All dielectric hard x-ray mirror by atomic layer deposition
Adriana Szeghalmi, Stephan Senz, Mario Bretschneider, Ulrich Gösele, and Mato Knez
Citation: Appl. Phys. Lett. 94, 133111 (2009); doi: 10.1063/1.3114402
View online: http://dx.doi.org/10.1063/1.3114402
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v94/i13
Published by the American Institute of Physics.
Related Articles
Submicron-diameter phase-separated scintillator fibers for high-resolution X-ray imaging
Appl. Phys. Lett. 102, 051907 (2013)
Full-field transmission x-ray imaging with confocal polycapillary x-ray optics
J. Appl. Phys. 113, 053104 (2013)
Large-field high-contrast hard x-ray Zernike phase-contrast nano-imaging beamline at Pohang Light Source
Rev. Sci. Instrum. 84, 013707 (2013)
Compensational scintillation detector with a flat energy response for flash X-ray measurements
Rev. Sci. Instrum. 84, 013103 (2013)
Endstation for ultrafast magnetic scattering experiments at the free-electron laser in Hamburg
Rev. Sci. Instrum. 84, 013906 (2013)
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 20 Feb 2013 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
APPLIED PHYSICS LETTERS 94, 133111 共2009兲
All dielectric hard x-ray mirror by atomic layer deposition
Adriana Szeghalmi,1,a兲 Stephan Senz,1 Mario Bretschneider,2 Ulrich Gösele,1 and
Mato Knez1,b兲
1
Max Planck Institute of Microstructure Physics, Weinberg 2, D-06120 Halle, Germany
IFG Institute for Scientific Instruments GmbH, Rudower Chaussee 29/31, D-12489 Berlin, Germany
2
共Received 8 January 2009; accepted 16 March 2009; published online 3 April 2009兲
Mirrors consisting of Al2O3 and Ta2O5 共⬃2 nm film thickness兲 nanolaminates for hard x-ray
wavelengths were produced by atomic layer deposition and characterized. Atomic force microscopy
and transmission electron microscopy 共TEM兲 proved extremely smooth surfaces of the mirrors,
which are critical for highest reflectance. TEM images showed sharp interfaces between the oxides.
The experimental x-ray reflectivity data were theoretically modeled and indicated minimal random
thickness variations in the individual layers. Additionally, a depth graded sample with a total
thickness of ⬃4 ␮m for focusing applications in transmission 共Laue兲 geometry and capillaries was
coated. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3114402兴
The rapidly increasing interest in x-ray spectroscopy and
the goal to reach diffraction limited resolution in x-ray microscopy set off the need for advanced optics.1,2 The fabrication of the required advanced optical elements is technically challenging. Suitable materials must have low x-ray
absorbance limiting the available choices. X-ray mirrors
based on Bragg diffraction of nanolaminates3 and zone
plates4 pose challenges to both coating technologies and lithography. The period of such optical elements has to be
very small 共often below 10 nm兲 with extreme thickness precision and smooth interfaces. Atomic layer deposition 共ALD兲
is a promising technology fulfilling these requirements. Metallic coatings for applications in x-ray spectroscopy were
already demonstrated.3,4
Up to date only a few materials mainly based on W, Pt,
Ta, and Ir have found applications in hard x-ray optics.1–5
The performance of such multilayer optics 共peak reflectivity
and full width at half maximum兲6,7 is constrained by interfacial smoothness, interlayer mixing/interdiffusion, internal
stress and nanocrystallinity, as well as thermal stability of the
multilayers. An increased variety of available materials could
possibly lead to more stable, efficient, and long-lasting op-
tics. In this manuscript nanolaminates with application potential in x-ray spectroscopy comprising two alternating dielectric materials 共Al2O3 and Ta2O5兲 deposited by ALD are
presented.
Extensive theoretical x-ray reflectivity 共XRR兲 modeling
of a variety of materials combined with the requirement of
high density and availability by standard ALD have led to
the choice of Al2O3 and Ta2O5. Based on calculations, the
film thicknesses were chosen to produce x-ray mirrors with
the first Bragg peak at ⬎0.5° grazing angle for the 1.54 Å
Cu K␣ line. Table I gives a compilation of the nanolaminate
samples fabricated on silicon wafers. Figures 1共a兲 and 1共b兲
show the cross-sectional transmission electron microscopy
共TEM兲 images of nanolaminates with a total of 21 and 81
共sample 2兲 alternating layers, respectively. The film thicknesses determined from TEM are 7.2 and 2.1 nm for Al2O3
and Ta2O5, respectively. The film thicknesses are in good
agreement with the data obtained from XRR of sample 2
共6.34 and 2.05 nm, see Table I兲. The reflectivity spectrum of
the sample with 21 layers is not shown here. The Bragg
peaks of the 21 layer sample overlap with the peaks of
the thicker sample, however, with lower intensity. The
TABLE I. Compilation and characteristics of the samples coated by ALD with the experimental thickness and surface roughness 共␴, nanometer兲 共area 1
⫻ 1 ␮m2兲 determined by AFM. The thickness of sample 1 was determined by spectroscopic ellipsometry with an effective medium approximation 共EMA兲
layer for modeling the surface roughness. The thicknesses of samples 2–4 correspond to data from the fit of the XRR spectra. The layer thicknesses of the
depth-graded sample 5 varied from 9 to 10.8 nm with a step size of ⬃0.1 nm. Superscript numbers correspond to the number of ALD cycles 共c兲 applied to
deposit the layer. The substrate is a single crystalline silicon wafer.
No.
Composition
1
2
3
4
共84.3 nm, EMA 1.3 nm兲
Ta2O1000c
5
21c
Ta2O100c
共Al2O55c
5
3 / / Ta2O5 兲 ⫻ 40, 81 layers
300c
51c
71c
Ta2O5 共Al2O3 / / Ta2O5 兲 ⫻ 30, 61 layers
51c
21c
Ta2O21c
5 共Al2O3 / / Ta2O5 兲
71c
⫻ 20/ / 共Al2O51c
3 / / Ta2O5 兲 ⫻ 20, 81 layers
91c
407 layers from 共Al2O71c
3 / / Ta2O5 兲 ⫻ 5
108c
to 共Al2O88c
/
/
Ta
O
兲
⫻
9
2 5
3
5
Thickness
共nm兲
␴ 共nm兲
84.3
10.0 共6.34/ / 2.05兲 ⫻ 40
34.0 共4.9/ / 5.4兲 ⫻ 30
0.43
0.37
0.56
9.7 共4.8/ / 2.3兲 ⫻ 20 共4.8/ / 5.4兲 ⫻ 20
0.51
共from ⬃9.0 to ⬃10.8兲 共total ⬃4 ␮m兲
0.77
a兲
Electronic mail: szeghalm@mpi-halle.mpg.de.
Electronic mail: mknez@mpi-halle.de. Tel.: ⫹49-345-5582642. FAX: ⫹49-345-5511223.
b兲
0003-6951/2009/94共13兲/133111/3/$25.00
94, 133111-1
© 2009 American Institute of Physics
Downloaded 20 Feb 2013 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
133111-2
Szeghalmi et al.
Appl. Phys. Lett. 94, 133111 共2009兲
FIG. 1. TEM images of nanolaminates with 共a兲 21 and 共b兲 81 共sample 2兲 layers. 共c兲 Cross-section FIB SEM image of a detail of sample 5; the inset shows
a higher magnification detail.
constructive interference of the reflections at the interfaces
was maximized in the sample with 81 layers.
The TEM images in Figs. 1共a兲 and 1共b兲 demonstrate the
existence of sharp interfaces between the Al2O3 and Ta2O5
layers. The outermost layers show a minimal increase in
roughness compared to the substrate surface. Atomic force
microscopy 共AFM兲 measurements give a root mean square
roughness of ⬍0.4 nm for sample 2, which is in good agreement with the values from TEM images and variable angle
spectroscopic ellipsometry measurements. AFM and TEM
measurements provide local information on the roughness
with investigated regions of up to 10⫻ 10 ␮m2, whereas
ellipsometry analyzes an area of several square millimeters.2
Smooth surfaces are critical for high XRR. It is important to consider both the quality of the outermost layer and
the quality of the inner layers. Roughness reduces the interference of the reflections at the interfaces and also facilitates
diffusion of the nanolaminate components under operating
conditions of the optical elements, thus reducing their longterm stability.
Normalized intensities were difficult to determine experimentally due to the sample curvature and sample illumination, even though silicon wafers with a diameter of 100
mm were coated and the beam size was reduced to less than
100 ␮m. Large area samples were necessary to ensure that
the x-ray beam was completely reflected by the surface even
at small grazing angles. However, the coating process can
also induce some bending of the wafers due to stress in the
layers. This bending could cause additional focusing of the x
rays. This would lead to a higher reflectivity in the Bragg
peaks compared to the total internal reflection intensity
around the 0.2° grazing angle. The normalized intensity of
the first Bragg peak for samples 2–4 amounted to a value of
⬎90% at grazing angles of 0.62° and 0.54°, respectively. In
contrast, the performance of standard metallic mirrors at 8
keV rapidly decreases with increasing grazing angles with
reflectance values in the range of 80%–95% below 1°.
The XRR spectra of samples 2–4 are presented in Figs. 2
and 3. The experimental spectra were fitted using the IMD
software package8 to determine the thickness of the layers.
The optical constants of Al2O3 and Ta2O5 were applied in
the calculations as included in the software based onto the
Centre for X-ray Optics 共CXRO兲 atomic scattering factors.
Additional fit parameters 共roughness, random thickness
variation兲 did not significantly improve the fit. The film
thicknesses were controlled by the number of ALD cycles.
The position of the first order Bragg peak could be shifted
toward higher grazing angles if the thickness of the Ta2O5
layer and the period of the multilayer were reduced. Sample
2 with a Ta2O5 thickness of ⬃2 nm shows the first order
Bragg peak at 0.62° while sample 3 with ⬃5.4 nm Ta2O5 at
0.54°. In sample 4, an additional multilayer with ⬃2.3 nm
Ta2O5 content was deposited below the sequence of sample
FIG. 2. XRR spectra 共experimental symbol and fit line兲 of sample 2.
Downloaded 20 Feb 2013 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
133111-3
Appl. Phys. Lett. 94, 133111 共2009兲
Szeghalmi et al.
FIG. 4. XRR spectrum of sample 5.
FIG. 3. XRR spectra 共experimental symbol and fit line兲 of samples 共a兲 3 and
共b兲 4.
3. As expected, the XRR spectrum of sample 4 showed secondary Bragg peaks at 0.68° and 1.28° compared to sample 3
共see Fig. 3兲. Their intensity is much lower and shows that the
interference from the “buried” layers was perturbed by the
top multilayer.
The effect of random thickness variations in the layers
was theoretically investigated. Random thickness variations
lead to broader and weaker Bragg peaks in the calculated
XRR spectra. The calculated spectra with 5 Å random thickness variation already lacked the second order Bragg peak
reflection. The experimental XRR data, however, showed
sharp and intense Bragg peaks up to the eighth order 共4.29°
grazing angle, see Fig. 2兲 and proved minimal random thickness variations.
We further tested the possibility of growing a larger
number of multilayers 共total thickness of ⬃4 ␮m with depth
grading from 9 to 10.8 nm and with 0.1 nm step increase兲 by
ALD. Such films with increasing period found applications
as x-ray optics of sectioned multilayers in Laue 共transmission兲 geometry. Not only the control of thickness and uniformity of the deposited films but also the sectioning of the
samples pose severe difficulties.9 By applying focused ion
beam 共FIB兲 cross sectioning damage-free samples 关see Fig.
1共c兲兴 of the ALD multilayer could be produced. The thickness of the cross section was ⬃600 nm. The lateral size of
the prepared cross section is ⬃20⫻ 4 ␮m2. However, larger
sections could also be achieved. The stability of the ALD
stack during the sectioning was ensured with a Pt layer deposited on top of the multilayer. The ALD stack consisting of
Al2O3 and Ta2O5 layers was very stable. This is also confirmed by the fact that sectioning to ⬃60 nm by ion beam
and mechanical polishing of sample 2 with a multilayer
thickness of ⬃400 nm could be performed without damaging the multilayer.
The XRR spectrum of the thick multilayer 共sample 5兲 is
presented in Fig. 4. The high intensity of the first order
Bragg peak 共⬎80% of the total internal reflectance value at
⬃0.2°兲 is of special interest.
Finally, preliminary efficiency data of coated capillaries
共sequence samples 1 and 2兲 with a diameter of 500 ␮m
共N16B glass兲 were recorded. The capillaries were cut at 9.5
cm length after ALD deposition. Compared with a standard
500 ␮m pinhole setup, the coated capillaries had 2.6 and 1.2
enhancement factors for the Cu K␣ and Mo K␣ lines, respectively. The pure Ta2O5 coating led to similar enhancement
factors as the nanolaminate comprising 81 layers within the
experimental accuracy.
In conclusion, nanolaminates consisting of Al2O3 and
Ta2O5 for applications in x-ray optics were produced and
characterized. The nanolaminates showed high x-ray reflectance with multiple Bragg peaks. FIB cross sectioning of a
nanolaminate with large total thickness was performed without damaging the interfaces. Further analysis of the focusing
efficiency of these sections as well as improvement of ALD
coated capillaries are in progress.
This work was funded by the German Federal Ministry
of Education and Research 共BMBF兲 in the framework of the
Nanofutur program 共Contract No. 03X5507兲. The authors
thank Sigrid Hopfe and Norbert Schammelt for sample
preparation for TEM investigations.
1
H. C. Kang, J. Maser, G. B. Stephenson, C. Liu, R. Conley, A. T.
Macrander, and S. Vogt, Phys. Rev. Lett. 96, 127401 共2006兲.
2
T. Weitkamp, O. Dhez, B. Kaulich, and C. David, Proc. SPIE 5539, 195
共2004兲.
3
T. Salditt, S. P. Krüger, C. Fuchse, and C. Bähtz, Phys. Rev. Lett. 100,
184801 共2008兲.
4
F. H. Fabreguette, R. A. Wind, and S. M. George, Appl. Phys. Lett. 88,
013116 共2006兲.
5
K. Jefimovs, J. Vila-Comamala, T. Pilvi, M. Ritala, and C. David, Phys.
Rev. Lett. 99, 264801 共2007兲.
6
L. Jiang, B. Verman, B. Kim, Y. Platonov, Z. Al-Mosheky, R. Smith, and
N. Grupido, Rigaku J. 18, 13 共2001兲.
7
F. E. Christensen, W. W. Craig, D. L. Windt, M. A. J. Garate, C. J. Hailey,
F. A. Harrison, P. H. Mao, J. M. Chakan, E. Ziegler, and V. Honkimaki,
Nucl. Instrum. Methods Phys. Res. A 451, 572 共2000兲.
8
D. L. Windt, Comput. Phys. 12, 360 共1998兲.
9
H. C. Kang, G. B. Stephenson, C. Liu, R. Conley, R. Khachatryan, H. Yan,
J. Maser, J. Hiller, and R. Koritala, Rev. Sci. Instrum. 78, 046103 共2007兲.
Downloaded 20 Feb 2013 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions