- Email: [email protected]
Depth resolution optimization for low-energy ERDA
NIM B Beam Interactions with Materials & Atoms
Nuclear Instruments and Methods in Physics Research B 261 (2007) 512–515 www.elsevier.com/locate/nimb
Depth resolution optimization for low-energy ERDA S. Giangrandi
a,b,*
, K. Arstila a,c, B. Brijs a, T. Sajavaara d, A. Vantomme c, W. Vandervorst a,b
a Imec, Kapeldreef 75, B-3001 Leuven, Belgium K.U.Leuven, INSYS, Kasteelpark Arenberg 10, B-3001 Leuven, Belgium Instituut voor Kern-en Stralingsfysica, K.U.Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium d Department of Physics, P.O. Box 35, FIN-40014, University of Jyva¨skyla¨, Finland b
c
Available online 11 April 2007
Abstract With the implementation of low-energy time-of-flight Elastic Recoil Detection Analysis (ERDA), routine analysis of thin films with high depth resolution becomes possible. The optimization of the measurement conditions is a key issue for an accurate sample characterization and is normally a compromise among depth resolution, mass resolution and sensitivity, for a given sample. In this work, we focus on the depth resolution optimization, presenting an extensive study of two different materials, SiO2 and TiN, representative of light and medium mass targets. The film thickness varies between 10 and 50 nm. The samples were measured with different beams (35Cl, 63Cu, 79Br and 127I), energies (from 2 to 16 MeV) and incident angles. The experimental results are supported and generalized by simulations run with the Monte Carlo code MCERD. The different contributions of the system resolution, straggling and multiple scattering are evaluated and discussed. The best surface resolution is obtained in the low-energy limit. On the other hand, at low-energy the resolution deteriorates rapidly and better results for thicker films are obtained with higher incident energies. The loss of resolution with increasing depth is dominated by multiple scattering and becomes more relevant for heavy ions and heavy target atoms. In order to maintain a good depth resolution throughout the film, reducing the incident angle is more efficient than acting on the beam energy. Ó 2007 Elsevier B.V. All rights reserved. PACS: 82.80.Yc; 24.10.Lx; 82.80.Rt Keywords: Elastic recoil detection; Time-of-flight; Depth resolution; Multiple scattering; Monte Carlo simulation
1. Introduction A significant number of time-of-flight (TOF) spectrometers combined with energy detectors have been developed in the past two decades and ERDA has become accepted as a quantitative depth profiling method for the analysis of thin films [1,2].
*
Corresponding author. Address: Imec, Kapeldreef 75, B-3001 Leuven, Belgium. Tel.: +32 16 288097; fax: +32 16 281576. E-mail address: [email protected] (S. Giangrandi). 0168-583X/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2007.03.093
Due to the shrinking dimensions in microelectronic devices, during the recent years there has been more interest towards TOF-ERDA in laboratories with smaller accelerators (terminal voltage < 2 MV) [3,4]. Low-energy TOF-ERDA can combine accurate data quantification with fast response time. The measurement conditions for a given sample are normally a compromise among depth resolution, mass resolution, sensitivity and sample damage due to ion beam irradiation. Previous works have shown high resolution results obtained with low-energy ERDA [5,6], but no comprehensive study of the topic has been so far presented. In this work, we investigate the contributions that limit the depth resolution and optimize the experimental
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515
conditions as a function of the film thickness for a light target (SiO2) and a medium mass target (TiN).
513
3. Results and discussion 3.1. Surface resolution
Two different materials were studied in this work: thermal SiO2 and TiN films. The SiO2 films were thermally grown on 12 in. silicon wafers. The film thickness was measured in 12 different points across the wafer by means of spectroscopic ellipsometry, performed with an Aset-F5 tool. Average values of 10.2, 20.0 and 48.4 nm, with a standard deviation smaller than 1%, were measured for the three samples studied. The estimated sample density is 2.27 g/cm3. The TiN films were grown by sputter deposition on top of 100 nm SiO2 on a silicon substrate. The films have nominal thicknesses of 10, 15, 20, 30 and 50 nm and density of 4.0 g/cm3. The film composition, as measured with TOFERDA using 16 MeV 63Cu beam, is Ti 45%–N 55%. A very limited amount of surface impurities (H, C and O) was observed for both materials. The different incident beams considered (35Cl, 63Cu, 79Br and 127I) were produced by a SNICS ion source and accelerated through a Pelletron accelerator (2 MV terminal voltage) at energies ranging between 2 and 16 MeV. The TOF spectrometer here used for the measurements is described elsewhere [7]. The telescope is placed at 38.4° with respect to the beam and the sample alignment can be freely chosen using a six-axis goniometer. The beam incident angle is defined with respect to the target surface. When not differently specified, the measurements were done in mirror geometry, with both incident and exit angles equal to 19.2°. The detector solid angle is 0.36 msr, while the TOF length and the time resolution are 57.4 cm and 550 ps, respectively. The experimental energy resolution was extracted by fitting the front edge of the energy profiles with an error function. The conversion to depth was obtained using the surface approximation [8]. A similar procedure was used at the back edge for the extraction of the depth resolution at the interface, having considered the energy lost by the incident ion in the layer. In addition to the experimental data, simulations were performed using the Monte Carlo code MCERD [9]. The simulation includes the experimental parameters associated to the setup as well as multiple scattering (MS) and energy straggling in the film. The depth resolution can be extracted from the output of the simulation considering the TOF distribution related to a specific depth, according to the following expression: DT ðd i Þ Ddðd i Þ ¼ ; ðot=odÞjd i where Dd(di) and DT(di) are the depth and the TOF resolution at depth di and t the average TOF as a function of the depth.
The surface depth resolution of the system is limited by several factors, i.e. time resolution, detector solid angle, straggling in the C foil, tandem effect, C foil thickness non-uniformity, length resolution of the time-of-flight and beam divergency, energy spread and spot size. All these contributions were analytically calculated for the current setup and compared to the experimental data. Fig. 1 shows the surface depth resolution as a function of the beam energy, in case of O atoms recoiled by a 35Cl beam incident on a 50 nm SiO2 layer. For all the ions and the energies studied, the main contribution is given by the kinematic spreading due to the detector solid angle. The surface depth resolution for different beams and energies is shown in Fig. 2. The experimental data are compared to the Monte Carlo simulations, showing that the best surface resolution is obtained for all beams in the low-energy limit. Moreover, at similar incident energies heavier beams provide better results, but it is not clear what is the optimal combination of beam and energy. Same conclusions can be drawn in case of TiN target, both for the N and Ti signals (not shown). 3.2. Depth dependence The picture becomes more complex when considering the depth resolution at different depths. In Fig. 3, the resolution for different film thicknesses is considered both for SiO2 (O signal) and TiN (N signal). The energies shown in the figure provide the best results for the two targets. The resolution deteriorates linearly with the depth and, as expected, more rapidly for heavier beams, lower energies and higher target mass. For example, in case of 6 MeV 127I on SiO2 it drops by 5 nm every 10 nm of film, while for same energy 63Cu this value is 18
Exp Total Solid angle (0.36 msr) Time resol (550 ps) C foil non-homog (60 %) Tandem effect (15 keV)
16
Surface depth resolution (nm)
2. Experimental details
14 12 10 8 6 4 2 0 0
2
4
6
8
10
12
Beam energy (MeV) Fig. 1. Experimental and calculated surface depth resolution versus ion energy. The O atoms are recoiled by a 35Cl beam incident on a 50 nm SiO2 layer. Calculated contributions to the energy resolution are also shown.
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515 16
35
Cl - exp Cl - sim 63 Cu - exp 63 Cu - sim 79 Br - exp 79 Br - sim 127 I - exp 127 I - sim
Surface depth resolution (nm)
35
14
12
10
8
6
4 2
4
6
8
10
12
14
16
Beam energy (MeV) Fig. 2. Experimental and simulated surface depth resolution versus beam energy in case of O recoiled by 35Cl, 63Cu, 79Br and 127I beams. The best results are obtained in the low-energy limit.
35
4 MeV Cl - exp 35 4 MeV Cl - sim 35 6 MeV Cl - exp 35 6 MeV Cl - sim 63 6 MeV Cu - exp 63 6 MeV Cu - sim 127 6 MeV I - exp 127 6 MeV I - sim
Depth resolution (nm)
30
25
20
15
3.3. Angle dependence
10
5 0
10
20
30
40
50
Depth (nm) 63
6 MeV Cu exp 63 6 MeV Cu sim 63 9 MeV Cu exp 63 9 MeV Cu sim 79 9 MeV Br exp 79 9 MeV Br sim 127 9 MeV I exp 127 9 MeV I sim
Depth resolution (nm)
30
25
20
15
5 10
In addition to the beam mass and energy, the role of the measurement geometry was also evaluated. Measurements were performed varying the angle between the incident beam and the target surface from 19.2° to 5°. The results for the Ti profiles obtained from TiN layers with 63Cu beam are shown in Fig. 5. Due to the increased path length of the ions inside the film, the surface depth resolution can be sensibly improved by reducing the incident angle. More important, the depth resolution deteriorates with similar slopes for different incident angles and, for the cases shown, even at glancing geometry the resolution remains improved up to 20 nm. On the other hand, at smaller angles the energy profile shows a long tail caused by MS. Due to this tail, any con500
10
0
few nanometers and are thus preferable to study the film quality at the interface. Similar considerations can be done about the energy dependence. Lower beam energies guarantee a better resolution in the near-surface region, but with a higher deterioration with increasing depth. Two different processes contribute to the loss of resolution with increasing depth: energy straggling and MS. These two contributions are shown in Fig. 4 for an O profile obtained from a 50 nm SiO2 layer with 6 MeV 35Cl beam. While the energy straggling modifies only slightly the profile simulated considering a constant surface resolution, the addition of multiple scattering is essential for a correct interpretation of the experimental data. It is important to note that the energy profile broadening due to multiple scattering is present not only at the interface, but to a smaller extent also at the surface. This effect becomes more visible with heavier beams and targets, and implies a different surface depth resolution for different film thicknesses.
20
30
40
50
Depth (nm) Fig. 3. Depth resolution as a function of film thickness, for different films and energies. (a) O signal from SiO2 films. (b) N signal from TiN films.
reduced to 2 nm every 10 nm. The dependence on the target mass can be deduced from the slopes of N and O depth resolution in case of 6 MeV 63Cu beam. Based on our data, when comparing similar energies, lighter beams result in better resolutions already after a
Yield (counts/channel)
514
EXP Surface Straggling MCERD
400
300
200
100
0 2.4
2.6
2.8
3.0
3.2
3.4
Energy (MeV) Fig. 4. Experimental O profiles recoiled from a 50 nm SiO2 film measured with 6 MeV 35Cl beam and simulated profiles considering (1) no energy straggling, (2) energy straggling and (3) multiple scattering.
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515 30
Depth resolution (nm)
25
20 63
9 MeV Cu α = 19.2o o α = 12.0 o α = 8.0 o α = 5.0
15
10
63
5
4 MeV Cu o α = 19.2
0 0
10
20
30
40
50
Depth (nm) Fig. 5. Angle dependence of the depth resolution for Ti recoiled from TiN films measured with 9 MeV 63Cu with different incident angles. The results obtained for 4 Mev 63Cu are also shown for comparison. Points and lines indicate respectively experimental and simulated results.
siderations about possible material diffusion in the underlying layer become impossible. It is also clear from the figure that a more constant depth resolution throughout the film is guaranteed by reducing the incident angle rather than by acting on the beam energy. 4. Conclusions In this work, we have studied the optimization of the depth resolution as a function of the film thickness for a light and a medium mass target, varying the beam mass and energy as well as the incident angle.
515
The best surface resolution is obtained in the low-energy limit and can be significantly improved by tilting the sample in order to increase the path length of the ions in the target. However, the resolution deteriorates rapidly as a function of the depth in this energy range and better results for thicker films are obtained with higher incident energies. The loss of resolution with increasing depth is dominated by multiple scattering and becomes more relevant for heavy ions or heavy target atoms, even for films only a few nanometers thick, thus demanding higher energies and lighter ion beams. In order to maintain a good depth resolution throughout the film, reducing the incident angle is more efficient than acting on the beam energy. References [1] H.J. Whitlow, G. Possnert, C.S. Petersson, Nucl. Instr. and Meth. B 27 (1987) 448. [2] P. Goppelt, B. Gebauer, D. Fink, M. Wilpert, Th. Wilpert, W. Bohne, Nucl. Instr. and Meth. B 68 (1992) 235. [3] J.K. Kim, Y.S. Kim, G.D. Kim, H.W. Choi, H.J. Woo, S.Y. Cho, C.N. Whang, Nucl. Instr. and Meth. B 140 (1998) 380. [4] B. Brijs, T. Sajavaara, S. Giangrandi, K. Arstila, A. Vantomme, W. Vandervorst, Microelectron. Eng. 80 (2005) 106. [5] M. Do¨beli, C. Kottler, F. Glaus, M. Suter, Nucl. Instr. and Meth. B 241 (2005) 428. [6] B. Brijs et al., Nucl. Instr. and Meth. B 249 (2006) 847. [7] S. Giangrandi, T. Sajavaara, B. Brijs, K. Arstila, A. Vantomme, W. Vandervorst, in preparation. [8] W.K. Chu, J.W. Mayer, M.A. Nicolet, in: Backscattering Spectrometry, Academy Press, 1978. [9] K. Arstila, T. Sajavaara, J. Keinonen, Nucl. Instr. and Meth. B 174 (2001) 163.
Nuclear Instruments and Methods in Physics Research B 261 (2007) 512–515 www.elsevier.com/locate/nimb
Depth resolution optimization for low-energy ERDA S. Giangrandi
a,b,*
, K. Arstila a,c, B. Brijs a, T. Sajavaara d, A. Vantomme c, W. Vandervorst a,b
a Imec, Kapeldreef 75, B-3001 Leuven, Belgium K.U.Leuven, INSYS, Kasteelpark Arenberg 10, B-3001 Leuven, Belgium Instituut voor Kern-en Stralingsfysica, K.U.Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium d Department of Physics, P.O. Box 35, FIN-40014, University of Jyva¨skyla¨, Finland b
c
Available online 11 April 2007
Abstract With the implementation of low-energy time-of-flight Elastic Recoil Detection Analysis (ERDA), routine analysis of thin films with high depth resolution becomes possible. The optimization of the measurement conditions is a key issue for an accurate sample characterization and is normally a compromise among depth resolution, mass resolution and sensitivity, for a given sample. In this work, we focus on the depth resolution optimization, presenting an extensive study of two different materials, SiO2 and TiN, representative of light and medium mass targets. The film thickness varies between 10 and 50 nm. The samples were measured with different beams (35Cl, 63Cu, 79Br and 127I), energies (from 2 to 16 MeV) and incident angles. The experimental results are supported and generalized by simulations run with the Monte Carlo code MCERD. The different contributions of the system resolution, straggling and multiple scattering are evaluated and discussed. The best surface resolution is obtained in the low-energy limit. On the other hand, at low-energy the resolution deteriorates rapidly and better results for thicker films are obtained with higher incident energies. The loss of resolution with increasing depth is dominated by multiple scattering and becomes more relevant for heavy ions and heavy target atoms. In order to maintain a good depth resolution throughout the film, reducing the incident angle is more efficient than acting on the beam energy. Ó 2007 Elsevier B.V. All rights reserved. PACS: 82.80.Yc; 24.10.Lx; 82.80.Rt Keywords: Elastic recoil detection; Time-of-flight; Depth resolution; Multiple scattering; Monte Carlo simulation
1. Introduction A significant number of time-of-flight (TOF) spectrometers combined with energy detectors have been developed in the past two decades and ERDA has become accepted as a quantitative depth profiling method for the analysis of thin films [1,2].
*
Corresponding author. Address: Imec, Kapeldreef 75, B-3001 Leuven, Belgium. Tel.: +32 16 288097; fax: +32 16 281576. E-mail address: [email protected] (S. Giangrandi). 0168-583X/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2007.03.093
Due to the shrinking dimensions in microelectronic devices, during the recent years there has been more interest towards TOF-ERDA in laboratories with smaller accelerators (terminal voltage < 2 MV) [3,4]. Low-energy TOF-ERDA can combine accurate data quantification with fast response time. The measurement conditions for a given sample are normally a compromise among depth resolution, mass resolution, sensitivity and sample damage due to ion beam irradiation. Previous works have shown high resolution results obtained with low-energy ERDA [5,6], but no comprehensive study of the topic has been so far presented. In this work, we investigate the contributions that limit the depth resolution and optimize the experimental
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515
conditions as a function of the film thickness for a light target (SiO2) and a medium mass target (TiN).
513
3. Results and discussion 3.1. Surface resolution
Two different materials were studied in this work: thermal SiO2 and TiN films. The SiO2 films were thermally grown on 12 in. silicon wafers. The film thickness was measured in 12 different points across the wafer by means of spectroscopic ellipsometry, performed with an Aset-F5 tool. Average values of 10.2, 20.0 and 48.4 nm, with a standard deviation smaller than 1%, were measured for the three samples studied. The estimated sample density is 2.27 g/cm3. The TiN films were grown by sputter deposition on top of 100 nm SiO2 on a silicon substrate. The films have nominal thicknesses of 10, 15, 20, 30 and 50 nm and density of 4.0 g/cm3. The film composition, as measured with TOFERDA using 16 MeV 63Cu beam, is Ti 45%–N 55%. A very limited amount of surface impurities (H, C and O) was observed for both materials. The different incident beams considered (35Cl, 63Cu, 79Br and 127I) were produced by a SNICS ion source and accelerated through a Pelletron accelerator (2 MV terminal voltage) at energies ranging between 2 and 16 MeV. The TOF spectrometer here used for the measurements is described elsewhere [7]. The telescope is placed at 38.4° with respect to the beam and the sample alignment can be freely chosen using a six-axis goniometer. The beam incident angle is defined with respect to the target surface. When not differently specified, the measurements were done in mirror geometry, with both incident and exit angles equal to 19.2°. The detector solid angle is 0.36 msr, while the TOF length and the time resolution are 57.4 cm and 550 ps, respectively. The experimental energy resolution was extracted by fitting the front edge of the energy profiles with an error function. The conversion to depth was obtained using the surface approximation [8]. A similar procedure was used at the back edge for the extraction of the depth resolution at the interface, having considered the energy lost by the incident ion in the layer. In addition to the experimental data, simulations were performed using the Monte Carlo code MCERD [9]. The simulation includes the experimental parameters associated to the setup as well as multiple scattering (MS) and energy straggling in the film. The depth resolution can be extracted from the output of the simulation considering the TOF distribution related to a specific depth, according to the following expression: DT ðd i Þ Ddðd i Þ ¼ ; ðot=odÞjd i where Dd(di) and DT(di) are the depth and the TOF resolution at depth di and t the average TOF as a function of the depth.
The surface depth resolution of the system is limited by several factors, i.e. time resolution, detector solid angle, straggling in the C foil, tandem effect, C foil thickness non-uniformity, length resolution of the time-of-flight and beam divergency, energy spread and spot size. All these contributions were analytically calculated for the current setup and compared to the experimental data. Fig. 1 shows the surface depth resolution as a function of the beam energy, in case of O atoms recoiled by a 35Cl beam incident on a 50 nm SiO2 layer. For all the ions and the energies studied, the main contribution is given by the kinematic spreading due to the detector solid angle. The surface depth resolution for different beams and energies is shown in Fig. 2. The experimental data are compared to the Monte Carlo simulations, showing that the best surface resolution is obtained for all beams in the low-energy limit. Moreover, at similar incident energies heavier beams provide better results, but it is not clear what is the optimal combination of beam and energy. Same conclusions can be drawn in case of TiN target, both for the N and Ti signals (not shown). 3.2. Depth dependence The picture becomes more complex when considering the depth resolution at different depths. In Fig. 3, the resolution for different film thicknesses is considered both for SiO2 (O signal) and TiN (N signal). The energies shown in the figure provide the best results for the two targets. The resolution deteriorates linearly with the depth and, as expected, more rapidly for heavier beams, lower energies and higher target mass. For example, in case of 6 MeV 127I on SiO2 it drops by 5 nm every 10 nm of film, while for same energy 63Cu this value is 18
Exp Total Solid angle (0.36 msr) Time resol (550 ps) C foil non-homog (60 %) Tandem effect (15 keV)
16
Surface depth resolution (nm)
2. Experimental details
14 12 10 8 6 4 2 0 0
2
4
6
8
10
12
Beam energy (MeV) Fig. 1. Experimental and calculated surface depth resolution versus ion energy. The O atoms are recoiled by a 35Cl beam incident on a 50 nm SiO2 layer. Calculated contributions to the energy resolution are also shown.
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515 16
35
Cl - exp Cl - sim 63 Cu - exp 63 Cu - sim 79 Br - exp 79 Br - sim 127 I - exp 127 I - sim
Surface depth resolution (nm)
35
14
12
10
8
6
4 2
4
6
8
10
12
14
16
Beam energy (MeV) Fig. 2. Experimental and simulated surface depth resolution versus beam energy in case of O recoiled by 35Cl, 63Cu, 79Br and 127I beams. The best results are obtained in the low-energy limit.
35
4 MeV Cl - exp 35 4 MeV Cl - sim 35 6 MeV Cl - exp 35 6 MeV Cl - sim 63 6 MeV Cu - exp 63 6 MeV Cu - sim 127 6 MeV I - exp 127 6 MeV I - sim
Depth resolution (nm)
30
25
20
15
3.3. Angle dependence
10
5 0
10
20
30
40
50
Depth (nm) 63
6 MeV Cu exp 63 6 MeV Cu sim 63 9 MeV Cu exp 63 9 MeV Cu sim 79 9 MeV Br exp 79 9 MeV Br sim 127 9 MeV I exp 127 9 MeV I sim
Depth resolution (nm)
30
25
20
15
5 10
In addition to the beam mass and energy, the role of the measurement geometry was also evaluated. Measurements were performed varying the angle between the incident beam and the target surface from 19.2° to 5°. The results for the Ti profiles obtained from TiN layers with 63Cu beam are shown in Fig. 5. Due to the increased path length of the ions inside the film, the surface depth resolution can be sensibly improved by reducing the incident angle. More important, the depth resolution deteriorates with similar slopes for different incident angles and, for the cases shown, even at glancing geometry the resolution remains improved up to 20 nm. On the other hand, at smaller angles the energy profile shows a long tail caused by MS. Due to this tail, any con500
10
0
few nanometers and are thus preferable to study the film quality at the interface. Similar considerations can be done about the energy dependence. Lower beam energies guarantee a better resolution in the near-surface region, but with a higher deterioration with increasing depth. Two different processes contribute to the loss of resolution with increasing depth: energy straggling and MS. These two contributions are shown in Fig. 4 for an O profile obtained from a 50 nm SiO2 layer with 6 MeV 35Cl beam. While the energy straggling modifies only slightly the profile simulated considering a constant surface resolution, the addition of multiple scattering is essential for a correct interpretation of the experimental data. It is important to note that the energy profile broadening due to multiple scattering is present not only at the interface, but to a smaller extent also at the surface. This effect becomes more visible with heavier beams and targets, and implies a different surface depth resolution for different film thicknesses.
20
30
40
50
Depth (nm) Fig. 3. Depth resolution as a function of film thickness, for different films and energies. (a) O signal from SiO2 films. (b) N signal from TiN films.
reduced to 2 nm every 10 nm. The dependence on the target mass can be deduced from the slopes of N and O depth resolution in case of 6 MeV 63Cu beam. Based on our data, when comparing similar energies, lighter beams result in better resolutions already after a
Yield (counts/channel)
514
EXP Surface Straggling MCERD
400
300
200
100
0 2.4
2.6
2.8
3.0
3.2
3.4
Energy (MeV) Fig. 4. Experimental O profiles recoiled from a 50 nm SiO2 film measured with 6 MeV 35Cl beam and simulated profiles considering (1) no energy straggling, (2) energy straggling and (3) multiple scattering.
S. Giangrandi et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 512–515 30
Depth resolution (nm)
25
20 63
9 MeV Cu α = 19.2o o α = 12.0 o α = 8.0 o α = 5.0
15
10
63
5
4 MeV Cu o α = 19.2
0 0
10
20
30
40
50
Depth (nm) Fig. 5. Angle dependence of the depth resolution for Ti recoiled from TiN films measured with 9 MeV 63Cu with different incident angles. The results obtained for 4 Mev 63Cu are also shown for comparison. Points and lines indicate respectively experimental and simulated results.
siderations about possible material diffusion in the underlying layer become impossible. It is also clear from the figure that a more constant depth resolution throughout the film is guaranteed by reducing the incident angle rather than by acting on the beam energy. 4. Conclusions In this work, we have studied the optimization of the depth resolution as a function of the film thickness for a light and a medium mass target, varying the beam mass and energy as well as the incident angle.
515
The best surface resolution is obtained in the low-energy limit and can be significantly improved by tilting the sample in order to increase the path length of the ions in the target. However, the resolution deteriorates rapidly as a function of the depth in this energy range and better results for thicker films are obtained with higher incident energies. The loss of resolution with increasing depth is dominated by multiple scattering and becomes more relevant for heavy ions or heavy target atoms, even for films only a few nanometers thick, thus demanding higher energies and lighter ion beams. In order to maintain a good depth resolution throughout the film, reducing the incident angle is more efficient than acting on the beam energy. References [1] H.J. Whitlow, G. Possnert, C.S. Petersson, Nucl. Instr. and Meth. B 27 (1987) 448. [2] P. Goppelt, B. Gebauer, D. Fink, M. Wilpert, Th. Wilpert, W. Bohne, Nucl. Instr. and Meth. B 68 (1992) 235. [3] J.K. Kim, Y.S. Kim, G.D. Kim, H.W. Choi, H.J. Woo, S.Y. Cho, C.N. Whang, Nucl. Instr. and Meth. B 140 (1998) 380. [4] B. Brijs, T. Sajavaara, S. Giangrandi, K. Arstila, A. Vantomme, W. Vandervorst, Microelectron. Eng. 80 (2005) 106. [5] M. Do¨beli, C. Kottler, F. Glaus, M. Suter, Nucl. Instr. and Meth. B 241 (2005) 428. [6] B. Brijs et al., Nucl. Instr. and Meth. B 249 (2006) 847. [7] S. Giangrandi, T. Sajavaara, B. Brijs, K. Arstila, A. Vantomme, W. Vandervorst, in preparation. [8] W.K. Chu, J.W. Mayer, M.A. Nicolet, in: Backscattering Spectrometry, Academy Press, 1978. [9] K. Arstila, T. Sajavaara, J. Keinonen, Nucl. Instr. and Meth. B 174 (2001) 163.