Hyperfine Interactions 136: 73–95, 2001.
`© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
`
`73
`
`Physical and Chemical Characterization of
`Therapeutic Iron Containing Materials: A Study of
`Several Superparamagnetic Drug Formulations with
`the β-FeOOH or Ferrihydrite Structure
`
`FELIX FUNK1,∗
`, GARY J. LONG2, DIMITRI HAUTOT2, RUTH BÜCHI1,
`ISO CHRISTL1 and PETER G. WEIDLER1
`1Institute of Terrestrial Ecology, Swiss Federal Institute of Technology Zurich, CH-8952 Schlieren,
`Switzerland; e-mail: funk@viforint.ch
`2Department of Chemistry, University of Missouri-Rolla, Rolla, MO 65409-0010, USA
`
`(Received 12 November 2001; accepted 11 December 2001)
`
`Abstract. The effectiveness of therapeutically used iron compounds is related to their physical and
`chemical properties. Four different iron compounds used in oral, intravenous, and intramuscular ther-
`apy have been examined by X-ray powder diffraction, iron-57 Mössbauer spectroscopy, transmission
`electron microscopy, BET surface area measurement, potentiometric titration and studied through
`dissolution kinetics determinations using acid, reducing and chelating agents. All compounds are
`nanosized with particle diameters, as determined by X-ray diffraction, ranging from 1 to 4.1 nm. The
`superparamagnetic blocking temperatures, as determined by Mössbauer spectroscopy, indicate that
`the relative diameters of the aggregates range from 2.5 to 4.1 nm. Three of the iron compounds have
`an akaganeite-like structure, whereas one has a ferrihydrite-like structure. As powders the particles
`−1.
`form large and dense aggregates which have a very low surface area on the order of 1 m2 g
`There is evidence, however, that in a colloidal solution the surface area is increased by two to three
`orders of magnitude, presumably as a result of the break up of the aggregates. Iron release kinetics
`by acid, chelating and reducing agents reflect the high surface area, the size and crystallinity of the
`particles, and the presence of the protective carbohydrate layer coating the iron compound. Within a
`physiologically relevant time period, the iron release produced by acid or large chelating ligands is
`small. In contrast, iron is rapidly mobilized by small organic chelating agents, such as oxalate, or by
`chelate-forming reductants, such as thioglycolate.
`
`Key words: colloidal iron oxyhydroxides, therapeutic use, X-ray diffraction, Mössbauer, BET sur-
`face area, dissolution kinetics.
`
`1. Introduction
`
`Iron oxyhydroxide particles of desired size and chemical properties are of increas-
`ing interest in technological applications, such as catalysts, pigments, ferrofluids,
`recording media, and magnetic resonance imaging contrast agents. In addition, they
`
`* Corresponding author. Present address: Vifor (International) Inc., Rechenstrasse 37, CH-9001
`St. Gallen, Switzerland.
`
`Pharmacosmos, Exh. 1026, p. 1
`
`

`

`74
`
`F. FUNK ET AL.
`
`have a potential importance in medical applications. Although iron is the second
`most abundant metal in the Earth’s crust, iron deficiency is the most common
`mineral deficiency disease worldwide. More than one billion people have iron
`deficiency and about 700 million people have iron deficiency anemia [1].
`The low bioavailability of non-haem iron reflects its distinct tendency to hydrol-
`yse and polymerize unless it is strongly complexed [2]. Nutritional iron deficiency
`may be treated, depending on the risk situation, either by nutrient fortification or by
`pharmaceutical supplementation. In the latter approach, iron can be administered
`by either an oral or a parenteral route. The most often used oral preparation is
`ferrous sulphate, whereas an iron dextran complex and an iron sucrose complex
`are the preferred forms for parenteral administration [1, 3]. The choice of an ad-
`ministrative route and the specific drug to be used depends upon several factors,
`including the diagnosis of the cause and impact of the iron deficiency, the effective-
`ness, safety and economy of the different treatments, and the compliance, toxicity
`and side effects of a specific drug formulation.
`Iron(III) oxyhydroxide complexes with (poly)saccharides form the basis for a
`group of drugs designed for oral as well as parenteral administration [1]. However,
`the specifications for the drugs meeting the pharmacokinetical factors required
`for the different administrative pathways differ widely. The specifications may be
`achieved through, in some cases, the selection of different carbohydrate coatings or,
`in other cases, by synthetic procedures which can change the physical and chemical
`properties of the iron oxyhydroxide complexes.
`In this paper we report an investigation of a series of iron(III) oxyhydroxide
`complexes which are kept in solution as colloidal particles by protection with
`different carbohydrate coatings. These complexes are prepared specifically for ad-
`ministration by different routes, i.e., by oral, intravenous, or intramuscular routes.
`In order to acquire information about the structure, size, and surface areas of these
`iron oxyhydroxy complexes, they have been characterized by powder X-ray dif-
`fraction, iron-57 Mössbauer spectroscopy, transmission electron microscopy, BET
`surface area measurements, and potentiometric titrations. The physico-chemical
`properties of the complexes have been related to their chemical stability and lability
`in terms of their dissolution by acid, chelating agents and reducing agents. The
`results of these studies provide insight into the effectiveness of iron release from
`the different iron preparations under different conditions, conditions which mimic
`physiological compartments, such as blood, stomach or intestine. In addition, these
`studies provide a foundation for assessing the properties necessary for improving
`preparations for specific drug administrative routes.
`
`2. Experimental
`
`2.1. MATERIALS
`The following iron compounds were investigated: Ferrum Hausmann® intramuscu-
`lar (iron(III) hydroxide dextran complex, Dexfer®; lots 375009A1, solution sam-
`
`Pharmacosmos, Exh. 1026, p. 2
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`75
`
`ple, and 521119M, powder sample), in the following called iron dextran; Ferrum
`Hausmann® intramuscular (high molecular weight iron(III) hydroxide complex
`with polymaltose, Amylofer®; lots 545009A1, solution sample, and 612209M,
`powder sample), called iron dextrin; Maltofer® (low molecular weight iron(III)
`hydroxide complex with polymaltose; lots 654009M, drops solution sample, and
`512219M, powder sample), called iron polymaltose; and Ferrum Hausmann® i.v.
`(iron(III) hydroxide sucrose complex, Venofer®; lot 630209, solution sample), called
`iron sucrose. Human apo-transferrin was purchased from Sigma Chemicals Co.,
`St. Louis, MO, USA. All other chemicals were of the highest purity commercially
`available. Solutions were prepared using double distilled water.
`
`2.2. POWDER X-RAY DIFFRACTION
`
`Powder X-ray diffraction patterns were obtained on a SCINTAG XDS 2000 dif-
`fractometer using Cu Kα 0.15418 nm radiation. This X-ray diffractometer used is
`a Bragg–Brentano camera equipped with a Peltier cooled, lithium drifted silicon
`detector. The powder samples of iron dextrin, iron dextran and iron polymaltose
`were suspended in acetone, gently ground in an agate mortar and transferred with
`a pipette onto a glass slide and then air dried. The concentrate of iron sucrose was
`air dried on a glass slide. The diffraction patterns were recorded over a 2θ angular
`◦
`◦
`range of 15 to 95
`with a step of 0.03
`in 2θ and a 10 second counting time per
`step at room temperature for iron dextrin, iron dextran, iron polymaltose and iron
`◦
`sucrose. For the akaganeite reference sample this range extended over 5 to 134
`2θ.
`The diffraction patterns were evaluated, when possible, with the Rietveld meth-
`od [4]. In this method a least-squares refinement is carried out until the best fit is
`obtained between the observed and calculated powder diffraction pattern, where
`the calculated pattern is based on a refined model for the crystal structure or struc-
`tures, i.e., the unit cell lattice and atomic positional parameters, and the diffraction
`optics and instrumental factor [5]. For these refinements to be successful the in-
`formation content of the diffraction pattern must be sufficient, i.e., the patterns
`should contain many diffraction lines with small line widths at high diffraction
`angles. The parameters containing information on the quality of the refinement are
`cited in Table I. Rwp and S are a measure of the agreement between the calculated
`diffraction patterns of the mineral phases in the powder mixture and the measured
`diffraction profile. Both values are equal to 1.0 in the ideal case. The Bragg factor,
`RB, expresses the agreement between the model used for each single phase and
`the measured diffraction pattern. The Durbin–Watson statistical parameter, Dwd,
`whose maximum is 2.0, reflects the goodness of refinement in total. The space
`group was I 4/M, and Z was equal to 8. The chemical formula used in the refine-
`ment was equal to FeOOCl with Cl occupying 5% of the possible sites, hence, a
`formula weight of 89.63 gram per mol can be deduced.
`The background was refined with a polynom of 3rd order.
`
`Pharmacosmos, Exh. 1026, p. 3
`
`

`

`76
`
`F. FUNK ET AL.
`
`Table I. Crystallite size, strain and quality of fit parameters
`
`Compound
`
`MCLa
`
`(nm)
`
`Strainb
`−1
`
` dd
`
`Quality of fit
`
`Rwp
`
`S
`
`RB
`
`Dwd
`
`iron dextrin
`4.1
`iron polymaltose
`1.9
`iron dextran
`1.8
`akaganeite
`5.2
`a MCL = mean coherence length.
`b d = lattice spacing.
`
`0.004
`0.013
`0.010
`0.006
`
`13.8
`12.0
`12.0
`21.4
`
`3.9
`3.6
`3.7
`5.8
`
`9.6
`4.8
`5.1
`13.1
`
`1.0
`1.1
`1.1
`0.3
`
`Number of
`reflections
`
`198
`182
`189
`366
`
`The mean coherence length (MCL), the average size of the crystals, was cal-
`culated with the method of Williamson and Hall [6]. From this method one also
`obtains information about the internal stress and strain experienced by a crystal.
`LaB6 (NIST # 660) was used as an internal line width standard.
`
`2.3. MÖSSBAUER SPECTRA
`
`−2 of powder and the spectra
`The Mössbauer spectral absorbers contained 44 mg cm
`were measured between 4.2 and 295 K on a constant-acceleration spectrometer
`which utilized a room temperature rhodium matrix cobalt-57 source and was cal-
`ibrated at room temperature with α-iron foil. The resulting paramagnetic spectra
`have been fit with the distribution method of Le Caër [7]. These fits, which used
`−1, revealed
`20 component doublets and a fixed component linewidth of 0.23 mm s
`no correlation between the quadrupole splitting and the isomer shift. An attempt
`to fit the spectra containing both superparamagnetic and antiferromagnetic spectral
`components with the method of Le Caër [7] and with the method of Wivel and
`Mørup [8] proved unsuccessful. As a result of the failure of these methods to fit
`the spectra with a distribution model, the spectra have been fit with the minimum
`number of sextets and doublets needed to reproduce the observed spectral absorp-
`tion profile. In these fits the components within each magnetic sextet have the same
`linewidth and an area ratio of 3 : 2 : 1 : 1 : 2 : 3.
`The estimated approximate errors for the hyperfine parameters are ±2 kOe for
`
`−1 for the isomer shifts, ±0.02 mm s−1 for the
`the hyperfine fields, ±0.01 mm s
`quadrupole splittings, and ±2 percent for the relative areas of the superparamag-
`netic and antiferromagnetic components. The errors in the parameters derived from
`the completely superparamagnetic spectra are somewhat smaller.
`
`Pharmacosmos, Exh. 1026, p. 4
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`77
`
`2.4. TRANSMISSION ELECTRON MICROGRAPHS
`
`Transmission electron micrographs were obtained on a PHILIPS CM30ST elec-
`tron microscope. Sample material was suspended in ethanol with the help of an
`ultrasonic bath. Subsequently the drops of the suspension were transferred onto
`Cu-grids and air dried.
`
`2.5. BET SURFACE AREAS
`
`The N2 adsorption isotherms were measured with a Micromeritics Gemini 2360
`◦
`C in a
`device. Prior to the N2 adsorption the samples were heated for 24 h at 90
`steady stream of nitrogen. Higher temperatures were not applied because of the
`possible dehydroxylation and subsequent change in the mineral structure of the
`iron oxyhydroxides [9]. For the calculation of the specific surface area we used the
`BET equation [10].
`
`2.6. POTENTIOMETRIC TITRATIONS
`The charging behavior of iron oxides was determined by potentiometric acid-base
`titration. Experiments were performed at 25±1
`◦
`C in a thermostated room using au-
`tomatic precision titration equipment [11]. Four burettes (Dosimat 605, Metrohm,
`Herisau, Switzerland), a glass electrode (Metrohm) and an AgCl reference elec-
`trode (Metrohm) were connected to a personal computer by a Microlink MF18
`interface (Biodata, Manchester). The burettes were filled with CO2-free deionized
`−3 HCl, ∼0.05 mol dm
`−3 KOH, and 2 mol dm
`−3 KCl. The
`water, 0.05 mol dm
`KOH and KCl solutions were prepared under a nitrogen atmosphere using CO2-
`free deionized water. To keep the solutions CO2-free during the experiments, all
`burettes were connected to the atmosphere through a glass tube filled with NaOH
`on granulated activated carbon. All experiments were carried out in a 250 cm3 glass
`vessel, which was continuously flushed with water-saturated, CO2-free nitrogen
`gas. Before starting an experiment, solutions were acidified to remove residual
`CO2. Typically, experiments were performed by titrating with base, the forward
`titration, prior to acid titration, the back titration. During the entire titration cycle
`the ionic strength was held constant within ±1% by adding either water or salt
`solution to correct for dilution effects resulting from the addition of acid or base.
`After one titration cycle the ionic strength was adjusted to the next higher level
`by adding salt solution. In this way a series of forward and backward titrations at
`different ionic strengths were obtained within a single experiment.
`The solution was stirred for two minutes after each addition of titrant. Acid and
`base dose sizes were automatically adjusted to yield data in steps of about 20 mV.
`−1
`Electrode readings were recorded, if the electrode drift was less than 0.05 mV min
`or after a maximum drift time of 20 min.
`Exact base concentration and electrode parameters were obtained by blank titra-
`tions of the electrolyte solution by using a least-square fitting routine. The calcu-
`
`Pharmacosmos, Exh. 1026, p. 5
`
`

`

`78
`
`F. FUNK ET AL.
`
`lation included activity corrections as well as diffusion potentials. The average
`−3. The derived
`deviation between fit and measurement was less than 15 µmol dm
`parameters, such as the dissociation constant of water and the activity coefficients
`were similar to literature values [12].
`The surface charge density was determined by subtracting theoretical blank
`titrations from the oxide titrations prior to division of the resulting data by the
`BET surface area. The common intersection point of the resulting titration curves
`at different ionic strengths represents the point of zero charge (PZC).
`
`2.7. DISSOLUTION KINETICS
`
`−3 HCl) as
`Acid dissolution was performed in a strongly acid solution (2 mol dm
`−3 HCl), pH 3 (1 mmol dm
`−3 HCl), pH 5 (buffered with
`well as at pH 1 (0.1 mol dm
`−3
`−3 acetic acid/sodium acetate) and pH 7 (buffered with 25 mmol dm
`25 mmol dm
`Hepes/sodium hydroxide). Dissolution was monitored periodically at 470 nm on an
`Uvikon 810 spectrophotometer (Kontron Instruments, Zurich, Switzerland). Decay
`−3,
`kinetics were measured at 470 nm with oxalic acid/sodium oxalate (0.2 mol dm
`−3 at pH 5 and 7 in the buffers described above), and β-
`pH 3), EDTA (10 mmol dm
`−3 acetic acid/sodium
`−3) at pH 3, pH 5 (50 mmol dm
`mercaptoethanol (0.2 mol dm
`−3 Hepes/sodium hydroxide buffer). Dis-
`acetate buffer) and pH 7 (150 mmol dm
`−3) was followed by monitoring the ab-
`solution by thioglycolic acid (0.2 mol dm
`−3 Hepes/sodium hydroxide buffer) and
`sorbance at 530 nm at pH 7 (150 mmol dm
`−3 acetic acid/sodium acetate buffer), and at 470 nm at pH 3.
`at pH 5 (50 mmol dm
`Formation
`of
`tiron
`(brenzcatechine-3,5-disulfonic
`acid
`disodium salt,
`−3) complexes was monitored at 660 nm at pH 3, 560 nm at pH 5
`10 mmol dm
`−3 acetic acid/sodium acetate buffer) and at 490 nm at pH 7
`(25 mmol dm
`−3 Hepes/sodium hydroxide buffer). Transfer of iron to apo-transfer-
`(50 mmol dm
`−3) and bovine serum albumin (50 g dm
`−3) was followed at pH 7
`rin (3 g dm
`−3 Hepes/sodium hydroxide) at 470 nm. The iron concentration was
`(50 mmol dm
`−3. All solutions prepared for kinetic determinations were kept at room
`125 mg dm
`temperature in the dark except during photometric measurements.
`
`3. Results and discussion
`
`3.1. POWDER X-RAY DIFFRACTION
`The powder X-ray diffraction patterns of iron dextrin, iron polymaltose and iron
`dextran are shown in Figure 1. The sharp peaks indicated by × in the pattern and
`the difference curve for iron polymaltose are due to the presence of NaCl in the
`sample. The patterns are similar to that obtained with β-FeOOH, akaganeite. This
`mineral crystallizes in the tetragonal crystal class with the space group I 4/M,
`No. 87, with unit cell parameters a = 1.048 nm and c = 0.3023 nm [13].
`The very broad lines observed in Figure 1 are typical of a substance consisting
`of small crystallites, although the line broadening could also be caused by internal
`
`Pharmacosmos, Exh. 1026, p. 6
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`79
`
`Figure 1. X-ray powder diffraction patterns of (a) iron dextrin, (b) iron dextran, (c) iron polymaltose
`(× = NaCl), and (d) iron sucrose. The points represent the experimental data and the line the
`calculated pattern. The vertical bars indicate the positions of the peaks and the curve beneath is
`the difference between the observed and calculated intensity.
`
`stress and strain. Because of the small number and the broad line widths of the
`observed diffraction peaks, and the use of Cu Kα radiation, a determination of the
`site occupancy, especially of the iron atoms on the corresponding lattice positions,
`was not done. The site occupancy of the Cl position at (0 0 0) was constrained to
`−1. The deviation between the calculated lattice parameters and those
`5% mol mol
`of akaganeite found in literature [13, 14] are small, see Figures 2(a) and 2(b). The
`fractional coordinates for Fe are similar to the ones reported by McKay [13] for
`akaganeite (Figure 2(c)). A somewhat larger deviation from literature data was ob-
`served for the oxygen positions of the three studied akaganeite samples, especially
`for the O1 positions (Figure 2(d)). This indicates a larger distortion in our samples
`which can be explained by the small crystallite sizes. These are usually more sus-
`
`Pharmacosmos, Exh. 1026, p. 7
`
`

`

`80
`
`F. FUNK ET AL.
`
`Figure 2. Comparison of the results of the Rietveld refinement with literature data. The lattice cell
`parameters, a (a) and c (b), and the fractional coordinates of the iron positions (c) and the two
`oxygen positions (d) are shown for iron polymaltose (Mal), iron dextran (Dex), iron dextrin (Amy), a
`synthetic akaganeite (Aka), and for two β-FeOOH investigated by McKay [13] (McKay) and Szytula
`et al. [14] (Szy). Error bars represent standard deviations; if not visible they are smaller than the
`symbols.
`
`ceptible to strain than larger crystallites. This interpretation is also supported by
`the values obtained for the strain parameter and given in Table I.
`The broad lines in the X-ray diffraction patterns suggest that the crystallites
`are very small, a suggestion which is supported by the mean coherence length
`values. There is a marked decrease from 4.1 nm for iron dextrin to 1.9 nm for
`iron polymaltose and to 1.8 nm for iron dextran. The values for the strain, see
`Table I, suggest an increase for the smaller crystallites. The measured parameters
`were compared with those of a synthetic akaganeite, a sample which is typical of
`most laboratory-produced akaganeite [15]. The particles of this sample [16] consist
`of approximately 100 nm long bars with a cross section of 25 to 100 nm2.
`The refinements of these structures are satisfactory if one considers the small
`number of observed diffraction peaks and their large line widths. The Rwp values of
`
`Pharmacosmos, Exh. 1026, p. 8
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`81
`
`12 to 21%, given in Table I, are typical for these X-ray diffraction patterns and an
`S value of 3 to 6 is good for such conditions. This is supported by the high values
`of Dwd for iron dextrin, iron dextran, and iron polymaltose. The low Dwd value
`for sample akaganeite can not be explained. The low RB values indicate the good
`agreement between the structural model and the measured data. The theoretical
`limit is never achieved for powder samples even under optimized conditions.
`If one assumes that each particle in suspension is composed of a single spherical
`crystallite, the upper limit for its surface area, SA, is given by
`SA = 3r
`−1ρ
`−1,
`−3.
`where r is the radius of the particle in m and ρ is the density of the particle in g m
`Then by using the specific density of akaganeite, ρ = 3.56 × 103 kg m
`−3, one
`−1 for iron dextrin, iron polymaltose
`obtains surface areas of 410, 890 and 940 m2 g
`and iron dextran, respectively.
`The X-ray diffraction pattern of iron sucrose is also shown in Figure 1. A strik-
`ing feature of this pattern is the high background upon which broad lines can
`be seen. This high background is due to the large amount of organic substances
`contained in the original suspension. The observed diffraction peaks are simi-
`lar to those of ferrihydrite, an iron hydroxide with a poorly ordered crystallo-
`graphic structure in which iron ions are randomly distributed in octahedral inter-
`stices formed by hexagonal close packed oxygen ions [17]. In contrast to the 2-line
`and 6-line ferrihydrites, whose diffraction patterns show only 2 or 6 peaks, iron
`sucrose may have an additional shift of the iron ions toward the oxygen planes.
`More detail about ferrihydrites and their possible structures may be found in Drits
`et al. [17].
`The mean coherence length for a iron sucrose crystallite is approximately 1 nm,
`and thus is comparable with the other three samples. Hence, a maximum surface
`−1 can be calculated by using the same assump-
`area of approximately 1690 m2 g
`tions as mentioned above.
`
`3.2. MÖSSBAUER SPECTRA
`
`The Mössbauer spectra of iron polymaltose, iron dextran, and iron dextrin, ob-
`tained between 4.2 and 295 K, are shown in Figures 3, 4, and 5, respectively, and
`a summary of the hyperfine parameters associated with the fits is given in Table II.
`The spectra of iron sucrose are very similar to those shown in Figure 3 for iron
`polymaltose. It is immediately clear from these figures and the parameters given in
`Table II that the spectra of all four compounds are very similar at 295 and 4.2 K. At
`295 K all the spectra exhibit a broadened doublet, whereas at 4.2 K they all exhibit
`a well resolved but broadened magnetic sextet.
`The 295 K Mössbauer spectra of all four compounds require two doublets for
`an adequate fit and the resulting isomer shifts and quadrupole splittings of these
`doublets are characteristic of iron(III) in a distorted octahedral environment. The
`
`Pharmacosmos, Exh. 1026, p. 9
`
`

`

`82
`
`F. FUNK ET AL.
`
`Figure 3. The Mössbauer spectra of iron polymaltose obtained at the indicated temperatures.
`
`results are essentially identical to those reported at room temperature by Cham-
`baere et al. [18] for akaganeite, β-FeOOH, and for ferrihydrite [15]. Unfortunately,
`there is not sufficient resolution of the two doublet components to obtain reliable
`hyperfine parameters or linewidths for each doublet, values which could be used to
`evaluate [18] the amount of chloride present in the “open” channels of β-FeOOH.
`In order to overcome this difficulty, the 295 K spectra have been fit [7] with a
`distribution of quadrupole splittings with a possible linear correlation between the
`isomer shift and the quadrupole splitting. No such correlation was found and the
`resulting average isomer shifts and quadrupole splittings are completely consis-
`tent [19] with what would be expected for β-FeOOH containing small amounts of
`chloride ion.
`The 4.2 K spectra shown in Figures 3–5 are all quite similar to that reported
`earlier [19] for β-FeOOH at 30 K, except that the spectra reported herein are
`somewhat more symmetric in both the intensities and widths of the six component
`lines. Indeed, as observed earlier, it was impossible to adequately fit the 4.2 K
`
`Pharmacosmos, Exh. 1026, p. 10
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`83
`
`Figure 4. The Mössbauer spectra of iron dextran obtained at the indicated temperatures.
`
`spectra with two sextets and at least three sextets are required, although the three
`resulting best fit sextets do not readily correlate with those found earlier [20, 21]
`or with the proposed electronic and structural properties of β-FeOOH. However,
`the 4.2 K weighted average isomer shifts, see Table II, agree exactly with the
`−1 obtained earlier [20, 21]. In con-
`weighted average isomer shift of 0.48 mm s
`trast, the 4.2 K weighted average hyperfine field of 482 kOe reported earlier is
`significantly higher than the weighted average 463, 470, and 474 kOe values found
`herein for iron polymaltose, iron dextrin and iron dextran, respectively, but slightly
`smaller than the 490 kOe obtained for iron sucrose. Chambaere and De Grave have
`concluded [20, 21] that the smaller hyperfine fields in β-FeOOH correspond to
`iron sites with additional hydroxyl ions in their coordination environment. Thus the
`reduced hyperfine fields in iron dextran, iron dextrin, and especially iron polymal-
`tose may indicate that they have higher levels of chloride present and, in contrast,
`there may be less chloride in iron sucrose. This finding is in accordance with the
`
`Pharmacosmos, Exh. 1026, p. 11
`
`

`

`84
`
`F. FUNK ET AL.
`
`Figure 5. The Mössbauer spectra of iron dextrin obtained at the indicated temperatures.
`
`proposed iron sucrose structure found by X-ray diffraction, i.e., a ferrihydrite-like
`structure which has no “channels” containing chloride.
`A careful comparison of the spectra shown in Figures 3–5 and the parame-
`ters given in Table II reveals that, at intermediate temperatures, the spectra of the
`four compounds are very different. These spectral results are all characteristic of
`compounds consisting of fine magnetic particles undergoing superparamagnetic
`relaxation [22–25].
`For compounds containing magnetically ordered iron, below a certain particle
`size, typically 15×103 to 150×103 nm3, depending upon the compound, individual
`particles may consist of single magnetic domains. Because of the small particle
`size, the magnetic anisotropy energy of the particle, which for simplicity is often
`given as
`E(θ ) = KV sin2 θ,
`
`Pharmacosmos, Exh. 1026, p. 12
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`85
`
`Compound
`
`iron sucrose
`
`Table II. A summary of the Mössbauer spectral parameters
`(cid:8)H(cid:9)
`(cid:8)δ(cid:9)a
`−1)
`(kOe)
`(mm s
`0.36
`0.45
`0.46
`0.46
`0.46
`
`T
`(K)
`
`295
`85
`50
`
`4.2
`
`
`
`(cid:8) EQ(cid:9)
`−1)(mm s
`0.72
`0.75
`0.71
`–
`–
`
`0
`0
`0
`346
`490
`
`Area
`(%)
`
`100
`100
`40
`60
`100
`
`100
`71
`29
`21
`79
`100
`
`100
`13
`87
`100
`100
`
`100
`1
`99
`100
`100
`100
`
`iron polymaltose
`
`iron dextran
`
`iron dextrin
`
`295
`78
`
`45
`
`4.2
`
`295
`78
`
`45
`4.2
`
`295
`78
`
`60
`30
`4.2
`
`0
`0
`297
`0
`330
`463
`
`0
`0
`278
`366
`474
`
`0
`0
`356
`381
`427
`470
`
`0.36
`0.44
`0.44
`0.45
`0.45
`0.48
`
`0.36
`0.47
`0.47
`0.47
`0.48
`
`0.37
`0.44
`0.44
`0.45
`0.46
`0.48
`
`0.74
`0.76
`–
`0.77
`–
`–
`
`0.74
`0.94
`–
`–
`–
`
`0.73
`0.79
`–
`–
`–
`–
`
`a The isomer shifts are given relative to room temperature α-iron foil.
`
`where V is the volume of the particle, K is the magnetic anisotropy constant, and θ
`is the angle between the magnetization vector and the easy axis of magnetization,
`may be comparable to the thermal energy, kT . As a consequence, the magnetic
`moments of a particle may fluctuate through a rotation of the spins about the
`easy magnetic direction. If this fluctuation occurs on a measurable time scale, the
`material is said to be exhibiting superparamagnetic behavior on that time scale. As
`was first noted by Néel [26], for KV (cid:1) kT , the temperature dependence of the
`magnetic relaxation time, τ , is given by
`τ = τ0 exp(KV / kT ),
`−10 to 10−12 s. Thus, at a given tempera-
`
`where τ0 is typically of the order of 10
`ture, large particles may exhibit slow magnetic relaxation, whereas, small particles
`
`Pharmacosmos, Exh. 1026, p. 13
`
`

`

`86
`
`F. FUNK ET AL.
`
`may exhibit fast or superparamagnetic relaxation. The superparamagnetic blocking
`temperature, TB, is defined as that temperature below which the superparamagnetic
`relaxation is slow on the time scale of the measurement technique.
`
`−7 to 10−8 s is ideal for the study
`The iron-57 Mössbauer-effect time scale of 10
`of superparamagnetism in many magnetic systems. If the magnetic material under
`study has a highly uniform distribution of particle sizes, then its blocking tem-
`perature will be well defined and there will be a narrow temperature range over
`which the observed Mössbauer spectrum will change from doublets, characteristic
`of superparamagnetic relaxation, to sextets, characteristic of slow magnetic relax-
`ation. Unfortunately, many fine particle materials do not posses this narrow size
`distribution and, thus, they show a distribution of blocking temperatures. In this
`case the average blocking temperature may be approximated as that temperature
`at which the Mössbauer spectral absorption area is one-half superparamagnetic
`doublets and one-half magnetic sextets. In general, the more uniform the particle
`size, the narrower will be the temperature range over which both components are
`observed. Further, at temperatures near the blocking temperature, the influence of
`magnetic relaxation on the Mössbauer-effect time scale may be observed as highly
`broadened magnetic sextets.
`The Mössbauer spectra shown in Figures 3–5, as well as those of iron sucrose,
`all exhibit the expected changes for superparamagnetic relaxation and the tempera-
`ture dependence of the percentage of the magnetically ordered spectral absorption
`area is shown in Figure 6. From this figure it is apparent that the blocking tempera-
`tures are ca. 55, 65, 115, and 215 K for iron sucrose, iron polymaltose, iron dextran,
`and iron dextrin, respectively. Thus it is clear from a magnetic point of view that,
`in the solid state, the average particle volume in these three materials must increase
`in the same order. The observed diameters of ∼1.0, 1.9, 1.8, and 4.1 nm for iron
`sucrose, iron polymaltose, iron dextran, and iron dextrin, respectively, obtained by
`X-ray diffraction correspond to spherical particle volumes of ∼0.5, 3.6, 3.1, and
`36.1 nm3. If one assumes that the magnetic anisotropy constant is the same for all
`
`Figure 6. The temperature dependence of the percentage of magnetically ordered components
`present in the Mössbauer spectra of iron sucrose ((cid:2)), iron polymaltose ((cid:3)), iron dextran ((cid:4)), and
`iron dextrin (•).
`
`Pharmacosmos, Exh. 1026, p. 14
`
`

`

`CHARACTERIZATION OF THERAPEUTIC IRON CONTAINING MATERIALS
`
`87
`
`four compounds, then the differing blocking temperatures yield relative volumes
`of 8.4, 11.0, 19.3, and 36.1 nm3 for the four compounds.
`The differences in the particle volumes found by X-ray diffraction and Möss-
`bauer spectroscopy are rather small. They may be due to very strong magnetic
`coupling between the collection of individual crystallites found in a given ag-
`gregate in the solid state. As is indicated below, BET surface area measurements
`indicate that these aggregates are very densely packed, a packing which could well
`lead to strong magnetic coupling within the aggregate. Indeed, such “superferro-
`magnetic” coupling has been proposed earlier [22] for densely packed goethite. In
`the presence of such intercrystalline magnetic coupling, the volume of the aggre-
`gate would determine the superparamagnetic blocking temperature and thus yield
`a larger average particle volume in the solid state. Because these aggregates do not
`possess long range crystallographic order, their X-ray diffraction patterns yield the
`smaller volume of the individual crystallites. Alternatively, the discrepancy may be
`due to the non-spherical shape of the particles. Further, the diameters observed by
`X-ray diffraction are measured for the crystallite dimension and not for the particle
`volume.
`The temperature dependence of the weighted average hyperfine fields in iron
`sucrose, iron polymaltose, iron dextran, and iron dextrin is shown in Figure 7. It
`should be noted that in no case does the plot of the hyperfine field versus tem-
`perature resemble the Brillouin-like plot typically expected of the magneti