(IUCr) Assessment of the X-ray diffraction-absorption method for quantitative analysis of largely amorphous pharmaceutical composites

Volume 36
Part 1
Pages 74-79
February 2003

Received 13 May 2002
Accepted 21 October 2002

Assessment of the X-ray diffraction-absorption method for quantitative analysis of largely amorphous pharmaceutical composites

P. Bergese,^a I. Colombo,^b D. Gervasoni^b and Laura E. Depero^a ^*

^aINSTM and Structural Chemistry Laboratory, University of Brescia, via Branze, 38-25123, Brescia, Italy, and ^bEurand International S.p.A., Physical Pharmacy, via Martin Luther King, 13-20060 Pessano con Bornago, Milano, Italy
Correspondence e-mail: depero@bsing.ing.unibs.it

Determination of the residual weight fraction of a crystalline drug in a largely amorphous pharmaceutical composite is still a challenging question. None of the quantitative X-ray diffraction (QXRD) methods found in the literature is suitable for these inclusion systems. The composite's diffraction patterns present a structured amorphous halo (arising from the amorphous matrix and drug molecular clusters) in which the crystalline drug peaks rise up. Moreover, the matrix traps a non-negligible quantity of water (which cannot be directly detected by X-ray diffraction) and the crystal structure of the drug may be unknown. In this work, a development of the QXRD analysis based on the diffraction-absorption technique is presented. The method is standardless, avoids the interpretation of the amorphous halo and the knowledge of the crystal structures of the phases, and takes into account the absorbed water. Results are in excellent agreement with those obtained by differential scanning calorimetry (DSC). The general features of the technique open its application to other classes of largely amorphous composite materials, like glass systems generated in the stabilization/solidification of toxic waste.

Keywords: X-ray diffraction-absorption method; quantitative phase analysis; amorphous pharmaceutical composites; drug composites.

1. Introduction

Dispersion into cross-linked polyvinylpirrolidone (crospovidone or PVPCL) is a widely employed method for enhancing the dissolution rate (i.e. the bio-availability) of poorly water-soluble drugs (Serajuddin, 1999), since the three-dimensional polymeric network physically constrains the formation of highly active amorphous and/or nanocrystalline drug phases (Chiou & Riegelman, 1971; Takayama et al., 1982). In past years, several techniques have been developed for achieving this process (Carli et al., 1986; Colombo & Pallado, 2001; Lovrecich, 1995). Nowadays, the state of the art is still evolving and pharmaceutical companies are expending much effort to overcome the weaknesses of the existing technologies and to look for more effective, easy and environmentally safe processes.

The main research goal is to design a process that allows the full transformation of the drug from microcrystals to molecular clusters (the most reactive phase), but the processed composites show a more complex microstructure. They consist of crospovidone microparticles in which both nanocrystals and molecular clusters of drug are embedded (Carli & Garbassi, 1985) and a non-negligible fraction of water is adsorbed. Moreover, depending on the materials, loading technique and manufacturing conditions, the grain's surface can be crowned by drug microcrystals, the size distribution of the dispersed drug particles can be sharp or broadened, and the matrix may stabilize different crystalline phases of the drug (Bergese et al., 2001). Full microstructural characterization is a necessary step for understanding the release properties of the composites and developing the production process. In particular, the determination of the residual weight fraction of a crystalline drug, which is the basic criteria for evaluating the effectiveness of the dispersion technique, is one of the more complex aspects of the matter and remains a problem without an adequate solution (Vippagunta et al., 2001).

In the pharmaceutical field, quantitative analysis is traditionally performed by differential scanning calorimetry (DSC) and X-ray diffraction (QXRD) (Carli et al., 1986; Theeuwes et al., 1974; Suryanarayanan, 1995). When a compound melt with decomposition and/or solid-phase interactions occurs during heating, the DSC technique cannot be applied. QXRD, which relates the intensity of the coherently diffracted X-rays to the phase abundance in the specimen, overcomes these problems. However, none of the QXRD methods found in the literature seems to be suitable for systems made up of drug embedded into an amorphous polymeric matrix, since the diffraction patterns present a structured halo (arising from the matrix and the drug molecular clusters) in which the crystalline drug peaks rise up (Fig. 1).

Figure 1
X-ray diffraction pattern of the griseofulvin/crospovidone composite G05; the ICDD reference pattern of griseofulvin (DiMarco, 1989

[DiMarco, J. (1989). The Polytechnic Institute, Brooklyn, New York, USA. JCPDS-ICDD database, card No. 40-1937.]

) is also presented.

The classic QXRD methodologies applied to pharmaceuticals (Suryanarayanan, 1995) do not take into account the coexistence of crystalline and amorphous drug, and/or microabsorption, the effect of which, depending on the difference in absorption of the X-rays between the diffracting phase and specimen, may be important in composite materials (Zevin & Kimmel, 1995a $[Zevin, L. S. & Kimmel, G. (1995a). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 35-45. Berlin: Springer.]$ ). The approaches developed for evaluating the crystallinity of polymers (Zevin & Kimmel, 1995b $[Zevin, L. S. & Kimmel, G. (1995b). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 211-217. Berlin: Springer.]$ ), successfully extended to partially amorphous pharmaceuticals (Surana & Suryanarayanan, 2000), cannot be used for composites, since the drug and matrix have different chemical compositions.

QXRD based on Rietveld theory (Rietveld, 1969; Bish & Howard, 1988) has recently been applied to crystalline pharmaceuticals (Iyengar et al., 2001; Yamamura & Momose, 2001) but the presence of a structured amorphous halo seriously complicates the analysis. Le Bail (1995) and Lutterotti (1998) proposed the application of Rietveld refinement by considering the amorphous phase as a nanocrystalline phase, i.e. a phase with a very short coherence. However, to build a structural model allowing for amorphous phases of both drug and polymer (in which drug and water are trapped) seems a difficult task. Another method (Orlhac et al., 2001) to evaluate the crystalline phase in partially amorphous composites is based on performing Rietveld refinement of the diffraction pattern from which the amorphous and background contributions have been subtracted (the sample has to be doped by an analytical standard). Nevertheless, all of the above enhancements of the Rietveld method are limited by the need for crystallographic data of the phases (totally unknown for some drugs), and for complementary information about microabsorption (which must be experimentally evaluated).

In this work, we present a development of the QXRD analysis based on the diffraction-absorption method (Zevin & Kimmel, 1995c $[Zevin, L. S. & Kimmel, G. (1995c). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 104-112. Berlin: Springer.]$ ). In this approach, the intensity scale factor and the microabsorption parameter are obtained by a calibration procedure based on the diffraction data of different mixtures of pure crystalline drug and polymer.

2. Theoretical basis

In the hypothesis of a powder sample with randomly oriented grains measured in Bragg-Brentano geometry, the basic equation which relates the integrated intensity of a phase diffraction peak to the phase abundance in the specimen is

$[I_{ij} = {{K_e K_{ij} c_j } / {\rho _j \mu^*}},\eqno (1)]$

where c_j is the weight fraction of crystalline phase j; K_e is a constant for a particular experimental system; K_ij is a constant for each diffraction peak i from the crystal structure of phase j; [rho] _j is the density of phase j; and [mu] * is the mass absorption coefficient of the specimen (Alexander & Klug, 1948). By considering the sum of all the diffraction peaks i, belonging to phase j, we obtain

$[\textstyle\sum\limits_i I_{ij} = \Big(K_e \textstyle\sum\limits_i K_{ij} c_j \Big)\big/ \rho _j \mu^*.\eqno (2)]$

Applying equation (2) to a pure crystalline phase j (c_j = 1) we have

$[\textstyle\sum\limits_i ( I_{ij} )_0 = \Big(K_e \textstyle\sum\limits_i K_{ij} \Big)\big/ \rho _j \mu _j^* .\eqno (3)]$

The diffraction-absorption method is based on the calculation of the crystalline phase abundance c_j as obtained by combining equations (2) and (3):

$[c_j = \Big[\textstyle\sum\limits_i I_{ij} \big/ \textstyle\sum\limits_i ( I_{ij} )_0 \Big]\big(\mu^* / \mu _j^*\big).\eqno (4)]$

Equation (4) must be corrected by a semi-empirical factor [alpha] , which takes into account microabsorption effects, thus becoming

$[c_j = \Big[\textstyle\sum\limits_i I_{ij} \big/ \textstyle\sum\limits_i ( I_{ij} )_0 \Big]\big( \mu^* / \mu _j^* \big)^\alpha . \eqno (5)]$

The factor [alpha] in equation (5) depends on the grain size distribution of the sample as well as on the difference in absorption coefficients between the analyte and the matrix.

Drug/crospovidone composites are made up of the polymeric amorphous matrix in which a known amount of drug is dispersed. The drug is present in amorphous and crystalline phases, the proportions of which have to be determined. Starting from equation (5), we will develop a model suitable for calculating the amount of crystalline drug in these composites.

We define the total drug weight fraction (also called drug content) [varphi] _D, the crystalline drug weight fraction c_D, and the amorphous drug weight fraction a_D, so that

$[\varphi_{D} = a_{D} + c_{D} .\eqno (6)]$

We also introduce the mass balance equation

$[\varphi_{D} + \varphi_{P} + \varphi_{W} = 1 , \eqno (7)]$

where [varphi] _P and _W are respectively the polymer and the adsorbed water weight fractions. Equations (6) and (7) give the working expression of the mass balance equation:

$[a_{D} + c_{D} + \varphi_{P} + \varphi_{W} = 1 .\eqno (8)]$

It must be noted that the X-ray technique does not allow one to evaluate the adsorbed water in the specimen, but if [varphi] _W is higher than 3% its contribution to the overall mass absorption coefficient ( [mu] *) cannot be neglected (Suryanarayanan & Herman, 1991). [mu] * is independent of the material's physical state and structure, since it is the weighted sum of the mass absorption coefficients of the atoms of the phases. Thus here it is given by

$[\mu^{*} = (a_{D} + c_{D})\mu_{D}^{*} + \varphi_{P}\mu_{P}^{*} + \varphi_{W}\mu_{W}^{*}\eqno (9)]$

where $[\mu_{D}^*]$ , $[\mu_{P}^*]$ and $[\mu_{W}^*]$ are the mass absorption coefficients of drug, polymer and water, respectively. By merging equations (5) and (9), we obtain

$[c_D = \Big[\textstyle\sum\limits_i I_{iD} \big/ \textstyle\sum\limits_i ( I_{iD} )_0 \Big]\big( c_D + a_D + \gamma \varphi _P + \delta \varphi _W \big)^\alpha ,\eqno (10)]$

where [gamma] = $[\mu_{P}^*]$ / $[\mu_{D}^*]$ and [delta] = $[\mu_{W}^*]$ / $[\mu_{D}^*]$ . Equation (10) relates the amount of crystalline drug to the peak's integrated intensity.

The unknown constants [sum] _i(I_iD)₀ and [alpha] can be calculated from the calibration procedure based on the measurement of several mixtures of raw crystalline drug and crospovidone. In mixtures, no drug amorphous phase exists, thus a_D = 0 and equation (10) becomes

$[\textstyle\sum\limits_i I_{iD} = c_D \textstyle\sum\limits_i ( I_{iD} )_0 ( c_D + \gamma \varphi _P + \delta \varphi _W )^{ - \alpha } . \eqno (11)]$

In order to use equation (11) to refine [sum] _i(I_iD)₀ and [alpha] in a plot of [sum] _iI_iD versus c_D, [varphi] _P and _W must be given as functions of c_D. The water is completely absorbed by the crospovidone and the amount of absorbed water may change in each crospovidone sample; thus we define a constant f:

$[f = {{\varphi _W }/({\varphi _P + \varphi _W })}.\eqno (12)]$

From equation (8), with a_D = 0, and equation (12) we obtain

$[\varphi_{W} = f(1-c_{D}),\qquad \varphi_{P} = (1- f\,) (1-c_{D}).\eqno (13)]$

Substituting equation (13) into equation (11) gives

$[\textstyle\sum\limits_i I_{iD} = c_D \textstyle\sum\limits_i ( I_{iD} )_0 \{ c_D + (1 - c_D) [ \gamma (1 - f) + \delta f\,] \}^{ - \alpha }. \eqno (14)]$

Equation (14) is the equation used to extrapolate [sum] _i(I_iD)₀ and [alpha] . From a theoretical standpoint, [sum] _i(I_iD)₀ could also be calculated from the diffraction pattern of the pure crystalline drug. On the other hand, it is more appropriate to evaluate it by the calibration procedure, since microabsorption (which mainly depends on the difference in X-ray absorption between drug and polymer and on grain size distribution) may vary from pure crystalline drug and largely amorphous drug/polymer composites.

After the refinement of [sum] _i(I_iD)₀ and [alpha] , equation (10) was used to calculate the crystalline drug weight fraction of specimens containing amorphous drug (a_D > 0).

3. Materials and methods

3.1. Materials

Griseofulvin (an antibiotic substance, C₁₇H₁₇ClO₆) and [beta] -piroxicam (a common non-steroidal anti-inflammatory, C₁₅H₁₃N₃O₄S) have been chosen as model drugs for testing the method. Griseofulvin was supplied by Welding GmbH (Germany) and [beta] -piroxicam by Dinamite Dipharma (Italy). The micronized cross-linked polyvinylpirrolidone ([C₆H₉NO]_n), usually called crospovidone or PVPCLM, was obtained from BASF (Germany).

The mixtures of pure crystalline griseofulvin and crospovidone (griseofulvin weight fraction: 5.3, 10.6, 20.9, 30.9, 40.0 and 50.3%), and pure crystalline [beta] -piroxicam and crospovidone ( [beta] -piroxicam weight fraction: 3.6, 6.6, 14.4, 22.7, 35.4, 47.3 and 66.9%) were prepared using the materials as received. They were weighed with a precision balance (±10^-4 g) and then mechanically gently mixed for about 5 min.

The composites were prepared by several different solid dispersion techniques. The griseofulvin/crospovidone composite G03 was prepared by swelling, under continuous mixing, the polymer powders with an N,N-dimethylformamide (DMF) solution of the drug (solvent-swelling technique). After swelling, the loaded particles were left in the open air for 60 min, and dried in a vacuum oven at 373 K for 12 h. G04 was prepared in the same way as G03, but with an additional step consisting of exposure of the sample to DMF vapours for 24 h.

The griseofulvin/crospovidone composites G01, G02 and G05 were prepared with a process assisted by subcritical CO₂ (Colombo & Pallado, 2001).

The [beta] -piroxicam/crospovidone composite BP01 was prepared by the mechano-chemical activation process (Lovrecich, 1995).

The [beta] -piroxicam/crospovidone composite BP02 was prepared by the solvent-swelling technique, using a chloroform solution of the drug. After swelling, the loaded particles were left in the open air for 20 min and dried in a vacuum oven at 323 K for 3 h. Then they were exposed to chloroform vapours for 3 h and dried again in the vacuum oven at 303 K for 1 h. BP03 was prepared in the same way as BP02, but with an additional exposure of the sample to acetone vapours for 12 h.

3.2. X-ray diffraction

The powder samples for XRD measurements were prepared, after gentle grinding, in a back-loading sample holder. The material's stability under X-ray radiation flux was preliminarily checked. The XRD experiments were carried out on two diffractometers: a Bruker D8 ( [theta] / [theta] geometry and glancing-incidence experiments) and a Philips PW1050 ( [theta] /2 [theta] geometry); in both cases Cu K [alpha] radiation ( [lambda] = 1.541 Å), produced by an X-ray sealed tube powered by 40 kV × 40 mA, was used.

Glancing-incidence XRD (GIXRD) measurements were performed on physical mixtures and composites to evidence possible preferred orientation of the particles, since this effect would lead to erroneous quantitative analysis. Actually it would produce a significant deviation of the relative peak intensities in the diffraction pattern with respect to those obtained for randomly oriented samples (Bergese et al., 2001).

The measurements for QXRD analysis were performed on the Philips PW1050 diffractometer. Full patterns were collected in order to choose the most suitable peaks for the analysis. The 101 (2 [theta] = 10.789°) and 102 (2 [theta] = 13.274°) (DiMarco, 1989) reflections were chosen for griseofulvin, since they do not overlap with other peaks, they are favourably positioned for separation from the amorphous halo, and their intensities are also significant in mixtures with a griseofulvin content of 5% (Fig. 1). Thus for the quantitative analysis, the angular range was selected as 6.5-15.5° (2 [theta] ) (Fig. 2). The step was 0.04° (2 [theta] ) for a counting time of 4 s. Similarly for [beta] -piroxicam, the 011 (2 [theta] = 8.633°) and 020 (2 [theta] = 11.671°) (Canfield & Sundeen, 1989) peaks were chosen and the patterns collected from 2 [theta] = 6.5 to 2 [theta] = 15.5°; the step was 0.04° (2 [theta] ) for a counting time of 4 s. In order to improve the precision of the analysis and to evaluate the integrated intensity coefficient of variation (CV = [sigma] /x_best), at least three samples from each mixture and composite were prepared and measured.

Figure 2
Decomposition of the X-ray diffraction pattern (R_wp = 2.3%) of a sample obtained from the griseofulvin/crospovidone composite G05.

3.3. X-ray diffraction pattern profile decomposition

Drug/crospovidone composites show an XRD pattern characterized by a structured amorphous halo and crystalline drug peaks. The peaks were fitted by using a split pseudo-Voigt function, in order to take care of the peak asymmetry caused by the instrumental axial divergence (Jenkins & Snyder, 1996 $[Jenkins, R. & Snyder, R. L. (1996). Introduction to X-ray Powder Diffractometry, pp. 187-191. Chichester: Wiley Interscience.]$ ); the values of the form parameters of each peak function have been fixed to that refined for the pure crystalline drug. This procedure was necessary for achieving the pattern decomposition of composites with low residual crystallinity (i.e. with small peaks). A combination of a straight line and a Gaussian function was used for fitting the amorphous halo and the instrumental background (Fig. 2). This choice was made by considering the best analytical fitting procedure (in terms of lower R_wp), since the approach does not assign a physical meaning to the fitted amorphous halo. For the same reason, the 111 peak (2 [theta] = 14.641°) of griseofulvin and the 100 peak (2 [theta] = 12.490°) of [beta] -piroxicam were fitted, but their integrated intensities were not used in crystallinity calculations. The refinements (performed by the software TOPAS P; Bruker, 1998) gave an R_wp value lower than 2.8%.

3.4. Differential scanning calorimetry

Calorimetric measurements were performed with a Perkin Elmer Pyris 1 differential scanning calorimeter at 10 K min^-1 scanning rate. The pure drugs, in about 2 mg samples, and composite materials, in about 8 mg samples, were analysed in aluminium pans. A nitrogen flux of 20 cm³ min^-1 was employed during each analysis.

3.5. Thermogravimetry

Measurements of the water content of the solid samples ( [varphi] _W and f) were carried out on a Perkin Elmer TGA-Pyris 1. The samples (about 8 mg) were weighed into an aluminium pan and heated under a stream of nitrogen. The samples were heated from 293 to 473 K at 10 K min^-1.

4. Results and discussion

GIXRD patterns obtained at different fixed angles confirmed the absence of preferred orientation effects in the powder samples.

In order to evaluate [sum] _i(I_iD)₀ and [alpha] , calibration graphs using pure crystalline drug/crospovidone mixtures were prepared. In the graphs, the sum of the integrated intensities of the chosen peaks was plotted against the drug weight fraction of the mixture (c_D) (Fig. 3).

Figure 3
Top: calibration graph prepared with pure crystalline griseofulvin/crospovidone mixtures. [I_(101)D + I_(102)D] of each measured sample has been plotted against the griseofulvin weight fraction of the relative mixture (c_D). The solid line is the refined calibration curve [r² = 0.989672; F = 1533.19 (1, 16, 0.05)]. Bottom: calibration graph relative to [beta]

-piroxicam/crospovidone mixtures [r² = 0.99375; F = 3021.82 (1, 19, 0.05)].

The f value of the crospovidone samples, as measured by thermogravimetry, was 0.0429. The [gamma] and [delta] values were calculated by means of the chemical composition of the drug, crospovidone and water ( [gamma] = 0.320 and [delta] = 0.590 for griseofulvin, and [gamma] = 0.369 and [delta] = 0.680 for piroxicam). Finally [sum] _i(I_iD)₀ and [alpha] values were determined from equation (14) and a least-squares minimization procedure based on the Levemburg-Marquardt algorithm; fitting results and parameters are presented in Table 1.

Table 1
Fitting results and parameters of the calibration curves

Physical mixture	_i(I_iD)₀ ±	±	r²	F statistic
Griseofulvin/crospovidone	3360 ± 134	0.95 ± 0.07	0.989672	1533.19 (1, 16, 0.05)
-Piroxicam/crospovidone	6824 ± 172	0.65 ± 0.06	0.99375	3021.82 (1, 19, 0.05)

The refined values of [sum] _i(I_iD)₀ and [alpha] were substituted in equation (10) and the residual crystalline drug weight fractions (c_D) of the composites were calculated; the results are listed in Tables 2 and 3. The experimental errors have been calculated by the general propagation theory applied to [sum] _iI_iD and [sum] _i(I_iD)₀ and [alpha] standard deviations (SD). In Tables 1 and 2, the residual crystallinity is also reported, i.e. the percentage of crystalline drug weight fraction with respect to the drug content of the compound [RCR = 100(c_D/ [varphi] _D)].

Table 2
QXRD and DSC analysis of the griseofulvin/crospovidone composites: drug content of the composite ( [varphi] _D), water content of the composite (_W), mean integrated intensity sum ( [sum] _iI_iD), crystalline drug weight fraction (c_D), coefficient of variation (CV), residual crystallinity [RCR = 100(c_D/_D)]

	QXRD						DSC
Composite	_D	_W	_iI_iD ±	c_D ±	CV (%)	RCR ± (%)	RCR ± (%)
G01	0.143	0.109	978 ± 42	0.130 ± 0.008	6.1	90.9 ± 5.5	92.5 ± 2.8
G02	0.142	0.109	1020 ± 16	0.135 ± 0.003	2.2	95.3 ± 2.1	97.8 ± 2.9
G03	0.167	0.045	522 ± 28	0.069 ± 0.005	7.6	44.4 ± 3.1	38.4 ± 1.2
G04	0.157	0.095	262 ± 30	0.035 ± 0.006	16.2	22.5 ± 3.6	19.3 ± 0.6
G05	0.223	0.075	917 ± 87	0.134 ± 0.018	9.4	59.5 ± 5.6	60.6 ± 1.8

Table 3
QXRD and DSC analysis of the [beta] -piroxicam/crospovidone composites: drug content of the composite ( [varphi] _D), water content of the composite (_W), mean integrated intensity sum ( [sum] _iI_iD), crystalline drug weight fraction (c_D), coefficient of variation (CV), residual crystallinity [RCR = 100(c_D/_D)]

	QXRD						DSC
Composite	_D	_W	_iI_iD ±	c_D ±	CV (%)	RCR ± (%)	RCR ± (%)
BP01	0.141	0.106	1513 ± 69	0.140 ± 0.009	6.7	99.0 ± 6.7	95.0 ± 5.8
BP02	0.136	0.066	1050 ± 67	0.095 ± 0.008	8.2	69.8 ± 5.7	65.1 ± 1.2
BP03	0.167	0.067	1261 ± 55	0.114 ± 0.008	6.5	81.2 ± 5.4	77.1 ± 1.8

The crystalline drug weight fraction was also determined by DSC, following the method proposed by Theeuwes et al. (1974). The linearity of the relationship between drug melting enthalpy and crystalline drug weight fraction in the physical mixtures was checked in the range 0.1-1.

QXRD and DSC data are in excellent agreement for both sets of samples, even if they show different X-ray absorption [griseofulvin/crospovidone composites do not present microabsorption ( [alpha] [asymptotically equal to] 1) unlike [beta] -piroxicam/crospovidone composites]. This accord is a striking result, since the two techniques are based on different physical phenomena, thus confirming the reliability of the proposed method. Quantitative analysis by means of DSC, with a coefficient of variation (CV) that ranges from 2.3 to 6.1%, is more precise than QXRD analysis, for which the CV ranges from 2.2 to 16.2%. The CV value of 16.2% is significantly higher than the values found in the literature (see, for example, Suryanarayanan, 1995), which are about 10%. However, the CV value of 16.2% refers to a composite (G04) with a crystalline drug weight fraction c_D = 0.035. This is a value close to the detection limit of the X-ray diffraction technique, since the peaks are hardly detectable from the amorphous background. In this case we must consider the low sampling reproducibility, the poor counting statistics, and the fact that the decomposition of the diffraction profile is numerically less reliable. Thus, the QXRD calculated value of c_D is about 10% bigger than the value obtained by DSC because of an inevitable overestimation of the peaks area. The above problem is not significant for the other composites, and the CV values from QXRD measurements of those composites are in agreement with the literature results.

5. Conclusions

The QXRD approach developed in this work is shown to be a consistent and accurate method to determine the residual crystalline drug weight fraction in largely amorphous pharmaceutical composites (QXRD and DSC analyses are in excellent agreement). The implemented diffraction-absorption method has the merits of being a standardless technique, avoiding the interpretation of the amorphous halo and knowledge of the crystal structures of the composite phases. Another important feature of the proposed technique is that it takes into account the absorbed water, which influences the X-ray absorption coefficient, and subsequently the intensity scale factor.

Finally, this approach is general and it can be applied to largely amorphous composite materials, thus finding relevance in other emerging research fields, such as glass systems generated in the stabilization/solidification of toxic waste.

Acknowledgements

We thank Dr Elza Bontempi for helping us in performing GIXRD measurements.

References

Alexander, L. & Klug, H. P. (1948). Anal. Chem. 20, 886-894.
Bergese, P., Bontempi, E., Colombo, I. & Depero, L. E. (2001). J. Appl. Cryst. 34, 663-665.
Bergese, P., Bontempi, E., Colombo, I., Gervasoni, D. & Depero, L. E. (2001). Proceeding of 7th European Conference on Advanced Materials and Processes. ISBN 88-85298-39-7.
Bish, D. L. & Howard, S. A. (1988). J. Appl. Cryst. 21, 86-91.
Bruker AXS (1998). TOPAS P V1.0. Profile Fitting Using an Analytical as well as a Fundamental Parameters Approach. Bruker AXS, Karlsruhe, Germany.
Canfield, D. & Sundeen, D. (1989). University of Southern Mississippi, Hattiesburg, MS, USA. JCPDS-ICDD database, card No. 40-1982.
Carli, F., Colombo, I., Magarotto, L., Motta, A. & Torricelli, C. (1986). Int. J. Pharm. 33, 115-124.
Carli, F. & Garbassi, F. (1985). J. Pharm. Sci. 74, 963-967.
Chiou, W. L. & Riegelman, S. J. (1971). Pharm. Sci. 60, 1281-1302.
Colombo, I. & Palladio, P. (2001). Process for the Preparation of Accelerated Release Formulations using Compressed Fluids. Patent Application No. PCT/EP01/02538.
DiMarco, J. (1989). The Polytechnic Institute, Brooklyn, New York, USA. JCPDS-ICDD database, card No. 40-1937.
Iyengar, S. S., Phadnis, N. V. & Suryanarayanan, R. (2001). Powder Diffr. 16, 20-24.
Jenkins, R. & Snyder, R. L. (1996). Introduction to X-ray Powder Diffractometry, pp. 187-191. Chichester: Wiley Interscience.
Le Bail, A. J. (1995). J. Non-Cryst. Solids, 183, 39-42.
Lovrecich, M. (1995). Supported Drugs with Increased Dissolution Rate and a Process for their Preparation. US patent 5449521.
Lutterotti, L. (1998). Mater. Sci. Forum, 278, 87-92.
Orlhac, X., Fillet, A., Deniard, P., Dulac, A. M. & Brec, R. (2001). J. Appl. Cryst. 34, 114-118.
Rietveld, H. M. (1969). J. Appl. Cryst. 2, 65-71.
Serajuddin, A. T. M. (1999). J. Pharm. Sci. 88, 1058-1066.
Surana, R. & Suryanarayanan, R. (2000). Powder Diffr. 15, 2-6.
Suryanarayanan, R. (1995). Physical Characterization of Pharmaceutical Solids, edited by H. G. Brittain, pp. 188-221. New York: Marcel Dekker.
Suryanarayanan, R. & Herman, C. S. (1991). Pharm. Res. 8, 393-399.
Takayama, K., Imazumi, H., Nambu, N. & Nagai, T. J. (1982). J. Pharm. Dyn. 5, S3.
Theeuwes, F., Hussain, A. & Higuchi, T. J. (1974). Pharm. Sci. 63, 427-429.
Vippagunta, S. R., Brittain, H. G. & Grant, D. J. W. (2001). Adv. Drug Delivery Rev. 48, 3-26.
Yamamura, S. & Momose, Y. (2001). Int. J. Pharm. 212, 203-212.
Zevin, L. S. & Kimmel, G. (1995a). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 35-45. Berlin: Springer.
Zevin, L. S. & Kimmel, G. (1995b). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 211-217. Berlin: Springer.
Zevin, L. S. & Kimmel, G. (1995c). Quantitative X-ray Diffractometry, edited by I. Mureinik, pp. 104-112. Berlin: Springer.

research papers