Comparative Studies of Perfluorocarbon Alternative Gas Plasmas for Contact Hole Etch

Jpn. J. Appl. Phys. Vol. 42 (23) pp. 5759 5764 Part 1, No. 9A, September 23 #23 The Japan Society of Applied Physics Comparative Studies of Perfluorocarbon Alternative Gas Plasmas for Contact Hole Etch Shingo NAKAMURA, Mitsushi ITANO, Hirokazu AOYAMA, KentaroSHIBAHARA 1, Shin YOKOYAMA 1 and Masataka HIROSE 2 Chemical Division, DAIKIN Industries, Ltd., 1-1 Nishi Hitotsuya, Settsu, Osaka 566-8585, Japan 1 Research Center for Nanodevices and Systems, Hiroshima University, 1-4-2, Kagamiyama, Higashi-hiroshima, Hiroshima 739-8527, Japan 2 Advanced Semiconductor Research Center, National Institute of Advanced Industrial Science and Technology, Central 4, 1-1-1 Higashi, Tsukuba, Ibaraki 35-8562, Japan (Received December 27, 22; accepted for publication May 14, 23) Saturated perfluorocarbons (PFCs) such as CF 4,C 2 F 6,C 3 and c- are used as dry-etch gases in the semiconductor industry. They have a significant greenhouse effect. Unsaturated fluorocarbons can be alternated with these PFCs because of their easy decomposition in the atmosphere. The authors have diagnosed the plasmas generated from straight-chain unsaturated gases such as,, and in an inductively coupled plasma reactor and have compared their etch properties. It was found that high selectivity has been obtained in a or plasma without mixing any specific gases. Fine contact holes of approximately 1 nm in diameter also have been obtained using or with or without adding Ar or O 2. These good etch properties of and have been achieved as a consequence of the appropriate balance between the lower density of fluorocarbon polymers and the dominant etching species CF þ with lower etching efficiency. It can be concluded that and can be used as PFC replacements for the dry-etch gas. [DOI: 1.1143/JJAP.42.5759] KEYWORDS: greenhouse effect, PFC alternative gas, straight-chain unsaturated fluorocarbon, inductively coupled plasma (ICP), etching, contact hole etch 1. Introduction Saturated perfluorocarbons (PFCs) such as CF 4, C 2 F 6, C 3 and c- are used as dry-etch gases in the fabrication of ultralarge-scale integrated circuits (ULSIs) in the semiconductor industry. These gases have long atmosphere lifetimes and strong absorptions of infrared radiation, exhibiting high global warming potentials (GWPs) and so promoting the green house effect. Recently, the replacement, decomposition or recycling of PFCs used for dry-etch gases has been attempted to reduce the greenhouse effect. 1) PFC such as octafluorocyclopropane (c- ) are still widely used to etch interlayer dielectrics and fabricate fine contact holes in ULSIs, while it will be difficult to achieve a contact hole etch with a high aspect ratio in the sub-65 nm range. 2) Table I summarizes the GWP 1 and atmospheric lifetime of PFC alternative candidates together with c-. The GWP 1 is defined as a relative GWP value calculated for the period of one hundred years against CO 2. 3,4) It should be noted that and have much lower GWP 1 values and atmospheric lifetimes than c-. These alternative candidates with a double bond in their molecules are subject Table I. Molecular structure, global warming potential (GWP 1 ) and atmospheric lifetime for octafluorocyclopropane(c- ), hexafluoropropene ( ), hexafluorobutadiene ( ), octafluoro-2-butene ( ) and octafluoropentadiene ( ). Molecular Structure aþ GWP 1 Atmospheric lifetime (years) c- F 2 C- j j 1 32 F 2 C- CF=CF CF= 1 =CFCF= 29.3 CF=CFCF= a) GWP 1 (global warming potential for one hundred years) to decomposition in the atmosphere through their reaction with hydroxyl radicals. This fragmentation mechanism leads to the much shorter atmospheric lifetime and much lower effective GWP 1 for the replacement candidates. 5) In addition, these alternative candidates have the potential of good etch properties because the double bond is broken selectively and the ratio of particular radicals and ions is predominantly controlled. Thus it is important to investigate the PFC alternative etch gases such as hexafluoropropene (, CF= ), hexafluorobutadiene (, = CFCF= ), octafluoro-2-butene (, CF=CF ) and octafluoropentadiene (, CF=CFCF= ). These are straight-chain unsaturated compounds with the double bond in the molecules. In the present study, we have diagnosed these gas plasmas and examined the etching ability to evaluate its potential as an alternative etching gas. 2. Experimental The schematic diagram of an inductively coupled plasma (ICP) reactor used in the present study is shown in Fig. 1. Etching gases were introduced from the outlets set at eight points on an inner wall. A single-turn antenna of 14 mm in diameter generates plasmas through a 9-mm-thick quart plate. A 2 inch wafer was clamped using an electrostatic chuck (ESC) assembly on a chiller-cooled stage kept at 11 C. A quadrupole mass spectrometer and a Langmuir probe were set on a chamber sidewall to diagnose the plasmas. The Langmuir probe used to measure electron densities and electron temperatures was set 2 mm above the center of the wafer. The probe tip was heated to prevent fluorocarbon polymer deposition. The relative amount of positive ions was evaluated by quadrupole mass spectrometry, and the density of CF x (x ¼ 1{3) fluorocarbon radicals was measured by appearance mass spectrometry (AMS). 6) The morphology of fluorocarbon polymers deposited on the wafer was measured by atomic force microscopy (AFM). Chemical bonding features were also analyzed by Fourier transform infrared (FT-IR) spectroscopy. A 2-mm-thick 5759

576 Jpn. J. Appl. Phys. Vol. 42 (23) Pt. 1, No. 9A S. NAKAMURA et al. ngle Turn Antenna Matching Box 13.56MHz ~ Electron Temperature (ev) 2 4 6 8 1 Quartz Plate QMS Electrostatic Chuck Inner Wall 1mm 1mm Matching Box 4kHz 2mm layer deposited by atmospheric pressure chemical vapor deposition (APCVD) was used as the substrate. The positive chemical amplification electron-beam resist used mainly consisted of highly sensitive novolak resin, which was manufactured for the trial by Hitachi Chemical Co, Ltd.. The electron beam resist with a thickness of 95 nm was patterned to form contact holes with 18 nm 4.5 mm diameters. The layer was etched with,,,c- and plasmas under almost the same etching conditions. Typical conditions were as follows: a pressure of 3 mtorr, a gas flow rate of 5 7 sccm, an ICP power of 6 W and a bias power of 2 W. 3. Results and Discussion 3.1 Plasma diagnostics The electron density and temperature of the c-,,, and plasmas are compared in Fig. 2. The electron temperatures of and are slightly higher than those of, and c- while the electron densities of and are almost the same as those of and c-. In addition, the electron density of is the highest in these plasmas. The amount of positive ion components in each plasma is shown in Fig. 3. A CF þ ion with low etching ability is the main ionic species and a CF þ 2 is a minor species in every plasma. It is suggested for and that a lot of CF þ 3 ions with high etching ability are efficiently produced from the CF fragment generated by the break of the double bond. This is consistent with the fact that the fraction of the CF þ 3 ion in the C 5 plasma is higher than in the plasma. A large amount of F þ ions in the and plasmas are attributed to high electron temperature, as shown in Fig. 2. Fluorocarbon plasma with high electron temperature in general tends to contain highdensity F þ ions and F radicals. nce the CF þ 2 /CFþ ratio is about.2 and is nearly equal for these plasmas, the CF þ 3 / CF þ ratio can be utilized as an index of etching efficiency. ~ Cb 13mm 2mm 2 Inch Wafer Langmuir Probe Fig. 1. Schematic diagram of an inductively coupled plasma(icp) reactor. Frequencies of source power and bias power are 13.56 MHz and 4 khz, respectively. The bottom of the inner wall (2 mm in diameter) is located at a distance of 13 mm from the quartz plate. The gap from the single-turn antenna of the ICP to the electrostatic chuck stage is 1 mm. Heating was not used in this study while the cylindrical inner wall with the heating function is set in an etching chamber. c- c-.87 5.2 Electron 5.1 Temperature 5.3 6.6 6..97 1.7 1.8 Electron Density 1.45 1 2 Electron Density ( 1 11 cm -3 ) Fig. 2. Electron temperature and electron density at 3 mtorr and 6 W source power. F C CF F C CF F C CF F C CF F C CF CF3 CF3 F F CF CF c- - CF CF2- C 3 F6 ( CF= ) CF CF ( CF=CF ) (CF2=CFCF= ) ( CF=CFCF= ) 2 4 6 8 Ion Count (%) Fig. 3. Relative amount of positive ions at 3 mtorr and 6 W source power. These ions were detected by quadrupole mass spectrometry. The CF þ 3 /CFþ ratio in each plasma is in the following order: (.81) > (.68) > c- (.33) > (.3) > (.1). Figure 4 shows the total CF x (x ¼ 1{3) radical densities of these plasmas. The total densities of CF x (x ¼ 1{3) radicals are highest in the c- plasma and lowest in the plasma. The density of the radical is the highest of all these plasmas and that of the CF radical is the lowest. Figure 5 shows the average roughness Ra measured by AFM and the deposition rate r depo of the fluorocarbon polymer films deposited on a substrate with each plasma without RF bias power. The observations in Fig. 5 show that the larger molecule of parent gas results in a higher deposition rate and a rougher surface of the polymer, except for. This implies that the polymeric radicals are

Jpn. J. Appl. Phys. Vol. 42 (23) Pt. 1, No. 9A S. NAKAMURA et al. 5761 CF c- - I I - CF= CF=CF =CFCF= CF=CFCF=.1 2.1 4.1 6.1 8.1 Total Radical Density ( 1 12 cm -3 ) Normalized Absorbance (arb.units) 3 2 1 2 v( ) 136-117cm -1 c- 1 1 Wave Number (cm -1 ) v(- -) 125-15cm -1 v(c-f) 14-1cm -1 Fig. 4. CF x (x ¼ 1{3) radical densities at 3 mtorr and 6 W source power. These radical densities were measured by appearance mass spectrometry (AMS). Fig. 6. Fourier transform infrared (FT-IR) spectroscopy of fluorocarbon polymer films. These films were deposited at 3 mtorr, 6 W source power and electrostatic chuck temperature of 11 C. The absorbance normalized by a film thickness indicates the density of the fluorocarbon polymer films. v( ), v(c F) and v( ) show the vibration absorption range. 5 1 1 c- 2 4 F 6 Ra(nm) r depo (nm/min).8 285 1.4 243 c- 1.8 259 2.8 262 6.1 317 these candidates (,,, ) and the conventional etching gas (c- ) are summarized in Fig. 7. Figure 8 shows the dissociation and etching model of these fluorocarbon gases in the plasma. The total CF x (x ¼ 1{3) radical density of the c- plasma is the highest, however the c- plasma deposits a lower-density fluorocarbonpolymer film than the and plasmas do. The fluorocarbon polymer surface of the c- plasma was rougher than those for the and plasma as shown in Fig. 5. These facts indicate that polymeric radicals such as ac x F y (x = 3, y = 5) deposit the rough and porous films of fluorocarbon polymers in the c- plasma while a large number of radicals rearranged from a CF fragment in the and plasmas deposit high-density fluorocarbon polymers in a similar manner to the radical, as illustrated in Fig. 8. The plasma had the Fig. 5. The surface roughness and deposition rate of fluorocarbon polymer films. The fluorocarbon polymer films were deposited at 3 mtorr, 6 W source power and electrostatic chuck temperature of 11 C. Ra is average roughness measured by AFM and r depo is the deposition rate of the fluorocarbon polymer films. the origins of surface roughness. Figure 6 shows the FT-IR spectra of the fluorocarbon polymer films deposited for 1.5 min on substrates in each plasma without RF bias power. Absorbance is normalized by the fluorocarbonpolymer film thickness measured by scanning electron microscopy (SEM). The absorbances for the, and c- plasmas are higher than those for the and plasmas, which indicates that, and c- plasmas deposit higher-density fluorocarbon-polymer films. This fact is consistent with the results in Fig. 5. A rough surface generally indicates a porous structure. It is also noteworthy that both and with the CF fragment deposit high-density fluorocarbon polymers. The results obtained by a series of plasma diagnostics of Density of Fluorocarbon Polymer (arb.units) 3. 2.5 2. 1.5 ( CF=CF ) ( CF= ) c- ( =CFCF= ) :Total CFx(x=1-3) Radical Density ( CF=CFCF= ) 2 4 6 8 Surface Roughness Ra (nm) Fig. 7. A summary of fluorocarbon polymer properties and radical densities. X-axis indicates a surface roughness Ra shown in Fig. 5. Y axis indicates the fluorocarbon polymer densities estimated by the normalized FT-IR spectrum shown in Fig. 6. Diameters for each circle correspond to the total CF x (x ¼ 1 3) radical densities shown in Fig. 4.

5762 Jpn. J. Appl. Phys. Vol. 42 (23) Pt. 1, No. 9A S. NAKAMURA et al. Selectivity 2 4 6 8 1 3 3 c- =CFCF= CF CF= 2 I I CF=CFCF= CF: CF=CFCF:. CF2 CF. C x F y (x 3,y 5) 2 C x F y (x 2,y ) CF x (x=1-3) :CFCF= plasma plasma CF CF 3 CF 2 CF CF CF CO C CF x,f x, CF3 CO x,f x, x F y CF COF x COF x - - - - - C x F y - C x F y - C x F y - C x F y O z reactive layer C x F y O z reactive layer c- c- c- 2.47 2.68 2.29 1.34 1.37 1.34 2.13 1.93 4.89 Selectivity( /) 6.11 Selectivity( /) 678 Etch Rtae 654 643 57 588 2 4 6 8 1 Etch Rate (nm/min) Fig. 9. etch rate and selectivity at 3 mtorr, 6 W source power and 2 W bias power. Main Etching Species : CF, Etching Efficiency : High Fluorocarbon polymer : High Density and Fine Surface Main Etching Species : CF Etching Efficiency : Low Fluorocarbon polymer : Low Density and Rough Surface Fig. 8. The image of dissociation and etching in the unsaturated fluorocarbon gas plasmas. CF þ 3 ions with high etching ability impinge on the high-density film of fluorocarbon polymers in the and plasmas. On the other hand, CF þ ions with low etching ability bombard the porous and low-density films of fluorocarbon polymer in the and plasmas. Etch reaction products such as CO 2, COF 2 and F 4 are released from the x O y F z reactive layer. lowest CF x (x ¼ 1{3) radical densities, and its fluorocarbonpolymer density was the lowest. The deposition rate of fluorocarbon polymers for the plasma is the highest, and the surface is the roughest. Hence, it is probable that more polymeric radical species other than CF x (x ¼ 1{3) radicals are the main precursor for these two plasmas. We speculate that the polymeric radicals are the large fragment ones which has another double bond stabilized by the break of one of the double bonds as illustrated in Fig. 8. Etching efficiency was also explained as illustrated in Fig. 8 based on the plasma diagnostics. The relative etching efficiency defined as the CF þ 3 /CFþ ratio is in the order > > c- > >, as described before. The plasma has the highest etching efficiency because of its high ion density due to a high electron density. In the and plasmas, CF þ 3 ions impinge on the high-density fluorocarbon-polymer film. On the other hand, in the and plasmas, CF þ ions with low etching ability bombard the porous and low-density films of fluorocarbon polymers. 3.2 Etch properties Figure 9 shows the etch rate and selectivity for the etch plasmas. The etch rates for the, and c- plasmas are slightly higher than those for the and plasmas. The selectivity against resist or is smaller in the, and c- plasmas than in the and Normalized Etch Rate 1.2 1.1 1..9.8.7 c-.6.1 1. 1. Contact Hole ze (µm) Fig. 1. Microloading effect at 3 mtorr, 6 W source power and 2 W bias power. plasmas. Figure 1 shows the microloading effect of contact holes in these plasmas. The etch rates of the contact holes are normalized by that of a 4.5-mm-diameter hole. In the case of plasma, the normalized etch rate for a 18 nm hole is about.8 and the microloading effect is not so severe. The normalized etch rate for the holes ranging from nm to 2.5 mm is larger than 1. except for the plasma. This is a reverse tendency of the ordinary microloading effect. As shown in Fig. 11, the etched holes have tapered shape except for in the case of the plasma. The tapered shape gives rise to an ion-focusing effect at the bottom. This ion-focusing effect and high-density fluorocarbon polymers are the origins of the reverse-microloading effect in the plasmas except. As shown in Fig. 11, in the case of the c- and plasmas, the widening at the top and narrowing at the bottom of the contact holes are significant. This is explained by the widening of the resist in the shape of a facet because of the low etch selectivity for resist of. The nearly vertical etching profile is obtained with the plasma. In this case, the etch rate is 57 nm/min and the selectivity against resist and are 2.1 and 6.1, respectively. The diameter and depth of the hole,

Jpn. J. Appl. Phys. Vol. 42 (23) Pt. 1, No. 9A S. NAKAMURA et al. 5763 c- 4mTorr, ICP4W, Bias15W / O 2 (1 %), ICP4W, Bias2W (a) φ max.49µm, D1.4µm (a) φ d.18µm ( φ max.12µm, D.95µm) (b) φ d.2µm ( φ max.1µm, D.95µm) F 6 / Ar(25 %), ICP4W, Bias2W C 5 / Ar(5%) 7mTorr, ICP6W, Bias2W (b) φ max.55µm, D1.8µm (c) φ max.43µm, D1.4µm F 6 C 5 (d) φ max.2µm, D1.32µm (c) φ d.2µm ( φ max.13µm, D 1.1) (d) φ d.17µm ( φ max.13µm, D1.6µm) (f) φ max.22µm, D1. Fig. 12. Etch profiles of the fine contact holes formed with, / O 2 (1%), /Ar (25%) and /Ar (5%) plasmas. d, max and D represent designed diameter, etched maximum diameter and etched depth of the contact holes, respectively. Fig. 11. Etch profiles of.2 mm contact holes at 3 mtorr, 6 W source power and 2 W bias power. max and D represent the maximum diameter and depth of etched contact holes, respectively. shown in Fig. 11(d) for, are 2 nm and 1.32 mm, respectively. The fluorocarbon polymer film deposited in the plasma has a rough surface and the lowest density as described previously. Hence, it seems that etch reaction products are released easily and the etch rate is consequently in an acceptable range. The good etch profile is due to the appropriate balance between the high-density CF þ ions with a low etching efficiency and the low-density fluorocarbon polymers. In the plasma, the etch profile exhibits a slight side etching since the ions reflected at the tapered resist are more than in the plasma. The microloading effect is also higher than in the plasma because the CF x (x ¼ 1{3) radical densities and the film density of fluorocarbon polymers in the plasma are higher than in the plasma. In the plasma, the etch profile is not shrunk at the bottom of the contact holes. This can be attributed to the higher electron density and the lower ion-focusing effect of the plasma than those of the c- and plasmas. 3.3 Fine contact hole The Ar gas dilutes the fluorocarbon gas and the O 2 gas decomposes the polymeric radical to depress excess deposition. These gases generally prevent etch stop. The contact holes of about 1 nm in diameter, which are smaller than the mask resist patterns, are obtained in the, or a mixture plasma with Ar or O 2. The hole-size reduction was realized by the local deposition of fluorocarbon polymer around the resist opening by increasing the pressure of the etching atmosphere from 3 mtorr to 4 7 mtorr. Figure 12 shows the SEM crosssection of the fine contact holes etched with, /O 2 (1%), /Ar (25%) and /Ar (5%) plasmas. Figure 13 shows the positive ion content in the, /O 2 (1%) and /Ar (25%) plasmas. The CF þ is the dominant etching species and the others are less significant in these plasmas. Figure 14 shows CF x (x ¼ 1{3) radical densities in the /O 2 (1%) and /Ar (25%) plasmas. Certain radicals do not contribute to etching the fine contact hole in both plasmas. It is possible that both the polymeric radicals except CF x (x ¼ 1{3) and the CF þ ion play an important role in etching the fine holes in the, /O 2 (1%) and /Ar (25%) plasmas. /Ar (25%) ICP4W,Bias2W /O2 (1%) ICP4W,Bias2W 4mTorr ICP4W,Bias15W Ar CF Ar CF CF CF C F O C C % 2% 4% 6% 8% 1% Ion Count (%) Fig. 13. The positive ion content in the, /O 2 (1%) and / Ar (25%) plasmas. F

5764 Jpn. J. Appl. Phys. Vol. 42 (23) Pt. 1, No. 9A S. NAKAMURA et al. /O 2 (1%) ICP4W Bias2W /Ar (25%) ICP4W Bias2W CF 1 CF 1 result of the plasma diagnostics and the etch properties have shown that the and gas can be used as the PFC alternatives, which can realize acceptable etch profiles for contact holes. A fine hole of approximately.1 mm in diameter can be formed in the, and the mixture gas plasma with Ar or O 2. These results are obtained as a consequence of the good balance between the low-density film of fluorocarbon polymers and the many CF þ ions with low etching efficiency. The straight-chain can also be a promising dry-etch gas if etch conditions such as bias voltage are optimized. 1 1 1 11 1 12 1 13 Radical Density (cm -3 ) Fig. 14. CF x (x ¼ 1{3) radical densities in the /O 2 (1%) and / Ar (25%) plasmas. 4. Conclusion The straight-chain unsaturated fluorocarbon compounds were compared with the conventional c- gas to evaluate them as PFC alternative candidates for dry-etch gases. The 1) Y. Matsushita: Oyo Buturi 69 (2) 35 [in Japanese]. 2) T. Mukai and S. Samukawa: Proc. Symp. Dry Process, Tokyo, 1999, p. 39. 3) Intergovernmental Panel on Climate Change: Climate Change 21: The Scientific Basis, eds. J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell and C. A. Johnson (Cambridge University Press, 21) Chap. 6, p. 3894. 4) SEMI 1997 PFC Forum (1997) p. 1. 5) R. Imasu, A. Suga and T. Matsuno: J. Meteo. Soc. Jpn. 73 (1995) 1123. 6) Y. Hikosaka, H. Toyoda and H. Sugai: Jpn. J. Appl. Phys. 32 (1993) L353.