Nanostructuring of Mo/Si multilayers by means of reactive ion etching using a three-level mask

Nanostructuring of Mo/Si multilayers by means of reactive ion etching using a three-level mask
of 6

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  Thin Solid Films 458  ( 2004 )  227–2320040-6090/04/$ - see front matter   2003 Elsevier B.V. All rights reserved.doi:10.1016/j.tsf.2003.09.070 Nanostructuring of Mo y Si multilayers by means of reactive ion etchingusing a three-level mask  L. Dreeskornfeld , G. Haindl , U. Kleineberg , U. Heinzmann , F. Shi , B. Volland , a a a a b  , 1 b I.W. Rangelow , E. Majkova , S. Luby *, Kostic , L. Matay , P. Hrkut , P. Hudek  , Hsin-Yi Lee b c c  ,  d d d d  , 2 e University of Bielefeld, Faculty of Physics, Universitatsstr. 25, Bielefeld D-33615, Germany a ¨ University of Kassel, Department for Technical Physics, Heinrich Plett Str. 40, Kassel D-34109, Germany b  Institute of Physics, Slovak Academy of Sciences, Dubravska cesta 9, Bratislava SK-84511, Slovakia c  Institute of Informatics, Slovak Academy of Sciences, Dubravska cesta 9, Bratislava SK-84237, Slovakia d  National Synchrotron Radiation Research Center, Hsin-Chu 300, Taiwan, ROC  e Received 20 March 2003; received in revised form 25 August 2003; accepted 11 September 2003 Abstract Recently, Mo y Si multilayer reflectors have been gaining industry interest as a promising choice for the next generation extremeultraviolet mask material for printing sub 70 nm feature size devices. A reactive ion etching system with optimized hardwareusing CHF  y Ar process regime shows the capability for highly anisotropic etching of sub ( 400 nm feature sizes in Mo y Si test 3 multilayers with ten periods and a bilayer thickness of 7.8 nm which were prepared by e-beam evaporation. A three-level-mask technique consisting of a top resist mask layer poly-methyl-meth-acrylate, a middle hard amorphous Si mask layer and a bottom-level polyimide layer is used to create the etch mask. The etch characteristics of the polyimide film is shown to be one of themajor factors determining the success of the described multilayer etching process. The developed etching technology demonstratessuperior process performance without facets, excellent uniformity and good profile control. No contamination, degeneration ordefect generation in the unetched multilayer structure could be detected. This non-conventional process results in minimumdeposition during the etching thus eliminating the need for a dry or wet cleaning. Sidewall angles in Mo y Si multilayers of 85 8 ,without undercut, bowing and ripples resulting in smooth sidewalls are achieved.   2003 Elsevier B.V. All rights reserved. Keywords:  Molybdenum; Silicon; Multilayers; Nanostructures 1. Introduction The extension of optical lithography into the extremeultraviolet range  ( 11–13 nm wavelength )  known asextreme ultraviolet lithography  ( EUVL )  is known to bea promising candidate for next generation lithographysupporting 70 nm critical dimensions and even below w 1 x .In order to realize EUVL a bunch of new technologiesranging from EUV photon sources, fabrication of extremely precise aspherical substrates, improved mul-tilayer thin film technology, reticle fabrication with *Corresponding author. Tel.:  q 421-2-5249-6131; fax:  q 421-2-5249-5689.  E-mail address:   ( S. Luby ) .Current affiliation: Fa. Infineon, Dresden. 1 Current affiliation: Fa. Leica, Jena. 2 pattern accuracy of a few tens of nanometers to newresist chemistry is currently worldwide under develop-ment. The reduction in wavelength places very stringentrequirements on optics and reticles and motivates theneed for extremely low levels of aberrations in the highefficiency reflective optics as well as very good unifor-mity and defect control in the reflective masks.Our interest is focused on the fabrication of lateralnanostructures into periodic multilayers acting as highlyreflecting Bragg-mirrors in the extreme ultraviolet andsoft X-ray regime at non-grazing incidence. The tech-nological applications of these patterned systems areespecially in the fields of binary or phase shiftingreflection masks for EUVL w 2 x as well as high-resolutionX-ray diffraction optics such as Bragg–Fresnel zoneplates  w 3 x  and multilayer reflection gratings  w 4 x . Thefuture potential of these elements depends severely on  228  L. Dreeskornfeld et al. / Thin Solid Films 458 (2004) 227–232 Fig. 1. Patterning scheme of the three-level mask process. the structural accuracy of the fabrication process,demanding the preparation of amorphous  ( or polycrys-talline )  periodic multilayers with individual film thick-ness of a few nanometers and atomically smoothinterfaces as well as the fabrication of lateral artificialnanostructures in these multilayers by nanolithographyand reactive ion etching  ( RIE ) .Physical vapor deposition  ( PVD )  like magnetronsputtering, ion beam sputtering and ultra-high vacuumelectron beam deposition are used to fabricate atomicallysmooth periodic multilayers with EUV peak reflectivitiesof up to 69.5% at 13 nm wavelength w 5 x .The dry etch pattern transfer into multicomponentmaterials underlies several unwanted secondary effectslike preferential sputtering or etching, resulting in under-cutting, bowing and rough sidewalls  ( ripples ) w 6 x . Thisis known to be a severe limitation for the fabrication of quantum dots of semiconductor heterostructures  w 7 x  ormagnetic nanodots  w 8 x  for ultradense storage devicesand multilayers. For EUV mask technology this draw-back and the lack of local repair mechanisms for defectsinside the multilayer stack resulted in the developmentof absorber mask patterns deposited onto multilayerreflectors instead of structuring the multilayer itself.Here some problems are resulting from multilayerdegeneration during absorber deposition, structuring andrepair.A main requirement for the nanofabrication of artifi-cial nanostructures into the multilayer systems with highstructural accuracy is a plasma resistant etch mask withalmost vertical sidewalls. While our previous experi-ments using conventional poly-methyl-meth-acrylate ( PMMA )  and also new multicomponent chemicallyamplified resists did not result in a satisfactory processmainly due to a low etch selectivity  ( typically 0.2 ) , nowwe have applied a layer stack of nonconventional mul-ticomponent mask materials consisting of PMMA, amor-phous-Si  ( a-Si )  and polyimide. Considering the differentetching reactivity of Mo and Si a sidewall passivationtechnique was applied to protect the sidewalls frometching. 2. Experimental details 2.1. Multilayer deposition Test substrates of planar Mo y Si multilayers  ( notoptimized regarding their EUV peak reflectivity )  wereprepared in a UMS 500 Balzers apparatus by electronbeam evaporation onto Si  ( 100 )  wafer with a thermallygrown oxide layer  ( thickness 1.3  m m ) . Ten bilayerswere deposited at a growth rate of 0.1 nm y s by alter-nating Si  ( nominal thickness 6.8 nm )  and Mo  ( nominalthickness 1 nm )  deposition. While Si is the first depos-ited layer, the stack is terminated by a Mo layer withoutany additional capping layer. The layer thickness wascontrolled by a quartz monitor. The vacuum prior thedeposition was 10 Pa and the substrate temperature y 7 during the deposition was below 60  8 C. Previouslypublished details of the deposition process are given in w 9 x . The bilayer thickness, periodicity and interfacequality of the multilayer stack were characterized byhard X-ray reflectivity measurements prior and aftereach process step in order to confirm that no degradationof the multilayer occurred. 2.2. Three-level mask technique A schematic representation of the process is displayedin Fig. 1. First, the 600 nm thick polyimide PI 2610 ( DuPont ) , the 40-nm thick a-Si and the 200-nm thick PMMA layers are deposited onto the multilayer-coatedsubstrate. Second, the PMMA resist is exposed byelectron beam lithography  ( as described below )  and theexposed areas are removed using standard wet post-exposure processes. These structures are then subse-  229  L. Dreeskornfeld et al. / Thin Solid Films 458 (2004) 227–232 quently transferred into the a-Si layer by RIE-etching inCF plasma and finally into the polyimide layer by an 4 oxygen plasma process. The preparation of the polyim-ide mask was performed in a multi-step process whichis described elsewhere w 10 x .Direct-write electron beam lithography was used toform a set of line-gratings with varying pitch ratio anda period of 800 nm in a 200-nm thick top layer of resistPMMA Elvacite 2041  ( DuPont ) . The machine used wasa modified ZBA10 y 1  ( Zeiss y Leica )  30 keV vector scanvariable shaped beam pattern generator, the minimumspot size and deflection being 25 nm and 50 nm,respectively. The exposure was made by rectangularshots of 5000 nm = 400 nm with a base dose of 600 m C y cm using a combined dose y geometrical proximity 2 correction.A 1 = 1 mm periodic array of lines 1-mm long and ( 400-nm wide was formed in one exposure cycle ( without moving the sample ) . This way the test struc-tures were prepared. The grating masks of the total areaof 5 = 5 mm corresponding to a grating period of 800nm  ( 1250 lines y mm )  were prepared by sequential repos-itioning of the sample.The e-beam pattern generator is equipped with a fineindividual shape-by-shape dynamically programmabledosage control  ( one increment 25 nC y cm  )  thus ena- 2 bling us to create resist masking patterns with a goodcritical dimension control over the whole grating areawith nearly vertical side-walls. This turns out to be animportant requirement for a subsequent successful sub-micron pattern transfer with vertical and controlled slopeprofiles into the substrate. 2.3. Reactive ion etching of Mo y Si multilayer  The etching profile formed in fluorine containingplasma depends due to four etch phenomena:  ( i ) increased etch rate at the bottom  ( vertical etch rate ) , ( ii )  decreased etch rate at the sidewalls  ( lateral etchrate ) ,  ( iii )  passivant deposition onto the sidewalls,  ( iv ) permanent removal of passivants from the bottom of thetrench w 11 x .The etch mechanism of fluorine plasmas with Mo y Simultilayer is mainly dominated by chemically activeneutrals which are provided by the electron impactdissociation of fluorine containing gases  ( SF , CHF , 6 3 etc. )  in the plasma. With the addition of physicalcomponent  ( even by ions with energy below 30 eV )  theetch rate of Mo y Si multilayer increases dramatically dueto the ion enhancement effect of the fluorine–siliconchemical reactions. In this matter, the etch rate of surfaces perpendicular to the ion beam  ( e.g. the bottomof a trench )  is increased over the etch rates of surfacesparallel to the ion beam  ( e.g. sidewalls ) . With gaseslike CHF or C F  ( passivant precursors )  added to the 3 2 6 SF plasma, a sidewall passivation layer can be depos- 6 ited which blocks the etching at the sidewalls, andsimilarly, the etch rate decreases with increasing concen-tration of the sidewall passivant. In order not to block the etching of the trench bottom, passivation layersdeposited on the bottom must be permanently scavengedby ion bombardment, allowing the chemically activefluorine radicals to form with Si and Mo volatileproducts.The pattern transfer into the Mo y Si multilayer wasperformed by fluorine RIE in a parallel plate plasmachamber  ( Oxford Plasmalab 80 )  equipped with an RFgenerator  ( 13.56 MHz ) . The temperature of the substrateelectrode is controlled in a y 70 to q 80  8 C temperaturerange by an external heater y chiller. The etch gases wereCF , SF , CHF , O and Ar. 4 6 3 2 The most important steps of the lithographic processand the final etching results were analyzed by means of a high-resolution, low electron energy scanning electronmicroscope  ( SEM Hitachi 4000 ) , atomic force micros-copy  ( AFM Digital Instruments Nanoscope III )  andsurface profilometry  ( DEKTAK 3030 ST )  to evaluatethe etch rates and the etching profile. High-resolutiontransmission electron microscopy images of samplecross-sections  ( XTEM Philips CM200 ST )  were takento analyze the accuracy of the pattern transfer afterremoval of the remaining resist by oxygen ashing. TheX-ray measurements were performed at wiggler beam-line BL-17 B at the National Synchrotron RadiationResearch Center  ( NSRRC ) , Hsinchu. 3. Experimental results and discussion It is well known that the etching reactivity of Si andMo with free fluorine atoms is quite different. We havefound that under the similar process conditions theetching rate of Mo can be lower than that of Si by afactor of 4  w 12 x . In the fluorine plasma SF and CF 6 4 produce F* radicals for the chemical etching of Siforming volatile SiF species via an etch reaction Si q 4 4F*. The reaction of atomic fluor radicals with siliconcan be monitored by optical spectroscopy and was usedfor implementing an endpoint detection w 12 x . The etch-ing chemistry of molybdenum is more complex. In apure CF plasma the main etch product is molybdenum 4 hexafluoride  ( MoF  )  which is volatile  w 13 x , but the 6 formation of other fluorides with lower F contentMoF  ( n - 6 )  cannot be excluded. In the presence of  n oxygen the reaction products MoF , MoO F or  x   2 2 MoOF molybdenum oxyfluorides can be formed  w 14 x 4 thus reducing the process selectivity to the polymermask. Two different fluorine RIE processes have beenexamined and compared regarding their Mo y Si etchingcharacteristics:  230  L. Dreeskornfeld et al. / Thin Solid Films 458 (2004) 227–232 Fig. 2. X-TEM image of one grating period etched into the Mo y SiML by SF  y CHF  y Ar. 6 3 Table 1Comparison of the RIE etch parameters for both processesSample Temperature Pressure DC-Bias SF 6  CHF 3  Ar Av. etching rate ( 8 C ) ( mTorr ) ( V ) ( sccm ) ( sccm ) ( sccm ) ( nm y min ) P1 15 30 120 0.5 2.0 6.8 5.5P3 16 20 150 0.0 1.5 6.6 0.7 3.1. Fluorine RIE process SF   y CHF   y  Ar  6 3 The first process is a combined SF  y CHF  y Ar RIE 6 3 process at a total pressure of 30 mTorr and a DC biasof 120 V  ( sample P1, Table 1 ) . We have measured anaverage etch rate of 5.5 nm y min and a selectivity of about one for this process. While SF generates free F* 6 radicals for chemical etching and the Ar ions provide q a physical sputter yield, CHF is known to form CF – 3 2 CHF compounds which may act as a main gas precursorfor polymer formation for sidewall passivation whichshould improve the etching anisotropy. The SEM inspec-tion after the resist removal by oxygen ashing showedsignificant line edge roughness as a result of the RIEprocess.Details of the etch profile regarding sidewall angle,structural underetching and sidewall roughening havebeen analyzed by XTEM as displayed in Fig. 2. Theimage shows a completely etched Mo y Si stack and anadditional 35 nm overetching into the SiO layer. The 2 different shape of the etch profile inside the SiO layer 2 can be explained by ion-induced charging which resultsin ion stream shadowing and sidewall reflection. Anegative sidewall slope of the etching profile of approx-imately 80 8  with an underetching of 10–15 nm can bedetected. The different etching rates for Mo and Siresults in prominent sidewall ripples, detected as lineetch roughness by SEM. Additionally, contaminationresidual can be detected at the sidewall edges. Althoughtheir chemical nature is not known, this can be causedby resist residuals not fully removed by the oxygenashing or contamination from non-volatile molybdenumreaction products created during RIE. 3.2. Fluorine RIE process CHF   y  Ar  3 To avoid the unwanted effect of underetchingdescribed above a similar process without SF addition 6 was performed. As expected, the lack of SF as fluorine 6 radical precursor results in significantly reduced etchrate  ( 0.7 nm y min )  while the selectivity remains one ( sample P3, Table 1 ) . Different features before  ( Fig. 3a ) and after  ( Fig. 3b )  resist removal have been analyzedby SEM. In contrast to process described in the sectionabove Fig. 3a displays the almost perfect conformity of the etch transfer from the patterned polyimide mask intothe multilayer. The line edge roughness of the gratingstructure, as displayed in Fig. 3b, was strongly improvedcompared to the SF  y CHF  y Ar process. 6 3 After resist removal the etch profile was again ana-lyzed by XTEM  ( Fig. 4 ) . The accuracy of the etchprofile has remarkably improved as mainly displayed bythe steep positive sidewall slope of approximately 85 8 .The sidewalls are nearly free from underetching and noripples srcinating from the different etch characteristicof Mo and Si are detectable. An excellent uniformity isalso noticeable. No multilayer degradation due to theetching procedure can be detected, however, the termi-nating Mo layer is almost fully removed during thewhole process. This can be explained by the formationof MoO which is water soluble and thus easily 3 removed. 3.3. Multilayer grating A 1250 lines y mm laminar multilayer grating has beenetched by the process described in Section 3.2. Thegrating structure has been deeply etched down to thesilicon substrate thus providing an amplitude grating.The grating diffraction efficiency has been characterizedby measuring rocking curves  ( fixed detector angle,rotating sample )  around the 1st and 2nd multilayerBragg order at SRRC.In Fig. 5 if the rocking curve measured around the1st multilayer Bragg order is shown. The position of thesatellites corresponds to the grating period of 800.5 nm.The second order satellites are smaller due to the line y  231  L. Dreeskornfeld et al. / Thin Solid Films 458 (2004) 227–232 Fig. 3. SEM images of different features transferred into the Mo y Si multilayered stack with a gas mixture CHF  y Ar:  ( a )  a corner structure still 3 covered with polyimide. The multilayered stack is displayed as a brighter stripe at the bottom of the etch profile.  ( b )  Grating structure after resistremoval  ( some resist residuals are still visible on the grating lines ) .Fig. 4. X-TEM image of one grating period etched into the Mo y SiML by CHF  y Ar. 3 Fig. 5. Rocking curve of a patterned multilayer laminar grating around1st Bragg order. groove ratio close to one. The rocking scans point atthe good quality of etching process. Smaller peaksbetween the satellites indicate the parasitic lateral super-periodicity several times larger than the regular gratingperiod. These features are caused by secondary effectsof the patterning procedure w 15 x . 4. Conclusion We have developed and analyzed a new reactive ionetching process for etching structures of several 100 nminto Mo y Si multilayer systems with an etch profileaccuracy of a few nanometers. The application of amultilevel resist in combination with an optimized flu-orine etch process providing high etching anisotropy
Related Documents
View more
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks