Redox reactions of K3[Fe(CN)6] during mechanochemically stimulated phase transitions of AlOOH

Article abstract of DOI:10.1016/j.jpcs.2011.03.016

Thermally induced redox reactions of K₃[Fe(CN)₆] (1) were investigated for a broad temperature range by thermal methods and structure analytical methods (ESR and M??bauer spectroscopy, X-ray Powder diffraction and XANES). Based on the influence of the mechanically activated and transforming matrices 2 and 3, redox processes can be tuned to form doped Al₂O₃ systems which contain either isolated Fe³ centres or redox active phases and precursors like (Al_1-xFe _x)₂O₃ (4), (Al_3-xFe _x)O₄ (5), Fe₃O₄, Fe ₂O₃ and Fe⁰. The phase Fe₃C and the chemically reactive C-species were detected during the reaction of 1. The final composition of the doped products of α-Al₂O₃ is mainly influenced by the chemical nature of the Fe doping component, the applied temperature and time regime, and the composition of the gas phase (N ₂, N₂/O₂ or N₂/H₂). From the solid state chemistry point of view it is interesting that the transforming matrix (2 and 3) possesses both oxidative and protective properties and that the incorporation of the Fe species can be performed systematically.

Full text of DOI:10.1016/j.jpcs.2011.03.016

Journal of Physics and Chemistry of Solids 72 (2011) 794–799
Contents lists available at ScienceDirect
Journal of Physics and Chemistry of Solids
journal homepage: www.elsevier.com/locate/jpcs
Redox reactions of K3[Fe(CN)6] during mechanochemically stimulated phase
transitions of AlOOH
Reinhard Sto¨ßer a, Michael Feist a,n, Kerstin Patzwaldt a, Michael Menzel b, Franziska Emmerling b
a Institut fu¨r Chemie der Humboldt-Universita¨t zu Berlin, Brook-Taylor-Straße 2, D-12489 Berlin, Germany
b Bundesanstalt fu¨r Materialforschung und -pru¨fung, Richard-Willsta¨tter-Straße 11, D-12489 Berlin, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Thermally induced redox reactions of K3[Fe(CN)6] (1) were investigated for a broad temperature range
by thermal methods and structure analytical methods (ESR and Mo¨ßbauer spectroscopy, X-ray Powder
diffraction and XANES). Based on the influence of the mechanically activated and transforming matrices
2 and 3, redox processes can be tuned to form doped Al2O3 systems which contain either isolated Fe3þ
centres or redox active phases and precursors like (Al1ꢀxFex)2O3 (4), (Al3ꢀxFex)O4 (5), Fe3O4, Fe2O3 and
Fe0. The phase Fe3C and the chemically reactive C-species were detected during the reaction of 1. The
Received 15 November 2010
Received in revised form
5 March 2011
Accepted 25 March 2011
Available online 6 April 2011
Keywords:
final composition of the doped products of a-Al2O3 is mainly influenced by the chemical nature of the
A. Inorganic compounds
C. Mo¨ssbauer spectroscopy
C. Thermogravimetric analysis
D. Electron paramagnetic resonance
Fe doping component, the applied temperature and time regime, and the composition of the gas phase
(N2, N2/O2 or N2/H2). From the solid state chemistry point of view it is interesting that the transforming
matrix (2 and 3) possesses both oxidative and protective properties and that the incorporation of the Fe
species can be performed systematically.
& 2011 Elsevier Ltd. All rights reserved.
1. Introduction
[4,5]. Based on careful chemical analysis and long term thermal
analysis, the authors determined the structures of several inter-
Potassium hexacyanoferrate(III), K3[Fe(CN)6] (1), continues to
be a chemical ‘‘evergreen’’ which is also true for Berlin Blue. This
could be deduced, last but not least, from the observation of a
never ending stream of reports on novel chemical and application
aspects found for 1. The present paper contributes to this by
reporting some unexpected spectroscopic and chemical findings
which were obtained by a combination of thermal and spectro-
scopic investigations of 1 in the form of the pure substance as well
as after mechanochemically performed doping of AlOOH xerogels
(2) with 1 and other Fe species, followed by a thermal treatment.
The aim of this study is deduced from the question whether a
targeted Fe doping of Al2O3 samples could be realized by
combining the competing mechanochemically and thermally
stimulated redox processes of iron with the transformation
processes occurring within an AlOOH/Al2O3 matrix. This question
is closely related to the preparation of supported aluminium
oxide catalysts with a defined FeII or FeIII content, as well as to
the preparation of magnetic materials [1–3] by primarily employ-
ing the chemical properties of the Al2O3 precursors and the
special properties of their AlOx matrices.
mediates. The investigations described here support the results of
the authors concerning pure 1, but include additional findings.
The synthesis and structural characterization of the pure crystal-
line phases of ternary and quaternary metal nitrides [6], carbides
[7] and carbometallates [8] and their thermochemistry [9] is well
documented in the literature and contain numerous novel and
unique results. The approach is comparable to that followed in
the current publication. Here, however, we focused on a more
phenomenological description with regard to the redox processes
of 1 competing with the chemistry of the AlO(OH) matrix
transformation.
2. Experimental
All chemicals were of analytical purity (p.A.) and used
as commercially available: K3[Fe(CN)6] (Laborchemie Apolda,
Germany), AlOOH (Nabaltec, USA).
The thermal behaviour was investigated by employing a
simultaneously coupled TA–MS device (Netzsch STA 409
C
Skimmers with a mass spectrometer Balzers QMG 421) in various
A first and important step regarding the investigation of the
thermal decomposition of 1 was given by Wolski and Porawski
gas atmospheres (Details in S5). The mechanical activation was
performed using
a planetary ball mill Pulverisette (Fritsch,
Germany; details in S7). The ESR spectra were recorded on an
X-Band spectrometer (ZWG-Magnettech GmbH, Berlin-Adlershof,
Germany) with MgO:Cr3þ as a standard both for intensity and g0
values (g0 ¼1.9796). The Mo¨ßbauer spectra were obtained with a
n Corresponding author. Tel.: þ49 30 2093 7346; fax: þ49 30 2093 7392.
E-mail addresses: feistm@chemie.hu-berlin.de (M. Feist),
franziska.emmerling@bam.de (F. Emmerling).
0022-3697/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2011.03.016

R. Sto¨ßer et al. / Journal of Physics and Chemistry of Solids 72 (2011) 794–799
795
Fig. 1. (a) TA–MS curves of K3[Fe(CN)6] (1) (34.89 mg) in N2 with the IC curves for the mass numbers m/z¼18 (H2Oþ), 27 (HCNþ), 44 (CO2þ), 52 (C2N2þ), (b) evolution of
the phase composition of K3[Fe(CN)6] samples tempered under N2 at twelve temperatures between 25 and 1100 1C in 100 K steps, (c) three characteristic reaction stages at
ca. 300, 700, and 1100 1C are visualized by their Mo¨ßbauer spectra.
Wissel Electronics spectrometer (Starnberg, Germany) at 295 and
77 K. All spectra were normalized to the standard -Fe foil
(GoodFellow, USA, details in S2). Details about the XRD and
diffraction (XRD), the X-ray absorption near edge spectroscopy
(XANES), and the electron spin resonance (ESR) spectroscopy
(Supporting information in S1–S5). Stable intermediates being
formed in defined temperature-time ranges could be identified
(Fig. 1b and c) and allow description of the evolution of the Fe3C
and Fe0 product phases. The thermal process is accompanied by
colour changes starting with yellow, followed by greenish-blue
to black.
a
XANES measurements can be taken from S3 and S4.
3. Results and discussion
The investigation of the thermal behaviour of pure 1 by
employing a thermal analyzer simultaneously coupled to a mass
spectrometer (TA–MS) is depicted in Fig. 1a together with the
results of the characterization of separately tempered (0.5 h/N2)
samples corresponding to the essential stages of the thermona-
nalytical (TA) curves obtained during the linear heating run. The
thermal activation causes an internal redox process in 1 that
results in the liberation of (CN)2 to yield the formation of lower-
valent Fe species. In this concern, the good agreement of the
TA–MS findings (Fig. 1a) to the Mo¨ßbauer data is obvious. They
were confirmed and completed by the results of the X-ray powder
Furthermore, one can deduce from Fig. 1 that the thermally
initiated reactions start above 250 1C and that the well separated
mass loss around 300 1C is caused by the exothermal (peak
temperature TP 338 1C) release of (CN)2 and, to a minor extent,
of HCN. This is accompanied by the formation of FeII species in the
solid residue, which is clearly indicated by the results of the
Mo¨ßbauer investigations together with XRD and XANES (cf. S2–S4).
The XANES measurements were especially helpful in the case of
sample 1 tempered at 300 1C. Here, the formation of FeII was
indicated by the Mo¨ßbauer data, but could not be confirmed by
XRD due to a low crystallinity of the solid phase formed. It could be

796
R. Sto¨ßer et al. / Journal of Physics and Chemistry of Solids 72 (2011) 794–799
Fig. 2. Simplified reaction scheme of the redox processes occurring in the systems
K3[Fe(CN)6], K3[Fe(CN)6]/AlOOH und Fe0/AlOOH. The reaction sequence supports
and extends the mechanism proposed in [12].
Fig. 3. (a) TA–MS curves of the sample {30% K3[Fe(CN)6]þ70% AlOOH} (8 h
co-milled) using mechanically pre-activated (50 h milling, 3) AlOOH. The IC curves
for the mass numbers m/z¼12(Cþ) and 27 (HCNþ) represent HCN, whereas m/z¼52
(C2N2þ) proves practically the absence of dicyane. (b) Mo¨ßbauer spectrum of the
residue of the TA heating run in (a). The Fe component is composed from: 1 Singlet
shown by XANES, however, that 21% FeII was contained in the
sample with, expectedly, a major part of FeIII (cf. S4). It is obvious
that structurally distorted precursors of Berlin Blue were formed
[10,11]. In the following temperature range between 400 and
650 1C, however, significant changes could not be established
neither by thermal nor by spectroscopic methods. In the solid
residues, K4[Fe(CN)6] is contained, as could be evidenced by XRD.
Above 700 1C, on the other hand, Fe3C is unambiguously detected
both by Mo¨ßbauer spectroscopy and XRD, even if K4[Fe(CN)6]
remains to be the major component of the product phase.
Above 900 1C, metallic Fe0 can be detected unambiguously in
the solid residue (Fig. 1b and c); it is coexisting with KHCO3
(cf. S3a), which is unusual, but in line with the mass spectro-
metric detection of CO2. This formation of carbonate or CO2 can
be explained by highly reactive carbon formed within the decom-
posing matrix of 1 and is obviously able to react with oxygen
traces in the carrier gas.1
(22.5%):
¼ ꢀ0.009 mm/s; Bhf¼ ꢀ33.07 T; 1 Sextet (15.4%):
mm/s; Bhf¼ ꢀ20.84 T.
d
¼ ꢀ0.056 mm/s;
G
¼0.399 mm/s;
1
Sextet (62.1%):
d
¼0.018 mm/s;
¼ ꢀ0.001
D
d
¼0.186 mm/s; D
would be both a support and a substantial extension of the
decomposition mechanism proposed by Raj and Danon [12].
A completely different decomposition proceeds if untreated 1
is mixed with a mechanically activated (50 h high-energy ball
milling), originally crystalline AlOOH (3), then co-milled for
further 8 h, and subsequently subjected to a TA heating run
(Fig. 3a). The TA–MS curves of the mixture {30% 1þ70% 3} and
the Mo¨ßbauer spectrum of the residue after 1200 1C under N2 in
Fig. 3 clearly show that the intermediates formed from 1 react
with the transforming matrix 3. A certain similarity with the
TA–MS curves in Fig. 1 might be deduced, but undoubtedly no (CN)2
is formed here and, as before, a low quantity of HCN (200–500 1C) is
released. Furthermore, the water release from 3 starts earlier
(ca. 160 1C) than in the case of matrix 2, but is intense and spread
over a large temperature range up to 500 1C as does HCN.
After the final tempering of 1 under N2 at 1100 1C, the phases
a-Fe und Fe3C (Cohenite) could be detected by XRD which directly
supports the Mo¨ßbauer findings (Fig. 1, cf. also S2c) and is in line
with the ESR data, i.e. the ferromagnetic resonances.
Mo¨ßbauer spectrum of the residue of the TA run comprises
two sextets caused by 62.1% Fe0 and 15.4% Fe3C, respectively, and
one singlet (22.5%) showing a comparably small isomer shift (IS)
that has to be related to FeIII (cf. S2). This FeIII species was formed
by an oxidative reaction of 1 with the matrix 3, even if internal
reduction processes of 1 are still predominating in this concen-
tration range (30% 1).
As
a mechanism for this kind of ‘‘self-reduction’’ of 1,
a reaction sequence can be proposed that begins with the release
of (CN)2, is then followed by transformations yielding FeI3I(CN)6,
and ends up with the formation of a-Fe, Fe3C, and C (Fig. 2). This
If the concentration of 1 in the mixture with matrix 3 is
lowered, e.g. down to 1 wt% such as when applied for the study on
1
Measurements with an oxygen sensor (Dr. J. Hanss, University of Augsburg,
Germany) yielded an oxygen content in the inert gas (Ar, N2) of ca. 80 ppm.
A comparing TA–MS measurements of non-milled Fe powder under N2 and the
same conditions as applied for 1 exhibited no mass gain at all, i.e. oxidation does
not occur under these conditions.
the transformation of AlOOH xerogels yielding a-Al2O3 [13], the
reactive interaction of the intermediates with the matrix is more
strongly expressed. The formation of Fe3C or Fe0 under N2, for

R. Sto¨ßer et al. / Journal of Physics and Chemistry of Solids 72 (2011) 794–799
797
Fig. 4. Mechanical activation and thermal treatment of the samples of K3[Fe(CN)]6 (1) and AlOOH (2) in different gas atmospheres.
Table 1
Peak temperatures TP (1C) of the crystallization peaks of
result of the milling or co-milling of the AlOOH xerogel with different dopants (1 wt% each).
a
-Al2O3 in the DTA heating curves in different gas atmospheres together with the percentage of decrease
D
TP as a
Sample
TP/DTP
Atmosphere during TA heating run
N2
N2/O2
N2/H2
1178
AlOOH xerogel, untreated
1176
1160
AlOOH xerogel, 8 h milled
1050/ꢀ11%
1146/ꢀ3%
1039/ꢀ12%
1083/ꢀ7.8%
1054/ꢀ10%
987/ꢀ16%
1073/ꢀ9%
1070/ꢀ9%
1048/ꢀ10%
1149/ꢀ1%
1038/ꢀ11%
1083/ꢀ7%
–
1041/ꢀ12%
1144/ꢀ3%
1070/ꢀ9%
1090/ꢀ9%
–
AlOOH xerogel, co-milled with K3[Fe(CN)6]
y co-milled with Fe(NO3)3 9 H2O
y co-milled with Al(NO3)3 9 H2O
y co-milled with Fe(acac)3
y co-milled with Fe2O3
–
–
y co-milled with FeCl3
y co-milled with Fe0
1082/ꢀ7%
1058/ꢀ9%
1070/ꢀ9%
1100/ꢀ7%
instance, is not detectable any more. Fig. 4 depicts schematically
the experimental approach to yield the mentioned results.
From a mechanochemical point of view it is of particular
importance to know that the influence of doping with 1 is weak,
even negative concerning the down-shift of the crystallization
the surface of the activated xerogels is blocked by 1 and its
decomposition products. This effect, which might be regarded as a
mechanochemical deactivation, could be achieved with other
dopants as well (cf. S5) which were unable to form appropriate
crystallization seeds such as, for example, Fe2O3. Up to now, this
kind of a netto activation effect is expressed in the strongest
manner in the case of 1 and has to be attributed to its specific
thermochemical behaviour, including the reaction with the
matrix.
temperature TP (formation of a-Al2O3). Compared with the aver-
age value of the TP decrease of 10.9% (see Table 1), one reaches
here only ꢁ20% of the activation effect being related to the
undoped matrix. This means that the crystallization starting from

798
R. Sto¨ßer et al. / Journal of Physics and Chemistry of Solids 72 (2011) 794–799
Fig. 5. Reference systems: Co-milling of AlOOH xerogels (2) with 1 wt% doping by different iron compounds as pretreatment for the TA heating runs up to 1200 1C in
various gas atmospheres. The phase compositions of the samples after heating comprise compounds of Fe0, Fe2þ, and Fe3þ in different structural environments, among
them magnetite (M) and haematite in two forms (Ha, Hb).
The investigation of several reference systems (Fig. 5) allowed
it to be demonstrated that (i) the results presented here can be
generalized to some extent, (ii) which kind of redox processes
occur in the systems, and (iii) how these redox reactions can be
influenced by the chemical nature of the Fe3þ dopant.
The mechanochemical reactions of the matrix are not restricted to
thin layers of the surface region only. XRD studies on mixtures of
pre-heated samples of 1 (30 min at 200, 400, 700, 900 1C under N2)
and 3 demonstrated that the formation of structurally distorted
corundum can be evidenced already before tempering (cf. ESR results
in S1) which means that parts of the bulk matrix were transformed
mechanochemically. Corundum particles formed this way are capable
to act as seeds for the crystallization of the entire sample and hereby
influences the peak temperature TP of the DTA effect as well
(see Table 1). This is in line with the observed ESR signals at g0 ꢁ4.3
[15] which, during the subsequent tempering step, change to the
As emphasized by Fig. 5, promising possibilities exist for the
reactive doping of Al2O3 matrices. The surface enrichment of the
a
-Al2O3 crystallites with the phase precursors (Al1ꢀxFex)2O3 (4)
and (Al3ꢀxFex)O4 (5) [14], symbolized in Fig. 5 by Fe3þ or Fe2þ
,
respectively, is possible due to competing processes initiated by
the type of Fe ligand, the oxidative power of the transforming
AlOOH matrix, as well as the chosen gas atmosphere in the TA
runs (cf. especially S1b, integrals in the range of 330 mT).
Interesting results were obtained after the doping of 2 with
Fe3O4, Fe2O3, or Fe0. In this concern, the comparably large
variation of the Fe0 fraction of the total Fe content in the product
typical FeIII–ESR fine structure for
a
-Al2O3:Fe3þ with g0 ꢁ13 and 5.2
[13,16,17] thus indicating a more pronounced ordering of the
structure. Furthermore, the protective (i.e. trapping) function of the
matrix for the incorporated FeII and FeIII species could be evidenced
by the ESR data (cf. S1).
a-Al2O3 is noteworthy. This fraction decreases down to ꢁ50% in
the case of reactive interaction with the matrix, whereas one
obtains 100% Fe0 for pure 1 under N2/10%H2. If the internal
oxidative processes were reinforced by employing Fe(NO3) ꢂ 9H2O,
despite the forming gas atmosphere, N2/10%H2, the Fe0 content
can be lowered down to ꢁ40%.
4. Conclusions
Summarizing the presented findings, it should be emphasized
that they represent: (a) the evidence of intermediates of the
thermal transformations of 1, (b) the demonstration of both the
oxidative and protective functions of the thermally transforming
matrix 2, together with (c) the oxidative and disperse incorpora-
tion of Fe3þ ions into the forming
a-Al2O3 matrix, (d) the proof of
the coexistence of phases with various desired oxidation states of
Finally the addition of Fe0 powder to 2, proves that, indeed, the
oxidation during interaction with the transforming matrix 2
proceeds stepwise. This is a further support of the proposed
interpretation of the findings obtained for the systems {1/Matrix}.
In fact, the formation of 59% FeII and even 21% FeIII due to
oxidation by the matrix (i.e. by in situ liberated water) could be
observed after co-milling of 2 with 1 wt% Fe0 and subsequent TA
heating run under N2. Besides these FeII or FeIII species, protected
inside the matrix in the form of the phase precursors 4 and 5, the
coexisting phase Fe3O4 being detected exclusively under these
conditions was formed as an alternative to the incorporation in
the Al-rich phases.
iron, and finally, (e) the demonstration of netto effects causing the
decrease of the crystallization temperatures of
a-Al2O3 after
mechanical activation of mixtures of 2 with FeIII or Fe0 species.
The choice of the doping component together with an appropriate
temperature-time regime, as well as the required composition of the
gas phase during the TA heating runs allows either distinct product
phases originating from the doping component with particular local
magnetical properties (i.e. Fe0, Fe2O3, Fe3O4, etc.) to be obtained or to

R. Sto¨ßer et al. / Journal of Physics and Chemistry of Solids 72 (2011) 794–799
799
force the incorporation of Fexþ into the phase precursors such as 4 or
) which would be a result of the
[3] A.L. Buchachenko, Russ. J. Phys. Chem. 83 (2009) 1637.
[4] W. Wolski, B. Porawski, J. Therm. Anal. 7 (1975) 139.
[5] W. Wolski, B. Porawski, J. Therm. Anal. 9 (1976) 181.
[6] R. Kniep, Pure Appl. Chem. 69 (1997) 185.
5 up to the x-Al2O3 phases (x¼a,g
surface-constituting influence of the transforming AlOx matrix.
[7] B. Davaasuren, H. Borrmann, E. Dashjav, G. Kreiner, M. Widom, W. Schnelle,
F.R. Wagner, R. Kniep, Angew. Chem. Int. Ed. 49 (2010) 5687.
[8] E. Dashjav, G. Kreiner, W. Schnelle, F.R. Wagner, R. Kniep, W. Jeitschko, J. Solid
State Chem. 180 (2007) 636.
Appendix A. Supplementary material
[9] A. Gudat, R. Kniep, A. Rabenau, Thermochim. Acta 160 (1990) 49.
[10] B. Mohai, L. Bagyin, J. Inorg. Nucl. Chem. 33 (1971) 3311.
[11] M. Verdaguev, G.S. Girolami, in: J.S. Miller, M. Drillon (Eds.), Magnetism:
Molecules and Materials, VCH, Weinheim, 2005, p. 284.
[12] D. Raj, J. Danon, J. Inorg. Nucl. Chem. 37 (1975) 2039.
[13] M. Nofz, R. Sto¨ßer, G. Scholz, I. Dorfel, D. Schultze, J. Eur. Ceram. Soc. 25
(2005) 1095.
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.jpcs.2011.03.016.
References
[14] R. Sto¨ßer, M. Menzel, M. Feist, Hyperfine Interact. 190 (2009) 43.
[15] R. Sto¨ßer, G. Scholz, J.Y. Buzare´, G. Silly, M. Nofz, D. Schultze, J. Am. Ceram.
Soc. 88 (2005) 2913.
[1] N. Hayashi, I. Sakamoto, H. Tanoue, H. Wakabayashi, T. Toriyama, Hyperfine
Interact. 141 (2002) 163.
[16] R. Sto¨ßer, M. Feist, J. Phys. Chem. C 112 (2008) 16438.
[17] R. Sto¨ßer, M. Feist, M. Menzel, M. Nofz, J. Am. Ceram. Soc. 93 (2010) 1447.
[2] R.A. Stukan, A.G. Knizhnik, V.E. Prusakov, A.L. Buchachenko, M.M. Levitskii,
A.A. Zhdanov, J. Struct. Chem. 32 (1991) 830.