Surface Smoothing and etching by gas cluster ion beam

J.H. Song and W.K. Choi

Thin Film Technology Research Center, Korea Institute of Science and Technology, Cheongryang P.O. Box 131, Seoul 136-791, Korea

Abstract. Ar and CO2 gas cluster ion beam with a few nm size were generated by an adiabatic expansion through Laval nozzle. The existence and the mean size distribution of the cluster were analyzed by time-of-flight measurement. Crater induced by Ar cluster ion beam and crown-like hillock by CO2 cluster ion impact on Si(100) were observed by an atomic force microcopy. CO2 cluster ion was irradiated on Si at 25 kV with the variations of ion dose from 1010 to 1013 cluster ions(CI)/cm2, at the flux of 109/cm2 s. Through this isolated cluster ion impact, the interaction mechanism between cluster ion with solid surface was suggested to be made of three steps: surface embossment, surface sputtering and smoothing, and surface etching. Another surface smoothing and etching experiment using CO2 cluster ion beam were carried out over ITO/glass and Cr-masked Si3N4 thin film surfaces at 25 kV.

I. Introduction

Recently gas cluster ion beam having a few nm size has been interesting since this nanoparticle ion beam known to be exclusively prominent in atomic-scale surface smoothing and useful in hard material etching like CVD diamond and SiC, due to large sputtering yield and lateral momentum transfer [1]. Besides improving the performance of TMR magnetic multilayer NiFe [2] and microwave resistance of high Tc cuprate superconducting YBaCuO [3] films through surface smoothing, its application for nano secondary ion mass spectroscopy (n-SIMS) is quite predictable in future. Related to semiconductor technology, new cluster ion beam source with high current and broad beam size is highly demanded for high speed and large area surface smoothing for CMP process. Also due to its capability of delivering large kinetic energy without much damage and negligible surface charging, it emerged as new candidate over new low energy ion implantation technique for fabricating shallow junction in VLSI [4]. In view of fundamental cluster ion-solid interaction, there is still controversy over the mechanism of formation of hillock or crater when the cluster ion beam is irradiated on solid surface [5-7].

In this article, surface interaction with cluster ion beam using Ar and CO2 cluster was investigated and surface smoothing and etching results are presented when CO2 cluster ion beam was irradiated on Si, ITO, and Si3N4.

II. Experimental Figure 1 illustrated the schematic diagram

of 150 kV cluster ion accelerator consisting of cluster source, acceleration, and main experimental chamber. Ar and CO2 cluster was generated by an adiabatic expansion through a Laval nozzle with the diameter of 0.1 mm throat at 4.5 and 5 bar, respectively. Expanded cluster was by skimmer and then ionized by electron impact. The cluster ion beam could be accelerated up to 150 kV and irradiated on the sample surface. Other experimental set up can be found in detail elsewhere [8,9]. From the time-of-flight measurement, the mean size of cluster distribution was about 1000 for Ar and 750 molecules for CO2 at room temperature. At 25 kV and the very low cluster ion flux, Ar and CO2 cluster were impacted on Si(100) surface and the formation of cluster ion induced structure was examined. For the study of evolution of surface morphology, CO2 cluster ion was irradiated on Si(100) surface, in which native oxide layer was removed by dilute HF solution, and the ion dose was varied in the range of 1010 - 1013 cluster ions(CI)/cm2. In addition, CO2 cluster ion beam was irradiated on commercial ITO surfaces and Cr-masked Si3N4 thin film grown by plasma enhanced chemical vapor deposition (PECVD) on Si

Fig. 1 Schematic diagram of a 150 kV cluster

Fig. 2 Time-of-flight spectra of Ar (P=4.5 bar) and CO2 cluster (P=5 bar) at room temperature.

III. Results and discussion

1. Isolated Cluster Impact

induced Structure In order to investigate the cluster ion-solid

interaction, isolated impact at the very low flux was carried out over Si surfaces. It is very interesting that two different kinds structures are formed, crater or hillocks at the low dose of Ar and CO2 cluster ion impact. In case of Ar cluster ion impact, crater is formed as it is expected from

the point of ballistic collision. As shown in Fig. 3(a), the shape of the crater is not symmetry and the sputtered particles are not isotropically redeposited. On the other hand, in case of CO2 cluster ion impact, outgrown hillocks are induced with the a few ten’s nm diameter and a few height. Among them, huge size of crater shaped structure with size of about 1 µm and a few tens nm height is rarely found as shown in Fig. 3(b).

Previously, observation of crater formation from the ion bombardment of 125 keV Bi+ and 250 keV Bi+ was reported by Merkle and Jager.[10] Beuhler and Friedman also presented that holes with a diameter of 6 nm formed in a 9.5 nm thick Pt-C film as (H2O)50 cluster ions at 250 keV.[11,12] Recently, Yamada et al, reported that an isolated Ar cluster ion impact induced not hillocks but craters, and which were observed using a scanning tunneling microscope (STM) on Au/sapphire [7] and HOPG (highly oriented pyrolitic graphite) surface [13].

Gspann reported that impacting on Si surfaces at the acceleration voltage of 100 kV, supersonic cluster ion beam induced hillocks with nm height instead of surface craters through atomic force microscope [5]. This unexpected result was explained in terms of the rebounce of elastic target materials for generated shock waves. According to the calculation [14], a hemispherical crater and two or three-layered shock waves were once created after the impact, but the created crater was immediately filled up with the fluidized hot carbon material due to elastic recovery before the arrival of reflected shockwaves.

From the results in this experiment, the difference of the cluster ion induced structure results from the different chemical reactivity. In case of Ar cluster ion impact, the cluster ion

Nozzle

Ionizer Reflectron

EinzelLens

Accelerationtube

Scanner

TMP TMP TMP

Sourcechamber

Processchamber

MBP

Expansionchamber

SkimmerPermanentmagnet

V

0 200 400 600 800

CO2, p=5.0 bar

Ar, p=4.5 bar

Inte

nsity

(arb

. uni

ts)

Flight time (µs)

Fig. 3 AFM images of isolated (a), (b) Ar cluster and (c) CO2 ion impacted Si surface.

collides with the substrate and Si atoms are sputtered and it coincides with the model of macroscopic ballistic collision. However, when CO2 cluster ion is impinged into the Si surface, a chemical reaction occurs and some silicon oxide species are easily composed in locally high pressure and very high temperature environment. This Si-O bonding is larger than Si-Si bonding and therefore the compound should be outgrown for the existence. Later a chemical analysis of the hillocks will be ready for the identification of chemical composition and structure through high

Fig. 4 AFM image of Si surface irradiated at the dose of 5x1014 CI/cm2 at 25 kV. resolution NEXAFS and TEM [15].

As the variation of ion dose, the Si surface roughness (not shown here) is increased from 0.4 nm for bare Si wafer cleaned by dilute HF to 1.2 nm after cluster ion impact 5x1011 cluster ions (CI)/cm2 by the increase of the number of induced hillocks. And it was saturated 1.22 nm after ion dose 5x1012 CI/cm2. Based upon these results of isolated cluster ion impact, it was already suggested [16] that cluster ion-solid interaction evolves with subsequent three-step processes. It was described into surface embossment, surface sputtering and smoothing, and surface etching. Firstly surface embossment happens at the beginning stage of low ion dose by the formation of protruding hillocks. And then secondly, surface sputtering begins over critical ion dose at which the area of induced hillock is equal to unirradiated area. Simultaneously, sputtered atoms from the hillocks migrate and fill the valleys, called surface smoothing. Lastly, modified surface region by cluster ion impact will be easily removed into the vacuum and in consequence so-called surface etching occurs.

Figure 4 shows the Si surface irradiated at the dose of 5x1014 CI/cm2 at 25 kV and was etched as deep as 6 nm and the surface roughness becomes 0.7 nm. Compared to unirradiated area, it shows very flat surfaces.

2. Cluster Ion Irradiation On

ITO and Si3N4 Figure 5 illustrates AFM images for bare and

irradiated ITO surfaces by CO2 monomer and cluster ions. On the scanned ITO area in Fig. 5(a)

(a)

(b)

(c)

Fig. 5 AFM images of (a) bare ITO, (b) monomer ion irradiated ITO, and (c) cluster ion irradiated ITO

about 10 hillocks with 15 nm in height and a few hundreds nm in width are observable. The root mean square roughness σrms of the bare ITO surface is 1.31 nm. Fig. 5(b) illustrates the ITO surface irradiated by the monomer ions formed at the inlet pressure of 1 bar and with the dosage of 1.5x1014/cm2. In this case, the density of the hillocks on the surface is not much changed. However, the shape of the hillocks turns into a spike-like one from a stalagmite-like one. Moreover, the surface roughness slightly increases up to σrms =1.6 nm. On the other hand, when cluster ions are irradiated, some different features are observed

Fig. 6 AFM image of cluster ion irriadiated Si3N4 surfaces from the ITO surfaces. After the cluster ion bombardment at the dosage of 5x1014/cm2, the irradiated surface becomes smoother σrms =0.94 nm than the bare ITO surface and the ITO surface irradiated by monomer ions as shown in Fig. 5(c).

In order to extend cluster ion beam in MEMS technology, one of the widely used thin film of strain-released Si3N4 was irradiated by cluster ion beam. Si3N4 thin film was deposited on Si by PECVD and Cr mask was deposited for the fabrication of some MEMS structure. As shown in Fig. 6, irradiated area was etched and the surface roughness became atomically flat as much as 0.16 nm, which is exceptionally smooth compared to 1 nm of as-deposited Si3N4.

IV. Conclusions Ar and CO2 cluster ion beam was generated by an adiabatic expansion through a Laval nozzle at room temperature and irradiated onto Si, oxide and nitride surface at 25-50 kV. In case of an isolated impact, two different kinds of induced structures were observed, hillocks and crater. When large CO2 cluster ion was impinged into Si, crown like denting crater was found at top of the

(c)

(a)

(b)

σrms=0.16 nm

σrms=1.1 nm

hillock. This difference feature of the cluster ion impact is expected from the difference in both size and chemical reactivity. After prolonged isolated cluster ion impact on clean Si surface, it is observed that cluster ion-solid interaction is phenomenologically evolved in three steps: surface embossment, surface sputtering and smoothing, and surface etching. When cluster ion beam was irradiated onto ITO surface where hillocks were preexisted, it shows different sputtering phenomena from monomer ion irradiation. From the cluster ion irradiation on Si, ITO, and Si3N4, all irradiated surfaces became very smooth and it was proved that cluster ion beam is very effective in surface smoothing and nano etching.

Acknowledgement This work is partially supported by the Tera-level Nano Device (TND) National Program.

References 1. I. Yamada, Eur. Phys. J.D9, 55 (1999). 2. J.A. Greeger, D.B. Fenner, J. Hautala, L.P.

Allen, V. DiFilippo, N. Toyoda, I. Yamada, J. Matsuo, E. Minami, H. Katsumata, Surf. Coat. Tech. 133/134, 274 (2000).

3. D. Fathy, O.W. Holland, R. Liu, J. Wosik, and W.K. Chu, Mater. Lett. 44, 248 (2000)

4. I. Yamada, J. Matsuo, and N. Toyoda, 13th International Conference on Ion Beam Modification of Materials (Kobe, Japan, 2002) 5. Gspann, J., Nucl. Instru. Meth. B112, 86 (1996). 6. J.H. Song and W.K. Choi, Nucl. Instru. Meth. B190, 792 (2002). 7. D. Takeukuchi, K. Fukushima, J. Matsuo, I. Yamada, Nucl. Instru. Meth. B 153, 264 (1999). 8. J.H. Song, S.N. Kwon, D.K. Choi, and W.K. Choi, Nucl. Instru. Meth. B179, 568 (2001). 9. J. H. Song, D.K. Choi, and W.K. Choi, Current Appl. Phys. 1, 521 (2001) 10. K.L. Merkle and W. Jager, Philo. Mag. A44, 741 (1981) 11. R. Beuhler and L. Friedman, Chem. Rev. 86, 521 (1986) 12. M.W. Matthew, R. Beuhler, M. Ledbetter, and L. Friedman, J. Phys. Chem. 90, 3152 (1986) 13. D. Takeuchi, T. Seki, T. Aoki, J. Matsuo,

and I. Yamada, Chem. Phys. 54, 76 (1998). 14. Y. Yamaguchi and J. Gspann, Eur. Phys. J.D.16, 103 (2001). 15. L.P. Allen, Z. Insepov, D.B. Fenner, C. Santeufemio, W. Brooks, K.S. Jones, and I. Yamada, J. Appl. Phys. 92, 3671 (2002). 16. J.H. Song, D.K. Choi, and W.K. Choi, Nucl. Instru. Meth. B, 196, 268 (2002).