applied

sciences
Article
Karl Fischer Water Titration—Principal Component
Analysis Approach on Bread Products
Gabriela Popescu 1 , Isidora Radulov 2 , Olimpia A. Iordănescu 3 , Manuela D. Orboi 4 ,
Laura Rădulescu 5 , Mărioara Drugă 6 , Gabriel S. Bujancă 5 , Ioan David 6 , Daniel I. Hădărugă 7 ,
Christine A. Lucan (Banciu) 6 , Nicoleta G. Hădărugă 6, * and Mircea Riviş 8
1 Department of Management and Rural Development, Banat’s University of Agricultural Sciences and
Veterinary Medicine “King Michael I of Romania” from Timişoara, Calea Aradului 119,
300645 Timişoara, Romania; gabrielapopescu@usab-tm.ro
2 Department of Soil Sciences, Banat’s University of Agricultural Sciences and Veterinary Medicine
“King Michael I of Romania” from Timişoara, Calea Aradului 119, 300645 Timişoara, Romania;
isidora_radulov@usab-tm.ro
3 Department of Horticulture, Banat’s University of Agricultural Sciences and Veterinary Medicine
“King Michael I of Romania” from Timişoara, Calea Aradului 119, 300645 Timişoara, Romania;
olimpiaiordanescu@usab-tm.ro
4 Department of Economics and Company Financing, Banat’s University of Agricultural Sciences and
Veterinary Medicine “King Michael I of Romania” from Timişoara, Calea Aradului 119,
300645 Timişoara, Romania; orboi@usab-tm.ro
5 Department of Food Control and Expertise, Banat’s University of Agricultural Sciences and Veterinary
Medicine “King Michael I of Romania” from Timişoara, Calea Aradului 119, 300645 Timişoara, Romania;
laura.radulescu@usab-tm.ro (L.R.); gabrielbujanca@usab-tm.ro (G.S.B.)
6 Department of Food Science, Banat’s University of Agricultural Sciences and Veterinary Medicine
“King Michael I of Romania” from Timişoara, Calea Aradului 119, 300645 Timişoara, Romania;
marioaradruga@usab-tm.ro (M.D.); ioandavid@usab-tm.ro (I.D.); alexalukan@gmail.com (C.A.L.)
7 Department of Applied Chemistry, Organic and Natural Compounds Engineering,
Polytechnic University of Timişoara, Carol Telbisz 6, 300001 Timişoara, Romania; daniel.hadaruga@upt.ro
8 Department of Dental Medicine, “Victor Babeş” University of Medicine and Pharmacy Timişoara,
Eftimie Murgu Sq. 2, 300041 Timişoara, Romania; rivis.mircea@umft.ro
* Correspondence: nicoletahadaruga@usab-tm.ro; Tel.: +40-256-277-423




Received: 18 August 2020; Accepted: 14 September 2020; Published: 18 September 2020


Featured Application: Classification of bread products as a specific type of food product according
to a standard database, using coupled Karl Fischer titration and multivariate analysis

Abstract: (1) Background: The water content and the way of bonding in the food matrices, including
bread, can be easily and simply evaluated by Karl Fischer titration (KFT). The goal was to identify
the main KFT parameters that influence the similarity/dissimilarity of commercial bread products,
using multivariate statistical analysis. (2) Methods: Various commercial bread samples were analyzed
by volumetric KFT and the water content, parameters from titration process and KFT kinetics were
used as input for principal component analysis (PCA). (3) Results: The KFT water content was in the
range of 35.1–44.2% for core samples and 19.4–22.9% for shell samples. The storage and transportation
conditions consistently influence the water content of bread. The type of water molecules can
be evaluated by means of KFT water reaction rates. The mean water reaction rates up to 2 min
are consistently higher for bread core samples, which indicates a high fraction of “surface” water.
PCA reveals the similarity of core samples and various bread types, as well as dissimilarity between
bread parts, mainly based on KFT kinetic parameters. (4) Conclusions: KFT kinetics can be a useful
tool for a rapid and simple differentiations between various types of bread products.

Appl. Sci. 2020, 10, 6518; doi:10.3390/app10186518 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 6518 2 of 15

Keywords: bread; water content; Karl Fischer titration; KFT kinetics; principal component analysis

1. Introduction
Bread is one of the most consumed food products all over the world. There are various types of
bread from the main ingredients and shapes points of view. Wheat bread is largely commercialized,
made even by wheat flour or mixed with other cereal flour types such as rye [1,2]. On the other
hand, rye bread has become more popular, as has whole grain bread [3,4]. Other bread types are
“special” or “functional” food products [2,5,6]. Generally, white bread is made by common wheat
flour (Triticum aestivum L. var. aestivum), water, yeast, shortening agents, sugar and salt. Nutritional
and organoleptic characteristics can be enhanced by addition of oils and fats, oilseeds, almonds or
legumes. On the other hand, gluten-free or low carbohydrate products can be obtained using rice,
corn or cassava flour. Depending on the desired characteristics of the final products, additives such as
enzymes, emulsifiers, gums modified starch or cellulose, as well as protein-rich ingredients (egg and
soy), can be used [1,7–18].
The composition of bread strongly influences its final properties, such as nutritional value, taste
and flavor, rheological properties and shelf-life [4,19]. In this regard, water plays an important role,
in both dough and bread processing steps [20,21]. Water molecules can migrate between starch and
gluten [22]. During these steps, starch gelatinizes and water can partially evaporate from the bread
surface [23–25]. Water molecules migrate from the core to the surface, but the diffusive flow of liquid
water is slower than the evaporation process. On the other hand, the partial water vaporization in
the bread provide a vapor pressure gradient. Water vapor migrates to the core and condensates
(mechanism of “evaporation–condensation”) [26]. Consequently, the outer zone of the bread starts
drying and forms the crust. These variations on water content and distribution influence the final
properties of bread such as softness of the crumb or the crispness of the crust [26]. Among starch
and gluten, which can range 45–58% in bread, proteins (~6%) also influence the water content and
mobility [6,20].
It is a challenge to determine the water content, water molecule types and their mobility in
such complex food matrices. There are many methods of water determinations including physical
or chemical methods [27]. Many factors influence the results, such as limit of detection, type of
water, compounds that bias the measurement method or sample solubility and diffusion of all water
molecules. Physical methods are based on weight loss, such as oven, infrared and microwave drying or
thermogravimetry, as well as spectral and other physical properties of the sample (e.g., conductimetry,
refractive index or capacity) [28–33]. The water determination in bread is generally based on such
methods. Other methods are based on chemical properties of samples. It is the case of Karl Fischer
water titration (KFT, volumetric or coulometric techniques) and other types of titration methods (acyl
halide, anhydride and calcium carbide). Moreover, water distribution between dough components,
water hydration capacity or other properties can be evaluated by farinograph method [22,34].
KFT is a useful method for water determination in various materials, including less soluble
or insoluble samples. The water content can be easily and selectively determined in complex food
matrices such as dairy and meat products, honey, sugars, chocolate, fruits and vegetables, spices and
cereal products [3,35–39]. There are many advantages for water determination by KFT: cover all
water content ranges (ppm to 100%, by selecting coulometric, single- or bi-component volumetric
techniques), determination of both free and “strongly retained” water, a wide range of solvent polarity
and temperature of analysis, which allows protecting less stable/degradable samples, rapid and simple
technique and equipment, increased sensitivity and accuracy and possibility of coupling with other
equipment such as oven or distillation apparatus [38–40]. On the other hand, some compounds can
interfere with the KFT reagents (such as aldehydes and ketones, which react with the alcohol used as
solvent, or compounds that react with iodine—reducing agents) [27,31,33].
Appl. Sci. 2020, 10, 6518 3 of 15

The KFT working parameters can be selected according to the composition and characteristics
of the food sample. If the food product has a high content of oil and/or fat and lower content of
water, the coulometric KFT technique can be used. On the other hand, these types of food products
having higher content of water, such as in the case of butter, require volumetric KFT working with
more hydrophobic solvent mixtures (e.g., methanol–decanol). The electrode coating for lecithin-based
samples can be prevented by adding chloroform and formamide. Soybean, rapeseed, sunflower,
sesame and pumpkin seed oils have been analyzed by coulometric KFT. The water content was in
the range of 0.098–0.689% by automated coulometric KFT and 0.079–0.684% for manual KFT [38].
On the contrary, butter oil had 0.061–0.108% and 0.079–0.099% water by volumetric KFT working in
methanol/chloroform and methanol/hexanol solvent mixtures, respectively [41].
Cereals, cereal-based products, sugars and dairy food ingredients and products have been widely
studied from the KFT water content point of view. They generally have higher water content. On the
other hand, the highly hydrophilic components (carbohydrates such as starch, mono- and disaccharides
such as sucrose and lactose and proteins such as gluten) strongly interact with water molecules.
Such components are less soluble in the KFT working medium and the extraction of water from the
sample particles are enhanced by increasing the temperature analysis, using the homogenization or
milling, and increasing the solvent polarity by adding polar solvents (e.g., formamide). Water content of
wheat and rye flour mixtures have been determined using volumetric KFT at elevated temperature [3,12],
while for sorghum the KFT water content at 65 ◦ C slowly decreased from 12.9% to 12.25% by increasing
the grinding time from 10 to 90 s [39]. Other studies used starch, chick-pea flour, dry alimentary small
paste or toasted ground barley for the determination of water content by volumetric KFT. For a better
accuracy, the determination temperature was increased to 50 ◦ C, where the water content was up to
0.7% higher than for the analysis temperature of 25 ◦ C [42]. Invert sugar and fructose syrups have been
compared for their water content and the corresponding mass loss at elevated temperature. They have
a water content of 26.67% and 29.46% for invert sugar and fructose syrups, respectively. On the other
hand, the mass loss was in the range of 26.4–26.7% for inverted sugar syrup dried at 105 ◦ C, while for
fructose syrup the mass loss had no significant variation by halogen drying at 100–115 ◦ C (mass loss of
29.06–29.46%) [32]. Similar observations were accounted for maltose (5.1–5.64% KFT water content or
halogen drying mass loss), glucose monohydrate (8.34% by KFT and 7.72% mass loss), sucrose (0.024%
by KFT and 0.055–0.062% mass loss) or isomalt- and sorbitol-based candies [30]. All determinations
required elevated temperature (50 ◦ C), formamide addition and internal homogenization. Sucrose I
and II and fructose had the KFT water content of 0.0552–0.0556%, 0.0464–0.0469% and 0.0800–0.0803%,
respectively, while for glucose monohydrate the determined water content was consistently higher
(8.84%). α-Lactose (lactose monohydrate), anhydrous β-lactose and the amorphous form have different
water content, which is difficult to determine. Volumetric KFT performed in fifteen laboratories at 40 ◦ C
provide relatively close results for water content of various commercial lactose ranging 4.48–5.19% [43].
A comprehensive study on the use of vaporization coulometric KFT technique have been performed
by Kestens and co-workers [44]. The robustness and good repeatability of this coupled technique was
demonstrated for various difficult food matrices, including toasted bread.
Dairy products and protein-containing food products have high content of water. Generally,
proteins are more hydrophilic and formamide can be added in order to enhance the water extraction
during volumetric KFT analysis [27]. Moreover, homogenization and enhancing analysis temperature
can be applied (for example, for meat products) [35,45,46]. On the other hand, high fat dairy products
need the addition of more hydrophobic solvents to the KFT working medium (such as chloroform or
decanol). Isengard and collaborators [29] also used 1-propanol, 1-butanol or tert-butyl-methyl-ether in
order to enhance the water extraction in the volumetric KFT analysis at 50 ◦ C for water determination
in dried milk products. When the gas extraction was coupled with an automated KFT technique,
using methanol–octanol solvent mixture at 40 ◦ C, valuable results for the water content of various
commercial butter samples have been obtained [40]. Products having high content of fats and sugars
need homogenization and cooling, as well as hydrophilic or more hydrophobic solvent mixture,
Appl. Sci. 2020, 10, 6518 4 of 15

depending of the characteristic of the main components—sugars and fats, respectively. Instant powders
such as coffee substitutes, coffee and cocoa-based instant powders have been analyzed by volumetric
KFT by addition of formamide [47].
The goal of the present study was to identify the main KFT parameters that influence
the similarity/dissimilarity between commercial bread products, using multivariate statistical
analysis—principal component analysis (PCA)—as a useful combined tool for a rapid and simple
differentiations between various types of bread products.

2. Materials and Methods

2.1. Materials
Bread samples were obtained from the local market (the flour used for obtaining the bread products
by various local producers was from the same manufacturer, Timişoara, Timiş county, Romania).
Three major groups of bread samples were used for KFT and multivariate PCA analyses: (1) White
wheat breads consisted of five sample types, which varied only in the shape and presence or absence
of different seeds on the top of the bread (these extra-materials were not included in samples for
analysis): “HC1”, white wheat bread, non-packed (core); “HC2”, white wheat bread, sliced and packed
in plastic bag (core); “HC3”, white wheat bread roll, non-packed (core); “HC4”, white wheat bread
roll, topped with various seeds, non-packed (core); and “HC5”, homemade white wheat bread (core).
A Moulinex OW 3022 Home Bread, having kneading (5, 20, 0.2 and 0.25 min, alternately with rests
or growing), rest (5 min after the first kneading), growing (39, 25.5 and 49.45 min after the second to
fourth kneading), baking (48 min, followed by 2.51 min), and maintaining at warm condition (60 min)
cycles, was used. The homemade white wheat bread was obtained from 455 g of white wheat flour
(from the same manufacturer as above), 250 mL of water, 30 g of sugar, 8 g of salt, 26 g of milk powder,
25 g of sunflower oil and 9 g of yeast. (2) Whole meal or Graham breads included codes “MC” and
“GC” (cores). (3) Brown wheat breads included “BC1” (non-packed, from the market) and “BC2” (from
the manufacturer). For comparison of KFT kinetic data, the shell (up to 1.5 mm from the outer surface
of the bread) and pre-dried bread samples were included in the study. The shell samples were coded
as “MS” or “BS” (whole meal and brown wheat bread shell samples, respectively), while the pre-dried
samples were coded as “Cd” and “Sd” for core and shell, respectively. Pre-drying was performed at
37 ◦ C in the oven with forced convection for 30 min, in order to evaporate the “surface” water and to
compare the KFT kinetic parameters. All bread samples were kept in the dark at room temperature,
in sealed bags prior analysis. Representative core and shell samples were prepared right before analysis
by cutting in cubic portions with a side of 1–1.5 mm. Weighted samples were immediately sealed in
the KF sample transfer unit until analysis.

2.2. Volumetric Karl Fischer Water Titration (KFT)

The water content of bread samples as well as the KFT kinetics were obtained using the volumetric
bi-component KFT method. A Karl Fischer 701 Titrando apparatus equipped with a 10-mL dosing
system and a 703 Ti Stand mixing system were used (Metrohm AG, Herisau, Switzerland). Component
1 was Titrant 5 apura® and Component 2 Solvent apura® (Merck&Co., Inc., Darmstadt, Germany).
The titer of 4.4286 mg/mL for Component 1 was obtained using Water standard 1% apura® , standard for
volumetric Karl Fisher titration (Merck&Co., Inc.). The sample amount was in the range of 0.057–0.419 g
(Table 1). The following KFT parameters were used: I(pol) 50 µA, end point 250 mV, maximum rate
5 mL/min, stop criterion of drift, stop drift 15 µL/min and temperature of 25 ± 1 ◦ C. Measurements
were done in triplicate (unless otherwise stated).
The KFT kinetics was performed for pseudolinear ranges from the titration curves. The mean
water reaction rate depends on the diffusion of water molecules to the sample surface during extraction
and analysis. The water reaction rate can be determined by knowing the variation of the titration
volume of Component 1 (iodine solution with known titer) on time, as well as the reaction volume
Appl. Sci. 2020, 10, 6518 5 of 15

(the volume of the working medium was 60 mL in all cases) (Figure 1). This variation of the iodine
concentration on time for three pseudolinear ranges was obtained from the Volume (normalized to the
sample mass) versus Time curve. This was the mean iodine or water reaction rate on the specified time
Appl. Sci. 2020, 10, x FOR PEER REVIEW 5 of 16
range (expressed as mmol/L/s or mM/s).

2 NH H3C OH SO2 H2O I2

N
methanol sulfur
imidazole water iodine
dioxide

O O
H3C S - + - +
O O H2N 2 I H2N
N N
imidazolium methylsulfate imidazolium iodide

Figure 1. The overall KFT redox reaction (in the first step, methanol, sulfur dioxide and imidazole from
Figure 1. The overall KFT redox reaction (in the first step, methanol, sulfur dioxide and imidazole
Component 2 react and provide imidazolium methyl sulfite; in the second step, the latter reacts with
from Component 2 react and provide imidazolium methyl sulfite; in the second step, the latter reacts
the water from the sample and iodine from Component 1, providing imidazolium methyl sulfate and
with the water from the sample and iodine from Component 1, providing imidazolium methyl sulfate
imidazolium iodide).
and imidazolium iodide).
2.3. Statistics and Principal Component Analysis (PCA)
2.3. Statistics and Principal Component Analysis (PCA)
The KFT water content and values for KFT kinetic parameters were provided as mean (±standard
The KFT
deviation, SD). water
Moreover, content and values
the results from KFT for analysis
KFT kinetic parameters
and kinetics werecore,
for bread provided
pre-dried as bread
mean
(±standard deviation, SD). Moreover, the results from KFT analysis and
core, raw and pre-dried bread shell samples were compared for significant difference using Tukey’s kinetics for bread core, pre-
dried(honestly
HSD bread core, raw anddifference)
significant pre-driedtest bread
from shell
the samples
one-way wereANOVA compared
modulefor significant7.1difference
in Statistica software
using Tukey’s
(StatSoft, HSD OK,
Inc., Tulsa, (honestly
USA). significant
KFT volume/sampledifference) testratio
mass from(V/mthe (mL/g)),
one-waywater ANOVA contentmodule
(W) (%),in
Statistica 7.1 software (StatSoft, Inc., Tulsa, OK, USA). KFT volume/sample
variation of the KFT volume/sample mass on time (∆(V/m)/∆t) (mL/g/s) and the mean KFT reaction mass ratio (V/m (mL/g)),
water
rate (v)content
(mM/s)(W) were(%), variation
used of the KFT
as dependent volume/sample
variables. mass on
The categorical time (Δ(V/m)/Δt)
predictor (factor) was (mL/g/s) and
the bread
the mean
sample codeKFT reaction
(HC (whiterate
wheat (v)bread
(mM/s) were
core), MC used as dependent
(whole meal breadvariables.
core), GC The categorical
(Graham bread predictor
core_, BC
(factor) wheat
(brown was the bread
bread sample
core), code (HC
MS (whole meal(white
breadwheat
shell, bread
BS (browncore),wheat
MC (whole meal bread
bread shell), core), GC
MCd (pre-dried
(Graham
whole mealbread
breadcore_,
core),BCBCd(brown wheatbrown
(pre-dried bread wheat
core), MSbread(whole
core),meal
MSdbread shell, whole
(pre-dried BS (brown
meal wheat
bread
bread shell), MCd (pre-dried whole meal bread core), BCd (pre-dried brown
shell) and BCd (pre-dried brown wheat bread shell)). The parameterization was sigma-restricted, while wheat bread core), MSd
(pre-dried
the confidence whole
limit meal bread shell)
and significance level and
were BCd
set at (pre-dried
0.95 and 0.05, brown wheat PCA
respectively. bread is a shell)).
multivariateThe
parameterization
statistical was sigma-restricted,
tool providing information onwhile the confidence limitbetween
similarity/dissimilarity and significance level were
bread samples, as wellset as
at
0.95 and 0.05, respectively. PCA is a multivariate statistical tool
the influence of independent variables for grouping. Both titration and kinetic parameters from theproviding information on
similarity/dissimilarity
KFT analysis were usedbetween for PCA. bread
This samples,
technique asapproximates
well as the influence
the dataofmatrix
independent variables
as a product for
of two
grouping.matrices
reduced Both titration andonly
that retain kinetic
the parameters from thePC
useful information. KFT1 analysis
(Factor 1, were
Principalused for
ComponentPCA. This
1) is
technique
that directionapproximates the dataspace
from the property matrix as a product
characterized by of two reduced
a maximum matrices
variance that
of the retain
data. PC2only the
(Factor
useful
2, information.
Principal Component PC1 (Factor
2) also has1, Principal
a maximum Component
variance1)ofisthe that direction
data from the
but restricted byproperty space
orthogonality
characterized
to by a maximum
the first component. Other variance
PCs can be of obtained
the data. inPCthe
2 (Factor
same 2, Principal
manner, butComponent
only few PCs 2) will
alsoretain
has a
maximum
the variance of from
main information the data but restricted
all KFT data. Theby orthogonality
scores to the will
matrix vectors first provide
component. Other PCs
information can
about
be obtained in the same manner,
similarity/dissimilarity but only(grouping
of the samples few PCs will retain
of the the main
cases). Theinformation
loadings matrix from all KFT data.
vectors will
The scores
provide matrix vectors will provide
the similarity/dissimilarity information
of properties, as wellabout
as thesimilarity/dissimilarity
influence of the properties of to
thethesamples
model.
(grouping
PC&CA of thefrom
module cases). The loadings
Statistica 7.1 package matrix vectors
(StatSoft), withwill provide
centered datatheandsimilarity/dissimilarity
cross-validation method, of
properties,
were used for as well
PCAas the influence
analysis of the KFTof the properties to the model. PC&CA module from Statistica 7.1
data.
package (StatSoft), with centered data and cross-validation method, were used for PCA analysis of
3.
theResults
KFT data.

3.1. Karl Fischer Water Titration and Kinetics for Bread Samples
3. Results
TheFischer
3.1. Karl water content of bread
Water Titration core
and samples
Kinetics variesSamples
for Bread in the range of 39–44.2% (Table 1). The highest
water content of white wheat bread samples was obtained for commercial non-packed samples (43.1%),
The awater
but with highercontent of bread
standard core (probably
deviation samples varies
due toina the rangeuniformity
reduced of 39–44.2%
of (Table 1). The
samples). Thishighest
aspect
water content of white wheat bread samples was obtained for commercial non-packed
was observed also for white bread roll samples, which were non-packed commercial products (39–39.4% samples
(43.1%), but with a higher standard deviation (probably due to a reduced uniformity of samples).
This aspect was observed also for white bread roll samples, which were non-packed commercial
products (39–39.4% water content, with relatively high SD). The lowest SD in the case of white bread
was obtained for the homemade samples; the KFT water content in this case was 41% (Table 1). The
Graham samples have the lowest water content (36.5%), while the whole meal and brown bread water
Appl. Sci. 2020, 10, 6518 6 of 15

water content, with relatively high SD). The lowest SD in the case of white bread was obtained for
the homemade samples; the KFT water content in this case was 41% (Table 1). The Graham samples
have the lowest water content (36.5%), while the whole meal and brown bread water content depends
on the sample source: the commercial whole meal and brown bread samples have higher water
content (~44%), in comparison with the samples obtained directly from the manufacturer (35.1%);
this difference can be explained by the possible use of other recipes and/or processing technique for the
product from the market (i.e., colorant additives are indicated on the label).
Completely different results were obtained after pre-drying of the white bread core samples
(Table 1). The water content after pre-drying is approximately the same for whole meal bread and brown
bread core samples (~31%), in comparison with the total water content of 44% without pre-drying.
A similar approach was applied for some shell samples, without or with pre-drying. KFT water
content of these samples was obtained only for comparison of the KFT kinetics and evaluating the
corresponding parameters for similarity/dissimilarity of samples by PCA. The water content of the
bread shell samples decreased to half in comparison with the core samples (22.9% and 19.4% for whole
meal and brown bread shell samples, respectively, in comparison with ~44% for the corresponding
core samples). On the other hand, the pre-drying of shell samples consistently decreased the water
content (from 22.9% to 1.1% for whole meal bread shell and from 19.4% to 4.7% for brown bread shell
samples, Table 1).

Table 1. KFT results for bread core samples, pre-dried bread core samples, as well as raw and pre-dried
bread shell samples. Values are expressed as mean (±standard deviation, SD) of triplicate analysis
(excepting * for duplicate or single analysis).

Sample Mass KFT Volume KFT Volume/Sample Mass Ratio Water Content
No. Code
(g) (mL) (mL/g) 1 (%) 1
1 HC1 * 0.0768 (±0.0149) 3.223 (±0.498) 42.16 (±1.71) a 43.12 (±6.66) a
2 HC2 0.0572 (±0.0070) 2.554 (±0.291) 44.66 (±0.84) a 39.56 (±0.74) a
3 HC3 0.0656 (±0.0178) 2.902 (±0.860) 43.16 (±0.18) a 39.04 (±1.00) a
4 HC4 0.4192 (±0.5131) 3.002 (±0.799) 44.05 (±1.85) a 39.37 (±1.47) a
5 HC5 * 0.0690 (±0.0141) 2.894 (±0.566) 41.69 (±0.71) a 41.06 (±0.34) a
6 GC 0.1288 (±0.0600) 5.328 (±2.491) 40.57 (±1.49) a 36.50 (±1.32) b
7 MC 0.0850 (±0.0032) 3.884 (±0.187) 46.45 (±0.97) a 44.23 (±1.14) a
8 BC1 0.0837 (±0.0018) 3.777 (±0.135) 44.06 (±0.59) a 44.11 (±0.61) a
9 BC2 0.0631 (±0.0376) 2.297 (±1.370) 37.86 (±1.70) ab 35.15 (±0.35) b
10 MCd * 0.0851 2.731 32.43 b 31.06 c
11 BCd 0.0851 2.752 32.74 b 31.30 c
12 MS 0.0909 (±0.0014) 2.154 (±0.081) 22.95 (±1.01) c 22.94 (±1.05) d
13 MSd * 0.0915 0.099 23.61 c 1.05 e
14 BS 0.0811 (±0.0084) 1.656 (±0.572) 17.56 (±4.04) d 19.44 (±4.72) d
15 BSd * 0.0926 0.452 5.35 e 4.72 e
1For a given parameter (KFT volume/sample mass ratio (V/m) and water content (W)), values with different letters
are significantly different, according to Tukey’s HSD (honestly significant difference) test (p < 0.05). All p-level
values are presented in Supplementary Materials (Tables S2 and S3).

The aspect of Volume/sample mass (V/m) versus Time titration curves from the KFT analysis are
very different for bread samples (Figures 2 and 3; see also Figures S1 and S2). Only two pseudolinear
ranges were observed for all white bread samples and Graham bread samples, while whole meal
and brown bread samples provide three pseudolinear ranges. The water reaction rates for white
bread core samples, corresponding to “surface” water molecules, are in the range of 2.4–3.2 mM/s,
the highest being determined for packed samples (code “HC2”, Table 2). The Graham bread samples
have similar behavior, but the water reaction rate for the first interval is lower (1.67 mM/s), while that
for the second interval is in the same range (0.06 mM/s). Different results were obtained in the case
of brown bread samples: the samples obtained from the market (containing colorants as additives)
have similar behavior such as white bread samples (code “BC1”, 2.17 mM/s for v1 and 0.097 mM/s for
v2 ). It can be due to different recipes used, in comparison with the similar sample obtained from the
Appl. Sci. 2020, 10, 6518 7 of 15

Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 16

manufacturer. Higher water reaction rates for both first and second interval (3.9–4.2 and 0.2–0.5 mM/s,
respectively)
were obtained were inobtained
the caseinofthe case of
whole whole
meal mealand
bread bread and brown
brown bread samples
bread samples obtained
obtained from fromthe
the manufacturer
manufacturer (code
(code “BC2”).
“BC2”). Even Even the core
the core samples
samples werewere pre-dried
pre-dried in theinoven
the (the
oven“surface”
(the “surface”
water water
was
was partially
partially removed),
removed), thesethese
waterwater reaction
reaction ratesrates
remainremain approximately
approximately the samethe same
(3.8–4(3.8–4
mM/s mM/s
for thefor the
first
first interval and 0.3–0.5 mM/s for the second one). Some KFT analyses were
interval and 0.3–0.5 mM/s for the second one). Some KFT analyses were performed on bread shell performed on bread shell
samples,
samples,butbut the
the results havehigh
results have highSDs SDsininmost
most cases
cases (probably
(probably duedue to the
to the higher
higher hygroscopicity
hygroscopicity of theof
the crust).
crust). TheThe water
water reaction
reaction ratesrates corresponding
corresponding to theto“surface”
the “surface”
waterwater
moleculesmolecules are relatively
are relatively low (0.8– low
(0.8–2.1 mM/s),
2.1 mM/s), while
while those
those corresponding
corresponding to “strongly
to “strongly retained”
retained” water
water molecules
molecules vary
vary in in a widerange
a wide range
(0.32–0.44
(0.32–0.44and
and0.06–0.3
0.06–0.3mM/s
mM/sfor for whole
whole meal and brown brown bread
breadsamples,
samples,respectively, Table2;2;v3vvalues
respectively,Table 3 values
arepresented
are presentedin inTable
Table S1).
S1).

26
50
24

22
40
20

18

30 16

14

V/m (mL/g)
V/m (mL/g)

12
20
10

8
10
6

0 2 V/m (MSa) (mL/g)

V/m (HC2a) (mL/g) V/m (MSb) (mL/g)
V/m (HC2b) (mL/g) 0
V/m (MSc) (mL/g)
V/m (HC2c) (mL/g) -2
-10 -50 0 50 100 150 200 250 300 350 400
-50 0 50 100 150 200 250 300 350
Time (s)
Time (s)

(a) (b)
Figure2.2.The
Figure TheVolume/sample
Volume/samplemass mass (V/m)
(V/m) versus
versus Time
Time titration
titrationcurves
curvesfrom
fromKFT
KFTanalysis
analysisofofthe
theselected
selected
breadsamples
bread samples(triplicates):
(triplicates): (a)
(a) white
white wheat
wheat bread,
bread, sliced
slicedand
andpacked
packedininplastic
plasticbag
bag(core), code
(core), code “HC2”;
“HC2”;
and (b) whole meal wheat bread (shell), code
and (b) whole meal wheat bread (shell), code “MS”. “MS”.

Table
4 2. The mean KFT water reaction rates (based on the 3.0
variation of the normalized KFT volume on
2.5
time,
3 ∆(V/m)/∆t) on the first two time ranges (pseudolinear)
2.0
for bread coreBC
samples,
HC
pre-dried bread
core
2
samples and raw GC and pre-dried bread shell samples Values are expressed as mean (±standard
1.5
GC GC BC
deviation, SD) of triplicate
HC
HC
analysis
HC
(except * for duplicate
1.0 or single analysis).
BC
1 BC
BC HC GC
BC 0.5 HC BC
Factor 3: 15.93%

HC HC HC MC HCHC
Factor 2: 28.24%

HCHC HCHC MC
0 Variation
HC
of the KFT Variation0.0of the KFT Mean KFT
MC HC
HC HC HC Mean KFT
HC BC
HC HC BC
No. Code Volume/Sample Mass on Time
BC Volume/Sample-0.5 Mass on Time
HCReaction Rate
HC Reaction Rate
-1 1 BC
[∆(V/m)1 /∆t1 , mL/g/s] BC [∆V/m)2 /∆t , mL/g/s] 1 v2 (mM/s) 1 v2 (mM/s) 1
-1.02 GC
1-2 HC1 * MC
MC
0.434 (±0.078) a (±0.0027) a
0.0039 -1.5 2.42 (±0.43) a
HC 0.022 (±0.015) a
2 HC2 MC 0.581 (±0.066) a (±0.0008) a
0.0090 -2.0 3.23GC
(±0.37) a 0.050 (±0.004) a
3-3 HC3 0.503 (±0.124) a (±0.0093) a
0.0192 -2.5 2.80 (±0.69) a 0.107 (±0.052) a
4-4 HC4 0.495 (±0.112) a 0.0127 (±0.0067) a 2.75 (±0.62) a 0.071 (±0.037) a
-3.0
5 HC5 * 0.483 (±0.107) a 0.0139 (±0.0015) a 2.68 (±0.59) a 0.077 (±0.008) a
ab -3.5 a ab a
6-5 GC 0.301 (±0.156) 0.0111 (±0.0093)
-5 -4 -3 -2 -11.67 (±0.87)
0 1 2 0.062
3 (±0.052)
4 5 6
-5 -4 -3 -2 -1 0 1 2 3 4 5 6
c 1: 44.90%c
7 MC 0.762 (±0.024)
Factor 1: 44.90% 0.0380 (±0.0056) b 4.24 (±0.13)
Factor 0.211 (±0.031) b
8 BC1 0.389 (±0.004) a 0.0175 (±0.0034) a 2.17 (±0.02) a 0.097 (±0.019) a
9 BC2 (a) (±0.083) c
0.708 0.0848 (±0.0086) c 3.94(b)
(±0.46) c 0.472 (±0.048) c
10 MCd * 0.717 c 0.0557
2.5 d 3.99 c 0.310 d
1.0
11 BCd * 0.682 c 0.0857 cd 44.90% 3.79 c 0.476 cd
12 MS 0.380 (±0.015) a 0.0571 (±0.0013)
2.0
d 2.12 (±0.09) a 0.318 (±0.007) d
13 MSd * 0.136 d 0.0785 c 0.76 d 0.436 c
14 0.5BS * 0.180 dv3 (mM/s) 0.0533 d 1.00 d 0.296 d
a 0.060 a
15 BSd * 0.205 d 1.5
0.0108 28.24% 1.14 d
Factor 2 : 28.24%

1
Eigenvalue

For a given parameter (variation of the KFT volume/sample mass on time (∆(V/m)/∆t) and mean KFT reaction
1.0
rate (v)),
0.0 values with different letters are significantly different, according to Tukey’s 15.93%
HSD (honestly significant
difference) Wtest
(%)
(p < 0.05). All p-level values are presented in the Supplementary Materials (Tables S4 and S5).
V/m (mL/g)
0.5
7.28%
v2 (mM/s) 3.65%
-0.5

0.0

v1 (mM/s)

-1.0 -0.5
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
-1.0 -0.5 0.0 0.5 1.0
Eigenvalue number
Factor 1 : 44.90%
 -10 -50 0 50 100 150 200 250 300 350 400
-50 0 50 100 150 200 250 300 350
Time (s)
Time (s)

(a) (b)
Figure 2. The Volume/sample mass (V/m) versus Time titration curves from KFT analysis of the selected
bread
Appl. Sci. samples
2020, 10, 6518(triplicates): (a) white wheat bread, sliced and packed in plastic bag (core), code “HC2”; 8 of 15
and (b) whole meal wheat bread (shell), code “MS”.

4 3.0

2.5
3
2.0 BC HC
GC
2 1.5
GC GC BC
HC
HC HC 1.0 BC
1 BC
BC
BC HC GC
HC HC HC 0.5 HC BC

Factor 3: 15.93%
Factor 2: 28.24%

HC HCHC MC HCHC
HC MC
0 HC 0.0 MC HC
HC HC BC HC HC HC BC
HC HC BC
BC HC
-0.5
-1 BC
-1.0 GC
MC
MC
-2
-1.5 HC
MC GC
-3 -2.0

-2.5
-4
-3.0

-5 -3.5
-5 -4 -3 -2 -1 0 1 2 3 4 5 6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Factor 1: 44.90% Factor 1: 44.90%

(a) (b)
2.5
1.0 44.90%

2.0

0.5 v3 (mM/s)
1.5 28.24%
Factor 2 : 28.24%

Eigenvalue
0.0 1.0
15.93%
W (%)
V/m (mL/g)
0.5
7.28%
v2 (mM/s)
-0.5 3.65%

0.0
v1 (mM/s)

-1.0
-0.5
-1.0 -0.5 0.0 0.5 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Factor 1 : 44.90% Eigenvalue number

(c) (d)

Figure 3. PCA results for the KFT data of bread core samples: (a) PC2 versus PC1 scores plot; (b) PC3
versus PC1 scores plot; (c) PC2 versus PC1 loadings plot; and (d) eigenvalues of the correlation matrix.
Meaning of color lines in (a,b): Green—brown wheat bread core (code “BC”); red—whole meal bread
core (code “MC”); blue—Graham bread core (code “GC”). Meaning of color lines in (d): Green numbers
– relevant eigenvalues.

3.2. Principal Component Analysis on KFT Data for Bread Samples

Valuable classifications were obtained in the case of bread samples using PCA technique. All white
bread core samples were located in the center of the scores plot, but less grouped (codes “HC”,
Figure 3a,b). The whole meal and brown bread core samples are well grouped, according to both PC2
versus PC1 and PC3 versus PC1 scores plots (codes “BC” (two groups) and “MC”). Some brown bread
samples are more similar with white bread samples (samples from the market). Graham bread samples
are partially similar with white bread; they are grouped in the upper side of the PC2 versus PC1 scores
plot (code “GC”).
Evident grouping was observed from PCA analysis of the KFT water reaction rate data for core,
“C” and shell “S” bread samples. The core samples are especially located in the center of the scores
plot, while shell samples were classified in the left of this plots (both PC2 versus PC1 and PC3 versus
PC1 , Figure 4a,b). These classifications are due to the KFT water content and normalized volume
(W and V/m), as well as KFT reaction rates corresponding to “strongly retained” water for PC1 (v2 and
v3 ) and to KFT reaction rates corresponding to both “surface” and “strongly retained” water for PC2
(v1 and v2 , Figures 3c and 4c). The water reaction rate corresponding to the “normal” drift in the KFT
process has weak influence in the multivariate analysis. The first two PCs are enough to retain the
important information from the KFT data, according to eigenvalues of the correlation matrix from the
PCA analysis (Figures 3d and 4d). These PCs explain 73.14% and 80.43% from the variance of the KFT
data corresponding to bread core samples and all bread core and shell samples, respectively. Other
PCA results are presented in the Supplementary Materials (Figures S3 and S4).
and v2, Figures 3c and 4c). The water reaction rate corresponding to the “normal” drift in the KFT
process has weak influence in the multivariate analysis. The first two PCs are enough to retain the
important information from the KFT data, according to eigenvalues of the correlation matrix from
the PCA analysis (Figures 3d and 4d). These PCs explain 73.14% and 80.43% from the variance of the
KFT
Appl. data corresponding
Sci. 2020, 10, 6518 to bread core samples and all bread core and shell samples, respectively.
9 of 15
Other PCA results are presented in the Supplementary Materials (Figures S3 and S4).

4
4

3
3 Sd
C
CdC C 2 C
2 C
Cd C C C
C
C 1 C

Factor 3: 15.06%
1 CC
Factor 2: 25.41%

C
C C C C
Sd S SS CC
C C 0 C CCC C
0 C C C
C C C Cd C CC
C C S S CC
S CCC C SS C
CC Cd
-1 C -1
C
Sd C
-2 -2
Sd

-3 -3

-4
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4
-4
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4
Factor 1: 55.02%
Factor 1: 55.02%

(a) (b)
3.5
1.0
v2 (mM/s)
v1 (mM/s)
3.0
55.02%

0.5 2.5
Factor 2 : 25.41%

2.0
v3 (mM/s)
Eigenvalue

V/m W
(mL/g)
(%)
0.0 1.5
25.41%

1.0
15.06%

-0.5 0.5
3.43%
1.08%
0.0

-1.0
-0.5
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
-1.0 -0.5 0.0 0.5 1.0
Eigenvalue number
Factor 1 : 55.02%

(c) (d)
Figure
Figure 4.
4. PCA
PCA results
results for
for the
the KFT
KFT data
data ofof all
all bread
bread core
core and
and shell
shell samples:
samples: (a)
(a) PC
PC2 versus
versus PC
PC11 scores
scores
2
plot; (b) PC 3 versus PC1 scores plot; (c) PC2 versus PC1 loadings plot; and (d) eigenvalues of the
plot; (b) PC3 versus PC1 scores plot; (c) PC2 versus PC1 loadings plot; and (d) eigenvalues of the
correlation matrix. Meaning
correlation matrix. Meaning of ofcolor
colorlines
linesinin (a,b):
(a,b): Green—raw
Green—raw bread
bread corecore (code
(code “C”);“C”); pink—pre-
pink—pre-dried
dried
bread core (code “Cd”); blue—raw bread shell (code “S”); red—pre-dried bread shell (code (code
bread core (code “Cd”); blue—raw bread shell (code “S”); red—pre-dried bread shell “Sd”).
“Sd”).
MeaningMeaning oflines
of color colorinlines
(d): in (d): Green
Green numbers—relevant
numbers—relevant eigenvalues
eigenvalues.

4. Discussion
4. Discussion
Water determination
Water determination in in bread
bread samples
samples is
is aa challenge.
challenge. The
The real
real water
water content
content of
of such
such complex
complex
food matrices is difficult to evaluate. There are many factors influencing the results, starting fromfrom
food matrices is difficult to evaluate. There are many factors influencing the results, starting the
the selection
selection of method
of the the method
to theto composition
the composition
and and sample
sample preparation.
preparation. The The
waterwater content
content ranges
ranges in
in the
the common values [20,26,48,49], but with differences between bread samples. However, the water
content of bread was close to the upper limit of the common range of water content values determined
with standard thermal methods or even higher, if the two-component volumetric KFT technique was
applied [48,49]. The explanation is that the KFT is sensitive to water, while other methods such as oven
drying determines the mass loss. This includes all volatiles that evaporate at the selected temperature,
especially water but also aroma compounds. On the other hand, drying at elevated temperature can
induces changes in the sample, sometimes with generation of volatiles [27,31–33,50]. KFT analysis
allows selecting the polarity of the solvent according to sample type and a temperature program
for a complete extraction and titration of water [3,37–39]. Generally, water content of non-packed
bread or fresh from the manufacturer is higher, most probably due the water absorption from the
environment during storage, transportation and commercialization. Ayub et al. (2003) evaluated the
moisture content by oven drying and water activity of bread samples during storage. The moisture
Appl. Sci. 2020, 10, 6518 10 of 15

content increased by a few percent (sometimes by 12%) after four days, while the water activity also
increased [20]. The water content of the bread crumb during the baking process can increase up to
2.5%, followed by a slow decrease up to the final baking step [51]. The maximum moisture content of
the core can reach 51%, depending on the composition and baking conditions [26].
The differences between KFT water content of core (crumb) and shell (crust) of breads are obvious,
the crust having consistently lower values (about half in comparison with the corresponding cores,
Table 1). On the other hand, pre-drying of core and shell samples reveals completely different
comparative values. Pre-drying lowered the water content of core samples only by a third, while the
shell samples by 75–95%. It is clear that most of the water in the core samples are strongly retained,
in comparison with that from shell samples, where a significant water fraction is “surface” water.
This behavior can be explained by the presence in the bread core of less degraded starch and gluten,
as well as other proteins, which absorb higher quantity of water. The absorbed water molecules are
strongly retained by such flour and bread components. On the other hand, the bread crust (shell) has
higher fraction of consistently degraded components and the absorbed water is mainly “surface” water.
These water molecules are easily evaporated [3,52–54]. Ureta and co-workers (2018) performed baking
tests and oven drying moisture determinations on various regions on the bread. The top crust had a
moisture content at the final of the baking process of 3.7–9.5%, depending on the baking temperature.
The upper, lower and core crumb regions had approximately constant moisture content during baking,
with a slight increase at the end of the baking process (up to 42–43%). On the contrary, the moisture
content of various crust regions consistently decreases during baking process [26,51].
There is a correlation between the water content determined by oven drying technique and water
activity of food products. This is due to the fact that both are related through the sorption isotherm
and the water content can be indirectly determined if the sorption isotherm and water activity are
known [31]. Such correlation can be observed for the bran-based bread crumb. An increasing of
the oven drying water content of bread crumb after the addition of bran fractions from 41% to 43%
is correlated with an increasing of the water activity from 0.96 to 0.97 [13,55]. This is particularly
due to the influence of bran on the starch–gluten–water interactions. Similar observations have been
made for bread obtained with potato fiber in order to improve the product characteristics during
storage [56]. On the other hand, the crust and under-crust of various traditional sourdough bread types
had a water content of 19% and 23%, respectively (determined by oven drying). The corresponding
crumb samples had a water content of ~46.4%. They are well correlated with the water activity.
The crust had a water activity of 0.76–0.77, while this parameter was consistently higher for the crumb
(~0.97). However, no significant changes on the water activity during eight days of storage have
been observed [57]. Such correlations, including the influence on the crispness, fragility and plasticity,
have also been obtained for flat extruded wheat and rye breads or bread spread samples containing
eggshell powders [58,59]. Regarding the bread crust and crumb, there are many studies related to
the influence of various ingredients on the water activity, water content and other physical chemical
characteristics [60–65]. No studies on the correlation of the water content determined by KFT and
water activity of bread have been made. However, such correlation cannot be simply made due to very
different conditions of measurements. Water activity is the ratio between the partial vapor pressure of
water in the food sample and the standard state partial vapor pressure of water, which is supposed
to be analyzed in gaseous state. On the contrary, the KFT water content is determined in solution or
suspension (the case of bread samples), which means water extraction in the liquid working medium.
To our knowledge, no significant studies on the application of KFT for evaluating the types of
water molecules in bread have been performed. The strength of physically bonding of water molecules
into a less soluble samples can be evaluated from the KFT curve. An efficient and simple way is
based on KFT kinetics, especially the mean water reaction rates on specific time ranges. Only few
studies on the KFT kinetics for estimating the water content in solids (i.e., magnesium gluconate
dihydrate or sucrose) or solutions have been published [66,67]. Our previous studies demonstrate the
possibility to evaluate the “surface” and “strongly retained” water molecules in solid samples by KFT
Appl. Sci. 2020, 10, 6518 11 of 15

kinetics [3,35–37,68–70]. The “surface” water molecules are weak bonded on the surface of the solid
particle, especially by hydrogen bonding. They appear by absorption during storage, transportation
and even prior analysis of a less soluble solid or semi-solid sample. They also exist in the raw samples.
“Surface” water molecules are easily disposable to the KFT working medium and react quickly. This is
observed on the first pseudolinear range of the KFT curve and are similar to the KFT water reaction in
completely soluble samples [31,33]. On the other hand, “strongly retained” water molecules (physically
bonded in the interior of the less soluble solid or semi-solid particles) react at a much slower rate, which
principally depends on the diffusion of these water molecules to the surface of sample particles. Thus,
this KFT water reaction rate consistently depends on the particle surface or dimensions. It is the case
of bread samples, which are heterogeneous particles. The reaction rate of the “strongly retained” water
molecules in the bread core samples is very different for white bread and whole meal or brown bread
samples due to the heterogeneity and level of degradation in the last case. Moreover, this “strongly
retained” water reaction rate can be easily determined for bread shell samples (Figure 2). Generally,
water reaction rates are more than twenty times higher for “surface” water molecules in bread core
samples, in comparison with the case of bread shell samples (v1 is up to seven times higher than v2 ,
Table 2).
The absolute values for the KFT kinetic parameters are difficult to evaluate in order to identify
the similarity/dissimilarity of the samples and the corresponding influence of these parameters.
The multivariate analysis techniques are the most appropriate for such evaluation. One of these
techniques is PCA, which is often used in food analysis. The bread samples are well classified if
KFT parameters have been applied to PCA, especially for whole meal and brown bread samples.
Moreover, core and shell samples are obviously grouped by means of KFT kinetic parameters. The KFT
water reaction rates corresponding to “surface” and “strongly retained” water molecules (i.e., v1 and
v2 , respectively) are more important for discriminating along the second PC, while the parameters
related to the final KFT water content are important for the first PC (Figures 3c and 4c). The latter
clearly influence the discrimination between core and shell samples (the “C” (core) and “S” (shell)
groups are separated along PC1 , Figure 4a), while the bread types are discriminated by KFT kinetic
parameters, along PC2 (Figure 3a). These two PCs retain the most useful information related for such
discriminations and classifications (Figures 3d and 4d). The behavior of bread samples are similar
to other less soluble samples such as cereal flour [3,36], meat products [35], cyclodextrins and their
complexes [37,68–70].

5. Conclusions
This is the first study on the coupling of Karl Fischer water titration technique with a multivariate
statistical analysis method in order to differentiate bread samples through water reaction kinetics.
The first two principal components in PCA analysis explain more than 70% of the variance of the
KFT data. The bread core and shell samples are completely dissimilar if the “surface” and “strongly
retained” KFT water reaction rates are considered. Moreover, whole meal and white wheat bread
samples are clearly differentiated through the same water reaction rates. These are correlated with
the composition and the way of bonding of water molecules in the bread particles used in the KFT
measurement, but the KFT–PCA technique does not need compositional information for discrimination.
KFT is selective for water, while other methods such as oven drying determine the overall mass change
(“moisture content”, which are also comprises of other volatile compounds).
Consequently, the overall KFT titration parameters could be used for discriminating by sample
region or parts, while KFT kinetics for classifications by sample types. However, such KFT–PCA
coupling technique needs appropriate calibrations and model systems, possibly a database for
every type of samples (e.g., bread products and specific solid ingredients for food such as flour,
dairy products, etc.).
Appl. Sci. 2020, 10, 6518 12 of 15

Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/10/18/6518/s1,

Figure S1: The Volume/sample mass (V/m) versus Time titration curves from KFT analysis of the bread samples,
Figure S2: The Volume/sample mass (V/m) versus Time titration curves from KFT analysis of the pre-dried core,
shell and pre-dried shell of bread samples, Table S1: The mean KFT water reaction rates on the third time range
(pseudolinear, the “normal” drift) for the core, pre-dried core, shell and pre-dried shell of bread samples, Table S2:
Significance levels (p-level) from the Tukey’s HSD (honestly significant difference) test for the KFT volume/sample
mass ratio (V/m) (mL/g) in the case of KFT analysis of all bread samples, Table S3: Significance levels (p-level) from
the Tukey’s HSD (honestly significant difference) test for the water content (W) (%) in the case of KFT analysis of
all bread samples, Table S4: Significance levels (p-level) from the Tukey’s HSD (honestly significant difference) test
for the mean KFT reaction rate for the first time range (v1 ) (mM/s) in the case of KFT analysis of all bread samples,
Table S4: Significance levels (p-level) from the Tukey’s HSD (honestly significant difference) test for the mean KFT
reaction rate for the second time range (v2 ) (mM/s) in the case of KFT analysis of all bread samples, Figure S3:
PCA results for the KFT data of bread core samples, Figure S4: PCA results for the KFT data of all bread core and
shell samples.
Author Contributions: Conceptualization, D.I.H. and N.G.H.; methodology, G.P., I.R., O.A.I., D.I.H., N.G.H.
and M.R.; validation, D.I.H. and N.G.H.; formal analysis, D.I.H. and N.G.H.; investigation, L.R., D.I.H., C.A.L.
and N.G.H.; resources, M.D.O., M.D., G.S.B. and I.D.; writing—original draft preparation, D.I.H. and N.G.H.;
writing—review and editing, G.P., I.R., O.A.I., D.I.H., N.G.H. and M.R.; supervision, N.G.H.; project administration,
N.G.H.; and funding acquisition, I.R. and N.G.H. All authors have read and agreed to the published version of
the manuscript.
Funding: This research was funded by Ministry of Research and Innovation of Romania, Executive Unit for
Financing Higher Education, Research, Development and Innovation (UEFISCDI), grant PNCDI III 2015–2020—ID
368, institutional development project “Ensuring excellence in R&D within USAMVBT”, from the institutional
performance subprogram 1.2, development of the R&D national system program 1.
Acknowledgments: The authors thank Heinz-Dieter Isengard (University of Hohenheim, Germany) for the help
with the KFT equipment and analysis, as well as to Simona Funar-Timofei (“Coriolan Drăgulescu” Institute of
Chemistry, Romanian Academy) for the help with Statistica 7.1 software.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to
publish the results.

References
1. Dewettinck, K.; Van Bockstaele, F.; Kühne, B.; Van de Walle, D.; Courtens, T.M.; Gellynck, X. Nutritional
value of bread: Influence of processing, food interaction and consumer perception. J. Cereal Sci. 2008, 48,
243–257. [CrossRef]
2. Goesaert, H.; Brijs, K.; Veraverbeke, W.S.; Courtin, C.M.; Gebruers, K.; Delcour, J.A. Wheat flour constituents:
How they impact bread quality, and how to impact their functionality. Trends Food Sci. Technol. 2005, 16,
12–30. [CrossRef]
3. Hădărugă, D.I.; Costescu, C.I.; Corpaş, L.; Hădărugă, N.G.; Isengard, H.-D. Differentiation of rye and wheat
flour as well as mixtures by using the kinetics of Karl Fischer water titration. Food Chem. 2016, 195, 49–55.
[CrossRef] [PubMed]
4. Mondal, A.; Datta, A.K. Bread baking—A review. J. Food Eng. 2008, 86, 465–474. [CrossRef]
5. Lopes Almeida, E.; Chang, Y.K.; Joy Steel, C. Dietary fibre sources in bread: Influence on technological
quality. LWT Food Sci. Technol. 2013, 50, 545–553. [CrossRef]
6. Xu, J.; Wang, W.; Li, Y. Dough properties, bread quality, and associated interactions with added phenolic
compounds: A review. J. Funct. Foods 2019, 52, 629–639. [CrossRef]
7. Birch, A.N.; Petersen, M.A.; Arneborg, N.; Hansen, A.S. Influence of commercial baker’s yeasts on bread
aroma profiles. Food Res. Int. 2013, 52, 160–166. [CrossRef]
8. Blandino, M.; Sovrani, V.; Marinaccio, F.; Reyneri, A.; Rolle, L.; Giacosa, S.; Locatelli, M.; Bordiga, M.;
Travaglia, F.; Coïsson, J.D.; et al. Nutritional and technological quality of bread enriched with an intermediated
pearled wheat fraction. Food Chem. 2013, 141, 2549–2557. [CrossRef]
9. Bosmans, G.M.; Lagrain, B.; Fierens, E.; Delcour, J.A. The impact of baking time and bread storage temperature
on bread crumb properties. Food Chem. 2013, 141, 3301–3308. [CrossRef]
10. Caballero, P.A.; Gomez, M.; Rosell, C.M. Improvement of dough rheology, bread quality and bread shelf-life
by enzymes combination. J. Food Eng. 2007, 81, 42–53. [CrossRef]
Appl. Sci. 2020, 10, 6518 13 of 15

11. Cho, I.H.; Peterson, D.G. Chemistry of Bread Aroma: A Review. Food Sci. Biotechnol. 2010, 19, 575–582.
[CrossRef]
12. Corpaş, L.; Hădărugă, N.G.; Codina, G.G.; Riviş, A.; Guran, E.; Baliţa, E.-N.; Hădărugă, D.I. Phospholipids
in homemade bread. J. Agroaliment. Process. Technol. 2012, 18, 336–340.
13. Curti, E.; Carini, E.; Bonacini, G.; Tribuzio, G.; Vittadini, E. Effect of the addition of bran fractions on bread
properties. J. Cereal Sci. 2013, 57, 325–332. [CrossRef]
14. de Lamo, B.; Gómez, M. Bread enrichment with oilseeds. A Review. Foods 2018, 7, 191. [CrossRef]
15. Goesaert, H.; Slade, L.; Levine, H.; Delcour, J.A. Amylases and bread firming—An integrated view. J. Cereal Sci.
2009, 50, 345–352. [CrossRef]
16. Jiang, D.; Peterson, D.G. Identification of bitter compounds in whole wheat bread. Food Chem. 2013, 141,
1345–1353. [CrossRef]
17. Moayedallaie, S.; Mirzaei, M.; Paterson, J. Bread improvers: Comparison of a range of lipases with a
traditional emulsifier. Food Chem. 2010, 122, 495–499. [CrossRef]
18. Shin, D.-J.; Kim, W.; Kim, Y. Physicochemical and sensory properties of soy bread made with germinated,
steamed, and roasted soy flour. Food Chem. 2013, 141, 517–523. [CrossRef]
19. Belz, M.C.E.; Ryan, L.A.M.; Arendt, E.K. The Impact of Salt Reduction in Bread: A Review. Crit. Rev. Food
Sci. Nutr. 2012, 52, 514–524. [CrossRef]
20. Ayub, M.; Wahab, S.; Durrani, Y. Effect of Water Activity (Aw) Moisture Content and Total Microbial Count
on the Overall Quality of Bread. Int. J. Agric. Biol. 2003, 5, 274–278.
21. Chen, G.; Öhgren, C.; Langton, M.; Lustrup, K.F.; Nydén, M.; Swenson, J. Impact of long-term frozen storage
on the dynamics of water and ice in wheat bread. J. Cereal Sci. 2013, 57, 120–124. [CrossRef]
22. Miś, A.; Krekora, M.; Niewiadomski, Z.; Dziki, D.; Nawrocka, A. Water redistribution between model bread
dough components during mixing. J. Cereal Sci. 2020, 95, 103035. [CrossRef]
23. Gil, M.J.; Callejo, M.J.; Rodriguez, G. Effect of water content and storage time on white pan bread quality:
Instrumental evaluation. Z. Lebensm. Unters. Forsch. 1997, 205, 268–273. [CrossRef]
24. Cappelli, A.; Oliva, N.; Cini, E. Stone milling versus roller milling: A systematic review of the effects on
wheat flour quality, dough rheology, and bread characteristics. Trends Food Sci. Technol. 2020, 97, 147–155.
[CrossRef]
25. Ziobro, R.; Korus, J.; Juszczak, L.; Witczak, T. Influence of inulin on physical characteristics and staling rate
of gluten-free bread. J. Food Eng. 2013, 116, 21–27. [CrossRef]
26. Wagner, M.J.; Lucas, T.; Le Ray, D.; Trystram, G. Water transport in bread during baking. J. Food Eng. 2007,
78, 1167–1173. [CrossRef]
27. Margolis, S.A.; Huang, P.H.; Hădărugă, N.G.; Hădărugă, D.I. Water determination. In Encyclopedia of
Analytical Science, 3rd ed.; Elsevier: Oxford, UK, 2019; Volume 10, pp. 382–390.
28. Yazgan, S.; Bernreuther, A.; Ulberth, F.; Isengard, H.-D. Water—An important parameter for the preparation
and proper use of certified reference materials. Food Chem. 2006, 96, 411–417. [CrossRef]
29. Isengard, H.-D.; Kling, R.; Reh, C.T. Proposal of a new reference method to determine the water content of
dried dairy products. Food Chem. 2006, 96, 418–422. [CrossRef]
30. Isengard, H.-D.; Präger, H. Water determination in products with high sugar content by infrared drying.
Food Chem. 2003, 82, 161–167. [CrossRef]
31. Isengard, H.-D. Water content, one of the most important properties of food. Food Control. 2001, 12, 395–400.
[CrossRef]
32. Heinze, P.; Isengard, H.-D. Determination of the water content in different sugar syrups by halogen drying.
Food Control. 2001, 12, 483–486. [CrossRef]
33. Isengard, H.-D. Rapid water determination in foodstuffs. Trends Food Sci. Technol. 1995, 6, 155–162. [CrossRef]
34. Sahin, A.W.; Wiertz, J.; Arendt, E.K. Evaluation of a new method to determine the water addition level in
gluten-free bread systems. J. Cereal Sci. 2020, 93, 102971. [CrossRef]
35. Zippenfening, S.E.; Hădărugă, D.I.; Negrea, I.; Velciov, A.; Hădărugă, N.G. Karl Fischer titration of salami
products: Variation of the water content during ripening and kinetics of the KFT process. Chem. Bull. Politeh.
Univ. Timişoara 2015, 64, 18–24.
36. Corpaş, L.; Hădărugă, N.G.; David, I.; Pîrşan, P.; Hădărugă, D.I.; Isengard, H.-D. Karl Fischer water
titration—Principal component analysis approach on wheat flour. Food Anal. Methods 2014, 7, 1353–1358.
[CrossRef]
Appl. Sci. 2020, 10, 6518 14 of 15

37. Hădărugă, N.G.; Hădărugă, D.I.; Isengard, H.-D. Water content of natural cyclodextrins and their essential
oil complexes: A comparative study between Karl Fischer titration and thermal methods. Food Chem. 2012,
132, 1741–1748. [CrossRef]
38. Felgner, A.; Schlink, R.; Kirschenbuhler, P.; Faas, B.; Isengard, H.-D. Automated Karl Fischer titration for
liquid samples—Water determination in edible oils. Food Chem. 2008, 106, 1379–1384. [CrossRef]
39. Schmitt, K.; Isengard, H.-D. Karl Fischer titration. A method for determining the true water content of
cereals. Fresenius’ J. Anal. Chem. 1998, 360, 465–469. [CrossRef]
40. Merkh, G.; Pfaff, R.; Isengard, H.-D. Capabilities of automated Karl Fischer titration combined with gas
extraction for water determination in selected dairy products. Food Chem. 2012, 132, 1736–1740. [CrossRef]
41. Isengard, H.-D.; Kerwin, H. Proposal of a new reference method for determining water content in butter oil.
Food Chem. 2003, 82, 117–119. [CrossRef]
42. Tomassetti, M.; Campanella, L.; Aureli, T. Thermogravimetric analysis of some spices and commercial
food products. Comparison with other analytical methods for moisture content determination (Part 3).
Thermochim. Acta 1989, 143, 15–26. [CrossRef]
43. Isengard, H.-D.; Haschka, E.; Merkh, G. Development of a method for water determination in lactose.
Food Chem. 2012, 132, 1660–1663. [CrossRef]
44. Kestens, V.; Conneely, P.; Bernreuther, A. Vaporisation coulometric Karl Fischer titration: A perfect tool
for water content determination of difficult matrix reference materials. Food Chem. 2008, 106, 1454–1459.
[CrossRef]
45. Rückold, S.; Grobecker, K.H.; Isengard, H.-D. The effects of drying on biological matrices and the consequences
for reference materials. Food Control. 2001, 12, 401–407. [CrossRef]
46. Kestens, V.; Charoud-Got, J.; Bau’, A.; Bernreuther, A.; Emteborg, H. Online measurement of water content
in candidate reference materials by acousto-optical tuneable filter near-infrared spectrometry (AOTF-NIR)
using pork meat calibrants controlled by Karl Fischer titration. Food Chem. 2008, 106, 1359–1365. [CrossRef]
47. Isengard, H.-D.; Färber, J.-M. ‘Hidden parameters’ of infrared drying for determining low water contents in
instant powders. Talanta 1999, 50, 239–246. [CrossRef]
48. Belitz, H.-D.; Grosch, W.; Schieberle, P. Food Chemistry; Springer: Berlin, Germany, 2009; pp. 670–745.
[CrossRef]
49. Bhatt, C.M.; Nagaraju, J. Studies on electrical properties of wheat bread as a function of moisture content
during storage. Sens. Instrum. Food Qual. Saf. 2010, 4, 61–66. [CrossRef]
50. Isengard, H.-D.; Heinze, P. Determination of total water and surface water in sugars. Food Chem. 2003, 82,
169–172. [CrossRef]
51. Ureta, M.M.; Diascorn, Y.; Cambert, M.; Flick, D.; Salvadori, V.O.; Lucas, T. Water transport during bread
baking: Impact of the baking temperature and the baking time. Food Sci. Technol. Int. 2018, 25, 187–197.
[CrossRef]
52. Hădărugă, D.I.; Birău (Mitroi), C.L.; Gruia, A.T.; Păunescu, V.; Bandur, G.N.; Hădărugă, N.G. Moisture
evaluation of β-cyclodextrin/fish oils complexes by thermal analyses: A data review on common barbel
(Barbus barbus L.), Pontic shad (Alosa immaculata Bennett), European wels catfish (Silurus glanis L.),
and common bleak (Alburnus alburnus L.) living in Danube river. Food Chem. 2017, 236, 49–58. [CrossRef]
53. Hădărugă, N.G.; Bandur, G.N.; David, I.; Hădărugă, D.I. A review on thermal analyses of cyclodextrins and
cyclodextrin complexes. Environ. Chem. Lett. 2019, 17, 349–373. [CrossRef]
54. Hădărugă, N.G.; Szakal, R.N.; Chirilă, C.A.; Lukinich-Gruia, A.T.; Păunescu, V.; Muntean, C.; Rusu, G.;
Bujancă, G.; Hădărugă, D.I. Complexation of Danube common nase (Chondrostoma nasus L.) oil by
β-cyclodextrin and 2-hydroxypropyl-β-cyclodextrin. Food Chem. 2020, 303, 125419. [CrossRef] [PubMed]
55. Curti, E.; Carini, E.; Tribuzio, G.; Vittadini, E. Effect of bran on bread staling: Physico-chemical characterization
and molecular mobility. J. Cereal Sci. 2015, 65, 25–30. [CrossRef]
56. Curti, E.; Carini, E.; Diantom, A.; Vittadini, E. The use of potato fibre to improve bread physico-chemical
properties during storage. Food Chem. 2016, 195, 64–70. [CrossRef]
57. Chiavaro, E.; Vittadini, E.; Musci, M.; Bianchi, F.; Curti, E. Shelf-life stability of artisanally and industrially
produced durum wheat sourdough bread (“Altamura bread”). LWT Food Sci. Technol. 2008, 41, 58–70.
[CrossRef]
58. Marzec, A.; Lewicki, P.P. Antiplasticization of cereal-based products by water. Part I. Extruded flat bread.
J. Food Eng. 2006, 73, 1–8. [CrossRef]
Appl. Sci. 2020, 10, 6518 15 of 15

59. Kobus-Cisowska, J.; Szymanowska-Powałowska, D.; Szymandera-Buszka, K.; Rezler, R.; Jarz˛ebski, M.;
Szczepaniak, O.; Marciniak, G.; J˛edrusek-Golińska, A.; Kobus-Moryson, M. Effect of fortification with calcium
from eggshells on bioavailability, quality, and rheological characteristics of traditional Polish bread spread.
J. Dairy Sci. 2020, 103, 6918–6929. [CrossRef]
60. Primo-Martín, C.; de Beukelaer, H.; Hamer, R.J.; Van Vliet, T. Fracture behaviour of bread crust: Effect of
ingredient modification. J. Cereal Sci. 2008, 48, 604–612. [CrossRef]
61. Purlis, E.; Salvadori, V.O. Modelling the browning of bread during baking. Food Res. Int. 2009, 42, 865–870.
[CrossRef]
62. Primo-Martín, C.; van de Pijpekamp, A.; van Vliet, T.; de Jongh, H.H.J.; Plijter, J.J.; Hamer, R.J. The role of the
gluten network in the crispness of bread crust. J. Cereal Sci. 2006, 43, 342–352. [CrossRef]
63. Pico, J.; Reguilón, M.P.; Bernal, J.; Gómez, M. Effect of rice, pea, egg white and whey proteins on crust quality
of rice flour-corn starch based gluten-free breads. J. Cereal Sci. 2019, 86, 92–101. [CrossRef]
64. Lind, I.; Rask, C. Sorption isotherms of mixed minced meat, dough, and bread crust. J. Food Eng. 1991, 14,
303–315. [CrossRef]
65. Wolter, A.; Hager, A.-S.; Zannini, E.; Czerny, M.; Arendt, E.K. Influence of dextran-producing Weissella cibaria
on baking properties and sensory profile of gluten-free and wheat breads. Int. J. Food Microbiol. 2014, 172,
83–91. [CrossRef] [PubMed]
66. Grünke, S.; Wünsch, G. Kinetics and stoichiometry in the Karl Fischer solution. Fresenius’ J. Anal. Chem.
2000, 368, 139–147.
67. Yap, W.T.; Cummings, A.L.; Margolis, S.A.; Schaffer, R. Estimation of water content by kinetic method in
Karl Fischer titration. Anal. Chem. 1979, 51, 1595–1596. [CrossRef]
68. Hădărugă, D.I.; Hădărugă, N.G.; Bandur, G.N.; Isengard, H.-D. Water content of flavonoid/cyclodextrin
nanoparticles: Relationship with the structural descriptors of biologically active compounds. Food Chem.
2012, 132, 1651–1659. [CrossRef]
69. Hădărugă, N.G.; Hădărugă, D.I.; Isengard, H.-D. “Surface water” and “strong-bonded water” in cyclodextrins:
A Karl Fischer titration approach. J. Incl. Phenom. Macrocycl. Chem. 2013, 75, 297–302. [CrossRef]
70. Ünlüsayin, M.; Hădărugă, N.G.; Rusu, G.; Gruia, A.T.; Păunescu, V.; Hădărugă, D.I. Nano-encapsulation
competitiveness of omega-3 fatty acids and correlations of thermal analysis and Karl Fischer water titration
for European anchovy (Engraulis encrasicolus L.) oil/β-cyclodextrin complexes. LWT Food Sci. Technol. 2016,
68, 135–144. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).