CHAPTER: DEPENDENCE of SECONDARY ION MASS SPECTROMETRY RELATIVE
SENSITITY FACTORS (RSF) on INSTRUMENT CONDITIONS: ACCURACY OF RSF
DETERMINATIONS
Secondary ion mass spectrometry (SIMS) relative sensitivity factors (RSF) derived from
ion implanted standards are used for quantification, the conversion of secondary ion intensity
to impurity atom density. Recognizing that more than one definition of RSF exits, we define
RSF in this work via the relationship: impurity density equals RSF multiplied by the ratio of
measured secondary ion intensity of the impurity to that of the matrix. Details of this definition
and working application of it can be found in references 1-3. The SIMS RSF does not vary
with any parameter or condition that affects equally (proportionally) the measured secondary
ion intensities of the matrix reference and the impurity ion, by definition. The transferrability
and reproducibility of RSFs among different instruments and laboratories was addressed by
Simons et al.4 Spiller and Ambridge,5 using a CAMECA instrument, showed that the
accuracy and reproducibility of SIMS RSFs depended on the location of the analysis area
within a target holder window and with the location of the window in the target holder,
concluding that analysis should be performed at least 0.5 mm away from a window edge and
near the holder center, in order to assure a variation in the value of RSF of less than 10 to
15%. They also determined that reproducibility (stability) was improved by closing slightly the
entrance slit. This 10 to 15% variation in RSF was determined for moderately controlled
conditions. They used implanted reference standards and define their RSF exactly as we do
here. They achieved a variation of only 3% among nine measurements in one case.
Satoh et al.,6 using a small Ga ion beam in a custom instrument, measured a
repeatibility of RSF of 20% and an absolute accuracy of a factor of 2 to 3. Chi and Simons, 7
using a CAMECA instrument, discuss RSF variations for 3 elements in Si. They define RSF
as we do and also show the importance of analysis location within a holder window. They
report an RSF variation for B and Cu of less than 10%, and a larger variation of 2 to 3 for Na,
which they attribute to the narrow secondary energy distribution of Na and its greater
dependence on location and instrument variables. They note the importance of the
magnitude of the band pass energy (energy slit) for Na. Lux, 8 using several CAMECA
instruments, studied RSF variation for 23 elements in Si. She describes two strictnesses of
protocol, achieving 21 and 11% variation in RSF. She studied the influence of primary beam
species and mass filtering of the beam, and the sample holder, and the detector type.
Kurosawa et al.,9 using a CAMECA (assumed from the conditions stated) instrument, report a
variation in RSF of 10%. They report RSF patterns versus ionization potential and electron
affinity. For negative ions (electron affinity), C, O, and As, lie on a lower (by our definition, not
theirs) RSF line than other elements, in a similar relationship as reported in reference 3.
Using an ESCALAB instrument and the same definition of RSF that we use, Strydom et al. 10
discuss RSF values for HgCdTe, but do not report variations or accuracies. However, they
compare their values with others1 which shows variations in RSF among the two independent
measurements (allowing for the instrument difference that they point out) of about 40%.
Newbury and Simons11 analyzed NBS standards using a CAMECA instrument and show
absolute RSF deviations for 15 elements that vary within ~20%, with 10 of the 15, varying
within ~12%. They discuss other issues regarding SIMS standards. Lodding and Odelius 12
discuss an RSF pattern, secondary ion energy distributions, effect of offset voltage, and
detection limits.
Here we discuss the influence of SIMS measurement conditions and instrument
parameters on RSF values for CAMECA IMS 3f, 4f, and 5f instruments using both oxygen and
cesuim primary ion beams. Conditions addressed include primary ion impact energy (and
associated angle of incidence), sputtering rate (primary ion current density), size (diameter) of
contrast diaphragm, secondary ion species and various associated widths of secondary ion
energy distribution, secondary ion intensity impurity or matrix), and location of analysis area
within a sample holder (nearness to a sample holder edge).
We have measured the influence of the secondary ion energy distribution of selected
elements, namely K (narrow energy distribution), Mn (medium energy distribution), and Si
(wide energy distribution), both as impurities and as matrix components on some of these
variables. These results assume the use of the full 128 eV energy band pass of the
electrostatic analyzer in the CAMECA instrument. This work is specifically quantitatively
applicable to a CAMECA IMS instrument (sector magnet, high extraction voltage, 128 eV
energy band pass, and mass independent detector system), and only qualitatively applicable
to quadrupole instruments.
We have measured weak systematic dependence of RSF on some of the above
variables, but the magnitudes of the variations in RSF are of the order of a few to ~20%,
individually [which are all within the ~30% accuracy that we quote for most RSFs]. The
notable exception is the location of the analysis area within a sample holder window
(nearness to a sample holder edge), which can cause variations of a factor of 50% to 2 or 3,
or even 5 if the crater is placed near the corner of the window. If RSF measurements and
quantified SIMS measurements are performed at least 1 mm away from a window edge,
RSFs should be measurable to within ~30%, from instrument-to-instrument (magnetic sector
with reproducible high fields) and time-to-time [for samples with accurate implant fluences or
known doping densities, or the same round robin samples, for full 128 eV band pass energy,
for conditions that do not saturate (more than a few percent) the secondary ion detector
used]. Possible exceptions are elements with high ionization potential, low electron affinity,
and that are easily desorbed, which include N, Hg, and the noble gases. These elements
may be ionized above the surface,13 which may cause irregular or not easily reproducible
secondary yields and therefore RSFs. RSFs for F (enhanced positive ion yield) are quite
reproducible and consistent.
Variation of RSF with primary beam energy and associated angle of incidence: This
issue is discussed in Sections 1.2 and 1.3 of reference 2. For our standard operating
condition of 6- or 8-keV O impact energy, the incidence angle is (40±1)°, and for 14.5-keV Cs
energy, it is 26°. All RSF "standards" work has been carried out under these conditions, as
stated in reference 2, so all of our RSFs and systematics are for these conditions. Variations
in ion yield with angle of incidence may result from variations in the degree of incorporation of
atoms of the primary beam and resulting changes in the surface work function, but if these
yield changes are the same for impurity species and matrix reference element, then the RSF
is not altered. This is essentially the case for O and Cs primary beams in the CAMECA
instrument because of the results of the following experiments. We measured the
dependence of RSF of B and F on primary O2 impact energies of 3.0, 5.5, 8.0, 10.5, and 13.0
keV in a CAMECA IMS 4f (corresponding angles of incidence of 52, 42, 9, 37, and 35°,
respectively). The results show that the RSF varies ~15% for B and 33% for F over that range
of values of primary impact energy (the full range on energies provided on the standard
CAMECA instrument). Similar mesurements were made for Cs primary impact energies of
9.5, 12.0, and 14.5 keV (corresponding to angles of incidence of about 26, 25, and 24°,
respectively) for S and As in Si. The results showed that the RSF varies by less than ~8%
over this range of values of Cs impact energy. However, we specify the use of only 6- or 8keV O2+ impact energy, and only 14.5 keV Cs impact energy for RSF reference standards.
Variation of RSF with sputtering rate: We measured the dependence of RSF on
sputtering rate for 8-keV O2 primary ions and a 250m raster and a Si matrix over the range
of sputtering rate that we normally use and that corresponds to primary O 2 beam currents
from 0.1 to 10 A, or two orders of magnitude in sputtering rate. A slight systematic increase
in RSF was measured over this range, but the total change was 12% over the two orders of
magnitude, or ~6% within the total range over which we collect data.
Variation of RSF with size (diameter) of contrast diaphragm (CD): We studied in detail
a number of elements that have different energy distributions, including K, which has a narrow
energy distribution, and Mn, which has a medium energy distribution, implanted into Si, with Si
as the matrix reference, and measured the RSF using constrast diaphragms (CD) with
diameters of 30, 85, 150, and 400 m (the four standard sizes in the CAMECA IMS
instrument). We installed a set of four new apertures for the mesurement. The results show
that the values of RSF measured for Mn for the three smaller CDs are within ~2%, and the
value for the largest CD (400m) is 20% lower. If all four values of RSF are combined, the
composite error is ~12%, which is still within the accuracy stated for RSF. [If only the three
smaller CDs are used, the error in RSF is negligible (within measurement accuracy - the
~2%.] For K, the values of RSF measured for the three smaller CDs varied ~13%, with no
consistent trend with CD size. The 400 m CD could not be used because the secondary
intensity of K+ from the reference standard would have saturated the electron muiltiplier
detector. This is nearly always the case, and the 400 m CD would almost never be used for
analysis of K+. Some errors in RSF measurement may result because some analysts may try
to use the 400 m CD or more intense secondary ion intensities and may not be careful to
check or understand the effect of detector dead time/detector saturation.
Variation of RSF with location of the analysis area within a sample holder window: The
effect of location of the analysis area within a sample holder window on RSF and on energy
distribution is very significant -- up to factors of 2 or 3, or even 5 if the analysis is performed in
the corner of a holder window. If analysis is performed more than 1 mm from a sample holder
edge, The variation (reproducibility) of RSF is within ~20%. Moving from the center toward an
edge of the window cuts off the low energy portion of the energy distribution and therefore
affects impurities with narrow energy distributions more than those with wide distributions
(determined from corresponding measurements of secondary ion energy distributions).
Deline14 has also reported this observation. For combinations like K in Si, the RSF is
affected.
This may explain why RSFs for combinations like K in Si vary among
workers/laboratories; analyses may not all have been performed more than 1 mm away from
a sample holder egde. Elements that may be ionized above the sputtering surface, like N, Hg,
and the noble gases, He, Ne, Ar, Kr, and Xe (high IP, low EA, and high vapor pressure), may
be more affected by this situation and exhibit less consistent RSFs.
In summary, the dependences of RFS on selected instrument/analysis conditions in a
CAMECA IMS instrument, in terms of error or uncertainty are:
primary ion energy: O2 - 10%; Cs - 15%
sputtering rate: 15%
CD size: 5 or 10%
analysis location: 10 to 300%
impurity species: 20%
Total (for centered analysis location and corresponding 10 or 20%, and the full range of the
above variables: 50 or 60%.
We15 have checked reproducibility of RSF for most elements from H to U.
Additional issues and comments that might be useful to a user
Effects may occur at the ion-mixed sputtering surface of the matrix from which both
matrix and impurity ions are sputtered. The extraction, focusing, and transmission optics must
be considered. Finally, detector issues must be considered. In the Cameca magnetic sector
instrument, in which the extraction and focusing fields are relatively large, these issues are
different from those in the weaker field quadrupole instruments. In reference 2, most of the
issues that affect the measurement accuracy of SIMS are discussed. However, we have
carried out additional measurements to check further or substantiate some of our statements - for the magnetic sector instrument -- for which the present work is relevant, and upon which
the prior RSF systematics of atomic and now molecular ions are based. 1-3 In reference 2, we
state that the primary cause of variations in RSF -- reproducibility from time-to-time, from
instrument-to-instrument, from sample-to-sample, etc., is the location of the analysis area
(crater) within a sample holder window. This has once again been demonstrated, but with
greater quantitative details. Location of the analysis area within the sample holder window is
the only experimental parameter that causes significant (>25%) variation in SIMS RSF values,
using the standard full 128 eV band pass energy, and with the energy slits centered, which is
the condition for our RSF measurements.
It is important to use certain matrix reference elements for consistent, reproducible
RSFs, for example, As for GaAs, P for InP, Te for HgCdTe, i.e., the column V or VI element
for III-V or II-VI materials, respectively, and Si for SiO2 or Si3N4. It is important not to use
elements that do not produce consistent RSFs as impurities (in Si, for example), namely N,
Hg, or other elements that have high IP, low EA, or are gases or are volatile.
It is important not to saturate the matrix reference signal during RSFs determinations,
or RSFs that are too low will be recorded.
It is important not to saturate the detection electronics, so elements with low ionization
potential are sometimes not good choices for positive secondary ion mass spectrometry,
elements such as Al, Ga, or In, which are common in III-V materials.
When measuring detection sensitivities, it is important that a low fluence implant be
used -- a fluence that yields a maximum impurity density that is no more than about four
decades above the actual background signal (instrument background), or a detection limit that
is too high will be recorded.
We generally use implant standards that have fluences of 1013 to 1014 cm-3 to avoid
creating a detection limit that is caused by sputtering memory in the instrument rather than the
true background at the mass of interest. In contrast, when studing an element that has low
detection sensitivity, we use implant standards that have higher fluences, namely 10 15 to 1016
cm-3, in order to detect secondary ions of the impurity.
We use implanted standards of the rare isotopes of some elements in order to
enhance the detection sensitivity and detection limits of certain elements, which certainly
include the ambient gas or atmospheric elements, and combinations of them or molecular
interferences at their masses (H, C, N, O, Si, S). The use of these masses also greatly
reduces the signal at the surface caused by surface contamination for some masses.
We use higher energies for implant standards to place the peak of the depth profiles
deeper than the SIMS equilibration depths, especially for higher atomic number elements.
All of the above mentioned issues account for some of the consistency and
reproducibility of RSFs that we measure and report. We believe that anyone who observes
these procedures and uses the full 128 eV energy band pass and does not perform an
analysis for quantification near a sample holder edge can achieve RSFs that are consistent
and reproducible within ±30%, except for the elements noted.
Issues such as the purity of the primary ion beam and dependence of collector
efficiency on ion mass for an electron multiplier are not addressed here. We assume that
adequate primary beam mass separation and calibrated electron multiplies (or well designed
Faraday cups) are used for RSF determinations. The issue of the effect of any charging
during the determination of RSFs for insulators is left for another treatment; this work applies
to targets that do not charge significantly during SIMS analysis. We have measured
secondary ion energy distributions under various conditions for several elements as both
impurities and as matrix elements (metals, semiconductors, and insulators) and for elements
with relatively narrow energy distribution (K), medium width distribution (Mn), and relatively
wide distributions (Si). K was measured as a matrix element (high intensity) for KTaNbO 3, Mn
for a Mn-Zr alloy, and Si from crystalline Si. We have measured energy distributions for
elements that have most of their distribution in negative potential region (Hg), most in the
positive potential region (Si, As), and fairly symmetrically split between positive and negative
regions (Te). The results for matrices like BN (an insulator) and HgCdTe were especially
interesting. These results, considered in total, show that the 128 eV energy band pass
passes >99% of secondary ions with narrow or medium widths, and >95% of the distributions
for elements with wide distributions, such as Si -- by integration of the energy distributions.
References:
1. R.G. Wilson, J. Appl. Phys. 63, 5121 (1988)
2. R.G. Wilson, F.A. Stevie, and C.W. Magee, Secondary Ion Mass Spectrometry
[Wiley, New York, 1989]
3. R.G. Wilson and S.W. Novak, J. Appl. Phys. 69, 446 (1991)
4. D.S. Simons, P.H. Chi, P.M. Kahora, G.E. Lux, J.L. Moore, S.W. Novak, C. Schwartz, S.A.
Schwarz, F.A. Stevie, and R.G. Wilson, Secondary Ion Mass Spectrometry
SIMS
VII, A. Benninghoven, C.A. Evans, K.D. McKeegan, H.A. Storms, and H.W.
Werner,
Eds. [Wiley, NY, 1990] pp. 111-114.
5. G.D.T. Spiller and T. Ambridge, Secondary Ion Mass Spectrometry SIMS V, A.
Benninghoven, R.J. Colton, D.S. Simons, and H.W. Werner, Eds. [Springer, Berlin,
1986] p. 127
6. H.Satoh, M. Owari, and Y. Hihei, Secondary Ion Mass Spectrometry SIMS VII, A.
Benninghoven, C.A. Evans, Jr., K.D. McKeegan, H.A. Storms, and H.W. Werner, Eds.
[Wiley, NY, 1990] p. 91
7. P.H. Chi and D.S Simons, SIMS VII, ibid., p.126
8. G. Lux, SIMS VII ibid, p. 123.
9. S. Kurosawa, Y. Homma, T. Tanaka, and M. Yamawaki, Secondary Ion Mass
Spectrometry SIMS IV, A. Benninghoven, J. Okano, R. Shimizu, and H.W. Werner,
Eds., [Springer, Berlin, 1984] p. 107
10. H.J. Strydom, A.P. Botha, C.A. Strydom, and J.S. Vermaak, SIMS VII, ibid., p.163
11. D.E. Newbury and D.S. Simons, SIMS IV, ibid., p. 101
12. A. Lodding, H. Odelius, and L.G. Petersson, SIMS IV, ibid., p.478
13. V.R. Deline, Analytical Electron Microscopy 1987 (Proceedings of the Microbeam
Analysis Society), D.C. Joy, Ed. [San Francisco Press, San Francisco, 1987], 269-70
14. R. Holland and G.W. Blackmore, Surface and Interface Analysis 4, 174 (1982)
15. F.A. Stevie is included in this RSF check.
Other Bibliography
Y. Homma, S. Kurosawa, Y. Kubota, Y. Nakamura, K. Nomura, M. Shibata, H. Shichi, J.
Takahashi, Y. Yoshioka, and T. Ogawa, SIMS VII, ibid., p.119
Y. Homma, Secondary Ion Mass Spectrometry SIMS VI, A. Benninghoven, A.M.
Huber, and H.W. Werner, Eds. [Wiley, NY, 1988] p. 717
MBernhein and G. Slodzian, SIMS VII, ibid., p.139
P. Roitman, D.S. Simons, P.H Chi, R.M. Lindstrom, G.E. Lux, S. Baumann, S.W. Novak,
R.G. Wilson, D. Farrington, J. Keenan, F.A. Stevie, J.L. Moore, R.B. Irwin, A.J. Filo,
C.W. Magee, R. Alcorn, and D. File, "Round-robin study of implants in Si and SiO2
using SIMS, RBS, and NAA," Secondary Ion Mass Spectrometry SIMS VII, A.
Benninghoven, C.A. Evans, K.D. McKeegan, H.A. Storms, and H.W. Werner, Eds.
[Wiley, NY, 1990], pp. 115-117