The DSC approach to study non-freezing water contents of hydrated hydroxypropylcellulose (HPC)
Journal of Thermal Analysis and Calorimetry
The DSC approach to study non-freezing water contents of hydrated hydroxypropylcellulose (HPC)
Przemysław Talik 0 1
Urszula Hubicka 0 1
0 Department of Inorganic and Analytical Chemistry, Faculty of Pharmacy, Jagiellonian University Medical College , 9 Medyczna St., 30-688 Krakow , Poland
1 & Przemysław Talik
The research consisted of investigation of non-freezable bound water (NFW) contents of raw hydroxypropylcellulose (HPC) Klucel Pharm HF, MF, and LF by differential scanning calorimetry. Polymers (HF, MF and LF) used in the research differed significantly in their molecular mass and viscosity. Mixtures of HPC and low soluble salicylic acid or highly soluble sodium salicylate were used to examine the effect of counter-ions presence and drug solubility on NFW contents. The dependencies of determined enthalpies of melting (DH) and respective contents of water Wc (water fraction in the dry mass of HPC or HPC mixture) DH = f(Wc) were linear. The sought values of non-freezing water were calculated (by extrapolation) as the concentration at which DH = 0. It was found that the viscosity and molecular mass of the raw HPCs did not affect the contents of non-freezing water, which ranged between 0.54 and 0.51 g g-1 for all types LF, MF, HF. Poorly soluble salicylic acid (SA) reduced the non-freezing water content to 0.19-0.17 g g-1. On the other hand, matrices composed of highly soluble sodium salicylate (NaSA) showed variable decrease of NFW content-0.44 g g-1 for HF and 0.54-0.51 g g-1 for LF and MF. It was also found that the maximum temperature Tmax of melting was strongly influenced by the solubility of the drug. Highly soluble NaSA shifted Tmax toward the lower negative values, while SA toward the higher positive values. Assessing NFW contents in hydrated HPC matrices may contribute to a better understanding of the drug release and dissolution mechanisms of tablet formulations.
Hydroxypropylcellulose (HPC) is a non-ionic,
water-soluble cellulose ether derivative. It is a semi-crystalline
polymer with relatively low Tg and a high degree of
amorphous content, which causes high molecular mobility
and plasticity [
]. Similarly to other structural analogs of
cellulose, HPC can be treated as a natural material, that is
biocompatible with human tissues (including blood),
nontoxic and relatively low cost [
]. Its structure with a molar
substitution of 3.0 is shown in Fig. 1.
The physical and chemical properties of HPC can be
modified by controlling side chain length and branching
during the synthesis process. This is possible, because the
secondary hydroxyl groups, shown in side chains, are
available for further etherification with propylene oxides
]. This results in formation of structures, containing more
than one propylene substituent. The flexibility of HPC
synthesis allows it to be obtained in several types
characterized by various molecular mass (from 80,000—Klucel
EF Pharm to 1,150,000—Klucel HF Pharm) and viscosity
(ranging from 25—Klucel LF Pharm to 6500—Klucel
MF Pharm mPas). For that reason, HPC, among the other
polysaccharides, is widely applied as a drug delivery
carrier for oral drug delivery systems in pharmacy [
]. It is
Fig. 1 The structure of hydroxypropylcellulose with a molar
substitution of 3.0
also frequently used as a thickener, stabilizer, emulsifier or
fat replacer in food industry [
]. HPC has also found
applications in material science and engineering [
Similarly as in the case of other hydrophilic
cellulosebased polymers (cellulose, chitosan, schizophyllan,
hyaluronan, CMC) [
], HPC–water interactions are
believed to be related to hydrogen bonding, which can lead
to formation of new stable structures [
]. These entities
are particularly interesting due to their ability to drastically
change their mechanical and physicochemical properties.
Modern explanation of that peculiar capacity assumes the
existence of three distinct fractions of water: free (or bulk)
water (FW), freezable bound water (FBW) and
non-freezable bound water (NFW) [
]. The first type of water—free
water—does not significantly differ in its
melting/crystallization temperature and enthalpy from normal (bulk)
water. The second type, called freezable bound water,
represents the water which is less closely associated with
polymer chains. It exhibits a crystallization phase transition
and a melting point lower than 0 C which distinguishes it
from the bulk water. This fraction is also characterized by
supercooling and significantly smaller enthalpies than the
free water. When first-order phase transitions of small
fractions of water (strongly associated with the polymer
matrix) cannot be observed calorimetrically, we speak
about non-freezable bound water which is the third type.
This type of water embedded in hydrophilic polymer does
not crystallize even when cooled down to - 100 C [
For low water contents, all water found in the matrix is
considered to be non-freezable [
]. The sum of the
freezing bound and non-freezing bound water fractions is
the bound water content.
Coexistence of those types of water in the polymer
matrix forms a complex, consisting of polymer chains, ice,
unfrozen water and air. Both ice and unfrozen water can be
embedded in the pores and/or cavities of the matrix. When
the dimensions of such a cavity do not exceed several
Angstroms, the crystallization of water is difficult or
impossible. Liu et al. [
] called such hollow spaces
‘‘nanocavities’’ and demonstrated that they could be an
important reason for the formation of non-freezing water.
This author concluded that hydrogen bonding is not the
sole factor that has influence over water crystallization.
Moreover, they suggested that it is more acceptable to
think that bound water bind with hydrogen bonds is just
one of the various physical states in the polymer.
The existing techniques that can be applied in the
research of bound water include, among the others, nuclear
magnetic resonance (NMR) [
], magnetic resonance
imaging (MRI) , time domain reflectometry (TDR)
] and differential scanning calorimetry (DSC)
14, 15, 22–25
]. However, the melting/freezing
experiments using DSC seem to be the most suitable to
investigate and quantify the NFW.
The method described in [
14, 22, 24
] was successfully
applied to determine the NFW contents of some raw
hydrophilic polysaccharides: cellulose, chitosan and
schizophyllan (0.14, 0.37 and 0.38 g g-1 respectively),
hyaluronan, CMC 90 kDa and CMC700 kDa (0.77, 0.77,
0.84 g g-1 respectively) [
As described above water holding properties are
virtually dependent on the chemical and physical structure of
polymer matrix. The objective of this paper was to
investigate the non-freezable bound water contents of raw
hydroxypropylcellulose Klucel Pharm HF, MF, and LF
by differential scanning calorimetry, three polymers
significantly different in their molecular mass and viscosity.
Physical mixtures of HPC and low soluble salicylic acid or
highly soluble sodium salicylate were used to examine the
effect of counter-ions presence and drug solubility on NFW
Assessing NFW contents in hydrated HPC matrices may
contribute to a better understanding of the drug release and
dissolution mechanisms of tablet formulations.
The pharmaceutical Klucel Pharm
hydroxypropylcelluloses of viscosity LF, MF and HF were purchased from
Hercules Incorporated, Aqualon Division. Their Brookfield
viscosity (H2O, cps), moles of substitution, molecular mass
and percent of hydroxypropoxy groups, as quoted by the
manufacturer, are collected in Table 1. The salicylic acid
(SA) was purchased from Pharma Cosmetic, Krakow,
Poland (batch 113383) and sodium salicylate (NaSA) was
from Amara, Krakow, Poland (batch 281113) both received
as a gift. The water solubility of salicylic acid is 2.2 g L-1,
while sodium salicylate is nearly 500 times higher.
The HPC material was gently mixed for 5 min with the
examined drugs using an agate mortar and pestle at a 1:1
polymer/drug w/w ratio to prepare 3 g of each of the
corresponding physical mixtures (PMs). Obtained PMs as well
as raw HPC samples were dried in a vacuum desiccator
over phosphorus pentoxide for about 4 weeks. Then,
portions of about 20–30 mg were accurately weighed and
placed into plastic pans. An excess of Milli-Q water was
added. The water was allowed to evaporate slowly at room
temperature until the desired water content Wc from 0.20 to
5.00 g g-1 was obtained. The Wc was defined as water
fraction mH2O related to dry mass mdry of HPC or HPC
MPs (Wc = mH2O 9 md-r1y). The aluminum sample pans
were exposed to steam in an autoclave at 120 C for 3 h, to
form a passive coat of Al2O3, for protection from
spontaneous reactions with water. Then, the homogeneous
material of about 8 mg was transferred to pans,
hermetically sealed and conditioned at room temperature for about
30 h to obtain an equilibrium state. The crucibles were
weighted before and after each measurement to insure that
there is no loss of mass.
The qualitative and quantitative studies of hydrated
polymers using DSC are based on freezing and thawing
experiments, in which the difference in physical properties
between FBW in form of ice and NFW is investigated. As
previously suggested in [
] and also successfully applied
], the measured enthalpies of melting were first
normalized to the mass of the dry raw HPC or their PMs
and then plotted against the respective water concentration
(Wc). In this way, the linear dependences DH = f(Wc) were
obtained and the NFW contents were estimated by
extrapolation to DH = 0.
DSC analysis was performed with DSC 7020 calorimeter,
Hitachi Inc., equipped with DSC 7020 electric cooling unit.
The apparatus was calibrated with indium, tin standards
(purity of 99.9999%) and Milli-Q water. The freezing and
heating measurements were performed in nitrogen
atmosphere with a flow rate of 50 mL min-1 in three steps:
cooling from 20 C to - 55 C at 3 C min-1, isothermal
at - 55 C for 2 min and heating back to 20 C at
3 C min-1. In order to reduce experimental errors, DSC
runs were repeated three times.
The obtained plots were interpreted using Muse
Measurement v 9.21U software. The determination of
enthalpies was carried out using a T-Slice Analysis (Integral
Tangential) taking into account the baseline shift and
nonlinearity. The melting temperatures of water absorbed
in the sample were determined at the maximum point of the
corresponding enthalpy peaks (Tmax).
Results and discussion
To achieve improved readability upper parts of curves
corresponding to crystallization were cut off, because
nearly all of them showed a large ‘‘loop’’. This artifact is
shown in Fig. 2 and is attributed to the exothermic
processes during water crystallization.
Figure 3 shows some representative examples of DSC
runs, obtained from hydrated, raw HPC HF samples of Wc
ranging from 0.56 to 2.46 g g-1. The first curve marked as
Wc = 0.56 can be considered a straight baseline because no
significant phase transition of water is observed. It can be
explained by the fact that for low concentrations of water,
nearly all absorbed molecules of non-freezable water
nd –40 –20
Fig. 3 The DSC melting curves, obtained from raw HPC HF/water
samples of Wc ranging from 0.56 to 2.46. The arrows point a
magnified small and broad peak obtained for sample with Wc = 0.99
(subfigure a) and two overlapped peaks obtained for samples with
Wc = 2.00 and 2.46 (subfigure b)
(0.56 g of water per 1.00 g of HPC HF) diffuse into the
amorphous region of the sample and each of them is being
restrained by the hydroxyl group of
hydroxypropylcellulose chain. When the quantity of water in the matrix is
slightly increased, the first-order transition of
crystallization is absent, while the melting peak can be seen. This
happens unless the water content exceeds a certain Wc
amount, which varies in accordance with the chemical
structure of polysaccharide and/or additives. When this
value is reached, typically two separate melting peaks are
observed, usually with a very distant Tmax. That
phenomenon is highlighted in Fig. 3 (subfigure a). The first
peak on Wc = 0.99 at the temperature Tmax = - 21.6 C
was very small and broad and the second, well formed at
Tmax = - 3.6 C. With higher contents of water, the peaks
were shifted to a higher temperature range. In most of
cases, only one broad endothermic event (two overlapped
peaks) could be seen [
]. However, rarely the peaks
were separated and that is shown on subfigure b. The water
molecules originating from ice which melts at higher
temperatures (around 0 C) are assigned to a free water,
and those that melt at proportionally lower temperatures
are assigned to the freezable bound water .
In Fig. 4, a relationship between maximum temperatures
Tmax of melting of water vs. appropriate Wc, obtained from
HPC and their mixtures is shown.
Figure 4A gathers LF, MF and HF types of raw
hydroxypropylcellulose. One can see that all lines are very
close to each other and form a cluster. This means that the
internal architecture of the samples was comparable and
nearly the same number of hydroxyl groups interacts with
similar water molecules. The effect of polymer mass and
viscosity is negligible. Further analysis of Fig. 4a shows
that the cluster passes the 0 C axis (red dashed line)
Water content Wc/gwaterg–1dry mass
Water content Wc/gwaterg–1dry mass
Water content Wc/gwaterg–1dry mass
between Wc around 1.75 and 1.85, after a rapid increase
from about - 9 C. Then, the lines reach a plateau and do
not pass 2 C until the end of the measurement range. The
sole exception is HPC MF which goes along 0 C axis.
Figure 4b collects all used types of HPC and low solubility
salicylic acid physical mixtures. As in the case of raw
material, the HPC MF/SA PM is slightly under the rest of
lines; however, the distance does not exceed 1.1 C. In the
whole range of Wc, the Tmax were positive and the cluster
gently rose from 0 C not exceeding 3 C
(DTmax & 3 C). This corresponds very well with the
observations of Faroongsarng and Sukonrat [
showed that the depression of melting temperature is not
only attributed to water–polymer interactions, but water
embedded in the pores and/or cavities of the matrix as well.
In HPC/SA mixtures, both phenomena take place. There
are less hydroxyl groups when compared to raw HPCs and
access to them is more restricted in the presence of SA
molecules. The opposite effect can be seen in Fig. 4c,
where HPC mixtures with high soluble sodium salicylate
(NaSA) were studied. The lines are not as bundled as raw
HPC or SA PMs samples, and they are crossing each other.
The maximum temperatures of the group begin between
- 16 and - 12 C and are negative in the whole
measurement range (DTmax & 12–16 C). Considering that the
architecture of a matrix (1:1 w/w HPC/NaSA) is similar or
even identical to that of salicylic acid, such interactions
must be related to a strong influence of an easily
dissociated sodium cation.
The determination of non-freezable bound water content
is possible using both the crystallization or melting
enthalpies. However, obtained values are not equal and the
NFW contents determined using crystallization enthalpies
are smaller than those calculated from melting enthalpies.
That phenomenon was formerly explained and discussed
by Mlcˇoch and Kucˇer´ık [
]. In their publication, the
authors demonstrated that the difference in enthalpy of
crystallization and melting is related to the properties of
water and is ‘‘hidden’’ in the segment of the curve where
the heating is switched into cooling part.
To complete these conclusions, we want to point at
mentioned above artifact (the ‘‘loop’’) that appeared during
the cooling. Certainly, this is another important factor
which affects the final measurements. To avoid this
problem, we tried to lower the cooling rate or decrease the
sample mass. However, it did not bring the expected result.
For that reason, only the cooling parts of the DSC curves
were used in further work.
The measured melting enthalpies of hydrated raw HPC
samples and their PMs were plotted against respective Wc
and a very good fit with the linear DH = f(Wc) regression
(as it is shown in Fig. 5) was obtained. The linear trend was
maintained throughout the measuring range.
The calculated values of NFW and accompanying
coefficients of determination R2 are listed in Table 2.
Fig. 5 The determined enthalpies of melting versus respective
contents of water Wc represented linear dependences. The sought
value of non-freezing water calculated as the concentration at which
DH = 0 (extrapolation) is marked with red triangle. The figure shows
an example obtained for HPC MF (DH = - 182.5 ? 337.7 9 Wc)
As seen in Table 2, the contents of non-freezing water in
different LF, MF, HF types of raw HPC are nearly equal
and range from 0.51 to 0.54 g g-1. This indicates that the
molecular mass and viscosity do not affect the content of
non-freezing water. In this case, the access to hydroxyl
groups and their amount, which is the same, plays the key
role (see Table 1). For the salicylic acid physical mixtures,
the non-freezing water contents were also similar, but
ranging from 0.17 to 0.19 g g-1 (3 times smaller
comparing to their raw analogs). This confirms earlier research
], where authors demonstrated, that apart from hydrogen
bonding, also the presence of ‘‘nanocavities’’ influences the
formation of non-freezing water. In other words, such a
threefold decrease may be explained by the fact that with
the lowering amount of hydroxyl groups, the number of
‘‘nanocavities’’ also decreases. That is because the
‘‘nanocavities’’ were occupied or there was limited access
to them due to small salicylic acid molecules.
The last group of samples under study was HPC and
NaSA matrices. What can be considered interesting is that
the NFW contents were similar to unmixed HPC. The LF
and MF types were nearly similar and equaled to 0.50 and
0.48 g g-1, respectively, HF was 0.44 g g-1. This clearly
shows that the deficiency of hydroxyl groups was
compensated by the strong influence of the counter ion Na?
]. High dissociation leads to a stronger interaction
between the dissociated ion pair and water molecules. This
phenomenon activates adsorption of bound water.
The number of water molecules per disaccharide unit
(nNFW) calculated for all types of raw HPC was in the range
of 21.7–22.9 and was comparable to nNFW obtained for
CMC 700 kDa 21.9 [
]. The other studied polymers such
as cellulose, chitosan, schizophyllan, hyaluronan and CMC
90 kDa have nNFW lower than 18.7 (2.4, 7.00, 14.1, 17.2,
In the present study, differential scanning calorimetry was
used to investigate the non-freezing water contents of
hydroxypropylcellulose of various molecular mass and
viscosities: HF, MF, LF and their matrices composed of
drugs with different solubilities such as salicylic acid and
sodium salicylate. It was found that under the measurement
conditions, the viscosity and molecular mass affects the
NFW contents in a raw polymer only negligibly. Poorly
soluble small molecule drug (salicylic acid) reduced the
NFW amount nearly 3 times. It was demonstrated that the
formation of non-freezable water is influenced by both the
reduced access to a smaller number of hydroxyl groups and
the spatial architecture (‘‘nanocavity’’) of the polymer. The
NFW contents calculated for matrices with highly soluble
sodium salicylate were comparable to those of raw HPC,
which shows how strong an influence of the counter Na?
ion is. During the studies, it was also found that the
maximum temperature Tmax of melting was strongly influenced
by the solubility of the drug. Highly soluble NaSA shifted
Tmax toward the lower negative values, while SA toward
the higher positive values.
This type of research may contribute to better
understanding of processes that influence dissolution behavior of
both highly and poorly soluble drugs from tablet
Open Access This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://creative
commons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
1. Picker-Freyer KM , Durig T. Physical mechanical and tablet formation properties of hydroxypropylcellulose: in pure form and in mixtures . AAPS PharmSciTech . 2007 ; 8 ( 4 ): E1 - 9 .
2. Alderman DA . A review of cellulose ethers in hydrophilic matrices for oral controlled-release dosage forms . Int J Pharm Technol Prod Manuf . 1984 ; 5 : 1 - 9 .
3. Klose RE , Glickman M. Gums . In: Furia TE, editor. Handbook of food additives. 2nd ed., Tom 1 . CRC Press: Boca Raton London, New York: Washington D.C.; 1973 . p. 323 .
4. Tanaka A , Furubayashi T , Tomisaki M , Kawakami M , Kimura S , Inoue D , Kusamori K , Katsumi H , Sakane T , Yamamoto A . Nasal drug absorption from powder formulations: the effect of three types of hydroxypropyl cellulose (HPC) . Eur J Pharm Sci . 2017 ; 96 : 284 - 9 .
5. Macchi E , Zema L , Pandey P , Gazzaniga A , Felton LA . Influence of temperature and relative humidity conditions on the pan coating of hydroxypropyl cellulose molded capsules . Eur J Pharm Biopharm . 2016 ; 100 : 47 - 57 .
6. Hughey JR , Keen JM , Bennett RC , Obara S , McGinity JW . The incorporation of low-substituted hydroxypropyl cellulose into solid dispersion systems . Drug Dev Ind Pharm . 2015 ; 41 ( 8 ): 1294 - 301 .
7. Bajdik J , Regdon G Jr, Marek T , Eros I , Su¨vegh K, Pintye-Ho´di K. The effect of the solvent on the film-forming parameters of hydroxypropyl-cellulose . Int J Pharm . 2005 ; 301 ( 1-2 ): 192 - 8 .
8. Ye R , Harte F. High pressure homogenization to improve the stability of casein-hydroxypropyl cellulose aqueous systems . Food Hydrocoll . 2014 ; 35 : 670 - 7 .
9. Ye R , Cheng Q , Cai J , Feng T , Wang G . Stable casein-hydroxypropyl cellulose complexes at low pH . J Food Qual . 2016 ; 39 ( 4 ): 292 - 300 .
10. Liu X , Zhou Y , Nie W , Song L , Chen P . Fabrication of hydrogel of hydroxypropyl cellulose (HPC) composited with graphene oxide and its application for methylene blue removal . J Mater Sci . 2015 ; 50 ( 18 ): 6113 - 23 .
11. Ledwon P , Andrade JR , Lapkowski M , Pawlicka A . Hydroxypropyl cellulose-based gel electrolyte for electrochromic devices . Electrochim Acta . 2015 ; 159 : 227 - 33 .
12. Barzic AI , Dimitriu DG , Dorohoi DO . New method for determining the optical rotatory dispersion of hydroxypropyl cellulose polymer solutions in water . Polym Eng Sci . 2015 ; 55 ( 5 ): 1077 - 81 .
13. Vueba ML , Batista de Carvalho LA , Veiga F , Sousa JJ , Pina ME . Influence of cellulose ether mixtures on ibuprofen release: MC25, HPC and HPMC K100 M. Pharm Dev Technol . 2006 ; 11 ( 2 ): 213 - 28 .
14. Mlcˇoch T , Kucˇer´ık J. Hydration and drying of various polysaccharides studied using DSC . J Therm Anal Calorim . 2013 ; 113 ( 3 ): 1177 - 85 .
15. Hatakeyama T , Inui Y , Iijima M , Hatakeyama H . Bound water restrained by nanocellulose fibres . J Therm Anal Calorim . 2013 ; 113 ( 3 ): 1019 - 25 .
16. Wolfe J , Bryant G , Koster KL . What is'unfreezable water', how unfreezable is it and how much is there ? CryoLetters . 2002 ; 23 : 157 - 66 .
17. Ping ZH , Nguyen QT , Chen SM , Zhou JQ , Ding YD . States of water in different hydrophilic polymers-DSC and FTIR studies . Polymer . 2001 ; 42 : 8461 - 7 .
18. Liu GL , Yao KD . What causes the unfrozen water in polymers: hydrogen bonds between water and polymer chains? Polymer . 2001 ; 42 : 3943 - 7 .
19. Watanabe K , Wake T. Measurement of unfrozen water content and relative permittivity of frozen unsaturated soil using NMR and TDR . Cold Reg Sci Technol . 2009 ; 59 : 34 - 41 .
20. Pru˚sˇova˘ A, Conte P , Kucˇer´ık J , Alonzo G . Dynamics of hyaluronan aqueous solutions as assessed by fast field cycling NMR relaxometry . Anal Bioanal Chem . 2010 ; 397 : 3023 - 8 .
21. Kulinowski P , Młynarczyk A , Jasin´ski K , Talik P , Gruwel ML , Tomanek B , We˛glarz WP, Doroz_yn´ski P. Magnetic resonance microscopy for assessment of morphological changes in hydrating hydroxypropylmethylcellulose matrix tablets in situ-is it possible to detect phenomena related to drug dissolution within the hydrated matrices? Pharm Res . 2014 ; 31 ( 9 ): 2383 - 92 .
22. Liu J , Cowman MK . Thermal analysis of semi-dilute hyaluronan solutions . J Therm Anal Calorim . 2000 ; 59 : 547 - 57 .
23. Dehabadi L , Udoetok IA , Wilson LD . Macromolecular hydration phenomena: an overview of DSC studies on sulfonated tetrafluoroethylene-based fluoropolymer-copolymer (Nafion) and cellulose biopolymer materials . J Therm Anal Calorim . 2016 ; 126 ( 3 ): 1851 - 66 .
24. Pru˚sˇova˘ A, Sˇ mejkalova ˘ D, Chytil M , Velebny ´ V, Kucˇer´ık J. An alternative DSC approach to study hydration of hyaluronan . Carbohydr Polym . 2010 ; 82 : 498 - 503 .
25. Hatakeyama T , Tanaka M , Hatakeyama H . Thermal properties of freezing bound water restrained by polysaccharides . J Biomater Sci Polym Ed . 2010 ; 21 : 1865 - 75 .
26. Berthold J , Desbrie`res J , Rinaudo M , Salme´n L. Types of adsorbed water in relation to the ionic groups and their counter-ions for some cellulose derivatives . Polymer . 1994 ; 35 ( 26 ): 5729 - 36 .
27. Faroongsarng D , Sukonrat P . Thermal behavior of water in the selected starch- and cellulose-based polymeric hydrogels . Int J Pharm . 2008 ; 352 ( 1-2 ): 152 - 8 .