Numerical and Experimental Investigations on the

Stability of a Wavefront Interferometer Support

Structure
An accurate thermal-mechanical stability predictive approach
Nicolas Jobert, Victor Pinty, Paulo Dasilva, Pascal Mercere
Synchrotron SOLEIL - Accelerators & Engineering Division
nicolas.jobert@synchrotron-soleil.fr — (33) 1 69 35 91 45

Abstract
In this paper we will cover the investigations efforts carried out in order to meet stability requirements of a
Results
Wave Front Sensor support structure, to be installed and used in the “Métrologie” Beamline of SOLEIL. The active
components (gratings, lens and camera) should not have more than 10nm motion relative to each other. Hence, the
Residual temperature fluctuation in the hutch have been recorded (about 0.03degC RMS), and the
design objective can be very simply stated: provide a support structure that allows alignment of active components, frequency content calculated (Fig2). Then, the (complex) thermal mechanical frequency response of
but also the possibility of maintaining position and allowing reproducibility of measurements within aforementioned the structure has been estimated, for all points and all frequency between 1 and 1440 cycles per day
limits over periods of months or years. So, the design objective is to obtain a rigid body, both from vibrational and (Figure 3).Finally, the alignement error between CCD and lens has been evaluated (Fig 4). Since air
thermal elastic point of view. The additional difficulty here is that the entire setup has to be easily removable, flow conditions are variable in the hutch, analysis has been repeated for low, intermediate, and high
placing stringent limits on total weight and volume, while still ensuring overall reproducibility.
An invar support structure has been designed and constructed, then tested in the Métrologie Beamline hard
convection coefficient values (5,10 and 15 W/m2/K).
X-rays hutch. Investigations efforts covered both dynamic and thermal elastic aspects, putting the emphasis on a
step-by-step comparison of analysis and testing results. This allowed to double check the major assumptions, and
confirm the absence of discrepancies between expected and actual results. Efforts to quantify significant sources
of uncertainties where made, covering numerical and experimental approaches. The external disturbances being
random, a frequency domain approach is applied, and the final reliability of the structure is found to be acceptable
without any specific insulation measure. This seemingly counter-intuitive conclusion exemplifies the usefulness of
relying on a consistent set of design rules and effective analysis tools.

Introduction
Located at the French National Synchrotron Radiation Source (SOLEIL) , the Métrologie beamline is
dedicated to the characterization of optical components ([1]). For 3rd and 4th generation sources, op- Figure 2: Thermal Environment
tics are limiting factors to the conditioning of X-ray beams. At high energy and small grazing angles,
optical path errors are in the order of picometers, and optical surfaces quality can only be assessed
based on the well known Shack-Hartmann principle.

Grating YAG Lens CCD

Pads

Baseplate

Figure 1: System Overview

Since this setup will be used as a calibration system, it will only be put in place at intervals of
months or years. This places a number of conflicting requirements on the design, since the setup must
be immune against external disturbances and simultaneously sufficiently lightweight. Figure 3: Thermal Mechanical Response at 1 (top) and 1440 (bottom) cycles per day (nm/K) for intermediate Hconv value
Additionnaly, it is not practical to directly measure displacements with amplitude of a few nanome-
ters, hence validation of the setup can only be made using an indirect approach, using combined
experimental and numerical investigations.

Main Objectives and Difficulties

Two major disturbance sources will govern: firstly, structural vibrations (passive control using flexi-
ble pads) and secondly, external temperature variations (ambiant temperature is actively controlled).
Hhere we focus exclusively on the second point. Our objectives are:
1. Estimate and understand thermally induced misalignment (CCD vs grating)
2. Quantify instrument reliability, estimate associated uncertainties Figure 4: Calculated misalignment thermal sensivity (left) and Experimental setup (right)
3. Design ”as light as reasonably achievable” isolation system The alignment error PSD not only gives access to the
RMS amplitude, but also to valuable information such as Hypothesis σerror [nm] T[min]
From a design point of view, this task is made arduous by two conflicting user’s requirements, firstly
a 10nm reproducibility and secondly a short settling time. Also, air velocity around the support is not the statistical period. Hence, it is possible to estimate the Low Hconv 0.38 5.2
precisely known, hence convection efficiency is expected to be time dependent. variance reduction obtained via averaging. In this case, Mid Hconv 0.71 5.3
the RMS amplitude is about 1nm, hence the 10nm crite- High Hconv 1.08 5.3
rion is expected to be respected.p The expectancy of po-
sitionnal error is E(error) = (5.2/Tavg )σerror where
Method Tavg is the averaging time (units: minutes). Using mea-
Table 1: Misalignment amplitude/period

General
1. Record ambiant temperature in room Text, estimate corresponding PSD ΦText
2. Compute thermal dynamic transmissibility HT , then statically apply temperature response to de-
rive structural transmissibility HT S
3. Estimate system misalignement response PSD ΦErrorij = |(HT Si − HT Sj )|2ΦText , then integrate
to obtain misalignement error RMS amplitude and statistical frequency
Mathematical Development
For a multi-DOF system, response to an harmonic excitation can be obtained using:
sured thermal transfer function, it was shown that system thermal time constant is about 10000s,
compared with 7910s as obtained by using thermal modal analysis ([2],[3]) for Hconv = 10W/m2/K.
Conclusions
• A high-fidelity tool was required to estimate the severity of ambiant temperature variations on the
alignment performance of a wavefront (Shack-Hartman) interferometer
• A general frequency domain analysis procedure has been developped, using ANSYS APDL Math
• This allowed to retain the detailed model while having fast results, hence making it possible to run
sensitivity analyses on uncertain physical parameters

(K + jωC)HT (ω) = P (1)

Where:
• K (resp. C) is the conductivity matrix (resp. capacitance) matrix (n × n)
• P is forcing function vector (thermal nodal power) (n × 1 column vector)
• HT is the thermal transmissibility vector (complex valued) (n × 1 column vector)
References
[1] Mourad Idir, Pascal Mercere, Thierry Moreno, Aurélien Delmotte, Paulo Dasilva, and Mo-
hammed H Modi. Metrology and tests beamline at soleil design and first results. In AIP Con-
ference Proceedings, volume 1234, pages 485–488. AIP, 2010.
[2] Nicolas Jobert. Working wonders using apdl math: Thermal modal analysis. PADT, 2017.
[3] Takeshi Morishima. Novel thermal error reduction techniques in temperature domain. 2016.