221 PARTICLE SIZE AND ROASTING ON WATER SORPTION IN CONILOCNo CrrOêaF, FP.E CE.  et al.
DURING STORAGE
Paulo Cesar Corrêa1, Gabriel Henrique Horta de Oliveira2, Ana Paula Lelis Rodrigues de Oliveira3, 
Guillermo Asdrúbal Vargas-Elías4, Fernanda Machado Baptestini5
(Recebido: 12 de agosto de 2015 ; aceito: 03 de novembro de 2015)
ABSTRACT: The aim of this work was to evaluate alterations on the water sorption of coffee due to the effect of roast, 
grind and storage in two temperatures (10 and 30 ºC) during 180 days. Crude grain coffee (Coffea canephora) with average 
initial moisture content of 12.61 % (d.b.) was used. Grain was roasted at two levels: medium light (ML) and moderately 
dark (MD). Afterwards, grain was processed in three different particle sizes: fine (0.59 mm), medium (0.84 mm) and coarse 
(1.19 mm), besides the whole coffee lot. Samples prepared were then stored in two temperatures (10 and 30 ºC). These were 
analyzed during six months, at five distinct times (0, 30, 60, 120 and 180 days) regarding moisture content and water activity. 
Furthermore, mathematical modeling and thermodynamic properties acquisition of the coffee moisture adsorption process 
were accomplished. A split plot design was used, in which plots consisted of storage period and split-plots consisted of a 2 x 
4 x 2 factorial (two roasting degrees, four particle sizes and two storage temperatures), with five repetitions.It was concluded 
thatparticle size did not significantly affectedmoisture content of coffee, independently of roast degree; Sigma-Copace model 
best represented hygroscopic equilibrium for sorption of roasted coffee; with moisture content reduction, an increase of 
differential enthalpy and entropy of sorption and Gibbs free energy occurs.
Index terms: Adsorption isotherms, mathematical modeling, thermodynamic properties, Coffea canephora.
GRANULOMETRIA E TORREFAÇÃO NA SORÇÃO DE ÁGUA EM CAFÉ CONILON 
DURANTE O ARMAZENAMENTO
RESUMO: Objetivou-se, nesse trabalho, avaliar as alterações na sorção de água de café, devido ao efeito da torrefação, 
moagem e armazenamento em duas temperaturas (10 e 30 ºC), durante 180 dias. Café cru (Coffea canephora), com teor de 
água inicial médio de 12,61 % (b.u.) foi utilizado. Os grãos foram torrados em dois níveis: média clara (MC) e moderadamente 
escura (ME). Posteriormente, os grãos foram processados em três diferentes granulometrias: fina (0,59 mm), média (0,84 mm) 
e grossa (1,19 mm), além do lote de café inteiro. As amostras foram armazenadas em duas temperaturas (10 e 30 ºC). Estas 
foram analisadas durante seis meses, em cinco diferentes tempos (0, 30, 60, 120 e 180 dias), acerca do teor de água e atividade 
de água. Posteriormente, a modelagem matemática e a aquisição das propriedades termodinâmicas do processo de adsorção 
foi realizada. Um esquema de parcelas subdivididas foi usado, em que as parcelas consistiram no tempo de armazenamento 
e as subparcelas um fatorial 2 x 4 x 2 (dois níveis de torrefação, quatro níveis de granulometria e duas temperaturas de 
armazenamento), com cinco repetições. Foi concluído que a granulometria não afetou significativamente o teor de água de 
café, independentemente da torra; o modelo de Sigma-Copace é o que melhor representa o equilíbrio higroscópico de sorção 
de café torrado; com a redução do teor de água há um aumento da entalpia e entropia diferenciais de sorção e da energia 
livre de Gibbs.
Termos para indexação: Isotermas de adsorção, modelagem matemática, propriedades termodinâmicas, Coffea canephora.
1 INTRODUCTION the water exchange between the product, which 
Storage of agricultural products is is hygroscopic, and the environment (HENAO; 
historically an important post-harvest procedure QUEIROZ; HAJ-ISA, 2009).
with the aim of achieve food safety and Agglomeration is an important issue 
economic strength for agribusiness. However, for industry, which may lead to complicating 
in order to ensure product quality and higher product handling and transport and, therefore, the 
shelf life, temperature and relative humidity of processing of the final product, directly influencing 
the environment surrounding the product are profit.Ground roasted coffee may absorb water 
primordial parameters to be considered during from the atmosphere because of its low moisture 
storage (CORRÊA et al., 2014). They determine content. Coffee agglomeration occurs when the 
the water activity (aw) of the product and, thus, moisture content ranges from 7 to 8% (OLIVEIRA 
1,5Universidade Federal de Viçosa - Departamento de Engenharia Agrícola - Av. PH Rolfs, s/n, Viçosa-MG - 36.570-000 
copace@ufv.br, fbaptestini@yahoo.com.br
2,3Instituto Federal do Sudeste de Minas Gerais - Campus Manhuaçu, BR 116, km 589,8 -  Distrito Realeza - Manhuaçu-MG 
36.905-000 - gabriel.oliveira@ifsudestemg.edu.br, ana.lelis@ifsudestemg.edu.br
4Universidad de Costa Rica, Centro de Investigaciones en Granos y Semillas (CIGRAS) - Facultad Ciencias Agroalimentarias 
San Pedro de Montes de Oca - San José - Costa Rica-  América Central - gvargase@gmail.com
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233, abr./jun. 2016
Particle size and roasting on water sorption in ... 222
et al., 2014; ROBERTSON, 1993; SILVA et al., Given the importance of knowledge of 
2006). the hygroscopicity of agricultural products, as 
Particle size and roasting also affects the well as the interaction of water with the product, 
hygroscopic capacity of the product and thus the objective of this study was to determine the 
agglomeration index (OLIVEIRA et al., 2014; adsorption isotherms of coffee in different roasting 
ROBERTSON, 1993; SILVA et al., 2006). Thus, and particle size degrees, stored at either 10 or 30 
studies on the influences of particle size and ºC. In addition, to determine the thermodynamic 
roasting level on water absorption and storage properties of water sorption as a function of aw.
are required to understand product-environment 
interactions. 2 MATERIALS AND METHODS
Grinding coffee grain aims to increase the Sample preparation
specific surface area for extraction, increasing 
components transfer related to flavor, affecting Dried raw coffee grain (Coffea canephora 
cup quality (ILLY; VIANI, 1996). Baptestini Pierre ex A. Froehner) were purchased at the local 
(2011) observed that smaller particle-sized coffees market in Zona da Mata, MG.The coffee grain 
had higher a  values after 120 days of storage. In were sorted to remove deteriorated, damaged, and w
addition, the rupture of coffee tissues and cells bored coffee grain to obtain a homogeneous raw 
caused by grinding leads to easier release of material with minimal defects.
the volatile compounds that perfume the drink The mean initial moisture content of the 
(ANDUEZA; DE PEÑA; CID, 2003). Therefore, coffee beans was 12.61% on a dry basis (db), 
coffee drink quality is expected to decrease which was determined gravimetrically using a 
with storage time because of the loss of volatile forced-air oven at 105 ± 1 °C for 24 h.
compounds from ground coffee. Coffee grain were subjected to the roasting 
Roasting also affects the degree of moisture process after sorting. A roaster of direct gas burn 
sorption, coffee being more (dark roasts) or (LPG), with rotary cylinder at 45 rpm, with 
less (light roasts) hygroscopic according to its capacity of 350 g, was used to roast (brand Rod-
degree of roasting. Different researchers stated Bel).The degree of roasting was determined by 
this trend (BICHO et al., 2012; SCHMIDT; a trained professional by monitoring the sample 
MIGLIORANZA; PRUDÊNCIO, 2008). color and comparing it with the Agtron/Specialty 
Sorption isotherms are indispensable to Coffee Association of America (SCAA) standard 
determine and analyze water sorption changes roast number (ASSOCIAÇÃO BRASILEIRA 
during storage. Mathematical models are used DA INDÚSTRIA DE CAFÉ - ABIC, 2015). Two 
to construct isotherms aiding the prediction and roasting degrees were obtained: medium light 
simulation of the performance of materials in (ML) and moderately dark (MD), corresponding 
a certain environment.Different works stated to Agtron numbers of SCAA#65 and SCAA#45, 
sorption isotherms of coffee beans, ground coffee, respectively (Figure 1).
dehulled coffee or roasted coffee (CORRÊA et The mass loss parameter was determined 
al., 2014; FURMANIAK et al., 2009; HENAO; to ensure roasting uniformity, and the coffee 
QUEIROZ; HAJ-ISA, 2009; IACCHERI et al., beans lost, on average, 15.85 and 18.74 gof 
2015; RAMÍREZ-MARTÍNEZ et al., 2013), mass under ML and MD roasting, respectively, 
however, research were not made regarding at a temperature of 285 ºC (VARGAS-ELÍAS, 
different particle sizes and roasting degrees of 2011). Roaster temperature and roasting time 
coffee. tests were performed to evaluate their influence 
Water sorptionand thermodynamic sorption on mass losses. The product was removed from 
properties of a materialmay be monitored by the roaster when reaching the aforementioned 
assessing the a . According to Corrêa, Oliveira degrees of roasting and immediately cooled to w
and Santos (2012), thermodynamic properties of room temperature. 
agricultural products is a key resource that can Following the roasting process, the coffee 
inform assessments of the effect of aw on storage beans were processed in a Mahlkönig mill 
and can inform the understanding of the adsorbed (Germany, model K32 S30LAB) at three different 
water properties and food microstructure. This particle sizes: fine (0.59 mm), medium (0.84 mm), 
information enables the study of the physical and coarse (1.19 mm), and a batch of coffee was 
phenomena that occur on food surfaces. maintained as whole beans(without milling).
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
223 Corrêa, P. C. et al.
FIGURE 1 - Roasting degrees employed: medium light (A) and moderately dark (B) (ABIC, 2015).
The prepared samples were then placed in     (7)
polypropylene bags and refrigerated at two storage 
temperatures (10 and 30 ºC)in BOD chambers. 
The treatments were sampled and analyzedduring 
storageat five period(0, 30, 60, 120, and 180 days). In which: MRE = mean relative error, %; 
Figure 2 summarizes the sample preparation SDE = standard deviation of the estimate, % d.b.; 
for further analyzes described below. Y = observed value; Ŷ = estimated value by the 
The changes in the water activity (aw) of the model; n = number of observed data; and, DF = 
roasted whole bean and ground coffee samples residue degrees of freedom (number of observed 
were assessed using an AquaLab 4TE(Decagon data minus number of model parameters).
Devices, USA) water activity meter with an Thermodynamic parameters: differential 
aw accuracy of ± 0.003. Five replicates were entropy of desorption (∆S), differential enthalpy 
measured.The equilibrium moisture content of a (∆H), Gibbs free energy (∆G) and enthalpy-
coffee sample was defined as the content assessed entropy relationship, were obtained by means of 
after each five storage times (0, 30, 60, 120 and a known methodology [1], which an approximate 
180 days) because the samples were stored in (1-α)100% confidence interval for isokinetic 
permeable plastic bags and the time elapsed was temperature was used. These parameters are 
deemed sufficient to reach equilibrium. expressed respectively by Equations 8-12:
Mathematical models commonly used     (8)
to describe sorption phenomena in agricultural 
products were fitted to the collected hygroscopic 
equilibrium data (Table 1).     (9)
The standard deviation of the estimate 
(SDE), mean relative error (MRE) values and 
residual plot were analyzed to assess the goodness       (10)
of the model fit. The SDE and MRE values of each 
model were calculated using Equations 6 and 7, 
respectively:       (11)
    (6)
      (12)
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233, abr./jun. 2016
Particle size and roasting on water sorption in ... 224
FIGURE 2 - Stages for sample preparation.
TABLE 1 - Mathematical models used to represent sorption isotherms.
Modelname Model
Copace (1)
Modified GAB (2)
Halsey (3)
ModifiedOswin (4)
Sigma-Copace (5)
in which: Ue = equilibrium moisture content, % d.b.;aw = water activity, dimensionless;a, b, c = model parameters 
which depends on the product; and,T = temperature, ºC.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
225 Corrêa, P. C. et al.
In which: ∆H = differential enthalpy of should not be used to predict moisture content 
sorption, kJ kg-1; ∆Hvap = latent heat of vaporization values as the aw tends toward zero. This is a 
of pure water, kJ kg-1; ∆Hst= net isosteric heat of limitation of the model.
sorption, kJ kg-1; ∆S = differential entropy of Figures 3 to 6 present the mean equilibrium 
sorption, kJ kg-1 K-1; and ∆G = Gibbs free energy, moisture contents and standard deviation at each 
kJ kg-1 mol-1; T = isokinetic temperature, K; m = storage period (0, 30, 60, 120, and 180 days), B 
number of data pairs of enthalpy and entropy; t obtained by adsorption,of the roasted conilon 
is the t value at (m-2) degrees of freedom; T  = coffee beans for the evaluated particle sizes, as hm
harmonic mean temperature, K; and, n = number well as their isotherms by Sigma-Copace model.
of temperatures utilized. As can be seen in Figures 3 to 6, water 
The “+” and “-”signs in Equation 8 and activity increased throughout storage, regardless 
others related to thermodynamic properties of roast and particle size degree. This trend is due 
express the direction of heat transfer, which is to adsorption process during storage, in which 
associated with the spontaneity of the process finer products are more hygroscopic, presenting 
studied. Thus, a positive sign corresponds to the higher potential to exchange moisture with the 
adsorption processes in the present study. environment; in this case, gaining moisture.
A split plot design was used, in which plots Storage temperature affects the 
consisted of storage period and split-plots consisted hygroscopicity of roasted coffee samples, regardless of particle size. At a given constant 
of a 2 x 4 x 2 factorial (two roasting degrees, four water activity, the equilibrium moisture content 
particle sizes and two storage temperatures), with decreases with a decrease in storage temperature. 
five repetitions. As the temperature increases, molecular vibrations 
3 RESULTS AND DISCUSSION also increase, thereby increasing the distance between molecules and, consequently, decreasing 
Tables 2 and 3 present the coefficients the attractions between molecules. 
of the models fitted to the observed data for the At higher water activities, the equilibrium 
hygroscopic equilibrium of coffee in the different moisture contents increase sharply, especially 
conditions used. for samples roasted to MD. This trend is related 
MRE values lower than 10% indicate a to roasting degree (more hygroscopic)which at 
good fit for practical purposes (SAMAPUNDO MDcoffee may adsorb a greater amount of water 
et al., 2007) and description of a specific physical from the environment. Similar increases in water 
process is inversely proportional to the SDE adsorption isotherms have been observed by 
value. A maximum SDE of the model estimate of Anese, Manzocco and Nicoli (2006), Baptestini 
0.5 % (d.b.) was considered acceptable. In order to (2011) and Corrêa, Afonso Júnior and Stringheta 
conclude the best model, suitability of the model (2000), for roasted ground coffee stored at 30 ºC 
to respond all variables (storage temperature, for 1 month and for roasted ground coffee stored 
roast degree and particle size degree) was taken in various packages.
into account. Among all the models tested, A good correspondence was observed 
the Sigma-Copace model had the best results between the data estimated using the Sigma-
regarding residual plots, MRE and SDE values Copace model and the experimentally observed 
at all storage temperatures, roast and particle size data. However, it should be noted the Sigma-
degree, followed by Halsey model (Tables 2and3). Copace model is an exponential model. Therefore, the inflection of the isotherm should not be 
This trend can be seen at coffee grain roasted at used to predict moisture content values as the 
moderately dark degree, with fine particle coffee a  tends toward zero, which is a limitation of 
(Table 3).Copace and Modified GAB models wthe model. Future studies are recommended to 
present biased residual plots whilst Modified GAB experimentally assess the equilibrium moisture 
and Modified Oswin presented MRE value higher contents for a  values ranging from 0.5 to 0.9 
than 10% (Table 3). wand analyze the resulting ground roasted coffee 
The Sigma-Copace model has also adsorption isotherms, especially for coffee roasted 
been reported to satisfactorily describe the to MD, because there may be an overestimation of 
hygroscopicity of soluble coffee (CORRÊA; the equilibrium moisture content values at these 
AFONSO JÚNIOR; STRINGHETA, 2000) and aw levels.These levels were not achieved at the 
conilon coffee cherries (CORRÊA et al., 2014). present study probably to the lower storage period 
However, because the Sigma-Copace model is an (six months), requiring higher period to achieve 
exponential model, the inflection of the isotherm higher levels of aw.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233, abr./jun. 2016
Particle size and roasting on water sorption in ... 226
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
TABLE 2 - Parameter estimates of hygroscopic equilibrium models of medium light roasted coffee, whole, fine, medium and coarse particle sizes and its 
respective determination coefficients (R2), standard deviation of the estimate (SDE), mean relative error (MRE) and residual plot (B – biased; R – random)at 
temperatures of 10 and 30 ºC.
WholeCoffee
Fittedparameters MRE SDE R2
Models Residual plot
a b c (%) (% d.b.) (%)
Copace 0.2217 - 0.0026 2.0712 3.87 0.17 92.64 R
Modified GAB 1.5018 58697490 1.2655 4.33 0.18 91.55 R
Halsey 1.3059 - 0.0035 1.3272 3.87 0.17 92.62 R
ModifiedOswin 3.4227 0.0010 2.1947 3.87 0.17 92.79 R
Sigma-copace - 1.1543 - 0.0026 1.4793 3.86 0.16 92.88 R
Fine Coffee
Copace 0.5888 0.0016 2.2271 6.92 0.33 89.20 R
Modified GAB 2.1534 1086 1.2614 7.31 0.35 87.85 R
Halsey 1.8295 0.0019 1.3063 6.58 0.32 89.84 R
ModifiedOswin 5.011 - 0.0057 2.4113 6.12 0.31 90.59 R
Sigma-copace - 1.0349 0.0017 1.6950 7.22 0.34 88.47 R
MediumCoffee
Copace 0.3490 0.0060 2.8808 6.69 0.25 93.62 R
Modified GAB 1.9438 265.6907 1.3547 6.95 0.26 93.02 R
Halsey 1.3829 0.0059 0.9827 6.71 0.25 93.55 R
ModifiedOswin 5.5280 - 0.0295 1.7415 6.77 0.26 93.19 R
Sigma-copace - 1.7023 0.0060 2.1542 6.64 0.25 93.76 R
CoarseCoffee
Copace 0.5589 0.0036 2.0645 4.37 0.18 93.57 R
Modified GAB 2.0411 826.4070 1.1647 4.80 0.20 92.48 R
Halsey 1.8138 0.0049 1.3789 4.48 0.19 93.26 R
ModifiedOswin 4.6036 - 0.0148 2.4715 4.78 0.19 92.59 R
Sigma-copace - 0.9290 0.0036 1.5581 4.27 0.18 93.93 R
227 Corrêa, P. C. et al.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233, abr./jun. 2016
TABLE 3 - Parameter estimates of hygroscopic equilibrium models of moderately dark roasted coffee, whole, fine, medium and coarse particle sizes and 
its respective determination coefficients (R2), standard deviation of the estimate (SDE), mean relative error (MRE) and residual plot (B – biased; R – random) at 
temperatures of 10 and 30 ºC.
WholeCoffee
Fittedparameters MRE SDE R2
Models Residual 
a b c (%) (% d.b.) (%) plot
Copace - 1.2570 - 0.0006 6.3450 9.02 0.27 93.08 R
Modified GAB 0.7449 3071165 2.0371 8.07 0.25 94.17 B
Halsey 0.4668 - 0.0003 0.4340 9.11 0.27 92.97 R
ModifiedOswin 6.1339 0.0049 0.7282 9.43 0.28 92.54 R
Sigma-copace - 5.5530 - 0.0005 4.5920 8.76 0.26 93.38 R
Fine Coffee
Copace - 0.6566 - 0.0058 4.7150 9.71 0.36 92.86 R
Modified GAB 1.0058 10259337 1.8935 10.09 0.41 90.51 R
Halsey 0.6369 - 0.0034 0.5909 9.86 0.36 92.75 R
ModifiedOswin 4.9883 0.0314 1.0159 10.11 0.37 92.52 R
Sigma-copace - 3.9327 - 0.0059 3.4666 9.57 0.36 92.92 R
MediumCoffee
Copace - 0.5796 - 0.0021 4.8426 8.75 0.25 95.09 B
Modified GAB 1.0019 30517074 1.9552 7.73 0.23 96.00 B
Halsey 0.6927 - 0.0012 0.5777 9.01 0.26 94.75 R
ModifiedOswin 5.6483 0.0121 1.0044 9.53 0.28 94.00 R
Sigma-copace - 3.9768 - 0.0022 3.5867 8.40 0.24 95.52 R
CoarseCoffee
Copace - 0.9357 - 0.0011 5.9178 6.30 0.27 94.89 R
Modified GAB 0.8648 4774015 2.0777 5.72 0.28 94.61 R
Halsey 0.5813 - 0.0005 0.4687 6.38 0.27 94.87 R
ModifiedOswin 6.7563 0.0064 0.8029 6.60 0.27 94.78 R
Sigma-copace - 5.0459 - 0.0012 4.3541 6.14 0.27 94.93 R
Particle size and roasting on water sorption in ... 228
FIGURE 3 - Observed and estimated values, obtained by means of Sigma-Copace model, of equilibrium adsorption 
moisture contents of the roasted conilon coffee beans, whole particle size, stored at 10 and 30 ºC.
FIGURE 4 - Observed and estimated values, obtained by means of Sigma-Copace model, of equilibrium adsorption 
moisture contents of the roasted conilon coffee beans, fine particle size, stored at 10 and 30 ºC
FIGURE 5 - Observed and estimated values, obtained by means of Sigma-Copace model, of equilibrium adsorption 
moisture contents of the roasted conilon coffee beans, medium particle size, stored at 10 and 30 ºC.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
229 Corrêa, P. C. et al.
FIGURE 6 -Observed and estimated values, obtained by means of Sigma-Copace model, of equilibrium adsorption 
moisture contents of the roasted conilon coffee beans, coarse particle size, stored at 10 and 30 ºC.
Brunauer (1945) classified sorption in the bond strength between water and the 
isotherms based on the van der Waals forces sorbent surface (coffee), which, according to 
of nonpolar gases adsorbed onto several non- Al-Muhtaseb, McMinn and Magee (2004), at 
porous solid substrates.A comparison analysis the beginning of sorption, has numerous highly 
of the results indicates that the whole bean and active polar sorption sites with high interaction 
ground roasted-coffee adsorption isotherms energies that, with time, are covered by water 
may be classified as type III isotherms. Type III molecules, forming a monomolecular layer. As 
isotherms,known as the Flory–Huggins isotherms, water molecules continue to bind chemically to 
mainly correspond to food consisting of crystalline the highly active sorption sites, sorption begins to 
components, a state characterized by a three- occur in less-active sites (high moisture content) 
dimensional regular arrangement of molecules with lower interaction energies; therefore, the 
based on their orientation (HASSINI et al., 2015). differential enthalpy of adsorption is lower. This 
Observed and estimated values of trend occurred for whole coffee (without grinding) 
differential enthalpies and entropies of adsorption and grinded coffee samples roasted at MD level 
are shown in Figures 7 and 8, respectively. (Figure 7-B). At these samples, adsorption 
Positive values of differential enthalpy occurred at vapor form, because the released 
represent a process with heat absorption energy of this process did not attained the value of 
(endothermic) and negative values of differential latent heat of condensation.
enthalpy, which is the case of the present study, However, coffee roasted to ML presented 
indicates an exothermic and energetically different behavior for grinded coffee (fine, 
favorable transformation process, in other words, medium, coarse), as can be seen in Figure 7-A. 
with heat release (CHU et al., 2004; SHAFAEI; Higher values of ∆H indicates that, during 
MASOUMI; ROSHAN, 2016). Furthermore, adsorption, there are more polar sites or sorption 
Chu et al. (2004) states that negative values of sites at the adsorbent surface of coffee. Viganó et 
differential entropy explains the exothermic nature al. (2012) reported that higher values reflect higher 
of the adsorption process, along with increase of interaction energies and a greater heterogeneity of 
molecules organization. water molecules, suggesting that these products 
It can be noticed that an increase in moisture are more strongly affected by changes in relative 
content, the differential enthalpy of adsorption humidity. 
tends to reach the latent heat value of pure water ∆H values as a function of the particle size 
(2442.45 kJ kg-1). This trend indicates that the of roasted coffeemay be analyzed by evaluating 
number of bonds formed between water molecules roasting. Coffee ground MD samples were 
and the active adsorption sites of the product are statistically similar (p < 0.05) but differed from the 
close to the potential maximum (saturation). pattern for whole roasted coffee. Baptestini (2011) 
This increase is explained by the change also concluded that arabica coffee samples that 
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233, abr./jun. 2016
Particle size and roasting on water sorption in ... 230
were roasted and ground to the fine, medium, and samples, regardless of roasting degree (Figure 9). 
coarse particle sizes were statistically equivalent. Lower absolute ∆G values are related to lower 
However, coffee roasted to ML presented random product hygroscopicity. 
pattern. The medium and coarse particle sizes of Throughout storage, is expected a decrease 
conilon coffee are statistically equal (p < 0.05). in ∆G values, because coffee samples adsorb 
Differential entropy of adsorption had the moisture from the environment, decreasing the 
same tendency as ΔH. A similar result was reported number of adsorption sites and, therefore, reducing 
by Al-Muhtaseb, McMinn and Magee (2004) for the potential for spontaneous water sorption. This 
starch powder and by Baptestini (2011) for ground trend can be noticed at Figure 9.
roasted coffee stored in various packages. Temperature affects ∆G, increasing its 
Coffee samples roasted to ML and grinded values at higher temperatures (Figure 9). This 
exhibited a pattern opposite to the above-described. trend is related to the higher level of excitation of 
According to Rizvi (2005), there are two opposite molecules comprising the product, accelerating 
patterns for the differential entropy of water gas exchange, increasing the rate and degree of 
adsorption in food: a loss of entropy because of spontaneity of water sorption.
the water location or an increase in entropy as a To validate the theory of enthalpy-entropy 
result of the formation of a solution, for example, compensation, the isokinetic temperature should 
during food solubilization and product expansion. be compared with the harmonic mean (Thm) of the 
Therefore, the increase in ∆S indicates the temperature range used to determine the sorption 
formation of water molecule layers that will be isotherms. The calculated Thmwas 292.82 K. The 
subsequently removed from the product surface. results (Table 4) indicate thatthe enthalpy-entropy 
This trend indicates that samples roasted to MD compensation theory may be applied because 
permits the formation of water layers. TB ≠ Thm, except for the ML roasted and finely 
The observed and estimated values of the ground conilon coffee sample.
change in the Gibbs free energy are shown in The enthalpy-entropy compensation theory 
Figure 9. is, most likely, not valid for this sample because 
Regardless of roast and grind degree, Gibbs the ∆S values recorded for this sample (Figure 8A) 
free energy values were negative. This is typical were close to zero.According to Table 4, the water 
of an exergonic reaction or spontaneous process, sorption mechanism of roasted coniloncoffee may 
which does not require the addition of energy from be controlled by either the enthalpy or entropy, 
the environment surrounding the product. Whole with a greater number of samples controlled by 
bean coffee had lower ∆G values than the ground entropy (5) than enthalpy (2).
FIGURE 7 - Observed and estimated values of differential enthalpy of adsorption(∆H) of conilon coffee roasted 
at medium light (A) and moderately dark (B), in different particle sizes.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
Particle size and roasting on water sorption in ... 231
FIGURE 8 - Observed and estimated values of differential entropy of adsorption(∆S) of conilon coffee roasted at 
medium light (A) and moderately dark (B), in different particle sizes.
FIGURE 9 -Observed and estimated values of Gibbs free energy (∆G) of conilon coffee roasted at medium light 
(A) and moderately dark (B), in different particle sizes, stored at 10 and 30 ºC
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
Particle size and roasting on water sorption in ... 232
TABLE 4 -Isokinetic temperature (TB) and Gibbs free energy at isokinetic temperature (∆GB) in order to evaluate 
the enthalpy-entropy compensation theory of Conilon coffee, in two roasting degrees (ML – medium light; MD – 
moderately dark) and in four particle sizes.
ML MD
Part.Size
T  (K) ∆G (kJ kg-1) T  (K) ∆G (kJ kg-1B B B B )
Whole
174.82 ± 13.95 -2469.36 25.82 ± 2.91 -2446.99
Fine 768.90 ± 516.31 -2710.43 183.15 ± 12.52 -2479.41
Medium 434.73 ± 41.46 -2547.20 112.58 ± 10.69 -2467.04
Coarse 490.92± 67.55 -2557.00 60.73 ± 4.81 -2454.45
4 CONCLUSIONS ASSOCIAÇÃO BRASILEIRA DA INDÚSTRIA DE 
Roasting coffee to moderately dark (MD) CAFÉ. Norma de qualidade recomendável e boas 
increased the hygroscopicity of these samples práticas de fabricação de cafés torrados em grão e 
compared to coffee roasted to medium light (ML). cafés torrados e moídos. Disponível em: <http://www.
The Sigma-Copace model best describes abic.com.br/publique/media/NMQ_LEGISLAcaO_
the hygroscopic equilibrium, followed by Halsey NORMAQUALIDADE.pdf>. Acesso em: 13 out. 
model. 2015.
The equilibrium moisture content of roasted 
coniloncoffee decreases with an increase in BAPTESTINI, F. M. Efeito da granulometria e da 
temperature at a given value of aw.
Differential enthalpy (ΔH)and differential embalagem na sorção de água pelo café torrado 
entropy (ΔS) values altered with moisture content e moído. 2011. 95 p. Dissertação (Mestrado em 
variation of coffee. Engenharia Agrícola) - Universidade Federal de 
The enthalpy-entropy compensation theory Viçosa, Viçosa, 2011.
may be satisfactorily applied to water sorption by 
coffee. BICHO, N. C. et al. Use of colour parameters for 
roasted coffee assessment. Ciência e Tecnologia de 
5 ACKNOWLEDGEMENTS Alimentos, Campinas, v. 32, n. 3, p. 436-442, 2011.
The authors express their gratitude to CNPq 
for the financial support to perform this work BRUNAUER, S. The adsorption of gases and vapors. 
(Edital Universal CNPq nº 14/2012, 483622/2012-5). Princeton: Princeton University Press, 1945.
6 REFERENCES CHU, B. S. et al. Separation of vitamin E from palm 
AL-MUHTASEB, A. H.; MCMINN, W. A. M.; fatty acid distillate using silica: I equilibrium of batch 
MAGEE, T. R. A. Water sorption isotherms of starch adsorption. Journal of Food Engineering, Kidlington, 
powders: part 2, thermodynamic characteristics. v. 62, p. 97-103, 2004.
Journal of Food Engineering, Kidlington, v. 62, p. 
135-142, 2004. CORRÊA, P. C.; AFONSO JÚNIOR, P. C.; 
STRINGHETA, P. C. Estudo do fenômeno de adsorção 
ANDUEZA, S.; DE PEÑA, M. P.; CID, C. Chemical de água e seleção de modelos matemáticos para 
and sensorial characteristics of espresso coffee as representar a higroscopicidade do café solúvel. Revista 
affected by grinding and torrefacto roast. Journal of 
Agricultural and Food Chemistry, Easton, v. 51, p. Brasileira de Produtos Agroindustriais, Campina 
7034-7039, 2003. Grande, v. 2, n. 1, p. 19-25, 2000.
ANESE, M.; MANZOCCO, L.; NICOLI, M. C. CORRÊA, P. C. et al. Isotermas de sorção de água de 
Modeling the secondary shelf life of ground roasted frutos de Coffea canephora. Revista Brasileira de 
coffee. Journal of Agricultural and Food Chemistry, Engenharia Agrícola e Ambiental, Campina Grande, 
Easton, v. 54, p. 5571-5576, 2006. v. 18, n. 10, p. 1047-1052, 2014.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016
Particle size and roasting on water sorption in ... 233
CORRÊA, P. C.; OLIVEIRA, G. H. H.; SANTOS, E. RIZVI, S. S. H. Thermodynamic properties of foods in 
S. Thermodynamics properties of agricultural products dehydration. In: RAO, M. A.; RIZVI, S. S. H.; DATTA, 
processes. In: ARANA, I. (Ed.). Physical properties of A. K. (Ed.). Engineering properties of foods. New 
foods: novel measurement techniques and applications. York: CRC, 2005. p. 239-326.
Boca Raton: CRC, 2012. p. 131-141.
ROBERTSON, G. L. Packaging of beverages. In: 
FURMANIAK, S. et al. Searching the most optimal ______. Food packaging: principles and practice. 
model of water sorption on foodstuffs in the whole range New York: M. Dekker, 1993. p. 588-621.
of relative humidity. Food Research International, 
Amsterdam, v. 42, n. 8, p. 1203-1214, 2009. SAMAPUNDO, S. et al. Sorption isotherms and 
isosteric heats of sorption of whole yellow dent corn. 
HASSINI, L. et al. Desorption isotherms and Journal of Food Engineering, Kidlington, v. 79, n. 1, 
thermodynamic properties of prickly pear seeds. p. 168-175, 2007.
Industrial Crops and Products, Amsterdam, v. 67, p. 
457-465, 2015. SCHMIDT, C. A. P.; MIGLIORANZA, E.; 
PRUDÊNCIO, S. H. Interação da torra e moagem do 
HENAO, J. D.; QUEIROZ, M. R.; HAJ-ISA, N. M. café na preferência do consumidor do oeste paranaense. 
A. Umidade de equilíbrio de café cereja descascado Ciência Rural, Santa Maria, v. 38, n. 4, p. 1111-1117, 
baseada em métodos estático e dinâmico. Revista 2008.
Brasileira de SHAFAEI, S. M.; MASOUMI, A. A.; ROSHAN, 
H. Analysis of water absorption of bean and 
Engenharia Agrícola e Ambiental, Campina Grande, chickpeaduring soaking using Peleg model. Journal of 
v. 13, n. 4, p. 470-476, 2009. the Saudi Society of Agricultural Sciences, Riyadh, 
2016. In press.
IACCHERI, E. et al. Different analytical approaches for 
the study of water features in greenand roasted coffee SILVA, F. S. et al. Ângulo de repouso, atrito interno 
beans. Journal of Food Engineering, Amsterdam, v. e efetivo dos grãos de café compergaminho. Revista 
146, p. 28-35, 2015. Brasileira de Produtos Agroindustriais, Campina 
Grande, v. 8, n. 1, p. 17-23, 2006.
ILLY, A.; VIANNI, R. Espresso coffee: the chemistry 
of quality. San Diego: Academic, 1996. 253 p. VARGAS-ELÍAS, G. A. Avaliação das propriedades 
físicas e qualidade do café em diferentes condições 
OLIVEIRA, G. H. H. et al. Caracterização física de de torrefação. 2011. 111 p. Dissertação (Mestrado 
café após torrefação e moagem. Semina. Ciências em Engenharia Agrícola) - Universidade Federal de 
Agrárias, Londrina, v. 35, p. 1813-1828, 2014. Viçosa, Viçosa, 2011.
RAMÍREZ-MARTÍNEZA, A. et al. Water transport VIGANÓ, J. et al. Role of enthalpy and entropy in 
in parchment and endosperm of coffee bean. Journal moisture sorption behavior of pineapple pulp powder 
of Food Engineering, Kidlington, v. 114, n. 3, p. 375- produced by different drying methods. Thermochimica 
383, 2013. Acta, Amsterdam, v. 528, p. 63-71, 2012.
Coffee Science, Lavras, v. 11, n. 2, p. 221 - 233 abr./jun. 2016