3.1. DRH Prediction of Single Components
The deliquescence phenomenon is directly related to the solubility of the component.
Figure 1a shows the temperature-dependent aqueous solubility of the example component fructose. The example system fructose/water is used to explain the use of the phase diagrams and serves as a proof of principle for the PC-SAFT predictions. In this diagram, pure fructose and pure water are present on the right and left axis of the diagram, respectively. The high solubility of fructose increases with temperature reaching the melting point of fructose at 353.2 K. Although the binary interaction parameter was fitted to the osmotic coefficient (related to the water activity), the PC-SAFT calculated solubility of fructose very well agrees with the experimental solubility data. Due to hydrate formation of fructose [
42], solubility data below 298.15 K is not considered in this work.
The two diagrams in
Figure 1 were calculated using (a) the solid–liquid equilibrium (Equation (1)) for fructose and (b) the solid–liquid–vapor equilibrium (Equations (1) and (4)). The only difference is that
Figure 1a studies the solubility of the crystallizing component in liquid water as a function of temperature, while
Figure 1b studies the influence of relative humidity on the state of the crystal. As shown in
Section 2 (Equations (1) and (4)), the crystal solubility is directly linked to the deliquescence via the solid–liquid–vapor equilibrium.
If a mixture of crystalline fructose and liquid water at a composition left of the solubility line (
Figure 1a) is stored at a fix temperature, e.g., 298.15 K, this results in an unsaturated solution (L). At fructose concentrations right of the solubility line, a saturated liquid phase evolves and coexists with crystalline fructose (fructose crystals + L region in
Figure 1a). The saturated liquid phase always has a composition located on the solubility line (w
fructose = 0.78 at 298.15 K). The water activity of that saturated liquid phase is determined by the RH in the vapor phase according to Equation (4) in thermodynamic equilibrium. This RH is the DRH of fructose at the considered temperature and is shown in
Figure 1b (61.5% RH at 298.15 K).
Figure 1b shows the phase transitions occurring starting from crystalline fructose as a function of RH. When crystalline fructose is exposed to an RH below the DRH, crystalline fructose remains stable. If the RH in the atmosphere reaches DRH, water can be absorbed leading to a saturated fructose solution (having the composition on the solubility line in
Figure 1a at that temperature) next to the fructose crystals. However, the equilibrium amount of absorbed water is not instantaneously reached, but the kinetics of this process depends on the driving force of the water sorption, whereas the latter depends on the ratio between the RH in the atmosphere and the DRH. If the RH is increased above the DRH, an unsaturated solution is formed (dissolution of all fructose crystals) with a water activity being identical to the RH of the equilibrium vapor phase. Only at DRH, crystalline fructose can coexist simultaneously with a vapor phase and a liquid phase in thermodynamic equilibrium. If the RH is decreased below the DRH, the liquid phase evaporates, and fructose crystallizes. The literature values for the DRH of fructose from 293.15 to 313.15 K agree with the PC-SAFT predicted DRH at room temperature. Above 303.15 K, the prediction slightly underestimates the experimental DRH values. This might result from deviations between the prediction and reality, but it might also be a result from kinetic inhibitions during the vapor-sorption measurement. The measured equilibrium water activity of a saturated solution at 298.15 K shows that this effect occurs at least at this temperature and therefore is likely to occur for higher temperatures as well.
Figure 2 shows the water sorption of crystalline fructose as a function of RH. In agreement with the observations from
Figure 1 (pure fructose crystals below DRH, liquid solution above DRH), the predicted water sorption below DRH is negligible, whereas at a DRH of 61.5%, the water sorption is predicted to suddenly increase to w
water = 0.22. This value corresponds to the concentration of the saturated fructose/water solution (
). Above the DRH, the amount of absorbed water further increases, and the additional water from the humid atmosphere further dilutes the fructose/water solution. At 100% RH, fructose is predicted to be infinitely diluted by water from the vapor phase (w
water→1).
The sorption measurements [
14,
45] shown in
Figure 2 indicate that the water sorption below 61% RH is almost zero and thus fructose crystals are stable below this RH. Above 61% RH, a water-sorption increase is observed. The experimental mass fraction of water slightly above DRH steadily increases from 0 to 0.23, while the water sorption is expected to jump abruptly to the water mass fraction of 0.22 according to thermodynamic phase equilibrium conditions. However, due to kinetic inhibitions, this abrupt jump to equilibrium water mass fractions is not observed in the measurements. Therefore, the prediction might indicate that the measured data points between 62 and 64% RH were not in thermodynamic equilibrium yet. Only the measured water sorption at 66% RH (highest investigated RH) seems to have reached thermodynamic equilibrium.
Equilibrium water-activity measurements were performed for high water concentrations and high water activities [
46]. These values are located at the top-right side of the diagram. In this part of the diagram, the water content increases steadily up to w
water = 1 at 100% RH. The equilibrium water-activity measurements of diluted fructose/water solutions are in almost quantitative agreement with the PC-SAFT predictions.
To further validate the prediction capability of PC-SAFT, DRH values of several single components from literature are compared to the predicted values listed in
Table 4. The experimentally determined DRH values are separated in DRH and
values according to their experimental source: DRH values were obtained from gravimetric vapor-sorption measurements, whereas
values were obtained from measuring the relative humidity above a saturated solution (equilibrium water-activity measurement). DRH values determined via gravimetric sorption techniques are usually higher than the
values, since the vapor sorption is kinetically inhibited at RHs slightly above DRH [
2]. This is also observed for the values given in
Table 4. The equilibrium water activity measurements thus allow for the determination of the DRH value without kinetic hindrance. However, they cannot measure the DRH of metastable crystals, as transformation to the thermodynamically stable form usually occurs. This can be better accomplished by gravimetric vapor sorption.
In case of lactose, the measured
is higher than the DRH determined via gravimetric vapor sorption due to hydrate formation at 298.15 K [
44]. The hydrate has a lower aqueous solubility than the anhydrate leading to a higher water activity in the hydrate-saturated liquid phase compared to the anhydrate-saturated liquid phase.
CA forms a monohydrate or anhydrate depending on the RH and temperature. The solubility-phase diagram of CA is shown in the supporting information (
Figure S1). CA anhydrate is not thermodynamically stable in water at 298.15 K, as the CA monohydrate is the stable form above 60.3% RH at 298.15 K [
33]. Thus, CA anhydrate in contact with water transforms to CA monohydrate at 298.15 K, and therefore, the water activity in equilibrium becomes the water activity of a saturated solution of CA monohydrate (78% RH). The measured equilibrium values
of CA anhydrate and CA monohydrate are equal (78% RH), because the CA anhydrate transforms to the CA monohydrate during measurement. Nevertheless, due to nonequilibrium conditions during the measurement, the DRH of CA anhydrate was measured using gravimetric vapor sorption.
Table 4 also compares the PC-SAFT predictions to those applying Raoult’s law. These two methods are based on solubility measurements of pure components in water. The difference between the predicted DRH values form Raoult’s law and PC-SAFT results from the water activity coefficient, which is only accounted for by PC-SAFT.
The average relative deviation ARD of
DRH values was calculated using the experimentally determined value
and the predicted value
in Equation (13).
The ARD values of DRH/ when applying Raoult’s is about 8%, whereas as the ARD of PC-SAFT predicted DRH values compared to the measured ones is only 2%. This indicates the importance of considering the activity coefficients for the prediction of the DRH values. The largest difference between the PC-SAFT predicted DRH value and the literature DRH value occurs for sucrose. For all other components, the deviation is below ±1.6% RH.
3.3. Predicting DRH as a Function of Temperature
The DRH of (a) pure CA, (b) AA/CA, (c) CA/fructose, (d) AA/fructose, (e) fructose/sucrose, and (f) AA/CA/fructose as a function of temperature is shown in
Figure 5.
The phase behavior of CA (anhydrate and hydrate) is shown in
Figure 5a. The equilibrium water activity measurements obtained from Reference [
33] show that CA hydrate is stable below a temperature of 309 K (squares in
Figure 5a). CA anhydrate is the stable form above this temperature (circles in
Figure 5a). The gravimetric vapor-sorption measurements of CA anhydrate obtained from Reference [
44] (stars in
Figure 5a) agree with the equilibrium water activity measurements above the transition temperature. Below the transition temperature, the gravimetric vapor-sorption measurements differ from the equilibrium water activity measurements. This is an indirect proof that hydrate formation indeed occurs in thermodynamic equilibrium, but the anhydrate did not fully transform to the hydrate during the gravimetric vapor-sorption measurement. We conclude that the DRH of the metastable form (CA anhydrate) was observed without knowing that this is the metastable form [
44]. The DRH values obtained from gravimetric vapor-sorption measurements (stars in
Figure 5a) agree with the predicted CA anhydrate deliquescence line above and below the transition temperature.
The predicted deliquescence lines of CA anhydrate and hydrate intersect at the transition temperature (309.45 K). Below the transition temperature, the predicted DRH of CA hydrate is higher than that of the CA anhydrate. Above the transition temperature, the predicted DRH of CA hydrate is lower than that of the CA anhydrate. At RHs above the DRH of the crystal mixture, a liquid forms (region L). It can be seen that the DRH values obtained from equilibrium water activity measurements (circles and squares in
Figure 5a) are in excellent agreement with the predicted DRH values for the respective form.
The CA anhydrate–hydrate solid–solid transformation RH was obtained for different temperatures from Reference [
33]. The measurements indicate that the anhydrate is stable at low RHs and at higher temperatures. The predicted solid–solid transformation line (grey dash-dotted line) separates the region of stable CA anhydrate from the CA hydrate formation region and is in agreement with the experimentally determined solid–solid transition between hydrate and anhydrate. The ARD of the predicted solid–solid transformation RH from the experimental one is 4%. According to the prediction, the hydrate is not thermodynamically stable above 309.45 K.
The RH influence on a CA/AA crystal mixture is studied in
Figure 5b. The gravimetric vapor-sorption measurement was performed with a CA anhydrate/AA crystal mixture at different temperatures [
44]. Although CA hydrate is the stable form below 309.45 K (compare grey dash-dotted line in
Figure 5b), we assume that the transformation of CA anhydrate to the CA hydrate did not occur during the performed measurements (hydrate formation of pure CA did not occur during measurement either, see
Figure 5a). The experimental data in
Figure 5b matches with the predicted deliquescence line of the CA anhydrate/AA crystal mixture. Comparing the predicted CA hydrate deliquescence line in
Figure 5a,b, it can be seen that the AA influence on the DRH of the CA hydrate is negligible.
The temperature-dependent deliquescence of the CA/fructose mixture is shown in
Figure 5c. The experimental data [
44] indicate that deliquescence occurs for RHs lower than 48.9% for temperatures above 293.15 K. This results from the low DRH value of fructose (61% RH at 298.15 K [
14]). In this case, the presence of fructose prevents the formation of CA hydrate above 293.15 K, since deliquescence occurs prior to hydrate formation (hydrate formation above 59% RH at 293.15 K [
33]). The predicted deliquescence line of CA anhydrate/fructose overestimates the measured DRH. Nevertheless, the course of the measured DRH values for the CA/fructose agree with the predicted deliquescence line of CA/fructose. The predicted CA hydrate deliquescence line in
Figure 5c is at significantly lower RH compared to the system without fructose (
Figure 5a). The CA hydrate in presence of fructose deliquesces prior to the CA hydrate in presence of AA (compare
Figure 5b). This results from the low DRH of fructose compared to AA and the resulting lower DRH
mix of the crystal mixture. The temperature of the intersection of CA anhydrate/fructose and CA hydrate/fructose deliquescence line explicitly does not mean that the CA hydrate cannot be present above this temperature. The CA hydrate might still be thermodynamically stable above the solid–solid transformation RH of hydrate formation as soon as fructose is dissolved entirely (compare presence of glucose at RH above DRH
mix in
Figure 3b).
The system AA/fructose (
Figure 5d) looks simpler, as hydrate formation does not occur. Thus, below the DRH, crystals of both AA and fructose are stable and above the deliquescence line a solution forms. The measured DRH value matches the predicted one at 303.15 K. Above this temperature, the predicted DRH is slightly higher than the measured one, and below this temperature, the predicted DRH is slightly lower than the measured one with an overall ARD of 4%. The system fructose/sucrose (
Figure 5e) does not form any hydrates either for the here-investigated conditions. The predicted deliquescence line agrees with the measured DRH values at temperatures above 303.15 K. Below that temperature, the prediction differs by 9% from the measured DRH values.
The temperature dependence of deliquescence for the AA/CA/fructose ternary crystal mixture is finally shown in
Figure 5f. The literature data show that the crystal mixture deliquesce above 48.8% at 293.15 K [
44]. This leads to the fact that, similar to the CA/fructose crystal mixture, hydrate does not occur (hydrate formation of CA hydrate above 59% RH [
33]). The predicted DRH of the ternary mixture is higher than the one obtained from Reference [
44] (overall ARD 11%). Particularly considering that all lines in
Figure 5 were full predictions using solubility data of each individual component in water only, the predictions seem valuable for estimating stability regions. Thus, the RH and temperature limits for stable food or pharmaceutical ingredients (also in crystal mixtures) can be predicted, and time-consuming experiments (vapor-sorption measurements or equilibrium water activity measurements) can be prevented. The predicted phase diagrams can be used to determine which component is responsible for the low DRH
mix in case of crystal mixtures, and the outcome of the deliquescence (complete dissolution of either of the components) can be predicted as a function of the crystal mixture composition.
The temperature-dependent deliquescence behavior of anhydrates as well as of hydrates and even of crystal mixtures thereof can be quantitatively predicted via PC-SAFT. The prediction explicitly differentiates between the solid state (anhydrate vs. hydrate, see Equations (1) and (2)). The DRH can be easily predicted as a function of temperature as well as for multicomponent crystal mixtures (the decrease in DRH of crystal mixtures does not depend on its composition).
Moreover, combining components with a low DRH and hydrate-forming components can prevent hydrate formation, because the crystals deliquesce before hydrate formation can occur.