Extreme-ultraviolet multilayer coatings with highspectral purity for solar imaging

Michele Suman,1,2,* Maria Guglielmina Pelizzo,1,2 David L. Windt,3

and Piergiorgio Nicolosi1,2

1National Research Council–National Institute for the Physics of the Matter, Laboratory for Ultravioletand X-ray Optical Research, via Gradenigo, 6B–35131 Padova, Italy

2Department of Information Engineering, University of Padova, via Gradenigo, 6B–35131 Padova, Italy3Reflective X-Ray Optics LLC, 1361 Amsterdam Avenue, Suite 3B, New York, New York 10027, USA

*Corresponding author: sumanmic@dei.unipd.it

Received 2 July 2009; revised 3 September 2009; accepted 8 September 2009;posted 10 September 2009 (Doc. ID 113417); published 1 October 2009

Future solar experiments designed to perform solar plasma diagnostics will also be based on extreme-ultravilet observations. Multilayer (ML) optics are essential in this spectral region since these coatingshave high reflectivity at normal incidence. Typically, the reflectivity curve of aML coating has a small butfinite bandwidth, and this can be a serious drawback when several spectral lines fall within the band-width. In fact, spectral lines emitted by different ion species can correspond to different plasma condi-tions. We present the design, realization, and characterization of an innovative ML structure with highreflectivity coupled with a strong rejection ratio for two adjacent spectral features. The key element is anoptimized capping layer structure deposited on top of the ML that preserves the performance reflectanceat the target wavelength and at the same time suppresses the reflectance at specific adjacent wave-lengths. Application to the Fe XV 3 × 106 K coronal emission line at 28:4nm with rejection of the HeII Lyman-α line at 30:4nm is presented. © 2009 Optical Society of America

OCIS codes: 310.4165, 310.0310, 350.6090.

1. Introduction

High normal incidence reflectivity in the extreme-ultraviolet (EUV) and soft x-ray spectral rangescan be obtained only with multilayer (ML) structuresthat consist of an alternating sequence of thin filmsof highly absorbing (absorber) and less absorbing(spacer) materials to achieve high optical contrastand designed in such a way as to allow the electricfield components reflected at the various interfacesto add in phase in a selected spectral range [1].The typical thickness of a single layer is a few nano-meters, depending on the spectral range. Because ofthe high throughput at near-normal incidence, MLoptics have been the enabling technology for solarimaging instruments in the EUV spectral range.

In the past decade, normal-incidence telescopes thatuse narrowband ML mirrors have been flown aboardrockets and satellites, as an EUV Imaging Telescope(EIT) onboard the Solar and Heliospheric Observa-tory (SOHO) [2], and Transition Region and CoronalExplorer (TRACE) [3], to image the Sun at some se-lected EUV narrowband spectral ranges. Further-more, future space missions, for example, the SolarOrbiter (SOLO)—European Space Agency (ESA)cornerstone, or the Geostationary Operational Envir-onmental Satellite (GOES)-R Series, within theframework of NASA missions dedicated to spaceweather analysis, will require the use of such tech-nology. MLs for solar line emission observations,working at Fe IX (17:1nm), Fe XII (19:5nm), Fe XV(28:4nm), and He II (30:4nm) have been fabricated.The performance of such MLs is mainly evaluated interms of peak reflectivity at a working wavelengthand rejection capability of unwanted lines; in fact,

0003-6935/09/295432-06$15.00/0© 2009 Optical Society of America

5432 APPLIED OPTICS / Vol. 48, No. 29 / 10 October 2009

the bandwidth of such coatings is not negligible;therefore the detected signal often includes the con-tribution of both target lines and adjacent spectrallines. A suitable dual-band Mo/a-Si ML solution thatcuts down the He I reflection at 58:4nmwhile havinggood reflectivity at He II 30:4nm has already beenproposed [4]. In this case the two spectral bands arequite separate. A different serious problem can de-rive from lines adjacent to the working wavelengththat can, consequently, pollute the detected signal.For example, in the case of a Mo/a-Si periodic struc-ture optimized for 28:4nm observation, the reflectiv-ity curve results are quite spectrally broad and thevery close and brilliant He II line falls within thebandwidth. This fact can seriously affect diagnosticsbased on the Fe XV line signal. The Mo/a-Si materialis generally used for the multilayer stack because ofits high stability, even though enhancement of peakreflectivity and rejection capability could, in princi-ple, be obtained by use of other material combina-tions such as Mg/SiC [5], B4C=Si=Mo [6], or B4C=Si[5]. In contrast, stability of such material combina-tions over time is a critical issue for long-term spacemissions and is under investigation [7]. Dependingon the different environments in space missions,MLs can be subject to high thermal load, oxidation,or ion/electron bombardment. Accordingly, a key ele-ment in such nanostructures is the capping layer(CL), i.e., the final layers deposited on top of theML that should protect the structure from environ-mental attack [8]. The CL structure has also beenstudied to provide additional reflectivity in differentspectral bands such as the UV and the visible. Thisfact allows us to create a more compact telescope inwhich different spectral regions share a common op-tical path and in which multiband observation is con-cretized at the detector level [9] and allows a morecomplete plasma diagnostics by deriving differentphysical parameters.

Here we present the design, realization, and char-acterization of Mo/a-Si ML coatings with improvedsignal-to-noise ratio. This property is obtained witha CL structure that can be either the same Mo/a-Simaterial combination or some other material for im-proved performance [10,11]. This structure is speci-fically designed to reject unwanted lines whilepreserving the peak reflectivity of the ML. The de-sign principle of the structure is presented, as wellas some examples of possible CL structures. Realiza-tion and characterization of a periodic Mo/a-Si MLcovered by a Mo/a-Si CL optimized for the observa-tion of the Fe XV line while rejecting the He II strongline is subsequently presented; rejection ratio cap-abilities better than 10−4 are proved.

2. Multilayer Design Method

When EUV radiation interacts with a ML structure,the superposition of the incident and reflected elec-tromagnetic waves generates a standing wave fielddistribution in the ML structure (Fig. 1). By tailoringthe distribution of this field inside the structure (i.e.,modifying the thickness of each layer) it is possible toattribute some specific properties to the ML.

Let us call λpeak the selected working wavelengthand λnoise the wavelength to be rejected. Hereafter,MLmeans the structure constituted by the repetitionof two or more materials optimized to achieve thebest reflectivity at λpeak; the CL is the structure ontop of the ML that suppresses the λnoise component.Like the ML, the CL structure consists of a sequenceof absorber and spacer material layers that can bethe same or different from those in the underlyingML, with a suitably optimized thickness.

For the optimization design of such a structure weused the following steps:

1. Design the periodic ML underneath the CL toachieve maximum reflectivity at λpeak.

2. Computation of the standing wave in the MLstructure for λpeak as well as computation of thestanding wave in the ML structure at λnoise.

Fig. 1. Schematic representation of a standing wave inside theML and the four-step optimization.

Table 1. ML with a Mo/a-Si Periodic Structure and aMo/a-Si CL for Example 1

CL Structure Thickness (nm) ML Structure Value

a-Si 14.7 Period d (a-Si/Mo) 15:15nmMo 2.2 Ratio a-Si/d 0.868a-Si 57.75 Nnumber of periods 35Mo 2

Table 2. ML with a Mo/a-Si Periodic Structure and anAl=MgF2 CL for Example 2

CL Structure Thickness (nm) ML Structure Value

MgF2 5 Period d (a-Si/Mo) 15:33nmAl 10.25 Ratio a-Si/d 0.77MgF2 3.5 Number of periods 35Al 25.4MgF2 3.5

Table 3. Reflectivity Computed at the λpeak (28:4 nm) and λnoise(30:4 nm) Wavelengths for Examples 1 and 2 for a Standard Periodic

a-Si/Mo ML Assuming 0:5 nm and Roughness at Interfaces

R28:4nm R30:4nm

Optimized MLþMo=Si CL 0.196 6:4 × 10−5

Optimized MLþ Al=MgF2

CL0.195 3:7 × 10−5

Standard periodic ML 0.25 0.045

10 October 2009 / Vol. 48, No. 29 / APPLIED OPTICS 5433

3. Optimization of thickness and number of CLlayers to obtain:

a. Absorber layers that correspond with the λpeakstanding wave nodes.b. Minimization of the distance between absorber

layer and antinodes of the λnoise standing wave; inthis way the absorption at λnoise is considerably en-hanced while the effect at λpeak is less important.4. Reoptimization of the ML period and γ ratio

parameters to improve the ratio RðλpeakÞ=RðλnoiseÞand preserve peak reflectivity.

The procedure is also applicable to aperiodic MLdesign; in this case step (4) means reoptimization ofeach layer thickness.

A. Designs with Different Capping Layers and MaterialCombinations

The design method can be used with different mate-rial combinations and applied to both periodic andaperiodic structures; moreover, the CL can be com-posed of different materials, therefore making it pos-sible to use CLs that can be used to protect thecoating in harsh environmental conditions or to re-flect visible and/or UV radiation.In the following, we present a Mo/Si ML that is

specifically targeted to reflect the 28:4nm line while

rejecting the strong 30:4nm line. CLs have been de-signed with both Mo/a-Si (Example 1) and Al=MgF2material combinations (Example 2).

Optimization and simulations have been per-formed with the IMD program [12] for a 5° incidenceangle and using the Center for X-Ray Optics (CXRO)database (a-Si 1997, and Mo 1997). The realizationof Example 1 is much simpler, since the materialsused in the CL structure are the same as the onesused for the ML; realization of Example 2 would re-quire the use of four different targets or a two-stepdeposition process. The two structures are listed inTables 1 and 2, respectively; the expected perfor-mances are listed in Table 3 and are compared withthose of a standard periodic a-Si/Mo ML without aCL and optimized for best peak reflectivity at28:4nm. Roughness at interfaces of 0:5nm has beenassumed for the calculation. Structures that endwith an a-Si layer are assumed to develop on topof a 2nm SiO2 oxide layer; nevertheless this samestructure can also be optimized with the topmostRuO2 layer for protection in harsh environments.Different from a standard periodic structure, bothExamples 1 and 2 have a rejection capability at30:4nm, which is considerably lower than 10−4.Figure 2 shows the simulated reflectivity curvesand compares them with those of a standard periodicstructure.

The standing wave distribution for Example 1is shown in Fig. 3. Mo has a relevant absorption

Fig. 2. (a) Reflectivity [log scale case, (b)] curve for the structuresof Example 1 (dotted curve), Example 2 (solid curve), and a stan-dard periodic structure (dash–dot curve).

Fig. 3. Standing wave pattern for the Example 1 structure at (a)28:4nm and (b) 30:4nm wavelengths.

Table 4. Reflectivity Computed at the λpeak (28:4 nm) and λnoise (30:4 nm) Wavelengths for Optimized a-Si/B4C and SiC/Mg MLs and for theCorresponding Standard Periodic MLs

Materials Structures Rð28:4nmÞ Rð30:4nmÞ

a-Si=B4C (0:5nm roughnessat interfaces)

Periodic ML [period d ¼ 154:8, ratio a-Si=d ¼ 0:24] 0.3 0.02Periodic ML ½periodd ¼ 154:8; ratioa-Si=d ¼ 0:24�þCL½22:4nm;a-Siþ 8nm;B4C�

0.26 7:8 × 10−4

SiC/Mg (1:7nm roughnessat interfaces)

Periodic ML d ¼ 145:6 ratio Mg=d ¼ 0:75 0.45 0.018Periodic ML ½periodd ¼ 145:6; ratio Mg=d ¼ 0:75�þCL½21:7nm;Mg þ 6nm;SiC�

0.4 2:7 × 10−4

5434 APPLIED OPTICS / Vol. 48, No. 29 / 10 October 2009

coefficient in this spectral region. By placing the Molayer to correspond with a node of the 28:4nm stand-ing wave [see Fig. 3(a)], the absorption of this radia-tion is minimized and the reflectivity is maximized.The sameMo layer also corresponds to an antinode ofthe 30:4nm radiation [see Fig. 3(b)], which causes aquasi-total absorption of such spectral components.To test the flexibility of the proposed method, it

has been applied to multilayers with different com-binations of material. Table 4 compares the perfor-mances of a-Si=B4C and SiC/Mg MLs that wereobtained by use of the proposed design criteria withthose of standard periodic MLs made of the samematerial combinations.

3. Tolerance Analyses

Material optical constants for thin films can dependon deposition techniques and procedures. Therefore,how the proposed design critically depends on opticalconstants has been a matter of investigation. The op-tical constants and imaginary coefficients of a-Si[Fig. 4(a)] and Mo [Fig. 4(b))] as provided by theCXRO and the IMD database are reported togetherwith that published by Tarrio et al., which were ex-perimentally determined in situ during a sputteringdeposition process [13]; as shown, there are no neg-ligible differences. The structure reported in Table 1has therefore been reoptimized by use of the con-stants from Ref. [13]; the newly derived structure(Example 1B) and its performance are reported in

Tables 5 and 6, respectively. The variation in opticalconstants results in a quite different design of the CLstructure, demonstrating how sensitive the design isto such parameters. Figure 5 shows the reflectivitycurves of Example 1 and Example 1B. The new struc-ture preserves a high peak reflectivity, although therejection ratio results are lower, which is due to dif-ferent optical contrast and radiation absorption.

In both Examples 1 and 1B the most critical layerseems to be the second one of a-Si, which has a thick-ness of 57:75nm if the IMD database optical con-stants are used and 41:2nm if the Tarrio et al. dataare used. Simulations show that a relative thicknesserror of only a few percent can considerably affect thefinal performance resulting in a lower reflectivitypeak and a shift in the peak and rejected wave-lengths, as reported in Fig. 6 for Example 1B. Asshown, if the layer thickness departs from the opti-mum value, the reflectivity at 30:4nm rises quicklyand diminishes the rejection ratio of the structure.

4. Experimental Results

The Example 1B structure has been deposited by Re-flective X-ray Optics (RXO) with a magnetron sput-tering facility [14]. After some preliminary sampledepositions and testing, the new optical constants re-ported in Ref. [13] were adopted into the design pro-cedure. The structure has been characterized at theRXO laser plasma facility just after deposition [15].Since this first measurement was limited by the

Fig. 4. Imaginary parts of the complex refractive index of (a) a-Siand (b) Mo: solid curve, optical constants from the CXRO database;dotted curve, optical constants from Ref. [13].

Table 5. ML Optimized with Ref. [13] Optical Constantsfor Example 1B

CL Structure Thickness (nm) ML Structure Value

a-Si 15.4 Period (a-Si/Mo) d 153nmMo 3.55 Ratio a-Si/d 0.768a-Si 41.2 Number of periods 40Mo 3.55

Table 6. Reflectivity Computed at the λpeak (28:4 nm) and λnoise w(30:4 nm) Wavelengths for the Example 1B Structure and for the

Corresponding Standard Periodic ML, Assuming 0:5 nmRoughness at Interfaces

R28:4nm R30:4nm

Optimized MLþMo=Si CL 0.174 3 × 10−3

Standard periodic ML 0.2 0.05

Fig. 5. (a) Dashed curve, reflectivity of the Example 1 structure(Table 1); solid curve, reflectivity of the Example 1B structure(Table 5). (b) Same as (a) but on a logarithmic scale.

10 October 2009 / Vol. 48, No. 29 / APPLIED OPTICS 5435

signal-to-noise ratio of the experimental facility, newmeasurements were performed at the AdvancedLight Source (ALS) Synchrotron Facility one monthafter deposition [16] to measure the rejection ratioaccurately at the 30:4nmwavelength. Results are re-ported in Fig. 7 together with the expected theoreti-cal curve. The measurements show a reflectivitypeak of approximately 0.19 and a minimum reflectiv-ity value of approximately 6 × 10−5; this result de-monstrates the possibility of a design with thepractical feasibility of a high peak reflectivity struc-

ture having a high rejection capability of an adjacentunwanted spectral line.

From the data in Fig. 7 one observes that the ex-perimental curve shows some slight differences com-pared with the computed curve: the reflectivity peakis higher and spectrally shifted by approximately0:1nm; the minimum of the curve is at a longerwavelength (32:5nm instead of 30:4nm) and an ad-ditional minimum appears on the short wavelengthside. As demonstrated in Section 3, the structure hastwo main critical tolerances: thickness of the seconda-Si layer and optical constant values. During the ex-perimental deposition session, the thickness deposi-tion errors must be evaluated to less than 0:1nm;therefore, they should not affect the final perfor-mance based on the critical tolerances shown inFig. 6. Consequently, the main cause of mismatch be-tween experimental and theoretical results must beattributed to the uncertainty of the optical constants.To tune the final deposition process it is, therefore,necessary to correct the layer thickness, which canbe done by having a better knowledge of the opticalconstants or by adopting a suitable metrology relatedto the standing wave distribution.

5. Conclusion

An innovative method for the design of multilayerstructures with improved spectral filtering per-formance has been presented. Structures with highreflectivity at the 28:4nm Fe XV line and high rejec-tion at the strong 30:4nm He II line have beendesigned. Preliminary samples have been achievedand tested, demonstrating the actual feasibility ofthe structures.

This research has been supported by an AgenziaSpatiale Italiana grant I/015/07/0. The researchhas also been performed within the framework ofthe European Cooperation in Science and Technology(COST) ActionMP0601, “Short wavelength radiationsources.” The authors gratefully acknowledge ErickM. Gullikson, Center for X-Ray Optics, LawrenceBerkeley National Laboratory, for the experimentalmeasurements at the ALS Synchrotron facility.

References1. E. Spiller, Soft X-Ray Optics (SPIE Press, 1994), Chap. 8.2. http://umbra.nascom.nasa.gov/eit/3. http://trace.lmsal.com/4. B. J. Zhu, Z. Wang, Z. Zhang, F. Wang, H. Wang, W. Wu,

S. Zhang, D. Xu, L. Chen, H. Zhou, T. Huo,M. Cui, and Y. Zhao,“High reflectivity multilayer for He-II radiation at 30:4nm,”Appl. Opt. 47, C310–C314 (2008).

5. S. Zuccon, D. Garoli, M. G. Pelizzo, P. Nicolosi, S. Fineschi, andD. Windt, “Multilayer coating for multiband spectral observa-tions,” in Proceedings of the International Conference on SpaceOptics (European Space Agency, 2006).

6. J. Gautier, F. Delmotte, M. Roulliay, F. Bridou, M. F. Ravet,and A. Jérome, “Study of normal incidence of three-componentmultilayer mirrors in the range 20–40nm,” Appl. Opt. 44,384–390 (2005).

7. A. D. Rousseau, D. L. Windt, B. Winter, L. Harra,H. Lamoureux, and F. Eriksson, “Stability of EUVmultilayersto long-term heating, and to energetic protons and neutrons,

Fig. 6. Reflectivity at (a) 28:4nm and (b) 30:4nm as a function ofthe thickness of the critical a-Si layer. The shaded areas corre-spond to the a-Si thickness range where the reflectivity at (a)28:4nm is higher than 0.125 and (b) 30:4nm is lower than 2 × 10−2.

Fig. 7. Experimental results derived fromALSmeasurements forthe optimized structure of Example 1B in comparison with the the-oretical simulation performed taking into account the optical con-stants of Ref. [13]. (b) Same as (a) but on a logarithmic scale.

5436 APPLIED OPTICS / Vol. 48, No. 29 / 10 October 2009

for extreme solar missions,” Proc. SPIE 5900, 590004(2005).

8. S. Bajt, N. V. Edwards, and T. E. Madey, “Properties of ultra-thin films appropriate for optics capping layers exposed tohigh energy photon irradiation,” Surf. Sci. Rep. 63, 73–99(2008).

9. M. G. Pelizzo, D. Gardiol, P. Nicolosi, A. Patelli, and V. Rigato“Design, deposition, and characterization of multilayer coat-ings for the Ultraviolet and Visible-Light Coronagraphic Im-ager,” Appl. Opt. 43, 2661–2669 (2004).

10. M. Suman, M. G. Pelizzo, D. L. Windt, G. Monaco, S. Zucconand P. Nicolosi, “Innovative design of EUV multilayer reflec-tive coating for improved spectral filtering in solar imaging,”in Proceedings of the International Conference on Space Optics(European Space Agency, 2008).

11. M. Suman, M.-G. Pelizzo, P. Nicolosi, and D. L. Windt, “Aper-iodic multilayers with enhanced reflectivity for extreme ultra-violet lithography,” Appl. Opt. 47, 2906–2914 (2008).

12. D. L. Windt, “IMD: Software for modeling the optical proper-ties of multilayer films,” Comput. Phys. 12, 360–370(1998).

13. C. Tarrio, R. N. Watts, T. B. Lucatorto, J. M. Slaughter, andC. M. Falco, “Optical constants of in situ-deposited films of im-portant extreme-ultraviolet multilayer mirror materials,”Appl. Opt. 37, 4100–4104 (1998).

14. D. L. Windt and W. K. Waskiewicz, “Multilayer facilities forEUV lithography,” J. Vac. Sci. Technol. B 12, 3826–3832(1994).

15. www.rxollc.com/16. http://www.als.lbl.gov/als/techspecs/bl6.3.2.html

10 October 2009 / Vol. 48, No. 29 / APPLIED OPTICS 5437