Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~slane/ref/ps/ps2_final.ps Äàòà èçìåíåíèÿ: Tue Feb 1 00:33:58 2011 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 06:55:03 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http www.badastronomy.com bad tv foxapollo.html

AXAF COATING PROCESS SELECTION STUDY:
REFLECTIVITY MEASUREMENTS
Prepared by: Patrick Slane
3 November 1992
1

Contents
1 Introduction 1
1.1 X­ray Reflectivity at AXAF En­
ergies : : : : : : : : : : : : : : : 1
1.1.1 Fundamentals of grazing
incidence reflectivity: a
summary : : : : : : : : : 2
1.1.2 Deposition techniques:
methods and considerations 3
1.1.3 Materials : : : : : : : : : 3
1.1.4 Previous AXAF reflectiv­
ity study results : : : : : 4
2 Data Acquisition and Analysis 5
2.1 SAO Reflectivity Apparatus : : : 5
2.2 Data Analysis : : : : : : : : : : : 7
3 Process Selection Study Samples
and Reflectivity Results 9
3.1 Samples : : : : : : : : : : : : : : 9
3.2 Results : : : : : : : : : : : : : : : 10
3.3 Comparisons and Recommenda­
tions : : : : : : : : : : : : : : : : 17
4 Establishment of Coating RFP
Specification 17
5 Vendor Selection Study 17
6 References 18
List of Figures
1 Iridium reflectivity vs. grazing
angle at 8.03 keV. : : : : : : : : 2
1 Iridium reflectivity vs. energy at
30 arcmin. : : : : : : : : : : : : : 2
3 Reflectivity vs energy for Ir, Pt,
and Au at 40 arcmin. : : : : : : : 4
4 SSS spectrum of x­ray beam at
SAO reflectivity facility. : : : : : 5
5 Schematic illustration of x­ray re­
flectivity apparatus at SAO. : : : 6
6 IPC data showing both slit and
mirror images. : : : : : : : : : : 7
7 IPC spectra of slit and mirror for
typical reflectivity measurement. 8
8 Comparison of x­ray reflectivity
for sputtered and evaporated Au
coatings. : : : : : : : : : : : : : 12
9 Comparison of x­ray reflectivity
for sputtered and evaporated Pt
coatings. : : : : : : : : : : : : : 13
10 Comparison of x­ray reflectivity
for sputtered and evaporated Ir
coatings. : : : : : : : : : : : : : 14
11 Comparison of x­ray reflectivity
for coatings from vendor C. : : : 15
12 Comparison of x­ray reflectivity
for sputtered and evaporated Ni
coatings. : : : : : : : : : : : : : 16
13 Comparison between representa­
tive samples of the best Au, Pt,
and Ir coatings. : : : : : : : : : : 19
14 Comparison of x­ray reflectivity
for sample psirdcd1. : : : : : : : 20
List of Tables
1 Mirror Inventory for CPSS : : : : 10
2 Reflectivities of Sputtered Iridium
Samples : : : : : : : : : : : : : : 17
i

1 Introduction
In an effort to maximize the x­ray throughput
of the AXAF­I HRMA, extensive studies of x­
ray reflectivity have been carried out at SAO, in
conjunction with personnel at MSFC and TRW
(Slane 1990a). These studies have included both
theoretical and experimental investigations of a
number of factors affecting x­ray reflectivity. Re­
sults of initial studies, which we summarize be­
low, led to the conclusion that the original base­
line HRMA, which had gold (Au) coatings for
the 3 inner mirrors and nickel (Ni) coatings for
the outer mirrors, would have increased effective
area by substituting iridium (Ir) for gold. Fur­
ther, the studies suggested that the nature of
the deposition process could strongly affect the
resulting x­ray performance of the coatings, and
that further study of available processes could
lead to an enhanced mirror throughput. The
AXAF Coating Process Selection Study (CPSS)
was conceived to address these issues by for­
mulating a matrix of test elements that would
lead to a final recommendation for the baseline
HRMA coating. The elements of the plan were
arrived at in a meeting of the Coating Work­
ing Group, at MSFC in December 1991 (Loer
and Hixson 1991). Herein we present the results
of this study which have led the Coating Work­
ing Group to recommend iridium as the base­
line coating for the AXAF­I mission, and have
allowed the specification of a minimum perfor­
mance criterion for the coatings, based upon lab­
oratory measurements of coated samples.
1.1 X­ray Reflectivity at AXAF Energies
From an astrophysics point of view, a significant
feature of AXAF is the high energy response;
in contrast to the Einstein and ROSAT obser­
vatories which extended only to ¸ 4 keV and
¸ 2 keV, respectively, the AXAF effective area
extends to 10 keV. This will allow new spectro­
scopic investigations to be carried out, includ­
ing studies of thermal spectra at temperatures
typical of clusters of galaxies as well as investi­
gations in the astrophysically important Fe­line
region. These goals require an effective area
which remains large at higher energies, and this
in turn places constraints on the mirror coatings;
only through careful selection of coating materi­
als and methods will the reflectivity of the coat­
ings be sufficient to meet the goals of the mirror
design.
One primary concern with regard to thin film
coatings centers on the effective density ae of the
coating material. Because thin films tend to
grow in discrete ``islands'' which eventually co­
alesce, there may be considerable open space in
the resulting coating. As discussed below (Sec­
tion 1.1.1), the optical behavior of the coatings
may be characterized by a complex dielectric
constant Ÿ. At x­ray energies, total external re­
flection occurs at a critical angle
` c ú
p
ff (1)
where 1 \Gamma ff is the real part of the dielectric con­
stant. Because ff / ae (see below), we see that the
critical angle ` c scales as ae 1=2 . Thus, identifica­
tion of the critical angle through measurements
of reflectivity as a function of grazing angle pro­
vides a relatively sensitive indicator of the coat­
ing density. This is illustrated in Figure 1 which
contains a plot of reflectivity vs grazing angle,
for 8.03 keV x­rays, for an ideal iridium coating
as well as one with 80% bulk density. Figure 2
illustrates the behavior of the x­ray reflectivity
as a function of energy (for a grazing angle of
30 arcmin) for an ideal coating and for a coating
with iridium density equal to 80% that of the
bulk material.
It is clear that the effects of reduced den­
sity are most pronounced at higher energies; as
a result, we have carried out measurements at
8.03 keV for the Au, Pt, and Ir coatings. For
Ni coatings we have carried out measurements
at 6.4 keV, which provides a good probe of the
critical angle region for this material.
1

Figure 1: X­ray reflectivity as a function of graz­
ing angle for iridium coatings with 100% (solid)
and 80% bulk density. X­ray energy is 8.03 keV.
1.1.1 Fundamentals of grazing incidence
reflectivity: a summary
Optical Constants
In order to calculate the theoretical reflectivity
for a given coating, it is necessary to know the
optical constants associated with the coating ma­
terial. A list of such constants, and a review
of reflectivity calculations, is given in Zombeck
(1983a). Elsner et al. (1991) have presented a
review and critical assessment of the standard
approach to deriving optical constants for use
in x­ray reflectivity calculations. Summarizing
their description, following that of Henke (1982),
the optical constants are most conveniently ex­
pressed in the form of a complex dielectric con­
stant:
Ÿ = 1 \Gamma ff \Gamma ifl: (2)
The real and imaginary parts of the dielectric
constant are associated with the atomic scatter­
ing factors f 1 and f 2 (which are derived from the
photoabsorption coefficient ¯ a (E)):
ff = 1
ú
/
e 2
m e c 2
! ` N 0 ae
A
'` hc
E
' 2
f 1 ;
Figure 2: X­ray reflectivity as a function of en­
ergy for iridium coatings with 100% (solid) and
80% bulk density. Grazing angle is 30 arcmin.
and
fl =
1
ú
/
e 2
m e c 2
! `
N 0 ae
A
'`
hc
E
' 2
f 2 : (3)
Here m e is the electron mass, N 0 is Avagadro's
number, ae is the density, A is the atomic weight,
and other fundamental constants are expressed
with standard symbols.
Often, the optical constants are expressed in
terms of a complex refractive index:
~ n = 1 \Gamma ffi \Gamma ifi: (4)
Since
~ n = Ÿ 1=2 ; (5)
ffi and fi may be derived directly from ff and fl:
ffi = 1 \Gamma
s
1
2
Ÿ q
(1 \Gamma ff) 2 + fl 2 + (1 \Gamma ff)
–
; (6)
fi = sgn(fl)
s
1
2
Ÿ q
(1 \Gamma ff) 2 + fl 2 \Gamma (1 \Gamma ff)
–
:
(7)
2

Fresnel Equations
Given the optical constants of the coating mate­
rial, the reflectivity may be calculated as a func­
tion of energy and grazing angle by using the
Fresnel equations:
R? = a 2 + b 2 \Gamma 2a cos ` + cos 2 `
a 2 + b 2 + 2a cos ` + cos 2 `
(8)
and
R k = a 2 + b 2 \Gamma 2a sin ` tan ` + sin 2 ` tan 2 `
a 2 + b 2 + 2a sin ` tan ` + sin 2 ` tan 2 `
(9)
where
2a 2 =
Ÿ i
n 2 \Gamma fi 2 \Gamma sin 2 `
j 2
+ 4n 2 fi 2
– 1=2
+
i
n 2 + fi 2 \Gamma sin 2 `
j
; (10)
and
2b 2 =
Ÿ i
n 2 \Gamma fi 2 \Gamma sin 2 `
j 2
+ 4n 2 fi 2
– 1=2
\Gamma
i
n 2 + fi 2 \Gamma sin 2 `
j
: (11)
Here ` is the angle of incidence (i.e. ú/2 ­ graz­
ing angle) and n = 1 \Gamma ffi. A comparison of the
various sets of constants has been carried out by
Elsner (1991), who has recommended the use of
constants compiled by Henke et al. (1991) for
AXAF­related calculations; such a convention
has been adopted for the remainder of this re­
port.
1.1.2 Deposition techniques: methods
and considerations
One factor which may play a considerable role
in determining the density profile of a deposited
thin film is the mobility of the atoms as they de­
posit on the substrate; higher mobility promotes
formation of more nucleation sites and, thus, a
closer­packed structure. High mobility may be
achieved by ensuring that the atoms being de­
posited are reasonably energetic upon arrival at
the substrate. This, in turn, is accomplished by
providing energetic atoms at the site of the tar­
get, and by using low ambient pressure to min­
imize collisional energy losses as the atoms mi­
grate from target to substrate. While evapora­
tive processes are generally carried out at pres­
sures below 10 \Gamma5 torr, the energy of the atoms as
they leave the target material is relatively low.
Thus, the mobility upon arrival at the substrate
is not sufficient to prevent formation of columnar
growth associated with the development of voids.
In some modes of operation an ion beam is used
to assist the deposition process by supplying ad­
ditional energy to the atoms via bombardment
with accelerated ions.
Sputtering processes, on the other hand, pro­
vide considerable energy to the target atoms via
bombardment with energetic ions of a noble gas.
Ambient pressures, however, tend to be higher
because of the presence of the sputtering gas.
Minimization of the sputter­gas pressure, as well
as of the target­substrate distance, is necessary
to achieve more densely packed coatings.
1.1.3 Materials
Reflective coatings for previous x­ray missions
have included a number of different materials: Ni
for Einstein, Au for ROSAT and BBXRT, and Ir
for SAX to name a few. For the range of graz­
ing angles appropriate for AXAF, Ni cannot pro­
vide high energy response; as a result, the base­
line coating for the AXAF­I mirrors originally
included Au for the innermost mirrors (Zombeck
1983b). While Au reflects well at the lower en­
ergies as well, there are absorption features that
reduce the reflectivity in the vicinity of 2.2 keV.
To compensate for this loss, the baseline plan
was to coat the outermost mirrors with Ni, which
would introduce absorption edges at 0.8 keV, but
partially fill in the Au edges.
While the choice of Au to provide high en­
ergy response has a secure basis (it has been used
successfully on previous missions, and consider­
able expertise in producing Au coatings exists
in the coating industry), platinum (Pt) and Ir
3

Figure 3: X­ray reflectivity vs energy for Ir, Pt,
and Au at a grazing angle of 40 arcmin.
coatings offer the potential for even better high
energy performance. The theoretical reflectiv­
ity for these materials is plotted as a function
of energy, for a grazing angle of 40 arcmin, in
Figure 3. Experience has shown, however, that
one often obtains less than the theoretically pre­
dicted reflectivity on actual samples. To deter­
mine whether a change from Au to another ma­
terial would be beneficial, coatings of Pt and Ir
were produced using both evaporative and sput­
tering techniques.
1.1.4 Previous AXAF reflectivity study
results
Studies of x­ray reflectivity measurements car­
ried out at SAO prior to the CPSS were summa­
rized by Slane et al. (1991). The studies involved
measurements of more than 75 mirrors, testing
such parameters as coating material, deposition
process, deposition angle, and coating thickness.
Measurements were carried out primarily at 8.03
and 6.4 keV, with some additional measurements
at 4.5 and 1.5 keV. Several distinct conclusions
were reached from these studies:
1. X­ray reflectivity may be strongly affected
by coating thickness; the thinner coatings
(¸ 200 š A) performed better than the thicker
coatings (¸ 750 š A). This difference was par­
ticularly significant for samples produced by
evaporative methods, and is likely to be
related to changes in film morphology as
thickness increases.
2. The effective density of the films increases
with depth into the coating. This is evi­
dent from measurements at different ener­
gies; measurements made at 8.03 keV sug­
gest a coating with higher density than mea­
surements made at 6.4 keV (Slane 1990b).
The interpretation is that the lower energy
measurements sample only the upper por­
tions of the coating, which apparently are
somewhat ``fluffy'' with low effective density
(e.g. 80% that of the bulk material), while
higher energy measurements sample deeper
into the coating where the film growth has
permitted fewer voids between the crys­
talline islands.
3. Deposition should be carried out at near­
normal incidence; evaporative coatings for
which the substrate presented a large (60 ffi )
angle relative to the direction of the evap­
oration boat provided considerably reduced
reflectivity. This is likely to be associated
with shadowing effects whereby the islands
which are formed early in the coating pro­
cess cast an effective shadow over adjacent
regions, resulting in coating voids and lower
overall density. These results are consistent
with those from similar studies carried out
for the ROSAT mirror coatings (Burkert et
al. 1986).
4. Sputtered coatings, on the average, outper­
formed evaporative coatings. This result
is expected based upon the higher mobil­
4

ity provided by the sputtering process, as
mentioned above.
5. Iridium coatings provide higher reflectivity
than gold coatings, as expected from theo­
retical calculations.
As a result of these studies, careful consideration
of dc­magnetron sputtering and Ir coatings was
deemed a significant element of the CPSS.
2 Data Acquisition and Analysis
2.1 SAO Reflectivity Apparatus
The Reflectivity Measurement Apparatus at
SAO consists of a long (11.25 m) vacuum pipe
equipped with an Imaging Proportional Counter
(IPC) (Gorenstein et al. 1981) at one end and an
x­ray generator at the other. A wide variety of
targets and filters are available for generation of
x­rays. For this study, measurements were pri­
marily made with a Cu anode followed by a Ni
filter. Such a combination provides strong ab­
sorption of the Cu­Kfi fluorescence line, as well
at the bremsstrahlung component to the spec­
trum, while transmitting the Cu­Kff line at 8.03
keV (Figure 4). At this energy, the critical an­
gle for iridium falls within the AXAF range of
interest (Figure 3). As noted above, reflectiv­
ity measurements in the vicinity of the critical
angle yield the most sensitive probe for coating
anomalies such as reduced density.
The IPC has an intrinsic spatial resolution of
approximately 0.25 mm (¸ 0:5 arcmin as viewed
from the mirror location) at 8.03 keV. The full
field of the IPC is approximately 7.5 cm \Theta 7.5
cm (¸ 2 ffi \Theta 2 ffi ) although, in practice, only the
central 3.8 cm in the vertical dimension are uti­
lized. The energy resolution of the IPC is ap­
proximately 16% (FWHM) at 8.03 keV, and the
limiting data rate is approximately 400 counts
s \Gamma1 . Each photon event is sent through process­
ing electronics and tagged with a position and
energy.
As illustrated in Figure 5, x­rays are incident
on a test mirror from a distant source and re­
Figure 4: SSS spectrum of x­ray beam at SAO
reflectivity facility, using a copper anode, with an
anode voltage of 17 kV. Lower: Unfiltered spec­
trum in which the Cu­Kff and Cu­Kfi lines are
clearly seen superimposed on the bremsstrahlung
continuum. Upper: Spectrum after a 50 ¯m Ni
filter is introduced to reduce the continuum and
to filter the Cu­Kfi line. Additional lines from
Fe, presumably resulting from florescence in the
x­ray pipe, are also seen.
flected to the IPC; for comparison and simulta­
neous beam­monitoring, a fixed slit is also im­
aged in x­rays. The mirror/slit combination is
mounted on a motor­driven turntable which may
be rotated in order to change the incident grazing
angle at the mirror. The turntable is mounted on
a translation stage which may be moved to pre­
vent images from falling on the IPC window sup­
port ribs. A data acquisition computer is used to
collect raw events from the IPC processing elec­
tronics as well as to control the positioning of the
mirror and slit assembly. In a typical reflectivity
run, a scan of approximately 30 angle positions
is executed, with finer sampling of the angular
dependence in the critical angle area. A typical
run requires approximately 3 hours of acquisition
time and generates ¸ 25 Mbytes of data.
5

d l
IPC
slit
mirror
s
y
d
a 2q-a
q a
q
Figure 5: Schematic illustration of x­ray reflectivity apparatus at SAO.
6

Figure 6: IPC data showing both slit and mirror
images. The rectangular box illustrates manner
in which the vertical extent of the images is lim­
ited. For each image, events in a 200 bin region
centered on the image centroid are used to form
the image spectrum.
2.2 Data Analysis
Analysis of the reflectivity data consists of a
three­step procedure:
1. Event Selection. The IPC spatial data
consists of a 2­d image containing the mir­
ror and slit, as shown in Figure 6. A rect­
angular subimage is extracted, as indicated
in the Figure, to ensure that identical verti­
cal sizes are used for both images. The lat­
eral centroids are then determined for each
image, and spectra are extracted from re­
gions extending horizontally \Sigma100 IPC bins
from each centroid, or approximately \Sigma4:2
arcmin about the specular direction (1 bin
ú 25¯m). Horizontal window support ribs
produce regions of x­ray absorption which
are identical for the two images, as seen in
the Figure.
2. Angle Determination. The grazing an­
gle of the mirror is determined analytically,
as indicated in Figure 5; the separation of
the mirror and slit images is determined by
the difference in the relative centroid posi­
tions, and all other geometrical factors are
measured directly. Using symbols defined in
Figure 5, we have
tan(2` \Gamma ff) = (s + y)
(l + ffi)
(12)
and
tan ff = s
d \Gamma ffi
: (13)
Thus, the grazing angle ` is given by
` = 1
2
Ÿ
tan \Gamma1
`
s
d \Gamma ffi
'
+ tan \Gamma1
`
s + y
l + ffi
'–
:
(14)
The values s = 2:31 cm, ffi = 0, l = 2:09 m,
and d = 9:26 m were fixed for the duration
of this study.
3. Reflectivity Calculation. The reflectiv­
ity for a given mirror of length x is deter­
mined by combining the calculated grazing
angle with the ratio of counts N s in the im­
age of a slit of width w with that of the
mirror (Nm ). Thus,
R = Nm
N s
w
x tan `
: (15)
The number of counts in each image are ex­
tracted from the associated spectrum. A
spectral window of 10 PHA bins centered on
the centroid of the spectral line is used to
provide a roughly monoenergetic measure­
ment (Figure 7). This spectral window is
slightly smaller than the FWHM of the spec­
tral response. The line­to­continuum ratio
in this band is reasonably high, as verified
by measurements carried out with a solid­
state spectrometer (Figure 4). In determin­
ing the number of counts from the spectral
region, endpoint interpolation is carried out
to account for noninteger centroid positions.
7

Figure 7: IPC spectra of slit and mirror for typical reflectivity measurement. The vertical lines
represent the spectral window from which events are extracted. Note the lack of high energy x­rays
in the x­ray spectrum, due to the decrease in reflectivity with increasing energy.
8

Measurement uncertainties are calculated
based upon estimated errors in geometrical mea­
surements (which translate into uncertainties in
the angle determination, and hence the pro­
jection factor which determines the acceptance
width of the mirror) and Poisson errors in the
number of counts extracted from the spectra.
Grazing angle uncertainties are of order 0.2 ar­
cmin or less. Typical relative measurement un­
certainties for the reflectivity are approximately
2%. As noted above, a small systematic error in
the overall normalization may exist. For compar­
isons with theoretical reflectivity results, there
are additional factors which must also be con­
sidered:
1. Because the mirror is illuminated by x­rays
over its entire length, the reflectivity is actu­
ally averaged over a range of grazing angles.
The difference in grazing angle at the two
extremes of the mirror is roughly 10% of the
nominal grazing angle. At small angles the
effect of this deviation is quite small (since
the reflectivity changes very slowly with an­
gle in this region); in the critical angle re­
gion, however, this variation corresponds to
a significant change in reflectivity at the ex­
treme edges of the mirror (though in a par­
tially compensatory way since the reflectiv­
ity increases at one end and decreases at the
other).
2. The finite spatial window around the im­
age centroid, from which events are selected,
results in the inclusion of x­rays which are
scattered into a finite angular band around
the spectral direction. Because the scatter­
ing associated with the surface roughness of
the mirror increases with grazing angle, this
effect reduces the measured reflectivity at
large angles (relative to that measured at
small angles). As a result, measured val­
ues in close agreement with theory may be
obtained at small angles, but not at large
angles.
3. While a small correction, the spectral se­
lection does not provide the absolute equiv­
alent of a monoenergetic measurement.
Rather, continuum x­rays with energies dif­
ferent from 8.03 keV, but within the spec­
tral resolution of the IPC, will be counted
as well. Because the x­ray generator is op­
erated such that the line­to­continuum ratio
is very large, this does not represent a large
contribution, however.
It must be stressed that these factors, while im­
portant, do not enter into the mirror­to­mirror
comparisons.
In order to directly compare reflectivity results
among different samples, a data­fitting proce­
dure was used to derive the reflectivity at two
fiducial angles (20 and 34 arcmin ­ at which di­
rect measurements were not necessarily made).
The procedure (O'Dell 1992) consists of using re­
flectivity data in the vicinity of the two fiducial
angles (data between 10 and 30 arcmin for the
20 arcmin value, and between 30 and 38 arcmin
for the 34 arcmin value) to perform a quadratic
(20 arcmin) or cubic (34 arcmin) fit to the log­
arithm of the data. The reflectivity values and
associated uncertainties at the fiducial angles are
then calculated from the best­fit parameters.
3 Process Selection Study Samples
and Reflectivity Results
3.1 Samples
The coated samples used for this study were pro­
duced by a variety of techniques. Samples of
each type were obtained from at least two differ­
ent coating manufacturers to provide some level
of consistency check on results from a given ma­
terial and coating process. The substrates for all
mirrors were identically prepared, polished zero­
dur pieces (3'' \Theta 4'' \Theta 3
4
'' dimensions). The sam­
ples were prepared at MSFC to a flatness spec­
ification of –=2 (verified by Zygo interferometry
at 6328 š A) and a microroughness specification
of better than 5 š A RMS in the spatial band­
9

pass 1¯m ­ 500¯m (verified by WYKO measure­
ments).
Four different coating materials were investi­
gated: Ir, Au, Pt, and Ni. To eliminate any
reflectivity effects associated with coating thick­
ness, the same thickness (350 š A) was used for
each coating; a binding layer of Cr (100 š A) was
used for each coating to ensure adhesion. Two
deposition processes were investigated: e­beam
evaporation, and dc­magnetron sputtering. Each
coating vendor was required to produce 3 sam­
ples coated to identical specifications, but in sep­
arate deposition runs. In addition, each ven­
dor provided complete information on the de­
position parameters used in producing the coat­
ings, as well as verification that witness samples
coated with the mirrors passed a tape­pull ad­
hesion test. A report demonstrating that the
deposition procedures used could be scaled to
AXAF HRMA size mirrors was also provided. A
total of 45 mirrors were procured for the study,
as summarized in Table 1.
3.2 Results
SAO performed these measurements from March
16 to May 19, 1992. The results of the reflectivity
measurements are summarized in Figures 8 ­ 14
below. In each Figure, data points are compared
with a curve representing the theoretical reflec­
tivity from a perfectly smooth, infinitely thick
coating with density identical to that of the bulk
material. Error bars are shown on one set of data
points for each Figure, and are representative of
those for the other points (whose errors are not
shown, for clarity). Gaps in the angular coverage
between 20 and 25 arcmin are the result of the
slit image blocking the mirror image.
The data show some indication of a system­
atic normalization uncertainty, as evidenced by
reflectivity values that are too large at the small­
est angles. This issue is thought to be associated
with uncertainties in the effective width of the
slit used in the normalization process, particu­
larly since the effective width is known to exceed
Table 1: Mirror Inventory for CPSS
Mirror Deposition
Name Material Technique Vendor
psaudca1 dc
psaudca2 Gold magnetron A
psaudca3 sputtering
psaudcc1 dc
psaudcc2 Gold magnetron C
psaudcc3 sputtering
psauebb1 e­beam
psauebb2 Gold evaporation B
psauebb3
psauebd1 e­beam
psauebd2 Gold evaporation D
psauebd3
psirdca1 dc
psirdca2 Iridium magnetron A
psirdca3 sputtering
psirdcc1 dc
psirdcc2 Iridium magnetron C
psirdcc3 sputtering
psirdcc4 dc
psirdcc5 Iridium magnetron C
psirdcc6 sputtering
psirdce1 dc
psirdce2 Iridium magnetron E
psirdce3 sputtering
psirebb1 e­beam
psirebb2 Iridium evaporation B
psirebb3
psirebd1 e­beam
psirebd2 Iridium evaporation D
psirebd3
psptdca1 dc
psptdca2 Platinum magnetron A
psptdca3 sputtering
psptdcc1 dc
psptdcc2 Platinum magnetron C
psptdcc3 sputtering
psptebb1 e­beam
psptebb2 Platinum evaporation B
psptebb3
psnidca1 dc
psnidca2 Nickel magnetron A
psnidca3 sputtering
psnidce1 dc
psnidce2 Nickel magnetron E
psnidce3 sputtering
psniebb1 e­beam
psniebb2 Nickel evaporation B
psniebb3
10

the mechanical width due to some x­ray pene­
tration at the slit edges. Mirror­to­mirror com­
parisons do not suffer from this uncertainty in
that the same slit was used for each measure­
ment. This is evident in comparisons of samples
produced by the same vendor in different deposi­
sion runs, where mirror­to­mirror variations are
observed to be negligible for nearly all of the
samples. As noted above, a deficiency in mea­
sured reflectivity at larger angles may be par­
tially due to x­rays which are scattered at angles
larger than the chosen spatial acceptance win­
dow.
We discuss the results for each coating mate­
rial separately:
1. Gold The reflectivity results for Au are
shown in Figure 8. Little difference is seen
between evaporated and sputtered coat­
ings, although it is significant that the mir­
rors with the lowest reflectivity were pro­
duced by evaporation.
2. Platinum Results for Pt coatings are
shown in Figure 9. Here there is a clear
difference between the results from evap­
orated coatings and coatings produced by
sputtering vendor A. Coatings produced by
sputtering vendor C, however, performed
no better (on average) than the evapo­
rated coatings. This poor performance was
attributed to less than optimal choice of
sputtering parameters (see discussion re­
garding Ir samples below). As shown in the
Figure, the best Pt results produced reflec­
tivity in excess of the theoretical limit for
Au coatings.
3. Iridium The results for Ir coatings are
plotted in Figure 10. Here there is a dra­
matic difference between the evaporated
and sputtered samples; the evaporated
samples produce reflectivities which fall
below the theoretical limit for Au ­ and far
below that for Ir ­ while the best sputtered
samples approach theoretical values for Ir.
As with the Pt samples, the sputtered sam­
ples from vendor C performed poorly in
comparison with other sputtered samples.
This vendor supplied a 2nd set of mirrors
for which the sputter gas pressure and the
target­substrate distance were reduced. As
discussed in Section 1.1.2, such modifica­
tions would be expected to increase the
mobility of the depositing atoms, thus in­
creasing the effective density ­ and, hence,
reflectivity ­ of the resulting coatings. As
illustrated in Figure 11, this modification
produced the desired results, with the new
set of coatings performing as well as the
remaining sputtered samples.
4. Nickel Reflectivity results for Ni coatings
are shown in Figure 12. While interference
effects associated with the finite thickness
of the coatings make comparisons difficult
at larger angles, the results from the dif­
ferent coatings are in excellent agreement
at smaller angles and, particularly, through
the critical angle region. For this material
there seems to be no indication of a pref­
erence between the sputtering and evapo­
ration techniques. While the scatter in the
interference feature position from vendor
to vendor are clearly significant (indicating
varying levels of success at obtaining the
correct film thickness), sample­to­sample
variations from any given vendor are very
small, indicating good repeatability.
The reflectivity results for all samples mea­
sured in this study are accessible on the CfA
HEAD LAN in the directory:
/home/reflec/ps—results
The files are in 5 column ASCII format, with
the columns containing `, d`, R, dR, and
the run number. Filenames are of the form
date.mirror.ref where date is the date of mea­
surement and mirror is the mirror name. All
data may be obtained via anonymous ftp by first
sending e­mail to
slane@cfa.harvard.edu
requesting that these data be placed in the ftp
directory.
11

Au
psaudca
psauebb
psaudcc
psauebd
Figure 8: Comparison of x­ray reflectivity for sputtered and evaporated Au coatings. Solid line
corresponds to theoretical reflectivity. Error bars shown on single data set are representative of all
data.
12

Pt
Au
psptdca
psptebb
psptdcc
Figure 9: Comparison of x­ray reflectivity for sputtered and evaporated Pt coatings. Curves cor­
respond to theoretical reflectivity for Pt (solid) and Au (dashed). Error bars shown on single data
set are representative of all data.
13

Ir
Au
psirdca
psirebb
psirdcc
psirebd
psirdce
Figure 10: Comparison of x­ray reflectivity for sputtered and evaporated Ir coatings. Curves
correspond to theoretical reflectivity for Ir (solid) and Au (dashed). Error bars shown on single
data set are representative of all data.
14

Ir
Au
psirdcc1­3
psirdcc4­6
Figure 11: Comparison of x­ray reflectivity for coatings from vendor C before (triangles) and after
(squares) adjustments to sputtering parameters as discussed above.
15

Ni
psnidca
psniebd
psnidce
Figure 12: Comparison of x­ray reflectivity for sputtered and evaporated Ni coatings. Solid line
corresponds to theoretical reflectivity. Error bars shown on single data set are representative of all
data.
16

3.3 Comparisons and Recommendations
An overall comparison of the results summarized
above lead to the conclusion that Ir should be
the coating material of choice. Figure 13 illus­
trates a comparison between representative sam­
ples of the best Au, Pt, and Ir coatings obtained
in the study. It is also clear from the Ir results
summarized in Figures 10 and 11 that the depo­
sition method of choice should be a sputtering
technique. The results of the study confirm ex­
pectations based upon the theoretical reflectivity
for each material and the physical principles in­
volved in the deposition processes. In addition,
they provide a reflectivity standard upon which
the HRMA coating specifications may be based
(Section 4.0). The Coating Working Group,
based upon the results of this study, has recom­
mended to the AXAF Project that the baseline
HRMA coating consist of Ir.
4 Establishment of Coating RFP
Specification
Using the Ir reflectivity results summarized
above, a minimum specification for the perfor­
mance of the actual HRMS coating has been de­
rived. The specification is based upon the mean
value of the x­ray reflectivity from the sput­
tered Ir samples (eliminating samples psirdcc1­3
which, as discussed above, were improved upon
with modifications to deposition parameters) at
two fiducial angles. As described in Section 2.2,
the reflectivity at 20 and 34 arcmin was deter­
mined for each mirror by performing a fit to
the reflectivity values at nearby angles. The re­
sults (O'Dell 1992) are summarized in Table 2.
The reflectivity specification recommended for
the RFP was established as a requirement that
the vendor produce a coating which meets or ex­
ceeds reflectivity values which are 2oe below the
mean, as determined above. This translates into
the requirement R[20 0 ] – 0:81 and R[34 0 ] – 0:49,
using the fitting procedure described above.
Table 2: Reflectivities of Sputtered Iridium Sam­
ples
Coating R[20 0 ] oe R [20 0 ] R[34 0 ] oe R [34 0 ]
irdca1 0.852 0.009 0.499 0.005
irdca2 0.815 0.008 0.496 0.005
irdcc4 0.842 0.007 0.525 0.005
irdcc5 0.855 0.009 0.508 0.005
irdcc6 0.858 0.008 0.522 0.005
irdce1 0.845 0.009 0.506 0.005
irdce2 0.824 0.008 0.510 0.005
irdce3 0.826 0.008 0.521 0.005
Mean 0.840 0.016 0.511 0.011
5 Vendor Selection Study
In support of the TRW vendor selection process,
one additional sample measurement was carried
out at SAO. Results from this sputtered Ir sam­
ple, supplied by a vendor that had only supplied
e­beam evaporation samples for the CPSS, are
plotted in Figure 14. It is clear from Figure 14
that the reflectivity from this sample is represen­
tative of that from good sputtered Ir mirrors, ex­
ceeding the minimum performance specification
as indicated by the dashed lines in the Figure.
17

6 References
Burkert, W. et al., ``Effects of Mirror Contamination Observed in the ROSAT Programme'', SPIE
Proceedings Vol. 733, p. 217­227 (1986)
Elsner, R.F. 1991, ``Dielectric and optical constants for AXAF reflecting materials based on data
from Henke et al. (1991),'' Memorandum W. E. Taylor/EJ31 (August 28, 1991).
Elsner, R.F, O'Dell, S.L., and Weisskopf, M.C. 1991, ``Effective Area of the AXAF X­Ray Tele­
scope: Dependence upon Dielectric Constants of Coating Materials,'' Journal of X­ray Science and
Technology, 3, 35­44 (1991)
Gorenstein, P. et al. 1981, Trans. IEEE Nuc. Sci., NS­28, 869.
Henke, B.L. et al. 1982, Atomic Data and Nuclear Data Tables 27(1).
Henke, B.L. et al. 1991, to be published.
Loer, S. and Hixson, S. 1991, TRW Interoffice Correspondence ­ AXAF­91­7­008.
O'Dell, S.L. 1992, Letter to TRW, MSFC, SAO (1992, June 11).
Slane, P.O. 1990a, ``AXAF Reflectivity Study Plan'', SAO­AXAF­90­031 ­ Prepared for
NASA/MSFC.
Slane, P.O. 1990b, ``AXAF X­Ray Reflectivity Studies: Reflectivity of Iridium Coatings'', AXAF
Interim Report ­ SAO­AXAF­90­049.
Slane, P.O. et al. 1991, ``Grazing Incidence X­ray Reflectivity: Studies for the AXAF Observatory'',
SPIE Proceedings Vol. 1546, p. 26­40 (1991), and AXAF Interim Report ­ SAO­AXAF­91­016.
Zombeck, M.V. 1983a, ``AXAF Effective Area Studies'', AXAF Interim Report ­ SAO­AXAF­83­
015.
Zombeck, M.V. 1983b, ``Optical Constants and Reflectivities for Nickel, Gold, and Platinum in the
X­ray Region of the Spectrum (0.1 ­ 10 keV)'', AXAF Interim Report ­ SAO­AXAF­83­016.
18

Ir
Pt
Au
psirdce
psptdca
psauebb
Figure 13: Comparison between representative samples of the best Au, Pt, and Ir coatings obtained
in CPSS. It is clear from these measurements that Ir coatings offer considerably higher reflectivity
than the baseline HRMA plan for use of gold.
19

psirdce1
psirdcd1
Figure 14: Comparison of x­ray reflectivity for sample psirdcd1, evaluated in support of vendor
proposal evaluation, and psirdce1 which is representative of good sputtered Ir samples. It is clear
from the Figure that the new sample performs as well as other sputtered samples. Dashed lines
represent minimum performance criterion as described in Section 4.
20