Ar Inductively Coupled Plasma

Download PDF
2 downloads 0 Views 570KB Size Report
Comment
Mar 13, 2016 - products are effectively removed, which results in there being a negligible amount of Cl element on the surface. The melting points of the HfCl4 ...
Home

Search

Collections

Journals

About

Contact us

My IOPscience

Dry Etching of TaN/HfO2 Gate Stack Structure by Cl2/SF6/Ar Inductively Coupled Plasma

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2005 Jpn. J. Appl. Phys. 44 5811 (http://iopscience.iop.org/1347-4065/44/7S/5811) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 147.32.96.164 This content was downloaded on 13/03/2016 at 16:32

Please note that terms and conditions apply.

Japanese Journal of Applied Physics Vol. 44, No. 7B, 2005, pp. 5811–5818 #2005 The Japan Society of Applied Physics

Dry Etching of TaN/HfO2 Gate Stack Structure by Cl2 /SF6 /Ar Inductively Coupled Plasma Myoung Hun SHIN, Sung-Woong N A, Nae-Eung LEE, Tae Kwan O H1 , Jiyoung K IM1 , Taeho LEE2 and Jinho A HN2 Department of Materials Engineering and Center for Advanced Plasma Surface Technolog, Sungkyunkwan University, Suwon, Kyunggi-do 440-746, Korea 1 Department of New Materials Engineering, Kookmin University, Seoul 861-1, Korea 2 School of Materials Science and Engineering, Hanyang University, Seoul 133-791, Korea (Received October 29, 2004; accepted April 21, 2005; published July 26, 2005)

The dry etching characteristics of the TaN/HfO2 gate stack structure using Cl2 /Ar, Cl2 /SF6 /Ar and Cl2 /SF6 /O2 /Ar inductively coupled plasmas (ICPs) were investigated and the etch rates of the TaN and HfO2 layers and TaN/HfO2 etch rate selectivities were compared. The results obtained for the TaN/HfO2 etching by varying the Cl2 /Ar gas mixing ratio, the top ICP electrode power, and the dc self-bias voltage (Vdc ) in the Cl2 /Ar plasmas showed that low etch selectivities were obtained, due to the high HfO2 etch rate. The effects of adding SF6 to the Cl2 /Ar plasmas and adding O2 flow to the SF6 /Cl2 / Ar chemistry were investigated for the purpose of improving the etch selectivity. Etch experiments performed by varying the Cl2 /SF6 /Ar gas mixing ratio and Vdc value in SF6 /Cl2 /Ar plasmas, combined with X-ray photoelectron spectroscopy measurements, showed that the etch rates were reduced compared to those in Cl2 /Ar chemistry, due to the heavy fluorination of the surface, however the etch selectivity was increased, due to a disproportionate decrease in the TaN and HfO2 etch rates. The addition of O2 flow to the SF6 /Cl2 /Ar plasma also increased the etch selectivity at an O2 flow rate of 5 sccm, due to the TaN etch rate being increased, while the HfO2 etch rate remained almost constant. [DOI: 10.1143/JJAP.44.5811] KEYWORDS: HfO2 , TaN, plasma etching, ICP (inductively coupled plasma), metal gate stack

1.

Introduction

For the integration of Hf-based high-k dielectric materials including HfO2 ,1–6) HfON,7,8) HfAlO,9) HfSiO,10) and HfSiON,11,12) the use of various metal gate electrode materials including TaN,8,12–15) TiN,6,15,16) HfN,5,17) WN,15) TaSiN4,18) and metal silicides15) is being widely studied for next generation nano-scale CMOS devices. The adoption of these new materials poses new integration problems, among which the selective etching of the metal gate electrodes and the high-k gate dielectrics over the Si substrate is expected to be among the critical steps in the process integration of the front-end of the line (FEOL). For the patterning of the metal gate electrode/high-k structure, there are two possible integration schemes which can be employed. In the first case, the metal gate electrode is selectively etched with a resist or hard-mask against a high-k dielectric layer and then the high-k dielectric layer is removed. In this scheme, a very high etch selectivity of the metal gate electrodes to the high-k materials in the range of 30:1– 50:1 is required, because of the very thin nature of high-k dielectric materials (typically less than 3 nm). After the metal gate electrode is etched, the remaining high-k dielectric layer needs to be removed using wet etch19–21) and/or dry etch processes21–29) with a high etch selectivity with respect to the Si substrate. In the second method, the metal gate electrode/high-k dielectric stack is selectively etched over the source/drain region in one step at a time with the same etch chemistry being used in the metal gate etch step. In this case, a very high etch selectivity of the gate stacks with respect to the Si substrate is also needed, in order to minimize the Si loss in the ultra shallow source/drain regions. In both methods, the etch selectivity of the etched layers (metal electrode or high-k dielectrics) to the underlayers (high-k dielectrics or Si substrate) is one of the most important parameters in the patterning of the gate stack structures. 

Corresponding author. E-mail address: [email protected]

Much research has been conducted into the plasma etching of typical high-k dielectric candidates, including ZrAlO,23) HfO2 ,21,22,25,27–29) HfO-based oxides,29) and ZrO2 ,24,26,27) using Cl2 or BCl3 /Cl2 ,21,23–26) HBr/Cl2 / O2 ,28,29) SF6 /Ar,21) and CHF3 /CF4 /Ar/O2 22,29) chemistry in RIE (reactive ion etching),22) ECR24–27) and ICP21,28,29) etchers. Many plasma etching studies of the high-k dielectrics showed that an etch rate higher than 100 nm/min can be obtained.22,23,28,29) Previous reports on the etch rate selectivities of the high-k dielectric layer to the Si substrate in the case of ZrO2 etching using SF6 /Ar and Cl2 /Ar ICPs21) in electron cyclotron resonance (ECR) BCl2 /Cl2 plasmas25,26) showed that etch selectivities of HfO2 to Si of less than 1 and 1.5, respectively, were obtained. These results indicate that it is generally very difficult to obtain a high etch selectivity of the high-k dielectric layer with respect to the Si substrate. As a result, the one-step plasma etching of metal gate electrode/ high-k dielectric stacks over the Si substrate is probably limited by the inherently similar or faster etch rate of the Si layer as compared with that of the high-k dielectrics. Since the high-k dielectric can be etched away by a wet and/or chemical dry etch process without leaving any residues on the Si substrate after the etching of the gate electrodes,20) the two-step etching of the metal gate electrode followed by the removal of the high-k dielectric might be a practical solution. For practical application during process integration, therefore, the low etch selectivity of the high-k dielectrics to the Si substrate and the difficulty in removing the etch residues after etching19,20) are the primary limiting factors in the gate patterning, rather than the etch rate itself. In this respect, the development of an etching process with a high etch selectivity of the metal gate electrode with respect to the high-k dielectric is required. In this work, as a model system for studying the etching characteristics of metal gate electrode/high-k dielectric stack structures, namely the TaN/HfO2 gate structure, was chosen. The plasma etching of various Ta containing layers, such as Ta,30–33) Ta2 O5 ,34) the TaN/Ta/TaN stack,35) and

5811

5812

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

TaN36) was investigated using various etch chemistries, such as Cl2 ,30–32,36) Cl2 /SF6 ,33) Cl2 /CF4 ,33) CH4 /H2 /Ar,34) SF6 ,34) and SiCl4 /NF3 35) chemistries in ECR,31–33) and ICP etchers.30,34–36) In these previous works, etch rates for the Tacontaining layers in the range of several tens to hundreds of nanometers per minute were achieved using chlorine- or fluorine-based chemistries and, in general, the etch rates of these materials were quite similar.33,34,36) Detailed investigations of the plasma etching of the metal gate electrode/high-k gate stack structure have rarely been reported. On the other hand, several studies of etching experiments employing high-k dielectric or metal gate electrodes have previously been reported.28,35) The plasma etching of the poly-SiGe/HfO2 stack structure using HBr/ Cl2 /O2 by Chen et al.28) showed that adding O2 to the HBr gas at reduced bias power was very effective in increasing the etch selectivity of the poly-SiGe layer to the HfO2 layer to a value as high as 70, possibly due to the formation of HfOx or HfOx Bry layer on the surface. Shimada et al. reported that, during the ICP etching of TaN/Ta/TaN multilayer electrodes on the SiO2 or Si3 N4 gate dielectric using SiCl4 /NF4 etch chemistries, the etch selectivity of the metal gate multilayers over the SiO2 and Si4 N4 gate dielectrics could be increased to over 50 and 150, respectively.35) These experimental results indicate that the formation of an inhibiting or passivation layer on the highk gate dielectric surface, without degrading the etch rate of the metal gate electrode materials, was critically important for achieving the required etch selectivity. In this report, the etch rates and etch selectivity of TaN/ HfO2 gate stack structures on an Si substrate were investigated by varying the process parameters including the etch gas mixing ratios (Cl2 /Ar, and Cl2 /SF6 /Ar), the additive O2 flow, the top electrode power, and the bottom bias electrode power in the ICP etcher. 2.

Experimental Procedure

The etching experiments were carried out in a modified commercial 8-inch ICP etching system having a 3.5-turn spiral copper coil on the top of the chamber separated by a 1cm-thick quartz window and pumped by a turbo molecular pump backed by a rotary pump. The substrate holder temperature was kept constant during etching by circulating cooling water at 18 C. The samples were fixed on an 8-inch wafer placed on the substrate holder using a heat-conductive paste, DC 340 (Dow Corning). A turbo molecular pump was used to reduce the residual gas pressure to below  ¼104 Torr before etching. RF power at 13.56 MHz was applied to the top coil electrode in order to induce the ICP. RF power at 13.56 MHz was fed to the bottom electrode in order to induce the DC self-bias voltage (Vdc ) to the wafer. For the sample preparation, an HfO2 layer with a film thickness of 80 nm was deposited on the Si(001) substrate by r.f. reactive sputtering of the Hf target using O2 /Ar sputtering gas. Herein, a relatively thick HfO2 layer was used for the evaluation of the etch selectivity. Also, a 300nm-thick TaN layer was deposited on the HfO2 /Si(001) substrate by d.c. reactive sputtering of the Ta target using N2 /Ar sputtering gas. The etching masks used for the TaN/ HfO2 /Si(001) samples were patterned by optical lithography using a positive photoresist (PR) with a thickness of 1.2 mm.

M. H. S HIN et al.

First, the TaN etch rates were measured by partially etching the TaN layers in the TaN/HfO2 stacks with the photoresist mask. Then, the TaN/HfO2 stack structures were etched, so that the samples were partially etched down to the HfO2 layer. Using the pre-measured TaN etch rates, the HfO2 etch rates and etch selectivity were estimated. The surface binding states of the etched TaN thin film in the Cl2 /SF6 /Ar plasmas were investigated using an AESXPS ESCA 2000 spectroscope, in order to identify the chemical binging states of the HfO2 surfaces. The Mg (K) source used for the XPS experiment provided non-monochromatic X-rays at 1253.6 eV. XPS narrow scan spectra of all the interesting regions were recorded with a pass energy of 20 eV. The take-off angle, i.e. the angle between the axis of the detector and the substrate normal, was kept at 0 and the angle of incidence of the X-rays on the substrate surface was 54.7 . The binding energy of 244.6 eV, corresponding to the C 1s spectra, was used as a reference peak position. The etched depth and profile of the TaN/HfO2 films were obtained by FE-SEM (field emission scanning electron microscopy). 3.

Results and Discussion

A.

TaN/HfO2 etching in Cl2 /Ar plasmas First, the etching characteristics of the TaN/HfO2 gate structures in the Cl2 /Ar ICP were measured by varying the Cl2 =ðCl2 þ ArÞ gas flow ratio. The etch rates and the etch selectivity of the TaN to the HfO2 layer (TaN/HfO2 etch selectivity) were measured and the results are shown in Fig. 1. During etching, the top electrode power and dc selfbias voltage, Vdc , were kept at 500 W and 200 V, respectively. The total gas flow was kept at 100 sccm. The etch rate data in Fig. 1 show that in each case the etch rates exhibit a similar trend as a function of the gas flow ratio. The etch selectivity decreased from 2.1 to 1.1 as the flow ratio increased from 20 to 60% and then leveled off when the gas flow ratio was further increased. This decrease is due to the increase in the HfO2 etch rate with increasing gas flow ratio. The etch rates and etch selectivity were also measured as a function of Vdc at the conditions of top electrode power of

Fig. 1. Etch rates of TaN and HfO2 and etch selectivity of the TaN layer with respect to the HfO2 layer as a function of the Cl2 =ðAr þ Cl2 Þ gas mixing ratio at a top electrode power of 500 W and Vdc of 200 V.

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

M. H. SHIN et al.

5813

Fig. 2. Etch rates of TaN and HfO2 and etch selectivity of the TaN layer with respect to the HfO2 layer as a function of the dc self-bias voltage, Vdc , at a top electrode power of 600 W and Cl2 =ðAr þ Cl2 Þ gas mixing ratio of 80%.

Fig. 3. Etch rates of TaN and HfO2 , etch selectivity of the TaN layer to the HfO2 layer, and dc self-bias voltage, Vdc , as a function of the top electrode power at a bottom bias electrode power of 500 W and Cl2 =ðAr þ Cl2 Þ gas mixing ratio of 80%.

600 W and a Cl2 =ðAr þ Cl2 Þ flow ratio of 80%, and the results are shown in Fig. 2. The TaN etch rate increased as Vdc increased. Even though the TaN etch rate data did not show an exact linear dependence on the square root of Vdc in the case of the ion-enhanced sputtering removal mechanism, following the relation in which the etch rate is given by AðjVdc j1=2  Vth1=2 Þ, where Vth is the threshold dc voltage for etching and A is a proportional constant,38,39) the observed increase in the etch rate with increasing Vdc indicates the importance of the ion-assisted chemical sputtering mechanism. However, the HfO2 etch rate shows different behavior with respect to Vdc . The HfO2 etch rate increased at smaller values of Vdc , but decreased again at Vdc  30 V. The etch selectivity increased from 0.35 to 2.8 as the Vdc value increased from  ¼ 0 to 90 V. The HfO2 etch rate, however, showed different behavior, as mentioned above, but the reason for this is not yet clear. The etch rates, etch selectivity, and Vdc obtained as a function of the top ICP electrode power at the bottom bias electrode power of 500 W and a Cl2 =ðAr þ Cl2 Þ gas mixing ratio of 80% are shown in Fig. 3. The Vdc value was decreased from 430 to 123 V, while increasing the top electrode power from 300 to 700 W. The TaN etch rate varied very little, due to the decrease in Vdc , in spite of the increased plasma density caused by the higher ICP power, whereas the HfO2 etch rate initially increased and then decreased with increasing ICP power. The observed variation in the HfO2 etch rate can be explained by the different etch process regimes which came into force, due to the changes in both the top electrode power and the Vdc (i.e. incident ion energy). The HfO2 etch rate was enhanced by the increased chlorination reaction (i.e. the etch process is in the radical-reaction limited regime) in the lower top power regime ( 500 W), due to the resulting increase in the Cl radical density. The HfO2 etch rate, on the other hand, was reduced by the decreased ion energy (i.e. the etch process is in the ion-flux-limited regime) in the higher top power regime ( 500 W), wherein Vdc is further decreased. As a result, the etch selectivity was decreased from 4.6 to 2.7 as

the top ICP electrode power increased from 300 to 500 W and then increased again to 4.0 when the top electrode power was increased to 700 W. The etch selectivity remained at a low value due to the high HfO2 etch rates. B.

TaN/HfO2 etching in Cl2 /SF6 /Ar plasma In order to investigate the effect of adding SF6 gas to the Cl2 /Ar plasma on the etch rates and etch selectivity, etch experiments were carried out by varying the SF6 =ðCl2 þ SF6 þ ArÞ gas flow ratio and Vdc , while keeping the total gas flow at 100 sccm and the Ar flow at 20 sccm. The top ICP electrode power was 300 W. The etch rate and selectivity results are shown in Figs. 4(a) and 4(b), respectively. As shown in Fig. 4(a), the etch rates increased monotonically with increasing jVdc j for a given gas composition. However, the etch rates of the TaN and HfO2 layers varied with different slopes as the value of jVdc j increased. Also, as the SF6 flow percentage increased, at a given Vdc , the TaN etch rate decreased, while the HfO2 etch rate slightly increased. As a result, the etch selectivity decreased as the SF6 flow percentage increased at SF6 flow percentages above 20%, as will be discussed later. Figure 4(b) shows the TaN/HfO2 etch selectivity obtained from the data in Fig. 4(a) as a function of Vdc for each gas composition. As the jVdc j value increases, the etch selectivity gradually decreases. The variation in the selectivity with jVdc j is more enhanced at lower SF6 percentages in the investigated SF6 flow rage of 20 – 80%. In this work, the highest etch selectivity of  ¼ 8:1 in the SF6 /Cl2 /Ar plasma was observed at a Vdc of  ¼ 50 V for an SF6 flow percentage of 20%. In addition to the effect of Vdc , the effect of the SF6 flow percentage on the etch rates and selectivity were investigated at a fixed Vdc of 50 V and the results are plotted in Fig. 5. The etch rates decreased abruptly as the SF6 percentage increased to 20% and then gradually decreased or slightly increased for TaN and HfO2 , respectively, as the SF6 percentage was further increased. Up to the SF6 percentage of 20%, the selectivity increased, while the rate of decrease

5814

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

M. H. S HIN et al.

Fig. 5. Etch rates of the TaN and HfO2 layers and TaN/HfO2 etch selectivity as a function of the SF6 =ðSF6 þ Cl2 þ ArÞ flow ratio at Vdc ¼ 50 V.

Fig. 4. (a) Etch rates of the TaN and HfO2 layers as a function of the dc self-bias voltage, Vdc , for each SF6 =ðSF6 þ Cl2 þ ArÞ flow ratio. The top ICP electrode power and the total ðSF6 þ Cl2 þ ArÞ flow were kept at 300 W and 100 sccm, respectively. (b) TaN/HfO2 etch selectivity as a function of the dc self-bias voltage, Vdc , for each SF6 =ðSF6 þ Cl2 þ ArÞ flow rate, as obtained from the etch rate data in Fig. 4(a).

in the HfO2 etch rate was larger than that in the TaN etch rate. The decreased selectivity with the SF6 percentage increased from 20 to 80%, however, is attributed to the decrease in the TaN etch rate, while the HfO2 etch rate remained almost constant. This disproportionate change in the etch rates of the TaN and HfO2 layers caused a large variation in the etch selectivity with increasing SF6 flow percentage. The possible reasons for this abrupt variation in the etch rates with SF6 percentage will be discussed later. In order to understand the etch mechanism and selective variation during SF6 /Cl2 /Ar etching, XPS measurements were obtained of the etched HfO2 surface in the HfO2 /Si sample with no pattern, and the results are shown in Fig. 6. During the etching performed for the sake of obtaining the XPS measurements, the top electrode power, Vdc , the chamber pressure and the total gas flow were 300 W, 200 V, 15 mTorr and 100 sccm, respectively. The Hf 4f, F 1s, Cl 2p, O 1s and C 1s spectra for various SF6 flow percentages are shown in Figs. 6(a), 6(b), 6(c), 6(d) and 6(e),

respectively. The significant intensities in the C 1s spectra shown in Fig. 6(e) were attributed to surface contamination. The peak positions are referenced to the binding energy of 244.6 eV corresponding to the C 1s spectra. No peaks were detected for the S (sulfur) element. In Fig. 6(a), the Hf 4f7=2 and Hf 4f5=2 peaks were observed at the binding energies of 16.3 and 17.8 eV corresponding to the Hf-O bonds, respectively.40) The observed binding energies of the Hf 4f7=2 and Hf 4f5=2 transitions/peaks were close to those of the HfO2 layers contaminated with the carbon on the surface.41) For the sample etched with 20% SF6 , the binding energies of the Hf 4f7=2 and Hf 4f5=2 peaks were shifted to 17.4 and 18.9 eV, respectively, representing a shift of  ¼ 1:1 eV toward the higher energy regime. This shift in the binding energy for the Hf 4f peaks for 20% SF6 chemistry is possibly due to the reduction in the amount of C on the etched surface due to there being less F on the surface,38) which is apparent from the decrease in the C 1s peak intensity [see Fig. 6(e)]. In the case of 20% SF6 chemistry, the chemical reaction of F with Hf is probably limited. With further increases in the SF6 =ðCl2 þ SF6 þ ArÞ gas flow ratio, the binding energies of the Hf 4f7=2 and Hf 4f5=2 peaks were shifted to 11.7 and 13.0 eV, respectively, representing a shift of  ¼ 4:6 eV toward the lower energy regime. The shifts in the binding energies of the Hf 4f7=2 and Hf 4f5=2 peaks are attributed to the fluorination of the HfO2 surface leading to the formation HfOx Fy on the surface, because a significant amount of F, rather than Cl, was observed on the surface, as will be described in the discussion of Fig. 6(b) and 6(c). These results are in contrast with those of a previous study, in which the CF4 etching of the HfO2 layer was reported to shift the Hf 4f7=2 and Hf 4f5=2 peaks to higher binding energies.29) In that work, the carbon from the CF4 plasma was incorporated into the film being etched and the peaks were shifted toward higher binding energies. The F 1s spectra obtained from the as-deposited and etched HfO2 /Si samples are shown in Fig. 6(b). The peak

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

M. H. SHIN et al.

5815

Fig. 6. XPS spectra obtained from the etched blanket HfO2 (70 nm)/Si(001) substrate; (a) Hf 4f, (b) F 1s, (c) Cl 2p, and (d) O 1s. The samples were etched by varying the SF6 =ðSF6 þ Cl2 þ ArÞ flow ratios at Vdc ¼ 200 V, a top ICP electrode power of 300 W, a chamber pressure of 15 mTorr, and a total ðSF6 þ Cl2 þ ArÞ gas flow of 100 sccm.

corresponding to the Hf-F bonding is observed at a binding energy of 685.4 eV, which is referenced from the data for the HfF4 compound.40) This clearly indicates the presence of a residue of Hf-F bonding elements, due to the formation of HfOx Fy by the reaction of F radicals and/or ion species with Hf in the HFO2 layer. The Cl 2p spectra were also obtained and are shown in Fig. 6(c). The satellite Hf 4d peak is positioned near the

binding energy of 202.5 eV. The two Cl 2p peaks at the binding energies of 198.2 and 199.4 eV observed in the Cl2 etched HfO2 layer25) are indicated in Fig. 6(c) as a reference. However, the Cl 2p peak intensity from the HfO2 layer etched in the Cl2 /SF6 /Ar plasmas is negligible. This result indicates the presence of a negligible amount of Hf chloride (HfClx ) residue on the etched surface, as in the case of several previous reports.25,29,36)

5816

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

The O 1s spectra obtained from the as-deposited and etched samples were obtained and are shown in Fig. 6(d). The O 1s peak corresponding to the Hf-O bonding in the presence of surface carbon was observed at a binding energy of 529.6 eV.40) The O 1s peak at the binding energy of 529.6 eV observed for the as-deposited sample indicates the presence of Hf-O bonding arising from the un-etched sample. For the etched sample, the O 1s peaks are shifted toward higher energies in the range of 530.2 – 530.4 eV, representing a shift of 0.6 – 0.8 eV. This shift in the binding energy is presumably due to the fluorination of the HfO2 layers, leading to the formation of HfOx Fy , as indicated in the Hf 4f spectra in Fig. 6(a). The observed drastic decrease in the TaN and HfO2 etch rates in the Cl2 /SF6 /Ar plasmas (Figs. 4 and 5) compared to that in the Cl2 /Ar plasmas (Fig. 2) needs to be explained in conjunction with the XPS results (Fig. 6). Even though there have been no reported experimental works involving the comparative measurements of the TaN and HfO2 etch rates in SF6 /Ar and Cl2 /Ar plasmas, the dependence of the Ta etch rate on the SF6 /(SF6 + Cl2 ) flow ratio has previously been reported.33) Since the etch rates and etch characteristics of Ta and TaN are very similar,36) a comparison can be made. Given that the etch by-products of TaCl5 and TaF5 have similarly high boiling points of 24232) and 230 C,37) respectively, the TaN etch rate would not be expected to change significantly with the addition of SF6 gas. Tsuchizawa et al.,33) however, reported that the addition of 25% SF6 flow reduced the Ta etch rate by a factor of  ¼ 2:8 compared to that without SF6 in SF6 /Cl2 ECR plasmas, and that increasing the SF6 flow ratio decreased the Ta etch rate further, which is similar to our own observation. It has been argued that the existence of oxygen in the plasma, resulting from the etching of the quartz window by F radicals and/or ions in the etcher, can decrease the Ta etch rate, due to the oxidation of Ta.33) Even though both Ta and TaN were also oxidized during Cl2 /Ar ICP etching,36) the addition of a small oxygen flow (< 6 sccm in this experiment) enhanced the TaN etch rates (see Fig. 7). Therefore, the residual oxygen in the chamber may not be responsible for the observed decrease in the TaN etch rate caused by the addition of SF6 . The observed decrease in the TaN etch rate, as the SF6 percentage increases from 0 to 20%, is presumably due to the reduction in surface chlorination which comes from increasing SF6 /Cl2 gas flow ratio. At SF6 percentages above 20%, the TaN etch rate might be limited by the surface passivation caused by the fluoride phase formed on the surface and, in turn, by the removal rate of the fluoride phase on the surface, as in the case of HfO2 etching. The reason for the observed decrease in the HfO2 etch rate as the SF6 percentage is increased from 0 to 20% is probably similar to that for the case of TaN. As indicated in the XPS measurements performed by other research groups,25,29) as well as the current experimental results, the HfCl4 etch byproducts are effectively removed, which results in there being a negligible amount of Cl element on the surface. The melting points of the HfCl4 and HfF4 compounds at 1 atm are 319 and 970 C, respectively, and the boiling point of HfCl4 is 432 C, as summarized in ref. 29. The boiling point of HfF4 is not available, however the volatility of HfCl4 is expected to be higher than that of HfF4 , considering a higher

M. H. S HIN et al.

Fig. 7. Etch rates of the TaN and HfO2 layers and TaN/HfO2 etch selectivity as a function of the O2 flow rate added to the ðSF6 þ Cl2 þ ArÞ flow rate of 100 sccm. The Vdc was kept at 50 V.

melting point of HfF4 .29) Above an SF6 percentage of 20%, the surface is passivated by the HfOx Fy formed during etching. In this condition, the HfO2 etch rate is determined by the removal rate of the HfOx Fy layer on the surface. The HfO2 etch rate is not expected to vary significantly with the SF6 flow percentage, due to the similar residual HfOx Fy layer thicknesses observed in each case, as indicated by the small variation in the Hf-F XPS peak intensity in Fig. 6(a). These experiments indicate that the fluorine-based etch chemistry suppresses the HfO2 etching. C. Effect of O2 addition on TaN/HfO2 etching in SF6 /Cl2 / Ar plasma To improve the TaN/HfO2 etch selectivity, an O2 flow was added to the SF6 /Cl2 /Ar chemistry. In order to investigate the effect of adding O2 to the SF6 (20 sccm)/ Cl2 (60 sccm)/Ar (20 sccm) chemistry on the etch selectivity, etch rate measurements were performed as a function of the O2 flow rate up to 15 sccm. The top ICP electrode power and the ðSF6 þ Cl2 þ ArÞ flow were kept at 300 W and 100 sccm, respectively. The dc self-bias voltage was kept at 50 V. The experimental condition giving the highest etch selectivity was selected from the etch results in the SF6 /Cl2 / Ar plasmas shown in Figs. 4 and 5. As can be seen from the results shown in Fig. 7, the TaN etch rate was affected very little by the additive O2 flow. However, the TaN etch rate and selectivity were affected significantly when O2 gas was added to the SF6 /Cl2 /Ar gas. The selectivity was increased from  ¼ 8 to 11.6 when an O2 flow of 5 sccm was added. When the additive O2 flow was further increased, however, the etch selectivity dropped to a lower value than that with no O2 addition, due to the decrease in the TaN etch rate. The etch selectivity can be enhanced by increasing the etch rate of the gate electrode materials, while decreasing that of the high-k gate dielectric under-layer. Chen et al.29) reported that the addition of O2 to the HBr/O2 ICP during poly-SiGe/HfO2 gate stack etching decreased the HfO2 etch rate and, as a result, increased the etch selectivity. In the

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

experimental conditions adopted in this work, however, the etch rate of HfO2 did not vary, but the etch selectivity was increased due to the increased TaN etch rate under certain etch conditions. An investigation of the reactive ion etching of the TaN layer in SF6 /O2 capacitively coupled plasmas showed that increasing the O2 flow rate, while keeping it at less than 10% of the total flow, reduced the TaN etch rate, although the number of data points obtained in these etching experiments was limited.41) The observed increase in the TaN etch rate with increasing O2 flow up to 5 sccm may be related to the variation in the plasma parameters caused by the addition of oxygen. Since the addition of O2 at a flow rate above a certain threshold reduces the TaN etch rate, as compared to the etch rate with no O2 addition, a detailed investigation of the effect of adding O2 on the etch rate and etch selectivity behaviors is needed. 4.

Conclusion

The etching characteristics of TaN(300 nm)/HfO2 (80 nm) gate stack structures on an Si substrate were investigated in Cl2 /Ar, Cl2 /SF6 /Ar, and Cl2 /SF6 /O2 /Ar plasmas, in an ICP etcher by varying the process parameters, viz. the etch gas mixing ratios, additive O2 flow rate, the top electrode power, and the bottom bias electrode power. The results obtained for the etch rates of the TaN and HfO2 layers and etch selectivity in Cl2 /Ar chemistry under the various process conditions showed that this etching system afforded low etch selectivity, due to the HfO2 etch rate being similar to the TaN etch rate. The high etch rate of the HfO2 layer in the Cl2 /Ar plasmas limited the etch selectivity to  ¼ 1{ 4:6. To improve the etch selectivity, etch experiments with additive SF6 flow were performed, by varying the Cl2 /SF6 / Ar gas mixing ratio and the self-bias voltage, Vdc . X-ray photoelectron spectroscopy (XPS) was used to investigate the chemical binding states of the etched surfaces with the Cl2 /SF6 /Ar chemistry. The combined results obtained from the etch rate measurements and XPS analyses showed that the etch rates of the TaN and HfO2 layers in the Cl2 /SF6 /Ar plasmas were significantly reduced, presumably due to the reduced chlorination on the surface at SF6 percentages 20% compared to those in the Cl2 /Ar plasmas. However, the etch selectivity was increased at reduced Vdc values and an SF6 flow percentage of 20%, due to the disproportionate changes in the etch rates of the TaN and HfO2 layers. An additional O2 flow was added to the Cl2 (60 sccm)/ SF6 (20 sccm)/Ar(20 sccm) plasma to enhance the etch selectivity. At certain O2 flow rate regimes, 5 sccm in this work, the TaN etch rate was increased, while the HfO2 etch rate remained almost constant, which resulted in the etch selectivity being increased up to  ¼11:6. Since the etch characteristics of HfO2 etching follow an ion-assisted chemical etching mechanism, due to the inherent low volatilities of the etch by-products, as in the case of metal gate electrode materials including TaN and TiN layers, only limited etch selectivity was obtained. Since for integration of the metal gate/high-k stack etching into the device the etch selectivity of the metal gate to the high-k gate dielectric layer of at least 30 – 50 would be required,29) the effect of various etch gases and additive gases on the etch rates of the metal gate electrodes and high-k dielectrics needs to be further investigated.

M. H. SHIN et al.

5817

Acknowledgment This work was supported through the Center of Excellency Program of the Korea Science and Engineering Foundation & the Ministry of Science and Technology (Grant No. R-11-2000-086-0000-0) and by the Ministry of Commerce, Industry and Energy. S.W. Na was also supported by the BK21 program funded by the Korea Research Foundation.

1) G. D. Wilk, R. M. Wallace and J. M. Anthony: J. Appl. Phys. 89 (2001) 5243. 2) R. M. Wallace: Appl. Surf. Sci. 231– 232 (2004) 543. 3) S. J. Lee, C. H. Choi, A. Kamath, R. Clark and D. L. Kwong: IEEE Electron Device Lett. 24 (2003) 105. 4) A. Vandooren, A. Barr, L. Mathew, T. R. White, S. Egley, D. Pham, M. Zavala, S. Samavedam, J. Schaeffer, J. Conner, B.-Y. Nguyen, B.-E. Jr. White, M. K. Orlowski and J. Mogab: IEEE Electron Device Lett. 24 (2003) 342. 5) C. Ren, C. H. Y. Yu, J. F. Kang, X. P. Wang, H. H. H. Ma, Y. YeeChia, D. S. H. Chan, M.-F. Li and D.-L. Kwong: IEEE Electron Device Lett. 25 (2004) 580. 6) A. Vandooren et al.: IEEE Trans. Nanotechnol. 2 (2003) 324. 7) C. H. Choi, T. S. Jeon, R. Clark and D.-L. Kwong: IEEE Electron Device Lett. 24 (2003) 215. 8) C. S. Kang, H.-Ju Cho, R. Choi, Y.-H. Kim, C. Y. Kang, Se J. Rhee, C. Choi, M. S. Akbar and J. C. Lee: IEEE Trans. Electron Devices 51 (2004) 220. 9) S. H. Bae, C. H. Lee, R. Clark and D. L. Kwong: IEEE Electron Device Lett. 24 (2003) 556. 10) T. Iwamoto et al.: Tech. Dig. Int. Electron Devices Meet. 2003, p. 27.5.1. 11) A. L. P. Rotondaro, M. R. Visokay, V. A. Shanware, J. J. Chambers and L. Colombo: IEEE Electron Device Lett. 23 (2002) 603. 12) X. Wang, J. Liu, F. Zhu, N. Yamada and D. L. Kwong: to be published in IEEE Trans. Electron Devices. 13) R. E. Nieh, C. S. Kang, H.-Ju Cho, K. Onishi, R. Choi, S. Krishnan, J. H. Han, Y.-H. Kim, M. S. Akbar and J. C. Lee: IEEE Trans. Electron Devices 50 (2003) 333. 14) K.-J. Choi and S. G. Yoon: Electrochem. Solid-State Lett. 7 (2004) G47. 15) J. K. Schaeffer et al.: J. Vac. Sci. & Technol. B 21 (2003) 11. 16) J. Westlinder, T. Schram, L. Pantisano, E. Cartier, A. Kerber, G. S. Lujan, J. Olsson and G. Groeseneken: IEEE Electron Device Lett. 24 (2003) 550. 17) H. Yu Yu, M.-F. Li and D.-L. Kwong: IEEE Trans. Electron Devices 51 (2004) 609. 18) P. Alen, T. Aaltonen, M. Ritala, M. Leskela, T. Sajavaara, J. Keinonen, J. C. Hooker and J. W. Maes: J. Electrochem. Soc. 151 (2004) G523. 19) M. A. Quevedo-Lopez, M. El-Bouanani, R. M. Wallace and B. E. Gnade: J. Vac. Sci. & Technol. A 20 (2002) 189. 20) J. Chen, W. J. Yoo, D. S. H. Chan and D.-L. Kwong: Electrochem. Solid-State Lett. 7 (2004) F18. 21) S. Norassetthekul, P. Y. Park, K. H. Baik, K. P. Lee, J. H. Shin, B. S. Jeong, V. Shishodia, D. P. Norton and S. J. Pearton: Appl. Surf. Sci. 187 (2002) 75. 22) J. A. Britten, H. T. Nguyen, S. F. Falabella, B. W. Shore, M. D. Perry and D. H. Raguin: J. Vac. Sci. & Technol. A 14 (1996) 2973. 23) K. Pelhos, V. M. Donnelly, A. Kornblit, M. L. Green, R. B. Van Dover, L. Manchanda, Y. Hu, M. Morris and E. Bower: J. Vac. Sci & Technol. A 19 (2001) 1361. 24) L. Sha, B.-O. Cho and J. P. Chang: J. Vac. Sci. & Technol. A 20 (2002) 1525. 25) L. Sha, R. Puthenkovilakam, Y.-S. Lin and J. P. Chang: J. Vac. Sci. & Technol. B 21 (2003) 2420. 26) L. Sha and J. P. Chang: J. Vac. Sci. & Technol. A 21 (2003) 1915. 27) L. Sha and J. P. Chang: J. Vac. Sci. & Technol. A 22 (2004) 88. 28) J. Chen, K. M. Tan, Nan Wu, W. J. Yoo and D. S. H. Chan: J. Vac. Sci. & Technol. A 21 (2003) 1210.

5818

Jpn. J. Appl. Phys., Vol. 44, No. 7B (2005)

29) J. Chen, W. J. Yoo, Z. Y. L. Tan, Y. Wang and D. S. H. Chan: J. Vac. Sci. & Technol. A 22 (2004) 1552. 30) Y. Iba, F. Kumasaka, H. Aoyama, T. Taguchi and M. Yamabe: Jpn. J. Appl. Phys. 37 (1998) L824. 31) T. Tsuchizawa, C. Takahashi, M. Shimada, S. Uchiyama, T. Ono and M. Oda: Microelectron. Eng. 53 (2000) 595. 32) S. G. Woo, S. H. Kim, S. Y. Ju, J. H. Son and J. Ahn: Jpn. J. Appl. Phys. 39 (2000) 6996. 33) T. Tsuchizawa, H. Iriguchi, C.u Takahashi, M. Shimada, S. Uchiyama and M. Oda: Jpn. J. Appl. Phys. 39 (2000) 6914. 34) K.-P. Lee, K. B. Jung, R. K. Singh, S. J. Pearton, C. Hobbs and

M. H. S HIN et al. P. Tobin: J. Electrochem. Soc. 146 (1999) 3794. 35) H. Shimada and K. Maruyama: Jpn. J. Appl. Phys. 43 (2004) 1768. 36) M. H. Shin, S. W. Na, N.-E. Lee and J. H. Ahn: to be published in Thin Solid Films. 37) A. J. van Roosmalen, J. A. G. Baggerman and S. J. H. Brader: Dry etching for VLSI (Plenum, New York, 1991) p. 121. 38) C. Streinbruchel: Appl. Phys. Lett. 55 (1989) 1960. 39) C. Streinbruchel: J. Vac. Sci. & Technol. A 8 (1990) 1663. 40) NIST, X-ray Photoelectron Spectroscopy Database, web version. 41) B. W. Smith, C. Fonseca, L. Zavyalova, Z. Alam and A. Bourov: J. Vac. Sci. & Technol. B 15 (1997) 2259.