Gas assisted focused electron beam induced etching of alumina

Gas assisted focused electron beam induced etching of alumina T. Breta兲 NaWoTec GmbH, A Carl Zeiss SMT Company, Industriestrasse 1, 64380 Rossdorf, Germany

B. Afra Institut de Microtechnique, Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland

R. Becker, Th. Hofmann, and K. Edinger NaWoTec GmbH, A Carl Zeiss SMT Company, Industriestrasse 1, 64380 Rossdorf, Germany

T. Liang Intel Corporation, 2200 Mission College Blvd., Santa Clara, California 95052

P. Hoffmann Institut de Microtechnique, Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland

共Received 8 July 2009; accepted 8 September 2009; published 2 December 2009兲 This study investigates focused electron beam induced etching for the removal of alumina particles on patterned extreme ultra violet 共EUV兲 mask using nitrosyl chloride 共NOCl兲 as assist gas. As potential contaminant, particles of aluminum oxide 共alumina, Al2O3兲 have been successfully removed, leaving the underlying layers undamaged. Particles were applied onto an EUV mask, consisting of a multilayer Bragg mirror capped with a thin ruthenium layer and a structured tantalum nitride 共TaN / TaON兲 absorber/antireflective film. Alumina particles were selectively etched using the chlorine-based gas, NOCl. Neither the Ru nor the absorber was significantly etched during the process in spite of a square area scanned by the focused electron beam being larger than the particle. The process resolution is discussed based on Monte Carlo electron scattering simulations. Thermodynamic driving forces for the electron-induced reactions and its selectivity are discussed and a chemical rationale is proposed. © 2009 American Vacuum Society. 关DOI: 10.1116/1.3243208兴

I. INTRODUCTION Nitrosyl chloride 共chemical formula, NOCl兲 is the active compound in “aqua regia.” This stable compound, gaseous at room temperature, was isolated in a pure form in the 19th century.1 This article discusses the usage of NOCl in selective gas assisted focused electron beam induced etching 共FEBIE兲 of alumina particles on an extreme ultraviolet 共EUV兲 mask, in an improved lithographic mask repair tool test vehicle. Focused electron beam induced processing 共FEBIP兲 is a nanofabrication technique.2 FEBIP uses an electron beam in order to locally induce chemical reactions on a substrate for deposition3–5 or etching.6 FEBIP is used in various applications such as deposition of high-purity SiO2,7 fabrication of three-dimensional high aspect ratio nanostructures,8,9 and EUV mask repair10,11 among other applications. The interaction of alumina material with a focused electron beam 共FEB兲 has been studied in the past.12,13 FEBIE of alumina particles using NOCl as assist gas, including various aspects of experimental methods and analysis of results, are presented in this work. The thermodynamics of chemical reactions for selective etching is presented and discussed. II. EXPERIMENT Etching experiments were performed on alumina particles placed on the EUV mask substrate. The EUV quartz mask a兲

Electronic mail: [email protected]

2727

J. Vac. Sci. Technol. B 27„6…, Nov/Dec 2009

used in these experiments had a structured TaN / TaOx film on a Ru-capped 40–50 Mo/ Si multilayer.14,15 Alumina nanoparticles were high-purity 共99.99%兲 alpha alumina with a diameter of 100– 150 nm 关as measured by scanning electron microscopy 共SEM兲兴 obtained from Taimei Chemicals Co., Ltd. 共TM-DAR® grade series produced via precipitation and thermal treatment兲. The amount of surfactants on alumina particles is estimated to be 0.25– 0.5 wt. % following thermogravimetric analysis 共TGA兲 measurements 共as a result, the contribution of surfactants to the particle volume may be neglected兲. A dispersion of alumina particles in water was prepared at the concentration of 18 ␮g ml−1 in a pH 3 nitric acid solution. An ⬃20 ␮l drop of this dispersion was placed on a known portion of an EUV mask. Another drop was also placed on an oxygen plasma cleaned sample of silicon substrate. Both the EUV mask and the silicon sample were wet instantaneously. Samples were dried in ambient air in less than 1 min. The presence of alumina particles on the silicon substrate was verified using SEM and energy-dispersive x-ray 共EDX兲 analysis 共with approximate surface density of 106 particles/ mm2兲. Nitrosyl chloride 关NOCl, CAS-No. 共2696-92-6兲兴 was used as the etch gas in these experiments. This once-commercial gas can be prepared by processing a mixture of concentrated hydrochloric acid and nitric acid.1 The gas handling was performed in a dedicated NaWoTec/ Zeiss proprietary gas admission system. All electron irradiation experiments and imaging were conducted at the energy of 1 keV. The Monte Carlo simulations were performed with

1071-1023/2009/27„6…/2727/5/$25.00

©2009 American Vacuum Society

2727

2728

Bret et al.: Gas assisted focused electron beam induced etching of alumina

2728

FIG. 1. 关共a兲 and 共b兲兴 SEM images of an alumina particle before and after the etching experiment using NOCl as assist gas. 共c兲 AFM image. 共d兲 AFM line profile corresponding to the line specified in image 共c兲.

MOCASIM software 共software developed by L. Reimer, purchased from Plano GmbH兲.

III. RESULTS An alumina particle was etched using e-beam in presence of NOCl 共see Fig. 1兲. The alumina particle shown in this figure is on the ruthenium capping layer of the EUV mask. The size of the alumina particle was reduced due to the etching process 关from the initial diameter of 100 nm to less than 50 nm, based on the atomic force microscopy 共AFM兲 cross section in Fig. 1共d兲兴, within 2 ⫻ 106 irradiation loops. The 225 nm square exposed area was larger than the particle. Meanwhile, the electron-exposed substrate was etched less than 10 nm, as can be seen from the AFM image 关Fig. 1共c兲兴. The surface outside the irradiated square in Fig. 1共b兲 appears brighter, as a round halo which decreases in intensity with the distance to the particle. This can be due to electroninduced surface chlorination of ruthenium, which would affect the secondary electron 共SE兲 yield. This surface chlorination is most likely reversible in a light N2 or O2 plasma or by exposure to air 共as will be further investigated with surface analysis methods兲. It is important that the EUV reflectivity remains unchanged by this process. Another surface modification factor that may have contributed to the change in SE yield is electron-induced contamination 共up to 5 nm thick outside the area irradiated by the primary beam, as can be seen in the AFM profile兲. With the sample preparation method used in this work 共liquid phase deposition of nanoparticles兲, the surface cannot be guaranteed to be free from organic contamination. In order to demonstrate the repeatability of the alumina etching process using NOCl, additional etching tests were undertaken. A sequence of images taken while etching two J. Vac. Sci. Technol. B, Vol. 27, No. 6, Nov/Dec 2009

facetted alumina particles connected with each other is shown in Fig. 2. The electron beam was scanned on the nondark regions in Fig. 2共b兲 共for 9 min兲 and Fig. 2共c兲 for 30 min, representing a total of 106 loops. The etch process was isotropic on the facetted particles. Although particles decreased in size, they remained attached during the etch process. This is likely due to the attraction forces being stronger than the repulsive Coulomb forces from electron beam induced charging. The AFM image and line profile 关Figs. 2共d兲 and 2共e兲兴 show damage of less than 10 nm to the

FIG. 2. 关共a兲–共c兲兴 SEM image sequence of two Al2O3 particles etched using NOCl as assist gas at various loop times. 共d兲 AFM image. 共e兲 AFM line profile corresponding to the line specified in image 共d兲. 关共f兲 and 共g兲兴 overview SEM images before and after the experiment.

2729

Bret et al.: Gas assisted focused electron beam induced etching of alumina

FIG. 3. 关共a兲 and 共b兲兴 SEM images of an Al2O3 particle sitting on the structured EUV absorber film before and after FEB-induced etching experiments using NOCl as assist gas.

ruthenium surface. A halo with decreasing intensity, when moving away from the etch-experiment area, is still visible. Particles which were not directly exposed to the beam 关top right of Figs. 2共f兲 and 2共g兲兴 were not affected. The etch process proved even more selective versus TaN / TaOx, as shown in Fig. 3. After 106 irradiation loops on a particle sitting on the structured layer of the EUV mask, the absorber material was completely free from side effects. Work is in progress to improve the etch rate and make this a practical process for the rapid and complete removal of alumina particles. IV. DISCUSSIONS The origin of the halo can be explained based on Monte Carlo simulations of electron trajectories. A model sample composed of an alumina sphere on top of an EUV mask was shot at with 1 keV electrons, as shown in Fig. 4. Three electron trajectories are shown: impinging at the top of alumina particle 共1兲, on the side of the alumina particle 共2兲, and on the EUV mask 共3兲. Most of the electrons impinging at the edge of the particle are scattered forward and can induce surface reactions on the substrate. This effect results in the halos seen in Figs. 1 and 2 and is mostly restricted to the Ru cap due to the narrow electron penetration depth.16 The halos will be avoided in the future by confining more carefully the exposure area away from particle edges. We propose the following rationale to account for alumina etching in the presence of the electron beam and NOCl. Electrons 共primary or secondary兲 provide the activation energy 共via various electron-induced processes兲 for all possible chemical reactions to take place 共based on the present compounds on the substrate surface and the adsorbed gasses兲.

FIG. 4. Monte Carlo simulation of electron trajectories using 1 keV electrons impinging at three different points 共1, 2, and 3兲 on an alumina particle on EUV mask 共1: 40 electron trajectories in the center of the particle; 2: 80 electron trajectories on the side of the particle; and 3: 40 electron trajectories on the EUV mask兲. JVST B - Microelectronics and Nanometer Structures

2729

However, reactions that advance are those resulting in an overall reduced energy of products. Tabulated standard enthalpies of formation17 for various compounds are used in the analysis below. In the presence of the electron beam, the oxygen in alumina can desorb from the surface 关see Eq. 共1兲, below兴, forming metallic aluminum.12 Adsorbed NOCl on the surface may dissociate into Cl* and NO 共or gaseous byproducts兲 in the presence of e-beam 关see Eq. 共2兲兴. The chlorine atom radicals can react exothermically with metallic aluminum to form AlCl3, as shown in Eq. 共3兲. Alternatively Eqs. 共1兲–共3兲 can be combined to obtain the enthalpy of direct surface chlorination, as in Eq. 共4兲. e-beam

Al2O3 ——→ 2Al + 3/2O2↑,

⌬ f H0 = 1676 kJ mol−1 , 共1兲

e-beam

NOCl ——→ 1/2N2↑+ 1/2O2↑+ Cl*, ⌬ f H0 = − 52.6 kJ mol−1 , 3Cl* + Al → AlCl3,

⌬ f H0 = − 704.2 kJ mol−1 ,

共2兲 共3兲

e-beam

Al2O3 + 6NOCl ——→ 2AlCl3 + 9/2O2↑+ 3N2↑, ⌬H0 = − 48 kJ mol−1 .

共4兲

The formation of other volatile compounds, such as aluminum oxychlorides, can also contribute to the electron beam assisted etching. The negative ⌬H0 of Eq. 共4兲 indicates that the reaction is slightly exothermic and, hence, thermodynamically possible. However, with such a low ⌬H0 absolute value 共as compared to the much larger energy required to break the Al–O bonds兲, the reaction cannot take place spontaneously. This limits the reaction to the areas activated by the electron beam. For a comparison, 48 kJ mol−1 represents roughly 0.5 eV atom−1, which is the typical energy deposited by the beam at an elastic collision event with a nucleus. The compound AlCl3 has a sublimation enthalpy of ⬃59 kJ mol−1.18 It sublimes as Al2Cl6 with a vapor pressure of ⬃3 ⫻ 10−4 mbar at room temperature,22 which is the driving force for the etch process since the equilibrium gets displaced. After sublimation of Al2Cl6, alumina from the layer below undergoes the same process and the etching continues in the presence of e-beam. Comparing the ⌬H0 values gives a quantitative idea of the relative bond strengths in the reactants and the products. We omitted here the quantitative evaluation of the entropic contributions 共formally necessary to calculate the change in Gibbs free energy兲 due to the lack of ⌬S0 values for some of the involved reactions. However, since Eq. 共4兲 results in the net production of gaseous compounds from solids, the entropic contributions should accelerate the reaction. A more detailed experimental study would involve, among others, investigating the role of the temperature on the etch rate, in which the entropic factors also play a larger role. The etch rate of particles may be different than that of films due to

2730

Bret et al.: Gas assisted focused electron beam induced etching of alumina

increased SE emission at edges and weaker thermal coupling. Further investigations will also involve experiments on flat samples. The etching process can be counteracted by side reactions. Hydrolyzing NOCl or AlCl3 with residual water vapor yields gaseous HCl. This decreases the amount of adsorbed chlorine and slows down the process. Partly chlorinated aluminum can also be reoxidized to alumina by O-containing gases 共such as oxygen, NOCl itself, or water兲, as shown below, e-beam

2AlCl3 + 3/2O2 ——→ Al2O3 + 3Cl2↑, ⌬H0 = − 267 kJ mol−1 ,

共5a兲

e-beam

2AlCl3 + 3NOCl ——→ Al2O3 + 3/2N2↑+ 9/2Cl2↑, ⌬H0 = − 425 kJ mol−1 ,

共5b兲

2AlCl3 + 3H2O → Al2O3 + 6HCl↑, ⌬H0 = − 979 kJ mol−1 .

共5c兲

With significantly negative enthalpy change values, these reactions 共when taking place兲 are efficient competitive pathways to the alumina etch process. This was confirmed in an experiment with conditions similar to the results presented in Figs. 1 and 2, in which a controlled H2O gas flow was added to the NOCl flow. In this experiment, the etch rate was then almost zero 共i.e., no measurable etching was observed兲. This demonstrated the detrimental role of oxygen-containing species on the etch process and partly explains the slow alumina etch rate observed. SEM images in Fig. 3 show a good etch process selectivity of alumina with respect to TaN / TaOx. The surface of TaN / TaOx is composed of tantalum oxide.15 Compounds that could be formed under e-beam irradiation on the TaOx surface in the presence of NOCl would be tantalum chloride 共TaCl5兲 or oxychlorides 共TaOxCly兲. The standard enthalpy of formation of Ta2O5 is reported as −2038 kJ mol−1 共Ref. 19兲 and that of TaCl5 as −860 kJ mol−1.20 Although the formation of volatile TaOxCly has been reported to only occur at temperatures above 240 ° C,21 TaCl5 could be generated exothermically at room temperature under the electron beam, as shown below, e-beam

Ta2O5 + 10NOCl ——→ 2TaCl5共ads.兲 + 15/2O2↑ + 5N2↑, ⌬H0 = − 208 kJ mol−1 ,

共6a兲

2TaCl5共ads.兲 + 5NOCl ——→ Ta2O5 + 15/2Cl2↑ + 5/2N2↑, 共6b兲

TaCl5 has a low volatility: its vapor pressure 共⬃5 ⫻ 10−5 mbar at 25 ° C, with a sublimation enthalpy of J. Vac. Sci. Technol. B, Vol. 27, No. 6, Nov/Dec 2009

⬃97 kJ mol−1兲 共Ref. 22兲 is five to ten times lower than that of Al2Cl6 共⬃3 ⫻ 10−4 mbar, see above兲, which removes 2 Al atoms per desorption event. The other reason for the low efficiency of an etch reaction like Eq. 共6a兲 is the oxidizing nature of NOCl. With the desorption of TaCl5 being slow, the competing reaction of O incorporation into a partly chlorinated surface takes place, as described by Eq. 共6b兲. This reaction is more exothermic than Eq. 共6a兲 and consumes half as much NOCl. In the same way as Al etching is slowed down by Al–O bond formation, the thermodynamically favored formation of Ta–O bonds as compared to Ta–Cl hampers the electron-induced surface effects. The vapor pressure data combined with the thermodynamics of chemical reactions explain the etch selectivity of Al2O3 with respect to TaN / TaOx. The etch selectivity of Al2O3 with respect to Ru, as observed in Figs. 1 and 2, can also be explained by the low volatility of ruthenium trichloride, which would sublime only in the range of 500– 800 ° C.23 As a summary, a new thermodynamic approach based on limited hypotheses accounts well for the observed phenomena. Etching of alumina 共or, by extension, of elemental Al兲 in the presence of NOCl is thermodynamically possible, thanks to the volatility of AlCl3: this opens the possibility of activating the particle etch locally with an electron beam since the reaction does not take place spontaneously. The type of electronic activation used in this study was not selective enough to resolve crystalline directions, so the etch reaction was isotropic. The chlorination reaction of Ta competes with efficient oxidation and leads to less volatile compounds, and Ru chlorides are even less volatile: the Al2O3 etch reaction is selective with respect to these substrate materials. V. CONCLUSIONS The successful etching of alumina nanoparticles as model contaminants on an EUV mask was demonstrated using gas assisted FEB-induced etching. The etch gas used was nitrosyl chloride, leading to an etch process selective with respect to Ru and TaN / TaOx. The observed surface effects and process resolution were explained based on the simulation of the electron trajectories. The etch process and the observed related effects 共such as isotropicity, limitation to the electronexposed area, drastic reduction in etch efficiency in the presence of water, and selectivity兲 were explained in light of chemical reaction thermodynamics. This study adds aluminum, as an element, to the list of FEB-induced etching possibilities. It also presents a rational method of design and interpretation of future etching processes. ACKNOWLEDGMENTS

e-beam

⌬H0 = − 581 kJ mol−1 .

2730

This work was partially funded by Intel Corporation. The authors thank Alke Finke at EPFL for preparation of alumina nanoparticle dispersions and Paul Bowen at EPFL for having carried out the TGA measurement of alumina particles. L. J. Beckham, W. A. Fessler, and M. A. Kise, Chem. Rev. 共Washington, D.C.兲 48, 319 共1951兲.

1

2731

Bret et al.: Gas assisted focused electron beam induced etching of alumina

2

I. Utke, P. Hoffmann, and J. Melngailis, J. Vac. Sci. Technol. B 26, 1197 共2008兲. 3 A. E. Ennos, Br. J. Appl. Phys. 5, 27 共1954兲. 4 A. G. Baker and W. C. Morris, Rev. Sci. Instrum. 32, 458 共1961兲. 5 R. W. Christy, J. Appl. Phys. 31, 1680 共1960兲. 6 J. W. Coburn and H. F. Winters, J. Appl. Phys. 50, 3189 共1979兲. 7 A. Perentes and P. Hoffmann, Chem. Vap. Deposition 13, 176 共2007兲. 8 I. Utke, A. Luisier, P. Hoffmann, D. Laub, and P. A. Buffat, Appl. Phys. Lett. 81, 3245 共2002兲. 9 P. Hoffmann, I. Utke, and F. Cicoira, Proc. SPIE 5023, 4 共2003兲. 10 T. Liang, A. R. Stivers, M. Penn, D. Bald, Ch. Sethi, V. Boegli, M. Budach, K. Edinger, and P. Spies, Proc. SPIE 5446, 291 共2004兲. 11 K. Edinger et al., Proc. SPIE 5256, 1222 共2003兲. 12 J. W. Park, A. J. Pedraza, and W. R. Allen, J. Vac. Sci. Technol. A 14, 286 共1996兲. 13 M. E. Mochel, I. A. Eades, M. Metzger, J. D. Meyer, and J. M. Mochel, Appl. Phys. Lett. 44, 502 共1984兲. 14 P. Y. Yan, G. Zhang, E. Spiller, and P. B. Mirkarimi, Proc. SPIE 5446, 824 共2004兲.

JVST B - Microelectronics and Nanometer Structures

15

2731

M. G. Lassiter, T. Liang, and P. D. Rack, J. Vac. Sci. Technol. B 26, 963 共2008兲. 16 T. Bret, I. Utke, P. Hoffmann, M. Abourida, and P. Doppelt, Microelectron. Eng. 83, 1482 共2006兲. 17 P. Atkins and J. de Paula, Atkins’ Physical Chemistry, 8th ed. 共Oxford University Press, Oxford, 2006兲, p. 995. 18 B. Brunetti, V. Piacente, and P. Scardala, J. Chem. Eng. Data 54, 940 共2009兲. 19 K. T. Jacob, C. Shekhar, and Y. Waseda, J. Chem. Thermodyn. 41, 748 共2009兲. 20 P. Gross, C. Hayman, D. L. Levi, and G. I. Wilson, Trans. Faraday Soc. 56, 318 共1960兲. 21 J. Aarik, K. Kukli, A. Aidla, and L. Pung, Appl. Surf. Sci. 103, 331 共1996兲. 22 A. Niskanen, T. Hatanpää, M. Ritala, and M. Leskelä, J. Therm Anal. Calorim. 64, 955 共2001兲. 23 H.-J. Cantow, H. Hillebrecht, S. N. Magonov, R. W. Rotter, M. Drechsler, and G. Thule, Angew. Chem., Int. Ed. Engl. 29, 537 共1990兲.