- Email: [email protected]
Initial performance results from the NV1002 high energy ion implanter
Nuclear
Instruments
and Methods
in Physics
Research
413
B55 (1991) 473-477
North-Holland
Initial performance results from the NV1002 high energy ion implanter E. McIntyre a, D. Balek a, P. Boisseau b, A. Dart a, A.S. Denholm a, H. Glavish ‘, C. Hayden L. Kaminski a, B. Libby a, N. Meyyappan a, J. O’Brien a, F. Sinclair a and K. Whaley a a Eaton Corporation, Semiconductor Equipment Division, 108 Cherry Hill Drive, Beverly, MA 01915, USA
a,
b PTC Inc., 259 Bishop’s Forest Drive, Waltham, MA 02154, USA ’ GMW Assoc., 1060 Lakeview Way, Redwood City, CA 94062, USA
ne Eaton NV1002 is a high energy ion implanter with beam current capability greater than 1 mA. Acceleration to energies between 80 and 2000 keV is achieved with a variable phase linear accelerator (linac). The first production NV1002 is being tested, and the initial results are reported. Currents of l-2 mA can be generated over an energy range of 40-1000 keV for boron, phosphorus, and arsenic. Useful currents are available at energies as low as 10 keV, and using doubly charged ions, as high as 2 MeV. Considerations for a commercial implanter such as ease of operation are discussed. Analyses of implanted wafers are presented, demonstrating uniformity, correct implant depth profiling, and freedom from contamination.
1. Introduction Several years ago, Eaton initiated a study to dethe best method for generating high energy ions for ion implantation. It was concluded that rf linear acceleration was the preferred approach because it allowed the use of all relevant dopants, was flexible in energy, and was simple to operate. During the course of the study, the variable phase linear accelerator concept was developed [l]. Major advantages over the alternative dc approaches exist since there are no voltages in the linac system above 80 kV and a conventional ion source can be used. Also, since the linac block is at ground potential, access is easy all along its length for pumping and the control of beam optics. These considerations led to the rapid development of the NV1000 implanter, which has been a technically successful system capable of generating beam currents up to 1 mA [2]. Subsequently, it was recognized that several improvements could be made to reduce the machine footprint.and increase beam current capability. This led to the development of the NV1002. The NV1002 linac has been designed primarily to accelerate singly charged ions in the mass range from 11 to 75 amu. It is, however, useful outside that range, though with lesser acceleration capability. For example, Hf should be accelerated to > 500 keV, and since doubly charged ions behave in the linac like particles of half their mass ions up to Sb2’ are also accelerated. The NV1002 ules the same sources as Eaton’s other implanters, and utilizes the Eaton NV20A robotic endstation. The automated software control of the source, injector, linac and robotics is consistent with the generic termine
0168-583X/91/$03.50
0 1991 - Elsevier Science Publishers
system developed for the NV20A. Such cross-model commonality has obvious advantages. The NV1002 combines high energy (1 MeV) and low energy (lo-80 keV) performance in a single machine. This article presents results from the initial performance tests of the first installed NV1002.
2. Beam current and energy capabilities The principle of rf linear acceleration with variable phase control and the design of the NV1002 have been described earlier [3-51. The present design specifications of the NV1002 include an energy range of 80-1000 keV for singly charged ions, current capability of 1 mA, and mass acceptance from B2+ to Sb”. These goals have been achieved and in some cases surpassed. Information from the initial tests are summarized below, beginning with the low energy range which uses only the dc injector stage of the system. In the energy range below 80 keV, the NV1002 performance is similar to that of a conventional medium-current ion implanter. The ion beam is accelerated to final energy in the dc injector stage of the machine. The beam passes through the linac with all of the rf cavities off. Radial focussing through the linac is maintained using the small electrostatic quadrupole lenses which are between each cavity. For energies from 10 to 80 keV sufficient current is always available from the injector to optimize transmission through the linac. Maximum beam currents are limited by increased space charge effects in the linac
B.V. (North-Holland)
V. MACHINES
474
E. McIntyre
et al. / The NV1002
which cannot be overcome by the electrostatic quadrupoles. For a given beam energy Ebeam and ion mass these space charge effects lead to final beam Mien> currents that vary more or less in accordance with (Ez&JMk&,). A series of beam tests were performed on the first NV1002. At each of several energies, and for three different ion species, the current injected into the linac was increased until no substantial increase in final transmitted current was observed. While there is more testing to be done, the currents observed to date increase as expected and are displayed in fig. 1. Ion energies above 80 keV are achieved through the use or one or more of the twelve cavities of the rf linac. The first cavities are used to form the 80 keV injected dc beam into ion “bunches” at the fundamental operating frequency of the machine (6.78 MHz). In the simplest bunching scheme, the conversion of the dc beam into bunches is done with a typical efficiency of about 25%. This bunching/transmission efficiency (E) is defined as the ratio of the total beam current at the entrance of the linac to the final accelerated and energy-analyzed beam current delivered to the endstation. The ratio reflects the capture efficiency of the bunching cavities, limitations on the focussing power of the quadrupole lenses and beam dynamical losses along the linac. In some instances it includes a small beam loss due to mechanical collimators at the linac entrance. In practice, these efficiencies for the NV1002 vary from 20-35%. The determination of optimized acceleration parameters (namely phases and voltage amplitudes for each cavity and quadrupole voltage settings) for the NV1002 has been made using our simulation program
q
q
0
0.
20.
0
I
1
I
40
60
80
100
Beam Energy (keVl
Fig. 1. The beam intensities (mk) of B+, P+ and AS+ delivered through the linac in low energy dc mode.
high energy ion implanter 0.35
+
0.3 0 + +
g 5 c .$ E w s
+
0.25
+
+ + q
73 q
q q
-t
0.2
‘D .B
0
ii E
0.15
g P E e
0.1
0 As+
m’ + ‘\
P+
80 keV
I
I
I
I
200
400
600
800
1
IO
Beam Energy (keV)
Fig. 2. The bunching/transmissionefficiency (c) of the linac as a function or energy from 10 to 800 keV. Below 80 keV, the transmission rises to almost 30%. Above 80 keV, the transmission falls, but recovers at about 200 keV, and remains essentially constant.
LINAC [2]. Data sets of the acceleration parameters are available for a wide range of species and energy and are stored for immediate recall in the control system. Additional bunching techniques are being developed to capture more of the dc beam. Higher bunching capture efficiency will be useful when the optimization of doubly charged ion currents is investigated. In the energy range just above 80 keV, where rf acceleration begins, the maximum current transmitted through the operating linac drops below the maximum current transmitted in the 80 keV dc mode. This effect is clearly seen in fig. 2 for PC and As+ beams and is due to the bunched nature of the rf beam, where the increased peak space charge density reduces transmission through the linac because of radial expansion of the bunches. Above 200 keV, the increased beam energy compensates for the high peak currents. In this higher energy range the maximum current limits remain fairly flat up to the highest energy of the machine, because the amount of charge which can be accommodated in a single beam pulse (and thus the current) is determined in the bunching process early in the linac. Fortunately, two techniques should allow significantly improved transmission in the 80 to 200 keV range. First, acceleration of doubly charged ions in the dc mode can provide relatively high currents up to 160 keV. Second, operation of the linac in an accelerate/decelerate mode may allow beams with energy greater than 200 keV to be decelerated at the end of the linac into the SO-200 keV
E. McIntyre
et al. / The NV1002 high energy ion irnplanter
range. The accel/decel mode has not been demonstrated to date and the outcome is likely to be dependent on the final energy spread that is acceptable. The first NV1002 has not yet been fully characterized, however table 1 shows the results of initial measurements for a variety of ions and energies. The injected and transmitted current, as well as the bunching/transmission efficiency (c) is shown. It should be noted that in most cases listed here, the injected current was not maximized, so that the table does not represent the maximum performance of the machine. Singly charged As’ beams of about 450 PA have been generated from 200 to 800 kev. This As+ data was collected
Table 1 Initial results of injected and transmitted currents for a variety of species and energies. In many cases more transmitted current could be obtained by increasing the injected current. Note that results for Pz+ and B2+ are given in charge mA. Species
Energy kV1
Injected current
Transmitted current
B/T efficiency
ImAl
LmAl
(6)
200 500 800 800 950
4.52 4.16 4.12 _ _
0.83 1.02 1.03 1.5 1.08
0.18 0.24 0.25
P* P+ P+ Pf P’ P+ P+ P+ P+ P’
200 280 300 350 400 500 600 770 800 880 900 1000 1000
2.6 2.47 2.5 2.90 4.37 5.5 _
0.20 0.27 0.26 0.24 0.27 0.20 _
4.15 10.0
0.527 0.666 0.650 0.702 1.19 1.1 1.0 1.37 1.1 1.88 1.00 1.28 1.8
AS+ AS+ As+ As’ AS’ As+ As+
200 300 400 500 600 700 800
1.98 1.99 2.4 2.0 2.0 2.0 2.06
0.350 0.580 0.500 0.443 0.516 0.416 0.466
ArC
1000
-
1.0
N;i
464
7.25
2.0
P2+ B2+
1900 1400
_ _
0.450 0.100
B+ B’ B+ B+ P+ ;I
5.7 4.4 9.1 _
0.24 0.25 0.20 0.21 0.18 0.17 0.29 0.21 0.22 0.26 0.21 0.23
0.28
415
with a relatively low injected current and higher transmitted values are obtainable with higher injected currents. B+ beams of at least 1 mA were produced at 500 and 950 keV and 1.5 mA was produced at 800 keV. Singly charged phosphorus beams of at least 1 mA and ranging up to 1.8 mA have been generated in the 400 to 1000 kev range. Although the general current specification of the NV1002 is 1 mA, the goal has been to develop stable 2 mA beams over a broad range of species and energies. A 2 mA beam of NT has been produced. At 770, 880 and 1000 keV, stable P+ beams up to 1.8 mA were generated by increasing the injected current. At these high currents (and corresponding high injected currents) transmission efficiency drops significantly and loading on the electrostatic quadrupole supplies becomes significant. (Compare the transmitted currents for 4.75 and 10 mA of injected P*). Achievement of 2 mA capability in normal operation will require modification of injection optics and an increase in the power handling capabilities of the quadrupole lenses. Finally, the use of doubly charged ions allows acceleration to energies above 1 MeV. Currents of 450 charge PA of 1.9 MeV Pzt and 100 charge I_LAof 1.4 MeV B2+ have been produced.
3. Linac o~r~tion At full energy, all twelve acceleration stages of the linac are under power. Each stage requires three control parameters: rf voltage, rf phase, and electrostatic quadrupole voltage. Thus for a given ion species and final energy, there are as many as 36 parameters to be set for proper operation. These parameters constitute a Linac Data Set. The NV1002 control system contains data sets for boron, phosphorus, and arsenic for energies from 100 to 1000 keV in 100 keV increments. Data sets at intermediate energies may be generated in a few minutes by manual adjustment of the machine parameters and then added to the database. In normal operation, the source/injector stage and linac are ramped to the desired settings simultaneously. The dc beam is injected into the linac, and high energy beam current is observed in the final Faraday cup. If source parameters and total current are similar to the conditions under which the stored data set was originally created, > 90% of the optimized beam current is typically available without fine tuning. Linac fine tuning to optimize final current is fully automated, allowing peak beam currents to be achieved in l-2 min. From the experience with the NV1000 linac and through the use of many proven NV20A components and subsystems, the NV1002 should be a very reliable implanter. The first period of operation is confirming this belief. V. MACHINES
476
E. McIntyre et al. / The NV1002 high energy ion implanter
4. Implanted wafer analysis Good implant u~fo~ty is a critical feature of any implanter. As a check on uniformity, implants were done on 200 mm wafers with 1.2 mA of 900 keV P+. Wafers were dosed to 5 x 1O*3 crnm2 using a tilt of 7O and a twist of O”. Fig. 3 shows a contour map of resistivity obtained from a Prometrix Omnimap RS20 for a pre~o~~zed wafer. This shows excellent uniformity, with e < 0.5%. Additionally, the mean sheet resistivity is as expected for the dose. The implant depth profile was examined through spreading resistance measurements as well as via SIMS (secondary ion mass spectrometry) analysis. Fig. 4 is a spreading resistance measurement for a wafer implanted with 500 keV BC. The peak of the concentration curve is just over 1 pm, consistent with theoretical and measured ranges of 500 keV Bt [6]. Fig. 5 is a SIMS analysis of a 500 keV Pt implant. It has a peak concentration at a depth of about 0.6 pm and a straggling width near 0.17 pm. This is consistent with measured values [6] of straggling. Wafers were also analyzed for impurities. Fig. 6 shows SIMS analysis of a wafer implanted with 1.4 MeV B+ for the presence of B, Na, Al, Cr, Fe and MO. The concentrations measures are accurate to about 20% and the depth to within 10%. All the impurities are small, have a flat
;__;.----y------_ ~....-+, > (.I’
..._’\A
--,. --....
J 3
OATE: 336,OO FILE Y sJEBo434 SOURCE: EATON
PROBE LOAD BEVEL AtaLE
7.5 0 Bon
OFOENTA~ON rlooa SI lO!lm !nEP 1EIcREr.4 sm SAMPLE t 7.6 ‘I 1014cm” spz xmlc! SHEET =
~pp.66o.a
Fig. 4. A spreading resistance measurement for a 500 keV B4 implant.
‘._,, ‘..,
I ..’
0
Fig. 3. A sheet resistivity map of a 900 keV P+ implanted wafer. Uniformity is good with (TQ 0.5%.
<
..,I,
2 DEFTH(microns)
*
3
I
Fig. 5. A SIMS analysis of a 500 keV P+ implant. The peak concentration is near 0.6 ym.
E. McIntyre
et al. / The NV1002
high energy ion implanter
411
tivity. Injected contaminants as well as contaminants created in the linac will be out of phase and gain little energy. Finally, the 51° post acceleration energy analysis will refine the mass purity further. 5. Conclusions The NV1002, an rf linac based implanter, is a robust, flexible, and simple to operate tool for high energy implantation. It has been demonstrated to generate milliampere beams of boron to arsenic in the MeV energy range. Low energy capability below 80 keV is similar to standard medium current implanters. Uniform, low contamination implants of the expected depth profile have been generated. Because of scheduling considerations, there was no prototype for the NV1002 and the first system to be made was recently shipped. Its initial performance characteristics are very encouraging and final testing will determine the full power of the NV1002. 1
2 DEPTH
3
4
5
References
(microns)
Fig. 6. A SIMS analysis of a 1400 keV Bf tamination is low.
implant. Con-
distribution in depth (and are probably dominated by instrumental backgrounds). The peak of the boron distribution is just over 2 pm deep, again consistent with measured and calculated values [6]. Such low contamination might be expected. The injector stage mass analysis prevents unwanted species from entering the linac. The linac itself has mass selec-
[l] H.F. Glavish, A. S. Denholm and G.K. Simcox, US Patent 4667111, 1987. [2] H.F. Glavish, D. Bernhardt, P.Boisseau, B. Libby, G. Simcox and A.S. Denholm, Nucl. Instr. and Meth. B21 (1987) 264. [3] H.F. Glavish, Nucl. Instr. and Meth. B21 (1987) 218. [4] J P. Boisseau, A.S. Denholm, H.F.Glavish and G. Simcox, Mater. Sei. Eng. B2 (1989) 223. [5] P.Boisseau, A. Dart, A.S. Denholm, H.F. Glavish, B.Libby, G. Simcox, Nucl. Instr. Meth. B37/38 (1989) 591. [6] J.F. Zeigler, J.P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon, New York, 1984).
V. MACHINES
Instruments
and Methods
in Physics
Research
413
B55 (1991) 473-477
North-Holland
Initial performance results from the NV1002 high energy ion implanter E. McIntyre a, D. Balek a, P. Boisseau b, A. Dart a, A.S. Denholm a, H. Glavish ‘, C. Hayden L. Kaminski a, B. Libby a, N. Meyyappan a, J. O’Brien a, F. Sinclair a and K. Whaley a a Eaton Corporation, Semiconductor Equipment Division, 108 Cherry Hill Drive, Beverly, MA 01915, USA
a,
b PTC Inc., 259 Bishop’s Forest Drive, Waltham, MA 02154, USA ’ GMW Assoc., 1060 Lakeview Way, Redwood City, CA 94062, USA
ne Eaton NV1002 is a high energy ion implanter with beam current capability greater than 1 mA. Acceleration to energies between 80 and 2000 keV is achieved with a variable phase linear accelerator (linac). The first production NV1002 is being tested, and the initial results are reported. Currents of l-2 mA can be generated over an energy range of 40-1000 keV for boron, phosphorus, and arsenic. Useful currents are available at energies as low as 10 keV, and using doubly charged ions, as high as 2 MeV. Considerations for a commercial implanter such as ease of operation are discussed. Analyses of implanted wafers are presented, demonstrating uniformity, correct implant depth profiling, and freedom from contamination.
1. Introduction Several years ago, Eaton initiated a study to dethe best method for generating high energy ions for ion implantation. It was concluded that rf linear acceleration was the preferred approach because it allowed the use of all relevant dopants, was flexible in energy, and was simple to operate. During the course of the study, the variable phase linear accelerator concept was developed [l]. Major advantages over the alternative dc approaches exist since there are no voltages in the linac system above 80 kV and a conventional ion source can be used. Also, since the linac block is at ground potential, access is easy all along its length for pumping and the control of beam optics. These considerations led to the rapid development of the NV1000 implanter, which has been a technically successful system capable of generating beam currents up to 1 mA [2]. Subsequently, it was recognized that several improvements could be made to reduce the machine footprint.and increase beam current capability. This led to the development of the NV1002. The NV1002 linac has been designed primarily to accelerate singly charged ions in the mass range from 11 to 75 amu. It is, however, useful outside that range, though with lesser acceleration capability. For example, Hf should be accelerated to > 500 keV, and since doubly charged ions behave in the linac like particles of half their mass ions up to Sb2’ are also accelerated. The NV1002 ules the same sources as Eaton’s other implanters, and utilizes the Eaton NV20A robotic endstation. The automated software control of the source, injector, linac and robotics is consistent with the generic termine
0168-583X/91/$03.50
0 1991 - Elsevier Science Publishers
system developed for the NV20A. Such cross-model commonality has obvious advantages. The NV1002 combines high energy (1 MeV) and low energy (lo-80 keV) performance in a single machine. This article presents results from the initial performance tests of the first installed NV1002.
2. Beam current and energy capabilities The principle of rf linear acceleration with variable phase control and the design of the NV1002 have been described earlier [3-51. The present design specifications of the NV1002 include an energy range of 80-1000 keV for singly charged ions, current capability of 1 mA, and mass acceptance from B2+ to Sb”. These goals have been achieved and in some cases surpassed. Information from the initial tests are summarized below, beginning with the low energy range which uses only the dc injector stage of the system. In the energy range below 80 keV, the NV1002 performance is similar to that of a conventional medium-current ion implanter. The ion beam is accelerated to final energy in the dc injector stage of the machine. The beam passes through the linac with all of the rf cavities off. Radial focussing through the linac is maintained using the small electrostatic quadrupole lenses which are between each cavity. For energies from 10 to 80 keV sufficient current is always available from the injector to optimize transmission through the linac. Maximum beam currents are limited by increased space charge effects in the linac
B.V. (North-Holland)
V. MACHINES
474
E. McIntyre
et al. / The NV1002
which cannot be overcome by the electrostatic quadrupoles. For a given beam energy Ebeam and ion mass these space charge effects lead to final beam Mien> currents that vary more or less in accordance with (Ez&JMk&,). A series of beam tests were performed on the first NV1002. At each of several energies, and for three different ion species, the current injected into the linac was increased until no substantial increase in final transmitted current was observed. While there is more testing to be done, the currents observed to date increase as expected and are displayed in fig. 1. Ion energies above 80 keV are achieved through the use or one or more of the twelve cavities of the rf linac. The first cavities are used to form the 80 keV injected dc beam into ion “bunches” at the fundamental operating frequency of the machine (6.78 MHz). In the simplest bunching scheme, the conversion of the dc beam into bunches is done with a typical efficiency of about 25%. This bunching/transmission efficiency (E) is defined as the ratio of the total beam current at the entrance of the linac to the final accelerated and energy-analyzed beam current delivered to the endstation. The ratio reflects the capture efficiency of the bunching cavities, limitations on the focussing power of the quadrupole lenses and beam dynamical losses along the linac. In some instances it includes a small beam loss due to mechanical collimators at the linac entrance. In practice, these efficiencies for the NV1002 vary from 20-35%. The determination of optimized acceleration parameters (namely phases and voltage amplitudes for each cavity and quadrupole voltage settings) for the NV1002 has been made using our simulation program
q
q
0
0.
20.
0
I
1
I
40
60
80
100
Beam Energy (keVl
Fig. 1. The beam intensities (mk) of B+, P+ and AS+ delivered through the linac in low energy dc mode.
high energy ion implanter 0.35
+
0.3 0 + +
g 5 c .$ E w s
+
0.25
+
+ + q
73 q
q q
-t
0.2
‘D .B
0
ii E
0.15
g P E e
0.1
0 As+
m’ + ‘\
P+
80 keV
I
I
I
I
200
400
600
800
1
IO
Beam Energy (keV)
Fig. 2. The bunching/transmissionefficiency (c) of the linac as a function or energy from 10 to 800 keV. Below 80 keV, the transmission rises to almost 30%. Above 80 keV, the transmission falls, but recovers at about 200 keV, and remains essentially constant.
LINAC [2]. Data sets of the acceleration parameters are available for a wide range of species and energy and are stored for immediate recall in the control system. Additional bunching techniques are being developed to capture more of the dc beam. Higher bunching capture efficiency will be useful when the optimization of doubly charged ion currents is investigated. In the energy range just above 80 keV, where rf acceleration begins, the maximum current transmitted through the operating linac drops below the maximum current transmitted in the 80 keV dc mode. This effect is clearly seen in fig. 2 for PC and As+ beams and is due to the bunched nature of the rf beam, where the increased peak space charge density reduces transmission through the linac because of radial expansion of the bunches. Above 200 keV, the increased beam energy compensates for the high peak currents. In this higher energy range the maximum current limits remain fairly flat up to the highest energy of the machine, because the amount of charge which can be accommodated in a single beam pulse (and thus the current) is determined in the bunching process early in the linac. Fortunately, two techniques should allow significantly improved transmission in the 80 to 200 keV range. First, acceleration of doubly charged ions in the dc mode can provide relatively high currents up to 160 keV. Second, operation of the linac in an accelerate/decelerate mode may allow beams with energy greater than 200 keV to be decelerated at the end of the linac into the SO-200 keV
E. McIntyre
et al. / The NV1002 high energy ion irnplanter
range. The accel/decel mode has not been demonstrated to date and the outcome is likely to be dependent on the final energy spread that is acceptable. The first NV1002 has not yet been fully characterized, however table 1 shows the results of initial measurements for a variety of ions and energies. The injected and transmitted current, as well as the bunching/transmission efficiency (c) is shown. It should be noted that in most cases listed here, the injected current was not maximized, so that the table does not represent the maximum performance of the machine. Singly charged As’ beams of about 450 PA have been generated from 200 to 800 kev. This As+ data was collected
Table 1 Initial results of injected and transmitted currents for a variety of species and energies. In many cases more transmitted current could be obtained by increasing the injected current. Note that results for Pz+ and B2+ are given in charge mA. Species
Energy kV1
Injected current
Transmitted current
B/T efficiency
ImAl
LmAl
(6)
200 500 800 800 950
4.52 4.16 4.12 _ _
0.83 1.02 1.03 1.5 1.08
0.18 0.24 0.25
P* P+ P+ Pf P’ P+ P+ P+ P+ P’
200 280 300 350 400 500 600 770 800 880 900 1000 1000
2.6 2.47 2.5 2.90 4.37 5.5 _
0.20 0.27 0.26 0.24 0.27 0.20 _
4.15 10.0
0.527 0.666 0.650 0.702 1.19 1.1 1.0 1.37 1.1 1.88 1.00 1.28 1.8
AS+ AS+ As+ As’ AS’ As+ As+
200 300 400 500 600 700 800
1.98 1.99 2.4 2.0 2.0 2.0 2.06
0.350 0.580 0.500 0.443 0.516 0.416 0.466
ArC
1000
-
1.0
N;i
464
7.25
2.0
P2+ B2+
1900 1400
_ _
0.450 0.100
B+ B’ B+ B+ P+ ;I
5.7 4.4 9.1 _
0.24 0.25 0.20 0.21 0.18 0.17 0.29 0.21 0.22 0.26 0.21 0.23
0.28
415
with a relatively low injected current and higher transmitted values are obtainable with higher injected currents. B+ beams of at least 1 mA were produced at 500 and 950 keV and 1.5 mA was produced at 800 keV. Singly charged phosphorus beams of at least 1 mA and ranging up to 1.8 mA have been generated in the 400 to 1000 kev range. Although the general current specification of the NV1002 is 1 mA, the goal has been to develop stable 2 mA beams over a broad range of species and energies. A 2 mA beam of NT has been produced. At 770, 880 and 1000 keV, stable P+ beams up to 1.8 mA were generated by increasing the injected current. At these high currents (and corresponding high injected currents) transmission efficiency drops significantly and loading on the electrostatic quadrupole supplies becomes significant. (Compare the transmitted currents for 4.75 and 10 mA of injected P*). Achievement of 2 mA capability in normal operation will require modification of injection optics and an increase in the power handling capabilities of the quadrupole lenses. Finally, the use of doubly charged ions allows acceleration to energies above 1 MeV. Currents of 450 charge PA of 1.9 MeV Pzt and 100 charge I_LAof 1.4 MeV B2+ have been produced.
3. Linac o~r~tion At full energy, all twelve acceleration stages of the linac are under power. Each stage requires three control parameters: rf voltage, rf phase, and electrostatic quadrupole voltage. Thus for a given ion species and final energy, there are as many as 36 parameters to be set for proper operation. These parameters constitute a Linac Data Set. The NV1002 control system contains data sets for boron, phosphorus, and arsenic for energies from 100 to 1000 keV in 100 keV increments. Data sets at intermediate energies may be generated in a few minutes by manual adjustment of the machine parameters and then added to the database. In normal operation, the source/injector stage and linac are ramped to the desired settings simultaneously. The dc beam is injected into the linac, and high energy beam current is observed in the final Faraday cup. If source parameters and total current are similar to the conditions under which the stored data set was originally created, > 90% of the optimized beam current is typically available without fine tuning. Linac fine tuning to optimize final current is fully automated, allowing peak beam currents to be achieved in l-2 min. From the experience with the NV1000 linac and through the use of many proven NV20A components and subsystems, the NV1002 should be a very reliable implanter. The first period of operation is confirming this belief. V. MACHINES
476
E. McIntyre et al. / The NV1002 high energy ion implanter
4. Implanted wafer analysis Good implant u~fo~ty is a critical feature of any implanter. As a check on uniformity, implants were done on 200 mm wafers with 1.2 mA of 900 keV P+. Wafers were dosed to 5 x 1O*3 crnm2 using a tilt of 7O and a twist of O”. Fig. 3 shows a contour map of resistivity obtained from a Prometrix Omnimap RS20 for a pre~o~~zed wafer. This shows excellent uniformity, with e < 0.5%. Additionally, the mean sheet resistivity is as expected for the dose. The implant depth profile was examined through spreading resistance measurements as well as via SIMS (secondary ion mass spectrometry) analysis. Fig. 4 is a spreading resistance measurement for a wafer implanted with 500 keV BC. The peak of the concentration curve is just over 1 pm, consistent with theoretical and measured ranges of 500 keV Bt [6]. Fig. 5 is a SIMS analysis of a 500 keV Pt implant. It has a peak concentration at a depth of about 0.6 pm and a straggling width near 0.17 pm. This is consistent with measured values [6] of straggling. Wafers were also analyzed for impurities. Fig. 6 shows SIMS analysis of a wafer implanted with 1.4 MeV B+ for the presence of B, Na, Al, Cr, Fe and MO. The concentrations measures are accurate to about 20% and the depth to within 10%. All the impurities are small, have a flat
;__;.----y------_ ~....-+, > (.I’
..._’\A
--,. --....
J 3
OATE: 336,OO FILE Y sJEBo434 SOURCE: EATON
PROBE LOAD BEVEL AtaLE
7.5 0 Bon
OFOENTA~ON rlooa SI lO!lm !nEP 1EIcREr.4 sm SAMPLE t 7.6 ‘I 1014cm” spz xmlc! SHEET =
~pp.66o.a
Fig. 4. A spreading resistance measurement for a 500 keV B4 implant.
‘._,, ‘..,
I ..’
0
Fig. 3. A sheet resistivity map of a 900 keV P+ implanted wafer. Uniformity is good with (TQ 0.5%.
<
..,I,
2 DEFTH(microns)
*
3
I
Fig. 5. A SIMS analysis of a 500 keV P+ implant. The peak concentration is near 0.6 ym.
E. McIntyre
et al. / The NV1002
high energy ion implanter
411
tivity. Injected contaminants as well as contaminants created in the linac will be out of phase and gain little energy. Finally, the 51° post acceleration energy analysis will refine the mass purity further. 5. Conclusions The NV1002, an rf linac based implanter, is a robust, flexible, and simple to operate tool for high energy implantation. It has been demonstrated to generate milliampere beams of boron to arsenic in the MeV energy range. Low energy capability below 80 keV is similar to standard medium current implanters. Uniform, low contamination implants of the expected depth profile have been generated. Because of scheduling considerations, there was no prototype for the NV1002 and the first system to be made was recently shipped. Its initial performance characteristics are very encouraging and final testing will determine the full power of the NV1002. 1
2 DEPTH
3
4
5
References
(microns)
Fig. 6. A SIMS analysis of a 1400 keV Bf tamination is low.
implant. Con-
distribution in depth (and are probably dominated by instrumental backgrounds). The peak of the boron distribution is just over 2 pm deep, again consistent with measured and calculated values [6]. Such low contamination might be expected. The injector stage mass analysis prevents unwanted species from entering the linac. The linac itself has mass selec-
[l] H.F. Glavish, A. S. Denholm and G.K. Simcox, US Patent 4667111, 1987. [2] H.F. Glavish, D. Bernhardt, P.Boisseau, B. Libby, G. Simcox and A.S. Denholm, Nucl. Instr. and Meth. B21 (1987) 264. [3] H.F. Glavish, Nucl. Instr. and Meth. B21 (1987) 218. [4] J P. Boisseau, A.S. Denholm, H.F.Glavish and G. Simcox, Mater. Sei. Eng. B2 (1989) 223. [5] P.Boisseau, A. Dart, A.S. Denholm, H.F. Glavish, B.Libby, G. Simcox, Nucl. Instr. Meth. B37/38 (1989) 591. [6] J.F. Zeigler, J.P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon, New York, 1984).
V. MACHINES