was sequentially prewashed with two 5 mL of acetone, two 5 mL of dichloromethane, and two 5 mL of n-hexane.This step was carried out to remove water from the sample.The extract was then concentrated to near dryness, then 5 mL of n-hexane was added before near-dryness concentration.The step was repeated one more time before the sample was concentrated to 1 mL.Afterward, the cleanup was carried out using a silica gel column.
For the flue gas samples, the PAHs and N-PAHs adsorbed on the XAD-2 were extracted via Soxhlet extraction for 24 hours using 350 mL hexane, with the internal standards added before the Soxhlet extraction.After extraction, the sample was concentrated to near dryness then 1 mL of n-hexane was added before the cleanup process.The cleanup procedure for both the flue gas and coffee bean samples was carried out using a column of silica gel (4.3 cc × 3) and sodium sulfate (2.8 cc × 1) prewashed sequentially with two 5 mL acetone, two 5 mL dichloromethane, and two 5 mL n-hexane.After, the column was flushed with two 3 mL and two 10 mL of n-hexane.PAHs were eluted using five 5 mL dichloromethane: hexane (1:4 v/v) followed by two 5 mL of n-hexane.Afterward, N-PAHs were extracted using three 10 mL of dichloromethane.The PAH and N-PAH extracts were concentrated at 0.5 mL and 50 µL under a stream of pure nitrogen gas before analysis.

PAH Instrumental Analysis
The sampled flue gas was analyzed for the U.S. EPA's 16 priority PAH compounds using gas chromatography (GC-Agilent 7890A) and triple quadrupole tandem mass (MS/MS-Agilent 7000D).Separation was accomplished using a ZB-PAH-EU 30 m × 0.25 mm × 0.20 µm column and helium gas at a flow rate of 1 mL min -1 .The injection temperature was 280°C.The oven temperature was maintained at 60°C for 0.5 min, then raised to 265°C at 15°C min -1 , and finally ramped to 305°C at 5°C min -1 , and held for 10 min.The ionization source was EI with a temperature of 250°C.Multiple Reaction Monitoring (MRM) was the monitoring mode used.The quality assurance and quality control information are presented in Table S4.

Nitrated PAH Instrumental Analysis
The same GC-MS/MS, column, and carrier gas used in PAH analysis were used to analyze fourteen N-PAH compounds.The temperature program was as follows: the oven temperature was maintained at 120°C for 1 min, then raised to 185°C at 30°C min -1 , and finally to 310°C at 8°C min -1 , and held for 5 min.The ionization source was EI with a temperature of 280°C.MRM was the monitoring mode used.The quality assurance and quality control information are presented in Table S5.

Statistical Analysis
The statistical differences between the roasting methods and correlation between PAH and N-PAH concentrations in the flue gas and coffee beans were carried out using IBM SPSS (Version 22).

Coffee Roasting Parameters
The coffee roasting parameters in this study, namely the temperature-time profiles, moisture content, and color are discussed in our recent pre-print publication (Cheruiyot et al., 2023) and presented in Tables S1-S3.In summary, both roasting methods had the same turning point at 1.4 minutes.The slow roasting method had a lower maximum rate of rise compared to the fast roasting method (25.5 vs. 21.2°Cmin -1 ) and had a more pronounced "first crack".After roasting, the moisture contents of the coffee formulations were lower in the slow roasting method, owing to advanced pyrolytic reactions signaled by the first cracks (Wilson, 2014).Finally, the additives, i.e., honey, sugar, and alcohol, resulted in a much darker roasted coffee, making the Agtron color reference scale not useful for identifying the roast level.This is because the additives affect the coffee formulation by introducing their color and potentially undergoing additional non-enzymatic browning reactions that influence the final color of the roasted coffee (Kontogiorgos, 2021).

Concentrations in the flue gas
To the best of our knowledge, there are no studies that have investigated PAH and N-PAH emissions from the stacks of coffee roasting machines.This knowledge gap is particularly relevant to independent coffee shops that often roast their own coffee in-house.PAH concentrations were detected in all the flue gas samples, as shown in Fig. 1.Notably, when comparing the fast and slow roasting methods, we found that the mass and BaP-TEQ concentrations were higher in the fast roasting method for all treatments.For example, in the case of formula A (only coffee), fast roasting resulted in mass and toxic concentrations approximately 1.6 and 3.2 times higher than slow roasting method (5,069 ng Nm -3 and 14.1 ng BaP-TEQ Nm -3 , respectively).However, a Student's ttest result revealed that only the mass concentrations between the two roasting methods were statistically different (p = 0.05), as shown in Table S4.This suggests that adopting the slow roasting method could potentially reduce PAH mass emissions into the atmosphere across the various coffee formulations.However, use of air pollution control devices like activated carbon would still be necessary to reduce emissions and protect the health of coffee machine operators.
The influence of additives (longan honey, granulated sugar, and alcohol) on the PAH mass concentrations was not obvious, as shown in Fig. 1(a).However, the toxic concentrations had a discernable trend: Samples C and D exhibited notably higher concentrations than A, and B consistently had the lowest concentrations (Fig. 1(b)).The intriguing disparity among the samples warrants a more in-depth examination.To shed light on these intriguing observations, it is imperative to consider the compositional variability of the additives.For instance, honey is composed of various organic compounds, including sugars and phenolic compounds, which could undergo complex reactions including Maillard reaction and caramelization during the roasting process (Kontogiorgos, 2021;Lee et al., 2022a).Likewise, granulated sugar can caramelize, potentially influencing PAH formation, while alcohol may introduce volatile compounds into the system, altering the roasting dynamics.Additionally, the specific concentrations of these additives and their interaction with coffee beans could play pivotal roles in the observed trends.Thus, further comprehensive investigations are necessary to unravel the intricate relationship between these additives, the coffee roasting process, and the resultant PAH emissions.
On the other hand, N-PAH concentrations were below the detection limit in all the coffee treatments.N-PAHs can form via either a primary pathway involving high temperatures and electrophilic nucleating with NO2 + or a secondary pathway involving homogenous gas-phase nitration reaction of PAH and gas-particle heterogenous pathways (Nagato, 2018;Lee et al., 2022b).However, these pathways have only been studied in combustion processes and in the atmosphere and are not verified in roasting processes.That said, the non-detection of N-PAHs could be attributed to the relatively low temperatures during roasting, which might not facilitate the necessary formation of nitrogen oxides and hydroxyl radicals required for N-PAH formation in the gaseous phase.

Concentrations in the Coffee Beans
PAHs in coffee could result from contamination from the environment (deposition from the air, contaminated soil, and water) or formation during roasting (Duedahl-Olesen et al., 2015;Aresta and Zambonin, 2023).Additionally, PAHs could be introduced to the coffee bean at various stages including cultivation, harvesting, storage, and transportation.To determine whether the coffee originally had PAHs, we analyzed PAH concentrations in the green coffee, as shown in Fig. S2.The total mass PAH and PAH4 (BaA, Chr, BbF, and BaP) concentrations in the green coffee bean (A) were 57.8 ng g -1 and 0.543 ng g -1 , respectively.Huang et al. (2023), who purchased green arabica coffee from a local coffee shop, found lower total mass PAH (∑ = 26) concentrations of 28.85 ng g -1 .Green arabica coffee from the same origin as this study, Indonesia, had total PAH concentrations of 47.3 ± 5.55 ng g -1 and PAH4 concentrations of 1.40 ± 0.29 ng g -1 (Ciecierska et al., 2019).That study showed that PAH concentrations in green coffee varied with origin, with the lowest found in Cuban green arabica coffee (8.66 ± 0.88 ng g -1 ) and the highest in Tanzanian green arabica coffee (59.6 ± 6.18 ng g -1 ).
Unlike in the flue gas, the difference between PAH concentrations in the fast and slow roasting methods was not clear, as shown in Fig. 2. In certain treatments (B-3 and D-1), the fast roasting methods resulted in higher concentrations while in all other cases the reverse was observed.Additionally, there was no statistical difference (p = 0.05) between the concentrations of the two roasting methods, as shown in Table S4.All the concentrations, except in C-1, D-2, and D-3 (in both roasting methods), B-3 and D-1 (in the fast roasting method), were lower than in the green coffee.Similarly, roasted coffee was reported to have lower concentrations than green coffee in several previous studies (Ciecierska et al., 2019).Contrarily, some studies found increased PAH concentrations after roasting (Houessou et al., 2008;Huang et al., 2023).Table 2 shows a comparison between PAH concentrations in this study and previous studies.According to the table, the PAH concentrations in A (only coffee) were 34.8 ng g -1 and 40.5 ng g -1 after fast and slow roasting respectively, which was much lower than in previous studies (Jimenez et al., 2014;Huang et al., 2023).For instance, Jimenez et al. (2014) reported that total PAH concentrations in three different brands of lightroasted coffee were in the range of 333-740 ng g -1 (∑ = 18).We also roasted the green coffee under dark roasting conditions to compare the concentrations with the light roast.The concentration in the dark roast was 37.5 ng g -1 , which was lower than in the light roast.This was contrary to   et al., 2015) other previous studies that reported an increase in concentrations with roast level.Huang et al. (2023) found that total PAH, BaP, and PAH4 concentrations were higher in darker than lighter roast.Furthermore, we do not expect the transfer of PAHs from the roasted bean to the brewed coffee to be significant because of the hydrophobic nature of the pollutants.The transfer from the roasted bean to the ready-to-drink coffee was found to be less than 35%, with extractability being higher in lighter than darker roasts (Houessou et al., 2006;Duedahl-Olesen et al., 2015;Aresta and Zambonin, 2023).Besides, mostly the less toxic low molecular weight PAHs, with lower toxicity, transfer into the brewed coffee (Huang et al., 2023).Additionally, the residual PAH concentrations in the cup of coffee would depend on the brewing method and how it is consumed.
Conversely, the concentrations of N-PAHs in the green coffee and roasted coffee beans were all below detection limits (LOQ = 0.1 ng g -1 ), except fast roasting of B-3 (0.503 ng g -1 ).This observation was different from Ko et al. (2018) who detected N-PAH during coffee roasting at 0 to 250°C for 5, 10, and 20 minutes.That study found that the total concentration (∑ = 7) increased with temperature and time.Based on the limit of quantification (LOQ), the analytical method used in this study was more sensitive than Ko et al. (2018)'s method (0.1 ng g -1 vs. 0.353-0.502ng g -1 ), suggesting that our study method is capable of detecting and quantifying much lower N-PAH concentrations, ruling out the possibility of non-detection because of inferior analytical method.Other several factors including varied roasting machine and conditions, coffee bean characteristics, and sample preparation could explain the disparity.

Congener Profile
The PAH compound profiles are presented in Fig. 3.The profiles of the flue gas were starkly different from the coffee bean.The flue gas profile was predominantly composed of naphthalene, a two-ringed PAH compound.However, the coffee bean profile had significant fractions of threeringed compounds.Generally, there was not much difference between fast and slow roasting methods besides certain formulations, e.g., B-3 and D-2.Jimenez et al. (2014) found acenaphthylene, fluorene, pyrene, chrysene, and indeno(1,2,3-cd)pyrene as the most abundant compounds during light roasting.Huang et al. (2023) found green arabica coffee had predominantly low molecular PAHs namely naphthalene, acenaphthylene, fluorene, phenanthrene, and anthracene.2-nitropyrene was the only N-PAH congener detected in the fast roasting of B-3.
The PAH diagnostic ratios were used to assess the potential source of PAHs in green coffee (Fig. S5).Only the anthracene/phenanthrene and fluoranthene/pyrene ratios were used for the diagnostic ratio analyses since the concentrations of benzo[a]anthracene, indeno[1,2,3-cd]pyrene, and benzo[g,h,i]perylene were below the detection limit.The Ant/(Ant+Phn) and Fla/(Fla+Pyr) ratios were 0.108 and 0.487 indicating that the source of PAH in green coffee was pyrogenic in nature and more specifically, petroleum combustion (Yang et al., 2014).Therefore, it is possible that these PAHs could be introduced to green coffee during transportation.

Emission Factors
Emission factors are important in estimating the amount of pollution emitted by a source.The emission factors during roasting were calculated using the concentrations, the dry exhaust flow rate, and the duration of the roasting, with the results presented in Fig. 4. In both total and BaP-TEQ emission factors, the fast roasting method had higher values than slow roasting.Therefore, the slow roasting method is better for reducing PAH emissions.The emission factors for the fast and slow roasting of green coffee were 1.36 mg kg -1 of coffee (7.49 µg BaP-TEQ kg -1 ) and 0.483 mg kg -1 (1.35 µg BaP-TEQ kg -1 ), respectively.

CONCLUSIONS
This study investigated PAH and N-PAH concentrations in the flue gas during coffee roasting and in the coffee bean.The results showed that PAH concentrations were detected in all flue gas samples and in the roasted coffee formulations.For the flue gas, the slow roasting method resulted in statistically (p = 0.05) lower PAH mass concentrations than fast roasting.For example, the PAH concentrations from fast and slow roasting green coffee were 8203 ng Nm -3 (45.1 ng Nm -3 ) and 5069 ng Nm -3 (14.1 ng Nm -3 ), respectively.Nevertheless, it is worth noting that additional measures, such as the use of air pollution control devices like activated carbon, might still be necessary for further emission reductions.Furthermore, the PAH concentrations in the roasted coffee were lower than in the green coffee.These results are particularly relevant to independent coffee shops that often roast their own coffee in-house.

Fig. 1 .
Fig. 1.Comparison of PAH (a) mass and (b) toxic concentrations in the flue gas of different coffee blends during the fast (orange) and slow (grey) roasting methods.

Fig. 2 .
Fig. 2. PAH concentrations in the roasted coffee bean.The fast and slow roasting methods are represented with orange-and gray-colored bars, respectively.

Fig. 3 .
Fig. 3. PAH congener profiles of the (a) flue gas and (b) coffee bean samples during fast and slow roasting methods.

Fig. 4 .
Fig. 4. Comparison of PAH emission factors based on (a) mass and (b) toxic concentrations from different coffee blends during the fast (black) and slow (grey) roasting methods.

Table 1 .
Details of the coffee formulations used in the study.

Table 2 .
Comparison of total PAH, BaP-TEQ, and PAH4 concentrations from literature.