632
J. Appl. Cryst.
(1993). 26, 632-635
A Mirror Furnace
for Neutron Diffraction up to
2300 K
By G. LORENZ, R. B. NEOER, J. MARXREITER, F. FREV AND J. SCHNEIOER
Institut ~r Kristallographie und Mineralogie, Theresienstrasse
41, D-8000
Miinchen 2, Germany
(Received
21
January
1993;
accepted 11 March
1993)
Abstract
This paper describes a mirror furnace that has been
developed for neutron diffraction work at tempera-
tures up to 2300K. It is based on a reflecting
rotational ellipsoid, in which the heating element, a
halogen lamp, is placed at one focus and the sample
at the other. It works in a normal,
i.e.
oxidizing,
atmosphere, but can also be used in a vacuum. It has
been developed for experiments of long duration. The
stability and reproducibility of the temperature are
better than 1% of the setting temperature. Further
main characteristics are applicability to single-crystal
and powder work, very low background, low power
consumption and very easy and cheap handling.
Several experiments have been carried out with the
furnace.
Introduction
Structural work at (very) high temperatures is im-
portant. Many naturally occurring compounds,
i.e.
minerals, have structures or microstructures or, more
generally, exhibit order/disorder phenomena that
reflect the thermal prehistory of the rock-forming
process itself. Therefore,
in situ
studies improve
understanding of basic problems of geoscientific
interest. In materials sciences, high-temperature
studies of structures in relation to material properties
are of importance for the practical use, and
understanding and optimization of, the material
processing itself.
Complementary X-ray and neutron scattering
methods both have well known advantages that need
not be discussed further. In particular, the relatively
low absorption of thermal neutrons by most elements
makes the use of neutron furnaces particularly easy,
whereas high-temperature studies at, say, T > 1300 K
with X-rays are usually cumbersome, at least if aimed
at structure refinement. Different methods in X-ray
and neutron high-temperature work are discussed by,
for example, Adlhart, Tzafaras, Sueno, Jagodzinski &
Huber (1982), Aldebert (1984), Ohsumi, Sawada,
Takeuchi & Sadanaga (1984) and Peterson (1992). A
neutron furnace for four-circle diffractometry with
very high temperature stability has recently been
© 1993 International Union of Crystallography
Printed in Great Britain - all rights reserved
reported by Kuhs, Archer & Doran (1993). A
neutron-scattering furnace for very high temperatures,
up to 2930 K, was developed at Harwell (Clausen
et
al.,
1984). It is clear from the literature and it has also
been our experience that different types of heating are
not equally well suited to different types of
(single-crystal, powder) diffraction experiments. A
particular type "of furnace and specific method of
temperature determination and control must be
chosen for a specific scientific problem.
We report here on a neutron mirror furnace that
was developed for use at weak- or medium-flux
neutron sources, for weak diffuse scattering, for
single-crystal or powder work and for use mainly in
an oxidizing atmosphere (Lorenz, 1988; Neder, 1990)
but also in a reducing atmosphere (Mursic, 1992). In
consequence, relatively large samples must be heated,
these must be oriented in the neutron beam and, last
but not least, the spurious background scattering from
the furnace should be kept at a very low level. The
basic principle of mirror heating was developed for
crystal-growth applications (Eyer, Zimmermann &
Nitsche, 1975; Watanabe & Shimazu, 1976). For
in
situ
scattering experiments, this furnace had to be
redesigned. It should be mentioned that the same
principle was also used for a new X-ray powder
furnace (Schneider, 1992; Schneider, Frey, Johnson &
Laschke, 1993).
I. Design
The mirror furnace uses the geometrical properties of
a closed rotational ellipsoid. With a heating element,
a halogen lamp, at one focus, the sample is placed at
the other. Radiation distributions in elliptic and
parabolic mirrors are considered by Hart (1958) and
allow one to find an optimum eccentricity of the
ellipsoid for the specific experimental conditions (sizes
of filaments, sample
etc.).
A more homogeneous
temperature distribution at the sample position and
a reduced maximum load of the lamp can be provided
by the use of two coaxial ellipsoidal mirrors
positioned in such a way that the sample is at one
common focus of both mirrors (Fig. 1). The rotational
axis is vertical to the scattering plane, which is defined
Journal of Applied Crystallography
ISSN 0021-8898 O 1993
G. LORENZ, R. B. NEDER, J. MARXREITER, F. FREY AND J. SCHNEIDER 633
by the incoming and the outgoing (diffracted) neutron
beams. If relatively large samples have to be used for
intensity reasons (see the Introduction), it is advanta-
geous for some axial defocusing, viz two displaced loci
of the two mirrors, to be tolerated. For this reason,
the furnace is made of three parts; the upper and
lower main-mirror parts and one central part. The
height of the latter matches the correct spacing of both
loci at the sample position. The three parts can
quickly be assembled after a change of sample without
the need for further alignment of the furnace. The
body of the furnace is made from aluminium. This
represents a good compromise between ease of
manufacturing and reflecting properties (see §V for
further comments). The central part covers the
scattering plane. Therefore, the aluminium walls of
this central part are as thin as 1.5 mm to avoid any
significant absorption and the diameter is as large as
140 mm to keep as low as possible spurious scattering
from the walls that might enter the detector. Note
that, except for these thin walls a long distance from
the sample position, no part of the furnace is in the
path of the neutrons. Both lamps may be inserted and
adjusted from outside through axial holes at the top
and bottom. A crucial point is the cooling of the lamp
jackets. The glass bodies of the light bulbs become
I
(,4).
/ jP
/ / i
/ /i
/
I - i - II
i
i I
I
Fig. 1. Schematic drawing of the mirror furnace.
mechanically unstable only at relatively high tempera-
tures. As mentioned, the lamps can and must be
adjusted in the focal position of the ellipsoidal mirror
body. There are, however, no important requirements
for accuracy of lamp positioning. Owing to the desired
extension of the focal image, some misalignment of
the extended filaments is of minor importance. A
further important concern is the quality of the
reflecting mirror surfaces. This is also a question of
material. Fortunately, aluminium is an excellent
choice, allowing for high-quality polishing. After
experiments with volatile samples, it is advisable to
repolish the mirrors. Essential improvements can,
however, be made if some advanced and sophisticated
methods of mirror surface treatment are used (see
Mursic, 1992). If necessary, the mirror bodies may be
cooled by simple water cooling. As may be concluded
from this design, there are no fundamental restrictions
with respect to the ambient atmosphere, in other
words, a normal 'oxidizing' atmosphere is as good as
vacuum conditions.
II. Specimen mounting and alignment
Sample support and adjustment are usually provided
by a thin ceramic tube or rod of
AI203 or ZrO2, to
which the sample may be glued with a ceramic cement.
Other possibilities are platinum cans for powder
samples or platinum/rhodium wires sintered around
a ceramic sample. Of course, some ofthis material acts
as an additional source of background scattering.
Minimization of background scattering can be
achieved with single-crystal materials, like MgO
needles, as supports for powder or ceramic samples.
A particularly sophisticated sample holder for
single-crystal work was used by Neder, Frey & Schulz
(1990). There, a cube was cut from a large (sample)
single crystal. On one side, a stem of the original
material was left, which was fixed with ceramic glue
to an Al20 3 rod. The ceramic glue was surrounded
by a BN cylinder that acted as an effective neutron
absorber. In this way, almost no spurious scattering
obscured the diffuse signals to be recorded. In the case
of powder or ceramic samples, only the correct
position at the focus of the mirrors is needed. For
work with single crystals, four-circle Eulerian
diffractometry can be provided with some restrictions
(Fig. 2). In the usual terminology, the crystal support
(ceramic tube or rod) is at a 7, -- 90° position. Tilting
around this position covers a range of _+ 22 ~ with an
accuracy of _+0.01 °. This tilting is performed by a
stepping motor and is computer controlled via an
encoder. The same holds for a q~ rotation around the
rod axis: here the whole range of _ 180 ° is accessible
with an accuracy of +_0.01 °. Because the whole
furnace can be adjusted via translational operations
and some tilt adjustments, the m axis of the
634
MIRROR FURNACE FOR NEUTRON DIFFRACTION UP TO 2300 K
diffractometer coincides with the vertical rotation axis
of the furnace. The full range of the o) position can be
utilized except for a 30 ° sector around the sample
holder. No restrictions apply to 20. Thus, arbitrary
sections from reciprocal space may be measured along
various pathways. Approximately 80% of reciprocal
space can be set into diffraction position (within the
limiting sphere). Alternatively, the furnace can be
placed in a large off-centre Eulerian cradle. This,
however, further limits the X rotation.
III. Temperature control
The temperature determination and control is usually
provided by Pt-Pt/Rh thermocouples of various
kinds, which can be used in an ambient atmosphere
up to 2100 K, and a PID temperature controller. The
thermocouples are placed in holes drilled in the
sample or are even sintered within a ceramic sample.
This procedure provides a precise temperature
measurement, except for gradients within the sample.
The accuracy of the temperature measurement is of
the order of 1% of the nominal one, the stability over
long time scales (of the order of 7 d) is _+ 1 K and the
reproducibility is _+ 5 K. Gradients within the sample
depend on specific factors such as the size or the
thermal conductivity of the sample. As an example,
the gradient within a ZrO2 single crystal, an 11 mm
cube, was about 30 K. It should be mentioned that
reflections from the Pt/Rh elements cause some
unwanted contamination of powder patterns, which
might be cumbersome in the analysis of minority
phases. In many cases, this influence may be ruled out
by excluding regions from these diagrams. Further
details and technical data are given in the Appendix.
+
• x@
Fig. 2. Schematic drawing of the angular degrees of freedom of the
sample within the furnace.
IV. Performance
The furnace was checked in a series of powder and
single-crystal experiments studying either Bragg
reflections or diffuse phenomena. To check the
background scattering by the furnace, an empty scan
was compared to a scan without the furnace. 'Empty
scan' means that the furnace was placed into the beam
together with the sample support,
i.e.
an A120 3 rod,
and a thermocouple, but without the sample crystal.
Fig. 3 shows an almost negligible smooth background
and only very weak reflections from the A120 3
ceramic. With a platinum can, correspondingly
spurious platinum reflections become visible that do
not affect the powder-data analysis in most cases.
They can even be avoided by using sintered samples.
High-temperature structural work with this type of
furnace was very successfully carried out on zirconia
single crystals (Lorenz, Frey, Schulz & Boysen, 1988;
Neder
et al.,
1990; Proffen, Neder, Frey, Keen &
Zeyen, 1993) at different instruments at reactors and
at the ISIS spallation source, and ZrSiO4 single
crystals were studied at temperatures up to 2000 K
(Mursic, Vogt, Boysen & Frey, 1992). Powder and
ceramic samples of, for example, ZrO 2 (Frey, Boysen
& Vogt, 1990; Boysen, Frey & Vogt, 1991), CeO 2
(Berber
et al.,
1991), ZrSiO 4 (Mursic, Vogt & Frey,
1992), NiTiO3 (Lerch, Boysen, Neder, Frey & Laqua,
1992) have been investigated in recent years. Their
great success with this type of furnace encouraged the
ILL at Grenoble to rebuild and sell mirror furnaces
of this type.
7
5 13 21 29 37 45 53 61 69 77 85
20
C~.
5 13 21 29 37 45 53 61 69 77 85
20
Fig. 3. Spurious background scattering at instrument MAN I/FRM.
Above without furnace; below with furnace including sample
support and thermocouple at sample position. Weak spurious
reflections are due to the AlzO3 ceramic support.
G. LORENZ, R. B. NEDER, J. MARXREITER, F. FREY AND J. SCHNEIDER
635
V. Further developments
Further developments of the mirror furnace towards
higher temperatures and various gas atmospheres
were investigated recently by Mursic (1992)(see also
Mursic, Vogt, Boysen & Frey, 1992). One essential
concern is the quality and the tolerable heat load of
the reflecting mirror surfaces. Coating with silver, gold
or even SiO2 is desirable. The main limiting factors
towards higher temperatures are, however, the
mechanical stability of the crystal support and the
cooling of the glass bodies of the light bulbs. Ceramic
refractory materials seem to be very promising. The
bulbs may be cooled in an air stream, which may
easily be realized in an ambient atmosphere. Another
nontrivial aspect concerns temperature measurement
at very high temperatures. In an oxidizing atmos-
phere, measurements by thermocouples are restricted
to temperatures below 2100K (PtRh30/PtRh6);
higher temperatures can only be measured by
pyrometric methods. Optical windows in the mirror
furnace for this purpose are therefore foreseen. In
vacuum, the temperature range up to 2500 K remains
accessible by the use of thermocouples (WRe5/
WRe26). Given this, it would be very interesting to
carry out experiments under different gas atmos-
pheres.
This work was supported by the BMFT, Germany,
project no. 03-B02A04.
APPENDIX
Technical data
Size~weight:
height 600mm; diameter <250mm;
weight 7 kg.
Mirror ellipsoids:
large/small half-axis 90/80 mm.
Central part:
A1Mg5, height 83mm; diameter
< 140 mm.
Material:
AIMg5.
Lamps:
halogen type FEL, 2 x 1000W, 120V
(Schahl Co., Miinchen, Germany); filaments
23 x 7 mm; life time dependent on temperature,
5 d at maximum load.
Cooling"
mirror bodies and lamp jackets water
cooled (1 l min-l); maximum temperature of the
jackets 555 K; maximum temperature of the light
bulbs 1000 K.
Temperature measurement:
thermocouples,
e.g.
EL18 (see text).
Temperature control:
PID-thyristor equipped con-
troller or Eurotherm 822 or 818; selling +_0.1 K,
reproducibility +_
1 K.
References
ADLHART, W., TZAFARAS, N., SUENO, S., JAGOI)ZINSKt, H. &
HUBER,
H. (1982).
J. Appl. C)'yst.
15, 236-240.
ALDEBERT, P. (1984).
Rev. Phys. Appl.
19, 649--662.
BERBER, K., MARTIN, U., MURSIC, Z., SCItNEII)ER, J., BOYSI-N,
H. & FREY, F. (1991).
Mater. Sci. Forum,
79-82, 685-690.
BOYSEN, H., FREY, F. & VO(;T, T. (1991).
Acta
(.'rvsl. B47,
881-886.
CLAUSEN, K., HAYES, W., HUTCHINGS, M. T.,
MACDONAI,D,
J. E., OSBORN, R. & SCHNABEI., P. (1984).
Rer. Phys. Appl.
19, 719-722.
EYFR, A., ZIMMERMANN, H. & NITSCHE, R. (1975).
European
Space Agency Spec. Publ.
114, 241-244.
FREY, F., BOYSEN, H. & VOGT, T. (1990).
Acta (_)')'st.
B46,
724-730.
HART, P. J. (1958).
J. Opt. Soc. Am.
48, 637-642.
KUHS, W. F., ARCHER, J. & DORAN, D. (1993).
J. Appl. Cryst.
Submitted.
LERCH, M., BOYSEN, H., NEDER, R., FREY, F. & LAQUA, W.
(1992).
J. Phys. Chem. Solids',
53, 1153-1156.
LORENZ,
G. (1988). PhD thesis, Ludwig-Maximilians-Univ.,
Miinchen, Germany.
LORENZ, G., FREY, F.,
SCHULZ,
H. & BOYSEN, H. (1988).
Solid
State lonics,
28-30, 497-502.
MURSlC, Z. (1992). PhD thesis, Ludwig-Maximilians-Univ.,
M tinchen, Germany.
MURStC, Z., VOGT, T., BOYSEN, H. & FREY, F. (1992).
J. Appl.
Cryst.
25, 519-523.
MURSlC, Z., VOGT, T. & FREY, F. (1992).
Acta Cryst.
B48,
584-590.
NEDER, R. B. (1990). PhD thesis, Ludwig-Maximilians-
Univ., Mtinchen, Germany.
NEDER, R. B., FREY, F. & SCHULZ, H. (1990).
Acta Cryst.
A46,
799-809.
OHSUMI, K., SAWADA, T., TAKEUCHI, Y. & SADANAGA, R.
(1984).
Material Science of the Earth's" Interior,
pp.
633-643. Tokyo: Terra Science.
PETERSON,
R. C. (1992).
J. Appl. Cryst.
25, 545-548.
PROFEEN, T., NEDER, R. B., FREY, F., KEEN, D. A. & ZEYEN,
C. M. E. (1993).
Acta Cryst.
B49, 599-604.
SCHNEIDER, J. (1992). 41st Annual Denver Conference on
Applications of X-ray Analysis, Colorado Springs, USA,
p. 134.
SCHNEIDER, J., FREY, F., JOttNSON, N. & LASCHKE, K. (1993).
Z. Kristallogr.
Submitted.
WATANABE, A. & SHIMAZU, M. (1976).
J. Appl. Cryst. 9,
466-469.