Hydration of Microcrystalline Cellulose and Milled Cellulose Studied

The analyzed relative pressures (P/P0) were in the range of 0.05−0.985. ... Grinding was performed in 10 min intervals, with 10 min pauses in betwee...
0 downloads 0 Views 262KB Size
Download PDF
3728

J. Phys. Chem. B 2008, 112, 3728-3734

Hydration of Microcrystalline Cellulose and Milled Cellulose Studied by Sorption Calorimetry Vitaly Kocherbitov,*,† Stefan Ulvenlund,‡ Maria Kober,‡ Kjell Jarring,‡ and Thomas Arnebrant† Biomedical Laboratory Science and Technology, Faculty of Health and Society, Malmo¨ UniVersity, SE-205 06 Malmo¨, Sweden, and AstraZeneca R&D Lund, S-22187 Lund, Sweden ReceiVed: December 7, 2007; In Final Form: January 11, 2008

The hydration of two different polymorphs of microcrystalline cellulose (cellulose I and II), as well as the hydration of amorphous cellulose was studied using water sorption calorimetry, gravimetric water vapor sorption, nitrogen sorption, and X-ray powder diffraction. Amorphous cellulose was prepared by means of ball-milling of microcrystalline cellulose (MCC). Whereas X-ray data showed the untreated MCC to consist of cellulose I, the amorphous cellulose was found to recrystallize into cellulose II after contact with water or water vapor at relative humidities (RHs) above 90%. Sorption isotherms show an increase of water sorption in the sequence cellulose I < cellulose II < amorphous cellulose. The enthalpy of water sorption becomes more exothermic in the same sequence. The specific area of cellulose is dramatically higher when calculated from the water adsorption than when calculated from nitrogen adsorption. A proposed mechanism of water sorption by MCC implies the adsorption of water molecules at solid-solid interfaces, i.e., between neighboring microfibrils, which explains the observed difference between water and nitrogen. The Brunauer-EmmettTeller (BET) model is therefore not appropriate for the description of the hydration of cellulose. Rather, the Langmuir model represents a more accurate description of water sorption by MCC at low RH. At higher RH, the water adsorption competes with capillary condensation. The thickness of microfibrils, as calculated using the fitting of the sorption isotherm of MCC with the Langmuir equation, is about 4 nm. This value compares favorably with literature data.

Introduction Cellulose is the most abundant organic substance on earth. Wherever it is used, it is in contact with the most abundant solvent on earth - water. In most of the applications, cellulose is not in direct contact with liquid water, rather it is in contact with water vapor that can be described by different values of relative humidities (RHs). Despite of the huge amount of publications on the topic of cellulose-water interactions (see, for example, refs 1 and 2 and the references therein), the exact mechanisms of those interactions at different levels of RH are not fully understood. One of the reasons for that is the wide variety of cellulose materials and sources, e.g., wood cellulose, bacterial cellulose, microcrystalline cellulose (MCC), etc. Another reason is the seemingly weak coordination of research efforts between scientists working on the structural characterization of cellulose (e.g., X-ray, spectroscopy) and scientists studying thermodynamic properties of hydration of cellulose (sorption isotherm studies, calorimetry). Below we will briefly describe the structure of cellulose and then provide a review of approaches used in the interpretation of water sorption isotherms of cellulose. Cellulose is an unbranched polysaccharide composed of β-D-glucopyranose units linked by 1f4 glucosidic bonds.3 Because of a large number of polar hydrogen and oxygen atoms, cellulose forms * Corresponding author. Address: Faculty of Health and Society, Malmo¨ University, SE-205 06 Malmo¨, Sweden. Ph: +4640 6657946. Fax: +4640 6658100. E-mail: [email protected]. † Malmo ¨ University. ‡ AstraZeneca R&D Lund.

both inter- and intramolecular hydrogen bonds. Two intramolecular hydrogen bonds, OH-3‚‚‚O5 and OH-2‚‚‚O6, can bind neighboring glucose units and thus provide high stiffness of natural cellulose chains.1 Intermolecular hydrogen bonds are different in different polymorphs of cellulose. Cellulose I is a natural polymorph, which can be divided into IR and Iβ polymorphs. In cellulose I (which has parallel arrangement of the chains), the hydrogen bonds exist only between chains belonging to the same sheet.4 Cellulose II (which features antiparallel arrangement of chains) can be obtained by recrystallization of native cellulose. In this polymorph, the hydrogen bonds are found also between sheets, i.e., they form a threedimensional (3D) network.5 Interestingly, it was suggested that, between sheets of cellulose I, hydrogen bonds of the C-H‚‚‚O type may exist.6 In natural cellulose samples, the cellulose chains are arranged in microfibrils that have different diameters in the range of 2-20 nm depending on their origin.1 The typical values of the widths of the microfibrils from higher plants are 3-4 nm. In some publications, it was proposed that a microfibril is not fully crystalline and consists of crystalline elementary fibrils7 of smaller widths (about 3.5 nm),8 but usually it is assumed that a microfibril contains a single crystalline core.9 When the structure of cellulose from wood is discussed, many authors use the model proposed by Fengel.10 According to this model, 16 elementary fibrils that have a width of 3 nm each are arranged in 12 nm fibrils that, in turn, form 25 nm fibrils. The structure of cellulose along its microfibril axis is also a topic for discussions. According to X-ray diffraction studies,

10.1021/jp711554c CCC: $40.75 © 2008 American Chemical Society Published on Web 02/29/2008

Hydration of MCC and Milled Cellulose

J. Phys. Chem. B, Vol. 112, No. 12, 2008 3729

there are light and dark areas along the cellulose microfibril, which can be attributed to crystalline and amorphous regions, respectively.3 Another interpretation implies continuously crystalline microfibrils that possess slight curvature, which brings specific domains in and out of Bragg diffraction conditions, producing light and dark domains along the microfibril axis.3 For the present study (which deals with the hydration of MCC) the presence or absence of amorphous domains in the natural cellulose is not important. During the process of preparation of MCC, possible amorphous domains are dissolved by an acid, and the resulting structure consists of crystallites in any case. The surface chains of crystallites, however, cannot be considered as fully crystalline since they differ in conformation from the interior chains.11 Hence, even in the absence of amorphous domains, no cellulose samples can be considered as fully crystalline. To quantify crystallinity, the so-called crystallinity index (usually determined from X-ray or NMR data) is used. The crystallinity (or amorphicity) of cellulose samples can be changed using a mechanical or a chemical treatment. Ballmilling in the absence of water produces cellulose samples with decreased crystallinity.12-19 The particular decrease of crystallinity of a sample depends on the rotation speed and the milling time.14 In the presence of water, the ball-milling of cellulose I can produce cellulose II. According to Ago et al.,12 the most effective transformation to cellulose II occurs when 30 wt % of water is present in the cellulose during the milling process. By addition of water, the amorphous cellulose obtained by dry ball-milling can be recrystallized back to cellulose I20,21 or to cellulose II.21 Crystallization to cellulose I or cellulose II apparently depends on the degree of amorphization. Iyer et al.17 suggested that crystallization into cellulose II does not start until the amorphous content is as high as 75% or more. A method to produce low-crystallinity cellulose by treatment with ZnCl2 was also reported.2,22 Water sorption isotherms of various types of cellulose have been studied for many decades.2,18,23-31 The experimental sorption isotherms of cellulose are often treated using BrunauerEmmett-Teller (BET)32 or Guggenheim-Anderson-de Boer (GAB)33 equations that assume the formation of multilayers of water on the surface of cellulose (i.e., at the solid-vapor interface). Some authors use corrections for the crystalline content of cellulose27,30 when treating the data obtained using BET or GAB approaches. Fractal-based theories were also applied for the description of the sorption of water on cellulose.34,35 In this paper we present a sorption calorimetric study of MCC, ball-milled cellulose, and cellulose recrystallized after ballmilling. The method of sorption calorimetry36 used here allows measurements of water sorption isotherm and the enthalpy of hydration in one experiment, thus providing the complete thermodynamic description of the studied system. The water sorption isotherm of MCC is interpreted from the point of view of the adsorption of water at a solid-solid interface rather than a solid-vapor interface. The obtained results are in agreement with the data on the structure of cellulose.

where Pvap and Psorp are thermal powers registered in the vaporization and sorption chambers, respectively, and Hvap w is the molar enthalpy of evaporation of pure water. For accurate calculation of the partial molar enthalpy of mixing of water, the sorption calorimeter was calibrated using magnesium nitrate as a standard substance.39 Gravimetric Vapor Sorption (GVS). In addition to the measurements of a sorption isotherm using sorption calorimetry, the sorption isotherm of MCC was also measured by GVS. The water sorption was determined in the RH range of 0-90% in steps of 10% RH at 25 °C using a dynamic vapor sorption system (Surface Measurement Systems). The system consists of a microbalance that records the mass change of the sample when a gas flow (N2) of controlled humidity passes over the sample. X-ray Powder Diffraction (XRPD). Diffractograms were obtained using a Scintag XDS 2000 equipped with a Cu KR radiation source (λ ) 1.54 Å). The diffraction angle was varied between 2° and 35° (2θ) with a scan speed of 1°/min. Nitrogen Sorption. The specific surface area was measured by an adsorption method with nitrogen gas applying the BET equation (multipoint method) on the results. The instrument used was a Tristar 3000 (Micromeritics). The samples were degassed for 3 h at 60 °C before analysis. The analyzed relative pressures (P/P0) were in the range of 0.05-0.985. Cellulose. MCCs Avicel PH-101 and PH-301 were received from FMC Biopolymers. According to the manufacturer, the degree of polymerization is 210-270 for PH-101 and 140180 for PH-301. The bulk densities are 0.26-0.31 and 0.340.45 g/cm3, respectively. Milled (amorphous) cellulose was prepared according to the following procedure. A Pulverizette 6 planetary ball mill (Fritsch, Germany) was loaded with 3 g of cellulose Avicel PH101. Grinding was performed in 10 min intervals, with 10 min pauses in between, for a total of 720 min. An 80 mL Syalon jar, together with five syalon balls (20 mm diameter), was used, and the rotation speed was 400 rpm. The temperature in the room was 23 °C, and, by using alternating milling and pausing periods, the sample temperature never exceeded 40 °C during the experiment. In order to produce recrystallized cellulose, the milled sample was mixed with water, and the mixture was kept at room temperature for at least 12 h. Then water was evaporated, and the remaining powder was dried in vacuum at room temperature.

Materials and Methods

Results

Sorption Calorimetry. Sorption calorimetric experiments were conducted at 25 °C in a 28 mm two-chamber sorption calorimetric cell inserted in a double-twin microcalorimeter.36,37 The samples under study were placed in the upper chamber, while pure water was injected in the lower chamber. In a sorption experiment, water evaporates from the lower chamber, diffuses through the tube that connects the two chambers, and

The water sorption isotherm of MCC PH-101 is shown in Figure 1. The curve represents the sorption calorimetric results, whereas the squares denote the GVS results. The data obtained by the two methods are in a good agreement. The good agreement between calorimetric and GVS results indicates that the results obtained using sorption calorimetry (a dynamic method) can be considered as equilibrium results. The sorption

is adsorbed by the sample in the upper chamber. The thermal powers released in the two chambers are monitored simultaneously. The activity of water aw in the sorption experiments was calculated from the thermal power of the vaporization of water in the lower chamber as described in ref 38. The partial molar enthalpy of mixing of water was calculated using the following equation: vap vap H mix w ) Hw + Hw

Psorp Pvap

(1)

3730 J. Phys. Chem. B, Vol. 112, No. 12, 2008

Figure 1. Sorption isotherm of MCC Avicel PH-101. The curve represents sorption calorimetric data, whereas the squares represent GVS data.

Figure 2. Sorption isotherms of MCC Avicel PH-101 (cellulose I; thick dashed curve), recrystallized cellulose (cellulose II, thin dashed curve), and amorphous (ball-milled) cellulose (thin solid curve).

isotherm of MCC PH-301 has the same shape as that of PH101, although the amount of water absorbed by PH-301 is slightly lower (data not shown). If it is not stated otherwise, the data on MCC discussed below were obtained using PH101. Figure 2 shows sorption isotherms of three different materials, namely MCC (Avicel PH-101), amorphous cellulose prepared by ball-milling of Avicel PH-101, and the ball-milled material recrystallized by an addition of water. The highest water uptake is observed for the milled MCC, and the lowest is observed for the original MCC. The sorption isotherm of the recrystallized material lies between the sorption isotherms of the original MCC and the milled material. At high RHs, the sorption isotherms of the milled MCC and the milled recrystallized cellulose converge, which indicates that the milled MCC actually recrystallizes during the course of the sorption experiment. The same conclusion follows from the GVS experiment with milled MCC, where the mass of water taken up by the sample at a fixed RH level of 90% first increases and then decreases (Figure 3). This type of behavior is often observed when a material sheds adsorbed water during a recrystallization process. The partial molar enthalpies of mixing of water Hm w with the three types of cellulose materials are shown in Figure 4. At

Kocherbitov et al.

Figure 3. GVS data on amorphous (ball-milled) cellulose. The upper curve shows water-to-cellulose mass ratio, whereas the lower curve represents the RH as a function of time. The decrease of water content at 90% RH is due to recrystallization of milled cellulose, and the concomitant shedding of adsorbed water.

Figure 4. The partial molar enthalpy of mixing of water as a function of water-to-cellulose mass ratio for MCC Avicel PH-101 (cellulose I; thick dashed curve), recrystallized cellulose (cellulose II; thin dashed curve), and amorphous (ball milled) cellulose (thin solid curve).

low water contents, the enthalpies observed in all three samples are about -18 kJ per mole of water. At higher water contents, the hydration enthalpies of amorphous and recrystallized materials are more exothermic than that of original MCC. The enthalpy curves of amorphous and recrystallized cellulose converge approximately at the same water content observed for the sorption isotherms of the same materials (Figures 2 and 4). The high level of noise seen at high water contents on the enthalpy curves is typical for sorption calorimetric experiments and is due to low calorimetric signals at the end of the experiments. However, recrystallization may contribute to the particularly high noise level observed in the case of amorphous cellulose at high water contents. XRPD results (Figure 5) show that the diffraction pattern of MCC is consistent with the structure of cellulose I.40,41 The diffractogram of the milled MCC consists of one broad peak that indicates that the material is in the amorphous state. The diffraction pattern of the milled recrystallized MCC is consistent with the structure of cellulose II.42 The observation that the amorphous cellulose recrystallizes to cellulose II, rather than to cellulose I, strongly indicates that the milling process effectively removes all crystallites of cellulose I. Hence, we find strong support for the idea that the material can be described as being entirely amorphous.

Hydration of MCC and Milled Cellulose

J. Phys. Chem. B, Vol. 112, No. 12, 2008 3731

Figure 6. The proposed scheme of hydration of MCC. The squares represent microfibrils, the circles depict water molecules.

calculate the surface areas A of the samples, according to the following equation:

A)

rw‚NA‚amol Mw

(2)

where Mw is the molar mass of water, rw is the water-to-cellulose mass ratio, NA is Avogadro’s number, and amol is the area of one water molecule, here taken as 0.105 nm2. It is noteworthy that the surface areas of the samples, as calculated from water sorption data, are at least 2 orders of magnitude higher than the surface areas measured by nitrogen adsorption. Discussion

Figure 5. XRPD data for the three types of cellulose.

The results of nitrogen adsorption studies are shown in Table 1. In the same table, the results of fitting of the water sorption isotherms with the Langmuir and BET32 equations are presented. The parts of the water sorption isotherms that correspond to water activity in the range of 0.05-0.4 were used for fitting. The calculated water-to-cellulose mass ratios corresponding to monolayer coverage are several times higher in the case of Langmuir model, as compared to the BET model. The mass ratios corresponding to monolayer coverage were used to

Microcrystalline Cellulose. The very large difference between the surface areas of MCC calculated using water and nitrogen sorption indicates a fundamental difference in the adsorption mechanism between the two cases. The difference may be rationalized if we assume that the adsorption of nitrogen occurs on the surface of cellulose particles (which have sizes in the micron range), while the adsorption of water also occurs inside the particles. In the latter case, adsorption would thus also occur on the interface between structural units of MCC that have characteristic sizes in the nanometer range. Since it is known that the main structural units of cellulose are microfibrils, which typically have sizes of 4 nm,1,3 it is reasonable to assume that adsorption of water occurs on the interface between the microfibrils (as shown in Figure 6). In most of the studies of water sorption on cellulose published before,26,27,29,30 the experimental sorption isotherms were treated using the BET32 and/or GAB33 models. These models assume the formation of multilayers of adsorbate on a surface, even before the completion of a monolayer (Figure 7.a). These models, especially the BET model, are widely used in the description of processes of the adsorption of gases on solid surfaces and for calculations of surface areas. Nevertheless, on the basis of our data, we suggest that these models are not suitable for description of the processes of adsorption of water on solid-solid interface. The formation of multilayers on two adjacent solid surfaces before the monolayer is almost complete would lead to large separations of the solid surfaces with extensive empty space between them. Since this scenario violates the principle of horror Vacui, it is clearly not realistic. We

3732 J. Phys. Chem. B, Vol. 112, No. 12, 2008

Kocherbitov et al.

TABLE 1: Data on the Sorption of Water and Nitrogen on Different Types of Cellulose N2 sorption sample

structure

area, m2/g, BET

MCC PH-101 MCC PH-301 milled milled recrystallized

cellulose I cellulose I amorphous cellulose II

1.04 0.55 0.88 0.2

H2O sorption

propose instead that the water molecules tend to be in contact with both surfaces until the monolayer is almost complete (Figure 7.b). This situation would prevent the formation of voids in the structure, and would also be expected to provide a much lower energy of the system, since water molecules may interact with both surfaces simultaneously. On the other hand, formation of the second layer probably starts at lower concentrations than those corresponding to the monolayer formation because the completion of the perfect monolayer would lead to a substantial decrease in entropy. Since the formation of multilayers is less pronounced at solid-solid interfaces, compared to solid-gas interfaces, the BET model would not be expected to apply for the description of water sorption by MCC. Instead, we propose that a simpler model that assumes only the formation of a monolayer (the Langmuir model) should be used. However, it has to be kept in mind that the model can only be applied to the part of the sorption isotherm that corresponds to water contents below the monolayer coverage. The Langmuir sorption isotherm can be written in the following form:

rw ) rmax w

Caw 1 + Caw

(3)

where rmax w is the water-to-cellulose mass ratio that corresponds to the monolayer coverage, and C is a constant related to the energy of adsorption. The monolayer coverage determined from eq 3 (or, analogously, from the BET equation) can be interpreted using two different approaches. The first approach is based on the assumption that water molecules adsorb only at specific sites, separated from each other. In this case, the amount of adsorbed water is a function of the surface density of the sites that interact strongly with water. Recently, we used this approach to calculate the surface density of silanol groups in calcined mesoporous silica.43 Another approach, used here, is based on an assumption of an even distribution of water molecules on the solid surface. In this case, the amount of adsorbed water is a function of a cross-section area of a water molecule. Previously, Khan and Pilpel24 used the first approach for interpretation of data on water vapor sorption on MCC. They assumed that the first water molecules bind only to the 6-hydroxyl group because it is “the most exposed hydroxyl group”. A similar approach was later used by Nilsson and Strømme,44 who reiterated the claim that it is energetically favorable for the water molecules to bind to the 6-hydroxyl groups. Here we use the second approach, and thus assume that water molecules do not adsorb only at specific sites, but can cover the whole solid-solid interface between cellulose microfibrils in a continuous fashion. The surface of MCC is an array of rather disordered OH groups that differ in conformations from the interior groups.11 The accessibility of those interfacial OH groups can be dependent on many factors, including hydrogen bonding between microfibrils. Regarding the energetic properties of OH groups, an ab initio study of the hydrogen bonding

mass ratio, Langmuir

mass ratio, BET

area, m2/g, Langmuir

area, m2/g, BET

0.0950 0.0819 0.2677 0.1943

0.0352 0.0308 0.0751 0.0636

334 288 940 682

124 108 264 223

Figure 7. Schematic representation of the adsorption of water on solid-gas interface (a) and on solid-solid interface (b).

between glucose and water45 indicates that the 6-hydroxyl group is not the most energetically favorable position for water interaction in a glucose ring. Assuming that the area of one water molecule is 0.105 nm2, we calculated the interfacial area A of cellulose PH-101 as 334 m2/g by means of eq 2. One should keep in mind that it represents the total solid-solid interfacial area. The area of the surface of microfibrils is twice larger (see Figure 7):

Asurface ) 2‚Ainterface ) 2‚ALangmuir

(4)

Furthermore, our approach allows for calculation of microfibrils size from the water sorption isotherms. Assuming a square cross-section of microfibrils with thickness w, the surface area can be expressed in the following way:

Asurface )

area 4wL 4 ) ) m Vdc wdc

(5)

where L is the length of microfibrils, and dc is the density of crystalline cellulose (1.6 × 106 g/cm3).46 Therefore, the thickness of microfibrils can be calculated using the data obtained from the fitting of the water sorption isotherm with the Langmuir equation:

w)

2 4 ) Asurfacedc ALangmuirdc

(6)

The microfibril size calculated using eq 6 is 3.7 nm for Avicel PH-101 and 4.3 nm for Avicel PH-301. This result is in good agreement with the commonly accepted value of about 4 nm,1,3 which lends further credibility to the approach used here.

Hydration of MCC and Milled Cellulose

J. Phys. Chem. B, Vol. 112, No. 12, 2008 3733

The curve of the partial molar enthalpy of mixing of water with MCC (the top curve in Figure 4) has a break at a water content of about 0.055 g/g. This value is higher than the BET monolayer coverage value and lower than the Langmuir monolayer coverage value (see Table 1). At the same water content, a smaller break on the water sorption isotherm of MCC can be seen after careful examination of the data in Figure 2. The water contents after this break are higher than those obtained by extrapolation of the Langmuir sorption isotherm. This indicates that additional processes contribute to the uptake of water. We suggest two different processes that may play a part. First, the formation of a second water layer between the microfibrils may start at this point. As was mentioned above, the formation of a perfect monolayer would lead to a dramatic decrease of entropy. Therefore the formation of the second layer may start at water contents below the value of 0.095 predicted by the Langmuir model. Second, a capillary condensation of water in cavities (pores) present in MCC may take place. Assuming that capillary condensation starts at aw ) 0.5 (which roughly corresponds to the break on the curves) one can calculate the size of the cavities using the Kelvin-Cohan equation:47

r-t)-

2γ cos θ‚Vm RT ln aw

(7)

In this equation, r is the radius, t is the thickness of the preadsorbed water layer, γ is the surface tension of water, Vm is the molar volume of water, and the contact angle θ is zero. The diameter of the pore (minus the thickness of the preadsorbed water layer) is then 3 nm. This value indicates that the smallest pores in MCC may be the missing microfibrils in the MCC structure (as shown in Figure 6). Since there is no pronounced capillary condensation regime with constant water activity on the sorption isotherm of MCC, the distribution of pore sizes may be rather broad. If present, the nanometer-scale pores in MCC must be assumed to be closed from both ends. This follows from the fact that nitrogen adsorption experiments show a very low surface area of MCC (Table 1). Another phenomenon that suggests the presence of closed pores in MCC was observed during drying of the cellulose samples in vacuum at room temperature. The powder that was used for vacuum drying was placed in an open container which was placed in a vacuum pistol connected to a vacuum pump. When the pressure in the vacuum pistol was substantially reduced, the powder started to “boil”. This process was so vigorous, that most of the powder was displaced from the container. We suggest that the “boiling” was caused by the release of gases adsorbed in the closed pores of MCC. Neither in the case of milled, nor in the case of milled recrystallized cellulose was this phenomenon ever observed. Amorphous Cellulose. As would be expected, the amorphous (ball-milled) cellulose takes up more water than MCC does (Figure 2). The water molecules can penetrate between lessordered cellulose chains of amorphous cellulose, which means that a larger number of sorption sites are available, as compared to MCC. However, the enthalpy of water-cellulose interactions is the same for the amorphous material and MCC at low water contents (Figure 4). This indicates that the nature of interactions of cellulose chains with water molecules does not change upon milling. At higher water contents, Hm w is more exothermic for the milled material than for MCC but this is probably due to a lower number of sorption cites in MCC. Again in agreement with expectation, the surface area calculated from water sorption

experiments is much higher in amorphous cellulose than in MCC (Table 1). However, we do not propose a structural interpretation of those values, since it is not clear whether microfibril-like structures are present in the milled material. At RHs above 90%, the milled cellulose recrystallizes to cellulose II, which leads to a decrease in the amount of absorbed water (Figure 3). The reason for this shedding of adsorbed water is the lowering of the number of sorption sites during the recrystallization process. More specifically, disordered cellulose chains that were available as water sorption sites in the amorphous material are arranged in a crystal structure after the recrystallization, and are thus no longer accessible. Recrystallized Cellulose. The sorption isotherm of the cellulose recrystallized after the boll milling lies between sorption isotherms of MCC and milled cellulose. This implies that the recrystallized cellulose is less crystalline than the initial MCC. The X-ray diffraction pattern of the recrystallized material also looks less ordered compared to that of the initial cellulose I. The crystallinity index of MCC calculated using intensities of X-ray diffraction peaks48 of cellulose I is 83.6%, while the crystallinity index of the recrystallized material calculated using an analogous method for cellulose II49 is 69.5%. This might be interpreted as a decrease of the size of microfibrils compared to MCC. However, the exact structure of the recrystallized sample is not known. In contrast to MCC, where the amorphous part is mostly the surface chains of microfibrils, in the recrystallized cellulose one can expect to find some bulk amorphous areas that were not able to crystallize because of entanglements of the cellulose chains of the ball-milled material. Those amorphous parts can absorb water as bulk rather than as surfaces (like in the case of MCC). The fact that at low water contents there are no breaks in the enthalpy curves of both amorphous cellulose and recrystallized cellulose (see Figure 4) suggests that the mechanism of hydration of the recrystallized cellulose is closer to that of amorphous cellulose rather than MCC. Conclusions The hydration of three types of cellulosesMCC, amorphous cellulose prepared by ball-milling, and recrystallized celluloses was studied using sorption calorimetry and other supplementary techniques. The main conclusions from the study are as follows: • MCC (cellulose I) can be turned into amorphous cellulose by means of ball-milling. The amorphous material recrystallizes into cellulose II at RH > 90%. The recrystallization process leads to a substantial loss of adsorbed water. • Sorption isotherms show an increase of water sorption according to the sequence original MCC < recrystallized cellulose < milled cellulose. • The enthalpy of hydration is -18 kJ/mol at zero water content for all three types of cellulose, while, at higher water contents, the enthalpy values are different. • The Langmuir model describes the water sorption by MCC more adequately than the BET model. • The thickness of microfibrils of MCC calculated from the water sorption isotherm using the Langmuir model and the assumption of a homogeneous distribution of water molecules between microfibrils is about 4 nm. • Adsorption data obtained from sorption calorimetry compares very favorably with those obtained by GVS measurements. However, the continuous character of the data provided by sorption calorimetry greatly facilitates elucidation of the finer details in the adsorption process, e.g., break points in the isotherm.

3734 J. Phys. Chem. B, Vol. 112, No. 12, 2008 • The break point observed in the adsorption isotherms obtained by sorption calorimetry may correspond to the onset of capillary condensation in closed pores. The calculated size of these pores suggest that they consist of missing microfibrils in the cellulose structure. Acknowledgment. This study was supported by a research grant from the Knowledge Foundation (KK stiftelsen, BiofilmsResearch Centre for Biointerfaces). The authors would like to thank Dr. Thomas Larsson and Dr. Lars-Erik Briggner at AstraZeneca R&D Lund for help in the milling experiments and for fruitful discussions, respectively. References and Notes (1) Dumitriu, S. Polysaccharides. Structural DiVersity and Functional Versatility; Marcel Dekker: New York, 2005. (2) Mihranyan, A.; Llagostera, A. P.; Karmhag, R.; Strømme, M.; Ek, R. Int. J. Pharm. 2004, 269, 433. (3) O’Sullivan, A. C. Cellulose 1997, 4, 173. (4) Nishiyama, Y.; Sugiyama, J.; Chanzy, H.; Langan, P. J. Am. Chem. Soc. 2003, 125, 14300. (5) Langan, P.; Nishiyama, Y.; Chanzy, H. J. Am. Chem. Soc. 1999, 121, 9940. (6) Jarvis, M. Nature 2003, 426, 611. (7) Frey-Wyssling, A. Science 1954, 119, 80. (8) Gardner, K. H.; Blackwell, J. J. Ultrastruct. Res. 1971, 36, 725. (9) Nieduszynski, I.; Preston, R. D. Nature 1970, 225, 273. (10) Fengel, D. J. Polym. Sci., Part C 1971, 36, 383. (11) Vietor, R. J.; Newman, R. H.; Ha, M. A.; Apperley, D. C.; Jarvis, M. C. Plant J. 2002, 30, 721. (12) Ago, M.; Endo, T.; Hirotsu, T. Cellulose 2004, 11, 163. (13) Howsmon, J. A.; Marchessault, R. H. J. Appl. Polym. Sci. 1959, 1, 313. (14) Ouajai, S.; Shanks, R. A. Cellulose 2006, 13, 31. (15) Maier, G.; Zipper, P.; Stubicar, M.; Schurz, J. Cell. Chem. Technol. 2005, 39, 167. (16) Iyer, P. B.; Sreenivasan, S.; Chidambareswaran, P. K.; Patil, N. B. Text. Res. J. 1986, 56, 509. (17) Iyer, P. B.; Sreenivasan, S.; Chidambareswaran, P. K.; Patil, N. B. Text. Res. J. 1984, 54, 732. (18) Stubberud, L.; Arwidsson, H. G.; Larsson, A.; Graffner, C. Int. J. Pharm. 1996, 134, 79. (19) Nakai, Y.; Fukuoka, E.; Nakajima, S.; Hasegava, J. Chem. Pharm. Bull. 1977, 25, 96.

Kocherbitov et al. (20) Wormald, P.; Wickholm, K.; Larsson, P. T.; Iversen, T. Cellulose 1996, 3, 141. (21) Liang, X. H.; Gu, L. Z.; Ding, E. Y. Wood Sci. Technol. 1993, 27, 461. (22) Heidarian, M.; Mihranyan, A.; Stromme, M.; Ek, R. Int. J. Pharm. 2006, 323, 139. (23) Sheppard, S. E.; Newsome, P. T. J. Phys. Chem. 1929, 33, 1817. (24) Khan, F.; Pilpel, N. Powder Technol. 1987, 50, 237. (25) Zografi, G.; Kontny, M. J. Pharm. Res. 1986, 3, 187. (26) Roskar, R.; Kmetec, V. Chem. Pharm. Bull. 2005, 53, 662. (27) Zografi, G.; Kontny, M. J.; Yang, A. Y. S.; Brenner, G. S. Int. J. Pharm. 1984, 18, 99. (28) Markova, N.; Sparr, E.; Wadso¨, L. Thermochim. Acta 2001, 374, 93. (29) Chirkova, J.; Andersons, B.; Andersone, I. J. Colloid Interface Sci. 2004, 276, 284. (30) Agrawal, A. M.; Manek, R. V.; Kolling, W. M.; Neau, S. H. J. Pharm. Sci. 2004, 93, 1766. (31) Jowitt, R.; Wagstaffe, P. J. The Certification of the Water Content of Microcrystalline Cellulose (MCC) at 10 Water ActiVities; Commission of the European Communities, Community Bureau of Reference, CRM 30, 1989. (32) Brunauer, S.; Emmet, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309. (33) de Boer, J. H. The Dynamic Character of Adsorption, 2nd ed.; Clarendon Press: Oxford, 1968. (34) Strømme, M.; Mihranyan, A.; Ek, R.; Niklasson, G. A. J. Phys. Chem. B 2003, 107, 14378. (35) Mihranyan, A.; Stromme, M. Chem. Phys. Lett. 2004, 393, 389. (36) Wadso¨, L.; Markova, N. ReV. Sci. Instrum. 2002, 73, 2743. (37) Wadso¨, I.; Wadso¨, L. Thermochim. Acta 1996, 271, 179. (38) Kocherbitov, V. Thermochim. Acta 2004, 414, 43. (39) Kocherbitov, V. Thermochim. Acta 2004, 421, 105. (40) Chen, R. Y.; Jakes, K. A.; Foreman, D. W. J. Appl. Polym. Sci. 2004, 93, 2019. (41) Bates, S.; Zografi, G.; Engers, D.; Morris, K.; Crowley, K.; Newman, A. Pharm. Res. 2006, 23, 2333. (42) Mansikkamaki, P.; Lahtinen, M.; Rissanen, K. Cellulose 2005, 12, 233. (43) Kocherbitov, V.; Alfredsson, V. J. Phys. Chem. C 2007, 111, 12906. (44) Nilsson, M.; Strømme, M. J. Phys. Chem. B 2005, 109, 5450. (45) Momany, F. A.; Appell, M.; Strati, G.; Willett, J. L. Carbohydr. Res. 2004, 339, 553. (46) Sun, C. Q. J. Pharm. Sci. 2005, 94, 2132. (47) Cohan, L. H. J. Am. Chem. Soc. 1938, 60, 433. (48) Segal, L.; Creely, J. J.; Martin, A. E., Jr.; Conrad, C. M. Text. Res. J. 1959, 29, 786. (49) Ingersoll, H. G. J. Appl. Phys. 1946, 17, 924.