You are on page 1of 10

Food Chemistry xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Comparison of methods for determining the deliquescence points


of single crystalline ingredients and blends
Matthew Allan, Lisa J. Mauer ⇑
Department of Food Science, Purdue University, 745 Agriculture Mall Drive, W. Lafayette, IN 47907, USA

a r t i c l e i n f o a b s t r a c t

Article history: Many crystalline food ingredients undergo dissolution at a critical relative humidity known as the deli-
Received 1 December 2014 quescence point (RH0). Blends of crystalline ingredients exhibit deliquescence lowering, resulting in a
Received in revised form 2 May 2015 deliquescence point (RH0mix) lower than the individual ingredient RH0s. To determine the effects of
Accepted 5 May 2015
method type and data collection parameters, these deliquescence measurement methods were com-
Available online xxxx
pared: saturated solution water activity, dynamic vapor sorption profiles (DVS), and dynamic dewpoint
sorption profiles (DDI). Water activity measurements were broadly applicable for measuring deliques-
Keywords:
cence points when 1–4 g sample, 50–125 lL/g water:solid ratio, and 24–48 h equilibration were used.
Deliquescence
Moisture sorption methods
DVS and DDI techniques should be limited to samples containing 1–2 ingredients due to crystal contact
Crystalline solids point effects on RH0mix measurement. Recommended DVS parameters are: 1% RH step, 0.01% equilibrium
Water activity criteria, and 3 h maximum time. DDI profiles were consistent with 50–150 mL/min vapor flow rates.
Ingredient mutarotation, solute interactions in solution, and contact point limitations affect the determi-
nation of deliquescence points.
Ó 2015 Published by Elsevier Ltd.

1. Introduction water activity (Mauer & Taylor, 2010a, 2010b; Zografi, 1988).
Eventually, complete dissolution will occur if the environmental
Deliquescence is the 1st order phase transformation of a crys- RH is maintained above the RH0, with increasing moisture sorption
talline solid to a solution that occurs above a critical relative kinetics beyond the RH0. The unsaturated solution will then be
humidity, known as the RH0. As the environmental RH increases diluted and the vapor pressure of the solution will approach equi-
from 0, but remains below the RH0, water molecules adsorb to librium with the atmosphere (Mauer & Taylor, 2010a, 2010b).
crystal surfaces in a semi-organized fashion or condense in tight Many crystalline food ingredients are deliquescent compounds,
spaces via capillary condensation, but little dissolution occurs and when two or more deliquescent ingredients are in contact with
(Thiel & Madey, 1987). At the RH0 phase boundary, both the crys- one another (as seen in many food systems and dry ingredient
talline solid and saturated solution are thermodynamically stable blends) a decrease in the deliquescence RH is observed. This low-
(Mauer & Taylor, 2010a, 2010b). Water condensation leading to ered deliquescence point is known as the RH0mix (Salameh,
formation of a saturated solution film at the crystal surface occurs Mauer, & Taylor, 2006). The RH0mix is lower than the individual
at RHs above the RH0 because there is a difference between the ingredients’ RH0s and the lowering effect can be justified using
chemical potentials of the solution on the crystal surfaces (ls) the Gibbs–Duhem equation (Eq. (2)). In a ternary system of two
and water vapor (l). The difference in chemical potential creates solutes and water (Eq. (2)), the chemical potential of water (dlw)
a driving force for water to collect on the surface of the crystal as decreases upon the addition of a second ingredient (dl2) to balance
long as the solute is present which can be expressed as: the equation (Wexler & Seinfeld, 1991). Eqs. (1) and (2) can be inte-
  grated to demonstrate that the addition of a second solute will
ps always lower the chemical potential of water (Wexler & Seinfeld,
ls  l ¼ RT  ln ¼ RT  lnðaw Þ ð1Þ
p0 1991).
n1 dl1 þ n2 dl2 þ nw dlw ¼ 0 ð2Þ
where R is the gas law constant, T is temperature, ps is vapor pres-
sure of the solution, p0 is vapor pressure of pure water, and aw is In blends of ingredients, the RH0mix is consistent regardless of
the ratio of ingredients because the solutes will go into solution
⇑ Corresponding author. at the same composition at RHs above the RH0mix, but below the
E-mail address: mauer@purdue.edu (L.J. Mauer). individual ingredient RH0s (Salameh et al., 2006). The ratio at

http://dx.doi.org/10.1016/j.foodchem.2015.05.042
0308-8146/Ó 2015 Published by Elsevier Ltd.

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
2 M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx

which the blend of ingredients initially deliquesces is known as the In previous studies, a wide range of method parameters were
eutonic composition, which can be approximated as the ratio of the used within dynamic vapor sorption (DVS), dynamic dewpoint iso-
compound solubilities, assuming no interactions between ingredi- therm (DDI), and aw measurements to determine the deliques-
ents occur (Salameh & Taylor, 2006). The RH0mix can be estimated cence points of single ingredients and blends. In order to
using equations such as the Ross equation (Eq. (3)) where a° is the standardize an approach for determining deliquescence points,
aw of the saturated solution of an ingredient in the blend (Ross, an understanding of the variables that significantly affect the mea-
1975): sured RH0 or RH0mix values is important. The objective of this study
was to document the effects of experimental variables, encompass-
aw ¼ ða Þ1 ða Þ2 ða Þ3 ð3Þ ing those used in previous publications, on the measurement of
The assumption that solutes do not interact, either in solid or deliquescence points using techniques that are commonly applied
solution states, leads to deviations between measured and pre- to this task (shown in bold in Table 1).
dicted RH0mix values. Deviations can be due to chemical reactions
(i.e. oxidation, hydrolysis, polymerization), enhanced or dimin- 2. Materials and methods
ished solubility resulting from blending the ingredients, mutarota-
tion, or electrostatic interactions (Kwok, Mauer, & Taylor, 2010; Li, 2.1. Materials
Gupta, Eom, Kim, & Ro, 2014; Salameh et al., 2006; Wu, He, Guan, &
Wu, 2010). Temperature also influences the RH0mix, and tempera- All compounds used were reagent grade materials that were
ture effects can be predicted; however, the predictive equations initially stored at 22 °C in a closed desiccator over indicating anhy-
require knowledge of the enthalpy of mixing or dissolution, infor- drous calcium sulfate (0%RH, Drierite™). The ingredients used
mation which may not be readily available for many ingredients were: NaCl (N), 98% D-fructose (F), KCl (K), monosodium glutamate
(Lipasek, Li, Schmidt, Taylor, & Mauer, 2013; Sereno, Hubinger, (M), and sucrose (S). N and M were purchased from Sigma–Aldrich
Comesana, & Correa, 2001). To avoid erroneous estimations of del- Chemical Co. (St. Louis, MO), S and K were purchased from
iquescence points, deliquescence measurements are necessary to Mallinckrodt Baker (Phillipsburg, NJ), and F was purchased from
accurately account for solute–solute or solute–water interactions Acros Organics (Waltham, MA). All ingredients were used as is,
in different ingredient blends. except for S which was milled to reduce the particle size so that
The RH0s of common ingredients have been published it was similar to the other ingredients, with an average size of
(Greenspan, 1977; Mauer & Taylor, 2010a, 2010b; Salameh et al., approximately 100–200 lm. The freshly milled sucrose was
2006; Young, 1967), but less information is available in the litera- annealed at 68%RH for 2–3 days to remove any amorphous regions
ture regarding the deliquescence points of less common ingredi- before being stored at 0%RH. Ingredient blends were prepared at
ents and many ingredient blends. Deliquescence measurements select ingredient ratios (ranging from 1:1, 1:1:1, 8.5:1.5, 8.5:1.5,
are important for determining the deliquescence behaviors of and 2.4:1.1:6.5 for equal ratio binary, equal ratio ternary,
ingredients and ingredient blends in order to design formulations, NS eutonic, KS eutonic, and MKS eutonic blends, respectively)
processing operations, and storage conditions to control the chem- and were geometrically mixed and stored at 0%RH prior to
ical and physical changes imparted by deliquescence (Hiatt, analysis.
Ferruzzi, Taylor, & Mauer, 2008). Multiple methods have been used
to measure the RH0 or RH0mix, and a summary of published meth-
ods is provided in Table 1. 2.2. Water activity of saturated solutions method

Table 1 The aw of prepared samples was measured using an AquaLab


Deliquescence measurement methods reported in literature. 4TE (Decagon Devices, Pullman, WA), in which a chilled mirror
dew point is used to calculate the vapor pressure at the set temper-
Method References
ature. In theory, the aw of the saturated solution will be the same as
Water activity of saturated solution Greenspan (1977), Horwitz and the RH0 assuming no polymorphic changes (e.g. hydrates) or chem-
(chilled mirror dewpoint, Latimer (2006), Marcolli, Luo, and
electrolytic sensor, capacitance Peter (2004), Salameh et al. (2006)
ical modifications (e.g. vitamin degradation) occur upon dissolu-
sensor, or manometric pressure) and Stubberud, Arwidsson, Larsson, tion and incubation. Equilibration of the sample and headspace is
The aw is the predicted RH0 and Graffner (1996) critical when measuring the aw, because the chemical potential
Gravimetric moisture sorption Cohen, Flagan, & Seinfeld, 1987; of the water in the sample is indirectly measured by analyzing
(scale, hanging or electrodynamic Colberg, Krieger, & Peter, 2004; Hiatt
the headspace RH above the sample. Factors such as the headspace
balance) Dynamic vapor sorption et al. (2008), Mauer and Taylor
to observe weight changes in (2010a), Salameh et al. (2006) and volume (volume of air to equilibrate), amount of water used to
respect to RH (aw) Salameh and Taylor (2005) make the saturated solution, and equilibration time could poten-
Dynamic dewpoint sorption profile Ghorab, Marrs, et al. (2014), tially affect the measurement results; therefore, the following
also known as a dynamic Mohamed Ghorab, Toth, Simpson, experimental parameters were controlled: amount of solid (0.5,
dewpoint isotherm Mauer, and Taylor (2013), Schmidt
and Lee (2012) and Yao, Yu, Lee,
1.0, and 2.0 g), ingredient ratios encompassing equal ratio blends
Yuan, and Schmidt (2011) and eutonic compositions (8.5:1.5, 8.5:1.5, and 2.4:1.1:6.5 for NS
Raman spectroscopy Colberg et al. (2004), Salameh et al. eutonic, KS eutonic, and MKS eutonic blends, respectively),
(2006) water–solid ratio (5, 25, 50, 125, 500 lL water/g solid), and equili-
Isopiestic method using salt slurries- Hiatt, Ferruzzi, Taylor, and Mauer
bration time (24, 48, 72 h). Samples were weighed into
gravimetric moisture sorption (2008) and Schmidt and Lee (2012)
and Mauer and Taylor (2010a) high-density polyethylene (HDPE) cups sourced from Decagon
FT-IR Han and Martin (1999) Devices, water was added by pipette, lids were applied to the cups,
Rapid single-particle mass Ge, Wexler, and Johnston, 1998 and the samples were incubated at 25 °C prior to aw measurement.
spectrometry All samples and treatments were prepared and analyzed in tripli-
Electrical conductivity Yang, Pabalan, and Juckett, 2006
RH controlled X-ray diffraction Linnow and Steiger, 2007
cate using the 1 measurement instrumental setting for aw determi-
Radius of crystal Colberg et al. (2004) nation. The AquaLab 4TE devices were verified or calibrated daily
Laser light scattering Braun and Krieger (2001), Colberg using 0.25, 0.50, 0.76, 0.92 aw standards purchased from Decagon
et al. (2004) Devices following the manufacturer’s directions (Decagon
Devices, 2013).

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx 3

2.3. Dynamic vapor sorption (DVS) method et al., 2006). All measurements were performed at 25 °C and the
amount of sample varied depending on the device due to the size
The dynamic vapor sorption (DVS) method is a common instru- of the sample holder: 8–15 mg in the VTI SGA-100, 2 g in the
mental approach for creating moisture sorption isotherms by mon- SPS, and 1 g in the VSA, unless otherwise stated. In the SPS, samples
itoring weight changes in tightly controlled RH and temperature were loaded into 15 mL, 3.9 cm diameter HDPE cups (sourced from
conditions. Moisture sorption isotherms of deliquescent crystalline Decagon Devices) that rested in 5.3 cm diameter aluminum pans
compounds have a sharp increase of moisture uptake above the designed for the SPS, and HDPE cups were also used in the VSA.
deliquescence RH, with increases in moisture sorption kinetics as The HDPE cups were used because ionic ingredients (e.g. NaCl)
the RH increases above the deliquescence point (Mauer & Taylor, could corrode the aluminum pans in the SPS or the steel cup in
2010a, 2010b). The DVS method has multiple parameters that the VSA, potentially causing pinholes or altering the measure-
can be adjusted, including RH step, equilibrium criterion, and max- ments. Preliminary testing of the HDPE cup moisture sorption
imum time. The RH step controls the increments of RH at which using the SPS documented 0.01% weight gain at 95%RH, and taking
the sample is equilibrated. The equilibrium criterion (%EC) is the this into account, the deliquescence measurements were not
maximum percent weight change between measurements in a altered due to the large weight changes that occur upon deliques-
set period of time that is allowed before proceeding to the next cence at lower RHs.
RH step. Upon a phase transformation, the weight change kinetics
are typically greater than the %EC setting and the RH is held for the
maximum time. The maximum time setting dictates the time 2.4. Dynamic dewpoint sorption profile (DDI) method
allowed at each RH step and overrides the %EC in cases where
the sample has not reached equilibrium within the specified The dynamic dewpoint isotherm (DDI) method developed by
amount of time, as is often the case for deliquescent samples held Decagon Devices for their VSA and the AquaSorp instruments is
at RHs near the deliquescence point. In previous studies, the DVS unique compared to the other methods since equilibrium is not
method parameters used to analyze deliquescent ingredients var- attempted. By definition, a moisture sorption isotherm is the equi-
ied from 1–5% RH steps, 0.01–0.001%EC, and 1.5 to 500 h maxi- librium relationship between the moisture content and the equili-
mum time settings (Ghorab, Marrs, Taylor, & Mauer, 2014; Li, brated RH (or aw) of a sample at a set temperature and pressure
Taylor, & Mauer, 2011; Lipasek et al., 2013; Salameh & Taylor, (Rouquerol, Rouquerol, Llewellyn, Maurin, & Sing, 2013; Sing,
2005). It is unclear how these different experimental settings 1985). The DDI is not a true isotherm because the sorption profile
might influence the measurement of the deliquescence points of does not attempt to reach equilibrium during the measurement,
single ingredients and blends, but differences between methods and thus the name has created some confusion as to what the
could contribute to reported discrepancies between studies. The moisture sorption profile actually is (Ghorab, Marrs, et al., 2014).
parameters chosen for this study were: RH step (1% RH, 2% RH, Despite the lack of equilibrium, the DDI technique has been used
and 5% RH step); equilibrium criteria (0.01, 0.1, and 1.0% w/w to profile the water uptake in amorphous foods, as well as for
weight change in 10 min); and maximum time (1.5 and 3 h). determining the RH0 in crystalline ingredients (Ghorab, Marrs,
Three different commercially available DVS instruments were et al., 2014; Ghorab, Toth, Simpson, Mauer, & Taylor, 2014; Yao,
used to compare the effects of different experimental parameters Yu, Lee, Yuan, & Schmidt, 2011).
on deliquescence measurements. The SPS Dynamic Vapor To generate a DDI profile, a sample is exposed to a controlled
Sorption Analyzer (Projekt Messtechnik, Ulm, Germany) and the flow rate of dry or moist air, a capacitance RH sensor continuously
SGA-100 symmetrical gravimetric analyzer (VTI Corp., Hialeah, monitors the aw, and when there is a change of 0.015aw or the scale
FL) are DVS devices in which a dry:wet air static mixer adjusts detects a change of weight, the air flow stops and the aw is mea-
the RH of the air before being pumped into the sample chamber sured using the chilled mirror dewpoint while the weight is
to the programmed RH step and the sample weight change is recorded (Decagon Devices, 2014). The flow rate is a programmed
recorded at each RH step. The Vapor Sorption Analyzer (VSA, parameter with the potential to influence sorption profiles. In the
Decagon Devices, Pullman, WA) uses similar principles, but adjusts VSA, flow rates can be programmed between 10–160 mL/min,
the RH by alternately pumping dry or wet air into the sample while in the AquaSorp the flow rate could range up to
chamber, the RH is continuously monitored by the capacitance 300 mL/min. The flow rates used in this study were 50, 100,
RH sensor, the airflow is periodically stopped, and the aw and 150 mL/min on both VSA and AquaSorp instruments, with single
weight measurements are determined by the chilled dew point tests performed on the AquaSorp at 200 mL/min to observe the
mirror and precision balance, respectively. Unlike the SPS and effects of higher flow rates on deliquescence measurements. The
VTI instruments in which the timing between weight measure- deliquescence points of N, F, S, K, M, KS (1:1 and eutonic ratio of
ments could be precisely programmed and used as part of the equi- 1.5:8.5), and MKS (1:1:1 and eutonic ratio of 2.4:1.1:6.5) were
librium criteria, the time between measurements in the VSA is determined using the DDI technique. For qualitative purposes,
dependent on the amount of time it takes for the aw of the chamber the following DDI sorption profiles were also collected using the
to stabilize, which can vary between sample types and between RH 100 mL/min flow rate: citric acid (anhydrous and monohydrate),
steps, but on average is around 5–6 min. In the VSA, the %EC was glucose (anhydrous and monohydrate), ribose, and NKM (1:1:1).
set as a maximum amount of weight change allowable between Approximately 1 g samples were weighed into HDPE cups for the
two consecutive measurements. Additionally, the VSA software VSA and stainless steel cups for the AquaSorp (which was unable
did not allow a 1.0%EC to be used, thus the VSA %EC criteria were to utilize the lighter HDPE cups). Although some corrosion byprod-
limited to 0.01% and 0.1% weight change between measurements. ucts were noticed in the stainless steel cup, the DDI profiles were
The deliquescence points of single ingredients (N and F), a bin- not significantly altered. The deliquescence point was determined
ary blend (NF at a 1:1 ratio), and a ternary blend (NFK at a 1:1:1 from the DDI profiles as the lowest aw after substantial weight
ratio) were prepared and analyzed in triplicates using the three change (>0.3%) from the previous measurement. Typically, the
DVS instruments and selected parameter treatments. To calculate DDI plot of weight change (y-axis) against aw (x-axis) turns 90°
the RH0 or RH0mix, the slopes of the baseline 3–4 points before at the RH0 and continues to increase in a vertical fashion until sat-
the deliquescence RH and 3 RH points above the deliquescence uration is surpassed (Ghorab, Toth, et al., 2014). If the aw does not
event were used, and the extrapolated intersection was identified change during deliquescence, the RH0 was the fixed aw value that
as the RH0 or RH0mix (Mauer & Taylor, 2010a, 2010b; Salameh was not changing after moisture uptake. For ternary blends and

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
4 M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx

sugars that undergo mutarotation, the DDI profiles became more some samples. As equilibration time increased, the standard devi-
complex. ation in aw measurements decreased. As the amount of water
added to the sample increased, longer equilibration times were
2.5. Statistical analysis needed.

All statistical analyses were conducted using Minitab 16 3.1.3. Ingredient ratios and aw
(Minitab Inc., State College, PN) and 2 or 3-way ANOVA models Differences in equilibration periods and water:solid ratios
depending on the number of factors in the model. The post hoc test needed for consistent aw readings were found between different
used was the Tukey HSD (a = 0.05). ingredient ratios. For example, a higher amount of water
(250 lL/g) in the KS, NS, and MKS blends containing equal ingredi-
ent masses resulted in significantly higher aw measurements than
3. Results and discussion when less water (125 lL/g) was present; however, when these
blends were prepared at their eutonic compositions no differences
3.1. Water activity measurement of saturated solutions in aw were found between these different water:solid ratios when
the samples were equilibrated for at least 48 h (Fig. 1e–j). The sig-
3.1.1. Sample size (mass) and aw nificantly greater aws in the equal w/w blends containing 250 lL
The amount of sample present in the water activity cup signifi- water/g could be attributed to one or more of the ingredients not
cantly affected the aw measurement and therefore deliquescence being saturated. The equilibration time significantly affected the
point determination, although the greatest variation in readings aws of the equal w/w blends of MKS containing 125 lL water/g,
was 0.019 aw in F. In samples for which the mass significantly wherein the aw after 24 h was greater than the aw after 48 and
affected the aw at a set water–solid ratio (62.5 lL/g) and equilibra- 72 h, while no significant differences occurred in the eutonic
tion time (24 h), the smallest sample size, 0.5 g, exhibited signifi- MKS blends with respect to equilibration time. In theory, blending
cantly lower aw readings and greater variability (standard at the eutonic composition will avoid having an excess of one solid
deviation) than larger masses. The small sample size creates a larger while another ingredient is no longer saturated (Kwok et al., 2010).
headspace in which the aw equilibrates and a smaller amount of Saturation and equilibration were easiest to manage when the
solution to contribute water vapor. The ambient RH in the laboratory ingredients were present at the eutonic ratio.
was monitored (using a VWR Traceable Hygrometer /Thermometer)
and was near 50%RH throughout the time course of these experi- 3.1.4. Recommendations for aw measurements to determine RH0 or
ments. Because the ambient RH was below the aw of the samples, RH0mix
if headspace was a factor in the aw measurements, then these condi- In order to maintain saturated solutions and standardize the use
tions would result in a decreased aw reading. A P 1 g sample size is of aw measurements for determining deliquescence points, the fol-
recommended to minimize headspace concerns while maintaining lowing sample preparation parameters are recommended: for sin-
the sample volume to less than half the sample cup as recommended gle ingredients use a 50–125 lL/g water:solid ratio in a 1–2 g
by the manufacturer (Decagon Devices, 2013). sample size, and equilibrate for 24 h prior to aw measurement;
for binary and ternary blends, combine the ingredients at approx-
3.1.2. Water:solid ratios, equilibration time, and aw imately the eutonic composition, use a 50–125 lL water/g in a 1–
The water:solid ratio and equilibration time significantly 2 g sample size, and equilibrate for 24 h. However, if the ratio of
affected the aw readings, although differences in water sensitivity ingredients in a blend is not at the eutonic composition, a mini-
were found between ingredients and blends. The lowest amount mum of 2 water:solid ratios (e.g. 75 and 100 lL/g) and an equili-
of water added, 25 lL/g, was not sufficient to produce stable aw bration time of 48 h is needed to verify that no difference in aw
readings in the ingredients or blends (Fig. 1), possibly because occurred between the different ratios. Previous studies used 60–
not enough water was present to form a saturated solution and/or 150 lL water/g solid to prepare samples (single ingredients and
equilibrate in the headspace of the aw meter, and/or minor water blends) for measurements when investigating deliquescence
loss during equilibration may have occurred. The aw readings for behaviors (Lipasek et al., 2013; Salameh et al., 2006), which fit
samples containing 25 lL water/g were significantly lower than within the recommended parameters of this study; however, there
the aws of samples that contained 125–250 lL water/g. are cases in which greater amounts of water were added
Conversely, too much water can be added which will also influence (Dupas-Langlet, Benali, Pezron, Saleh, & Metlas-Komunjer, 2013).
the aw. For example, F is highly soluble, 3.96–4.32 g/mL at 25 °C It is important to ensure saturation of all ingredients when deter-
(Goldberg & Tewari, 1989), and therefore the maximum amount mining deliquescence points. If stable aw readings are obtained in
of water that can be added to F while maintaining a saturated solu- samples prepared using different water:solid ratios, this could be
tion is 231–252 lL/g. For each deliquescent ingredient and blend, an indication of saturation.
there will be a range of water contents that supports stable aw By following the recommended sample preparation guidelines
readings while maintaining saturation. This appeared to fall within for single ingredients, the aw readings for N and K were within
50–125 lL water/g for the single deliquescent ingredients studied; 0.005 aw of the values reported in the Greenspan reference tables
however, careful consideration of solubility limits for each ingredi- (Greenspan, 1977), and the aws for N, K, and S were with within
ent should be used when designing experiments to ensure the sat- 0.01aw of the values listed in the review article by Mauer and
uration of all ingredients. Taylor (2010a, 2010b). In most published studies reporting the
Equilibration of the saturated solution is necessary for deliques- aw of saturated solutions of blends of deliquescent ingredients,
cence measurement, and the water:solid ratio was found to influ- the ingredient ratios were based on equal masses and the equili-
ence the equilibration time needed. Single ingredients and binary bration period was 624 h. A footnote in one study stated the aws
blends prepared using 50–125 lL water/g reached equilibrium of the ternary and quaternary blends were still decreasing after
(no further significant change in aw) within 24 h (Fig. 1a–d). 24 h (Salameh et al., 2006), a trend which was also observed for
With a greater number of ingredients, the viscosity of the solution blends containing two or more deliquescent ingredients at an
may increase, and longer equilibration times were needed to pro- equal w/w ratio and 125 lL water/g solid (Fig. 1f, h, and j). This
duce stable aw readings, requiring 48 h for many ternary blends. could be attributed to the lack of equilibration, indicating that a
Stirring samples may slightly reduce the equilibration period for longer equilibration time is needed in viscous samples. The

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx 5

Fig. 1. The effects of water–solid ratios, equilibration periods, and blend ratio on the aw of 2 g samples at 25 °C. The samples were: (A) fructose (F), (B) sucrose (S), (C)
monosodium glutamate (M), (D) KCl (K), (E) KS 1:1, (F) KS eutonic, (G) NS 1:1, (H) NS eutonic, (I) MKS 1:1:1, and (J) MKS eutonic. The statistical grouping within each
compound/blend (Tukey HSD a = 0.05) is presented.

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
6 M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx

measured aw values for the ternary blends in the study by Salameh As the RH step increased from 1% to 5%, the RH0 and RH0mix were
et al. (2006) were also higher than predicted using the Ross overestimated for many samples. For example, the extrapolated
Equation which could suggest the solutions were not saturated RH0 for N determined using the SPS device was 75.2%RH at a
with respect to all of the compounds. By using the recommended 1%RH step, 75.9%RH at a 2%RH step, and 77.9%RH at a 5%RH step.
parameters to measure the aw (RH0mix) of the eutonic blends used For NF samples, increasing the RH step from 1% to 2% in the VTI
in this study, the results deviated less than 0.01aw from the Ross (at 0.01%EC and 3 h maximum time) increased the RH0mix by
Equation predictions. 3.7%RH. The 1% RH step resulted in an estimated RH0 that was clos-
est to the data reported by Greenspan (1977). The 1%RH step is
3.2. Dynamic vapor sorption: SPS, VTI, VSA advantageous over larger RH steps because sorption kinetics are
measured at RHs 3–4%RH past the RH0, allowing for extrapolation
DVS techniques have been used to determine the deliquescence to use data points closest to the deliquescence event.
points of single crystalline ingredients and blends (Table 1); how-
ever, the reported deliquescence points between studies can vary 3.2.2. Equilibrium criterion
by several RH units. A range of experimental setting parameters The EC was a significant factor in determining the RH0 and
have been used in these studies, including 1%RH–5%RH steps, RH0mix for all samples analyzed using the SPS, for the RH0mix of
0.001%EC–0.01%EC, and 0.5–500 h maximum time (Ghorab, Toth, all blends analyzed using the VSA, and for only the RH0 of N using
et al., 2014; Hiatt et al., 2008; Li et al., 2011; Salameh & Taylor, the VTI (Table 2). When EC was a significant factor, decreases in the
2005). An objective of this study was to determine the effects of EC increased the accuracy of the measurement (e.g., the data were
these parameters in different moisture sorption instruments on closest to that reported by Greenspan (1977) and others). When
the extrapolated deliquescence points. the EC was set at 0.01%, samples were held at RHs just above the
deliquescence RH for the maximum time setting because the deli-
quescence kinetics at those RHs are slow, but greater than the EC.
3.2.1. RH step
No difference would be expected with a 0.001% EC, a parameter
Controlling the RH step was the largest significant factor in
used in previous publications, because again the maximum time
using DVS methods to determine the RH0 and RH0mix (Table 2).
setting would override the EC. However, as the EC was increased,
Differences in RH steps significantly affected the RH0 and RH0mix
the slow kinetics of deliquescence at the RHs just above the deli-
for most samples analyzed by the SPS method (excluding N), and
quescence point were less than the EC. In such cases, the samples
for some samples in both the VTI and VSA methods. The SPS instru-
were held at RHs just above the deliquescence RH for the minimum
ment had the most sensitive balance which likely contributed to
amount of time and the isotherm curves shifted on the x-axis
the differences in F-values between this DVS method and others.
resulting in an overestimation of the deliquescence point. For
example, increasing the EC from 0.01 to 0.1%EC in the VTI (with
a 1%RH step and 3 h maximum time) resulted in a 4.6%RH increase
Table 2 in the extrapolated RH0mix for NF.
F-values of factors and interactions of measured RH0 and RH0mix and MSError using the
VTI, SPS, and DVS (3-way ANOVA). Significant factors with a P value <0.001 are
indicated by *** or a P value <0.05 are indicated by *. 3.2.3. Maximum time settings
In theory, the maximum time setting should not affect the
Method Factor F-value for sample type
extrapolation of the deliquescence point since deliquescence is a
N F NF NFK 1st order reaction unless the maximum time exceeds the timescale
VTI RH step 42.36⁄⁄⁄ 9.62⁄ 3.05⁄ 2.82 of the deliquescence event. To use the extrapolation procedure for
%EC 15.93⁄⁄⁄ 0.78 1.52 1.25 determining the deliquescence point from a moisture sorption iso-
Max time 0.41 1.63 0.00 0.00
therm, multiple points above the deliquescence RH are desirable
RH step * %EC 4.64⁄ 1.62 0.82 1.08
RH step * time 2.12 1.93 0.04 0.02 prior to dilution beyond saturation of any ingredient present. Our
%EC * time 2.34 0.83 3.83 1.78 study used two maximum time settings (1.5 and 3 h) and found
3 Way 4.40⁄ 0.10 0.44 0.31 no significant differences in deliquescence point determinations
interaction between them for most samples and methods. Exceptions were F
SPS RH step 3.55 165.77⁄⁄⁄ 2646.79⁄⁄⁄ 669.01⁄⁄⁄ and NF samples analyzed by the SPS, which varied by as much as
%EC 16.48⁄⁄⁄ 23.55⁄⁄⁄ 1243.01⁄⁄⁄ 411.62⁄⁄⁄ 0.4%RH and 1.2%RH, respectively, when using a 1%RH step and
Max time 0.17 177.21⁄⁄⁄ 118.63⁄⁄⁄ 1.94
RH step * %EC 9.17⁄⁄⁄ 51.51⁄⁄⁄ 287.04⁄⁄⁄ 26.52⁄⁄⁄
0.01%EC.
RH step * time 0.12 10.81⁄⁄⁄ 15.25⁄⁄⁄ 86.78⁄⁄⁄ Two possible reasons for the significant effect of the maximum
%EC * time 2.18 37.4⁄⁄⁄ 18.54⁄⁄⁄ 74.58⁄⁄⁄ time setting on the SPS instrument for samples containing F are:
3 Way 4.64⁄ 1.91 95.64⁄⁄⁄ 5.33⁄ (1) the balance in the SPS device was the most sensitive in the study
interaction
creating the lowest MSError, and/or (2) F undergoes mutarotation
Size 0.01 3.57 29.49⁄⁄⁄ 0.25
(covariate) upon dissolution (deliquescence) and the ratio of tautomers will
change with time (Flood, Johns, & White, 1996), and this could
VSA-DVS RH step 0.84 30.86⁄⁄⁄ 1.22 0.75
%EC 2.02 1.31 10.64⁄ 10.16⁄ influence the moisture sorption kinetics resulting in time-
Max time 3.22 0.20 1.82 0.37 dependent changes in the moisture sorption profile. The slope of
RH step * %EC 0.01 0.35 0.18 0.09 F moisture uptake kinetics increased by 22.6% from the 1st quartile
RH step * time 0.06 0.61 0.00 2.52 to the 4th quartile of a 3 h equilibration period at 62%RH (just above
%EC * time 0.30 1.15 0.51 0.76
3 Way 2.97 0.16 0.18 3.83
the 61%RH0). There was a strong trend of accelerating moisture
interaction sorption kinetics for F with time (R2 = 0.99), and was not evident
Adjusted MSError for sample type
for N, which does not undergo mutarotation.

N F NF NFK
3.2.4. Recommendations for DVS measurements to determine RH0 or
VTI Error 0.22 1.75 1.55 4.22 RH0mix
SPS Error 0.04 0.06 0.01 0.08
When using a DVS technique to determine a deliquescence
VSA-DVS Error 0.26 0.26 1.75 2.44
point, the following experimental parameters are recommended

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx 7

for single ingredients and ingredient blends: a 1%RH step, 0.01%EC, (Fig. 2a). However, some deliquescent samples had different DDI
and either a 3 or 1.5 h maximum time setting. When following sorption patterns, including hydrate forming ingredients (e.g.,
these recommendations, the measured RH0 of N, F, and the citric acid and glucose), ternary blends, and sugars that undergo
RH0mix of NF blends correlated with previous studies (Greenspan, mutarotation (e.g., fructose and glucose). Hydrate forming ingredi-
1977; Mauer & Taylor, 2010a, 2010b; Salameh et al., 2006). ents will begin to sorb moisture during the anhydrate-hydrate
Significant differences in the RH0 or RH0mix of some samples were transition that might occur before the RH0, which can make inter-
found between the different instruments (Table 3), ranging up to a pretation of the deliquescence event challenging. In sugars that
2.6%RH discrepancy, but no technique consistently produced a undergo mutarotation and ternary blends, a drop in aw is observed
higher or lower deliquescence point than any other across all sam- during deliquescence forming an ‘‘S’’ shaped sorption profile with
ples. Of the three instruments used in the study, the SPS had the no clear deliquescence point. Mutarotation occurs when a crys-
greatest precision, as reflected in the MSError values which ranged talline reducing sugar dissolves and there is a shift in the ratio of
from 0.045–0.079 for the SPS, 0.216–4.217 for the VSA, and tautomers and anomers. As F dissolves, the a-pyranose concentra-
0.256–2.441 for the VTI. The low deviation between replicates on tion decreases while both the a and b-furanose concentrations
the SPS resulted in more factors to be found significant compared increase, with up to 21% furanose structure in solution (higher than
to the other devices. The largest variation between the deliques- many other reducing sugars) (Flood et al., 1996; Fennema,
cence point determinations for all DVS devices was found in the Damodaran, & Parkin, 2008). The shift in the tautomer ratio could
RH0mix of the NFK ternary blend, and of the three methods the vari- interact with water differently, which could cause a shift in aw as
ability was largest in the VTI data, likely due to the smaller sample seen in DDI profiles where F had a large decrease in aw as it deli-
size. As ingredient blends increase in complexity, the accurate quesced. To support this theory, DDI profiles of ribose, glucose,
determination of RH0mix can be hindered by the lack of uniform and S were compared, and the reducing sugars that undergo
contact between all of the ingredients (Salameh et al., 2006). mutarotation exhibited a drop in aw during deliquescence and ‘S’
shape, while the DDI profile of sucrose, a nonreducing sugar that
3.3. Dynamic dewpoint sorption (DDI) profiles does not mutarotate, did not exhibit the same pattern.
Determining the RH0 from an ‘‘S’’ shaped DDI profile is difficult,
DDI profiles of F, K, KS (1:1), MKS (1:1:1), and a eutonic blend of because the deliquescence point could be: (1) the aw at which the
MKS (2.4:1.1:6.5) were collected at flow rates ranging from 50 to initial onset of weight change occurred, or (2) the lowest aw after
200 mL/min are presented in Fig. 2. The RH0 or RH0mix in a DDI pro- weight gain. For the reporting purposes in this study, the lowest
file is identified as the aw at which a sharp uptake in moisture sorp- aw after weight gain was considered to be the RH0 as it was closer
tion occurs. Typically, no extrapolation is necessary for to the aw of the saturated solution. Since the DDI is a scanning
determining deliquescence points in a DDI profile because the aw method where the solid is continuously dissolving and the tau-
at which the sample begins to sorb water is evident, and many tomer ratio of an equilibrated solution is never reached, the aw
ingredients (including N and S) exhibit a vertical line comprised does not drop as low as the aw of an equilibrated saturated solu-
of numerous points at the same deliquescence aw (Ghorab, Toth, tion. The inability to drop to the same aw in the presence of excess
et al., 2014; Schmidt & Lee, 2012); however, linear extrapolation solids without equilibration (evident in Fig. 2C) suggests the DDI is
or the 2nd derivative of the sorption data have also been used to not an appropriate method for measuring the deliquescence point
determine the deliquescence point using a DDI method (Yao of reducing sugars.
et al., 2011). The aw during the deliquescence event is constant if The RH0mix of ternary blends determined using the DDI had
the kinetics of deliquescence does not exceed solute dissolution large variability, as with other measurement methods; however,
and the saturated solution is maintained. Upon complete dissolu- the DDI provided additional insight into what occurred during dis-
tion or when the sorption kinetics exceed the dissolution rate, solution. All ternary blends had a decrease in aw after the initial
the aw begins to increase because the saturated solution concentra- weight change, even when the blends were prepared at the eutonic
tion is no longer present. composition. The drop in aw could be attributed to the presence of
enough condensed water to initiate dissolution of all three ingredi-
3.3.1. Flow rate ents. To further explore the role of crystal contact in ‘‘S’’ shaped
The measured RH0 or RH0mix was not greatly affected by the DDI profiles, a ternary solution of 2 g MSK (1:1:1) with 100 lL of
vapor flow rate in the DDI technique for most ingredients and bin- water was incubated for 72 h, then dried at 11%RH for one month
ary blends (e.g. K, S, KS Eutonic, KS, N) even though the sorption to create an intimately mixed ternary system prior to DDI analysis.
rate was dependent on the water vapor flow rate. However, faster The resulting DDI profile did not have the ‘‘S’’ shaped pattern
vapor flow rates tended to result in higher RH0 or RH0mix values for (Fig. 2b), indicating that intimate and equal contact between crys-
some samples and fewer data points near the deliquescence event. tals results in a DDI profile without the ‘‘S’’ shape. Therefore, the
For example, the RH0 of N measured at 200 mL/min was 0.760 ‘‘S’’ shape in DDI profiles of ternary blends is likely results from
while at speeds of 100 and 50 mL/min the RH0 was measured at uneven crystal contact in the physical mixture.
0.750 and 0.751, respectively. The measured RH0 at 50 and
100 mL/min was closest to the reported values of the aw of a satu- 3.3.3. Recommendations for DDI measurements to determine RH0 or
rated N solution (0.753) (Greenspan, 1977). The inaccuracy of the RH0mix
200 mL/min vapor flow rates was not observed with K or S. The DDI is a relatively new method for measuring sorption pro-
Overall, using vapor flow rates of 50–150 mL/min to generate files. The DDI has been used to measure the RH0 of S and
DDI profiles resulted in consistent deliquescence points that devi- a-D-glucose, using a 300 mL/min flow rate, and the reported RH0
ated less than the confidence of the instrument (0.003aw) (Decagon for S was 0.852 (Schmidt & Lee, 2012) which is comparable to
Devices, 2014), but using faster flow rates (e.g., 200 mL/min) can results from this study and previous documented values
cause a slight elevation in the measured deliquescence point. (Dupas-Langlet et al., 2013; Mauer & Taylor, 2010a, 2010b;
Salameh et al., 2006). The RH0 of a-D-glucose using the same
3.3.2. Discrepancies in the DDI moisture sorption profiles DDI parameters was 0.905 (Scholl, 2014), consistent with the
In most deliquescent ingredient DDI profiles, the aw will stop reported RH0 of glucose monohydrate (Mauer & Taylor, 2010a,
increasing during deliquescence, and many samples exhibited a 2010b; Rüegg & Blanc, 1981; Salameh et al., 2006). Another study
near vertical moisture sorption profile during deliquescence used the DDI with an 80 mL/min flow rate to evaluate the

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
8 M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx

Table 3
Method comparison of deliquescence points determined for single deliquescent ingredients and blends using the recommended method parameters. Groupings within columns
indicate significant differences (n = 3, except KS and MSK using the VTI n = 1, Tukey HSD a = 0.05).

Method Average RH0 or RH0mix for sample type


N F NF KS NFK MSK
VSA-DVS 75.9 ± 0.17 (A) 62.2 ± 0.61 (AB) 41.1 ± 0.77 (B) 71.0 ± 0.07 (A) 40.5 ± 1.06 (A) 65.2 ± 0.83 (A)
VTI-DVS 75.6 ± 0.30 (AB) 61.4 ± 0.53 (AB) 42.5 ± 1.65 (B) 69.6 (C) 43.1 ± 1.95 (A) 64.2 (B)
SPS-DVS 75.6 ± 0.01 (B) 61.9 ± 0.19 (AB) 43.6 ± 0.11(B) 69.9 ± 0.01 (B) 42.7 ± 0.03 (A) 66.7 ± 0.03 (B)
VSA-DDI 75.1 ± 0.10 (B) 62.9 ± 1.25 (A) 46.6 ± 1.36(A) 70.2 ± 0.00 (B) 43.2 ± 2.46 (A) 64.2 ± 0.31 (B)
aw 75.4 ± 0.15 (B) 60.7 ± 0.20 (B) 37.5 ± 0.10 (C) 70.1 ± 0.20 (B) 34.9 ± 0.29 (B) 61.7 ± 0.07 (C)

Fig. 2. DDI profiles of single deliquescent ingredients and blends collected at different flow rates and with different sample sizes and compared to the aw of the corresponding
saturated solutions (dashed lines) at 25 °C. (A) DDI profiles of K and KS with different vapor flow rates. (B) DDI profiles of MSK blends with different ingredient ratios (1:1:1
and Eutonic blend) with different vapor flow rates. (C) DDI profiles of F collected at different flow rates and with different sample sizes (large samples were 4 g while the other
samples were approximately 1 g).

deliquescence point of NaCl in the presence of amorphous ingredi- consistent DDI profiles of deliquescent ingredients, but with sev-
ents (Ghorab, Toth, et al., 2014). A wide range of flow rates have eral caveats. The DDI method is not ideal for reducing sugars and
been used in published DDI sorption profiles of deliquescent ingre- blends with three or more ingredients, due to the ‘‘S’’ shaped sorp-
dients. For the most comparable results to other methods, this tion profiles, complicating deliquescence point determination.
study recommends 50–150 mL/min flow rates to generate

Fig. 3. Moisture sorption profiles of KS and KMS samples generated using (A) SPS-DVS and (B) DDI techniques and comparisons to the aw of the corresponding saturated
solutions (dashed lines, KS = 0.70 and KMS = 0.61) at 25 °C. In the KS blend, the aw is the same as the RH0mix measured by the DDI and SPS-DVS methods because it is not
limited by contact points. In the KMS blends, the aw is lower than the SPS-DVS RH0mix and the DDI, which the DDI has a drop in aw as a binary crystal contact point undergoes
deliquescence and the condensed water brings in the third ingredient, lowering the aw.

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx 9

3.4. DVS, DDI, and aw method comparison for determining RH0 and Cohen, M. D., Flagan, R. C., & Seinfeld, J. H. (1987). Studies of concentrated
electrolyte solutions using the electrodynamic balance. 1. Water activities for
RH0mix
single-electrolyte solutions. Journal of Physical Chemistry, 91(17), 4563–4574.
Colberg, C. A., Krieger, U. K., & Peter, T. (2004). Morphological investigations of
Significant differences were found between the RH0 and RH0mix single levitated H2SO4/NH3/H2O aerosol particles during
determined using different techniques even when the recom- deliquescence/efflorescence experiments. Journal of Physical Chemistry A,
108(14), 2700–2709.
mended method parameters developed in this study were followed Decagon Devices. (2013). Aqualab 4TE-Operator’s Manual. Pullman, WA.
(Table 3). The differences in RH0 between methods ranged from a Decagon Devices. (2014). Vapor Sorption Analyzer-Operator’s Manual. Pullman,
low of 0.8%RH for N to a high of 2.2%RH for F. Differences in WA.
Dupas-Langlet, M., Benali, M., Pezron, I., Saleh, K., & Metlas-Komunjer, L. (2013).
RH0mix ranged from a low of 1.4%RH for KS to a high of 9.1%RH Deliquescence lowering in mixtures of NaCl and sucrose powders elucidated by
for NF samples. When significant differences in deliquescence modeling the water activity of corresponding solutions. Journal of Food
point determinations between methods were present, the aw Engineering, 115(3), 391–397.
Fennema, O. R., Damodaran, S., & Parkin, K. L. (2008). Fennema’s food chemistry. CRC.
method produced the lowest values. Using the aw method, a worst Flood, A. E., Johns, M. R., & White, E. T. (1996). Mutarotation of D-fructose in
case scenario for the moisture sensitivity of deliquescent ingredi- aqueous-ethanolic solutions and its influence on crystallisation. Carbohydrate
ents could be developed; however, there are cases wherein the Research, 288, 45–56.
Ge, Z. Z., Wexler, A. S., & Johnston, M. V. (1998). Deliquescence behavior of
aw method would not reflect the observed deliquescence point of multicomponent aerosols. Journal of Physical Chemistry A, 102(1), 173–180.
the solid. For samples that contain a reducing sugar, the DVS Ghorab, M. K., Marrs, K., Taylor, L. S., & Mauer, L. J. (2014). Water–solid interactions
method results would represent the deliquescence point of the between amorphous maltodextrins and crystalline sodium chloride. Food
Chemistry, 144, 26–35.
solid form of the sugar. Upon dissolution, the sugar undergoes
Ghorab, M. K., Toth, S. J., Simpson, G. J., Mauer, L. J., & Taylor, L. S. (2014). Water–
mutarotation and it appears the solution–solid phase boundary solid interactions in amorphous maltodextri-crystalline sucrose binary
of the crystalline form is at a higher aw than the equilibrated satu- mixtures. Pharmaceutical Development and Technology, 19(2), 247–256.
rated solution. It is recommended to use a DVS method to measure Goldberg, R. N., & Tewari, Y. B. (1989). Thermodynamic and transport properties of
carbohydrates and their monophosphates: The pentoses and hexoses. Journal of
the deliquescence point of any compound or blend that undergoes Physical and Chemical Reference Data, 18(2), 809–880.
chemical changes upon dissolution. Greenspan, L. (1977). Humidity fixed-points of binary saturated aqueous-solutions.
The observed RH0mix of a ternary blend using a DVS or DDI Journal of Research of the National Bureau of Standards Section A-Physics and
Chemistry, 81(1), 89–96.
method is higher than the aw of the saturated solution (Fig. 3), Han, J. H., & Martin, S. T. (1999). Heterogeneous nucleation of the efflorescence of
which is consistent with previous reports of the RH0mix of blends (NH4)(2)SO4 particles internally mixed with Al2O3, TiO2, and ZrO2. Journal of
with 3 or more ingredients (Mauer & Taylor, 2010a, 2010b; Geophysical Research-Atmospheres, 104(D3), 3543–3553.
Hiatt, A. N., Ferruzzi, M. G., Taylor, L. S., & Mauer, L. J. (2008). Impact of
Salameh et al., 2006). A aw measurement may reflect the true del- deliquescence on the chemical stability of vitamins B1, B6, and C in powder
iquescence point for an intimately mixed blend, but in a typical blends. Journal of Agricultural and Food Chemistry, 56(15), 6471–6479.
free flowing dry blend, the ternary RH0mix is limited by contact Horwitz, W., & Latimer Jr, G. W. (2006). Method 978.18 Water Activity of Canned
Vegetables. In AOAC Official Methods of Analysis 18th Ed. Arlington VA: AOAC
points (Fig. 3) and the observed deliquescence point would be International.
higher than the measured aw. In such cases, perhaps both an aw Kwok, K., Mauer, L. J., & Taylor, L. S. (2010). Phase behavior and moisture sorption of
and an additional deliquescence point determination (such as a deliquescent powders. Chemical Engineering Science, 65(21), 5639–5650.
Li, N., Taylor, L. S., & Mauer, L. J. (2011). Degradation kinetics of catechins in green
DVS or DDI profile) would be useful for comparison purposes.
tea powder: Effects of temperature and relative humidity. Journal of Agricultural
and Food Chemistry, 59(11), 6082–6090.
Li, X., Gupta, D., Eom, H. J., Kim, H., & Ro, C. U. (2014). Deliquescence and
4. Conclusions efflorescence behavior of individual NaCl and KCl mixture aerosol particles.
Atmospheric Environment, 82, 36–43.
Linnow, K., & Steiger, M. (2007). Determination of equilibrium humidities using
The deliquescence point of a single ingredient or binary blend
temperature and humidity controlled X-ray diffraction (RHARD). Analytica
can be determined following the experimental conditions recom- Chimica Acta, 583(1), 197–201.
mended in this study using a DVS method, the DDI method, or Lipasek, R. A., Li, N., Schmidt, S. J., Taylor, L. S., & Mauer, L. J. (2013). Effect of
temperature on the deliquescence properties of food ingredients and blends.
by measuring the aw of a saturated solution. When measuring
Journal of Agricultural and Food Chemistry, 61(38), 9241–9250.
the RH0mix of a ternary blend, the aw of a saturated solution will Marcolli, C., Luo, B. P., & Peter, T. (2004). Mixing of the organic aerosol fractions:
provide a lower limit of deliquescence, which will complement Liquids as the thermodynamically stable phases. Journal of Physical Chemistry A,
data from a DVS profile that documents the moisture sensitivity 108(12), 2216–2224.
Mauer, L. J., & Taylor, L. S. (2010a). Deliquescence of pharmaceutical systems.
in the crystalline state. To use the aw method to determine the Pharmaceutical Development and Technology, 15(6), 582–594.
RH0 or RH0mix, use P1 g of solid, blend at the eutonic composition, Mauer, L. J., & Taylor, L. S. (2010b). Water–solids interactions: Deliquescence.
use a water:solid ratio of 50–125 lL/g, and equilibrate for 24 h. For Annual Review Food Science Technology, 1, 41–63.
Ross, K. D. (1975). Estimation of water activity in intermediate moisture foods. Food
DVS methods, use a 1%RH step, 0.01%EC, and 1.5–3 h maximum Technology, 29(3).
time setting. DDI profiles generated using 50–150 mL/min flow Rouquerol, J., Rouquerol, F., Llewellyn, P., Maurin, G., & Sing, K. S. (2013). Adsorption
rates provide useful data for determining the RH0 of single ingredi- by powders and porous solids: Principles. Methodology and Applications:
Academic press.
ents that do not undergo change upon dissolution and the RH0mix Rüegg, M., & Blanc, B. (1981). The water activity of honey and related sugar
of binary blends thereof. As a reference technique, measuring the solutions. Lebensmittel-Wissenschaft & Technologie, 14(1), 1–6.
aw of a saturated solution of a single ingredient (or a blend pre- Salameh, A. K., Mauer, L. J., & Taylor, L. S. (2006). Deliquescence lowering in food
ingredient mixtures. Journal of Food Science, 71(1), E10–E16.
pared near the eutonic composition) is recommended not only
Salameh, A. K., & Taylor, L. S. (2005). Deliquescence in binary mixtures.
for the precision of results, but also because it eliminates the Pharmaceutical Research, 22(2), 318–324.
experimental error associated with extrapolating the RH0 or Salameh, A. K., & Taylor, L. S. (2006). Role of deliquescence lowering in enhancing
chemical reactivity in physical mixtures. The Journal of Physical Chemistry B,
RH0mix from moisture sorption profiles and aw machines are pre-
110(20), 10190–10196.
sent in many food research facilities. Schmidt, S. J., & Lee, J. W. (2012). Comparison between water vapor sorption
isotherms obtained using the new dynamic dewpoint isotherm method and
those obtained using the standard saturated salt slurry method. International
References Journal of Food Properties, 15(1–2), 236–248.
Scholl, S. (2014). Investigation of glucose hydrate formation and loss: Parameters,
Braun, C., & Krieger, U. K. (2001). Two dimensional angular light scattering in mechanisms and physical stability. Urbana, Illinois: University of Illinois.
aqueous NaCl single aerosol particles during deliquescence and efflorescence. Sereno, A., Hubinger, M., Comesana, J., & Correa, A. (2001). Prediction of water
Optics Express, 8(6), 314–321. activity of osmotic solutions. Journal of Food Engineering, 49(2), 103–114.

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042
10 M. Allan, L.J. Mauer / Food Chemistry xxx (2015) xxx–xxx

Sing, K. S. (1985). Reporting physisorption data for gas/solid systems with special Yang, L. T., Pabalan, R. T., & Juckett, M. R. (2006). Deliquescence relative humidity
reference to the determination of surface area and porosity (Recommendations measurements using an electrical conductivity method. Journal of Solution
1984). Pure and Applied Chemistry, 57(4), 603–619. Chemistry, 35(4), 583–604.
Stubberud, L., Arwidsson, H. G., Larsson, A., & Graffner, C. (1996). Water solid Yao, W., Yu, X., Lee, J. W., Yuan, X., & Schmidt, S. J. (2011). Measuring the
interactions II. Effect of moisture sorption and glass transition temperature on deliquescence point of crystalline sucrose as a function of temperature using a
compactibility of microcrystalline cellulose alone or in binary mixtures with new automatic isotherm generator. International Journal of Food Properties,
polyvinyl pyrrolidone, International Journal of Pharmaceutics, 134(1–2), 79–88. 14(4), 882–893.
Thiel, P. A., & Madey, T. E. (1987). The interaction of water with solid surfaces: Young, J. F. (1967). Humidity control in laboratory using salt solutions – A review.
Fundamental aspects. Surface Science Reports, 7(6), 211–385. Journal of Applied Chemistry, 17(9), 241–245.
Wexler, A. S., & Seinfeld, J. H. (1991). 2nd-Generation inorganic aerosol model. Zografi, G. (1988). States of water associated with solids. Drug Development and
Atmospheric Environment Part A-General Topics, 25(12), 2731–2748. Industrial Pharmacy, 14(14), 1905–1926.
Wu, X. Q., He, W., Guan, B. H., & Wu, Z. B. (2010). Solubility of calcium sulfate
dihydrate in ca–mg–k chloride salt solution in the range of (348.15 to 371.15) K.
Journal of Chemical and Engineering Data, 55(6), 2100–2107.

Please cite this article in press as: Allan, M., & Mauer, L. J. Comparison of methods for determining the deliquescence points of single crystalline ingredients
and blends. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.05.042

You might also like