22-nm Half-pitch extreme ultraviolet node development at the SEMATECH Berkeley microfield exposure tool

Please download to get full document.

View again

of 8
13 views
PDF
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.
Document Description
22-nm Half-pitch extreme ultraviolet node development at the SEMATECH Berkeley microfield exposure tool
Document Share
Document Tags
Document Transcript
  22-nm Half-pitch extreme ultraviolet node development at the SEMATECHBerkeley microfield exposure tool Patrick P. Naulleau a, * , Christopher N. Anderson a , Jerrin Chiu a , Paul Denham a , Simi George a ,Kenneth A. Goldberg a , Michael Goldstein b , Brian Hoef  a , Russ Hudyma e , Gideon Jones a , Chawon Koh b ,Bruno La Fontaine c , Andy Ma d , Warren Montgomery b , Dimitra Niakoula a , Joo-on Park f  , Tom Wallow c ,Stefan Wurm b a Center for X-ray Optics, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA b SEMATECH, Albany, NY 12203, USA c  Advanced Micro Devices, Sunnyvale, CA 94088, USA d Intel Corporation, Santa Clara, CA 95052, USA e Hyperion Development LLC, San Ramon, CA 94582, USA f  Samsung Electronics, Hwasung-city, Gyeonggo-Do 445-701, Korea a r t i c l e i n f o  Article history: Received 10 November 2008Accepted 2 March 2009Available online 11 March 2009 Keywords: Extreme ultravioletLithographyPhotoresistLine-edge roughness a b s t r a c t Microfield exposure tools continue to play a dominant role in the development of extreme ultraviolet(EUV) resists. Here we present an update on the SEMATECH Berkeley 0.3-NA microfield exposure tooland summarize the latest test results from high-resolution line-space printing. Printing down to20-nm is presented with large process latitude at 22-nm half-pitch lines. Also presented are line-edgeroughness results along with a discussion of the importance of mask contributors to line-edge roughnessmeasured in resist. Finally we briefly describe an upgrade to the tool that will enable EUV resist devel-opment at the 16-nm half-pitch node and beyond. (This paper was presented in MNE 2008 conference,<http://www.mne08.org>, <http://www.mne-conf.org>). Published by Elsevier B.V. 1. Introduction As extreme ultraviolet (EUV) lithography approaches commer-cialization, resist issues remain animportant challenge. In particu-lar, simultaneously meeting resolution, line-edge roughness (LER),and sensitivity specifications is the primary challenge facing resistdevelopers. Excellent progress has been made in the areas of reso-lution and sensitivity, however, LER still needs much improve-ment. Moreover, the LER performance remains far from target nomatter the dose, and in practice non-resist contributors can beimportant.Microfield exposure tools (METs) [1–3] play a particularlyimportant role in the area of resist development. This is becausethe relative simplicity of such tools, in general, enables them toprovide higher resolution capabilities than production-scale alphatools. For example, preproduction/alpha tools just nowbeing exer-cised [4,5] have numerical apertures (NA) of 0.25 as compared tothe 0.3-NA available from the latest EUV METs. Moreover, theSEMATECH Berkeley MET [1,6,7] reported on here has the addi-tional benefit of a lossless programmable pupil fill illuminator [8]allowingaggressive k 1  factors to be achievedwithmodifiedillumi-nation settings such as dipole and quadrupole.The SEMATECH Berkeley exposure tool utilizes a 5  -reduction,0.3-NA optic [9] fabricated by Zeiss and designed, aligned, andcoated by Lawrence Livermore National Laboratory. It has a well-corrected field of view of 1  3mm at the reticle plane and200  600 l m at the wafer plane. The model shown in Fig. 1 de-picts the major components of the exposure station as well asthe EUV beam path (the system is described in detail in Ref. [1].With a NA of 0.3, the MET optic has a resolution of 22.5nm at a k 1  factor of 0.5 and a resolution of 16-nm at a  k 1  factor of 0.35.Examples of the capabilities of the programmable illuminatorcan be seen in Fig. 2 where we depict a variety of pupil fills andthe resulting modeled aerial image contrast transfer functions forequal lines and spaces. With conventional annular illuminationas used in the commercial implementations of the MET [2,3] wesee the roll half-pitch to be at 25nm. With the 45   dipole settingswe are able to push the role off to 20nm half-pitch and evenslightly smaller with relatively uniform performance throughpitch.Toachieveevenhigherresolutionswecanuse  x or  y  orienteddipoles (for vertical or horizontal lines) allowing resolutions assmall as 12-nm to be attained, however at the cost of strong vari-ations through pitch. 0167-9317/$ - see front matter Published by Elsevier B.V.doi:10.1016/j.mee.2009.03.013 *  Corresponding author. E-mail address:  PNaulleau@lbl.gov (P.P. Naulleau).Microelectronic Engineering 86 (2009) 448–455 Contents lists available at ScienceDirect Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee  Our ability to accurately control the pupil fill is shown in Fig. 3where we compare printing performance on radial gratings to pre-dicted aerial image performance for a 45   dipole setting. At thelargest feature size shown (34nm half-pitch) the dipole showsprinting failure in the orientation of the dipole owing to the dif-fractedorderinthatorientationfallingintothecentralobscurationof the projection optics (see diagram in Fig. 4a). As we push to thesmallest half pitch shown (28nm) four-fold printing degradationappears due the diffracted order simultaneously starting to falloutside the pupil in the direction orthogonal to the obscurationdirection (Fig. 4b). 2. EUV resist sensitivity  SensitivityofEUVresistsisofparticularconcernowingtoitsdi-rect impact on source power requirements. Due to difficulties in-volved in wafer-plane absolute dose calibration, most currentEUV exposure tools have been calibrated against a resist standardwith the actual calibration of the standard resist dating back toEUV exposures at Sandia National Laboratories in the mid 1990s.The resist standard itself, however, is calibrated against an evenolder generation EUV exposure tool that is no longer operationaland cannot be verified in terms of dose accuracy. To address thepotential uncertainty in the calibration of the EUV reference re-sists,asynchrotron-basedprocedurehasbeenimplementedallow-ing the absolute sensitivity of EUV photoresists to be measured[10].The resist sensitivity measurement method relies on the use of theEUVCalibrationsandStandards BeamlineoperatedbytheCen-ter for X-ray Optics and located at the Advanced Light Source syn-chrotron facility at Lawrence Berkeley National Laboratory [11].Using this beamline, a beam with an extremely well characterizedspatial and spectral profile can be achieved at the wafer plane. Aself-calibrated [10] photodiode can then be used to measure thetotalabsolutephotonfluxatthewaferplane.Notethattheself-cal-ibration procedure has also been verified against a NIST-calibratedphotodiode demonstrating the effectiveness of the self-calibrationprocedure. After measurement of the total wafer-plane dose, thewafer is inserted and subjected to a series of exposures spanninga region above and below the expected dose to clear. The waferis translated after each exposure. During this exposure procedure,a secondary  in situ  dose sensor is used to track the relative dose of the beam. With the exposure series complete, the wafer is pro-cessed and conventional methods are used to determine the expo-sure corresponding to dose to clear thereby determining theabsolute sensitivity of the photoresist.Using this procedure, it has been found that the long-standingresist calibration standard was incorrect by a factor of 1.9 with Fig. 1.  CAD model of the Berkeley MET exposure tool. Fig. 2.  Modeled aerial image contrast transfer function performance for equal lines and spaces with various illumination conditions. Images of the corresponding pupil fillsare shownabovethe plot. The outer circlein eachpupil fill imagerepresents the edge of the pupil ( r =0.1) and the inner circlerepresents the central obscurationof the METprojection optic which is located at  r =0.3. P.P. Naulleau et al./Microelectronic Engineering 86 (2009) 448–455  449  the actual sensitivity number being twice as fast as previouslythought. The implication is that all EUV resist sensitivity numbersreported over the past decade have been too lowby a factor of 1.9.With this result, the goal of an EUV resist with 10mJ/cm 2 sensitiv-ity, and the ability to simultaneously meet resolution and LER specifications is now much closer to reality. Very good performingresist with sensitivities in the 10–20mJ/cm 2 sensitivity range arenow fairly commonplace. 3. EUV resist resolution improvement Significant improvement in resist resolution has been achievedover the past few years. Numerous chemically amplified resistswithresolutionsof 25-nmor betterarenowavailable. Fig. 5showsa sampling of three such resists all with sensitivities of better than15mJ/cm 2 . InthecaseofresistC, goodfidelityatahalfpitchof 22-nm is observed. Fig. 6 shows printing results from the highest res-olution chemically amplified EUV resist characterized to date. Ithas a sensitivity of 15.2mJ/cm 2 and shows the ability to print20-nm half pitch lines, however, it does suffer from pattern col-lapse. Nevertheless, as demonstrated by the corresponding cross-sections,thisresistdoesresolve20-nmlines.NotethattheprintingresultsinFigs.5and6wereachievedusing45  dipoleillumination.To support semiconductor manufacturing, it is not sufficient fora photoresist to simply resolve a feature of a given size, it must beable to do so with some latitude in the focus and dose delivered tothe wafer, thereby relaxing the requirements on the exposure tooland/or optimizing the yield. The focus latitude of the resist shownin Fig. 6 is demonstrated in Fig. 7 where we show printed images through focus in 50nm steps for feature sizes ranging from26nm down to 20nm. A workable focus latitude down to 22nmis evident. Fig. 8 depicts the exposure latitude showing printing Fig. 3.  Direct comparison of printing and predicted aerial image performance for radial gratings of varying pitches printed with 45   dipole illumination with  r =0.15 andoffset=0.57. Fig. 4.  Diagrams of pupil fill with first diffracted order in the pupil for (a) 34-nm half-pitch radial grating and (b) 28-nm half pitch radial grating.450  P.P. Naulleau et al./Microelectronic Engineering 86 (2009) 448–455  results for the same resist through dose in 5% steps for the samefeature sizes.Improvement of theprocess windowperformance overtimeforEUV resists is demonstrated in Fig. 9 where we show bossungcurvesfromthebestperformingchemicallyamplifiedresistsunderannular illumination over the past 3 years. Annular illuminationresolution has improved from 40nm in 2006 to 32nm in 2008with a 150-nm depth of focus and an exposure latitude of 8%. 4. Line-edge roughness LER remains a significant challenge for EUV resists. As we pushto smaller and smaller targets for LER commensurate with targetCD reductions, non-resist effects become increasingly importantin experimentally measured LER results. Perhaps the most impor-tant non-resist effect is LER arising from the mask [12]. The maskcan directly contribute in two ways: one is the obvious source of LER on the mask (depicted in Fig. 10) being reproduced at the wa-fer and the second is from phase roughness on the mask (depictedin Fig. 11) couplingto speckle at the wafer and ultimatelyLER. Thelater is highly dependent on both illumination conditions anddefocus.In this section, we review the importance of these mask effectson LER as measured from exposures on the SEMATECH BerkeleyMET. A complete discussion of this topic can be found in Ref.[12]. Fig. 12 shows a plot of the predicted aerial image LER for Fig. 5.  Examples of three chemically amplified resists with resolutions of 25nm or better. Fig. 6.  Imaging results from a chemically amplified resist with a sensitivity of 15.2mJ/cm 2 . P.P. Naulleau et al./Microelectronic Engineering 86 (2009) 448–455  451  the SEMATECH Berkeley MET given the properties of the currentEUV mask in the system. The results are broken out to separatelyview the effect of mask LER as well as mask phase roughness andtwodifferent illuminationconditionsareconsidered. For thelowercoherence annular illumination setting, at best focus mask LER isseen to be the dominant term, whereas phase roughness domi-nates as we add defocus. For the higher coherence dipole setting,phase roughness is the dominant term throughout the focal range.The magnitude of these effects indicates that corrections to exper-imental LER data are indeed required as we approach values of 2nm if we wish to isolate resist LER effects.Using the modeling data above to compensate for mask effectsin our LER data, we can generate an improved estimate of the re-sist-limited LER performance of EUV resists over time. The correc-tion is applied by assuming the mask contributors to add inquadrature with resist LER. Fig. 13 shows a scatter plot of the cor-rected LER as a function of sensitivity for a wide variety of resistscharacterized on the SEMATECH Berkeley MET over the past sev-eral years. Thecorrectionisapplied assumingannular illuminationand operation at best focus. Under these assumptions, the maskcontributor is 1.43nm. The best observed results are an LER of 1.5nm at a sensitivity of 17mJ/cm 2 . 5. Future plans As resist resolutions approach 20nm, optical limitations of the0.3-NA MET become of increasing concern. Although as shown in Fig. 7.  Printed images through focus for the resist in Fig. 6. Fig. 8.  Printed images through dose for the resist in Fig. 6.452  P.P. Naulleau et al./Microelectronic Engineering 86 (2009) 448–455
Similar documents
View more...
Search Related
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