Microelectronic Engineering 110 (2013) 173–176

Contents lists available at SciVerse ScienceDirect

Microelectronic Engineering

journal homepage: www.elsevier .com/locate /mee

High resolution HSQ nanopillar arrays with low energy electron beam lithography

Y. Guerfi ⇑, F. Carcenac, G. LarrieuLAAS, CNRS, Univ de Toulouse, 7 avenue du colonel Roche, F-31400 Toulouse, France

a r t i c l e i n f o a b s t r a c t

Article history:Available online 22 March 2013

Keywords:Electron beam lithographyLow energyNanopillars1D nanostructure arrayHSQInorganic resistNanofabrication

0167-9317/$ - see front matter � 2013 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.mee.2013.03.055

⇑ Corresponding author at: LAAS, CNRS, 7 avenueToulouse, France. Tel.: +33 (0) 5 61 33 79 84; fax: +3

E-mail addresses: yguerfi@laas.fr (Y. Guerfi), glarri

Electron beam lithography (EBL) is commonly used for the fabrication of nanostructures by top-downapproach with precise control of size, shape, aspect ratio, and location. In this article, we demonstratethe realization of high aspect ratio nanopillars (7.5) with 20 nm diameter in 150 nm Hydrogen Sils-esQuioxane (HSQ) thickness based on 20 keV energy exposure. A detailed study of design strategieshas been conducted in order to correlate the design of the nanopillars and their practical realization.We describe an original way of scanning for HSQ nanopillar arrays fabrication with almost perfect aniso-tropic sidewalls (98.5%) and good circularity shape (1.62 nm-1r) without any residual resist remainingaround the nanostructures.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Following the down-scaling of MOSFETs to the ultimate scale,novel architectures were proposed as tridimensional multigatearchitecture [1]. In that context, gate all around nanowire transis-tors [2] present the optimum configuration for electrostatic controlof the gate over the channel [3]. The starting point of such a deviceis the formation of nanowire array with a perfect control of thedimension, location and parasitic contamination. For silicon nano-structure realization, conventional top-down approach is the mostused because of its suitability with CMOS fabrication tools andtechniques. The definition of the hard mask pattern is the criticalstep because it determines the form of the final structure. For ulti-mate scaling, electron beam lithography is one option in order toreach narrow dimension combined with high density.

The high resolution negative tone resist Hydrogen SilsesQuiox-ane (HSQ), is widely used for EBL processing, because of its capabil-ity to perform high resolution nanostructures [4]. The HSQ, beingan inorganic resist, has a good resistance to plasma etching and agood mechanical strength [5,6]. Finally, its good sensitivity to elec-trons permits to apply low doses during electron beam writing.

Outstanding results have been demonstrated using high energyelectron beam writers operating at 100 keV [7,8] or 50 keV [9],with dense arrays of 20 nm diameter nanopillars. However, ma-chines enable to work at such voltages are not widely used becauseof their high total cost of ownership (TCO). Moreover, a longerexposure time is expected when high voltage is used because of

ll rights reserved.

du colonel Roche, F-314003 (0) 5 61 33 62 08.eu@laas.fr (G. Larrieu).

the linear dependency of the dose with the energy. To circumventthese potential limitations, nano-patterning of resist at relativelylow energy is of prime interest but challenging due to its highersensibility to the proximity effects. Several studies at low voltageshave been conducted [10–13], but exclusively using thin resistthicknesses, which limit their application as etching hard-mask.The results generally show residual resist at the bottom of nano-structures [11] due to the proximity effects, limited aspect ratioor severe size dispersions [12] which should induce variability inthe nanostructure-based electronic device. A relevant work havebeen done at 5 keV with a multi electron beam (13000 beams)from Mapper Company (Netherland) at the CEA LETI (France) butstill in thin resist thickness [13].

In this work, a dedicated design strategy has been conducted inorder to demonstrate circular shaped nanopillars. Vertical nano-structures with high aspect ratio have been patterned using lowenergy electron beam lithography, with a good control of proxim-ity effects.

2. Experiments

The HSQ resist is identified with a general formula of (HSiO3/2)n,based on cage like structure which is transformed into networkstructure after e-beam exposure by the scission of Si–H bondsand the formation on Si–O–Si bonds (siloxane bonds), as explainedby Namatsu et al. [14]. In our experiments, HSQ from Dow Corning(FOx-15) have been diluted in Methyl Iso Butyl Keton (MIBK) in 1:1proportion and spun on Si (100) wafers for 60 s at 5000 rpm, in or-der to obtain 150 nm HSQ thickness. Then the sample has beensoftly baked at 80 �C for 1 min in order to evaporate the solvent

174 Y. Guerfi et al. / Microelectronic Engineering 110 (2013) 173–176

while avoiding the crosslinking of the resist before e-beamexposure.

The EBL was carried out with a RAITH 150 writer at 20 keV en-ergy exposure with a base dose of 300 lC/cm2 and a beam currentof 120 pA. Casino simulations [15] were performed to find voltageacceleration suitable to obtain height of nano-pillars sufficient toact as an hard mask in a plasma etching process. For that purpose,it was concluded that 20 keV is the suitable energy exposure tolimit electron forward scattering which degrade the side-wallanisotropy of the patterns. The development of the resist after

(b) (a)

(d)

(e) (c)

Fig. 1. (a) Illustration of RAITH circle mode design, (b and c) SEM top view of Raithcircle mode on HSQ of small and large nanopillars respectively, (d) SEM tilted viewof nanopillar array with (e) zoom on a single nanopillar.

Fig. 2. Circle design strategies: (a) Concentrically circle design, where the circles are desigstar like design, where the lines start and finish at the center, with a maximum distanceconcentrically circle design.

exposure was performed by manual immersion in high concen-trated (25%) TetraMethylAmmonium Hydroxide (TMAH) to in-crease the contrast [16]. Finally the sample was rinsed indeionized water then in methanol solution before a soft dry withnitrogen flux in order to reduce the surface tension and minimizethe nanopillar collapse.

3. Design strategy study

To create circular HSQ nanopillars, the first sets of experimentswere performed using the RAITH circle mode schematized inFig. 1a, that supposed to be in concentrically circle mode. Whenaddressing diameter less than 100 nm, the nanostructures showeda square shape (Fig. 1b).Moreover, a stitching defect was clearlynoticed when a larger diameter has been patterned (Fig. 1c). Wesuppose that this defect is issued by the connection of the concen-trically circles Raith based mode. From SEM titled view (Fig. 1d), itwas observed a bowed sidewalls and the stitching defect men-tioned above is transferred along the nanopillars (Fig. 1e). Sincethe results obtained by the conventional tools available in theRAITH software’s equipment give unexpected results, i.e. high dis-persion of nanostructures circularity, bowed sidewalls and thus, donot allow a good transfer of the nanostructures to the substrate, adedicated design strategy studies became a necessity to find out anefficient way for circle design. In this scope, we propose several de-signs: (i) ‘‘concentric circles’’ which consists in drawing circles oneinside another, using the resist resolution (5–10 nm) as the pitchbetween two circles (Fig. 2a), (ii) ‘‘star like’’ consists in drawinglines starting and finishing at the center of the pattern, with a max-imum distance between two adjacent lines is the HSQ resolution(Fig. 2b), and finally (iii) a combination of the two previous designs(Fig. 2c).

4. Results and discussion

Starting with the concentrically circles mode, we firstly decidedto tighten the circles in the peripheral region in order to emphasizethe effect of the exposure on sidewalls’ shape. We obtained circularpatterns (Fig. 3a) but the tilted inspection revealed nanostructureswith conic shapes (Fig. 3b). A similar result has been obtainedwhen tightening the circles on the center of the designed pattern(not shown). The conic shape obtained is probably due of thesuperposition of the energy distribution of each designed circlethat wraps all the Gaussians distributions into a larger one. More-over, the combination of concentrically circles and star like designsdoes not improve the situation where higher dispersion of nano-structure size was obtained coupled with a conic shape, suggestedby (Fig. 3c). Finally, the star like design leaded to the most interest-ing results where nanopillars with better circularity were observedby top SEM inspection with 1.62 nm 1r dimension dispersion(Fig. 3d). We achieved nanopillars with almost perfect vertical

ned one inside another, with the HSQ resolution as the pitch between two circles (b)between two adjacent lines is the HSQ resolution (c) combination of star like and

Fig. 3. SEM pictures: (a) top and (b) tilted views of concentrically circle design on HSQ with circular pattern but conic shaped nanopillars, (c) top view of combined designshowing a high dispersion in circularity, (d) top view of circular nanopillar obtained by start like design and (e) titled view of anisotropic HSQ nanopillar array with 20 nmdiameter and a pitch of 170 nm.

Y. Guerfi et al. / Microelectronic Engineering 110 (2013) 173–176 175

sidewalls anisotropy (98.5%), 20 nm diameter in 150 nm HSQthickness leading to an aspect ratio of 7.5. Moreover, a perfectlyclean surface without residual resist around the nanostructures isshown (Fig. 3e).

One additional advantage of the ‘‘star like’’ design is its lowersensitivity to a residual astigmatism of the beam compared to aconventional dot mode where an egg shaped nanopillars can be ob-

Fig. 4. SEM images of HSQ nanopillar arrays (spacing 150 nm) of 150 nm height andvarious diameters: (a) 22 nm, (b) 40 nm, (c) 78 nm and (d) 162 nm, (e) Demon-stration of low voltage capabilities to obtain high density nanostructure array(34 nm diameter, 26 nm spacing).

tained. The ‘‘star like’’ design forces the symmetry of the nanopil-lars with a homogenous distribution of the energy which resultsin a better reproducibility in the circle formation.

The ‘‘star like’’ design can be applied to several nanopillar diam-eters, as demonstrated in Fig. 4a–d, where well defined HSQ nano-pillar arrays with vertical sidewalls and nanostructures diameterranging from 20 to 160 nm. Finally, high density nanopillar arrayscan be achieved with, for example in the Fig. 4e, 34 nm diameterand 26 nm spacing. It is worth noting that all these demonstrationswere obtained at 20 keV energy exposure.

5. Conclusion

In this article, we demonstrated the capability of low energyelectron beam lithography to produce high resolution nanostruc-tures on HSQ resist. We demonstrated a unique design strategyto obtain good pattern circularity, by solving the irregularities in-duced by the traditional tools writer’s software. We achieved highaspect ratio HSQ nanopillar arrays with perfectly anisotropic side-walls and good circularity shape, without any residual resist at thebottom of the nanostructure. We showed also the ability of this de-sign to reproduce high density nanostructures with a precise con-trol of size and location, ready to be transferred into the substrateby anisotropic etching for the realization of ultimate architectureof vertical nanowire transistor [17].

Acknowledgements

This work was partly supported by the French RENATECH net-work (French national nanofabrication platform).

References

[1] B. Doyle, S. Datta, M. Doczy, S. Hareland, B. Jin, J. Kavalieros, T. Linton, A.Murthy, R. Rios, R. Chau, IEEE Electron Device Lett. 24 (2003) 263–265.

[2] J. Goldberger, A.I. Hochbaum, R. Fan, Nano Lett. 6 (2006) 973–977.[3] R. Chau, B. Doyle, S. Datta, J. Kavalieros, K. Zhang, Nat. Mater. 6 (2007) 810–

812.[4] L. Pain, C. Higgins, B. Scarfogliere, S. Tedesco, B. Dal’Zotto, C. Gourgon, M.

Ribeiro, T. Kusumoto, M. Suetsugu, R. Hanawa, J. Vac. Sci. Technol. B 18 (6)(2000) 3388–3396.

176 Y. Guerfi et al. / Microelectronic Engineering 110 (2013) 173–176

[5] K.A. Lister, B.G. Casey, P.S. Dobson, S. Thoms, D.S. Macintyre, C.D.W. Wilkinson,J.M.R. Weaver, Microelectron. Eng. 73–74 (2004) 319–322.

[6] H. Namatsu, T. Yamaguchi, M. Nagase, Y. Yamazaki, K. Kurihara, Microelectron.Eng. 41–42 (1998) 331–334.

[7] X.L. Han, Microelectron. Eng. 88 (2011) 2622–2624.[8] X.L. Han, G. Larrieu, E. Dubois, J. Nano, Nanotechnology 10 (2010) 7423–7427.[9] Z. Min, C. Baoqin, X. Changqing, L. Ming, N. Jiebing, Microelectron. Eng. 83

(2006) 788–791.[10] H. Yang, A. Jin, Q. Luo, C. Gu, Z. Cui, Y. Chen, Microelectron. Eng. 83 (2006) 788–

791.[11] A. Jamieson, C.G. Wilson, Y. Hsu, A.D. Brodie, J. Micro/Nanolithigr. MEMS

MOEMS 3 (2004) 442–449.

[12] V.R. Manfrinato, L.L. Cheong, H. Duan, D. Winston, H.I. Smith, K.K. Berggren,Microelectron. Eng. 88 (2011) 3070–3074.

[13] D. Rio, C. Constancias, M. Martin, B. Icard, J. van Nieuwstadt, J. Vijverberg, L.Pain, J. Vac. Sci. Technol. B 28 (2010). C6C14–C6C19.

[14] H. Namatsu, Y. Takahashi, K. Yamazaki, T. Yamaguchi, M. Nagase, K. Kurihara, J.Vac. Sci. Technol. B 16 (1) (1998) 69–76.

[15] Monte carlo simulation, Casino Software V2, Sherbrook University, 2000.[16] F. Fruleux-Cornu, J. Penaud, E. Dubois, M. François, M. Muller, Mater. Sci. Eng.,

C 26 (2006) 893–897.[17] G. Larrieu, X.-L. Han, Nanoscale 5 (6) (2013) 2437–2441.