Effects of forged stock and pure aluminum coating on cryogenic
performance of heat treated aluminum mirrors
Ronald W. Toland 3 , Raymond G. Ohl a , Michael P. Barthelmy b , S. Wahid Zewari 3 , Matthew A.
Greenhouse 3 , John W. MacKenty 0
a NASA/Goddard Space Flight Center, Greenbelt, MD
b Swales Aerospace, Inc,, 5050 Powder Mill Rd, Greenbelt, MD
c Space Telescope Science Institute, 3700 San Martin Dr, Baltimore, MD
We present the results of an on-going test program designed to empirically determine the effects of
different stress relief procedures for aluminum mirrors. Earlier test results identified a preferred heat
treatment for flat and spherical mirrors diamond turned from blanks cut out of A1 6061-T651 plate
stock 1 . Further tests have been performed on mirrors from forged stock and one set from plate stock
coated with Alumiplate™ aluminum coating to measure the effect of these variables on cryogenic
performance. The mirrors are tested for figure error and radius of curvature at room temperature and
at 80 K for three thermal cycles. We correlate the results of our optical testing with heat treatment
and metallographic data.
KEYWORDS: mirrors, cryogenic, aluminum, heat treatment, stress relief
The Infrared Multi-Object Spectrograph (IRMOS) is a facility instrument for the Kitt Peak National Observatory
(KPNO) Mayall Telescope (3.8 meter) that will see first light in the spring of 2003. The project is a collaboration of
NASA/Goddard Space Flight Center (GSFC), the Space Telescope Science Institute (STScI), and KPNO. IRMOS is a
low- to mid-resolution (R = _/ = 300 — 3000), near-IR (0.8 — 2.5 _m) spectrograph which produces simultaneous
spectra of ~100 objects in its 2.8 • 2.0 arcmin field of view. The instrument uses a Texas Instalments, Inc. micro-
electromechanical system (MEMS) multi-mirror array (MMA) device as a real-time programmable slit mask. The
spectrograph operating temperature is ~80 K and the design is athermal: The optical bench and mirrors are machined
from aluminum (Al) 6061-T651.
The optics for IRMOS are single-point diamond turned (SPDT) from Al 6061-T651 plate stock with figure error
requirements of < 0. 1A RMS (X = 632.8 nm and microroughness of <100 Angstroms RMS. These numbers are at the
limits of what can currently be achieved by SPDT. To ensure good performance at operating temperature of the IRMOS
mirrors, we first ran a set of tests to find a heat treatment process which could relieve the residual stress in Al 6061-T651
and thus produce a mirror which underwent minimal distortion from 293K to 80K 1 .
However, to meet the microroughness requirements, the IRMOS mirrors will be coated with Alumniplate™ high-purity
aluminum. Also, the mirrors as originally tested were all from plate stock; none were from forged or extruded Al stock.
Thus the first part of our testing program had no data on how these variables might affect cryogenic performance.
We present the results of an on-going test program designed to empirically determine the effects of different stress relief
procedures for aluminum mirrors. Earlier test results identified a preferred heat treatment for flat and spherical mirrors
diamond turned from blanks cut out of Al 6061-T651 plate stock 1 . Further tests have been performed on mirrors from
forged stock and one set from plate stock coated with Alumiplate™ aluminum coating to measure the effect of these
variables on cryogenic performance. The mirrors are tested for figure error and radius of curvature at room temperature
and at 80 K for three thermal cycles. We correlate the results of our optical testing with heat treatment and
Previously we reported the results of a mirror testing program designed to select the best heat treatment process for the
IRMOS mirrors. We tested 6 pairs of mirrors — flats and spheres — representing 6 different heat treatment procedures.
Evidence from the cryogenic testing led us to choose process ‘SR5’ for the IRMOS mirrors, based on the repeatably
small figure error distortion of its two representative mirrors.
However, these tests were performed on mirrors all taken from plate stock. A16061-T651 is available in forged and
extruded stock, as well, and it is not obvious that each type should behave identically under cryogenic conditions, even
after having undergone similar heat treating. In addition, at a later date we decided to add a coating of Alumniplate™ to
the IRMOS mirrors to reduce the overall microroughness of the mirror surfaces, a factor not present in our previous tests.
To measure the effects of these variables, we cryogenically tested two new sets of mirrors. Mirrors 008 and 008A were
cut from A16061-T651 forged stock and underwent our SR4 heat treatment process prior to cryogenic testing. We sent
mirrors 004 and 004A, made from plate stock and treated with SR4 1 , to be coated with Alumniplate™. Mirrors 004 and
004A were tested as part of our previous investigations, and thus could provide a direct comparison of cryogenic
performance with and without the Alumniplate™ coating.
As with the previous test mirrors', each has a 94 x 100 mm aperture. The flat mirrors are 17.3 mm thick. The spherical
mirrors have a radius of 400 mm (concave) and are cut such that they would be 22.9 mm thick at the comers of the
aperture, were the comers “sharp” and not “racetrack.” The mirrors do not have mounting features. Janos Technology,
Inc. single point diamond turned the mirror figures to < 0. 1 X RMS ( X = 632.8 nm), with a radius tolerance of ±1% for
Alumniplate™ is a nearly (99.9%) pure aluminum coating developed by Alumniplate, Inc. to improve the surface finish
of diamond turned aluminum optics. Electroless nickel plating, a more traditional method, leads to CTE mismatch at
cryogenic temperatures, causing unwanted figure distortion. Alumniplating has been used successfully to reduce
microroughness in aluminum diamond turned optics to between 30 and 40 angstroms — well below the IRMOS
requirements 2 . However, cryogenic performance for these coated mirrors has not been rigorously demonstrated.
These tests are part of a general investigation of the effect of various heat treatment procedures on aluminum mirror
performance at 80K. As such, we used the same method to test the mirrors described herein as in our previous work.
This procedure is described in detail elsewhere.
To summarize, each mirror is free-mounted inside a cryogenic chamber with a 6” fused silica window to allow optical
testing through the chamber wall. We use a Fizeau-type phase-shifting interferometer to obtain figure data on the mirror
both at ambient and 80K through the chamber window. For the spheres, the dewar is mounted on a rail for radius of
curvature measurements. Moving the dewar from the best focus position for interferometric testing to the “cat’s eye”
position and measuring the distance traveled gives the current radius of curvature of the mirror. We compare the radius
of curvature as measured at ambient to that measured at cryogenic temperatures to further characterize the changes in the
mirror from warm to cold.
Mirror 008A was tested in a slightly different manner than described above. To help us characterize the effects of the
window on our measurements, we placed diodes on the center and edge of the window while testing 008A. This blocked
certain portions of the mirror from being seen by the interferometer. As a result, Peak-to-Valley data for the mirror may
be exaggeratedly large. We do not believe RMS data on 008A is affected.
For miiTor 008, problems with test execution and data collection obliged us to test the mirror through five cold cycles
instead of the usual three. Each of the five tests, however, was conducted normally.
Data acquired from the interferometer is reduced via custom routines in Interactive Data Language (IDL). The phase
maps taken at 293K and 80K are carefully registered to avoid subtraction errors. We then subtract the warm data from
the cold data. Piston and tilt are fitted to this delta and subtracted out. The results in Tables 1 and 2 are based on this
delta wavefront error map, which is displayed in Figure 1. The power term listed is twice the Zemike power coefficient,
for reasons explained earlier 1 .
Window effects are assumed to be minimal due to the use of a shutter to screen off the window from the interior of the
dewar 1 .
It should be noted that the scale for the contour maps shown is not constant. That is, identical grayscale tones do not
correspond to identical heights from map to map. The various scales have been chosen to highlight certain features of
delta wavefront error map.
0.04 IX RMS 0.060X RMS
0.066X RMS 0.058XRMS
Figure 1 : Contour maps of change in mirror figure error from warm to cold. Each map represents one cold cycle. X-632.8 imi
1 st Cycle
2 nd Cycle
I" 1 Cycle
Table 1 : Data for 3 cold cycles of mirror 008A. All values given in waves.
Table 2: Data for 5 cold cycles of mirror 008. All values given in waves (radius change given as percentage of radius as measured at
We used an ADE Phase Shift MicroXAM white light interferometer to measure the microroughness at 3 locations on the
aperture of each test mirror. The interferometer has a spatial resolution of 0.3 nm and an amplitude sensitivity of 0.01 nm
over a 0.30 x 0.41 mm field of view. Its software uses a fringe-fitting algorithm to calculate surface error. For powered
mirrors, low order, spherical and cylindrical terms are removed by subtraction of a least squares fit to the surface error
For completeness, we present the data for all mirrors used in this test program. The microroughness of the SR 1 — 6 plate
stock test mirrors is virtually identical in amplitude (slO nm RMS) and character. The data appear to be dominated by
the interaction of the SPDT tool with the impurities associated with A1 6061 . The grain structure and tool marks give the
roughness a well-defined orientation for various spatial periods. The microroughness of the mirrors cut from forging
stock is about the same amplitude, but much more random in character. Table 3 gives the Microroughness data for each
RMS @ Point A
RMS @ Point B
RMS @ Point C
Table 3: Microroughness data for the mirrors under test. Tilt, Sphere, and Cylinder have been removed. All values given in
The test results lead to slightly ambiguous conclusions. The average distortion of the plate stock SR4 mirrors was: PV
0.477 A , RMS 0.064 A , for the flat and PV 0.703 A , RMS 0. 120 A for the sphere. The forged stock flat thus performed
noticeably worse than its plate stock equivalent, while the forged sphere did better. However, operator error when
testing mirror 004 (without the Alumniplate™ coating) could have exaggerated the figure error present 1 , making the
forged spherical mirror’s improved performance somewhat dubious. We can tentatively say that our results support the
thesis that similar heat treatment processes will produce similar mirrors for different types of A16061 stock.
We are indebted to Roger A. Paquin for his help in planning this experiment. For technical support and useful
discussions, we gratefully acknowledge: Dr. Charles Bowers, Dr. David A. Content, William L. Eichhom, Sandra M.
Irish, Mark B. Mann, and Armando Morell of GSFC; Russell B. Makidon and Robert S. Winsor of STScI; Thomas J.
French and Joe McMann of ManTech International Corp.; Christopher May of Orbital Science Corp.; Shelly Conkey and
Rita Rosenbaum of Swales Aerospace, Inc.; Steven J. Conard of the Johns Hopkins University Instrument Development
Group; and Alex Sohn of the North Carolina State University Precision Engineering Center. This work is supported by
the James Webb Space Telescope project at GSFC.
1 . R. Ohl et. al. “Comparison of stress relief procedures for cryogenic aluminum mirrors,” Proc. SPIE 4822, 5 1 -
2. D. Vukobratovich, K. Don, R. Sumner, “Improved cryogenic aluminum mirrors,” Proc. SPIE 3435, 9-18, 1998