Biometals
`DOI 10.1007/s10534-015-9845-9
`
`Physico-chemical properties of the new generation IV iron
`preparations ferumoxytol, iron isomaltoside 1000 and ferric
`carboxymaltose
`
`Susann Neiser • Daniel Rentsch • Urs Dippon • Andreas Kappler •
`Peter G. Weidler· Jiirg Giittlicber ·Ralph Steininger· Maria Wilhelm·
`Michaela Braitsch • Felix Funk • Erik Philipp • Susanna Burckhardt
`
`Received: 2 March 2015 /Accepted: 4 March 2015
`©Springer Science+Business Media New York 2015
`
`Abstract The advantage of the new generation N
`iron preparations ferric carboxymaltose
`(FCM),
`ferumoxytol (FMX), and iron isomaltoside 1000 (IIM)
`is that they can be administered in relatively high doses
`in a short period of time. We investigated the physico(cid:173)
`chemical properties of these preparations and compared
`them with those of the older preparations iron sucrose
`(IS), sodium ferric gluconate (SFG), and low molecular
`weight iron dextran (LMWID). Mossbauer spec(cid:173)
`troscopy, X-ray diffraction, and Fe K-edge X-ray
`absorption near edge structure spectroscopy indicated
`akaganeite structures (~FeOOH) for the cores ofFCM,
`IIM and IS, and a maghemite (y-Fei~) structure for
`that of FMX. Nuclear magnetic resonance studies
`confirmed the structure of the carbohydrate of FMX as
`a reduced, carboxymethylated, low molecular weight
`
`Electronic supplementary material The on1ine version of
`this article (doi:10.1007/s10534-015-9845-9) contains supple(cid:173)
`mentary material, which is available to authorized users.
`
`S. Neiser · M. Wilhelm · M. Braitsch ·
`F. Funk· E. Philipp · S. Burckhardt (181)
`Chemical and Preclinical Resean:h and Development,
`Vifor (International) Ltd., St. Gallen, Switzerland
`e-mail: susanna.burckhardt@viforpharma.com
`
`D. Rentsch
`Swiss Federal Laboratories for Materials Science and
`Technology (Empa), Diibendorf, Switzerland
`
`U. Dippon · A. Kappler
`Geomicrobiology, Center for Applied Geosciences,
`University of Tiibingen, Tiibingen, Germany
`
`dextran, and that of IIM as a reduced Dextran 1000.
`Polarography yielded significantly different fingerprints
`of the investigated compounds. Reductive degradation
`kinetics of FMX was faster than that of FCM and IIM,
`which is in contrast to the high stability ofFMX towards
`acid degradation. The labile iron content, i.e. the amount
`of iron that is only weakly bound in the polynuclear iron
`core, was assessed by a qualitative test that confirmed
`in
`the order
`decreasing
`labile
`iron contents
`SFG ~ IS > LMWID ~ FMX ~ IIM ~ FCM. The
`presented data are a step forward in the characterization
`of these non-biological complex drugs, which is a
`prerequisite to understand their cellular uptake mechan(cid:173)
`isms and the relationship between the structure and
`physiological safety as well as efficacy of these
`complexes.
`
`P. G. Weidler
`Institute of Functional Interfaces, Karlsruhe Institute of
`Technology (KIT), Eggenstein-Leopoldshafen, Germany
`
`J. Gottlicher . R. Steininger
`ANXA-Synchrotron Radiation Facility, Karlsruhe
`Institute for Technology (KIT),
`Eggenstein-Leopoldshafen, Germany
`
`Published unline: 24 M~rch 2015
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 1
`
`

`
`Keywords
`Intravenous iron · Iron sucrose ·
`Ferric carboxymaltose · Iron isomaltoside 1000 ·
`Ferumoxytol · Low molecular weight iron dextran
`
`EDTA
`FCM
`FDA
`FMX
`GFC
`Glc
`GOF
`HMBC
`
`HMWID
`HSQC
`
`IIM
`IS
`ISS
`N
`LMWID
`Mn
`Mw
`Mz
`NBCD
`NMR
`NTBI
`p
`PSC
`
`Abbreviations
`DQF-COSY Double quantum filtered correlation
`spectroscopy
`Bthylenediaminetetraacetic acid
`Ferric carboxymaltose
`U.S. Food and Drug Administration
`Ferumoxytol
`Gel-filtration chromatography
`Glucose
`Goodness of fit
`Heteronuclear multiple-bond
`correlation
`High molecular weight iron dextran
`Heteronuclear single quantum
`correlation
`Iron isomaltoside 1000
`Iron sucrose
`Iron sucrose similar
`Intravenous
`Low molecular weight iron dextran
`Number average molecular weight
`Weight average molecular weight
`z-average molecular weight
`Non-biological complex drugs
`Nuclear magnetic resonance
`Non-transferrin bound iron
`Polydispersity
`Polyglucose sorbitol
`carboxymethylether
`Quadrupole splitting
`Standard deviation
`Selected area electron diffraction
`Carboxymethylation substitution
`degree
`Sodium ferric gluconate
`Standard hydrogen electrode
`Transmission electron microscopy
`Total correlation spectroscopy
`Tris(hydroxymethyl)aminomethane
`United States Pharmacopeia
`X-ray absorption near edge structure
`X-ray diffraction
`
`QS
`s
`SAED
`SDCM
`
`SFG
`SHE
`TEM
`TOCSY
`TRIS
`USP
`XANES
`XRD
`
`~Springer
`
`Biometals
`
`Introduction
`
`Intravenous (N) iron therapy is widely used to treat
`iron deficiency and iron deficiency anemia (Auerbach
`and Ballard 2010). The indications include absolute
`iron deficiency, when there is a need for fast iron
`repletion or an intolerance to oral iron, as well as the
`therapy of anemia of chronic disease (ACD, also
`called iron sequestration syndromes) or functional iron
`deficiency (Goodnough et al. 2010). Under inflamma(cid:173)
`tory conditions (anemia of chronic disease) or when
`there is a high iron demand for erythropoiesis (func(cid:173)
`tional iron deficiency), such as during therapy with
`erythropoiesis stimulating agents, oral iron therapy is
`not effective and N iron is recommended (Goodnough
`et al. 2010; Qunibi 2010). Therefore, the therapeutic
`areas for N iron are widespread and, among others,
`include nephrology (Besarab and Coyne 2010; Mac(cid:173)
`dougall et al. 2012), cardiology (Avni et al. 2012;
`Macdougall et al. 2012; von Haehling et al. 2012),
`oncology (Gafter-Gvili et al. 2013), gastroenterology
`(Gomollon and Gisbert 2013), and gynecology (Brey(cid:173)
`mann et al. 2010; Haththotuwa et al. 2011), as well as
`patient blood management (pre-/postoperative ane(cid:173)
`mia) (Shander et al. 2012).
`All iron compounds used for N iron therapy consist of
`a polynuclear Fe(III)-oxyhydroxide/oxide core surround(cid:173)
`ed by carbohydrates which stabilize the core and protect
`the nanoparticles against further polymerization (Auer(cid:173)
`bach and Ballard 2010; Kastele et al. 2014; Macdougall
`and Geisser 2013; Qunibi 2010). Despite their similar
`structure, N iron compounds have distinct properties. In
`particular, they display a wide range of stability and,
`depending on the type of stabilizing carbohydrate, may
`have antigenic potential (Chertow et al. 2006). N iron
`compounds belong to the class of so-called non(cid:173)
`biological complex drugs (NBCD) which, in contrast to
`small molecules, cannot be fully characterised by
`physico-chemical methods and which are largely defined
`by the manufacturing process (Crommelin et al. 2014).
`A main goal of the recent developments in the field
`of N iron therapy was to provide a preparation that
`can be administered in higher doses and in a short
`period of time. Some of the older products, such as
`sodium ferric gluconate (SFG) and iron sucrose (IS)
`are not very stable and thus contain a higher percent(cid:173)
`age of labile, weakly-bound iron (Van Wyck et al.
`
`Pharmacosmos, Exh. 1035, p. 2
`
`

`
`J
`
`Biometals
`
`2004; Van Wyck 2004). This property, together with
`the high osmolarity and the high pH of the solutions,
`limits the maximum single doses allowed for SFG and
`IS to 62.5-125 and 200-500 mg iron (Fe), respective(cid:173)
`ly (Macdougall and Geisser 2013). In contrast, the
`three recently introduced IV iron preparations Ferin(cid:173)
`ject® /Inj ectafer® (active ingredient: ferric carboxy(cid:173)
`maltose,
`FCM),
`Feraheme®/Rienso®
`(active
`ingredient: ferumoxytol, FMX), and MonoFer® (ac(cid:173)
`tive ingredient: iron isomaltoside 1000, IlM), are more
`stable and can all be administered in comparatively
`high single doses (from 510 up to more than 1000 mg
`Fe) (Macdougall and Geisser 2013). An overview on
`selected properties and recommended dose regimen of
`the different preparations is given in Table 1.
`FCM consists of a polynuclear Fe(III)-oxyhydroxide
`core surrounded by carboxymaltose which is derived
`from maltodextrin, an oligosaccharide produced from
`starch. IlM contains a polynuclear Fe(III)-oxyhydroxyde
`core which is stabilized by a hydrogenated (reduced)
`Dextran 1000 (isomaltoside 1000) and a low amount of
`citrate (Andreasen and Christensen 2001; Medice Phar(cid:173)
`ma GmbH&Co. KG, Iserlohn, Germany 2011 ; Nordfjeld
`et al. 2012; Pharmacosmos A/S, Holbaek, Denmark
`2009). A different, matrix-like structure has been
`proposed for IIM (Jahn et al. 2011), but has recently
`been shown to be based on an incorrect interpretation of
`nuclear magnetic resonance (NMR) data (Kiistele et al.
`2014 ). FMX was originally developed as a contrast agent
`for magnetic resonance imaging and has been described
`as a superparamagnetic ferric oxide coated with polyglu(cid:173)
`cose smbitol carboxymethylether (PSC), a reduced and
`carboxymethylated dextran (Simon et al. 2006). FMX
`contains mannitol and further PSC as excipients (AMAG
`Pharmaceuticals Inc., Waltham, MA, USA 2009; Euro(cid:173)
`pean Medicines Agency 2012).
`In this work, the physico-chemical properties of
`these three new preparations for parenteral iron
`therapy were characterized and compared. In par(cid:173)
`ticular, we show that despite the similarity of the
`investigated compounds, they show unique properties
`and important differences which may have an impli(cid:173)
`cation for their therapeutic application. Comparable
`techniques have been used previously to characterize
`some of these compounds, but direct comparison is not
`always available and there is not always a good
`agreement among the results (Balakrishnan et al.
`2009; Funk et al. 2001 ; Fiitterer et al. 2013; Jahn et al.
`2011; Kudasheva et al. 2004).
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 3
`
`

`
`Materials and methods
`
`Materials
`
`The following IV iron preparations were obtained
`from a pharmacy or directly from the manufacturer:
`Ferahem.e® (30 mg Fe/m.L, AMAG Pharmaceuticals,
`Inc., Lexington, MA) and Rienso® (Takeda Global
`Research and Development Centre (Europe) Ltd.,
`Aldwych-London, UK), active ingredient ferumoxytol
`(FMX); MonoFer® (100 mg Fe/m.L, Pharmacosmos
`AJS, Holbaek, Denmark), active ingredient iron iso(cid:173)
`maltoside 1000 (IIM); Ferinject® (50 mg Fe/m.L,Vifor
`(International) Ltd., St. Gallen, Switzerland), active
`ingredient ferric carboxymaltose (FCM); Venofer®
`(20 mg Fe/mL, Vifor (International) Ltd., St. Gallen,
`Switzerland), active ingredient iron sucrose (IS);
`Cosmofer® (50 mg Fe/m.L, TEVA GmbH, Radebeul,
`Germany), active ingredient iron dextran, and Fer(cid:173)
`rlecit® (12.5 mg Fe/m.L, Sanofi-Aventis Deutschland
`GmbH, Frankfurt am Main, Germany), active ingre(cid:173)
`dient sodium ferric gluconate (SFG). Since different
`lots were used, the lot numbers are indicated for each
`method separately.
`
`Biometals
`
`above) and IIM (MonoFer® lot 949171-1) solutions in
`phosphate buffer: Product solution corresponding to
`120 mg iron was mixed with 50 mL phosphate buffer
`(4.43 g KH2PO.J0.143 g Na2HP04 per 500 mL) and
`heat treated at 132 °C and pH 5.4 for 90 min. The iron
`phosphate precipitate was separated from the carbohy(cid:173)
`drate-containing supernatant by centrifugation ( 10 min
`at 500 U/min). The heat treatment and centrifugation
`was repeated with the supernatant (132 °C, 15 min for
`FMX and 132 °C, 90 min for IIM). The filtrate of the
`supernatant was lyophilized and dissolved in distilled
`water [l.17 % (m/m)]. The resulting carbohydrate(cid:173)
`containing solution from FMX was dialyzed with a
`regenerated, ethylenediaminetetraacetic acid (EDT A)(cid:173)
`treated cellulose membrane (nominal cut-off 1000 Da,
`V-series by ZelluTrans Roth, Karlsruhe, Germany, lot
`190119) to remove phosphate and again lyophilized.
`The lyophilisate from FMX most likely contains both
`the core-stabilizing PSC and PSC added as an excipient.
`Due to the lower molecular weight of isomaltoside
`1000, the carbohydrate-containing solution from IIM
`was not purified by dialysis but by ion exchange (Merck
`ion exchanger V, no. 104836. Lot L701436 101) to
`remove the phosphate.
`
`Separation of mannitol from the FMX solution
`
`Mossbauer spectroscopy
`
`To remove mannitol, 18 mL FMX solution (Ferahe(cid:173)
`me® lot 09060402) were diluted with 7.2 mL distilled
`water and filtered with an Amicon Ultra-15 centrifugal
`filter device (IVD ultracel-10 K regenerated cellulose,
`molecular weight cut-off 10 kDa, Millipore, Ireland).
`The retentate was taken up six times with distilled
`water and centrifuged. The final retentate contained
`the mannitol-free FMX. Although the weight-average
`molecular weight of the carboxymethylated dextran in
`FMX is above the molecular weight cut-off of the
`filter, low-molecular weight fractions of the unbound
`carboxymethylated dextran may also have been partly
`removed from the solution by this method. The
`resulting material was used for X-ray diffraction
`(XRD) measurements and as starting material for the
`isolation of the FMX carbohydrate.
`
`Isolation of the carbohydrates from the FMX
`and IIM solutions
`
`The iron-carbohydrate nanoparticles were destroyed by
`heat treatment of the mannitol-free FMX (see section
`
`An aliquot of 400 µL (FMX and IIM) or 450 µL
`(FCM) of liquid sample material was pipetted under
`oxic conditions into a Teflon sample container, sealed
`with Kapton tape and frozen at -30 °C. The frozen
`samples were transferred into a closed cycle helium
`cryostat and Mossbauer spectra were recorded at
`temperatures between 245 and 5 K. The Mossbauer
`spectra were recorded in transmission geometry using
`a constant acceleration drive system equipped with a
`57 Co source in rhodium matrix and a proportional
`counter coupled to a CMCA-550 1024 multichannel
`analyzer (WissEL, Germany). All spectra were
`calibrated against a room temperature spectrum of a
`7 µm. alpha-Fe foil. The spectra were fitted using
`Voigt based spectral lines (RECOIL software suite,
`University of Ottawa, Canada). During fitting, the
`half-width-half-maximum of the peaks was kept at
`0.097 mm/s and the Gauss sigma parameter was
`varied to account for line broadening. The blocking
`temperature was defined as the temperature at which
`50 % of the material was paramagnetic and 50 %
`magnetically ordered (Funk et al. 2001 ; Murad 1996).
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 4
`
`

`
`Biometals
`
`The following lots were analyzed: MonoFer® lot
`949171-1, Feraheme® lot 09060402, Ferinject® lot
`062201.
`
`Iron(II) quantification
`
`The amount of Fe(II) was determined by cerimetric
`titration with cerium sulfate and potentiometric end(cid:173)
`point determination (Jander et al. 2003). The following
`lots were analyzed: MonoFer® lot 949171-1, Ferabe(cid:173)
`me® lot 09060402, and 9 consecutive lots ofFerinject®.
`
`X-ray diffraction (XRD)
`
`The samples were measured on different instruments
`(Bruker, Karlsruhe, Germany) as indicated in the
`Supplementary data (Table Sl). All samples were
`transferred to a flat disk sample holder with a 1 mm
`deep round indention with 20 mm in diameter. ex-Ali~
`(corundum) was applied as an internal standard to all
`samples. Liquid samples were applied onto a Si single(cid:173)
`crystal chip (diameter 1 inch) and air-dried at 50 °C.
`The diffractograms were analyzed with the soft(cid:173)
`wares DiffracPlus EV A 15.0 for a first identification of
`the crystalline phases, and TOPAS 4-2 (both Bruker
`AXS GmbH, Karlsruhe, Germany) for a more ad(cid:173)
`vanced determination of lattice constants by the
`Pawley method (Pawley 1981). Space group informa(cid:173)
`tion was obtained from literature. The profile function
`applied to all refinements was a modified Thompson(cid:173)
`Cox-Hastings pseudo-Voigt function (Young 1993).
`The goodness of fit parameter (GOF) was also
`calculated. A GOF of 1.0 corresponds to a perfect
`agreement between model and data. The domain size
`was determined from the Scherrer equation (Klug and
`Alexander 197 4) and consideration of the instrumental
`contribution to line broadening determined from LaB6
`measurements.
`
`X-ray absorption near edge structure spectroscopy
`(XANES)
`
`Fe K-edge XANES spectroscopy was performed at the
`X-ray beamline of the Synchrotron Radiation Labora(cid:173)
`tory for Environmental Studies (SUL-X) of ANKA
`(Karlsruhe Institute for Technology, Eggenstein(cid:173)
`Leopoldshafen, Germany) using a Si(lll) crystal pair
`in a fixed exit monochromator. To avoid radiation
`damage the beam was collimated. Pellets of fine
`
`lepidocrocite
`(cx-FeOOH),
`grained goethite
`(y(cid:173)
`FeOOH), 2-line and 6-line ferrihydrite, as well as
`akaganeite (~FeOOH) were prepared with cellulose
`in order to achieve optimal absorption for the spectra
`that were used as references. The liquid sample
`solutions (MonoFer® lot 224121-3; Venofer® lot
`076201) were measured in a custom-made cell for
`liquid samples. The cell mainly consists of two
`Kapton-foils with an effective diameter of 10 and
`0.7 mm separation, which results in a volume of
`0.55 mm3
`• The foils are sealed with 0-rings, and the
`parts are kept together by a metallic frame equipped
`with in- and outlets. The 0.7 mm spacing has been
`chosen because at that distance the transmission signal
`is not disturbed by the undiluted sample solution. Two
`cell fillings of each substance were measured, and
`each measurement was performed twice at the same
`spot to detect instrumental influences or changes
`within the sample that may occur due to high photon
`flux density. Energy step width across the edge was
`0.3 eV. The energy has been calibrated to 7112 eV at
`the first maximum of the first derivative of a XANES
`spectrum of elemental Fe (Fe foil). All measurements
`were done in transmission mode using ionization
`chambers as detectors for the incident and absorbed
`beams of sample, reference substances and Fe foil.
`Spectra were added and averaged. Their pre- and post(cid:173)
`edge ranges were approximated by linear and polyno(cid:173)
`mial fit-functions. The edge jump was normalized to 1
`in order to compare spectra with each other. Both steps
`of data processing were done with the Athena program
`of the IFEFFIT package (RAVEL and Newville 2005).
`Combinatorial Linear Combination Fits (LCF) have
`been performed with the Athena program of the
`IFEFFIT package for the Fe K-edge XANES spectra
`with five reference spectra: akaganeite, goethite,
`lepidocrocite, 2L-ferrihydrite (2L-Fh) and 6L-ferrihy(cid:173)
`drite (6L-Fh) (RAVEL and Newville 2005).
`
`NMR spectroscopy
`
`The 1H and 13C NMR spectra of the IIM and FMX
`carbohydrates were recorded at400.13 and 100.61 MHz
`on a Bruker Avance 400 NMR spectrometer (Bruker
`Biospin AG, Fiillanden, Switzerland). The 1H, the 1D
`diffusion-edited 1H, and the 13C NMR spectra, as well as
`the 1H-13C HSQC, 1H-13C HMBC, 1H-13C HSQC(cid:173)
`TOCSY, and 1H-1H DQF-COSY 2D correlation NMR
`experiments were performed at 298 K on a 5 mm
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 5
`
`

`
`broadband inverse probe equipped with z-gradient
`
`(100 % gradient strength of 53.5 Gcm-1) applying 90"
`pulse lengths of6.8 µs (1H) and 14.5 µs (13C). All NMR
`in D20
`experiments were performed
`solutions
`(c = 50-200 mglmL) using the Bruker standard pulse
`programs and parameter sets selecting coupling con(cid:173)
`stants of 145 Hz (HSQC), 10 Hz (HMBC) and mixing
`times of 150 ms (HSQC-TOCSY). The 1H and 13C
`chemical shifts were externally referenced relative to the
`signals of 3-trimethylsilyl-2,2,3,3-tetradeutero sodium
`propionate dissolved in DiO at 0.0 and -1.6 ppm,
`respectively. The carbohydrates of following lots were
`analyzed: MonoFer® lot 949171-1, Feraheme® lot
`()1)()6()4()2 and lot 10012802; Dextran 3-4000 (Amer(cid:173)
`sham Biosciences, lot 303820) and Dextran 1000
`(Sigma-Aldrich Switzerland, Lot BCBD4347V-02-
`002) were used as reference samples.
`
`Molecular weight distribution
`
`The molecular weight distribution was determined by
`gel-filtration chromatography (GFC) as described
`earlier (Geisser et al. 1992). The lot numbers of the
`investigated samples are given in the results section.
`For the determination of the molecular weight of
`carbohydrate samples, glucose and Dextran 1000 were
`used in addition to the pullulans PS, PlO, and P20 as
`calibration standards (Geisser et al. 1992).
`
`Polarographic analysis
`
`The reduction potentials were measured by differential
`pulse polarography based on the US Pharmacopeia
`(USP) method described in the monograph for iron
`sucrose injection (U.S. Food and Drug Administration
`2012b). The measurements were done at pH 7 in 15 %
`m/v acetate buffered solutions on a Metrohm 7'.17 VA
`Computrace polarograph (Metrohm AG, Herisau,
`Switzerland) with a multi-mode working electrode,
`reference electrode Ag/ AgC1/c(KO) = 1 mol/L, and
`with a platinum auxiliary electrode. The values reported
`in this work have been corrected by +236 mV to
`express the potentials relative to the standard hydrogen
`electrode (SHE). 1be iron concentrations in the sample
`solutions were 15, 50, 20, and 25 µg/mL for FMX, IlM,
`IS, and FCM, respectively. All iron formulations were
`measured immediately after opening the container. The
`following lots were analyzed: Feraheme® lot 10061002,
`
`~Springer
`
`Biometals
`
`MonoFer® lot 042838-3, Venofer® lot 901201, Ferin(cid:173)
`ject® lot 169001.
`
`Reductive degradation kinetics
`
`The reductive degradation kinetics were measured at pH
`2.50--2.60 and 25 °C in solutions containing ascorbic
`acid, citric acid, H3P04 , Na2HP04 (all 0.08 M), FeS04
`(0.008 M), and sorbitol (1 M) (Erni et al. 1984). The
`analyzed lots are given in the results section.
`
`"Tea test"
`
`The "tea test" was used as a qualitative assay to
`visualize the content of labile, weakly-bound iron by
`reaction with polyphenols. Five bags of Lipton white
`tea (lot L23320D023; exp 1112014. Unilever Schweiz,
`8240 Thayingen, Switzerland) were placed in 2 L of a
`0.9 % NaCl solution at 90--100 °C and allowed to
`steep for 1 min. White tea was chosen because of its
`light color combined with a high content of polyphe(cid:173)
`nols. After the infusion had cooled to <37 °C, the iron
`preparations were added to result in an iron concen(cid:173)
`tration of 0.1 mg/mL and stirred (Ferinject® lot
`258001; MonoFer® lot 042838-3; Feraheme® lot
`A56996A; Venofer® lot 133001; Cosmofer® lot
`1226801-3; Ferrlecit® lot D2A046A). No pH adjust(cid:173)
`ment was done. The solutions were filled into cuvettes,
`and photographs were taken 1-2 h after sample
`preparation had started.
`
`Results and discussion
`
`Characterization of the polynuclear iron oxide/
`hydroxide cores
`
`Mossbauer spectroscopy
`
`The Mossbauer spectra of FMX could be modeled
`with parameters characteristic for Fe(III), with no
`indication of the presence of Fe(II). Based on hyper(cid:173)
`fine field parameters of the 5 K spectrum, FMX was
`identified as nano-maghemite (Fig. 1; Table 2) (Trone
`et al. 2000; Tucek et al. 2006). The transition from the
`superparamagnetic to the magnetically-ordered state
`spanned a wide temperature range of more than 100 K.
`While the material is completely magnetically ordered
`at 5 K, first signs of magnetic relaxation were detected
`
`Pharmacosmos, Exh. 1035, p. 6
`
`

`
`77 K
`
`SK
`
`Biometals
`
`Fig. 1 Mossbauer spectra
`of FMX, IIM and FCM
`recorded at 77 K (left) and
`5 K (right). Col<Jred dots
`represent the Mossbauer
`signal, dark grey limis the fit,
`and light gray lines the
`single models used for
`fitting of the spectra.
`Modeling parameters are
`given in Table S2
`(Supplementary data).
`(Color figure online)
`
`c
`0
`li
`0 .,
`~ ..
`"' Cl>
`~
`<;;
`E
`0 z
`
`c:
`
`.S? a.
`j
`"' "' -~ c;;
`
`E
`0
`z
`
`· 12
`
`·B
`
`-4
`
`4
`0
`Velocity [mm/s )
`
`8
`
`12 -12
`
`·B
`
`·4
`
`4
`0
`Velocity [mm/s]
`
`8
`
`12
`
`Table 2 Overview of the results from XRD, XANES spectroscopy, and Mossbauer spectroscopy
`
`Sample
`
`XRD
`
`XANES
`
`Mossbauer
`
`Core mineral
`
`Domain size (nm)
`
`Core mineral
`
`Core mineral
`
`Blocking
`temperature (K)
`
`FMX
`FCM
`IIM
`IS
`
`Maghemite (y-Fei03)
`Akaganeite (P,.FeOOH)
`Akaganeite (P,.FeOOH)
`n/aa
`
`10
`4-5
`3
`nJa•
`
`n.d.b
`n.d.b
`Akaganeitc
`Akaganeite
`
`Maghemite
`Akaganeite
`Akaganeitc
`n.d.b
`
`73
`114
`56
`n.d.b
`
`• n/a (not applicable): akaganeite most likely but no clear identification of the core mineral by XRD, see text
`b not determined
`
`at 45 K.. For higher temperatures an increasing
`paramagnetic contribution was found. At 245 K, the
`superparamagnetic doublet was still strongly broad(cid:173)
`ened, showing that the material was not completely
`superparamagnetic although no distinct sextet features
`were visible. The coexistence of doublet and sextet
`features over a large temperature range has been
`described for nanoparticle suspensions of maghemite
`(Trone et al. 2000). The magnetic blocking tem(cid:173)
`perature of FMX was found to be 73 K. The blocking
`temperature depends mainly on the particle size,
`crystal structure and interparticle interactions. Based
`on the assignment as a maghemite suspension, the low
`blocking temperature of FMX suggests that the iron
`cores are in the range of 5-10 nm (Morup and Trone
`1994; Trone et al. 2000).
`
`The Mossbauer spectra of IIM could also be
`modeled with parameters characteristic for Fe(lll),
`with no indication of the presence of Fe(II). With an
`average hyperfine field of 47.6 T, JIM showed a
`significantly weaker hyperfine field at 5 K than FMX
`(Table S2, Supplementary data). Additionally, some
`magnetic relaxation, visible as innerline broadening,
`was present even at 5 K.. These spectral features
`suggest a low crystallinity of the iron cores. The
`hyperfine field of 47.6 T combined with a center shift
`of 0.48 mm/s are indicative for an akaganeite structure
`(Barrero et al. 2006; Bigham et al. 1990). The blocking
`temperature was 56 K (Table S2, Supplementary
`data). The spectra between 30 and 60 K showed an
`increase in paramagnetic contribution, and, at 77 K,
`66 ± 1.5 % of the spectral area showed magnetically-
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 7
`
`

`
`Biometals
`
`ordered material (Fig. 1). As reported earlier (Jahn
`et al. 2011), the material is completely paramagnetic at
`150 K.
`Also the spectra of FCM suggested an akaganeite(cid:173)
`like structure without indication of the presence of
`Fe(II). A second sextet was necessary to model the
`spectra between 5 and 77 K. The two models used for
`the spectrum at 5 K had a similar average of the center
`shift and a slightly weaker magnetic hyperfine field
`than that of IIM (Fig. 1; Table S2, Supplementary
`data). The hyperfine field strengths of the sextets were
`slightly smaller than expected for akaganeite. This
`could arise from iron on the surface of the mineral
`cores with a different binding environment compared
`to bulk material, defects in the akaganeite structure
`such as chloride vacancies or some additional lepi(cid:173)
`docrocite in the sample (Bigham et al. 1990; Murad
`and Cashion 2004). Despite the visible magnetic
`relaxation at 5 K, which suggests a low crystallinity,
`the blocking temperature of 114 K was the highest of
`the three samples. Between 30 and 120 K, both
`magnetically-ordered material and superparamagnetic
`
`material were present in the sample, while at 245 K
`the sample was completely paramagnetic, lacking the
`broadening observed for FMX (data not shown).
`
`X-ray diffraction (XRD)
`
`XRD diffractograms of FMX, recorded for the manni(cid:173)
`tol-free sample, allowed the identification of the FMX
`core as a maghemite (y-Fe20 3 space group #96 P43212)
`without any signals of magnetite (Fig. 2a). Models of
`ak:aganeite and magnetite worsened the agreement
`between model and raw data. The best GOF found with
`maghemite was 6.8, which still represents a lower
`quality fit. The lattice constants (Table S3, Supplemen(cid:173)
`tary data) were in good agreement with reference data
`for maghemite (Pecharroman et al. 1995), and the
`domain sizes were determined to lie around 10 nm,
`which is in agreement with the Mossbauer data
`(5-10 nm) and close to the earlier reported core
`diameter of 6.4 nm (Jahn et al. 2011).
`The diffractogram of IS appeared feature-poor with
`broad peaks, which resulted in an arguable fit with the
`
`Ca>1.5 ~--------~ (b) 300 - - - - - - - - - -
`FMX
`i
`.2.. 1.0
`1:,
`,....
`>C 0.5
`
`"
`
`'
`
`'
`
`11111 11111 1111111111111111111111111
`I 11111 •
`I 11111
`• • •111•111
`
`II
`
`v; 200
`Q.
`.2..
`~ 100
`'iii
`-·
`c
`-·
`~ i 0
`~
`- '~~· ~!"!f_J,~
`~~~~--~IW
`B ~~·
`30
`40
`30
`40
`20 c·>
`20 c·>
`(c) --------~ (d) 2
`___ _____ ~
`llM
`FCM
`v; 4
`Q.
`.2..
`...
`1:, 2
`>C
`~o
`'iii
`c
`J!!
`B
`
`10
`
`20
`
`50
`
`60
`
`70
`
`10
`
`20
`
`50
`
`60
`
`70
`
`0
`
`10
`
`20 c·>
`
`20
`
`30
`
`10
`
`20
`
`50
`
`60
`
`70
`
`30
`40
`20 c·>
`
`Fig. 2 X-ray diffractograms ofFMX (a), IS (b), FCM (c), and
`IIM (d). The diffractograms show the raw data (grey dots), fitted
`model (red), sub-patterns of the model phases (blue and green),
`and in the FMX spectrum (a) also the internal standard
`corundum (light blue). The model phases are maghemite for
`
`FMX (a), goethite (blue) and akaganeite (green) for IS (b), and
`akaganeite for FCM and IIM (c, d). Below the diffractograms,
`the difference curves are shown (gray) and the peak positions of
`the indicated phases are marked by vertical lines. (Color figure
`online)
`
`~Springer
`
`Pharmacosmos, Exh. 1035, p. 8
`
`

`
`Biometals
`
`different possible crystallite structures. A mixture of
`goethite (a:-FeOOH, space group #62 Pbnm) and
`akaganeite (~-FeOOH, space group #12 I2/m) could
`be postulated (calculated GOF value 1.2) (Fig. 2b).
`Previously published XRD analyses were interpreted
`in a variety of different ways: the core structure of IS
`was assigned as a ferrihydrite (Funk et al. 2001), a
`ferrihydrite with possibly other structures mixed in
`such as akaganeite (Jahn et al. 2011 ), as a lepidocrocite
`or ferrihydrite (Fiitterer et al. 2013), or as akaganeite
`(Kudasheva et al. 2004). The inconsistency of these
`results may partly arise from different experimental
`details such as the drying conditions. Moreover, a very
`small crystallite size or a low crystallinity in the IS core
`makes XRD, a technique suitable for the characteriza(cid:173)
`tion of long-range order, not an ideal method for the
`investigation of IS. In fact, earlier XRD results suggest
`a core diameter of between about 1 and 3 nm (Funk
`et al. 2001 ; Jahn et al. 2011; Kudasheva et al. 2004).
`The mineral core of FCM could clearly be identi(cid:173)
`fied as akaganeite (J3-Fe00H, space group #12 12/m)
`(Fig. 2c; Table 2). The lattice constants (Table S3,
`Supplementary data) were in good agreement with
`literature data (Post and Buchwald 1991). The GOF
`reached 1.4, and the domain size for the akaganeite lay
`around 4-5 nm, which is in agreement with the value
`of 4.3 nm reported earlier (Jahn et al. 2011).
`The mineral core of IIM could most likely be
`attributed to an akaganeite--like phase (J3-FeOOH,
`space group #12 I2/m) (Fig. 2d; Table 2). Though
`accompanied by a large standard deviation, the
`calculated lattice constants (Table S3, Supplementary
`data) were in good agreement with reference data for
`akaganeite (Post and Buchwald 1991). The domain
`sizes derived were around 3 nm. These results agree
`well with the previously published description of the
`IIM core as being similar to akaganeite with a core
`diameter derived by XRD of 4.2 nm (Jahn et al. 2011).
`As described for IS, also IIM showed the limitation of
`XRD for poorly crystalline and/or very small particles.
`The agreement between model and raw data was poor
`for all structures tested, which were goethite, lepi(cid:173)
`docrocite, maghemite and magnetite. The best GOF
`was obtained by an akaganeite structure with a value
`of 16.6. The high GOF value and the large variation in
`the difference curve reflect a low confonnity between
`the applied model and the data, which was observed
`previously (Jahn et al. 2011).
`
`X-ray absorption near edge structure (XANES)
`spectroscopy
`
`The relatively broad peaks in the XRD diffractograms
`and the high GOF values of the described substances
`show the limitation of XRD for poorly crystalline or
`very small particles, especially for IS and JIM. Fe
`K-edge XANES spectroscopy was chosen to better
`characterize IS and JIM, because it gives information
`about the short range order of an absorbing element,
`here Fe, and hence can be applied also for low- or
`non-crystalline materials. Furthermore, because Fe
`K-edge XANES measurements were done with liquid
`samples, artifacts from the drying process, which may
`play a role especially for the XRD of IS, can be
`excluded.
`The XANES spectra with details of their pre-edge,
`ma