ISSN 1063-7842, Technical Physics, 2007, Vol. 52, No. 6, pp. 770–775. © Pleiades Publishing, Ltd., 2007. Original Russian Text © V.A. Baturin, S.A. Eremin, S.A. Pustovoœtov, 2007, published in Zhurnal Tekhnicheskoœ Fiziki, 2007, Vol. 77, No. 6, pp. 93–98.
ELECTRON AND ION BEAMS, ACCELERATORS
Secondary Ion Mass Spectrometer Based on a High-Dose Ion Implanter V. A. Baturin, S. A. Eremin, and S. A. Pustovoœtov Institute of Applied Physics, National Academy of Sciences of Ukraine, ul. Petropavlovskaya 58, Sumy, 40030 Ukraine e-mail:
[email protected] Received July 27, 2006; in final form, November 7, 2006
Abstract—A secondary ion mass spectrometer built around a modified high-dose ion implanter is used to study secondary ion emission in metals over a wide range of primary beam energies. The implanter generates ion beams with energies of up to 150 keV and a substrate current to 30 µA. A modified MX7304A monopole mass spectrometer is applied as an analyzer of secondary ions with mass numbers of up to 400 with a resolution of 1 M at a level of 10% of the peak height. The detection limit for iron is 6.5 ppm. The analyzer is equipped with a small-size filter separating secondary ions in energy. The relative emission intensities of the secondary monatomic and cluster ions of copper for different primary ion beam parameters are studied. PACS numbers: 82.80.Ms DOI: 10.1134/S1063784207060163
INTRODUCTION Much attention has recently been given to methods of ion analysis, which are considered to be among the most informative and sensitive spectroscopic techniques. Collection and mass analysis of secondary ions produced in a solid by an accelerated ion beam constitute the basis for the method of secondary ion mass spectrometry (SIMS). The current of secondary ions of a particular chemical element is proportional to the secondary ion emission (SIE) ratio, which is the ratio between the number of secondary ions of a certain sort and the number of primary ions incident on the surface. Since this ratio is a function of the primary beam energy, it is reasonable to use the range of primary ion energies where the SIE ratio is maximal in order to raise the sensitivity to secondary ions of a certain sort. The energy dependences of this ratio for different targets were studied with the use of SIMS in [1, 2]. In all the targets, the SIE ratio increased with primary beam energy to a certain value depending on the sort of primary ions and the target material. This dependence is similar to the energy dependence of the sputtering ratio [3]. Specifically, for widely used Ar+ primary ions, the SIE ratio reaches a maximum in the range from 15 to 120 keV depending on the target material. It is therefore desirable to increase the primary ion energy to 100 keV and above in SIMS instruments. However, in conventional SIMS instruments, this is hardly feasible, since an increase in the primary ion energy entails a considerable complication of the ion formation column design
because of the need to use bulky isolators, sealed-off leads, and electrode gaps. Therefore, the primary beam energy is usually restricted to 20–30 keV [4]. In this study, we present a secondary ion mass spectrometer based on an ion implanter that is intended for high-dose implantation of semiconductors. The use of an ion implanter as a source of primary ions makes it possible to vary the ion energy over a wide range from 20 to 150 keV. The respective current density on the sample is on the order of 1 mA/cm2 . This ensures the dynamic, i.e., atom-by-atom, sputtering of the sample in contrast to the static sputtering, when at low primary current densities (of the order of 10–9–10–6 A/cm2), the species being sputtered are surface compounds and adsorbates. Impurities are removed from the primary ion beam by mass separation in an electromagnet, and neutrals are removed by deflecting the beam by a small angle in an electric field. The mass and energy analysis of secondary ions was performed with a modified MX7304A monopole mass spectrometer (AO SELMI, Sumy, Ukraine) equipped with a small-size energy filter. In the modified spectrometer, the residual gas ionizer was replaced by a sample holder. An advantage of using a monopole mass spectrometer in this case is its compactness and linear mass scale. In such an arrangement, ions with mass numbers of up to 400 can be analyzed, so that cluster ions can also be studied.
770
SECONDARY ION MASS SPECTROMETER
771
EXPERIMENTAL SETUP To apply the implanter as a source of primary ions for SIMS, it was modified. Figure 1 schematically shows the experimental setup. Ions are generated and a primary ion beam is formed in a duoplasmatron with incandescent cathode 1. For mass separation, the beam passes through a 90° sector electromagnet 2 with a radius of the central trajectory of 300 mm. For a 20-keV primary beam, the magnetic induction of the electromagnet allows separation of singly charged ions with mass numbers of up to 40. Separated ions are focused again with long-focal-length lens 3 operating in the acceleration mode to suppress spherical and chromatic aberrations. The ions are accelerated to desired energies in acceleration tube 4 (the ion source, as well as ion-separating and beam-forming systems, is under a high potential). The accelerated ion beam comes into chamber 5, which is under the Earth potential (the second chamber, which is intended to contain samples to be implanted, is not used in the modified version and is therefore omitted). The ion beam current is measured by Faraday cup 6 placed in the chamber. Diffusion pump 7 with a liquid-nitrogen trap evacuates the chamber to a residual pressure of 2 × 10−5 Pa. To improve the measurement accuracy, neutrals, which are not detected by the current meter but cause sputtering, should be cut off. For this purpose, the beam was deflected by a small angle in the horizontal plane with the help of plates 8 placed at the entrance to the chamber. The deflected beam strikes sample 9, which is placed at an angle of about 45° to the incident beam. Resulting secondary ions are extracted, focused into a beam by the ion optics of mass spectrometer 10, and applied to energy and mass filters. The detector of the mass spectrometer is screened from the beam plasma and X-ray radiation, which considerably increase the background and noise level and thereby lower the sensitivity of the device. In more detail, the path of the accelerated ions and the scheme of secondary ion formation and analysis are shown in Fig. 2. Before coming to sample 1, primary ions pass through input diaphragm 2, which forms a round beam with a diameter from 1 to 3 mm depending on the diameter of the aperture. The ion current incident on the sample is measured by a mobile Faraday cup placed behind the input diaphragm. This cup also acts as a shutter against the primary ions. After being extracted and focused by extracting and focusing electrodes 4 and 5, respectively, secondary ions enter into the energy filter. Both the input (6) and output (10) diaphragms of the energy filter are under a zero potential. The diameters of the input and output apertures are 1.6 mm. Energy separation is accomplished using deflection plates 8 and 9 shaped as 90° corners with filter-bandwidth-controlling slot diaphragm 7 in between. The diaphragm is under a zero potential. The distances from the centers of the input and output apertures to the diaTECHNICAL PHYSICS
Vol. 52
No. 6
2007
1
7
4
9
6
3
2
150 kV 10
5
8
Fig. 1. Schematic of the experimental setup (dashed lines show the trajectories of primary and secondary ions).
3 2 1 4 6
5
9
7
8 10
11
Fig. 2. Schematic of the secondary beam optics and the position of the sample (dashed lines show the primary and secondary ion trajectories).
phragm plane are 6 mm. The energy passband of the filter is tuned by varying positive potentials Udefl applied to the deflection plates. The characteristics of the energy filter were studied in [5]. In particular, the dependence of the energy to which the filter is tuned on deflecting potential Udefl was derived and the FWHM of the transmission spectrum was estimated for a width of the slot in the intermediate diaphragm of 1 mm (11% of the ion energy). Having passed through the energy filter, the secondary ions fall into a monopole mass analyzer 11, where they are detected by a VÉU-6 secondary electron multiplier (SEM) operating in the analog or count mode. The spectrometer is mounted on the side flange of the reception chamber. Computer control of
772
BATURIN et al.
Signal, V Ti+
TiO+
8 TiC+ 4 Fe+ 0 40
60
TiO+3 80
100
Ti2O+ Ti2O+2
2C 0 σ ( U ) -, C min = ---------------------U
120 M/Z
Fig. 3. SIMS spectrum taken of the titanium sample with 0.9% of iron impurity bombarded by Ar ions with energy EAr = 90 keV (IAr = 15 µA, the sample potential is 15 V, the analog mode of SEM operation). Some of the peaks considerably exceed the upper limit of the ADC scale, 10.24 V.
the mass spectrometer and spectrum recording is provided with application-specific software. THE PARAMETERS OF THE INSTRUMENT The main characteristics of a mass spectrometer are the resolving power and sensitivity threshold. It is known that the resolving power of quadrupole and monopole mass spectrometers decreases, while their sensitivity nonlinearly grows with an increase in the ion energy [6]. Generally, a 1-M resolution of peaks at a level of 10% of the peak height is sufficient except for special cases when very closely spaced peaks are needed to be resolved. In our case, the ion energies corresponding to this resolution lie in the range from 20 to 30 eV. The energy spectrum of secondary ions is continuous with a maximum at approximately 10 eV and a tail extending to several hundreds of electron volts. Therefore, for the maximum of the energy distribution to fall into the range from 20 to 30 eV, the metallic sample under study was kept under a positive potential varying from 10 to 20 V. For secondary ions of a given element that fall into the detector, their current is defined as +
I = I 0 Sβ Cf ,
The effect of the matrix makes quantitative analysis difficult and deteriorates its accuracy. Most often, such effects are estimated by semiempiric methods, because of which the error may exceed 100%. The situation is more favorable for quantitative analysis of microimpurities. Since the matrix properties are virtually constant over a wide range of impurity concentrations from 10–8 to 1 at % (and, hence, so is the matrix effect), the current of secondary ions generated by impurity atoms is proportional to the impurity concentration [8]. Therefore, the sensitivity threshold for a given element in a given matrix can be estimated by the formula
(1)
where I0 is the primary ion current, S is the sputtering ratio, β+ is the probability of ionization, C is the relative atomic concentration of the element, and f is the total transmission factor of the secondary ion channel. Since the secondary ion current is proportional to the probability of emitted particle ionization, the sensitivity of the method significantly depends on the chemical element and surrounding matrix. When positive ions are analyzed, the probability of ionization varies 104 to 105 times in passing from alkali metals to inertial gases and by a factor of 5 from matrix to matrix [7].
(2)
where C0 is the concentration of the given element in the matrix (C0 ≤ 1%), σ(U) is the noise level determined as the rms deviation of the signal at the point on the mass scale where a peak is absent (we can speak of the peak when the signal intensity is more than twice as high as the rms noise), and U is the signal from the element with concentration C0 . The noise level in an MX7304A mass spectrometer is specified largely by the discreteness of the analog-todigital converter (ADC) and SEM-induced noise. At relatively low SEM voltages (below 2.5 kV), the ADC noise prevails. Owing to the energy filter, off-axis position of the SEM, and screening of the detector, the noise component due to photons, high-energy ions, and neutrals falling into the detector is negligibly small. Figure 3 shows the secondary ion spectrum for a titanium sample with a 0.9% of iron impurity. Quantitative analysis was carried out using a RÉMMA-102 electron microscope. The intensity of the Fe peak (M/Z = 56) was 1.4 V (the output voltage of the direct-current amplifier) at a noise level of about 0.5 mV. The respective sensitivity threshold calculated by formula (2) is 6.5 ppm. These values were obtained at an SEM voltage of 2.5 kV. A further rise in the SEM voltage did not lead to a significant improvement of the signal-to-noise ratio because of an increase in the noise level. RESULTS AND DISCUSSION To investigate the potential of the instrument for operation with high-energy ion beams, we studied the emission of the secondary monatomic and polyatomic (cluster) ions of copper at different parameters of the primary ion beam. Figure 4 plots the spectrum obtained from a copper sample bombarded by 100-keV Ar ions. The spectrum shows groups of peaks corresponding to copper isotopes Cu+ (M/Z = 63 and 65) and isotope combinations + + of cluster ions Cu 2 and Cu 3 . When the range of mass +
numbers was extended to 400, cluster ions up to Cu 6 TECHNICAL PHYSICS
Vol. 52
No. 6
2007
SECONDARY ION MASS SPECTROMETER
were observed. Notably, as number n of atoms in a cluster grows, so does the number of possible combinations of 63Cu and 65Cu isotopes and, accordingly, the number of peaks related to the given cluster ion. Another noteworthy feature is that the emission intensity of cluster + ions Cu n does not decrease monotonically with increasing n (like in most metals) but varies periodically for clusters with an odd and even number of atoms [9–11] and depends on the cluster dissociation energy [12]. For the given copper sample and primary ion beam, the current of singly charged ions Cu+ and the relative + + content of cluster ions Cu 2 and Cu 3 were measured as a function of the primary current density at a fixed ion energy and vice versa (as a function of the ion energy at a fixed current density). The dependences were obtained at fixed parameters of the mass spectrometer to ensure a constant transmission factor of the secondary ion optics (f = const). Using formula (1) for the secondary ion current and putting C = 1 for a single-component sample, we have I = I0Ki f ,
773
Signal, 103 cps Cu+ 20
10 Cu+3
Cu+2 0 60
100
140
180 M/Z
Fig. 4. Fragment of the SIMS spectrum taken of the copper sample exposed to 100-keV Ar+ ions (the current density on the surface is 0.5 mA/cm2 , the SEM operates in the count mode under a voltage of 2.5 kV, the sample potential is 15 V, and the energy filter is tuned to 25 eV).
j, mA/cm2 0.2 0.3
0.1
(3)
0.4
(a)
30
In order to suppress the influence of the target temperature on the SIE intensity, the substrate was held at a constant temperature of 200°C during the experiment. Heating of the target decreases the number of adatoms and residual air molecules on the target surface, which affect the SIE intensity. Because of the presence of the energy filter, which transmits secondary ions only within a definite energy band, the dependences obtained describe the behavior of low-energy secondary ions with an average energy of about 10 eV. This energy corresponds to a maximum in the secondary ion energy spectrum and is most often used in SIMS instruments based on quadrupole (monopole) mass spectrometers. Figure 5 plots the intensity of the secondary copper ion signal, I(63Cu+), and the relative fractions of cluster + + ions, I( Cu 2 )/I(Cu+) and I( Cu 3 )/I(Cu+), versus the primary current (and current density) of argon ions at an argon ion energy of 100 keV. The current incident on the sample was varied by focusing the beam with a single lens placed in front of the acceleration tube. Since a diaphragm 2 mm in diameter positioned in front of the sample let only a certain (minor) part of the primary beam reach the sample surface, this current was measured by a mobile Faraday cup placed between the diaphragm and the sample and the current density was determined by the formula TECHNICAL PHYSICS
Vol. 52
No. 6
2007
20
10
0
4
8
12 I, µA
16
j, mA/cm2 0.2 0.3
0.1
20
0.4
(b)
0.4 I(Cu+)/I(Cu+n)
is the SIE ratio relating the number of where Ki = primary ions to the number of secondary ions sputtered from the surface. This factor is a function of the primary ion energy, Ki = Ki(E). Thus, for constant parameters of the secondary ion channel, I ~ I0Ki(E).
Signal, 103 cps
Sβ+
0.3 Cu+3
0.2
Cu+2
0.1
0
4
8
12 I, µA
16
20
Fig. 5. (a) Intensity of the secondary copper ion signal, I(63Cu+), and (b) the relative fractions of cluster ions, +
+
I( Cu 2 )/I(Cu+) and I( Cu 3 )/I(Cu+), vs. the primary current and current density (Ar+ primary ions, EAr = 100 keV, the other parameters are the same as in Fig. 4).
774
BATURIN et al. (a)
Signal, 103 counts/s
25 20 15 10 5 0
50
75
100
125
150 E, kV
125
150 E, kV
(b)
0.3 I(Cu+)/I(Cu+n)
Cu+3 0.2 Cu+2 0.1
0
50
75
100
Fig. 6. (a) Intensity of the secondary copper ion signal, I(63Cu+), and (b) the relative fractions of cluster ions, +
cluster ions of copper were obtained in [9], where the target was bombarded by 8-keV ions of rare gases. Figure 6 plots the signal of secondary copper ions, I(63Cu+), and relative fractions of cluster ions, + + I( Cu 2 )/I(Cu+) and I( Cu 3 )/I(Cu+), versus the energy of primary argon ions in the interval 50–150 keV at a fixed primary current. Since the acceleration tube has a focusing effect on the ion beam (the higher the accelerating voltage, the greater the focusing effect), use of primary beams with energies below 50 keV seems to be problematic because of a considerable decrease in the primary current. The dependences in Fig. 6 are plotted under the assumption that the angular distribution of secondary ions remains the same for primary beam energies in the interval 50–150 keV. According to the curves in Fig. 6, the SIE intensity decreases as the energy of bombarding ions increases from 50 to 150 eV. As was mentioned at the beginning of this paper, the dependence of the SIE ratio on the primary ion energy is similar to the energy dependence of the sputtering ratio. In the case of a copper sample exposed to argon ions, the sputtering ratio grows when the argon energy diminishes from 100 to 30 keV and gradually drops at higher energies. Such behavior accords with the run of the curve in Fig. 6a. The relative intensities of the cluster ions (Fig. 6b) also decrease, which may be explained by a decrease in the sputtering ratio and probability of sputtered atom recombination.
+
I( Cu 2 )/I(Cu+) and I( Cu 3 )/I(Cu+), vs. the primary ion energy (Ar+ primary ions, IAr = 10 µA, the other parameters are the same as in Fig. 4).
4I cos α -, j = ----------------2 πd
(4)
where d is the diameter of the limiting diaphragm (2 mm) and α is the angle of incidence of the primary beam on the sample (45°). When calculating the relative fractions of cluster ions, we added up the intensities of the peaks corresponding to different copper isotopes and also the intensities of isotope combinations corresponding to a given cluster ion. According to the plots in Fig. 5a, the secondary current of copper ions is generally proportional to the primary current incident on the sample. The linearity of the plot indicates the dynamic mode of sputtering, i.e., the sputtering of the sample material. Deviation from linearity observed at low current densities is associated with surface compounds and adsorbates remaining on the surface at a low density of the primary current. The relative intensities of cluster ions (Fig. 5b) monotoni+ cally grow with current density, the Cu 3 cluster intensity growing faster. Similar results for the emission of
CONCLUSIONS A modification of the secondary ion mass spectrometer is proposed that, in contrast to traditional SIMS instruments, uses a high-dose ion implanter as a primary ion source. This innovation allows the spectrometer to operate with high-energy (up to 150 keV) primary beams. The main parameters of the instrument, such as the resolution and sensitivity threshold, are determined. Its efficiency is demonstrated with the secondary emission of monatomic and cluster copper ions. The results obtained agree with the present-day theory of secondary ion emission in metals, as well as with the published experimental data. It is intended to extend the functionality of this instrument for detection of secondary neutrals. For this purpose, the instrument will be configured with a compact Nier-type electron ionizer, which will make it possible to study the sputtering of single- and multicomponent targets exposed to high-energy primary ions. In addition, neutrals present in the sputtered particles are much less affected by the so-called matrix effects (unlike secondary ions), so that quantitative analysis of the samples becomes feasible. REFERENCES 1. J. F. Hennequin, J. Phys. 29, 957 (1968). 2. K. Wittmaack, Surf. Sci. 53, 626 (1975). TECHNICAL PHYSICS
Vol. 52
No. 6
2007
SECONDARY ION MASS SPECTROMETER 3. N. V. Pleshivtsev, Physical Problems of Cathode Sputtering (Inst. At. Énergii im. Kurchatova, Moscow, 1979) [in Russian]. 4. V. T. Cherepin, Ion Microprobe Analysis (Naukova Dumka, Kiev, 1992) [in Russian]. 5. V. A. Baturin and S. A. Eremin, Prib. Tekh. Éksp., No. 2, 120 (2005). 6. G. I. Slobodenyuk, Quadrupole Mass Spectrometers (Atomizdat, Moscow, 1975) [in Russian]. 7. Methods of Surface Analysis, Ed. by A. Czanderna, (Elsevier, Amsterdam, 1975; Mir, Moscow, 1979), p. 276.
TECHNICAL PHYSICS
Vol. 52
No. 6
2007
775
8. D. V. Klyachko V. V. Ugarov, Poverkhnost, No. 8, 5 (1991). 9. N. Kh. Dzhemilev and R. T. Kurbanov, Izv. Akad. Nauk SSSR, Ser. Fiz. 43, 606 (1979). 10. P. Joyes, J. Phys. B 4, 15 (1971). 11. P. Joyes, J. Phys. Chem. Solids 32, 1269 (1971). 12. G. Staudenmaier, Radiat. Eff. 13, 87 (1972).
Translated by A. Sidorova