Optimization of high tunability barium strontium ... - Semantic Scholar
Optimization of High Tunability Barium Strontium Titanate Thin Films Grown by RF Magnetron Sputtering Nadia K. Pervez, Peter J. Hansen, and Robert A. York Department of Electrical and Computer Engineering and Materials Science Department University of California, Santa Barbara, California 93106 Abstract— Barium strontium titanate is a solid solution perovskite with a field-dependent permittivity. At microwave frequencies, its tunable dielectric constant and low loss make it a competitive choice for varactors and other tunable circuit elements. Much attention has been focused on the production of low-loss films for such applications, with little emphasis on how the film properties contribute to the circuit loss. When tunable elements are implemented in circuits, electrode loss dominates over film loss. In applications such as phase shifters, where cascaded tuning elements provide a predetermined amount of tuning, circuit designs using high tunability films minimize the number of tuning elements required, resulting in an overall reduction in circuit loss. When growth conditions are optimized for superior electrical properties, tunability and film loss are the two quantities of interest. By changing the oxygen partial pressure during growth, the amount of excess Ti incorporated into the film is changed. Films with higher excess Ti contents exhibit lower losses, higher breakdown voltages, and lower permittivities than more stoichiometric films. While all of the films approach the same high-field capacitance limit, the total tunability is determined not only by the zero-field permittivity but also by the breakdown voltage; a device must be able to tolerate sufficient applied bias to reach its high-field capacitance limit. By balancing these factors, we have produced capacitors with an unprecedented 13.71:1 (92.7%) tuning ratio at an applied field of 4.7 MV/cm.
I. I NTRODUCTION In recent years there has been much interest in thin-film barium strontium titanate (BST) for tunable components for RF and microwave applications [1], [2]. The applications for tunable dielectrics range from applications where films with low loss-tangents and small tunabilities are required to applications where films with high tunabilities and moderate loss-tangents are desirable. While recent work has focused on optimization of BST for low-loss applications [3], the advantages of growth optimization for high tunability applications have been largely unrealized. At high frequencies, ohmic losses from electrodes and circuit metalization dominate circuit loss, reducing the benefit of a material with an intrinsic low loss-tangent. High tunability applications require cascaded tuning elements. Designs employing high tunability films require fewer tuning elements, thereby reducing the amount of circuit metalization and associated ohmic loss. Phase shifters are an example of one such application, where a predetermined amount of phase delay must be implemented by cascading phase shifting units.
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
An increase in tunability results in an increase in the amount of phase shift per unit, and therefore a reduction in the total number of units required for the circuit implementation [4]. In this experiment we evaluated the effect of variations in the partial pressure of oxygen during sputter deposition on the electrical properties of thin film BST capacitors with the goal of maximizing tunability and dielectric strength. II. E XPERIMENTAL A series of films were grown on both platinized and bare c-plane sapphire substrates with backside Ti layers. The platinized sapphire substrates were for the fabrication of devices for electrical measurements whereas films grown on the bare sapphire substrates were used for Rutherford backscattering (RBS) analysis. During each growth, half-wafers of both types of substrates were used. The surface temperature of the platinized sapphire was 700 ◦ C during deposition for all growths. The sputtering targets were of two slightly different compositions: Ba0.5 Sr0.5 TiO3 and Ba0.5 Sr0.5 Ti1.02 O3 . The RF power on each gun was 150 W. The sputtering ambient for all growths was a mixture of Ar and O2 at a total pressure of 45 mTorr and a combined total flow rate of 100 sccm. The Ar/O2 flow rates for the series were 90/10, 80/20, 70/30, 60/40, and 50/50 sccm corresponding to oxygen partial pressures of 4.5, 9.0, 13.5, 18.0, and 22.5 mTorr. The growth time for all depositions was 64 minutes, resulting in film thicknesses ranging from 113 nm to 145 nm as measured by surface profilometry. The lower oxygen partial pressures corresponded to higher deposition rates. Parallel plate capacitors were fabricated using the films deposited on platinized sapphire. The 200 nm Pt ground planes on the substrates were deposited by e-beam evaporation prior to film growth. Buffered HF was used for a mesa isolation etch, after which top electrode and ground plane contacts were formed by lift-off. The top electrode metalization layer consisted of 150 nm of Pt deposited by e-beam evaporation. Capacitance and Q-factor (inverse loss-tangent) measurements were performed on 40 x 50 µm2 devices using an Agilent 4294A impedance analyzer for low voltage measurements (≤ 40 V) at 1 MHz and 100 MHz. High voltage capacitance measurements were performed at 100 MHz using an HP 8722D network analyzer with an external bias tee. A Keithley 6517A electrometer was used to source the high
278
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
600
350
90/10 80/20 70/30 60/40 50/50
εr
400 300
Fig. 1.
250 200 150
200
100
100
50
0
-3
-2
-1 0 1 E (MV/cm)
2
90/10 80/20 70/30 60/40 50/50
300
Q
500
0 -4 -3 -2 -1 0 1 2 E (MV/cm)
3 Fig. 3.
1 MHz r -E tuning curves for different Ar/O2 ratios (sccm).
C
Q
-11
140
7 10
90/10 500 13.71:1 @ 80/20 80/20 breakdown 400 11.36:1 @ 300 90/10 breakdown 200
-11
6 10
-11
5 10
C (F)
maximum tunability 10.15:1
-11
4 10
120
useful tunability 5.36:1
100 80
Q
εr
4
1 MHz Q-E curves for different Ar/O2 ratios (sccm).
600
-11
60
-11
40
-11
20
3 10 2 10
1 10
100 0 -5
3
0 -35
-4
-3 -2 E (MV/cm)
-1
-30
-25
-20
-15
-10
-5
0
0
V
0 Fig. 4.
Maximum and useful tunabilities.
Fig. 2. 100 MHz maximum tunability curves for 90/10 and 80/20 sccm Ar/O2 films.
voltages while an external series resistance of approximately 50 kΩ was used for current limiting. An Agilent 4155B semiconductor parameter analyzer was used for DC leakage current measurements. Compositional data was determined by RBS analysis on the samples grown on bare sapphire. III. R ESULTS AND D ISCUSSION Permittivity tuning curves in Figure 1 show that the film with the lowest oxygen partial pressure had the highest zerobias permittivity. All of the films in the series were observed to approach the same high-field permittivity limit, but some of the devices failed before reaching this limit. In the absence of such device failure, we would this expect the film with the highest zero-bias permittivity to have the highest tunability. The highest tunabilities were actually measured for the film with the second highest permittivity. In this case the increase in the film’s dielectrics strength compensated for the reduction in permittivity, as seen in Figure 2. Figure 3 shows how the Q-factor changed with bias for the films with the four lowest oxygen partial pressures during growth. When interpreting tuning curves, it is important to differentiate between the maximum tuning range (device destruction) and the useful tuning range — the range useful for circuit implementations. Figure 4 shows how the tuning curves together with the Q-factors can be used to define a maximum tunability and useful tunability for devices from
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
each film. The useful tuning range is limited by Q-rollover, which is associated with an exponential increase in leakage currents preceding device failure. From the results in Figure 3, we can see that the useful tuning range increases with higher oxygen partial pressures. From Figure 1 we see that the zero-bias permittivity drops significantly when the oxygen partial pressure is increased above 9.0 mTorr (80/20 sccm). Thus we are not surprised when we find in Table 1 that the 80/20 film has a higher useful tunability than the 90/10 film. The data in Table 1 shows that for the lower permittivity films the useful tunability approaches the maximum tunability. This is because the rough measure we use to define useful tunability fails in these cases because catastrophic device failure can occur without the Q-rollover
279
TABLE I E LECTRICAL PROPERTIES FOR THE FIVE FILMS GROWN WITH DIFFERENT A R /O2 FLOW RATES AT THE SAME TOTAL GROWTH PRESSURE .
Ar/O2 (sccm) 90/10
r
Q
Maximum Tunability
Useful Tunability
571
(1 MHz) 41
(100 MHz) 11.36:1 (91.2%)
(100 MHz) 6.79:1 (85.3%)
80/20
455
75
13.71:1 (92.7%)
9.38:1 (89.3%)
70/30
241
161
7.88:1 (87.3%)
4.70:1 (78.7%)
60/40
194
159
4.90:1 (79.6%)
4.82:1 (79.3%)
50/50
134
42
3.74:1 (73.3%)
3.74:1 (73.3%)
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
2
J (A/cm )
10
-5
10
-6
10
-7
10
-8
0.1 Fig. 5.
ACKNOWLEDGMENT 90/10 80/20 70/30 60/40 50/50
E=200 kV/cm 1
t (sec)
10
100
Transient leakage currents under E = 200 kV/cm. TABLE II
C OMPOSITIONAL PROPERTIES FOR THE FIVE FILMS GROWN WITH DIFFERENT A R /O 2 FLOW RATES AT THE SAME TOTAL GROWTH PRESSURE . Ar/O2 (sccm)
PO2 (mTorr)
Ba/Sr
90/10 80/20
4.5
0.94 0.93
This research was supported by the ARO through the Multifunctional Adaptive Radio Radar and Sensors program (MARRS MURI) under award number DAAD19-01-1-0496, by the ONR through the Center for Advanced Nitride Electronics program (CANE MURI) under award number N0014-01-1-0764, and by DARPA and DMEA through the Center for Nanoscience Innovation for Defense program (CNID) under award number DMEA90-02-2-0215.
(Ba+Sr)/Ti
R EFERENCES [1] B. Acikel, T. R. Taylor, P. J. Hansen, J. S. Speck, and R. A. York, IEEE Microwave Wireless Comp. Lett. 12, 237 (2002). [2] S. W. Kirchoefer, E. J. Cukauskas, N. S. Barker, H. S. Newman, and W. Chang, Appl. Phys. Lett. 80, 1255 (2002). [3] J. Im, O. Auciello, S. K. Streiffer, Thin Solid Films 413 243 (2002). [4] A. S. Nagra and R. A. York, IEEE Tran. Microwave Theory Tech. 49, 1705 (1999). [5] S. Stemmer, S. K. Streiffer, N. D. Browning, and A. I. Kingon, Appl. Phys. Lett. 74, 2432 (1999).
1.00 0.93 0.88
70/30
9.0 13.5
0.92
60/40
18.0
0.89
0.82
50/50
22.5
0.87
0.78
we used to define the useful tuning range. Figure 5 shows that there was no further decrease in leakage current for oxygen partial pressures above 13.5 mTorr (70/30 sccm). In these films the failure mechanism is likely different, and independent from the onset of high leakage currents that we traditionally associate with Q-rollover. According to the RBS analysis results in Table 2 the best film in the series (80/20) has a small amount of excess Ti. A possible explanation for the observed improvement is that the excess Ti passivates the film’s grain boundaries. The low Q-factor of the 50/50 (22.5 mTorr) film may be due to the accommodation of significant quantities of excess Ti not only on the grain boundaries, but also within the grains [5]. Further studies are currently underway to determine the effect of this significant titanium non-stoichiometry upon the film crystallinity and microstructure. IV. C ONCLUSION In this work we have investigated the effects of the oxygen partial pressure during film deposition, and the relationship between this variation in growth conditions and the electrical properties of the resulting BST parallel plate capacitors. We have found that by slightly increasing the oxygen partial pressure beyond what would be required for the most stoichiometric films we were able to produce devices with superior electrical properties. Our optimized condition of 80/20 Ar/O2 produced films with a 100 MHz maximum tunability of 13.71:1 (92.7%) at 4.7 MV/cm and a 100 MHz useful tunability of 9.38:1 (89.3%) at 2.8 MV/cm.
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
280
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
I. I NTRODUCTION In recent years there has been much interest in thin-film barium strontium titanate (BST) for tunable components for RF and microwave applications [1], [2]. The applications for tunable dielectrics range from applications where films with low loss-tangents and small tunabilities are required to applications where films with high tunabilities and moderate loss-tangents are desirable. While recent work has focused on optimization of BST for low-loss applications [3], the advantages of growth optimization for high tunability applications have been largely unrealized. At high frequencies, ohmic losses from electrodes and circuit metalization dominate circuit loss, reducing the benefit of a material with an intrinsic low loss-tangent. High tunability applications require cascaded tuning elements. Designs employing high tunability films require fewer tuning elements, thereby reducing the amount of circuit metalization and associated ohmic loss. Phase shifters are an example of one such application, where a predetermined amount of phase delay must be implemented by cascading phase shifting units.
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
An increase in tunability results in an increase in the amount of phase shift per unit, and therefore a reduction in the total number of units required for the circuit implementation [4]. In this experiment we evaluated the effect of variations in the partial pressure of oxygen during sputter deposition on the electrical properties of thin film BST capacitors with the goal of maximizing tunability and dielectric strength. II. E XPERIMENTAL A series of films were grown on both platinized and bare c-plane sapphire substrates with backside Ti layers. The platinized sapphire substrates were for the fabrication of devices for electrical measurements whereas films grown on the bare sapphire substrates were used for Rutherford backscattering (RBS) analysis. During each growth, half-wafers of both types of substrates were used. The surface temperature of the platinized sapphire was 700 ◦ C during deposition for all growths. The sputtering targets were of two slightly different compositions: Ba0.5 Sr0.5 TiO3 and Ba0.5 Sr0.5 Ti1.02 O3 . The RF power on each gun was 150 W. The sputtering ambient for all growths was a mixture of Ar and O2 at a total pressure of 45 mTorr and a combined total flow rate of 100 sccm. The Ar/O2 flow rates for the series were 90/10, 80/20, 70/30, 60/40, and 50/50 sccm corresponding to oxygen partial pressures of 4.5, 9.0, 13.5, 18.0, and 22.5 mTorr. The growth time for all depositions was 64 minutes, resulting in film thicknesses ranging from 113 nm to 145 nm as measured by surface profilometry. The lower oxygen partial pressures corresponded to higher deposition rates. Parallel plate capacitors were fabricated using the films deposited on platinized sapphire. The 200 nm Pt ground planes on the substrates were deposited by e-beam evaporation prior to film growth. Buffered HF was used for a mesa isolation etch, after which top electrode and ground plane contacts were formed by lift-off. The top electrode metalization layer consisted of 150 nm of Pt deposited by e-beam evaporation. Capacitance and Q-factor (inverse loss-tangent) measurements were performed on 40 x 50 µm2 devices using an Agilent 4294A impedance analyzer for low voltage measurements (≤ 40 V) at 1 MHz and 100 MHz. High voltage capacitance measurements were performed at 100 MHz using an HP 8722D network analyzer with an external bias tee. A Keithley 6517A electrometer was used to source the high
278
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
600
350
90/10 80/20 70/30 60/40 50/50
εr
400 300
Fig. 1.
250 200 150
200
100
100
50
0
-3
-2
-1 0 1 E (MV/cm)
2
90/10 80/20 70/30 60/40 50/50
300
Q
500
0 -4 -3 -2 -1 0 1 2 E (MV/cm)
3 Fig. 3.
1 MHz r -E tuning curves for different Ar/O2 ratios (sccm).
C
Q
-11
140
7 10
90/10 500 13.71:1 @ 80/20 80/20 breakdown 400 11.36:1 @ 300 90/10 breakdown 200
-11
6 10
-11
5 10
C (F)
maximum tunability 10.15:1
-11
4 10
120
useful tunability 5.36:1
100 80
Q
εr
4
1 MHz Q-E curves for different Ar/O2 ratios (sccm).
600
-11
60
-11
40
-11
20
3 10 2 10
1 10
100 0 -5
3
0 -35
-4
-3 -2 E (MV/cm)
-1
-30
-25
-20
-15
-10
-5
0
0
V
0 Fig. 4.
Maximum and useful tunabilities.
Fig. 2. 100 MHz maximum tunability curves for 90/10 and 80/20 sccm Ar/O2 films.
voltages while an external series resistance of approximately 50 kΩ was used for current limiting. An Agilent 4155B semiconductor parameter analyzer was used for DC leakage current measurements. Compositional data was determined by RBS analysis on the samples grown on bare sapphire. III. R ESULTS AND D ISCUSSION Permittivity tuning curves in Figure 1 show that the film with the lowest oxygen partial pressure had the highest zerobias permittivity. All of the films in the series were observed to approach the same high-field permittivity limit, but some of the devices failed before reaching this limit. In the absence of such device failure, we would this expect the film with the highest zero-bias permittivity to have the highest tunability. The highest tunabilities were actually measured for the film with the second highest permittivity. In this case the increase in the film’s dielectrics strength compensated for the reduction in permittivity, as seen in Figure 2. Figure 3 shows how the Q-factor changed with bias for the films with the four lowest oxygen partial pressures during growth. When interpreting tuning curves, it is important to differentiate between the maximum tuning range (device destruction) and the useful tuning range — the range useful for circuit implementations. Figure 4 shows how the tuning curves together with the Q-factors can be used to define a maximum tunability and useful tunability for devices from
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
each film. The useful tuning range is limited by Q-rollover, which is associated with an exponential increase in leakage currents preceding device failure. From the results in Figure 3, we can see that the useful tuning range increases with higher oxygen partial pressures. From Figure 1 we see that the zero-bias permittivity drops significantly when the oxygen partial pressure is increased above 9.0 mTorr (80/20 sccm). Thus we are not surprised when we find in Table 1 that the 80/20 film has a higher useful tunability than the 90/10 film. The data in Table 1 shows that for the lower permittivity films the useful tunability approaches the maximum tunability. This is because the rough measure we use to define useful tunability fails in these cases because catastrophic device failure can occur without the Q-rollover
279
TABLE I E LECTRICAL PROPERTIES FOR THE FIVE FILMS GROWN WITH DIFFERENT A R /O2 FLOW RATES AT THE SAME TOTAL GROWTH PRESSURE .
Ar/O2 (sccm) 90/10
r
Q
Maximum Tunability
Useful Tunability
571
(1 MHz) 41
(100 MHz) 11.36:1 (91.2%)
(100 MHz) 6.79:1 (85.3%)
80/20
455
75
13.71:1 (92.7%)
9.38:1 (89.3%)
70/30
241
161
7.88:1 (87.3%)
4.70:1 (78.7%)
60/40
194
159
4.90:1 (79.6%)
4.82:1 (79.3%)
50/50
134
42
3.74:1 (73.3%)
3.74:1 (73.3%)
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
2
J (A/cm )
10
-5
10
-6
10
-7
10
-8
0.1 Fig. 5.
ACKNOWLEDGMENT 90/10 80/20 70/30 60/40 50/50
E=200 kV/cm 1
t (sec)
10
100
Transient leakage currents under E = 200 kV/cm. TABLE II
C OMPOSITIONAL PROPERTIES FOR THE FIVE FILMS GROWN WITH DIFFERENT A R /O 2 FLOW RATES AT THE SAME TOTAL GROWTH PRESSURE . Ar/O2 (sccm)
PO2 (mTorr)
Ba/Sr
90/10 80/20
4.5
0.94 0.93
This research was supported by the ARO through the Multifunctional Adaptive Radio Radar and Sensors program (MARRS MURI) under award number DAAD19-01-1-0496, by the ONR through the Center for Advanced Nitride Electronics program (CANE MURI) under award number N0014-01-1-0764, and by DARPA and DMEA through the Center for Nanoscience Innovation for Defense program (CNID) under award number DMEA90-02-2-0215.
(Ba+Sr)/Ti
R EFERENCES [1] B. Acikel, T. R. Taylor, P. J. Hansen, J. S. Speck, and R. A. York, IEEE Microwave Wireless Comp. Lett. 12, 237 (2002). [2] S. W. Kirchoefer, E. J. Cukauskas, N. S. Barker, H. S. Newman, and W. Chang, Appl. Phys. Lett. 80, 1255 (2002). [3] J. Im, O. Auciello, S. K. Streiffer, Thin Solid Films 413 243 (2002). [4] A. S. Nagra and R. A. York, IEEE Tran. Microwave Theory Tech. 49, 1705 (1999). [5] S. Stemmer, S. K. Streiffer, N. D. Browning, and A. I. Kingon, Appl. Phys. Lett. 74, 2432 (1999).
1.00 0.93 0.88
70/30
9.0 13.5
0.92
60/40
18.0
0.89
0.82
50/50
22.5
0.87
0.78
we used to define the useful tuning range. Figure 5 shows that there was no further decrease in leakage current for oxygen partial pressures above 13.5 mTorr (70/30 sccm). In these films the failure mechanism is likely different, and independent from the onset of high leakage currents that we traditionally associate with Q-rollover. According to the RBS analysis results in Table 2 the best film in the series (80/20) has a small amount of excess Ti. A possible explanation for the observed improvement is that the excess Ti passivates the film’s grain boundaries. The low Q-factor of the 50/50 (22.5 mTorr) film may be due to the accommodation of significant quantities of excess Ti not only on the grain boundaries, but also within the grains [5]. Further studies are currently underway to determine the effect of this significant titanium non-stoichiometry upon the film crystallinity and microstructure. IV. C ONCLUSION In this work we have investigated the effects of the oxygen partial pressure during film deposition, and the relationship between this variation in growth conditions and the electrical properties of the resulting BST parallel plate capacitors. We have found that by slightly increasing the oxygen partial pressure beyond what would be required for the most stoichiometric films we were able to produce devices with superior electrical properties. Our optimized condition of 80/20 Ar/O2 produced films with a 100 MHz maximum tunability of 13.71:1 (92.7%) at 4.7 MV/cm and a 100 MHz useful tunability of 9.38:1 (89.3%) at 2.8 MV/cm.
0-7803-8410-5/04/$20.00 (c)2004 IEEE.
280
2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference
