TABLE I. Parameters of the structures studied.
Mirror type
No. 1 (a-C:HI /a-C:H II
Structure parameters
)
No. 2 ͑Ni/C͒
͑1͒ Structure period ͑Å͒
͑2͒ Number of period
94
59
72
30
͑3͒ Layer thickness ͑Å͒
͑4͒ Difference in decrements
of refraction of the layers ͑⌬␦͒
47/47
1.2ϫ10Ϫ6
24/48
18ϫ10Ϫ6
Nϭ118, the best agreement between the theoretical curve
and experimental data was achieved for the difference ⌬␦
ϭ1.2ϫ10Ϫ6 in the layer decrements of refraction, which
corresponds to the layer density difference ⌬
ϭ0.38 g/cm3. Calculations for the Ni/C mirror were made
using the tabular values of the decrements of refraction and
absorption coefficients for nickel and graphite. The param-
eters of the mirrors of both types are listed in Table I.
Comparison of the experimental data for different struc-
tures demonstrates that for MCIS having maximal reflectiv-
ity (Rmaxϭ40.7%) half as large as for the Ni/C mirror we
obtained the resolution that is twice as large ͑the half-width
of the first Bragg reflection peak was ⌬⌰ϭ0.02°͒. We note
that the smaller value of R for MCIS is, to a large extent, due
to the large roughness of quartz substrates. For MCIS the
curve Rϭf͑⌰͒ is below the reflectivity curve for the Ni/C
structure. In fact, it is seen from Fig. 1͑a͒ that in the region of
total external reflection, R for the Ni/C mirror approaches
98%, whereas for MCIS it is only 90%. It should be noted
that the half-width of the Bragg peak of the theoretical
Rϭf͑⌰͒ curve is about half as large as that of the experimen-
tal curve. We believe that this is due to the fact that the
experimental curve is the convolution of the true curve and
the instrumental function; for this reason the resulting width
is greater than the true width. This broadening effect is ex-
pected to be especially large for MCIS, as these structures
have small peak half-widths.
FIG. 2. Experimental dependencies Rϭf͑⌰͒ for ϭ8.33 Å. ͑1͒ Multilayer
carbon interference structure ͑Nϭ118, dϭ94͒; ͑2͒ Ni/C mirror ͑Nϭ60,
dϭ51 Å͒.
films with ⌬ϭ0.8–1 g/cm3 is quite realistic. In this case
Rmax would reach 90% at wavelength 1.54 Å and dϭ94 Å.
We note one more important advantage of MCIS when
used in x-ray spectroscopy. There are no even peaks in the
reflection spectrum of the structure studied ͓Fig. 1͑a͒, curve
1͔; this indicates that the different layer thicknesses are the
same and layer thicknesses are almost constant in any period.
One can quench the ‘‘even’’ reflection peaks for Ni/C mir-
rors as well as by the corresponding increase of the Ni layer
thickness. However, the calculations show that this decreases
the reflectivity of the structure by about 10% and broadens
the Bragg peak.
The data Iϭf͑⌰͒ for the MCIS and Ni/C mirror in the
soft x-ray range at wavelength ϭ8.33 Å ͑Al anode͒ are
shown in Fig. 2. The Ni/C mirror with period Lϭ51 Å hav-
ing reflectivity of 63% at ϭ1.54 Å was chosen as a refer-
ence sample.
The reflection coefficient from the MCIS at the maxi-
mum of the first Bragg peak at ϭ8.33 Å was calculated
*
using the formula Rmaxϭ(Imax /I0) g, where Imax is the inten-
sity of reflected beam at the maximum, I0 is the intensity of
the incidence beam, and g is the geometry factor. The latter
was introduced to take account of the sample size. For MCIS
we have gϭ3.4 but for Ni/C—mirror gϭ1.1. In this case
Since the main parameters that affect reflection at wave-
length 1.54 Å are the difference in layer densities and sub-
strate surface roughness, we have also calculated Rmax which
can be achieved by increasing ⌬ and decreasing . These
Rmax are listed in Table II.
R
maxϭ10%, which is half as much as the reflectivity for the
Ni/C mirror ͑Rϭ22%͒. Just as in the short wavelength range,
the half-width of the Bragg peak for MCIS was appreciably
smaller ͑⌬⌰ϭ0.12°͒ than for Ni/C mirrors ͑⌬⌰ϭ0.4°͒. The
error in the determination of R is mainly due to the small size
of experimental structures, the small angle of incidence of
x-rays, and the design of the experimental setup.
The resolution ͑⌬/͒ and the half-width of the peak
͑⌬⌰͒ are related by ⌬/ϭ⌬/tg͑⌰͒, where ⌰ is the angular
position of the Bragg peak.2 Using this relation, we evaluated
the resolutions for the MCIS and the Ni/C mirror to be 0.043
and 0.085, respectively.
Measured and calculated Rmax in the range from 7.33 to
44.7 Å are shown in Table III. The calculations were made
for the MCIS, whose parameters are listed in Table II ͑No.
1͒.
The measured reflectivity for MCIS is seen to decrease
from 12% to 0.2% as the wavelength increases from 7.33 to
44.7 Å. This is the result of the increase in the carbon ab-
sorption coefficient as approaches the carbon absorption
As is seen from Table II, increasing ⌬ to 0.5 g/cm3 at
fixed leads to an increase in Rmax up to 61%. Decreasing
to 4 Å at fixed ⌬ϭ0.5 g/cm3 leads to a further increase in
R
max , namely, by 11%. A further increase in ⌬ to 0.6
g/cm3 enables one to reach the value of 79% for the coeffi-
cient of reflection from MCIS. It is important to note that
⌬ϭ0.6 g/cm3 is practically attainable in the process of car-
bon film preparation, as the carbon film density usually var-
ies from 1.2 to 3 g/cm3.6 Therefore, the goal of preparing
TABLE II. MCIS parameters used in the calculations.
⌬
͑g/cm3)
͑Å͒
Rmax
%
No.
1
2
3
4
0.38
0.5
0.5
0.6
8
8
4
4
40.7
61
72
78
306 Appl. Phys. Lett., Vol. 69, No. 3, 15 July 1996 Kondrashov et al.
130.239.20.174 On: Mon, 24 Nov 2014 11:19:52