Effect of fermentation on the physicochemical characteristics and sensory quality of Arabica coffee

Abstract

This work aims to assess the physicochemical characteristics and final sensory quality of Yellow Catuai IAC 62 Arabica coffee fermented with Saccharomyces cerevisiae. For such a purpose, a Composite Central Rotational Design (CCRD) was performed to investigate how fermentation time,temperature and pH conditions, moisture content and concentration of sugars and organic acids affect its sensory quality on two different roast levels in accordance with Specialty Coffee Association (SCA) protocols. It was found that fructose concentration decreased from 12 g/L to around 5 g/L during fermentation, regardless of temperature condition. Furthermore, longer fermentation times and higher temperatures have lowered sucrose and glucose concentrations from 4 to 2 g/L and 7 g/L to zero, respectively. Glycerol concentration was higher as time and temperature increased, and optimal conditions ranged at temperatures between 24 °C and 32 °C from 35 to 45 h of fermentation time. pH decreased as fermentation time elapsed, but there was a more significant reduction due to higher temperatures, starting at around pH 5 and, lower than 4 under extreme conditions. Contents of organic acids such as acetic, propionic, succinic, and lactic acids, were measured at the final stage of each fermentation process under studied conditions. It was observed that coffee samples achieved final scores ranging from 81 to 85 (SCA score), even in longer times and extreme temperature conditions, thus all samples have been classified as specialty coffees. This work described the initial step towards parameterizing fermentation processes, given that the response variables of temperature and fermentation time, were optimal and enhanced the sensory quality of coffee as beverage. Saccharomyces cerevisiae, a commercial product which has already been made available for producers, can ensure an increase in the sensory quality of coffee.

Introduction

Coffee is a popular beverage worldwide, and Brazil was the leading coffee-producing country from 1994 to 2019, according to the Food and Agriculture Organization of the United Nations (FAOSTAT 2019). Between 2019 and 2020 crop year, Brazil produced over 58 million 60 kg jute bags of coffee, in addition to exporting over 17 million 60 kg jute bags from May to October 2021 alone (ICO 2022).

Its price as commodity can be significantly affected by its quality. Specialty coffee quality as a beverage is rather extensive, once it depends on the chemical composition of beans, and several factors, such as harvesting, processing, preparation, and roasting methods (Pimenta et al. 2018). Elhalis et al. (2020a) believe that post-harvest processing significantly alters its final taste, thus directly affecting the concentration of alcohols, sugars and acids within it: nonetheless effects of the entire process are still unknown.

Its post-harvest processing stages occur in two different ways, i.e., either through wet or dry processes. The wet process results in naturally pulped or washed coffee; after harvesting and cleaning, its cherries and green fruits are separated mainly by peeling the fruit's outer skin, resulting in a parchment bean. The dry process consists in simultaneously fermenting and drying coffee right after harvesting, which usually takes around 20 days (Brando and Brando 2014; Silva 2014). Generally, this process is thoroughly aerobic and can retain the highest glucose and fructose concentration within fruits (Knopp et al. 2006). These chemically resemble raw coffee and have higher soluble solid and total sugar concentrations (Ribeiro et al. 2011). Consequently, its body is fuller, and it has more remarkable sweetness. In both cases, there might be a fermentation stage which, under controlled conditions, can also produce chemical transformations, thus positively affecting its quality.

Several authors evidenced the use of microorganisms to initiate fermentation (Evangelista et al. 2014b, 2015; Pereira et al. 2015; Ribeiro et al. 2017; Martinez et al. 2017; Elhalis et al. 2020a) as conductors of an appropriate fermentative process towards producing unique coffees whose cupping scores are above 80, according to Specialty Coffee protocols established by the SCA (Specialty Coffee Association of America 2020). During fermentation, microorganisms produce several metabolites. The microbial activity and extent of fermentation determine the concentration of free sugars (fructose and glucose, for instance) and free amino acids, which remain in the bean and subsequently lead to the production of compounds through Maillard reactions and volatile compounds during the roasting process (Haile and Kang 2019b). Several studies also demonstrate that the addition of selected yeasts to wild yeasts from the coffee cherry microbiota affects its aroma.

Its fermentation process can occur in a submerged medium or solid state. Bioreactors can provide controlled environmental conditions for fermentation enhancement (Tang et al. 2021). Solid fermentation is a process carried out using solid materials, either in the absence of or having low moisture, so that they can act as physical supports and be a source of nutrients for microorganisms. Due to low moisture content, a limited number of microorganisms can develop in solid fermentation. Saccharomyces cerevisiae is an important microorganism able to adapt well to solid fermentation conditions which can, therefore, be used to produce food and beverages worldwide (Santos da Silveira et al. 2019). S. cerevisiae has been reported to improve the sensory quality of coffee, as a fruity aroma is produced at the end of the roasting process (Bressani et al. 2018; Wang et al. 2020).

Despite scientific evidence about the benefits of controlled fermentation for enhancing its quality, most coffee fermentation processes are still carried out worldwide with no environmental control and coffee beans are spontaneously produced with inconsistent and unpredictable quality (Elhalis et al. 2020b). Farmers are often unaware of or may find it complex to control fermentation conditions to produce superior quality coffee. Therefore, this study aimed to assess the effects of Arabica coffee fermentation, i.e., the Yellow Catuaí strain, on its physical–chemical characteristics (pH, sugars, glycerol, organic acids) and sensory quality. The results found through this study are the initial steps taken to apply the bucket method of coffee fermentation in farming environments so as to optimize the process and enhance its quality.

Material and methods

Material

The rural property selected to harvest coffee fruits is located in the greater coffee production region of Carmo do Paranaíba, Minas Gerais—Brazil, found through the following geographical coordinates: Lat.: 18 57′37″ S/Long.:46 36′17″ O at an altitude of 1080 m. The Arabica coffee cultivar, Yellow Catuaí strain, IAC 62, is a variety able to produce the largest number of fruits whose beans are called flat, medium size, retained in a #16 sieve, with no mochas, and only large flat beans were selected for sensory evaluation. Generally, over 82% are flat beans with excellent beverage quality.

Samples were collected during the 2019/2020 crop year by randomly harvesting 500 L of coffee fruits, out of which 220 L were ripe. The harvesting process occurred at approximately five different intervals, and there was consistent coverage across the coffee field.

Sample collection and preparation

Coffee was harvested manually to select only ripe fruits having a more significant potential for quality enhancement as cherries, i.e., yellow fruits having a brass color and no greenish color whatsoever. The harvested fruits were placed in plastic containers a new classification was carried out after harvesting to sort out fruits of higher quality. All fruits were submerged in water using a 500 L tank to simulate the same process performed by coffee washers, who typically separate dried fruits or those presenting granulation problems. Dried fruits or those having grainy problems (afloat) are less dense than water and float, which were thus discarded. Immature fruits were selected manually and discarded due to their astringency. Cherry fruits considered for fermentation were those having a bronze-yellowish color throughout the skin's length. The selected fruits were then stored in plastic containers with cold water and placed in a cooler with ice so that temperature remains low (< 4 ºC) according to the method used by Neto et al. (2018). This procedure avoids the fermentation process to occur as samples were transported from the farm to the laboratory where experiments and analyses would take place.

Incubation

The commercial yeast S. cerevisiae, brand Lallemand ORO™ (10¹⁰ UFC/g), was kindly provided by Lallemand. The ratio used for incubation was 12 g (0,012 kg) of yeast for every 10 kg coffee based on manufacturer´s recommendations. The yeast was weighed on an analytical balance (Shimadzu, model BL-3200H), dissolved in distilled water, and applied in a mass ratio of 10 g of commercial yeast for every 500 mL of distilled water. The Neubauer chamber method was used for cell counting with a previously prepared diluted solution of 1 ml of sample for every 100 ml distilled water, according to Eq. 1, where C₁ and C₂ were quadrant mean counts: the Neubauer chamber cell counting resulted in about 2.55 × 10⁸ cell/mL of solution.

$$\text{Cell}\text{count}\left(\frac{\text{Cell}}{\text{ml}}\right)=\left(\frac{C_1+C_2}{2}\right)\times 2.5\times {10}^5\times \frac{1}{100}$$ 1

Solid fermentation

Fermentation was carried out through a batch process. Each batch was fermented using 6 L of coffee divided into open 2L beakers and approximately 2.4 kg of sample mass. An incubator with temperature control was used for fermentation (Tecnal, model TE-421). Each experiment temperature was determined, according to the experimental design (CCRD) 200 g samples were collected every 4 h to measure moisture, pH, and concentration of sugars and organic acids. The bean mass was manually turned to ensure aerobic fermentation environment while sampling.

Drying and treatment of final samples

The beans were dried in an oven with forced-air circulation (Ethik Technology, Model 400/8D) at controlled temperature and intermittent drying processes, as it is traditionally adopted in conventional coffee processing. After 50 percent of the drying process (fruit peel darkening), the process was carried out in drying cycles lasting 8 h at 37.5 ºC, and the bean mass was measured at room temperature within the remaining 8-h intervals (Isquierdo et al. 2013).

After drying, samples were adequately packed in high-strength low-density polyethylene bags, and then labeled and packed in a hermetically sealed BOD Incubator at 25 °C (Solab, Model SL-225/364) to prevent the entry of light and contamination. Coffees were stored for 60 days and removed for milling afterwards. A mechanical coffee mill was used to completely remove the exocarp and dry endocarps of samples found on the farm (Model DRC; Pinhalense). This stage transforms dry cherries into green raw coffee beans with no classification (this is raw coffee without proper separation by size and removal of defects, just as it is performed in commercial coffee evaluation). Afterwards, samples were completely placed in plastic packaging traditionally used to store coffee samples. Thus, samples were sent directly to classification to sort out those having defects (impurities and imperfect beans) and ensure that only large beans (beans retained in the 16-coffee sieve without mochas) and no shelled beans remained for performing the sensory analysis (Specialty Coffee Association of America 2020).

Sample classification

Samples were classified based on the standards established by IN 08/2003 (Brasil 2003). All intrinsic and extrinsic defects were thoroughly removed. For the separation of large beans, #11 oblong test sieves (separation of mocha beans) and round sieves sized 16/64″ were used (the same as Sieve #16 used in coffee classification). Only beans retained in the 16/64″ sieve without straws were selected for the classification stage.

Physicochemical analysis of moisture and pH conditions during the fermentation process

Moisture assessment consisted in placing samples in previously weighed crucibles, which were taken to an oven (Nova Ética, model 402-3N) at 105 ºC for 24 h and then placed in desiccators to be cool weighed (Brasil Ministério da Agricultura 2009).

Part of the coffee fruit sample was crushed using an industrial blender for a minute (Camargo, 2L 800W/22000 rpm) until a uniform paste was formed and subjected to pH measurement using an adequately calibrated (Instrutherm, model PH-1700), according to the methodology described by Velmourogane (2013). All analyses were performed in triplicate.

CCRD to assess the influence of temperature and fermentation time on sugars and residual acids selected for the sensory analysis

A Central Composite Rotatable Design (CCRD) was carried out to analyze the coffee fermentation process and find optimal process conditions in studied ranges. Studied variables were process temperature ${(x}_1)$ and fermentation time ${(x}_2)$.

The temperature selected for the CCRD ranged from 15 to 30 ºC, considering the average temperature conditions in the Cerrado Mineiro Region and other works found in literature (Evangelista et al. 2014b). Fermentation time ranged from at least 12–48 h as in Avallone et al. (2001) and Pereira et al. (2015).

The experimental design consisted of 11 central replicates and α 1.4142. The variables of temperature (°C) and fermentation time (hours) were studied in the range of 11.89 (-α) to 33.10 (α) and 4.42 (-α) to 55.50 (α), respectively. Dependent variables were carbohydrates (sucrose, glucose, and fructose), organic acids (acetic, propionic, succinic and lactic acids), glycerol concentration and sensory analysis. Multiple regression modeling was performed to assess the influence of studied independent variables ($x_1$ and $x_2)$ on dependent variables, and parameters having significance level of over 10% were disregarded, i.e., in the hypothesis test with Student's t-table, it was considered a maximum error probability of 10%. In the regression analysis of coffee fermentation results, independent variables were transformed into their dimensionless form according to Eq. 2 for temperature (T) and Eq. 3 for fermentation time (t). In the fast-drying process, fermentation was stopped just after the fermentation time defined in the experimental design.

$$x_1=\frac{T\left({}^{\circ }C\right)-22.5}{7.5}$$ 2

$$x_2=\frac{t\left(h\right)-30}{18}$$ 3

Determination of sugars, glycerol, and organic acids

Sugars and organic acids were analyzed by placing the fermented coffee mass (entire fruit: exocarp, mesocarp, endocarp, and bean) into aluminum trays which were then exposed to the sun’s radiation for natural drying until it was dry enough to be ground using an analytical mill (about 12 h of sun drying, to facilitate mass processing and posterior extraction) considering average daytime temperature of 23.2 °C, once the oven was full of fruits with fermented coffee mass, oven moisture close to100% so that no dry coffee fruits remained in the sensory analysis. The used mill was the Willye knife (STAR FT-50, Fortinox) and the sample was ground using a 20 mesh sieve to separate the material used for extraction. Crushed samples of the fermented dough were weighed (6 g) and placed in 125 mL Erlenmeyer flasks with 40 mL of ultrapure water from a purifier (Gehaka, Model Master System All) to maintain the same proportion as that described by Ribeiro et al. (2017). The mixture was agitated using a magnetic stirrer for 10 min at room temperature to extract the samples. The extract was decanted and centrifuged (at 10,000 rpm and 4 °C for 10 min) in a centrifuge (Heal Force, Model Neofuge 15R) according to the methodology described by Ribeiro et al. (2017). The supernatant was filtered through a 0.22 μm cellulose acetate filter (for particles smaller than 0.22 μm) to be submitted to High Precision Liquid Chromatography (HPLC) after centrifuging.

Sugars and glycerol concentrations were determined using a HPLC (Shimadzu LC-20A) equipped with a refractive index detector, a Hi-plex Ca column (7.0 × 300 mm, Agilent, CA, USA) operated at 85 °C using ultra-pure water as mobile phase at a flow rate of 0.6 mL min⁻¹. The HPLC (Shimadzu LC-20A) was equipped with a diode array detector, a Shim-pack VP-ODS C8 phenyl column (150 × 4.6 mm) and operated at 30 °C and 0.6 mL min⁻¹ flow rate to determine organic acid concentrations. Mobile phase A consisted of a 0.01 M potassium dihydrogen phosphate buffer solution (H₃PO₄ at pH 2.50), and phase B was acetonitrile. Elution was set as follows: 0.00–3.00 min, 0% B; 3.00–5.00 min, 0–15% B; 5.00–8.00 min, 15—0% B; 8.00–10.10 min, 0% B; 10.10–15.00 min, 0–15% B.

Sensory analysis

Sensory analysis was conducted once the project was approved by the University’s Human Research Ethics Committee under protocol number 55909622.0.0000.5152. As regards cupping (sensory analysis of coffee), samples were classified and judged according to Specialty Coffee protocols (Specialty Coffee Association of America 2015). Two Q-Graders (Specialty Coffee Professional Grader) panels were formed for the sensory analysis, composed of eight certified judges. One of which took place in the town of Patos de Minas at Farroupilha Group’s Coffee Laboratory which was kindly allowed to be used by the group’s Board. It counted on the participation of 3 trained judges. The second panel was in Carmo do Paranaíba town—at Veloso Coffee’s Coffee laboratorywhich was kindly allowed to be used by the group’s Board, relying on the participation of 5 trained judges.

Samples were roasted on a Specialty Coffee Roaster (LABORATTO, Carmomaq, São Paulo, Brazil) before 24 h of coffee cupping. Each sample was roasted on two different levels: light–medium roast and medium roast (#65 and #55 Agtron Measurement, as described by ABIC, 2018), and tasted at each roast level in every experiment. Roasted coffee beans had been ground just before they were brewed in a coffee mill (ML1-NA, Pinhalense, São Paulo, Brazil) so as to grind a small amount of each sample before grinding all samples.

Coffee tasting was performed by blinded tasters; each sample had a code, and none of the tasters knew what process each code would represent. Ground coffee was evaluated first to evaluate its fragrance, brewed using hot water at around 90 °C and then the aroma of brewed coffee was assessed. After an olfactory analysis, cups were cleaned (excess of ground material withdrawn), and samples were tasted several times to evaluate gustatory analysis. Sensory analyses were performed according to the Specialty Coffee Association (SCA) Protocol, which evaluates 10 coffee attributes (Fragrancy and Aroma, Flavor, Aftertaste, Acidity, Body, Balance, Sweetness, Clean Cup, Uniformity, Overall, scored on a scale of 0 to 10, and its final score is the sum of all features) (Specialty Coffee Association of America 2020) and different coffee nuances, in accordance with the Coffee Taster’s Flavor Wheel (Association and World Coffee Research 2016).

Results and discussion

Moisture and pH

There was no moisture in the fermentative mass during the process. The samples' average initial moisture was 70.62 ± 1.03, but 68.67 ± 1.39 was found at the end of the process (wet basis). Figure 1 shows the initial and final pH conditions of the fermentation process for each CCRD experimental condition. Initial pH values of the fermentative mass were 5.4 on average and 0.25 of standard deviation was found.

Fig. 1.

Comparison of initial and final pH values of rotational central composite design experiments

The pH values observed here in were similar to those found by Elhalis et al. (2020a) in spontaneous fermentation (without inoculation), i.e., around 5.5. Other studies have shown that the initial pH of coffee beans ranges from 5.0 to 7.0 (Kwak et al. 2018; Haile and Kang 2019a).

As the fermentative process progressed, the metabolism of microorganisms allowed using the available sugars to form soluble organic acids as metabolites (Avallone et al. 2001; Kwak et al. 2018), thus promoting a reduction in pH values, a fact which had also been observed by Haile and Kang (2019b). Moreover, it was also evidenced herein that pH reduction varied in accordance with test temperatures. This fact was also verified by Avallone et al. (2001) who found a relationship between a decrease in pH and temperature, as lower pH values were found at night due to lower temperature.

Figure 1 shows the results of tests 5 and 6 (same fermentation time), through which it is possible to observe the temperature effect on pH, given that greater temperatures lead to decreased pH. Furthermore, tests 8 and 9 (same temperature) reveal that longer times also reduce pH, thus time and temperature affect pH conditions.

Using the Tukey mean difference test at 5% significance to the experiments at the central point of the experimental design, under the conditions of these experiments there was a significant difference, and the final pH value was lower. Also, by multiple regressions applied to the results of the entire set of experiments of the central composite design, it was verified that temperature and fermentation time influenced the initial pH (R² = 0.9872) and the final pH (R² = 0.9262). In the initial pH value, the effect of temperature was more pronounced compared to the effect of fermentation time, and in general, higher temperatures tended to lead to lower initial pH values. For the final pH, it is noted that the two factors influenced with similar intensity of their effects, and in general, the final pH was lower when combining higher temperatures and fermentation times.

Influence of temperature and fermentation time variables on carbohydrates, residual acids, and sensory analysis

The CCRD results are shown in Table 1. The evaluated response variables were sensory analyses of coffee fermented at light–medium and medium roast level. Other evaluated responses were the concentrations of sugars (sucrose, glucose, and fructose), glycerol and final organic acids affecting the score in the sensory analysis of fermented coffee.

Table 1.

Experimental planning with the evaluated responses (sugars, glycerol, organic acids and final grade of sensory analysis) concerning the variables studied

	Real value (coded value)		Carbohydrates			Alcohols	Organic acids				Final grade
	Temperature ($x_1$) (°C)	Time Fermentation ($x_2$) (h)	Sucrose (g/L)	Glucose (g/L)	Fructose (g/L)	Glycerol (g/L)	Acetic (mg/g)	Propionic (mg/g)	Succinic (mg/g)	Lactic (mg/g)	Light medium roast (SCA)	Medium roast (SCA)
1	15.00 (-1)	12.00 (-1)	4.5382	6.6934	10.5473	0.1634	2.0000	3.2733	27,667	0.0000	81.30	81.19
2	15.00 (-1)	48.00 (1)	3.0540	7.0421	6.0881	0.5641	9.3333	4.2200	18.000	0.0000	83.55	83.11
3	30.00 (1)	12.00 (-1)	2.6863	7.2341	11.1752	0.3914	13.333	3.3600	0.0000	0.0000	83.53	83.88
4	30.00 (1)	48.00 (1)	0.5159	1.8150	5.4760	0.7682	0.0000	4.6133	0.0000	0.0000	82.06	81.00
5	11.89 (-α)	30.00 (0)	4.0506	8.0709	8.8953	0.1800	48.000	0.0000	4.2133	5.8867	83.54	82.01
6	33.10 (α)	30.00 (0)	1.7347	4.6283	9.5563	0.9095	42.000	0.0000	0.0000	7.5933	84.42	82.98
7	22.50 (0)	4.42 (-α)	4.4759	7.1148	12.2048	0.0000	73.333	0.0000	0.0000	9.4000	82.25	83.44
8	22.50 (0)	55.50 (α)	0.6205	3.0667	5.1005	0.7244	28.000	0.0000	0.0000	3.0667	82.56	82.54
9	22.50 (0)	30.00 (0)	3.8526	5.0698	8.9730	0.7844	22.666	0.0000	0.0000	8.4000	84.47	82.31
10	22.50 (0)	30.00 (0)	3.1957	6.1205	8.7715	0.7071	28.000	0.0000	0.0000	9.1333	84.32	82.41
11	22.50 (0)	30.00 (0)	3.8671	6.7429	9.3328	0.7846	25.333	0.0000	0.0000	8.1333	84.39	82.27

Sugars and glycerol

The reduced regression models for fructose and glucose concentrations at the end of the fermentation process are shown in Fig. 2a and b, respectively. Figure 2a reveals that temperature has not influenced fructose concentration results at the end of the process. As for fermentation time, it was observed that a reduction in fructose concentration is evident at longer times. On the other hand, response of glucose concentration showed a different profile than fructose concerning the variables under study. As shown in Fig. 2b, the response surface resembled a saddle. It is possible to observe that the highest glucose concentration values are found at higher temperatures and shorter fermentation times, but at longer fermentation times and lower temperatures.

Fig. 2.

Respose surface and contour curve for the effect of the temperature and the time at the end of the fermentation process for: a fructose concentration, b glucose concentration, c sucrose concentration, d glycerol concentration, e SCA sensory assessment for light medium roasting (#65, SCA), f SCA sensory assessment for medium roasting (#55, SCA)

Initial sugar content was essential to determine the consumption and generation of mono and disaccharides. Fructose is known to have greater initial concentration, whose average values were 10.92 g/L (σ = 1.29), 8.09 g/L (σ = 0.97) and 4.90 g/L (σ = 1.13) for fructose, glucose, and sucrose, respectively. In the fermentative process of coffee, sugar reduction occurs due to the metabolic activity of microorganisms, thus forming metabolites, such as acids and alcohols, among others (De Bruyn et al. 2017; Martinez et al. 2017; Elhalis et al. 2020b). At the end of the fermentation process, the main sugars found here in were fructose, glucose, and sucrose; the latter was found at lower concentrations.

Figure 2c and d show the surfaces of the reduced regression model for sucrose and glycerol concentration at the end of the fermentation process. In the experimental design (Fig. 2d), the response surface showed an optimum tendency for glycerol concentration between 30 and 50 h of fermentation time at temperatures above 24 °C. The highest concentrations of glycerol were found at central points for both variables (22.5 °C and 30 h). Moreover, it was observed in experiment 6 that the highest temperature was selected for all experimental designs (33.1 °C) at the central point of fermentation time (30 h).

Fructose and glucose reached the highest concentrations as a result of sucrose hydrolysis by the yeasts present, mainly S. cerevisiae, thus transforming glucose and fructose into monosaccharides. Yeast consumes glucose first and then fructose due to its metabolism.

As shown in Fig. 2c, sucrose concentration reached higher values at lower fermentation times and temperatures. Its degradation is due to microbiological activity, which might form monosaccharides. Sucrose is the main sugar used in yeast metabolism.

The results of sugar concentration reached the lowest values in experiments at longer fermentation times and higher temperatures. Elhalis et al. (2020b) reported that sucrose and monosaccharides such as fructose and glucose were practically degraded at the end of fermentation process, conducted for 24 h at room temperature (between 10 and 30 °C). These sugar concentration reductions were also observed in other studies, such as in De Bruyn et al. (2017) and Martinez et al. (2017a).

In some of the tests conducted in this work allowed finding high sugar concentration, thus evidencing that fermentative mass sugars were not entirely degraded. Ribeiro et al. (2017) observed an increase in glucose and fructose at the end of the fermentative process which is related to enzymatic or hydrolytic reactions from the fermentative process. Polygalacturonase and pectinase enzyme activities, mainly produced by yeast fermentative processes, degrade pectinolytic polysaccharides (Masoud and Jespersen 2006). This process has a remarkably positive effect on coffee quality (Lee et al. 2015).

According to Avallone et al. (2001), the most easily metabolized sugars are prioritized by microorganisms before the hydrolysis of polysaccharides. This means that the coffee fermentation process is dynamic in consumption kinetics and sugar generation. De Bruyn et al. (2017) found that despite the concentration of sucrose and monosaccharides having decreased, glucose and fructose reached peaks during the fermentation process, which also corroborate the hypothesis that these sugars are both substrate and metabolites in biochemical reactions occurring in the process fermentation. Other authors also evidenced an increase in monosaccharide levels, such as Ribeiro et al. (2017). They justify it due to the breakdown of polysaccharides. This fact was also discussed by Haile and Kang (2019b) while describing that microbial activity and fermentation time determines the final concentration of sugars such as glucose and fructose.

Glycerol concentration at the end of the process is another interesting response in the study of coffee fermentation. Glycerol is an essential metabolite for quality improvement since it has a sweet taste and smooth mouth sensation (Swiegers et al. 2005). Glycerol concentration was zero in experiment 7 (shown in Table 1). The shortest fermentation time was found in this experiment, i.e., just 4.42 h. It can be explained by the fact that glycerol is a metabolite from sugar degradation by yeasts. It was also not detected in mechanically husked beans without fermentation (Elhalis et al. 2020a, b). Elhalis et al. (2020b) detected glycerol after 24 h of fermentation, which may be explained by their process, which partially removes the exocarp and mesocarp, thus reducing polysaccharides content.

The results section presents the central points and experiment 6 (both in Table 1) which reveal that time and temperature allowed reaching the highest glycerol concentrations. Similar results were observed by De Bruyn et al. (2017), who found spikes in glycerol concentration during the beginning of dry fermentation, but not at its end.

Sensory analysis

The sensory evaluation of coffees was performed at two different bean roast levels and tasted according to the SCA classification. The sensory analysis results of fermented coffees, as shown in Table 1, showed a slight variation in their score, i.e., between 81.33 and 84.47 for coffees tasted at light–medium roast level (Agtron # 65/SCA). At medium roast level (Agtron # 55 / SCA), coffee scores ranged from 81.00 to 83.88. Its results regarding different roasting methods showed that there was a significant dependence on the coffee scores for the two variables, temperature ${(x}_1)$ and fermentation time ${(x}_2)$, regardless of the used roasting level. Figure 2.e shows the responses of the reduced regression model and the contour surface selected for scoring average clear roasting according to the SCA concerning the studied variables.

The central levels of both variables showed optimal sensory results. Fermentation time ranged from 20 to 40 h and temperatures from 16 ºC to 28 °C for reaching optimal sensory scores. Table 2 presents some works on the inoculation of microorganisms in coffee fermentation.

Table 2.

Inoculation in different cultivars and fermentation conditions

Author	Cultivar	Temperature	Fermentation time	Method	Inoculation	Final grade
Elhalis et al. (2020a)	Bourbon	Air temperature (25–30 °C—day e 10–15 °C night)	36 h	Wet process	Spontaneous	89.50
					Natamycin (anti-Yeast)	84.75
					T. delbrueckii 084	85.50
					Spontaneous	91.50
Bressani et al. (2018)	Yellow Catuai	Air temperature	16 h	Dry process	S. Cerevisiae 0543	84.00
					C. parapsilosis 0544	81.50
Carvalho Neto et al. (2018)	Catuí	30 °C	12 h aerobic and 12 h anaerobic	Wet process	Spontaneous	80.67
					Lactobacilus Plantarum	80.00
Ribeiro et al. (2017)	Mundo Novo	Air Temperature	Over drying process 284 h	Semi-dry process	S. Cerevisiae 0200	80.13
					S. Cerevisiae 0543	82.63
					Spontaneous	-
	Yellow Ouro				S. Cerevisiae 0200	83.25
					S. Cerevisiae 0543	82.88
					Spontaneous	81.38
Martinez et al. (2017)	Yellow Catuaí	Air temperature 14.6 °C—28.2 °C	352 h	Semi-dry process	S. Cerevisiae 0543	81.40
					C. parapsilosis 0544	81.30
					T. delbrueckii 084	81.00
					Spontaneous	81.40
Pereira et al. (2015)	Catuí	Air temperature (24–32 °C—day and 12–15 °C night)	24 h	Wet process	P. fermentaris YC.2	89.00
					P. fermentaris YC.2 Sup	87.50
					Spontaneous	89.00
Evangelista et al. (2014b)	Acaiá	Air Temperature	Over Drying process	Semi-dry process	Controle	80.93
					S. cerevisiae YCN 724	79.33
					P. guillermondii YCN 731	74.17
					C. parapsilosis YCN448	80.00
					S. cerevisiae *YCN 727	81.08

*YCN 727 = CCMA 0543

Figure 2f shows the responses of the reduced regression model and the contour surface selected for the average roasting score of variables under study.

Before the sensory analysis, a physical evaluation of raw beans was carried out, as recommended by the Ministério da Agricultura Pecuária e Abastecimento (MAPA) Normative Instruction, nº 8 of 2003 (Brasil 2003), by the main author who is a certified Brazilian Grader. The physical evaluation consisted in assessing the appearance, color, and percentage of burnt beans so as to detect signs of undesirable fermentation processes.

The response surface shows that shorter fermentation times are required for carrying out a better sensory evaluation at higher temperatures. When the fermentation temperature is lower, it takes a long time to reach desirable values in the sensory evaluation. Optimal results were found in experiments 9, 10 and 11 at central point (22.50 °C and 30 h) and experiment 6 (33.10 °C and 30 h), and sensory scores were greater than 84 points. A control experiment was performed to make a comparison of sensory evaluation scores with experimental design results. In this experiment, inoculation using the yeast S. cerevisiae was not performed for the control. Fermentation control was not performed since the light–medium roast level is a commercial standard. According to the SCA, control (without fermentation) reached a sensory score of 82.16. This result indicates that the experiments carried out here in showed sensory results superior to those found in control. Bressani et al. (2018), Ribeiro et al. (2017) and Martinez et al. (2017a) obtained better results with coffees inoculated using S. cerevisiae if compared to their control results.

Similar results were observed in experiments on the inoculation of microorganisms by several authors. Bressani et al. (2018) found a maximum score of 84 points by inoculating the S. cerevisiae strain CMA0543 through the dry process. Pereira et al. (2015), Neto et al. (2018), and Elhalis et al. (2020b) conducted 24 h experiments (12 aerobic and 12 anaerobic fermentation hours) and 26 h of fermentation only by controlling temperature at 30 °C, with the two other experiments at room temperature, thus obtaining exceptional coffees through wet processes. These studies allowed observing that fermentation performed using inoculated microorganisms at room temperature can enhance the sensory quality of coffee.

Illy and Viani (2005) state that soluble carbohydrates, proteins and chlorogenic acids are degraded into melanoidins through Maillard and caramelization reactions in the roasting process. Moreover, other volatile compounds are released. The best relationship is generally found for aroma precursors in coffee during medium–light roasting, which has also been evidenced through this work. During the evaluation of coffee potentials after fermentation, medium-clear roasting reached better sensory results than medium roasting.

The best sensory results (for medium roast) observed were in fermentation times ranging between 25 and 35 h at temperatures above 28 °C. The response surface (Fig. 2.f) shows that shorter fermentation times can achieve excellent sensory evaluation results at higher temperatures. The best result was found in experiment 3 (30 °C and 12 h) as sensory scores were close to 84 points (Table 1). A comparison with the control was not performed since the roasting medium is not considered commercial and was only used to analyze Q-Graders. Based on the results of the two types of roasting methods, it can be seen that medium–light roast showed better sensory results while performing medium roast in lower temperature and longer fermentation time conditions.

Lee et al. (2016) used different roasting levels to evaluate the sensory quality of fermented green dark and light roast coffee and found scores ranging between 0 and 5 in addition to various attributes (sweet, fruity, buttery, caramel, chestnuts, roasted, smoky, spicy sulfurous coffee). Their sensory analysis results revealed a higher quality of coffee prepared through dark roasting than in light roasting, unlike what has been evidenced in this work.

Organic acids

Figure 3 shows that acetic and lactic acids were the primary acids found in the analysis, just as it was found by other authors who had fermented coffee using S. cerevisiae (Evangelista et al. 2014b, 2015; Pereira et al. 2015; Bressani et al. 2018; Neto et al. 2018). Succinic acid was found at lower concentrations and less representativity (except for tests 1 and 2, in which succinic is the main acid found).

Fig. 3.

Final organic acids concentration (Acetic, Propionic, Succinic and Lactic) for each experiment of the Composite Central Rotational Design as a function of temperature and fermentation time

Acetic acid shows concentrations from 0 to 73.33 mg/g, and lower concentration was found in experiment 4 at 30 °C within 48 h, and higher concentration was at a shorter fermentation time (4.42 h). Values between 0 and 73.33 mg/g were found in at least 7 tests at succinic acid concentrations of over 20 mg/g. It also showed adverse concentrations in different experiments which are not correlated with time or temperature. However, greater succinic acid concentration was found in experiment 1 (27.67 mg/g), but it was not so in 8 out of the 11 tests performed. This acid was found in all other experiments in literature, but also at lower concentrations (Evangelista et al. 2014b; Martinez et al. 2017; Bressani et al. 2018).

Lactic acid was the second most commonly found at the end of the process and experiments, except for Experiments 1 to 4. Its concentrations varied from 3.1 mg/g (longer) to 9.4 mg/g (shorter time). As it can be seen in Fig. 3, at the end of the fermentation process, a considerable concentration of acetic and lactic acid was detected in the fermentation process. Nonetheless, other authors have not found such significant acetic acid concentrations at the end of the fermentation process. Bressani et al. (2018) and Evangelista et al. (2014b) found nearly 10 mg/g of acetic acid, and Martinez et al. (2017a) found around 2 mg/g. Concerning acetic acid, Neto et al. (2018) observed that lactic acid increased as the fermentation process progressed. However, other authors found lactic acid concentrations close to 2 mg/g (Evangelista et al. 2015; Martinez et al. 2017).

Propionic acid was found only in the initial four experiments (1 to 4), whose concentrations varied from 3.2 to 4.6 mg/g. Propionic acid may be accountable for the taste and smell of onions (Pimenta 2003). Evangelista et al. (2014b) found concentrations close to 2 mg/g, and lower values were found by Martinez et al. (2017a). However, propionic acid was not found in many other works (Evangelista et al. 2015; Ribeiro et al. 2017; Elhalis et al. 2020b).

Conclusion

A slight variation in the sensory quality of coffee opens up great possibilities for producing higher quality coffee and can therefore add value to the final product at hand. Initiating microorganisms in fermentation can lead to maintaining its features, thus showing that these microorganisms can inhibit undesirable metabolite production. The yeast S. cerevisiae (Lallemand ORO™) was well suited to the cultivar Yellow Catuaí, IAC 62, which achieved good sensory results.

Environmental conditions and fermentation time for coffee standardization must be studied to obtain a high-quality product. These process results could be easily employed in fermentation through light–medium roast at coffee properties, as they are consistent with temperatures found for these producing regions. This fact is already becoming more complex for coffee evaluated at medium roast level, since optimal sensory results cannot be reached due to its typical temperatures, thus requiring investments that can make the process unfeasible.

However, these data can be used in regions having different temperatures where the coffee is typically produced. The concentrations of sugars, glycerol, and acids in fermented coffee showed values close to those found in literature, in addition to decreased pH, thence indicating fermentation completion.

By simulating different farm temperature conditions and at different fermentation times, it was possible to estimate how long it takes to ferment coffee fruits and optimize field conditions, once temperature control at farms is not easily controllable and always requires investments or even incurs high energy costs. Furthermore, it was possible to apply new fermentation methods at farm level without having to make significant investments in infrastructure and bear energy expenditures, in addition to improving beverage quality, thus increasing Brazilian competitiveness in the global coffee market. Unlike sugars found in the sensory analysis, organic acids did not match concentrations found in literature and showed no correlations with time/temperature parameters.

Author contributions

C.J.T.V.: Writing, Research, Methodology. L.S.M.: Writing, Review, Editing. R.C.S.: Validation, Software. M.A.S.: Formal analysis, Supervision, Validation, Review, Editing. M.F.Z.: Formal analysis, Supervision, Validation, Review, Editing. C.Z.G.: Formal analysis, Supervision, Validation, Review, Editing, Software. All authors reviewed the manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

Not applicable for that section.

Declarations

Conflict of interest

Carlos Johnantan Tolentino Vaz declares that he has no conflict of interest. Larissa Soares de Menezes declares that she has no conflict of interest. Ricardo Corrêa de Santana declares that he has no conflict of interest, Michelle Andriati Sentanin declares that she has no conflict of interest. Marta Fernanda Zotarelli declares that she has no conflict of interest and Carla Zanella Guidini declares that she has no conflict of interest.

Ethical standards

The present work is consonance with the ethical regulations of the institution and the country, having been approved by the Ethics Committee in Research with Human Beings under N^o. CAAE 55909622.0.0000.5152.