- Email: [email protected]
Effects of Ag doping on the crystallization properties of Sb-rich GeSb thin films
Thin Solid Films 519 (2011) 5323–5328
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Effects of Ag doping on the crystallization properties of Sb-rich GeSb thin films Nam Hee Kim a, Hyung Keun Kim a, Kyu Min Lee a, Hyun Chul Sohn a, Jae Sung Roh b, Doo Jin Choi a,⁎ a b
Department of Materials Science and Engineering, Yonsei University, 134 Shinchon-dong, Seodaemun-gu, Seoul 120-752, Republic of Korea Memory R&D Division, Hynix Semiconductor Inc., San 136-1, Amiri, Bubal-eup, Ichon-si, Gyeonggi-do, 467-701, Republic of Korea
a r t i c l e
i n f o
Article history: Received 8 April 2010 Received in revised form 27 January 2011 Accepted 8 February 2011 Available online 24 February 2011 Keywords: Sb-rich GeSb Ag doping Phase change Transmission Electron Microscopy Static test
a b s t r a c t Ag-doped and un-doped Sb-rich GeSb thin films were deposited by DC magnetron co-sputtering. The electrical, structural, and optical properties of the thin films phase change were investigated using 4-point probe measurement, X-ray diffraction (XRD), transmission electron microscopy (TEM), and a static tester. With increasing Ag doping content, the crystallization temperature and sheet resistance of crystalline state decreased from 325 °C to 283 °C and from 187.33 Ω/□ to 114.62 Ω/□, respectively. XRD patterns of the films showed a Sb hexagonal structure, and the calculated grain size increased from 13.9 nm to 17 nm as the Ag concentration increased. Grain sizes of the Ag-doped thin films were larger than the grain sizes of un-doped thin films, as determined by TEM images. A static tester verified the decreased crystallization speed and optical contrast. Un-doped GeSb crystallization took 160 ns and 16 at.% Ag-doped GeSb crystallization took 200 ns when the laser power was 13 mW. Based on a power–time-effect diagram, the 12.6 at.% Ag-doped GeSb showed good thermal stability in a crystalline state. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Phase change materials, such as compact disks, digital versatile disks, and blu-ray disks, have been used extensively for optical storage as they store information by optical changes between amorphous and crystalline phases [1]. These phase change materials can also be applied to phase change random access memory (PRAM) technology, which stores the information by the electrical resistivity differences between the two phases. Recently, PRAM has become a promising candidate for the next generation of non-volatile memory due to its scalability, fast operation speed, and compatibility with the complementary metal– oxide–semiconductor manufacturing process [2]. The most widely adopted phase change material is Ge2Sb2Te5 (GST225). However, there are some reports that Te can easily diffuse and interact with adjacent elements, deteriorating the reliability of the PRAM device [3]. Moreover, Te is toxic and, therefore, is not environmentally friendly. In this study, Te-free Sb-rich GeSb was chosen as a phase change material due to the fast crystallization speed as it is a grain growth dominant material [4], with high amorphous phase stability and an adequate archival life time [5] due to high crystallization temperature, as compared to commonly used phase change materials, like GST225 (156 °C), Sb2Te (114 °C), Ag/In doped Sb2Te (AIST, 163 °C) [6], and Ge1Sb4Te7 (GST147, 150 °C) [7].
⁎ Corresponding author. E-mail address: [email protected] (D.J. Choi). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.02.031
Significant work has been performed to improve the operating performance of PRAM by changing the device structure [8], adding a heating layer [9] and doping with various elements, such as N [10], O [11], Sn [12], and Si [13] to phase change the materials. In this work, Ag was chosen as a dopant, as it has already shown improved properties of GST225 [14] and Sb2Te3 (ST23) [15] phase change materials containing chalcogen Te elements. However, Ag doping of Te free Ge–Sb phase change materials for PRAM applications have been rarely reported. Sb-rich GeSb was doped with Ag, and the property changes of un-doped Sb-rich GeSb (GS) and Ag-doped GeSb (AGS) thin films were observed. Hereafter, 6.9 at.% Ag-GeSb, 12.6 at.% Ag-GeSb and 16 at.% Ag-GeSb will be represented as 6.9AGS, 12.6AGS and 16AGS, respectively. 2. Experiment One hundred-nm-thick GS and AGS thin films were deposited by DC magnetron co-sputtering from Ag (99.99%, SMC Tech Co. Ltd., Korea) and Ge20Sb80 (99.99%, SMC Tech Co. Ltd., Korea) targets at room temperature on glass and Si(100) substrates. The doped Ag content was altered by 0, 6.9, 12.6, and 16 at.% in the thin films by maintaining the Ag sputtering power at 0, 2.75, 5.6 and 8.55 W, while the sputtering power of Ge20Sb80 was maintained at 24 W. The concentration of each element in the film was verified by an electron probe X-ray microanalyzer (EPMA, EPMA 1600). The film thicknesses were measured by a surface profiler AS500 (KLA-Tencor Co.). The base pressure was less than 2.67 × 10− 4 Pa (2.0 × 10− 6 Torr), and the working pressure was 1.07 × 101 Pa (8.0 × 10− 3 Torr), which was the proper pressure in this study for generating plasma with an Ar flow of
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
40 SCCM. A substrate holder was rotated to obtain uniform thickness. In order to crystallize the films, the specimens were annealed by a halogen heater in a chamber at a ramp rate of 10 °C/min in an Ar atmosphere. A 4-point probe (CMT-SR 2000N, Korea) was used to confirm changes in the sheet resistance of each film. The crystal structure was analyzed by X-ray diffraction (XRD, D/MAX-2500H, Rigaku, Japan) using CuKα (λ = 0.15405 nm) radiation with a twotheta range of 20°–55°, in which the main peaks are located. Grain sizes were calculated by the Debye–Scherer formula using the XRD patterns and observed by transmission electron microscopy (TEM, JEM-2100, JEOL) images using a 0.00197 nm of electron wavelength. A static tester (Nanostorage Co. Ltd., Korea) was conducted by laser irradiation (λ = 650 nm) in the nanosecond temporal scale in order to observe the crystallization behavior.
a Sheet resistance (Ohm/ sq.)
5324
10
10
10
10
10
10
3. Results and discussion
GS 6.9AGS 12.6AGS 16AGS
6
5
4
3
2
1
50
0
100 150 200 250 300 350 400 450 500
Temperature (OC)
concentration (atomic %)
b 2000
dR/dT
0
GS
-2000
6.9AGS 12.6AGS
-4000
16AGS
-6000 O
0
50
200
250
300
350
400
450
190 180 170 160 150 140 130
80
110 0
60
150
c
120
Sb
100
315 C O 325 C
O Temperature ( C)
90
70
O
283 C O 288 C
-8000
Rset (Ohm/ sq.)
Fig. 1 shows the EPMA result of GS and AGSs. The x-axis represents the Ag sputtering power, and the y-axis represents the concentration of each element. The concentration of the sputtered GS thin film was similar to the target composition of Ge20Sb80. The Ag doping concentration increased with increasing Ag sputtering power, and the ratio of Ge:Sb was maintained as 1:3.09–3.26. The error bar was indicated at each concentration point in Fig. 1, and the standard deviation of the concentration was approximately ± 1.6 at.%. Therefore, this EPMA result was pretty reliable. Fig. 2a shows the sheet resistance as a function of the temperature for GS and AGS thin films. The films were annealed at each temperature point for 20 min at a ramp rate of 10 °C/min. The sheet resistance was high at the lower annealing temperature and abruptly decreased after a specific annealing temperature, likely due to film crystallization. The sheet resistance and crystallization temperature in the amorphous and crystalline states decreased as the Ag doping content increased in the films. For more specific analysis, the changes in resistance of the crystalline state, the set state (Rset), and crystallization temperature were shown in Fig. 2b and c. The average sheet resistance in the crystalline state, Rset, was 187.33 Ω/□, 158.98 Ω/□, 116.14 Ω/□ and 114.62 Ω/□ for GS, 6.9AGS, 12.6AGS, and 16AGS, respectively. In order to define the crystallization temperature, the sheet resistance graph was differentiated, and the minimum value of the derivative of each graph was regarded as the crystallization temperature. The crystallization temperature was approximately 325 °C, 315 °C, 288 °C, and 283 °C for GS, 6.9AGS, 12.6AGS, and 16AGS, respectively. The decreased crystallization temperature by doping Ag can be explained by a total bond enthalpy. Lankhorst [16] reported that the glass transition temperature corresponded to the total bond enthalpy, which was mostly related to the bond enthalpy and the number of bonds between atoms. The bond
3
6
9
12
15
18
Ag concentration (at.%) Fig. 2. (a) Sheet resistance changes as a function of temperature, (b) average sheet resistance in the crystalline state (Rset), and (c) differentiated sheet resistance change as a function of temperature for GS, 6.9AGS, 12.6AGS, and 16AGS.
50 40 30 Ge
20
Ag 10 0 0
3
6
9
Ag sputtering power (W) Fig. 1. EPMA results of each element as a function of the Ag sputtering power.
enthalpy of Ag–Ag is −50 kJ/mol, which is the average of the two values in Ref. [16]. This value was too low, compared to the bond enthalpies of Ge–Ge (186 kJ/mol) and Sb–Sb (175 kJ/mol) [16]. Therefore, AGS has a low glass transition temperature since the Ag atom leads to a low, total bond enthalpy. We also calculated glass transition temperature in accordance with the equations from the paper of Lankhorst. The crystallization temperature can be determined by experimental fact that the glass transition temperatures of the materials are mostly 80% of the
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
crystallization temperature [17]. Hence, we obtained the calculated crystallization temperatures of 301 °C and 276 °C for GS and 16AGS, and these values were similar with our experimental crystallization temperatures of 325 °C and 283 °C for GS and 16AGS. Alternatively, in the case of ST and GST phase change materials that contain a group VI atom, an increased crystallization temperature with doping Ag [14,15] occurred, although the enthalpy of Ag–Ag (−50 kJ/mol) was low,
compared to that of Ge–Ge (186 kJ/mol), Sb–Sb (175 kJ/mol) and Te–Te (197 kJ/mol). The number of Ag–Te and Ag–Sb bonds potentially compensated the reduced number of Ge–Te and Sb–Te bonds [16]. Therefore, doping Ag could increase the glass transition temperature of ST and GST systems. Fig. 3 shows XRD patterns of the films. Fig. 3a is an as-deposited XRD pattern, and Fig. 3b, c, and d shows the XRD patterns of the films
b
40
50
20
Intensity (arb. unit)
12.6AGS 30
40
50
6.9AGS 20
50
30
40
50
30
40
50
40
50
12.6AGS 20
6.9AGS 30
40
50
20
GS
GS 20
40
30
40
50
20
30
2θ(degree)
30
40
50
12.6AGS 20
30
40
50
30
40
50
Sb (012) 30
40
50
30
40
50
30
40
50
40
50
12.6AGS 20
20
GS
GS 20
20
6.9AGS
6.9AGS 20
16AGS
Intensity (arb. unit)
20
Sb (202)
16AGS
d Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Sb (012)
c
2θ(degree)
Sb (202)
20
30
Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Intensity (arb. unit)
30
Sb (202)
16AGS
16AGS 20
Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Sb (012)
a
Intensity (arb. unit)
5325
30
40
2θ(degree)
50
20
30
2θ(degree)
Fig. 3. XRD patterns of GS, 6.9AGS, 12.6AGS, and 16AGS; (a) as-deposited, annealed at (b) 330 °C, (c) 360 °C and (d) 420 °C.
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
Dhkl =
0:9λ βcos θ
where Dhkl is the grain size, λ is wavelength of CuKα radiation (λ = 0.15405 nm), β is the FWHM value in the radian scale, and θ is the diffraction angle. The calculated grain sizes decreased from 17 nm
17.5
0.64
17.0
0.62
16.5
0.60 FWHM grain size
16.0
0.58
15.5 0.56 15.0 0.54
14.5
0.52
14.0 13.5
0.50 0
3
6
9
12
15
18
Ag doping contents (at.%) Fig. 5. Average FWHM of Sb(012), Sb(110) and Sb(202) peaks and calculated grain size as a function of Ag doping content.
to 13.9 nm. The grains of GS and 16AGS thin films were observed by the TEM images depicted in Fig. 6a and b show crystalline phases of GS and 16AGS annealed at 420 °C for 20 min. The GS film had similar or smaller grain sizes than 16AGS. Since the smaller grain size leads to increased grain boundary scattering, the GS film would have a higher sheet resistance, as shown at Fig. 2a. A static test was performed in order to investigate the optical properties and phase transition behavior in the nano-second scale. The static tester uses pulsed laser irradiation for the phase transition, and the wavelength of the laser was 650 nm. Fig. 7 shows a power– time-effect (PTE) diagram, which was obtained by the static test of GS and AGS thin films. The PTE diagram indicates the optical contrast with different colors for the induced pulse width (0–350 ns) and laser power (0–35 mW). The reflectivity change and optical contrast were defined using the following equation;
R=
Rafter −Rbefore Rbefore
420 OC Ag 3Sb (111)
Sb (104)
16AGS
Intensity (arb. unit)
38.5
12.6AGS 38.5
39.0
39.5
Ag 3Sb (111)
39.0
39.5
40.0
40.5
41.0
40.5
41.0
40.5
41.0
40.5
41.0
Sb (104)
40.0 Sb (104)
6.9AGS 38.5
39.0
39.5
40.0 Sb (104)
GS 38.5
39.0
39.5
40.0
2θ θ (degree) Fig. 4. Narrow scanned XRD patterns of Ag3Sb (111) and Sb (104).
FWHM (deg.)
annealed at 330 °C, 360 °C, and 420 °C for 20 min at a ramp rate of 10 °C/min. In Fig. 3a, all the films showed amorphous states. In Fig. 3b, all films except for GS were crystallized. In Fig. 3c and d, all of the films were crystallized, and the intensity of the peaks increased as the Ag doping contents increased. After crystallization, although the single element Sb had a rhombohedral structure [18], the XRD patterns showed an Sb hexagonal structure. In order to explain this difference, the crystallization of GeSb was considered. In general, GeSb does not have any compounds and they exist as a solid solution of Ge in the Sb [19,20], thus Sb atoms crystallize first, and Ge atoms distort the Sb structure. Therefore, the Sb hexagonal structure was observed in this study. The lattice constants of GS with a Sb hexagonal structure (a= 4.26 Å and c = 11.45 Å) were measured, and they were slightly different than those of the rhombohedral Sb structure [21] (a = 4.31 Å and c = 11.27 Å) due to the Ge atoms. In the case of the AGS films, peaks not observed on the GS XRD pattern appeared that were identified as an orthorhombic Ag3Sb compound. In Fig. 4, we depicted a narrow scan of Fig. 3d with a 2θ range of 38°–41° wherein the Ag3Sb (111) peak appears. Increased Ag content increased the intensity of the Ag3Sb peak, while the intensity of Sb (104) was similar for all compositions. In the case of 16AGS, the intensity of Ag3Sb was larger than the Sb (104) peak, considering the high concentration of Ag in the film. Fig. 5 shows the average calculated grain size and full width at half maximum (FWHM) of Sb (012), Sb (110) and Sb (202) peaks. The Sb (104) peak was excluded, as it overlapped with the Ag3Sb (111) peak, which could affect the FWHM value. The grain sizes were calculated by the Debye–Scherer formula;
Calculated grain size (nm)
5326
Fig. 6. TEM images of (a) GS and (b) 16AGS annealed at 420 °C.
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
5327
Fig. 7. PTE diagram of (a) GS, (b) 6.9AGS, (c) 12.6AGS, and (d) 16AGS.
where Rafter is the reflectivity of the film after laser irradiation, and Rbefore is the reflectivity of the film before laser irradiation. In general, the PTE diagram has amorphous (A), crystalline (B), and ablation (C) areas with different colors, which designates an optical contrast change [22]. The three A, B, and C areas were observed in Fig. 7. The region with a light blue color located at a low laser power and short pulse width indicates the amorphous state which shows almost no optical contrast changes, since the applied laser power and pulse width were not sufficient for film crystallization. In B region, a positive optical contrast change represented by a color change occurred with increasing laser power and pulse width. As a result, the film was crystallized since phase change thin films generally have low reflectivity in an amorphous state and high reflectivity in a crystalline state [18]. As shown in the ablation region of C, negative optical contrasts with the striped dark blue color were caused by excessive laser power and pulse width. The stability of crystalline state of the films can be inferred from the ablation area in the PTE diagram. As such, 12.6AGS had the smallest ablation area, as compared to other films; therefore, it can be concluded that the 12.6AGS film had good thermal stability in the crystalline state. On the contrary, 6.9AGS showed the largest ablation area among the films. In addition, the crystallization speed decreased as the Ag concentration increased, except when the laser power was 20 mW. For a more detailed analysis, the optical contrast changes of GS and 16AGS thin films were plotted when the laser power was 13 mW (Fig. 8a), 20 mW (Fig. 8b) and 30 mW (Fig. 8c). The minimum crystallization time of GS and 16AGS was 160 ns and 200 ns when the laser power was 13 mW, and 60 ns and 60 ns when the laser power was 20 mW. When the laser power was 30 mW,
the minimum crystallization time was 30 ns and 40 ns for GS and 16AGS. It was believed that the doped Ag atoms decreased the crystallization speed of the films. When a dopant was added to a phase change material, it is captured in a growing crystal during crystallization. The dopant induces strain in the lattice, increasing the internal free energy. Therefore, the crystalline material would have decreased crystallization speed [23]. In addition, the 16AGS film showed a lower optical contrast value, as compared to the GS film, which corroborates a study of D. Z. Dimitrov et al. [18].
4. Conclusion GS and AGS thin films were investigated with Ag concentrations of 0, 6.9, 12.6, and 16 at.%. The sheet resistances of AGS thin films were less than that of the GS thin film in both amorphous and crystalline states. The crystallization temperature decreased as the Ag concentration increased since the total bond enthalpy was lowered by doping Ag which has the Ag–Ag bond enthalpy of −50 kJ/mol. X-ray diffraction was analyzed to determine the microstructure of GS and AGS thin films. XRD patterns showed a Sb hexagonal structure, and the average grain sizes were calculated. The calculated grain sizes increased from 13.9 nm for GS to 17.0 nm for 16AGS. The grain sizes of GS thin film and 16AGS thin films were verified by the TEM analysis. The decreased crystallization speed and optical contrast due to Ag atoms were confirmed by a PTE diagram, and the 12.6AGS thin film showed good thermal stability in the crystalline state.
5328
a
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
Acknowledgements
0.06 GS
Optical contrast
This work was supported by the second stage of the Brain Korea 21 project in 2010 and Hynix Semiconductor, Inc., Korea.
16AGS
0.05 0.04
References
0.03
[1] [2] [3] [4] [5] [6] [7] [8]
0.02 0.01 0.00
Laser power: 13 mW
-0.01 0
50
100
150
200
250
300
350
Pulse width (ns)
b GS
0.025
16AGS
Optical contrast
0.020 0.015
[21] [22] [23]
0.010 0.005 0.000
Laser power: 20 mW -0.005 0
50
100
150
200
250
300
350
Pulse width (ns)
c GS 0.02
Optical contrast
[9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]
16AGS
0.01
0.00
-0.12 -0.18
Laser power: 30 mW
-0.24 0
50
100
150
200
250
300
350
Pulse width (ns) Fig. 8. Optical contrast changes of GS and 16AGS as a function of pulse width when the laser power was (a) 13 mW, (b) 20 mW, and (c) 30 mW.
N. Yamada, E. Ohno, K. Nishiuchi, N. Akahira, J. Appl. Phys. 69 (1991) 2849. J. Maimon, E. Spall, R. Quinn, S. Schnur, Proc. IEEE Aerosp. Conf. 5 (2001) 2289. L. Krusin-Elbaum, C. Cabral, Appl. Phys. Lett. 90 (2007) 141902. S. Raoux, R.M. Robert, J. Jordan-Sweet, Microelectron. Eng. 85 (2008) 2330. L. van Pieterson, M.H.R. Lankhorst, J. Appl. Phys. 97 (2005) 083520. S. Raoux, C.T. Rettner, J. Appl. Phys. 102 (2007) 094305. H.K. Kim, S.Y. Lee, D.J. Choi, J. Kor. Phys. Soc. 55 (2009) 1896. A. Pirovano, F. Pellizzer, A. Redaelli, Proceedings of ESSDERC 2005: 35th European Solid-.State Device Research Conference, Grenoble, France, September 12–16, 2005, 313 F. Rao, Z. Song, Appl. Phys. Lett. 92 (2008) 223507. S.Y. Lee, H.K. Kim, D.J. Choi, J. Mater. Sci. 44 (2009) 4354. J. González-Hernández, P. Herrera-Fierro, Thin Solid Films 503 (2006) 13. R. Kojima, N. Yamada, Jpn. J. Appl. Phys. 40 (2001) 5930. J. Feng, Y. Zhang, B.W. Qiao, Appl. Phys. A 87 (2007) 57. D.H. Kim, M.S. Kim, R.Y. Kim, Mater. Charact. 58 (2007) 479. J. Xu, B. Liu, Z. Song, S. Feng, B. Chen, Mater. Sci. Eng. B 127 (2006) 228. M.H.R. Lankhorst, J. NonCryst. Solids 297 (2002) 210. S. Raoux, J.L. Jordan-Sweet, A. Kellock, J. Appl. Phys. 101 (2007) 044909. D.Z. Dimitrov, C. Babeva, S.T. Cheng, W.C. Hsu, Proc. SPIE 5380 (2004) 487. B.C. Giessen, C. Borromee-Gautier, J. Solid State Chem. 4 (1972) 447. D. Shakhvorostov, R.A. Nistor, L. Krusin-Elbaum, Proc. Nat. Acad. Sci. U.S.A. 106 (2009) 10907. International Center for Diffraction Data File No. 85-1322. S.K. Kim, S.Y. Choi, Kor. J. Mater. Res. 19 (2009) 203. A. Hirotsune, M. Terao, Y. Miyauchi, M. Miyamoto, Jpn. J. Appl. Phys. 46 (2007) 6652.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Effects of Ag doping on the crystallization properties of Sb-rich GeSb thin films Nam Hee Kim a, Hyung Keun Kim a, Kyu Min Lee a, Hyun Chul Sohn a, Jae Sung Roh b, Doo Jin Choi a,⁎ a b
Department of Materials Science and Engineering, Yonsei University, 134 Shinchon-dong, Seodaemun-gu, Seoul 120-752, Republic of Korea Memory R&D Division, Hynix Semiconductor Inc., San 136-1, Amiri, Bubal-eup, Ichon-si, Gyeonggi-do, 467-701, Republic of Korea
a r t i c l e
i n f o
Article history: Received 8 April 2010 Received in revised form 27 January 2011 Accepted 8 February 2011 Available online 24 February 2011 Keywords: Sb-rich GeSb Ag doping Phase change Transmission Electron Microscopy Static test
a b s t r a c t Ag-doped and un-doped Sb-rich GeSb thin films were deposited by DC magnetron co-sputtering. The electrical, structural, and optical properties of the thin films phase change were investigated using 4-point probe measurement, X-ray diffraction (XRD), transmission electron microscopy (TEM), and a static tester. With increasing Ag doping content, the crystallization temperature and sheet resistance of crystalline state decreased from 325 °C to 283 °C and from 187.33 Ω/□ to 114.62 Ω/□, respectively. XRD patterns of the films showed a Sb hexagonal structure, and the calculated grain size increased from 13.9 nm to 17 nm as the Ag concentration increased. Grain sizes of the Ag-doped thin films were larger than the grain sizes of un-doped thin films, as determined by TEM images. A static tester verified the decreased crystallization speed and optical contrast. Un-doped GeSb crystallization took 160 ns and 16 at.% Ag-doped GeSb crystallization took 200 ns when the laser power was 13 mW. Based on a power–time-effect diagram, the 12.6 at.% Ag-doped GeSb showed good thermal stability in a crystalline state. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Phase change materials, such as compact disks, digital versatile disks, and blu-ray disks, have been used extensively for optical storage as they store information by optical changes between amorphous and crystalline phases [1]. These phase change materials can also be applied to phase change random access memory (PRAM) technology, which stores the information by the electrical resistivity differences between the two phases. Recently, PRAM has become a promising candidate for the next generation of non-volatile memory due to its scalability, fast operation speed, and compatibility with the complementary metal– oxide–semiconductor manufacturing process [2]. The most widely adopted phase change material is Ge2Sb2Te5 (GST225). However, there are some reports that Te can easily diffuse and interact with adjacent elements, deteriorating the reliability of the PRAM device [3]. Moreover, Te is toxic and, therefore, is not environmentally friendly. In this study, Te-free Sb-rich GeSb was chosen as a phase change material due to the fast crystallization speed as it is a grain growth dominant material [4], with high amorphous phase stability and an adequate archival life time [5] due to high crystallization temperature, as compared to commonly used phase change materials, like GST225 (156 °C), Sb2Te (114 °C), Ag/In doped Sb2Te (AIST, 163 °C) [6], and Ge1Sb4Te7 (GST147, 150 °C) [7].
⁎ Corresponding author. E-mail address: [email protected] (D.J. Choi). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.02.031
Significant work has been performed to improve the operating performance of PRAM by changing the device structure [8], adding a heating layer [9] and doping with various elements, such as N [10], O [11], Sn [12], and Si [13] to phase change the materials. In this work, Ag was chosen as a dopant, as it has already shown improved properties of GST225 [14] and Sb2Te3 (ST23) [15] phase change materials containing chalcogen Te elements. However, Ag doping of Te free Ge–Sb phase change materials for PRAM applications have been rarely reported. Sb-rich GeSb was doped with Ag, and the property changes of un-doped Sb-rich GeSb (GS) and Ag-doped GeSb (AGS) thin films were observed. Hereafter, 6.9 at.% Ag-GeSb, 12.6 at.% Ag-GeSb and 16 at.% Ag-GeSb will be represented as 6.9AGS, 12.6AGS and 16AGS, respectively. 2. Experiment One hundred-nm-thick GS and AGS thin films were deposited by DC magnetron co-sputtering from Ag (99.99%, SMC Tech Co. Ltd., Korea) and Ge20Sb80 (99.99%, SMC Tech Co. Ltd., Korea) targets at room temperature on glass and Si(100) substrates. The doped Ag content was altered by 0, 6.9, 12.6, and 16 at.% in the thin films by maintaining the Ag sputtering power at 0, 2.75, 5.6 and 8.55 W, while the sputtering power of Ge20Sb80 was maintained at 24 W. The concentration of each element in the film was verified by an electron probe X-ray microanalyzer (EPMA, EPMA 1600). The film thicknesses were measured by a surface profiler AS500 (KLA-Tencor Co.). The base pressure was less than 2.67 × 10− 4 Pa (2.0 × 10− 6 Torr), and the working pressure was 1.07 × 101 Pa (8.0 × 10− 3 Torr), which was the proper pressure in this study for generating plasma with an Ar flow of
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
40 SCCM. A substrate holder was rotated to obtain uniform thickness. In order to crystallize the films, the specimens were annealed by a halogen heater in a chamber at a ramp rate of 10 °C/min in an Ar atmosphere. A 4-point probe (CMT-SR 2000N, Korea) was used to confirm changes in the sheet resistance of each film. The crystal structure was analyzed by X-ray diffraction (XRD, D/MAX-2500H, Rigaku, Japan) using CuKα (λ = 0.15405 nm) radiation with a twotheta range of 20°–55°, in which the main peaks are located. Grain sizes were calculated by the Debye–Scherer formula using the XRD patterns and observed by transmission electron microscopy (TEM, JEM-2100, JEOL) images using a 0.00197 nm of electron wavelength. A static tester (Nanostorage Co. Ltd., Korea) was conducted by laser irradiation (λ = 650 nm) in the nanosecond temporal scale in order to observe the crystallization behavior.
a Sheet resistance (Ohm/ sq.)
5324
10
10
10
10
10
10
3. Results and discussion
GS 6.9AGS 12.6AGS 16AGS
6
5
4
3
2
1
50
0
100 150 200 250 300 350 400 450 500
Temperature (OC)
concentration (atomic %)
b 2000
dR/dT
0
GS
-2000
6.9AGS 12.6AGS
-4000
16AGS
-6000 O
0
50
200
250
300
350
400
450
190 180 170 160 150 140 130
80
110 0
60
150
c
120
Sb
100
315 C O 325 C
O Temperature ( C)
90
70
O
283 C O 288 C
-8000
Rset (Ohm/ sq.)
Fig. 1 shows the EPMA result of GS and AGSs. The x-axis represents the Ag sputtering power, and the y-axis represents the concentration of each element. The concentration of the sputtered GS thin film was similar to the target composition of Ge20Sb80. The Ag doping concentration increased with increasing Ag sputtering power, and the ratio of Ge:Sb was maintained as 1:3.09–3.26. The error bar was indicated at each concentration point in Fig. 1, and the standard deviation of the concentration was approximately ± 1.6 at.%. Therefore, this EPMA result was pretty reliable. Fig. 2a shows the sheet resistance as a function of the temperature for GS and AGS thin films. The films were annealed at each temperature point for 20 min at a ramp rate of 10 °C/min. The sheet resistance was high at the lower annealing temperature and abruptly decreased after a specific annealing temperature, likely due to film crystallization. The sheet resistance and crystallization temperature in the amorphous and crystalline states decreased as the Ag doping content increased in the films. For more specific analysis, the changes in resistance of the crystalline state, the set state (Rset), and crystallization temperature were shown in Fig. 2b and c. The average sheet resistance in the crystalline state, Rset, was 187.33 Ω/□, 158.98 Ω/□, 116.14 Ω/□ and 114.62 Ω/□ for GS, 6.9AGS, 12.6AGS, and 16AGS, respectively. In order to define the crystallization temperature, the sheet resistance graph was differentiated, and the minimum value of the derivative of each graph was regarded as the crystallization temperature. The crystallization temperature was approximately 325 °C, 315 °C, 288 °C, and 283 °C for GS, 6.9AGS, 12.6AGS, and 16AGS, respectively. The decreased crystallization temperature by doping Ag can be explained by a total bond enthalpy. Lankhorst [16] reported that the glass transition temperature corresponded to the total bond enthalpy, which was mostly related to the bond enthalpy and the number of bonds between atoms. The bond
3
6
9
12
15
18
Ag concentration (at.%) Fig. 2. (a) Sheet resistance changes as a function of temperature, (b) average sheet resistance in the crystalline state (Rset), and (c) differentiated sheet resistance change as a function of temperature for GS, 6.9AGS, 12.6AGS, and 16AGS.
50 40 30 Ge
20
Ag 10 0 0
3
6
9
Ag sputtering power (W) Fig. 1. EPMA results of each element as a function of the Ag sputtering power.
enthalpy of Ag–Ag is −50 kJ/mol, which is the average of the two values in Ref. [16]. This value was too low, compared to the bond enthalpies of Ge–Ge (186 kJ/mol) and Sb–Sb (175 kJ/mol) [16]. Therefore, AGS has a low glass transition temperature since the Ag atom leads to a low, total bond enthalpy. We also calculated glass transition temperature in accordance with the equations from the paper of Lankhorst. The crystallization temperature can be determined by experimental fact that the glass transition temperatures of the materials are mostly 80% of the
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
crystallization temperature [17]. Hence, we obtained the calculated crystallization temperatures of 301 °C and 276 °C for GS and 16AGS, and these values were similar with our experimental crystallization temperatures of 325 °C and 283 °C for GS and 16AGS. Alternatively, in the case of ST and GST phase change materials that contain a group VI atom, an increased crystallization temperature with doping Ag [14,15] occurred, although the enthalpy of Ag–Ag (−50 kJ/mol) was low,
compared to that of Ge–Ge (186 kJ/mol), Sb–Sb (175 kJ/mol) and Te–Te (197 kJ/mol). The number of Ag–Te and Ag–Sb bonds potentially compensated the reduced number of Ge–Te and Sb–Te bonds [16]. Therefore, doping Ag could increase the glass transition temperature of ST and GST systems. Fig. 3 shows XRD patterns of the films. Fig. 3a is an as-deposited XRD pattern, and Fig. 3b, c, and d shows the XRD patterns of the films
b
40
50
20
Intensity (arb. unit)
12.6AGS 30
40
50
6.9AGS 20
50
30
40
50
30
40
50
40
50
12.6AGS 20
6.9AGS 30
40
50
20
GS
GS 20
40
30
40
50
20
30
2θ(degree)
30
40
50
12.6AGS 20
30
40
50
30
40
50
Sb (012) 30
40
50
30
40
50
30
40
50
40
50
12.6AGS 20
20
GS
GS 20
20
6.9AGS
6.9AGS 20
16AGS
Intensity (arb. unit)
20
Sb (202)
16AGS
d Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Sb (012)
c
2θ(degree)
Sb (202)
20
30
Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Intensity (arb. unit)
30
Sb (202)
16AGS
16AGS 20
Ag3Sb (101) Ag3Sb (020) Ag3Sb (111) Sb (104) Sb (110)
Sb (012)
a
Intensity (arb. unit)
5325
30
40
2θ(degree)
50
20
30
2θ(degree)
Fig. 3. XRD patterns of GS, 6.9AGS, 12.6AGS, and 16AGS; (a) as-deposited, annealed at (b) 330 °C, (c) 360 °C and (d) 420 °C.
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
Dhkl =
0:9λ βcos θ
where Dhkl is the grain size, λ is wavelength of CuKα radiation (λ = 0.15405 nm), β is the FWHM value in the radian scale, and θ is the diffraction angle. The calculated grain sizes decreased from 17 nm
17.5
0.64
17.0
0.62
16.5
0.60 FWHM grain size
16.0
0.58
15.5 0.56 15.0 0.54
14.5
0.52
14.0 13.5
0.50 0
3
6
9
12
15
18
Ag doping contents (at.%) Fig. 5. Average FWHM of Sb(012), Sb(110) and Sb(202) peaks and calculated grain size as a function of Ag doping content.
to 13.9 nm. The grains of GS and 16AGS thin films were observed by the TEM images depicted in Fig. 6a and b show crystalline phases of GS and 16AGS annealed at 420 °C for 20 min. The GS film had similar or smaller grain sizes than 16AGS. Since the smaller grain size leads to increased grain boundary scattering, the GS film would have a higher sheet resistance, as shown at Fig. 2a. A static test was performed in order to investigate the optical properties and phase transition behavior in the nano-second scale. The static tester uses pulsed laser irradiation for the phase transition, and the wavelength of the laser was 650 nm. Fig. 7 shows a power– time-effect (PTE) diagram, which was obtained by the static test of GS and AGS thin films. The PTE diagram indicates the optical contrast with different colors for the induced pulse width (0–350 ns) and laser power (0–35 mW). The reflectivity change and optical contrast were defined using the following equation;
R=
Rafter −Rbefore Rbefore
420 OC Ag 3Sb (111)
Sb (104)
16AGS
Intensity (arb. unit)
38.5
12.6AGS 38.5
39.0
39.5
Ag 3Sb (111)
39.0
39.5
40.0
40.5
41.0
40.5
41.0
40.5
41.0
40.5
41.0
Sb (104)
40.0 Sb (104)
6.9AGS 38.5
39.0
39.5
40.0 Sb (104)
GS 38.5
39.0
39.5
40.0
2θ θ (degree) Fig. 4. Narrow scanned XRD patterns of Ag3Sb (111) and Sb (104).
FWHM (deg.)
annealed at 330 °C, 360 °C, and 420 °C for 20 min at a ramp rate of 10 °C/min. In Fig. 3a, all the films showed amorphous states. In Fig. 3b, all films except for GS were crystallized. In Fig. 3c and d, all of the films were crystallized, and the intensity of the peaks increased as the Ag doping contents increased. After crystallization, although the single element Sb had a rhombohedral structure [18], the XRD patterns showed an Sb hexagonal structure. In order to explain this difference, the crystallization of GeSb was considered. In general, GeSb does not have any compounds and they exist as a solid solution of Ge in the Sb [19,20], thus Sb atoms crystallize first, and Ge atoms distort the Sb structure. Therefore, the Sb hexagonal structure was observed in this study. The lattice constants of GS with a Sb hexagonal structure (a= 4.26 Å and c = 11.45 Å) were measured, and they were slightly different than those of the rhombohedral Sb structure [21] (a = 4.31 Å and c = 11.27 Å) due to the Ge atoms. In the case of the AGS films, peaks not observed on the GS XRD pattern appeared that were identified as an orthorhombic Ag3Sb compound. In Fig. 4, we depicted a narrow scan of Fig. 3d with a 2θ range of 38°–41° wherein the Ag3Sb (111) peak appears. Increased Ag content increased the intensity of the Ag3Sb peak, while the intensity of Sb (104) was similar for all compositions. In the case of 16AGS, the intensity of Ag3Sb was larger than the Sb (104) peak, considering the high concentration of Ag in the film. Fig. 5 shows the average calculated grain size and full width at half maximum (FWHM) of Sb (012), Sb (110) and Sb (202) peaks. The Sb (104) peak was excluded, as it overlapped with the Ag3Sb (111) peak, which could affect the FWHM value. The grain sizes were calculated by the Debye–Scherer formula;
Calculated grain size (nm)
5326
Fig. 6. TEM images of (a) GS and (b) 16AGS annealed at 420 °C.
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
5327
Fig. 7. PTE diagram of (a) GS, (b) 6.9AGS, (c) 12.6AGS, and (d) 16AGS.
where Rafter is the reflectivity of the film after laser irradiation, and Rbefore is the reflectivity of the film before laser irradiation. In general, the PTE diagram has amorphous (A), crystalline (B), and ablation (C) areas with different colors, which designates an optical contrast change [22]. The three A, B, and C areas were observed in Fig. 7. The region with a light blue color located at a low laser power and short pulse width indicates the amorphous state which shows almost no optical contrast changes, since the applied laser power and pulse width were not sufficient for film crystallization. In B region, a positive optical contrast change represented by a color change occurred with increasing laser power and pulse width. As a result, the film was crystallized since phase change thin films generally have low reflectivity in an amorphous state and high reflectivity in a crystalline state [18]. As shown in the ablation region of C, negative optical contrasts with the striped dark blue color were caused by excessive laser power and pulse width. The stability of crystalline state of the films can be inferred from the ablation area in the PTE diagram. As such, 12.6AGS had the smallest ablation area, as compared to other films; therefore, it can be concluded that the 12.6AGS film had good thermal stability in the crystalline state. On the contrary, 6.9AGS showed the largest ablation area among the films. In addition, the crystallization speed decreased as the Ag concentration increased, except when the laser power was 20 mW. For a more detailed analysis, the optical contrast changes of GS and 16AGS thin films were plotted when the laser power was 13 mW (Fig. 8a), 20 mW (Fig. 8b) and 30 mW (Fig. 8c). The minimum crystallization time of GS and 16AGS was 160 ns and 200 ns when the laser power was 13 mW, and 60 ns and 60 ns when the laser power was 20 mW. When the laser power was 30 mW,
the minimum crystallization time was 30 ns and 40 ns for GS and 16AGS. It was believed that the doped Ag atoms decreased the crystallization speed of the films. When a dopant was added to a phase change material, it is captured in a growing crystal during crystallization. The dopant induces strain in the lattice, increasing the internal free energy. Therefore, the crystalline material would have decreased crystallization speed [23]. In addition, the 16AGS film showed a lower optical contrast value, as compared to the GS film, which corroborates a study of D. Z. Dimitrov et al. [18].
4. Conclusion GS and AGS thin films were investigated with Ag concentrations of 0, 6.9, 12.6, and 16 at.%. The sheet resistances of AGS thin films were less than that of the GS thin film in both amorphous and crystalline states. The crystallization temperature decreased as the Ag concentration increased since the total bond enthalpy was lowered by doping Ag which has the Ag–Ag bond enthalpy of −50 kJ/mol. X-ray diffraction was analyzed to determine the microstructure of GS and AGS thin films. XRD patterns showed a Sb hexagonal structure, and the average grain sizes were calculated. The calculated grain sizes increased from 13.9 nm for GS to 17.0 nm for 16AGS. The grain sizes of GS thin film and 16AGS thin films were verified by the TEM analysis. The decreased crystallization speed and optical contrast due to Ag atoms were confirmed by a PTE diagram, and the 12.6AGS thin film showed good thermal stability in the crystalline state.
5328
a
N.H. Kim et al. / Thin Solid Films 519 (2011) 5323–5328
Acknowledgements
0.06 GS
Optical contrast
This work was supported by the second stage of the Brain Korea 21 project in 2010 and Hynix Semiconductor, Inc., Korea.
16AGS
0.05 0.04
References
0.03
[1] [2] [3] [4] [5] [6] [7] [8]
0.02 0.01 0.00
Laser power: 13 mW
-0.01 0
50
100
150
200
250
300
350
Pulse width (ns)
b GS
0.025
16AGS
Optical contrast
0.020 0.015
[21] [22] [23]
0.010 0.005 0.000
Laser power: 20 mW -0.005 0
50
100
150
200
250
300
350
Pulse width (ns)
c GS 0.02
Optical contrast
[9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]
16AGS
0.01
0.00
-0.12 -0.18
Laser power: 30 mW
-0.24 0
50
100
150
200
250
300
350
Pulse width (ns) Fig. 8. Optical contrast changes of GS and 16AGS as a function of pulse width when the laser power was (a) 13 mW, (b) 20 mW, and (c) 30 mW.
N. Yamada, E. Ohno, K. Nishiuchi, N. Akahira, J. Appl. Phys. 69 (1991) 2849. J. Maimon, E. Spall, R. Quinn, S. Schnur, Proc. IEEE Aerosp. Conf. 5 (2001) 2289. L. Krusin-Elbaum, C. Cabral, Appl. Phys. Lett. 90 (2007) 141902. S. Raoux, R.M. Robert, J. Jordan-Sweet, Microelectron. Eng. 85 (2008) 2330. L. van Pieterson, M.H.R. Lankhorst, J. Appl. Phys. 97 (2005) 083520. S. Raoux, C.T. Rettner, J. Appl. Phys. 102 (2007) 094305. H.K. Kim, S.Y. Lee, D.J. Choi, J. Kor. Phys. Soc. 55 (2009) 1896. A. Pirovano, F. Pellizzer, A. Redaelli, Proceedings of ESSDERC 2005: 35th European Solid-.State Device Research Conference, Grenoble, France, September 12–16, 2005, 313 F. Rao, Z. Song, Appl. Phys. Lett. 92 (2008) 223507. S.Y. Lee, H.K. Kim, D.J. Choi, J. Mater. Sci. 44 (2009) 4354. J. González-Hernández, P. Herrera-Fierro, Thin Solid Films 503 (2006) 13. R. Kojima, N. Yamada, Jpn. J. Appl. Phys. 40 (2001) 5930. J. Feng, Y. Zhang, B.W. Qiao, Appl. Phys. A 87 (2007) 57. D.H. Kim, M.S. Kim, R.Y. Kim, Mater. Charact. 58 (2007) 479. J. Xu, B. Liu, Z. Song, S. Feng, B. Chen, Mater. Sci. Eng. B 127 (2006) 228. M.H.R. Lankhorst, J. NonCryst. Solids 297 (2002) 210. S. Raoux, J.L. Jordan-Sweet, A. Kellock, J. Appl. Phys. 101 (2007) 044909. D.Z. Dimitrov, C. Babeva, S.T. Cheng, W.C. Hsu, Proc. SPIE 5380 (2004) 487. B.C. Giessen, C. Borromee-Gautier, J. Solid State Chem. 4 (1972) 447. D. Shakhvorostov, R.A. Nistor, L. Krusin-Elbaum, Proc. Nat. Acad. Sci. U.S.A. 106 (2009) 10907. International Center for Diffraction Data File No. 85-1322. S.K. Kim, S.Y. Choi, Kor. J. Mater. Res. 19 (2009) 203. A. Hirotsune, M. Terao, Y. Miyauchi, M. Miyamoto, Jpn. J. Appl. Phys. 46 (2007) 6652.