Transcript Josep Nicolas ALBA synchrotron light Facility Optics

Optics Metrology at ALBA
Josep Nicolas
ALBA synchrotron light Facility
[email protected]
www.cells.es
Optics Metrology at ALBA
Josep Nicolas
ALBA synchrotron light Facility
Outline
1. The Optics Lab
2. The NOM instrument
3. Some applications
[email protected]
www.cells.es
Mirrors and gratings must be measured…
In a 3rd generation light source, slope
error of optics is usually the limit for
optical performance of the beamline:
• Flux density on SAMPLE
• Spectral resolution
• spot size on sample
Cff=2 D4 = 4
20000
Resolution Limit
Mirrors and gratings for beamlines are
critical:
• custom made for the beamline.
• have long delivery time.
• Expensive.
• One cannot have spares.
M3 ast coma
15000
10000
5000
250 l/mm
400 l/mm
150 l/mm
2000
100
500
1000
1500
2000
Energy (eV)
Contribution to the spectral
resolution limit for a PGM, with
slope errors of 0.5 μrad.
www.cells.es
…
Most mirror polishers cannot measure
mirror with accuracy better than 0.5 μrad
There are sources of error that canot be
controlled by the polisher
• Gravity sag
• Holder clamping
• Cooling manifolds
• Calibration of mirror bender
• Thermal bump
• Bending approximation
Twist introduced by the clamping
onto the mirror surface.
Also, a metrology lab opens the possibility to improve optics.
“If you can’t measure it, you can’t make it.”
www.cells.es
Mirror 400 mm (toroidal)
Mirror 1200 mm (plane elliptic)
Diffraction grating (170 mm)
37 mirrors and diffraction gratings
14 bent onto plano elliptic
7 toroidal
7 longer than 600 mm
>85% better than 1 μrad
>50 % better than 0.5 μrad
Maximum radius specified 22 km
Minimum radius (sagittal) 35 mm
Best measured error 55 nrad
About 30% were
initially out of specs
www.cells.es
17 m2
40 m2
Metrology Lab
21±0.5ºC
Optics lab.
The optics and metrology lab is one
of the perimetral labs of the
expermiental hall.
STM+AFM
50 m2
Optics Lab
21±0.2ºC
Class
10.000
Annex
www.cells.es
FIZEAU
INTERFEROMETER
NOM enclosure
NOM
www.cells.es
Fizeau interferometer
Alba-NOM
Provides 2D information quickly.
Difficult to measure curved surfaces.
Limited accuracy.
Provides 1D information
Can measure curved surfaces
Ultimate accuracy
instrument angle error
0.2
R=75 m
36 nrad RMS
angle error (rad)
0.1
0
-0.1
-0.2
-40
-30
-20
-10
0
10
position (mm)
20
30
40
www.cells.es
The Alba NOM
www.cells.es
The Alba NOM
A scanning deflectometer
meridional slope only
One line only
Versatile (any figure)
Ultimate accuracy
www.cells.es
The Alba NOM
Sources of error
- Unstability. Vibrations, turbulence
- Guidance error and pentaprism
alignment
- Autocollimator errors
www.cells.es
The NOM is placed a passive isolation enclosure, with 100 mm thick high density
polyuretane walls. Power disipation inside is minimized.
The lab itself is stablized to <0.3ºC (upgrade pending)
ΔT < 50 mK in 4 days
www.cells.es
Ground
vibration
(vertical)
2
10
0
10
Spectral Amplitude (nm RMS)
Amplitude spectral density (nm/Hz)
Although the amplitude of the vibration in the ground is 50 nm RMS, the
measurement on top of the NOM is 13 nrad RMS
100
10
1
0.1
20
40
60
80
Frequency (Hz)
100
20
40
60
80
Frequency (Hz)
100
NOM
(pitch
between
platforms)
13 nrad RMS in 6:40 min
www.cells.es
The pentaprism compensates most of the noise induced by vibrations, being
turbulence the remaining source of stability
Noise along the optical path of the beam,
and the corresponding spectrum
www.cells.es
The use of a differential interferometer allows an accurate characterization of the
motion performance of the bench
Position
Backlash
Accuracy
Repeatability
Resolution
100 nm
3.5 um
77 nm
20 nm
Resolution, a 40 nm step
Positioning error
1.5
40 nm
Y position error (m)
1
0.5
0
-0.5
-1
-1.5
-2
0
1
2
3
4
5
motor position (kec)
www.cells.es
6
4
x 10
Backlash
Accuracy
Repeatability
pitch
z
2 nrad
7.1 urad
200 nrad
yaw
x
390 nrad
10.1 urad
94 nrad
<115 nrad
6.1 urad
<200 nrad
z
14 nm
1.1 um
42 nm
straightness x
2 nm
2.1 um
47 nm
roll
flatness
3
10
0.4
0.5
0
-1
-2
-3
6
4
0.2
Flatness error (m)
1
Flatness error (m)
8
Yaw error (rad)
Pitch error (rad)
2
0
-0.2
-0.4
-0.6
2
0
-0.5
-1
-0.8
-4
0
0.2
0.4
0.6
0.8
1
motor position (m)
Pitch
1.2
0
0.2
0.4
0.6
0.8
1
motor position (m)
Yaw
1.2
0
0.2
0.4
0.6
0.8
1
motor position (m)
Flatness
1.2
0
0.2
0.4
0.6
0.8
1
motor position (m)
1.2
Straightness
www.cells.es
50
50
0
y
50
Roll
Yaw
50
0
-50
-50
z
0
-50
-50
z
Pitch
z
z
50
0
y
50
0
-50
-50
0
y
Flatness
Double pass raytracing of the scanning
pentaprism is used to determine the influence of
guidance error on the measurement
0
-50
-50
0
y
50
www.cells.es
50
R = 100 m
150
Roll - 0 nrad
Pitch - 2 nrad
Yaw - 0 nrad
Straightness - 0 nrad
Flatness - 5 nrad
Position - 17 nrad
Total - 17 nrad
tracing error (nrad)
100
50
0
-50
0
200
400
600
800
Position (mm)
1000
1200
Contribution of the guidance error to the LTP error
www.cells.es
Pentaprism error – R=100 m
Different spheres
250
200
tracing error (nrad)
tracing error (nrad)
R=100 m 17 - nrad RMS
150 R=200 m 8 - nrad RMS
100 R=500 m 3 - nrad RMS
50
0
-50
-100
A =-0.50 mrad - 57 nrad
A =-0.25 mrad - 34 nrad
A =+0.00 mrad - 17 nrad
 =+0.25 mrad - 23 nrad
100 A
A =+0.50 mrad - 45 nrad
200
R=50 m 33 - nrad RMS
0
-100
-200
0
200
400
600
800
Position (mm)
1000
0
1200
200
400
600
800
Position (mm)
1000
1200
Pentaprism roll – R=100 m
Different spheres – Position LUT
250
150 R=200 m 3 - nrad RMS
tracing error (nrad)
tracing error (nrad)
R=100 m 6 - nrad RMS
100 R=500 m 1 - nrad RMS
50
0
-50
-100
roll=-2.00 mrad - 14 nrad
roll=-1.00 mrad - 9 nrad
roll=+0.00 mrad - 17 nrad
100 roll=+1.00 mrad - 32 nrad
roll=+2.00 mrad - 66 nrad
200
R=50 m 11 - nrad RMS
200
0
-100
-200
0
200
400
600
800
Position (mm)
1000
1200
0
200
400
600
800
Position (mm)
1000
www.cells.es
1200
Pentaprism error – R=100 m
Different spheres
250
200
tracing error (nrad)
tracing error (nrad)
R=100 m 17 - nrad RMS
150 R=200 m 8 - nrad RMS
100 R=500 m 3 - nrad RMS
50
0
-50
-100
A =-0.50 mrad - 57 nrad
A =-0.25 mrad - 34 nrad
A =+0.00 mrad - 17 nrad
 =+0.25 mrad - 23 nrad
100 A
A =+0.50 mrad - 45 nrad
200
R=50 m 33 - nrad RMS
0
-100
-200
0
200
400
600
800
Position (mm)
1000
0
1200
200
400
600
800
Position (mm)
1000
1200
Pentaprism roll – R=100 m
Different spheres – Position LUT
250
150 R=200 m 3 - nrad RMS
tracing error (nrad)
tracing error (nrad)
R=100 m 6 - nrad RMS
100 R=500 m 1 - nrad RMS
50
0
-50
-100
roll=-2.00 mrad - 14 nrad
roll=-1.00 mrad - 9 nrad
roll=+0.00 mrad - 17 nrad
100 roll=+1.00 mrad - 32 nrad
roll=+2.00 mrad - 66 nrad
200
R=50 m 11 - nrad RMS
200
0
-100
-200
0
200
400
600
800
Position (mm)
1000
1200
0
200
400
600
800
Position (mm)
1000
www.cells.es
1200
Redundant-independent datasets
We use an error model to combine redundant-independent measurements of one
mirror, and that allow to estimate the slope profile and the linearity error of the
instrument.
• The results are as accurate as the error model.
• We have estimated by Monte Carlo simulation the dependence on
measurement parameters
www.cells.es
A 100 mm, R=75 m sphere has been used to cover 12 mrad range of the NOM.
Each reconstruction is based on 54 scans
Residual aberration + periodic subpixel interpolation error found for two different
roll positions ofbestthe
pentaprism.
fit surface slope
10
roll A
roll B
slope (rad)
5
0
-5
-10
-50
0
position (mm)
50
R=75 m
36 nrad RMS
www.cells.es
Facing up
Facing side
Facing down
The ALBA NOM can be set to measure mirrors
in working orientations
It is integrated to the TANGO control system
Fully automated.  1700 scans in 2010 in
about 1600 hours (2.5 km)
www.cells.es
The Alba NOM
Applications
1. Site acceptance tests of purchased optics
2. Collaboration with mirror manufacturers
3. Installation of optics on their holders
4. Calibration of mechanical benders
5. Optimization of figure error
www.cells.es
Plano elliptic
Facing down
R0 = 129 m
L = 800 mm ( 6 mrad)
0.234 μrad RMS
Profile Error
20
The mirror figure was optimized by spring
actuators.
Accuracy is required for convergence of the
optimization process, on both the
measurement and on the spring adjustment
Profile Error (nm)
15
10
5
0
-5
-10
-15
-500
0
position (mm)
www.cells.es
500
By introducing a deformation of the mirror bulk, one can partially compensate the polishing
error of the mirror.
Mirror before optimization
Polished surface
with slope error
Mirror bulk
It requires:
• Accurate metrology
Mirror after optimization
• Accurate deformation model
Pulling forces
• Accurate mechanics
Elastically deformed
mirror bulk
Pushing forces
www.cells.es
• Separate bending (motorized) from
correction (weak adjustable springs)
• Optimize forces and positions
• Use as few actuators as possible
• Minimization
boundaries
conditions)
using simplex
and
random
Final Slope error
• Based on the elastic beam theory.
0.048
0.1 N error
0.046
Without Error
0.044
0.042
0.04
0.038
(with
initial
• Iterative (because spring force cannot
be directly set)
0.036
0
2
4
6
Number of actuators
8
10
Achieved slope as a function of the
applied actuators, with error on the
applied forces.
www.cells.es
300 mm long mirror by InSync
on a 2-couple bender by Irelec.
Springs
www.cells.es
Before optimization
After optimization
Center stripe improved from 0.242 μrad RMS to 0.055 μrad RMS with 2 actuators
Agreement between prediction and measurement is 0.18 nm RMS
www.cells.es
A second mirror was optimized:
•600 mm long mirror
•Horizontal deflection
•4 spring actuators (pulling and pushing)
Profile Error
•
•
•
15
Profile Error (nm)
10
5
0
-5
Initial error: 0.211 μrad RMS
Final error: 0.083 μrad RMS
Limited by the poor control of the
applied force by the springs
(designed for corrections in the order
of 0.5 μrad)
-10
-15
-300
-200
-100
0
position (mm)
100
200
300
www.cells.es
Beamline measurements indicate that the spot can be expanded vertically
by a factor 20 with very little inhomogeneities.
Defocused beam
10 μm
110 μm
Focused beam
50μm
220 μm
The YAG diagnostics does not allow an accurate measurement of the spot
size when focused.
www.cells.es
The slope error of the VFM was measured using the pencil beam method. Showing
excellent agreement with the figure measured in the laboratory 2 years before and
after many bending-unbending, cycles.
𝑧
𝑢=
𝑐𝑜𝑠𝛼
Residual Profile (nm)
∆𝑧 = 2𝑞 ∆𝑛(𝑢)
4
NOM
In situ
3
0.055 μrad
0.045 μrad
2
1
0
-1
-2
-150
-100
-50
0
Position (mm)
50
100
www.cells.es
150
Bending actuators
Bending actuators include high
resolution strain gauge to provide
reliable feedback on the bent ellipse.
•
Stepper motors
•
Encoder+homing
•
Strain Gauges
Spring actuators
•
Correction independent of
ellipse
•
High force resolution
•
High dynamic range
•
Minimum distance 22 mm
•
Pulling or pushing.
•
Motorized for optimization
at the NOM
www.cells.es
First prototype finished last week.
Measurements in progress
www.cells.es
Optical metrology is essential to guarantee the optimal
performance of beamlines. About 30% of our optics have
benefited from our Metrology.
The Alba-NOM instrument, has been carefully characterized
in terms of sensitivity and accuracy. For this we combined,
accurate motion metrology, ray-tracing of the measurment
process and numerical methods.
A =-2.00 mrad - 14 nrad
A =-1.00 mrad - 9 nrad
A =+0.00 mrad - 17 nrad
 =+1.00 mrad - 32 nrad
100 A
A =+2.00 mrad - 66 nrad
0
-100
-200
0
200
400
4
600
800
Position (mm)
1000
1200
The measurements provided by the NOM are accurate
enough to optimize the mirror figure. We are running a
project to develop subnanometer accuracy for long mirrors.
Bent
Flat
In situ
3
Residual Profile (nm)
tracing error (nrad)
200
2
1
0
-1
-2
-150
-100
-50
0
Position (mm)
50
100
150
www.cells.es
Optics Group Juan Carlos Martínez
Team Josep Vidal
Igors Sics
Iulian Preda
Juan Campos (UAB)
We are a small group but we have
strong support from many….
Other facilities…
Frank Siewert (HZB)
Daniele Cocco (LCLS)
Mourad Idir (NSLS II)
Muriel Thomasset (Soleil)
Amparo Vivo (ESRF)
Valeriy Yashchuck (ALS)
Ray Barrett (ESRF)
…
Administration Laura Campos
…
Computing Zbigniew Reszela
and Controls Sergi Blanch
Guifré Cuní
Sergi Pusó
Ramon Escribà
…
Engineering Carles Colldelram
Edmundo Fraga
José Ferrer
Ricardo Valcárcel
…
Claude Ruget
IRPI (France)
Alba All beamline scientists
beamlines …
www.cells.es