Characterization and control of laser plasma fluxparameters for soft-x-ray projection lithography

Martin Richardson, William T. Silfvast, Howard A. Bender, Art Hanzo, Victor P. Yanovsky,Feng Jin, and Jerry Thorpe

Laser plasmas are intrinsically an attractive soft-x-ray source for projection lithography. Compact,flexible, and small enough to be dedicated to a single installation, they offer an alternative to costlymulti-installation synchrotron sources. For laser plasmas to provide ideal sources of soft x rays forprojection lithography, their properties must be tuned to optimize several critical parameters. Highx-ray conversion in the spectral band relevant to projection lithography is obviously required and hasalready received the attention of several studies. However, other features, such as the spectral contentand direction of the x-ray emission, the plasma and particulate emission, the technology of the target, andefficient laser design, must also be optimized. No systematic study of all these features specifically forprojection lithography has yet been made. It is our purpose to optimize these parameters in acoordinated approach, which leads to the design of a source that satisfies all the demanding requirementsof an operating lithographic installation. We make an initial investigation of the plasma and particleemission of plasmas that have previously been shown to be good x-ray converters to the 13-nmband. The importance of the results reported may well force new approaches to the design of laserplasma soft-x-ray sources for projection lithography.

1. Introduction

In current projection x-ray lithography schemes,1-3the use of a bright soft-x-ray source operating in thewavelength region of 13 nm, where efficient highreflective mirrors have already been demonstrated, isenvisaged. Several possible candidate sources arecurrently under consideration, including compactsynchrotrons, 4 free-electron lasers,5 laser plasmas,6

and various types of discharge device.7 There ismuch interest in the concept of using bright laserplasma x-ray sources for soft-x-ray projection lithog-raphy. The use of a laser plasma instead of acompact synchrotron offers the advantages of flexibil-ity, cost, single-source dedication, and operationalconvenience. Compared with other possible pulsedx-ray sources, such as free-electron lasers and dis-charge devices, laser plasma. sources are fairly welldeveloped and understood and should be able toprovide the required soft-x-ray fluxes. Laser plasmasources, therefore, in principle, present a low-risk

The authors are with the Center for Research in Electro-Opticsand Lasers, University of Florida, Orlando, Florida 32826-357.

Received 16 July 1992.0003-6935/93/346901-10$06.00/0.© 1993 Optical Society of America.

path toward a compact, low-cost, soft-x-ray sourcesuitable for a single-stepper production facility foradvanced integrated circuits.

Although laser plasmas have been under intenseinvestigation for many years, the investigation beingdriven by their application to such projects as laserfusion and coherent x-ray laser generation, onlylimited studies have been made with sufficient spec-tral resolution in an operating regime approachingthat which might be optimum for soft-x-ray projec-tion lithography.6'8' 0 All these studies have beenmade with available, generally commercial, laser sys-tems and were performed with simple planar solid-material targets. Although recent studies of thesoft-x-ray emission spectra from plasmas producedunder these conditions have estimated conversionefficiencies of the laser light into x-ray emissionwithin the bandwidth of typical normal-incidencex-ray mirrors to be in the 1% range,' 0 to our knowl-edge no systematic study has yet been done to opti-mize all the parameters of a laser plasma specificallyfor the high-repetition-rate, long-duty-cycle opera-tional environment of a high-throughput wafer pro-duction facility.

Several unique and, in some respects, mutuallyconflicting characteristics are required of a laserplasma ideally configured as an x-ray source for

1 December 1993 / Vol. 32, No. 34 / APPLIED OPTICS 6901

projection lithography. Here we briefly examine themost important of these characteristics.

A. X-Ray ConversionThe laser plasma source must display high conversionefficiency to x rays in the region of 13 nm within anarrow bandwidth. This requirement is set, at thepresent time, by the availability of high reflectivenormal-incidence x-ray mirror coatings."

Because all projection lithography designs utilize alarge number of reflective surfaces in series (usually

7), the R7 dependence of the system transmittancewill strongly influence the choice of operating wave-length. This is currently restricted to a narrowspectral region near 13 nm, where multilayer Mo-Simirrors can be fabricated to near their theoreticalreflectivities (R 70%).12 Other characteristics oflaser plasma x-ray emission not usually found inthese sources would also be desirable. Since only asmall spectral bandwidth of radiation is used forprojection lithography, one would ideally want thesource to be spectrally narrow, thereby limiting theamount of unused x-ray radiation that will be ab-sorbed by the first surface mirror. The absorbedflux could be so high as to cause degradation ordamage to this mirror. The presence of this un-wanted flux would suggest ideally the creation of apredominantly x-ray line emission in the plasma.Unfortunately, the spectrum of laser plasma emis-sion usually comprises a broad continuum of Planck-ian radiation from free-bound plasma collisions, onwhich a host of emission lines at many wavelengthsfrom many ionic transitions are superimposed.13Another aspect relates to the geometric match be-tween the emission direction of the x-ray emissionand the numerical aperture (NA) of the first collectormirror (typically NA 0.15).14 An ideal sourcewould be one in which the x rays are emitted from thelaser plasma with an angular distribution correspond-ing to this aperture, thereby limiting the amount ofunused x-ray emission. X-ray emission from laserplasmas created with planar targets is usually conicalin nature, but with a solid angle much greater thanthis aperture would require.' 5 Last, since there areno ultrafast time-dependent characteristics requiredof projection lithography sources, ideally long-dura-tion emitting plasmas would be preferable, especiallyif this meant that an increase in the x-ray conversionefficiency would be gained. Most types of laserplasma designed for x-ray generation have a veryshort duration emission, typically a few nanosecondsat most (these are required for such applications aslaser fusion and x-ray laser generation). Intrinsically,since a laser plasma is inertially confined, its x-rayemission lifetime will be limited, but times largerthan these few nanoseconds would be beneficial.Thus the greater the degree to which all theseparameters can be manipulated to provide efficientlygenerated useful 13-nm x-ray emission, the lower theoutput power requirements, complexity, and cost ofthe laser will be. Given that this source must oper-

ate for long periods of time without major mainte-nance, improved reliability will result.

B. Plasma and Particulate EmissionA laser plasma source for projection lithography mustnot contaminate the laser focusing optics and thex-ray collecting optics. It is unreasonable and wouldbe too costly to assume that these optics can bereplaced after short periods of operation. Laserplasma x-ray sources inherently produce ballisticparticles emanating from the plasma in all directions.The magnitude, size, and velocity of these particles is,of course, dependent on the target and the irradiationconditions. All this flying particulate emission ispotentially hazardous to these optical elements, ei-ther in its ability to degrade their surface quality or inthe progressive degradation of their performance thatis due to overcoating with target material. Althoughit has long been recognized that plasma debris is apotential problem for high-speed lithography,' 6 it hasnot been the strong focus of previous investigations oflaser plasmas. Single-shot applications, such as la-ser fusion and x-ray lasers, are not impaired by thisdebris problem. Laser ablation material-coatingstudies actually depend on such particulate emissionbut are done in a laser intensity regime that is farremoved from that in which significant x rays aregenerated.17 However, to our knowledge no system-atic studies have so far been made in which thespecific requirements relative to soft-x-ray projectionlithography are addressed. No attempts have beenmade to minimize the particulate emission while atthe same time ensuring that strong x-ray emission ismaintained at 13 nm. In addition, for lithographyapplications it may well be necessary to devise tech-niques and devices that actually inhibit or capture theflow of these particles. Little work has been re-ported so far in this domain.

C. Target SystemA third requirement of a laser plasma x-ray source forprojection lithography is an inexpensive, continu-ously sequencing target system. Current designs fora stand-alone projection lithography installation callfor a laser plasma source operating at a frequency inthe vicinity of 1000 Hz.' 8 Assuming the need fornonstop operation for an 8-h operating period and therequirement of a fresh target per shot implies anoninterfering shot sequence of more than 2 x 107targets. The design of a system with such a capabil-ity is, of necessity, dependent on knowledge of theoptimum target design and irradiation conditions.Although workers at Sandia National Laboratorieswere the first to use a high-speed tape target systemspecifically for laser plasma soft-x-ray projection li-thography experiments,19 a target system suitable fora commercial projection lithography installation is farfrom being developed.

D. High-Repetition-Rate LaserA fourth technological requirement for a high-repetition-rate x-ray source for projection lithogra-

6902 APPLIED OPTICS / Vol. 32, No. 34 / 1 December 1993

phy is, of course, the need for a suitable laser systemto produce the plasma. The specification of its ulti-mate output characteristics will have to await theresults of the experimentation referred to above;however, some comments can be made on the re-quired features of this laser system. The strictconstraints on long-term high-repetition-rate opera-tion with high reliability at a minimal operating costwill strongly limit the choice of laser candidates.Factors such as laser wavelength and pulse shape willprobably also play a role. At the present time, thetwo leading candidates would appear to be high-repetition-rate, diode-pumped, Nd:YAG lasers operat-ing at their fundamental wavelength (1064 nm) ortheir second harmonic (532 nm), or high-repetition-rate, short-wavelength (248-nm) KrF excimer lasers.Were one forced to specify a suitable laser system atthe present time on the assumption of the conversionefficiencies to soft x rays so far achieved at 13 nm, itwould be at or beyond the limits of these two technolo-gies. Clearly, improvements in target design, conver-sion efficiency, and satisfactory solutions to the debrisissue could, in principle, bring the required specifica-tions closer to or within the limits of present technol-ogy.

In this paper we begin to address the issues raisedabove. It is clear that no one issue can be solvedwithout considering its relationship to the others.It is therefore our intent, ultimately, to address allthese issues in a self-consistent manner. Withoutsuch a systematic approach, the choice of a laserplasma as a suitable source for soft-x-ray projectionlithography might be inadvertently bypassed in favorof some competing technology, such as synchrotrons,and the overall scheme of projection lithographychanged as a consequence. In our first studies ofthese problems, reported here, we concentrate ondefining some of the principal limitations of laserplasmas, as they are currently used, to the require-ments of a projection lithography source. In particu-lar, we report some novel characterizations of theparticulate emission from these laser plasmas andmake some logical extrapolations of these measure-ments to the conditions necessary for a lithographyinstallation. Although the study is preliminary innature, several important conclusions can be safelydrawn from this work. These are summarized inSection 6.

2. Experimental ConditionsThe emphasis of the present study was on making aninitial characterization of the particulate emissionfrom laser plasmas similar to those that have beenrecently used to demonstrate high soft-x-ray conver-sion for soft-x-ray projection lithography. Theseearly studies are restricted to measurements madewith solid-state lasers. The conditions closely re-semble the conditions used in current x-ray conver-sion measurements at Lawrence Livermore NationalLaboratory. The particulate emission has been mea-sured by several techniques. Clear demarcations

have been drawn between different types of particu-late emission. In this section we describe the condi-tions under which this particulate emission wascharacterized.

Two laser target installations have been estab-lished at the Laser Plasma Laboratory at the Centerfor Research in Electro-Optics and Lasers for thepurpose of investigating those properties of laserplasmas of importance to soft-x-ray projection lithog-raphy. The principal features of these installationsare shown schematically in Fig. 1. The first facility,which is shown in Fig. 1(a), utilizes a commercialQ-switched oscillator-amplifier Nd:YAG laser thatproduces pulses at up to repetition rate 10 Hz with

750 mJ of energy and 10 ns in duration.2 0 Thelaser is multimode with a beam divergence of 6 x 10-4

rad. The laser operates at a wavelength of 1064 nmor, with second-harmonic conversion, can provideenergies of as much as 400 mJ at 532 nm. Plasmasare produced from solid targets by the radiation ofthis laser focused with an f = 12-cm lens into a targetchamber with a base vacuum of 2 x 10-4 Torr. Thelaser intensity in the target plane of this laser wasaccurately characterized by infrared photography ofthe equivalent target plane intensity distributionproduced by a 50-cm focal-length lens with an opticalwedged image rattle plate, as shown in the layoutillustrated in Fig. 2(a). A typical isodensity contourplot of the radiation distribution in the equivalenttarget plane is shown in Fig. 2(b). The minimumFWHM spot diameter of the radiation distribution inthe chamber of Fig. 1(a) is 80 jim. All the experi-ments reported with this system were performed with

(a)

Target chamber

Single-mode injection-seededNd:YAG oscillator X-ray diagnostic

K.. Passive four-pass amplifier

(b)

Fig. 1. Two experimental laser target facilities for laser plasmastudies directed at soft-x-ray projection lithography. (a) Facilityequipped for the analysis of the particulate emission from plasmasproduced with a commercial Nd:YAG Q-switched laser system.(b) Target facility built around an oscillator-amplifier Nd lasersystem with well-controlled temporal and spatial beam characteris-tics.

1 December 1993 / Vol. 32, No. 34 / APPLIED OPTICS 6903

300

240

180

120

60

0

Fig. 2. Measthe laser tarttechnique usesity contour Econtours rang

a rotating.target regii

studies employed a novel laser configuration, which isshown in Fig. 1(b). This laser was designed specifi-cally to provide a uniform, well-characterized inten-sity distribution on target. This intensity distribu-tion should help ensure the production of an x-raysource with a uniform spatial distribution of x-rayemission. This has recently been shown to be an

COATED important requirement of a projection lithographyFILTER WE"OE CAMERA x-ray source. In most published optical schemes for

the mask illumination, a ring-field image of thisdistribution, created by a multilens configuration ofthe condenser optics, is raster scanned across thereflection mask. Uniformity of the x-ray distribu-tion of the x-ray target is therefore an importantrequirement for uniform illumination of the mask.

(a) The laser system used in this installation, which isshown schematically in Fig. 3, attempts to ensurethis. When a single-longitudinal-mode Gaussianbeam distribution in a smoothly varying form Gauss-

> SA C_ ~ ian laser pulse is generated, a well-characterized laserintensity distribution is produced in the target plane.This distribution is achieved through the develop-

| o): a/ S Rho x, ment of a single-mode injection-seeded Q-switchedoscillator with a diode-pumped single-mode Nd:YAGoscillator,22 producing a stable single-mode output.The 10-ns-duration, 10-mJ pulse from this laser is

' B y 2 y then amplified in an imaged relayed amplifier system.The first part of this amplifier system employs a novelpassive four-pass amplifier design. This type of am-plifier has been investigated by Andreev et al.

2 3 andalso formed the subject of a detailed study by Hunt.24

The present design is shown in Fig. 3. The oscillator0 60 120 180 240 300 output is fed into this amplifier system through a

polarizer and a Faraday rotator. The pulse thenPosition in pm makes four successive passes through a 460-mm-

(b) long, 16-mm-diameter, Nd:glass amplifier and a 3-m-surement of the laser beam focal spot distribution for long, one-to-one vacuum spatial filter. The overall,et facility shown in Fig. 1(a). (a) Schematic of the amplification is 400, with a final output energy ofd to measure the focal plane distribution. (b) Isoden- 2.7 J. The pulse duration and the uniform beamlot of the distribution at best focus. The isodensity quality are preserved through this amplification pro-ge betweenO0.5 and 3.0. cess. The output of this laser is focused with a

100-mm focal-length lens onto a target chamber witha vacuum base pressure of 2 x 10-5 Torr. The

-disk target system, permitting a fresh target assembly used in this facility is a rotating rodon to be presented in the focal region for assembly. This system permits the angular depen-

each shot. 'Ihe rotation o this target could beautomated for multishot operation. The target wasadjusted to this target position on each shot bymaximizing the hard-x-ray emission (that greaterthan 1 keV) emanating from the target. This emis-sion was monitored with a p-i-n x-ray diode in combi-nation with a 25-jim-thick Al filter. This targetfacility has several methods of assaying the plasmaemissions from the target. The velocity and therelative number of charged particles are measuredwith Faraday cup charge collectors. Neutral atomicparticulate emission and the charged particles arecaptured on acetate collectors, and large clusters ofmaterial, so-called "hot rocks," are analyzed with theuse of a high-repetition-rate (14-kHz) Cu-vapor laser.These diagnostics are discussed in detail below.

The second laser target installation used for these

----------- ---------------------------- ___

i P.W Xcl4 V4 607 T.adoo \ lu Ocla esap

Mod0Mhig

T.I.-cp.

! ^ ~~~~~~M.rrl------------....... } Osdlato

- I-lriny

| 4-Pass II.PllmpIlflerPolarizer i

i=_114 Spa-ial Pol-e Faradayr

L _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __Fl _ _ _ _ __r _ _ _ _ __la _ _ _ a__r

Fig. 3. Injection-seeded Nd:YAG oscillator with a passive four-pass Nd:glass amplifier. O.C., optical condenser.

6904 APPLIED OPTICS / Vol. 32, No. 34 / 1 December 1993

S

0

.0

arm

dences of plasma emissions to be easily analyzed.Attached to the target chamber is a large diagnosticchamber for x-ray instrumentation. In the presentstudies, the only diagnostic used is fast planar soft-x-ray photodiodes filtered with the use of a multilayermirror to analyze the x-ray emission at 13 nm fromthe plasma.

3. Characterization of Particulate EmissionThe particulate emission emanating from a laserplasma created off a solid target has several forms.These forms and their relative flux change as afunction of time during and after the interaction.Three specific forms of particulate matter can atpresent be identified. These are shown schemati-cally in Fig. 4. During the interaction of the laserlight with the plasma, highly stripped, energetic ionsare formed by collisional ionization resulting frominverse bremsstrahlung absorption of the laser lightby the plasma, the primary absorption mechanism inthis interaction regime. These ions stream ballisti-cally from the plasma region with velocities in therange 106-107 cm/s. Their total number can bequite large (>1016 ions), which accounts for themajority of the mass ablated during the laser plasmainteraction. This absorption mechanism and elec-tron thermal transport of the absorbed energy willablate several hundred nanometers of the targetduring this process. Immediately after the laserpulse, the plasma cools as further ablation of materialoccurs from the target. Thus a number of neutralatoms are ablated at this time. In addition, neutralatomic flux may also be generated on the periphery ofthe plasma during the laser plasma interaction and asa consequence of electron-ion recombination in theplasma. At still later times, after the interaction ofthe plasma with the target, hot clumps or clusters oftarget material are boiled off of the target. Thisoccurs within a crater that is formed in the targetsurface by the interaction. The three-dimensionalnature of this crater creates turbulent flow of themolten matter from the crater. The form of this

Q~~~~D 1

matter, as far as the size, velocity, and relativenumber of clusters are concerned, is probably depen-dent on the form of the crater made in the target andon the material characteristics of the target used.

In the present studies the clusters or hot rockswere studied optically. Early measurements haveshown that these clusters have sufficient tempera-tures such that they emit blackbody emission in thevisible part of the spectrum and can be photographedin flight. Their trajectory then appears as a streak oflight, generally of decreasing luminosity as the clus-ter cools. This approach, however, provides no infor-mation on size or velocity of the clusters. We haveused a photographic approach in which the clustersare illuminated with a rapid sequence of bright visiblelaser pulses. The light scattered from the clustersthen provides an indication of their size and, moreimportantly, an instantaneous measure of their veloc-ity during their trajectory. We are therefore able toestimate their momentum and impulse. The latterinformation is vital to estimating the potential theseparticles have for damaging the x-ray and visibleoptical systems situated in their path.

The setup used for these studies is shown in Fig. 5.Most measurements have been made so far on Sntargets. The plasmas are produced on a solid rotat-ing wheel target assembly irradiated at 450 by theNd:YAG laser shown in Fig. 1. The particles emanat-ing from the target are illuminated across a 25-mm-diameter field running parallel to and immediately infront of the target by the output of a 511-nm Cu-vapor laser producing 50-ns pulses with an individualenergy of 3 mJ at a frequency of 14 kHz. Theclusters passing through this field are then illumi-nated every 71.4 ms. They are photographed inflight by a film camera located in a direction orthogo-nal to the plane containing the Nd:YAG and Cu-vaporlaser beams. A typical photograph of the particlesemanating from a Sn laser plasma crater taken bythis method is shown in Fig. 6. The trajectories ofall the particles in the illumination field of theCu-vapor laser are visible as lines of bright dots, each

Atomic Flux

14 kHz

ClustersHot rocks>Y-_

Fig. 4. Plasma and particulate matter emanating from a laserplasma produced from a solid target.

0.5J1.064 ±mlons1011 W/cm

2

Fig. 5. Experimental arrangement used to estimate the velocityand the size of clusters emanating from laser plasmas.

1 December 1993 / Vol. 32, No. 34 / APPLIED OPTICS 6905

-

\\ :YAG Laser Beam

Cu-VaporLaser

Target Structure

Fig. 6. Photograph of high-repetition-rate (14-kHz) pulses of a511-nm-wavelength, Cu-vapor laser light scattered from clustersemanating from a Sn laser plasma.

dot resulting from the particle being illuminated bythe Cu-vapor laser. Measurements of the size andthe separation of the dots in each trajectory thenprovide a measure of the mass and the velocity of eachparticle. From these data the angular distributionand total mass and the energy and impulse resident inthis form of particulate emission can be estimated.The typical angular distribution of the velocitiesclusters emanating from Sn target craters is shown inFig. 7. As can be seen, the velocities of these par-ticles range over nearly an order of magnitude, from200 to 1000 cm/s. Particles with velocities as highas 2500 cm/s have been observed with this technique.The sizes of these particles range from less than 10 togreater than 200 mm. The latter size is comparablewith the spot size of the irradiating laser and thesoft-x-ray emitting zone. Thus these particles haveenergies of 1 x 10-3 to 4 x 10-2 g cM2/s 2 andspecific impulses per unit area in the range 3-100mJ/cm s. This is sufficient to cause cold particlecratering of most materials of the type used toconstruct visible and x-ray optical components and iscapable of puncturing any thin-film x-ray filters usedin the 13-nm region.

Although a full parametric study has yet to be madewith this diagnostic, one significant feature is appar-

s

Fig. 7. Angfrom a Sn lascrater had be(

a)

3)Z

02)z0

60

Number of shotsFig. 8. Increase in the cluster emission per shot as a function ofthe number of shots irradiating a single point on the target.

ent for the plasmas created in this study. This isillustrated in Fig. 8, which shows the number ofclusters created across a given field from an Sn targetas a function of the number of shots incident upon thesame position of the target. In this case we see thatfor the laser irradiation and the target conditionsused in this study, no particles detectable with thistechnique were observed for clean target conditions.As discussed below, this observation has a significantimpact on possible high-repetition-rate soft-x-raysource schemes for projection lithography. Anotherobservation from Fig. 8 is the fact that the number ofclusters emanating from the target increases with thenumber of shots on the same target position. Thisimplies that the cluster emission increases with thedepth of the crater formed in the solid target byprogressive laser irradiation. With successive shotson the same target position, the walls of the craterwill progressively steepen. Laser light absorbed onthe wall of this crater will increasingly boil off morematerial. This effect also has serious implicationsfor the use of solid targets in an x-ray source forprojection lithography and is discussed in Section 5.

4. Measurement of Atomic Flux EmissionAtomic and ionic flux measurements from Sn targetswith the laser target facility illustrated in Fig. 1(a)were made primarily with the technique that cap-tured the angular distribution of this flux on acetateor glass substrates. This is illustrated in Fig. 9(a).The 3 cm x 5 cm acetate collector was positioned inan arc around the target, as shown in Fig. 9(a), insuch a manner as to collect all the particulate emis-sion from the target. The acetate was positioned

2.5 cm from the target. Typically the debris frommany shots were accumulated to acquire sufficient

0 -20 0 20 40 60 80 100 to deposition of target mass on the collector. To avoidthe generation of clusters, a virgin target surface was

ANGLE FROM LASER BEAM AXMS (DEG) used for each shot. The latter was facilitated byilar distribution of the velocity of cluster emission the incorporation of an automatic rotational drive toer plasma target created from a solid target after a the target wheel assembly shown in Fig. 1(a).n produced by -40 shots. Quantitative data were extracted from the recorded

6906 APPLIED OPTICS / Vol. 32, No. 34 / 1 December 1993

virginsolidplanartarget

(a)

360 shots1.6 10 4 Torr

(b)Fig. 9. Experimental approach to measuring atomic flux fromlaser plasma targets: (a) use of collector plates, (b) typical massdeposit on curved collector plate.

deposition by densitometering of the distribution.Then, from known optical opacity-thickness character-istics of the target material, the thickness or massdistribution of the deposited material was determinedon the assumption that the target material wassputtered uniformly. The angular distribution ofthe atomic flux emanating from the target can beobtained from these data. Measurements were madeon Sn, Fe, and Au targets. For Sn targets, the totalmass deposited on the collector per shot was 1 mg.This implies that, for an irradiation spot diameter of

200 pm, the thickness of material ablated away asatomic material (in the absence of clusters and with-out the formation of a crater) was 3 [um. Theplasma ablation depth during the laser irradiation isonly 0.3 ,m; this implies that 10 times moremass is ablated from the target after the laser pulse asneutral or slightly ionized matter. An example ofdata obtained on an acetate collector is shown in Fig.9(b). It can be seen that, despite the fact that aplanar solid target is used, the atomic flux has astrong angular distribution. For Sn targets, thepeak atomic flux is 2 x 10-9 g/sr/pulse. Theangular distributions of the mass deposition per shotabout an axis normal to the target surface for differ-ent target materials is shown in Fig. 10. The massdeposition for Fe is significantly lower than that forAu or Sn by a factor of 3. All these data wereobtained with a background pressure in the targetchamber of 1.5 x 10-4 Torr. Under this condition,the particles are ballistic. They have a negligiblechance of undergoing collisions with background gasmolecules (mostly N2). The implications of this fluxfor possible x-ray sources for projection lithographyare discussed in Section 5. One approach commonlysuggested for reducing the impact of this flux onoptical components is the use of a background gas ofHe with a pressure high enough to impede the atomicflux particles collisionally, yet of a pressure lowenough to ensure minimal opacity to the soft-x-rayradiation. These conditions are met for the presentirradiation conditions with a He background pressureof - 0.2 Torr. Atomic flux collector measurementswere made from Sn targets under this pressure of He.The results are summarized in the data shown in Fig.10(d), in which the angular distribution of the atomicflux in this case is shown. As can be seen, there is

now no noticeable angular dependence of the cap-tured particles. In addition, under the conditions ofstatic background gas flow used in these experiments,the deposition thickness is reduced by a factor of 5.The use of a background gas as a debris mitigatingtechnique is discussed further in Section 5. Underthese background gas conditions, we would expect nochange in the character of the 13-nm soft-x-rayemission from the plasma. Figure 11 shows theapproach that is used to measure the emission at 13nm from a laser plasma. Two multilayer coatedmirrors reflect radiation from the target at 45°. Themultilayer mirrors are Mo-Si coated with a 9.06-nmlayer separation, providing a high-contrast bandpassfilter for 13-nm radiation. The radiation is detectedwith a fast, calibrated planar Al cathode photodiode,in front of which is a thin metal film filter to blockvisible and extreme ultraviolet from adding to thediode signal. It is estimated that the bandpass has awidth of - 3% at the optimum mirror wavelength.

5. Implications for High-Repetition-Rate X-Ray Sourcesfor Projection LithographyThis study in no way attempts to present a comprehen-sive treatment of the optimum requirements of alaser plasma soft-x-ray source for projection lithogra-phy. Its scope is limited. It presents some data onthe particle and plasma emission from laser plasmascurrently considered to be operating in the regimethat provides high enough x-ray conversion efficiencyto be feasible for soft-x-ray lithography. In thisstudy we have made an initial examination of thevarious types of particulate emission from the plasma,drawing a distinction between ionized particles, neu-tral atomic flux, and so-called clusters or hot rocks orlarge globular clumps of target material ejected fromthe target zone. Partly as a consequence of thesensitivities of the diagnostics that we use in thisstudy, we analyze separately the large clusters, downto 10 pum in size, and the atomic flux, which isassumed to be much smaller in particle size, evenmonatomic. So far we have no diagnostic that iso-lates particles that are smaller than 10 pLm but largerthan single atoms separately from the total atomicflux. This covers more than 3 orders of magnitudein particle impulse energy. However, to the best ofour knowledge this study does quantify two particu-lar parameters fairly accurately for the first time:an assessment of the mass ablated as atomic flux anda measure of the cluster emission from the plasma.From this study several conclusions that have pro-found implications for the use of laser plasmas insoft-x-ray projection lithography can be drawn.

The observation of cluster emission from laserplasmas produced from target materials that areefficient converters of laser radiation to 13-mm x rayshas significant implications for the use of a high-repetition-rate x-ray source in an operational lithogra-phy system. As noted above, the specific impulse ofthese particles is high enough to puncture availablethin-film x-ray filters and to damage irrevocably thex-ray and the optical components that would be in

1 December 1993 / Vol. 32, No. 34 / APPLIED OPTICS 6907

-90 0

Angle(a)

(degrees)

without background gas

0M(0

L_a)0L

a)

00a-a)

~0(0

0(0C

-90 0

Angle (degrees)(c)

Fig. 10. Angular distribution of atomic flux deposition per shot

their path in an operating lithography system. Inprinciple, it may be possible to impede their pathbetween the plasma and these delicate and costlyoptical components. Several possible techniques havebeen suggested. These include (1) the use of spin-ning apertures, (2) the use of pulsed gas or particlejets to deflect these particles to benign collectors, (3)the use of magnetic deflection after these particleshave been charged with a high-current electron beam,and (4) the use of laser light ablation of the particlesto deflect them away from sensitive optics. How-ever, none of these approaches is technologicallyelegant. For example, the use of a spinning aperturewould imply some undesirable technical requirements.

0.07

0.06

0.05

0.04

0.03'

0.02'

0.01

Angle (degrees)(b)

with background gas

-I 0.00 _90 -90 0 90

Angle (degrees)(d)

for (a) Fe (b) Au (c) Sn, (d) Sn under a background of 0.2 Torr of He.

Consider the use of a rotating aperture designed toinhibit the particles measured in this study. Theseparticles have velocities in the range 200-2500 cm/s.A simple configuration would be one as shown in Fig.12. Assuming that the aperture has a diameter of 1cm and is spaced 1 cm from the target, to inhibit theseparticles emanating from a 200-Hz x-ray sourcewould require the spinning disk to have a diameter of> 10 cm and to rotate with a speed of > 10,000 rpm.Although this capability is within current technology,the desirability of situating such a continuouslyoperating system in close proximity to expensive,critical optical components without maintenance forperiods of months may be questioned.

6908 APPLIED OPTICS / Vol. 32, No. 34 / 1 December 1993

Fe target Au target

0

a)0~

la(0

0.

a)-o(0E0

(00)C

90

0

0* 0 * 0 0

S 0 ~~ 0 0

S 0 0* a 0

130 A emission (zuns/div)

Fig.11. Top: Experimental setup used to measure 13-nm radia-tion from laser plasma. Bottom: Examples of the laser pulseshape (left) and the x-ray emission pulse (right) are shown.

An alternative to the above approach is to takeadvantage of the observation made in this study thatfor virgin target irradiation, no cluster formation wasobserved. Before discussing possible consequencesof this observation, several caveats should be made.First, although in this study we consistently observedno cluster formation for virgin target irradiationfrom a variety of targets, this may be dependent onthe irradiation conditions. It has been noted, forinstance, that in the use of longer pulses at shorterwavelengths (248 nin), there is some evidence ofcluster formation.l9 Moreover, since the techniquewe are currently using has a spatial resolution limit of- 10 [lm, it is possible that particles of smaller size

are not being detected in our experiments.The economic costs of avoiding the generation of

clusters in the x-ray source may not be very attractive.Consider the following implications for a 400-Hz laserplasma x-ray source operating with some type of solidtarget system, such as a rotating drum or a tape drive,which presents a new target surface to the laser beamfor each shot. If these target points are separated by

1 mm the target surface must move at a speed of 40in/s. The minimum new target surface area re-quired for each operating period of a lithographicfacility, which is assumed to be 8 h, will be 120 M2

.

The logistics of the material feedthrough would notbe insignificant. Moreover the cost may be prohibi-tive. Accepting the estimated cost per shot require-ment from the system analysis for an operation laserplasma x-ray source-based lithographic installationmade by Ceglio and Hawryluk,' 8 that is, a cost of

Required operating conditionsfor 200-Hz source

x-ray collecting aperture size 1cm

optics offset 2.5cm

target separation 1cm

disk diameter 12cm

Speed 11,000 rpm !

Fig. 12. Rotating-aperture cluster-inhibiting system for a high-repetition-rate soft-x-ray laser plasma source.

$10-6 per shot, would place an upper limit of approxi-mately $1 for the cost of each 8-h target element.This implies a maximum target area cost of <1O/cm

2. If a tape target were used, for instance, thecost for the target tape must be <$1/km. Theserequirements may be difficult to achieve. They mayimply that the use of a solid, planar moving targetproviding a virgin target surface per shot is not afeasible option for a soft-x-ray lithography laserplasma source.

The consequences of considering the mass of targetmaterial ablated as atomic flux are also disturbing.From the measurements reported here, a logicalextrapolation to a 400-Hz source would imply exceed-ingly high mass flows and unacceptable mass deposi-tion on x-ray and optical surfaces over a reasonableoperating period. Ceglio and Hawryluk,18 for ex-ample, assume that the principal optical elements inthe vicinity of the source would be exchanged everythree months, that is, after some 3 x 109 shots. Forexample, from measurements shown in Fig. 10, thetarget mass deposition on a surface, such as an x-raycollector mirror, situated 50 cm from the target invacuum, would be 300 mg/cm 2, or 400 plm thick.One approach to reducing this deposition rate with-out significantly reducing the x-ray emission wouldbe to use targets with a thickness equal to the laserablation depth ( 300 nm under the present condi-tions). Here we do not address how such a targetcould meet the cost requirements stated above. Thereduction in the mass deposition rate would be afactor of 20. Moreover, from the measurementsreported here, the use of a He background gas reducesthe atomic flux sticking rate by a factor of 10. Theadoption of these two approaches would reduce thedeposition rate to - 2 pLm per 3-month interval, still afactor of 103 more than is tolerable. A flowing Hesystem might be more efficient in extracting thismetallic aerosol. Such a system will require sometype of filtering system for isolating the ablatedmaterial.

6. SummaryThe measurements reported here and the conclusionsdrawn from them raise important questions regard-ing the feasibility of using laser plasma sources, asthey are configured at present, on the solid target forsoft-x-ray projection lithography. They imply thatfresh directions should be pursued toward satisfyingthe requirements of a high-repetition-rate laserplasma source. Central to these is the reduction ofparticulate matter emission from the source. Thisimplies the employment of an interaction regime inwhich the energy imparted to kinetic motion ofparticulate matter, particularly large clusters of neu-tral atoms, must be avoided. In addition to thisrequirement is the observation that manipulation ofthe plasma conditions to maximize the conversion oflaser light into radiation useful to projection lithogra-phy will have an impact on the overall requirementsin two ways. First, it will reduce the level of nonuse-

1 December 1993 / Vol. 32, No. 34 / APPLIED OPTICS 6909

targetmultilayer s-ray

i reflectors\\ Mo/SI 90.6A

filter planar soft XRD

laser pulse (lOns/div)

rotatingtarget

rotatingaperture

ful x-ray radiation and lessen radiation damage onthe first x-ray condenser mirror. Second, an in-crease in the useful x-ray conversion efficiency willreduce the overall laser output requirements. Fi-nally, the high-repetition-rate and long continuouslifetime requirements of the source will place strictlevels of cost and performance on target and lasersystems.

The authors gratefully acknowledge the loan of aCu-vapor laser from Oxford Lasers, x-ray detectorsfrom J. Cobble of Los Alamos National Laboratory, atarget chamber from J. Seely and C. Brown of theSpace Sciences Division of the Naval Research Labo-ratory, plasma detectors from A. Ng of the Universityof British Columbia, the provision of multilayermirrors from G. J. Kortright of the Center for X-RayOptics at the Lawrence Berkeley Laboratory and D.Windt of AT&T Laboratories, and the assistance ofKathy Abbott of the Naval Research Laboratory fordigitizing some of our photographic data. Usefuldiscussions with N. Ceglio, R. Stulen, R. L. Kauf-mann, and G. Kubiak are acknowledged. This workwas supported by the State of Florida and LawrenceLivermore National Laboratory contract B192600.

References and Notes1. W. T. Silfvast and 0. B. Wood, "Tenth micron lithography

with a 10Hz 37.2nm sodium laser," Microelectron. Eng. 8,3-11 (1988).

2. A. M. Hawryluk and L. G. Seppala, "Soft x-ray projectionlithography using an x-ray reduction camera," J. Vac. Sci.Technol. 6, 2162-2166 (1988).

3. K. Hoh and H. Tanino, Bull. Electrotechnol. Lab. (Jpn.) 49, 47(1987).

4. J. B. Murphy, "Electron storage rings as x-ray lithographysources: an overview," in Electron-Beam, X-Ray, and Ion-Beam Technology: Submicrometer Lithographies IX, D. J.Resnick, ed., Proc. Soc. Photo-Opt. Instrum. Eng. 1263,116-123 (1990); T. Tmimasu, "Review of Japanese compactelectron storage rings and their applications (invited)," Rev.Sci. Instrum. 60, 1622-1627 (1989).

5. B. E. Newnam, "Extreme ultraviolet free-electron laser-basedprojection lithography system," Opt. Eng. 30, 1100-1109(1991).

6. H. Pepin, P. Alaterre, M. Chaker, R. Fabbro, B. Faral, I.Toubhans, D. J. Nagel, and M. Pekerar, "X-ray sources formicrolithography created by laser radiation at I = 0.26 mm,"J. Vac. Sci. Technol. B 5, 27-32 (1987).

7. G. D. Lougheed, M. M. Kekez, J. H. W. Lau, and R. P. Gupta,"Solenoid gas puff imploding liner x-ray source," J. Appl.Phys. 65, 978-992 (1989); I. Okada, Y. Saitoh, S. Itabashi, andH. Yoshihara, "A plasma x-ray source for x-ray lithography,"J. Vac. Sci. Technol. B 4,243-247 (1986); D. A. Hammer, D. H.Kalantar, K. C. Mittai, and N. Qi, "X-pinch soft x-ray sourcefor microlithography," Appl. Phys. Lett. 57,2083-2085 (1990).

8. M. Chaker, H. Pepin, V. Bareau, B. Lafontaine, L. Toubhans,R. Fabbro, and B. Faral, "Laser plasma x-ray sources formicrolithography," J. Appl. Phys. 63, 892-899 (1988).

9. G. D. Kubiak, D. A. Outka, C. M. Rohlfing, J. M. Zeigler, D. L.Windt, and W. K. Waskiewicz, "Extreme ultraviolet resist andmirror characterization: studies with a laser plasma source,"J. Vac. Sci. Technol. B 8, 1643-1647 (1990).

10. R. L. Kauffman and D. L. Phillion, "X-ray production effi-ciency at 130 A from laser-produced plasmas," in Soft-RayProjection Lithography, J. Bokor, ed., Vol. 12 of OSA Proceed-ings Series (Optical Society of America, Washington, D.C.,1991), pp. 68-71.

11. D. G. Stearns, R. S. Rosen, and S. P. Vernon, "High-performance multilayer mirrors for soft x-ray projection lithog-raphy," in Multilayer Optics for Advanced X-Ray Applica-tions, N. M. Ceglio, ed., Proc. Soc. Photo-Opt. Instrum. Eng.1547, 2-13 (1991).

12. A. Rosenbluth, "Reflecting properties of x-ray multilayerdevices," Ph.D. dissertation (University of Rochester, Roches-ter, New York, 1980).

13. J. F. Seely, J. 0. Ekberg, C. M. Brown, U. Feldman, W. E.Behring, J. Reader, and M. C. Richardson, "Laser-producedspectra and QED effects for Fe-, Co-, Cu-, and Zn-like ions ofAu, Pb, Bi and U," Phys. Rev. Lett. 57, 2924-2926 (1986);J. F. Seely, U. Feldman, C. M. Brown, M. C. Richardson, andW. E. Behring, "High-resolution XUV spectroscopy using theOMEGA laser," inX-Rays from Laser Plasmas, M. C. Richard-son, ed., Proc. Soc. Photo-Opt. Instrum. Eng. 831, 25-29(1987).

14. N. M. Ceglio and A. M. Hawryluk, "Soft-x-ray projectionlithography system design, " in Soft-X-Ray Projection Lithogra-phy, J. Bokor, ed., Vol. 12 of OSA Proceedings Series (OpticalSociety of America, Washington, D.C., 1991), pp. 5-10.

15. F. E. Irons, R. W. P. McWhirter, and N. J. Peacock, "The ionand velocity structure in a laser-produced plasma," J. Phys. B5, 1975-1987 (1972).

16. M. C. Peckarar, J. R. Greig, D. J. Nagel, R. E. Pechacek, andR. R. Whitlock, "High-speed x-ray lithography with radiationfrom laser-produced plasmas," in Proceedings of the Sympo-sium on Electron and Ion Beam Science and Technology(Electrochemical Society, Princeton, N.J., 1978), Vol. 78-5, pp.432-436.

17. The laser intensity used for laser ablation material coating isusually well below intensities of 10ll W/cm2.

18. N. M. Ceglio and A. M. Hawryluk, "Soft x-ray projectionlithography system design and cost analysis," in MultilayerOptics for Advanced X-Ray Applications, N. M. Ceglio, ed.,Proc. Soc. Photo-Opt. Instrum. Eng. 1547, 82-101 (1991).

19. G. Kubiak, Sandia National Laboratory, Livermore, California94551 (personal communication, 1992).

20. QuantaRay, Inc., Model DCR1-A.21. Oxford Lasers, Ltd., Model CU-10.22. T. J. Kane and R. L. Byer, "Monolithic unidirectional single-

mode Nd:YAG ring laser," Opt. Lett. 10, 65-67 (1985).23. N. F. Andreev, G. A. Pasmanik, P. P. Pashinin, S. I. Shk-

lowsky, and V. P. Yanovsky, "Multipass amplifier with fullutilization of the active element aperature," Sov. J. QuantumElectron. 13, 641-643 (1983).

24. J. T. Hunt, "High peak power Nd:glass facilities for endusers," in Solid State Lasers II, G. Dube, ed., Proc. Soc.Photo-Opt. Instrum. Eng. 1410, 2-9 (1991).

6910 APPLIED OPTICS / Vol. 32, No. 34 / 1 December 1993