- Email: [email protected]
Plasma pretreatment of the Cu seed layer surface in Cu electroplating
Materials Chemistry and Physics 73 (2002) 227–234
Plasma pretreatment of the Cu seed layer surface in Cu electroplating Junhwan Oh a , Jaegab Lee b , Chongmu Lee a,∗ a
Department of Metallurgical Engineering, Inha University, 253 Yonghyeondong, Inchon 402-751, South Korea b Department of Metallurgical and Materials Engineering, Kookmin University, Seoul 136-702, South Korea Received 8 December 2000; received in revised form 28 April 2001; accepted 2 May 2001
Abstract Effects of plasma pretreatment to the Cu seed layer on copper (Cu) electroplating were investigated. Copper seed layers were deposited by magnetron sputtering onto tantalum nitride barrier layers before electroplating copper in the forward pulsed mode. The Cu seed layer was cleaned by plasma H2 or N2 prior to electroplating a copper film. Cu films electroplated on the copper seed layer with plasma pretreatment have shown better electrical and physical properties such as electrical resistivities, surface morphologies, levels of impurities, adhesion and surface roughness than those without plasma pretreatment. It is shown that carbon and metal oxide contaminants at the sputtered Cu seed/TaN surface can be effectively removed by plasma H2 cleaning. The degree of the (1 1 1) preferred orientation of the pulsed plated Cu film with plasma H2 pretreatment is as high as that without plasma pretreatment. Also, plasma H2 precleaning is more effective in enhancing the Cu electroplating properties onto the Cu seed layer than plasma N2 precleaning. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Plasma pretreatment; Electroplating; Seed layer; Copper
1. Introduction Copper has drawn much attention as a new interconnect material for deep submicron ultra-large scale integrated circuits (ULSIs), because of lower resistivity (1.7 cm) and higher resistance to electromigration compared with aluminum and its alloys [1–4]. Cu films can be obtained by several different deposition techniques including physical vapor deposition (PVD), chemical vapor deposition (CVD), and plating such as electroless plating and electroplating. Especially, electroplating is a very inexpensive process in principle and offers high deposition rate and a number of research groups have successfully used it to fill damascene structures. Therefore, the most popular technique today includes Cu electroplating onto a trench or a via [5]. Electroplating is an attractive alternative deposition method for copper with the need for a conformal and conductive seed layer [6]. The seed layer provides a low resistance conductor for the plating current that drives the electrodeposition process and also facilitates nucleation of the plated copper film. In addition, the Cu seed layer should be highly pure so as not to compromise the effective resistivity of the filled copper interconnects structure. This seed layer requires low electrical resistivity, low levels of impu∗ Corresponding author. Fax: +82-32-862-5546. E-mail address: [email protected] (C. Lee).
rities, smooth interface, good adhesion to the barrier metal and low thickness concurrent with coherence for ensuring void-free fill [7]. The electrical conductivity of the surface plays an important role in the formation of initial Cu nuclei and the surfaces with higher electrical conductivities shows much better formation of Cu nuclei than those with lower electrical conductivities. It is also known that the nucleation processes of Cu are very sensitive to surface condition [8]. Pulse plating has been around the industry for a very long time. The primary attraction was its control of the cost normally associated with precious metal plating. This was achieved through grain refinement, improved porosity and minimization of overplating. Also, pulse plating techniques reduced the formation of voids because the rate of metal deposition deep inside a trench becomes nearly the same as the rate at the upper portion. When Cu is deposited on the Cu seed/TaN substrate by electroplating, precleaning of substrate surface like plasma treatment is essential. Plasma cleaning, employing hydrogen as reactive gas, has been used for many years. The idea behind the utilization of the plasma is the creation of a very reactive gas environment, often enclosed in a vacuum system [9]. Hydrogen discharge cleaning of vacuum vessels for particle accelerators and magnetic fusion devices is a well-known technique in the field of nuclear and particle physics for the removal of light impurities from metallic materials. Hydrogen plasma can be used to remove most
0254-0584/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 2 5 4 - 0 5 8 4 ( 0 1 ) 0 0 3 7 8 - 9
228
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
materials provided that volatile hydrides are formed at the operating temperature. The plasma pretreatment was used to remove oxides and etch residues that contaminate the metal surface at the bottom of a via. In this communication, we report the investigation of the effects of plasma pretreatment to Cu seed/tantalum nitride (TaN)/borophosphosilicate glass (BPSG) samples on electroplating of copper (Cu) films using four point probe measurement, X-ray diffractometry (XRD), scanning electron microscopy (SEM), atomic force microscopy (AFM) and X-ray photoelectron spectroscopy (XPS) focused on levels of impurities in the electroplated Cu film and surface chemical bonding state of the Cu seed layer and the electroplated Cu film.
2. Experimental Copper seed layers (500 Å) were deposited onto TaN barrier layer by magnetron sputtering using a Cu target in Ar gas ambient before electroplating of copper in the forward
pulsed mode. Prior to electroplating a copper film, the Cu seed layer surface was cleaned with hydrogen and nitrogen plasma treatments. A sputtered TaN film 500 Å. thick was used as a barrier and adhesion promoter. The samples were rinsed with DI water for 3 min, followed by filtered nitrogen gas drying before electroplating. Plasma treatments were conducted in a parallel-type cold wall reactor of a plasma enhancement chemical vapor deposition (PECVD) system after pumping down to the base pressure of 10−5 Torr. Plasma exposure time during the plasma treatment was varied from 1 to 20 min and rf-power was from 20 to 140 W and the flow rate and the deposition temperature of plasma pretreatment were 100 sccm and 298 K, respectively. After plasma cleaning, Cu electroplating was conducted onto each substrate. The electroplating system mainly consists of a copper plating solution (a sulfuric acid–copper sulfate chemical solution without organic additives), a soluble anode, a wafer (which serves as a cathode) and a duty power supply. Forward pulsed plating was performed to improve via/trench filling capability. Then copper was electroplated in the copper plating solution onto the sub-
Fig. 1. SEM micrographs of copper films electroplated in a copper sulfate/sulfuric acid bath: (a) DC plating; (b) DC plating (vacuum annealing); (c) pulse plating at 60 mA cm−2 (on: 80 ms, off: 20 ms); (d) pulse plating at 60 mA cm−2 (on: 60 ms, off: 40 ms); (e) pulse plating at 60 mA cm−2 (on: 40 ms, off: 60 ms); (f) pulse plating at 60 mA cm−2 (on: 20 ms, off: 80 ms).
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
strate metal surface, which acts as the negative electrode of the electrochemical cell. Copper plating rate was in the range of 0.5–1 m min−1 and the current density was kept at 60 mA cm−2 . Effects of plasma pretreatments to Cu seed/TaN/BPSG samples on electroplating of copper (Cu) films were investigated. Atomic force microscopy (AFM, TopoMetrix ACCUREX) and scanning electron microscopy (SEM, Hitach S-4300) were used to examine the morphology of the substrate and the sputtered Cu seed. Resistivity of each sample was measured by the four-point probe measurement technique. The crystallographic orientations of Cu films were investigated by X-ray diffraction (XRD, Philips X’pert MPD). Levels of impurities and surface chemical bonding states of the Cu seed layer surfaces were investigated by X-ray photoelectron spectroscopy (XPS, SPESE). 3. Results and discussion Fig. 1 shows SEM micrographs of the copper film that was deposited by DC or pulsed electroplating in a sulfuric acid–cupric sulfate solution. The figures show that the pulse plating technique produces a relatively smooth fine-grained structure, while DC plating (see Fig. 1 (a) and (b)) offers rougher surface composed of large columnar grains. It should be noted that a fine-grained structure is obtained only when the pulse plating current densities have been used. The optimum pulse sequence is a complex function of the pulse current density, duty cycle, and polarity reversal. Resistivities of the pulse electroplated Cu at various off times are shown in Fig. 2. The pulse current (PC) plated Cu film shows lower resistivity than the direct current (DC) plated Cu film. Fig. 3 shows SEM micrographs of the samples, which have been given hydrogen and nitrogen plasma treatment at various plasma powers followed by Cu electroplating for 0.5–1 m. For nitrogen plasma treatment as the plasma power increases, the size of electroplated Cu particles also increases and the surface becomes very rougher. In the case
229
Fig. 2. Resistivities of Cu films electroplated at various on-off times.
of plasma H2 pretreatment the Cu grain size nearly does not change with increasing the rf-power. The Cu film surface became smoother with plasma H2 pretreatment. However, the smoothness was not improved with increasing the rf-power further beyond 140 W. As the rf-power increases higher than 140 W in the H2 plasma pretreatment, cracks and pores were found to form in the Cu film. Especially, these effects are strong for the plasma N2 pretreatment. The resistivities of the samples after Cu film deposition which have been pretreatmented with plasma H2 at the power of 20–140 W was as low as 1.9–2.1 cm. These may be due to the cause of the reduction of CuO and surface impurities. Effects of the plasma exposure time are also presented in Fig. 4. As the plasma exposure time increases, the Cu film surface is observed to get rougher. A distinct boundary between electroplating and sputtered Cu seed layers was not observed on SEM micrographs (Fig. 5), which indicates a homoepitaxial interface relation between the two. Fig. 6 shows the AFM micrographs of Cu seed layers on TaN given the plasma pretreatment at 100 W for 10 min. Figs. 7 and 8 show the RMS roughness of the
Fig. 3. SEM micrographs (×5000) of Cu films electroplated with plasma pretreatments at various rf-powers (exposure time: 10 min).
230
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
Fig. 4. SEM micrographs (×10,000) of Cu films electroplated with plasma pretreatment at various plasma exposure times (rf-power: 100 W).
Cu seed layer on TaN for various rf-powers and plasma exposure times in the plasma H2 pretreatment. The surface roughness tends to decrease with increasing rf-power, which implies that particulate contaminants on the seed layer surfaces have been removed by plasma treatment. Also, the surface roughness decreases rapidly with increasing the plasma exposure time. It is worth noting that the surface roughness decreases slowly in the case of nitrogen plasma cleaning, where as it decreases rapidly in the case of hydrogen plasma cleaning. Effects of H2 and N2 plasma pretreatments at 100 W for 10 min on the surface roughness of the electroplated Cu film are shown in AFM images (Fig. 9) and the surface of the Cu film electroplated on the Cu seed layer given the plasma H2 cleaning is smooth, where as that electroplated on the seed layer given plasma N2 cleaning is rough (root mean square (RMS) roughness = 14.75 nm with H2 plasma, 27.96 nm with N2 plasma). From this AFM observation it may be
Fig. 5. A cross-sectional SEM micrograph (×20,000) of the Cu film electroplated with the plasma H2 pretreatment (rf-power: 100 W, exposure time: 10 min).
said that the plasma H2 pretreatment offers the Cu film with smoother surface than the plasma N2 pretreatment. Figs. 10 and 11 present dependence of the surface roughness of the electroplated Cu on various plasma powers and plasma exposure times, respectively. The plasma H2 pretreatment at the power of 100 W for 10 min gives the lowest RMS roughness and resistivity. In this experiment, the plasma exposure time was fixed at 10 min. Also in the case of the plasma H2 pretreatment the surface roughness of the electroplated Cu film tends to decrease slowly, while in the case of the plasma N2 pretreatment it increases rapidly with increasing the rf-power of the plasma pretreatment. The plasma H2 pretreatment is favorable in obtaining smoother surface of the electroplating Cu film. Fig. 12 shows degrees of preferred orientation for three types of electroplated Cu films. The degree of the (1 1 1) preferred orientation of the Cu film with plasma H2 pretreatment is as high as the pulse plated Cu film without plasma pretreatment. In general Cu films with a stronger (1 1 1) texture have higher electromigration resistance. Sputtered Cu seed layers contain various kinds of impurities like F, O, C, N, S, Cl, etc. Plasma pretreatment minimizes contamination on the Cu seed layer/TaN sample. The C 1s peak (284.5 eV) and the O 1s peak (531.0 eV) in the XPS spectra of the Cu seed layer on the TaN substrate are shown in Figs. 13 and 14, respectively. The carbon and oxygen impurity concentrations in the top surface area of the Cu seed layer/TaN sample can be found in these figures. Fig. 15 shows the XPS of Cu 2p peak of the Cu seed layer on the TaN substrate. Cu (in CuO) 2p3/2 and Cu (in Cu2 O) 2p1/2 appeared at 933.6 and 953.5 eV, respectively. Figs. 13–15 were obtained after plasma H2 and N2 cleaning (100 W, 10 min). They show that the heights of the C, O and Cu–O peaks of the Cu seed layer/TaN sample are decreased by the plasma H2 pretreatment. Enrichment of oxygen and carbon was found at the interface of the electroplated Cu film and the Cu
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
231
Fig. 6. AFM micrographs of the Cu seed layer/TaN samples with (a) no pretreatment; (b) plasma H2 pretreatment; (c) plasma N2 pretreatment (rf-power: 100 W, plasma exposure time: 10 min).
Fig. 7. RMS roughnesses of the Cu seed/TaN film deposited with plasma H2 and N2 pretreatments at various rf-powers (plasma exposure time: 10 min).
Fig. 8. RMS roughnesses of the Cu seed/TaN film deposited with plasma H2 and N2 pretreatments at various plasma exposure times (rf-power: 100 W).
232
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
Fig. 9. AFM micrographs of the Cu films electroplated with (a) no pretreatment, (b) plasma H2 pretreatment and (c) plasma N2 pretreatment (rf-power: 100 W, plasma exposure time: 10 min).
Fig. 10. RMS roughnesses of the Cu films electroplated with plasma H2 and N2 pretreatments at various rf-powers (plasma exposure time: 10 min).
Fig. 11. RMS roughnesses of the Cu films electroplated with plasma H2 and N2 pretreatments at various plasma exposure times (rf-power: 100 W).
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
233
Fig. 14. The XPS O 1s peak (531.0 eV) of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time : 10 min). Fig. 12. XRD spectra for various types of electroplated Cu films.
seed/TaN/SiO2 /Si substrate. It can be due to the cause of poor wetting and peeling of the electroplated Cu film. Electroplated copper films with high purity were obtained. To understand the chemistry of removal of contaminants such as hydrocarbon and oxygen, CH4 and H2 O were added to the discharge in a separate experiment. From the result of this experiment we may draw a conclusion that a major mechanism for removing hydrocarbon is the formation of species with m/e = 15, 16, 17, 18 and 27, whereas the ion
Fig. 13. The XPS C 1s peak (284.5 eV) of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time: 10 min).
density of m/e = 19 is mainly related to the formation of H2 O. Cu–O + H2 → Cu + H2 O C + 2H2 → CH4 Oxygen atoms react with hydrogen atoms to form H2 O and some of carbon atoms in the Cu seed layer surface react with hydrogen atoms to form CH4 during the plasma H2 treatment. As a result of these two reactions only free Cu atoms remain since volatile H2 O and CH4 are evaporated. Hence, there are lots of free Cu atoms at the surface of the Cu seed layer/TaN sample after hydrogen plasma treatment. In contrast, little change occurs at the top surface of
Fig. 15. The XPS of Cu 2p peak of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time: 10 min).
234
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
the sample given a nitrogen treatment. Only a small number of free Cu atoms exist after the nitrogen plasma treatment. Therefore, the plasma H2 treatment is more effective in Cu nucleation and growth than the plasma N2 treatment, since the former forms free Cu atoms more effectively by removing carbon atoms and Cu-oxide at the top surface of the Cu seed layer/TaN sample.
cleaning procedure, resulting in substrate surfaces free of carbon and oxygen contaminants. The degree of the (1 1 1) preferred orientation of the Cu film with plasma H2 pretreatment is as high as pulse plated Cu film without plasma pretreatment. Plasma H2 precleaning of the Cu seed layer surface is more effective in enhancing Cu electroplating properties than plasma N2 precleaning.
4. Conclusions
Acknowledgements
Effects of plasma pretreatment to Cu seed/TaN/BPSG samples on the properties of the copper (Cu) film which is subsequently electroplated on the samples were studied and the following conclusions were obtained. The plasma H2 pretreatment offers a better condition for Cu electroplating than the plasma N2 treatment or no pretreatment. It is because only the plasma H2 treatment produces free Cu atoms by removing Cu oxide and carbon in the surface region. The plasma H2 is effective in removing carbon impurities and results in Cu-rich films and smooth morphologies. Our results show that Cu films electroplated on plasma pretreated copper seed layers give better electrical resistivities, surface morphologies, adhesion and low levels of impurities than those on copper seed layers without plasma pretreatment. The plasma treatment removes oxide, carbon and other residues from the substrate surface. XPS spectra show that volatile hydrogen compounds form during the
This work was supported by Grant No. 1999-1-301-002-5 from the Basic Research Program of the Korea Science and Engineering Foundation. References [1] J. Musil, A.J. Bell, M. Cepera, Czchoslovak, J. Phys. 45 (1995) 249. [2] J.C. Chiou, K.C. Juang, M.C. Chen, J. Electrochem. Soc. 142 (1995) 177. [3] V.M. Donnelly, M.E. Gross, J. Vac. Sci. Technol. A 11 (1993) 66. [4] J.B. Webb, D. Northcott, I. Emesh, Thin Solid Films 270 (1995) 483. [5] V. Dubin, S. Lopatin, R. Cheung, in: Electrochemical Society Proceedings, Vol. 98, 1998, p. 6. [6] R.L. Jackson, E. Broadbent, T. Cacouris, A. Harrus, M. Buberger, E. Patton, T. Walsh, Solid State Technol. 5 (1998) 49. [7] B. Chin, P. Ding, B. Sun, T. Chiang, D. Angelo, I. Hashim, Solid State Technol. 7 (1998) 141. [8] Y.S. Kim, D. Jung, S.K. Min, Thin Solid Films 349 (1999) 36. [9] N. Korner, E. Beck, A. Dommann, N. Onda, J. Ramm, Surf. Coat. Technol. 76 (1995) 731.
Plasma pretreatment of the Cu seed layer surface in Cu electroplating Junhwan Oh a , Jaegab Lee b , Chongmu Lee a,∗ a
Department of Metallurgical Engineering, Inha University, 253 Yonghyeondong, Inchon 402-751, South Korea b Department of Metallurgical and Materials Engineering, Kookmin University, Seoul 136-702, South Korea Received 8 December 2000; received in revised form 28 April 2001; accepted 2 May 2001
Abstract Effects of plasma pretreatment to the Cu seed layer on copper (Cu) electroplating were investigated. Copper seed layers were deposited by magnetron sputtering onto tantalum nitride barrier layers before electroplating copper in the forward pulsed mode. The Cu seed layer was cleaned by plasma H2 or N2 prior to electroplating a copper film. Cu films electroplated on the copper seed layer with plasma pretreatment have shown better electrical and physical properties such as electrical resistivities, surface morphologies, levels of impurities, adhesion and surface roughness than those without plasma pretreatment. It is shown that carbon and metal oxide contaminants at the sputtered Cu seed/TaN surface can be effectively removed by plasma H2 cleaning. The degree of the (1 1 1) preferred orientation of the pulsed plated Cu film with plasma H2 pretreatment is as high as that without plasma pretreatment. Also, plasma H2 precleaning is more effective in enhancing the Cu electroplating properties onto the Cu seed layer than plasma N2 precleaning. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Plasma pretreatment; Electroplating; Seed layer; Copper
1. Introduction Copper has drawn much attention as a new interconnect material for deep submicron ultra-large scale integrated circuits (ULSIs), because of lower resistivity (1.7 cm) and higher resistance to electromigration compared with aluminum and its alloys [1–4]. Cu films can be obtained by several different deposition techniques including physical vapor deposition (PVD), chemical vapor deposition (CVD), and plating such as electroless plating and electroplating. Especially, electroplating is a very inexpensive process in principle and offers high deposition rate and a number of research groups have successfully used it to fill damascene structures. Therefore, the most popular technique today includes Cu electroplating onto a trench or a via [5]. Electroplating is an attractive alternative deposition method for copper with the need for a conformal and conductive seed layer [6]. The seed layer provides a low resistance conductor for the plating current that drives the electrodeposition process and also facilitates nucleation of the plated copper film. In addition, the Cu seed layer should be highly pure so as not to compromise the effective resistivity of the filled copper interconnects structure. This seed layer requires low electrical resistivity, low levels of impu∗ Corresponding author. Fax: +82-32-862-5546. E-mail address: [email protected] (C. Lee).
rities, smooth interface, good adhesion to the barrier metal and low thickness concurrent with coherence for ensuring void-free fill [7]. The electrical conductivity of the surface plays an important role in the formation of initial Cu nuclei and the surfaces with higher electrical conductivities shows much better formation of Cu nuclei than those with lower electrical conductivities. It is also known that the nucleation processes of Cu are very sensitive to surface condition [8]. Pulse plating has been around the industry for a very long time. The primary attraction was its control of the cost normally associated with precious metal plating. This was achieved through grain refinement, improved porosity and minimization of overplating. Also, pulse plating techniques reduced the formation of voids because the rate of metal deposition deep inside a trench becomes nearly the same as the rate at the upper portion. When Cu is deposited on the Cu seed/TaN substrate by electroplating, precleaning of substrate surface like plasma treatment is essential. Plasma cleaning, employing hydrogen as reactive gas, has been used for many years. The idea behind the utilization of the plasma is the creation of a very reactive gas environment, often enclosed in a vacuum system [9]. Hydrogen discharge cleaning of vacuum vessels for particle accelerators and magnetic fusion devices is a well-known technique in the field of nuclear and particle physics for the removal of light impurities from metallic materials. Hydrogen plasma can be used to remove most
0254-0584/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 2 5 4 - 0 5 8 4 ( 0 1 ) 0 0 3 7 8 - 9
228
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
materials provided that volatile hydrides are formed at the operating temperature. The plasma pretreatment was used to remove oxides and etch residues that contaminate the metal surface at the bottom of a via. In this communication, we report the investigation of the effects of plasma pretreatment to Cu seed/tantalum nitride (TaN)/borophosphosilicate glass (BPSG) samples on electroplating of copper (Cu) films using four point probe measurement, X-ray diffractometry (XRD), scanning electron microscopy (SEM), atomic force microscopy (AFM) and X-ray photoelectron spectroscopy (XPS) focused on levels of impurities in the electroplated Cu film and surface chemical bonding state of the Cu seed layer and the electroplated Cu film.
2. Experimental Copper seed layers (500 Å) were deposited onto TaN barrier layer by magnetron sputtering using a Cu target in Ar gas ambient before electroplating of copper in the forward
pulsed mode. Prior to electroplating a copper film, the Cu seed layer surface was cleaned with hydrogen and nitrogen plasma treatments. A sputtered TaN film 500 Å. thick was used as a barrier and adhesion promoter. The samples were rinsed with DI water for 3 min, followed by filtered nitrogen gas drying before electroplating. Plasma treatments were conducted in a parallel-type cold wall reactor of a plasma enhancement chemical vapor deposition (PECVD) system after pumping down to the base pressure of 10−5 Torr. Plasma exposure time during the plasma treatment was varied from 1 to 20 min and rf-power was from 20 to 140 W and the flow rate and the deposition temperature of plasma pretreatment were 100 sccm and 298 K, respectively. After plasma cleaning, Cu electroplating was conducted onto each substrate. The electroplating system mainly consists of a copper plating solution (a sulfuric acid–copper sulfate chemical solution without organic additives), a soluble anode, a wafer (which serves as a cathode) and a duty power supply. Forward pulsed plating was performed to improve via/trench filling capability. Then copper was electroplated in the copper plating solution onto the sub-
Fig. 1. SEM micrographs of copper films electroplated in a copper sulfate/sulfuric acid bath: (a) DC plating; (b) DC plating (vacuum annealing); (c) pulse plating at 60 mA cm−2 (on: 80 ms, off: 20 ms); (d) pulse plating at 60 mA cm−2 (on: 60 ms, off: 40 ms); (e) pulse plating at 60 mA cm−2 (on: 40 ms, off: 60 ms); (f) pulse plating at 60 mA cm−2 (on: 20 ms, off: 80 ms).
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
strate metal surface, which acts as the negative electrode of the electrochemical cell. Copper plating rate was in the range of 0.5–1 m min−1 and the current density was kept at 60 mA cm−2 . Effects of plasma pretreatments to Cu seed/TaN/BPSG samples on electroplating of copper (Cu) films were investigated. Atomic force microscopy (AFM, TopoMetrix ACCUREX) and scanning electron microscopy (SEM, Hitach S-4300) were used to examine the morphology of the substrate and the sputtered Cu seed. Resistivity of each sample was measured by the four-point probe measurement technique. The crystallographic orientations of Cu films were investigated by X-ray diffraction (XRD, Philips X’pert MPD). Levels of impurities and surface chemical bonding states of the Cu seed layer surfaces were investigated by X-ray photoelectron spectroscopy (XPS, SPESE). 3. Results and discussion Fig. 1 shows SEM micrographs of the copper film that was deposited by DC or pulsed electroplating in a sulfuric acid–cupric sulfate solution. The figures show that the pulse plating technique produces a relatively smooth fine-grained structure, while DC plating (see Fig. 1 (a) and (b)) offers rougher surface composed of large columnar grains. It should be noted that a fine-grained structure is obtained only when the pulse plating current densities have been used. The optimum pulse sequence is a complex function of the pulse current density, duty cycle, and polarity reversal. Resistivities of the pulse electroplated Cu at various off times are shown in Fig. 2. The pulse current (PC) plated Cu film shows lower resistivity than the direct current (DC) plated Cu film. Fig. 3 shows SEM micrographs of the samples, which have been given hydrogen and nitrogen plasma treatment at various plasma powers followed by Cu electroplating for 0.5–1 m. For nitrogen plasma treatment as the plasma power increases, the size of electroplated Cu particles also increases and the surface becomes very rougher. In the case
229
Fig. 2. Resistivities of Cu films electroplated at various on-off times.
of plasma H2 pretreatment the Cu grain size nearly does not change with increasing the rf-power. The Cu film surface became smoother with plasma H2 pretreatment. However, the smoothness was not improved with increasing the rf-power further beyond 140 W. As the rf-power increases higher than 140 W in the H2 plasma pretreatment, cracks and pores were found to form in the Cu film. Especially, these effects are strong for the plasma N2 pretreatment. The resistivities of the samples after Cu film deposition which have been pretreatmented with plasma H2 at the power of 20–140 W was as low as 1.9–2.1 cm. These may be due to the cause of the reduction of CuO and surface impurities. Effects of the plasma exposure time are also presented in Fig. 4. As the plasma exposure time increases, the Cu film surface is observed to get rougher. A distinct boundary between electroplating and sputtered Cu seed layers was not observed on SEM micrographs (Fig. 5), which indicates a homoepitaxial interface relation between the two. Fig. 6 shows the AFM micrographs of Cu seed layers on TaN given the plasma pretreatment at 100 W for 10 min. Figs. 7 and 8 show the RMS roughness of the
Fig. 3. SEM micrographs (×5000) of Cu films electroplated with plasma pretreatments at various rf-powers (exposure time: 10 min).
230
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
Fig. 4. SEM micrographs (×10,000) of Cu films electroplated with plasma pretreatment at various plasma exposure times (rf-power: 100 W).
Cu seed layer on TaN for various rf-powers and plasma exposure times in the plasma H2 pretreatment. The surface roughness tends to decrease with increasing rf-power, which implies that particulate contaminants on the seed layer surfaces have been removed by plasma treatment. Also, the surface roughness decreases rapidly with increasing the plasma exposure time. It is worth noting that the surface roughness decreases slowly in the case of nitrogen plasma cleaning, where as it decreases rapidly in the case of hydrogen plasma cleaning. Effects of H2 and N2 plasma pretreatments at 100 W for 10 min on the surface roughness of the electroplated Cu film are shown in AFM images (Fig. 9) and the surface of the Cu film electroplated on the Cu seed layer given the plasma H2 cleaning is smooth, where as that electroplated on the seed layer given plasma N2 cleaning is rough (root mean square (RMS) roughness = 14.75 nm with H2 plasma, 27.96 nm with N2 plasma). From this AFM observation it may be
Fig. 5. A cross-sectional SEM micrograph (×20,000) of the Cu film electroplated with the plasma H2 pretreatment (rf-power: 100 W, exposure time: 10 min).
said that the plasma H2 pretreatment offers the Cu film with smoother surface than the plasma N2 pretreatment. Figs. 10 and 11 present dependence of the surface roughness of the electroplated Cu on various plasma powers and plasma exposure times, respectively. The plasma H2 pretreatment at the power of 100 W for 10 min gives the lowest RMS roughness and resistivity. In this experiment, the plasma exposure time was fixed at 10 min. Also in the case of the plasma H2 pretreatment the surface roughness of the electroplated Cu film tends to decrease slowly, while in the case of the plasma N2 pretreatment it increases rapidly with increasing the rf-power of the plasma pretreatment. The plasma H2 pretreatment is favorable in obtaining smoother surface of the electroplating Cu film. Fig. 12 shows degrees of preferred orientation for three types of electroplated Cu films. The degree of the (1 1 1) preferred orientation of the Cu film with plasma H2 pretreatment is as high as the pulse plated Cu film without plasma pretreatment. In general Cu films with a stronger (1 1 1) texture have higher electromigration resistance. Sputtered Cu seed layers contain various kinds of impurities like F, O, C, N, S, Cl, etc. Plasma pretreatment minimizes contamination on the Cu seed layer/TaN sample. The C 1s peak (284.5 eV) and the O 1s peak (531.0 eV) in the XPS spectra of the Cu seed layer on the TaN substrate are shown in Figs. 13 and 14, respectively. The carbon and oxygen impurity concentrations in the top surface area of the Cu seed layer/TaN sample can be found in these figures. Fig. 15 shows the XPS of Cu 2p peak of the Cu seed layer on the TaN substrate. Cu (in CuO) 2p3/2 and Cu (in Cu2 O) 2p1/2 appeared at 933.6 and 953.5 eV, respectively. Figs. 13–15 were obtained after plasma H2 and N2 cleaning (100 W, 10 min). They show that the heights of the C, O and Cu–O peaks of the Cu seed layer/TaN sample are decreased by the plasma H2 pretreatment. Enrichment of oxygen and carbon was found at the interface of the electroplated Cu film and the Cu
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
231
Fig. 6. AFM micrographs of the Cu seed layer/TaN samples with (a) no pretreatment; (b) plasma H2 pretreatment; (c) plasma N2 pretreatment (rf-power: 100 W, plasma exposure time: 10 min).
Fig. 7. RMS roughnesses of the Cu seed/TaN film deposited with plasma H2 and N2 pretreatments at various rf-powers (plasma exposure time: 10 min).
Fig. 8. RMS roughnesses of the Cu seed/TaN film deposited with plasma H2 and N2 pretreatments at various plasma exposure times (rf-power: 100 W).
232
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
Fig. 9. AFM micrographs of the Cu films electroplated with (a) no pretreatment, (b) plasma H2 pretreatment and (c) plasma N2 pretreatment (rf-power: 100 W, plasma exposure time: 10 min).
Fig. 10. RMS roughnesses of the Cu films electroplated with plasma H2 and N2 pretreatments at various rf-powers (plasma exposure time: 10 min).
Fig. 11. RMS roughnesses of the Cu films electroplated with plasma H2 and N2 pretreatments at various plasma exposure times (rf-power: 100 W).
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
233
Fig. 14. The XPS O 1s peak (531.0 eV) of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time : 10 min). Fig. 12. XRD spectra for various types of electroplated Cu films.
seed/TaN/SiO2 /Si substrate. It can be due to the cause of poor wetting and peeling of the electroplated Cu film. Electroplated copper films with high purity were obtained. To understand the chemistry of removal of contaminants such as hydrocarbon and oxygen, CH4 and H2 O were added to the discharge in a separate experiment. From the result of this experiment we may draw a conclusion that a major mechanism for removing hydrocarbon is the formation of species with m/e = 15, 16, 17, 18 and 27, whereas the ion
Fig. 13. The XPS C 1s peak (284.5 eV) of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time: 10 min).
density of m/e = 19 is mainly related to the formation of H2 O. Cu–O + H2 → Cu + H2 O C + 2H2 → CH4 Oxygen atoms react with hydrogen atoms to form H2 O and some of carbon atoms in the Cu seed layer surface react with hydrogen atoms to form CH4 during the plasma H2 treatment. As a result of these two reactions only free Cu atoms remain since volatile H2 O and CH4 are evaporated. Hence, there are lots of free Cu atoms at the surface of the Cu seed layer/TaN sample after hydrogen plasma treatment. In contrast, little change occurs at the top surface of
Fig. 15. The XPS of Cu 2p peak of the Cu seed layer on the TaN substrate (rf-power: 100 W, exposure time: 10 min).
234
J. Oh et al. / Materials Chemistry and Physics 73 (2002) 227–234
the sample given a nitrogen treatment. Only a small number of free Cu atoms exist after the nitrogen plasma treatment. Therefore, the plasma H2 treatment is more effective in Cu nucleation and growth than the plasma N2 treatment, since the former forms free Cu atoms more effectively by removing carbon atoms and Cu-oxide at the top surface of the Cu seed layer/TaN sample.
cleaning procedure, resulting in substrate surfaces free of carbon and oxygen contaminants. The degree of the (1 1 1) preferred orientation of the Cu film with plasma H2 pretreatment is as high as pulse plated Cu film without plasma pretreatment. Plasma H2 precleaning of the Cu seed layer surface is more effective in enhancing Cu electroplating properties than plasma N2 precleaning.
4. Conclusions
Acknowledgements
Effects of plasma pretreatment to Cu seed/TaN/BPSG samples on the properties of the copper (Cu) film which is subsequently electroplated on the samples were studied and the following conclusions were obtained. The plasma H2 pretreatment offers a better condition for Cu electroplating than the plasma N2 treatment or no pretreatment. It is because only the plasma H2 treatment produces free Cu atoms by removing Cu oxide and carbon in the surface region. The plasma H2 is effective in removing carbon impurities and results in Cu-rich films and smooth morphologies. Our results show that Cu films electroplated on plasma pretreated copper seed layers give better electrical resistivities, surface morphologies, adhesion and low levels of impurities than those on copper seed layers without plasma pretreatment. The plasma treatment removes oxide, carbon and other residues from the substrate surface. XPS spectra show that volatile hydrogen compounds form during the
This work was supported by Grant No. 1999-1-301-002-5 from the Basic Research Program of the Korea Science and Engineering Foundation. References [1] J. Musil, A.J. Bell, M. Cepera, Czchoslovak, J. Phys. 45 (1995) 249. [2] J.C. Chiou, K.C. Juang, M.C. Chen, J. Electrochem. Soc. 142 (1995) 177. [3] V.M. Donnelly, M.E. Gross, J. Vac. Sci. Technol. A 11 (1993) 66. [4] J.B. Webb, D. Northcott, I. Emesh, Thin Solid Films 270 (1995) 483. [5] V. Dubin, S. Lopatin, R. Cheung, in: Electrochemical Society Proceedings, Vol. 98, 1998, p. 6. [6] R.L. Jackson, E. Broadbent, T. Cacouris, A. Harrus, M. Buberger, E. Patton, T. Walsh, Solid State Technol. 5 (1998) 49. [7] B. Chin, P. Ding, B. Sun, T. Chiang, D. Angelo, I. Hashim, Solid State Technol. 7 (1998) 141. [8] Y.S. Kim, D. Jung, S.K. Min, Thin Solid Films 349 (1999) 36. [9] N. Korner, E. Beck, A. Dommann, N. Onda, J. Ramm, Surf. Coat. Technol. 76 (1995) 731.