IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS

The application of interference fits for overcoming limitations in clamping

methodologies for cryo-cooling first crystal configurations in x-ray
monochromators
To cite this article: J Stimson et al 2017 IOP Conf. Ser.: Mater. Sci. Eng. 278 012055

View the article online for updates and enhancements.

This content was downloaded from IP address 206.223.250.2 on 29/12/2017 at 12:21

CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

The application of interference fits for overcoming limitations

in clamping methodologies for cryo-cooling first crystal
configurations in x-ray monochromators

J Stimson1, P Docker2, M Ward1, J Kay2, L Chapon2 and S Diaz-Moreno2

1
Birmingham City University, Curzon Street, Birmingham, United Kingdom, B4 7XG
2
Diamond House, Harwell Science and Innovation Campus, Fermi Ave, Didcot OX11
0DE

joshua.stimson@bcu.ac.uk

Abstract. The work detailed here describes how a novel approach has been applied to overcome
the challenging task of cryo-cooling the first monochromator crystals of many of the world’s
synchrotrons’ more challenging beam lines. The beam line configuration investigated in this
work requires the crystal to diffract 15 Watts of 4-34 keV X-ray wavelength and dissipate the
additional 485 watts of redundant X-ray power without significant deformation of the crystal
surface. In this case the beam foot print is 25 mm by 25 mm on a crystal surface measuring
38 mm by 25 mm and maintain a radius of curvature of more than 50 km. Currently the crystal
is clamped between two copper heat exchangers which have LN2 flowing through them. There
are two conditions that must be met simultaneously in this scenario: the crystal needs to be
clamped strongly enough to prevent the thermal deformation developing whilst being loose
enough not to mechanically deform the diffracting surface. An additional source of error also
occurs as the configuration is assembled by hand, leading to human error in the assembly
procedure. This new approach explores making the first crystal cylindrical with a sleeve heat
exchanger. By manufacturing the copper sleeve to be slightly larger than the silicon crystal at
room temperature the sleeve can be slid over the silicon and when cooled will form an
interference fit. This has the additional advantage that the crystal and its heat exchanger become
a single entity and will always perform the same way each time it is used, eliminating error due
to assembly. Various fits have been explored to investigate the associated crystal surface
deformations under such a regime

1. Introduction
A synchrotron device is a cyclic particle accelerator. Electrons are raised to considerable speed by a
linear accelerator before being injected into an initial “booster ring”, a circular beam path with pulsed
magnetic fields in which the particle is further accelerated. Once the electron has reached near-
relativistic speeds, it is ejected into a larger “storage ring”, in which the velocity of the electron is
maintained. At intervals around the storage room there are insertion devices, each of which consists of
a series of high strength opposed dipoles, which undulate the electron’s path. As the electron is moving
at relativistic speeds, whenever it is accelerated it emits the excess energy in the form of a photon. The
high velocity of the electron coupled with the minor permutations of the path induced by the insertion

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

device causes the electron to remain close to the emitted photons, resulting in the emitted light falling
into the X-ray range. (1–3)
This raw light is produced as a broad spectrum, from which one wavelength is selected via Bragg
diffraction (4) using a series of single crystal silicon monochromator as shown in figure 1. These silicon
crystals diffract the selected X-ray wavelength, dependent on the angle of incidence, and absorb the rest
of the spectrum. As these synchrotron facilities are upgraded these first crystals are expected to absorb
more power to the point that the current cooling methods have reached a plateau where any further
increase to beam power results in significant thermal deformation of the diffracting surface (5).

Monochromatic Light

Broad Spectrum Light

Silicon Monochromator Crystals

Figure 1. The path of a beam through the monochromator series.

There are two main approaches to cooling the silicon crystals: they are most often cooled indirectly,
through contact with a heat exchange unit with a through flow of liquid nitrogen; in some cases they are
cooled directly, by flowing liquid nitrogen through the body of the crystal itself. Indirect cooling has
become the industry standard mainly because heat exchangers are easily constructed, whereas direct
cooling requires delicate seals which are only functional for a finite number of cycles before they must
be replaced, resulting in far greater costs (6). Until now indirectly cooling the monochromators has
been sufficient, but with rising power levels many facilities are looking for new solutions to handle
higher heat loads.
The efficiency of indirectly cooling the crystals is dominated by the force applied (7), so that the best
way to improve indirect cooling efficiency is to increase the force applied. This is due to only a small
number of asperities on the two surfaces actually coming into contact; increasing the force applied
deforms these asperities, allowing more to contact. Thus, in order to cool the crystal in the most efficient
fashion we need to apply greater and greater forces. However, these forces can in themselves
mechanically deform the diffracting surface, potentially causing a greater issue than the heat itself. This
means that increasing the clamping forces reduces thermal deformation, but increases mechanical
deformation. This suggests that there would be a point where the mechanical deformation overcame the
thermal as the dominant deformation, although a literature search found no documentation on such an
occurrence.
The final factor to consider for such a cooling system is the distribution of the stress field. If the
stress field is applied in a non-uniform pattern, it would cause a non-uniform deformation of the
monochromator crystal before any thermal load was even applied (8). As such it is vital that the stress
field is uniform.
A current limitation of the system is that the physical assembly of the monochromator imparts
multiple stress vectors of varying magnitude and direction, through tightening bolts, attaching the liquid
nitrogen manifold and differing thermal expansion coefficients of different materials. Any new system
should be adapted to avoid these shortfalls, by not tightening the assembly with bolts, applying the liquid
nitrogen manifold parallel to the diffracting surface and minimizing or utilizing the varying thermal
expansion characteristics of the materials used.

2
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

2. Method
In order to produce a high intensity, uniform stress field an interference fit was modelled using the
COMSOL MultiPhysics Finite Element Modelling software, with the copper heat exchanger designed
as a sleeve around the silicon crystal as shown in figure 2.

Figure 2. An interference fit monochromator design with copper sleeve as the hub component and
single crystal silicon shaft as the shaft component.

An interference fit is created when two objects attempt to occupy the same space, interfering with
each other. In most cases it is achieved by heating the hub component, expanding it thermally, and then
slotting it over the shaft component. As the hub cools, it shrinks, bringing it into contact with the shaft
component (9). However, with two dissimilar materials such as copper and silicon, it becomes more
complex. First, copper has a thermal expansion coefficient an order of magnitude greater than that of
silicon. Secondly, silicon experiences a negative thermal expansion coefficient beneath 120 K, causing
it to expand when the silicon is cooled. These thermal expansion coefficients are shown in figure 3 (10–
12).
Thermal Expansion Coefficients of Copper and Silicon
18
16
Expansion Coefficient (10-6/K)

14
12
10
8
6
4
2
0
-2 0 50 100 150 200 250 300 350
Temperatiure (K)

Silicon Copper

Figure 3. Thermal expansion coefficients of Silicon and Copper.

3
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

Due to the extreme variation between the two coefficients, it is possible to leave a gap at room
temperature and still achieve a high pressure interference fit after cooling. In order to identify whether
there should be a gap, and if so what size, the equation below (9) was used to model the pressure
achieved during cooling at bare contact and at a range of room temperature gap values analytically in
SciLab, the output of which is shown in figure 4.

𝑑𝑠𝑜 − 𝑑ℎ𝑜 + Δ𝑇(𝛼𝑠 − 𝛼ℎ )𝑑ℎ𝑖

𝑃= 2 2 2 2
𝑑ℎ𝑖 𝑑ℎ𝑜 + 𝑑ℎ𝑖 𝑑𝑠𝑜 𝑑𝑠𝑜 + 𝑑𝑠𝑖
𝐸ℎ ∗ ( 2 2 + 𝜈ℎ ) + 𝐸𝑠 ∗ ( 2 2 − 𝜈𝑠 )
𝑑ℎ𝑜 − 𝑑ℎ𝑖 𝑑𝑠𝑜 − 𝑑𝑠𝑖

(1)
Where:
𝑑ℎ𝑖 is the inner diameter of the hub 𝜈ℎ is the Poisson’s ratio of the hub
𝑑ℎ𝑜 is the outer diameter of the hub 𝜈𝑠 is the Poisson’s ratio of the shaft
𝑑𝑠𝑖 is the inner diameter of the shaft Δ𝑇 is the temperature difference between assembly
𝑑𝑠𝑜 is the outer diameter of the shaft and operation
𝐸ℎ is the Young’s modulus of the hub 𝛼𝑠 is the thermal expansion coefficient of the shaft
𝐸𝑠 is the Young’s modulus of the shaft 𝛼ℎ is the thermal expansion coefficient of the hub

4e08

3e08

2e08
Pressure, Pa
1e08

0e00
100
80
60 250 300
40 200
20 100 150
Gap, μm 0
Temperature, K

Figure 4. Pressure as a function of temperature and gap at room temperature.

To determine the pressure we wanted to induce in the system we considered three factors. First we
compared the ultimate compressive stresses of silicon and copper in literature, coming to values of
700 MPa (11) and 358 MPa (12) respectively. As the value for silicon is nearly twice that of copper,
we can be confident that the copper sleeve will plastically deform around the silicon, rather than crushing
it.
Next, in order to maintain a flat diffracting surface we needed to ensure that the force of linear
expansion produced by the silicon’s expansion, as shown in equation 2, would be greater than the friction
caused by the copper clamping on the silicon, as shown in equation 1 above. (13). Equation 2 allows us
to calculate the force generated from the thermal deformation of either material should they be
constrained and unable to move. By solving equation 2 simultaneously for the silicon and copper
components we are able to find the axial clamping force. If the axial clamping force exceeds the linear
expansion force, the interference face of the silicon would be restrained while the core is relatively free,
resulting in a bulge in the center of the diffracting face; whereas if the expansion force were to be greater

4
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

the silicon interference face would slip against the copper. We can tune this axial clamping force by
leaving a gap between the two materials at room temperature, as shown in Figure 4 above.
𝐹 = 𝛼𝑌𝐴Δ𝑇
(2)
Where:
𝐹 is the force produced 𝑌 is the Young’s Modulus of Elasticity
𝛼 is the coefficient of thermal expansion 𝛥𝑇 is the change in temperature

Finally we had to consider the performance of the current system; to produce a comparable thermal
transport we would have to at least match the clamping pressure of the existing system, which is set at
190 N.

3. Results
Solving equations 1 and 2 together we came to an optimal room temperature gap value of 60 μm in order
to give us the highest possible pressure without mechanically constraining the interference face of the
silicon. We then built another FMA model in COMSOL using this gap, the boundary conditions of
which are shown in Figure 5. By assigning the bottom-most surface as unable to move in the z-direction
and a central axis as unable to move in the x- or y- directions, we produce a stable and well-constrained
model that is subject to thermal expansion. We then held the inner surfaces of the hollow copper sleeve
constant at 80 K to simulate the flow of liquid nitrogen, and all outer surfaces of the copper as radiating
to a temperature of 80 K to simulate a heat shield.

(a)

(b)

Figure 5. Copper and silicon interference fit model with a 60 μm gap at room temperature.
a) Mechanical boundary conditions, left to right: Shaded surface subject to prescribed displacement
(z = 0); Bold line subject to prescribed displacement (x & y = 0).
b) Thermal boundary conditions, left to right: Shaded surfaces subject to fixed temperature (80 K);
Shaded surfaces subject to diffuse surface (to 88 K).

5
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

After the model reached a steady state, effectively cooling from 300 K to 80 K, the resulting surface
profile is shown in Figure 6. The entire surface is pushed up 5.2 μm, and the edge is raised a further
0.1 μm. The surface is also very uniformly deformed, due to the circular stress field acting on a circular
crystal.
Surface: Displacement field, Z component (mm) Max/Min Surface: Total displacement (mm)

x10-3

max: 0.00533413

min: 0.00520101
5.2

Figure 6. Deformation of diffracting surface after applied mechanical pressure.

We compared this surface deformation of our new design to one generated by a model of the existent
system, as shown in Figure 7.

(a)

(b)

Figure 7. Copper and silicon standard assembly model.

a) Mechanical boundary conditions, left to right: Shaded surface subject to fixed constraint; Shaded
surface subject to prescribed displacement (z = 0); Shaded volume pretensioned (190 N).

6
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

b) Thermal boundary conditions, left to right: Shaded surface subject to fixed temperature (80 K);
Shaded surface subject to diffuse surface (to 88 K).
Figure 7 shows the first crystal monochromator assembly currently used at the I20 beamline at
Diamond Light Source in Oxford, with a copper base-plate and heat exchange attached to two silicon
crystals by way of two stainless steel bolts. Matching boundary conditions were used where possible,
with the bottom surface secured in the z-direction and the fastening bolt for the assembly treated as a
true fixed domain. The bolts in this model were pre-tensioned with a 190 N force to match standard
operating procedure. Again, the inside of the heat exchangers were fixed at 80 K to simulate the flow
of liquid nitrogen, and all outer surfaces were treated as radiating to 80 K to simulate a heat shield.
The surface profile of one of the silicon crystals is shown in Figure 8. The surface is bent in a “V”
shape by the differing expansion coefficients of the various materials, with the center being pulled down
by 17.8 μm and the outer corners being raised by 2 μm. The deformation varies significantly along the
x-axis, due to the differing materials exerting different stress profiles on the crystal.

Surface: Displacement field, Z component (mm) Max/Min Surface: Total displacement (mm)

-3
x10
0.2
0
-0.2
-0.4
min: -0.0178104
-0.6
max: 0.00203977
-0.8
-1
-1.2
-1.4
-1.6

Figure 8. Deformation of diffracting surface after applied mechanical pressure.

4. Discussion
The results show that an interference fit is capable of applying enough force to parallel that used in the
existing system without applying such force that the diffracting surface would be significantly
mechanically deformed. This means an interference cooled crystal will be capable of giving us a
reduced surface deformation compared to a standard rectangular crystal and heat exchange assembly,
due to the far more uniform stress field and higher force between the crystal and the heat exchange. This
would allow us to use higher beam powers without deforming the diffracting surface such that the beam
is lost.
In addition, by clamping the crystal using an interference fit we would not be introducing strains and
rotations to the assembly as are conventionally incurred by manually assembling the system using bolts.
It will also give us a far more precise control over the clamping pressure than achieving it by turning
nuts. The use of fewer materials, and the application of the differing material properties to create the
clamping, results in a more uniform temperature profile.

5. Further Work
In light of the models presented in this work, we are now in the process of manufacturing heat
exchangers to fit cylindrical silicon crystals. A Fizeau interferometer will be used to measure the surface
deformation imparted by interference fits of varying magnitudes.

7
CEC 2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 278 (2017) 012055 doi:10.1088/1757-899X/278/1/012055
1234567890

In addition, we will be exploring the application of the interference fit to produce a gas-tight seal by
decreasing the gap, possibly even to the extent of interference fitting above room temperature. The
pressures involved should be able to produce a gas-tight seal without significantly deforming the
diffracting surface of the crystal. To test this theory, the heat exchange sleeves will be tested to see if
they are able to hold a gas-tight seal.

6. References
1. McMillan EM. The Synchrotron – A Proposed High Energy Particle Accelerator. Phys Rev.
1945;68(143):1–4.
2. Wilson EJN. Fifty years of synchrotrons. In: 5th European Particle Accelerator Conference.
Spain; 1996. p. 135–9.
3. Sham TK, Rivers ML. A Brief Overview of Synchrotron Radiation. Rev Mineral
Geochemistry. 2002;49(1):117–47.
4. Bragg WH, Bragg WL. The Reflexion of X-rays by Crystals. In: Proceedings of the Royal
Society of London. The Royal Society; 1913. p. 428–38.
5. Zhang L, Sánchez Del Río M, Monaco G, Detlefs C, Roth T, Chumakov AI, et al. Thermal
deformation of cryogenically cooled silicon crystals under intense X-ray beams: Measurement
and finite-element predictions of the surface shape. J Synchrotron Radiat. 2013;20(4):567–80.
6. Marion P, Zhang L, Goirand L. Cryogenic cooling of monochromator crystals: Indirect or
direct cooling? In: Proceeding MEDSI 2006. Grenoble; 2006.
7. Salerno LJ, Kittel P. Thermal contact conductance. J Mater Process Technol. 2003;135(2–
3):204–10.
8. Malvern LE. Introduction to the mechanics of a continuous medium. Prentice-Hall; 1969. 713
p.
9. Bower AF. Applied Mechanics of Solids. 1st ed. 2011. 820 p.
10. Shah JS, Straumanis ME. Thermal expansion behavior of silicon at low temperatures. Solid
State Commun. 1972 Jan;10(1):159–62.
11. Petersen KE. Silicon as a Mechanical Material. Proc IEEE. 1982;70(5):420–57.
12. National Institute of Standards and Technology (NIST). Properties of Copper and Copper
Alloys at Cryogenic Temperatures. 1992;Monograph:270. Available from:
http://nvlpubs.nist.gov/nistpubs/Legacy/MONO/nistmonograph177.pdf
13. Moorefield D. Precision machine design. Vol. 15, Precision Engineering. 1993. p. 51–2.