Professional Documents
Culture Documents
Beam Interactions
with Materials & Atoms
Nuclear Instruments and Methods in Physics Research B 258 (2007) 218–225
www.elsevier.com/locate/nimb
Surface Physics Division, Saha Institute of Nuclear Physics, 1/AF Bidhannagar, Kolkata 700 064, India
Abstract
Excellent detection sensitivity, high dynamic range and good depth resolution make the SIMS technique extremely powerful for the
analysis of surfaces and interfaces. However, a serious problem in SIMS analysis is its ‘‘matrix effect’’ that hinders the quantification of a
certain species in a sample and consequently, probing the composition of surfaces or interfaces by SIMS is greatly hindered. Appropriate
corrective measures are therefore, needed to calibrate the secondary ion currents into respective concentrations for accurate composi-
tional analysis. Working in the MCs+-SIMS mode (M – element to be analyzed, Cs+ – bombarding ions) can circumvent the matrix
effect. The quantitative potential of the MCs+-SIMS method is understood by assuming that an MCs+ ion is generated by the combi-
nation of a secondary neutral M0 atom with a re-sputtered Cs+ ion in the near-surface region. The emission process for the species M0 is
thus decoupled from the subsequent MCs+ ion formation process, in analogy with the ion formation in secondary neutral mass spec-
trometry (SNMS), resulting in a drastic decrease in matrix effect. Although this technique has found its applicability in direct quantifi-
cation, it generally suffers from a low useful yield. In such cases, detection of MCsþ
n (n = 2, 3, . . .) molecular ions offers a better sensitivity
as the yields of such molecular ion complexes have often been found higher than that of MCs+ ions. This is true in most of the cases
where the elements are strongly electronegative with respect to cesium. Several works have been reported on the emission of MCsþ n
molecular ions in the SIMS process, but a complete understanding on the formation mechanism of these ion complexes is still lacking.
The kinetic energy distributions of secondary MCsþ n molecular ion complexes has been found to be an effective approach to estimate the
local instantaneous surface work function changes under various surface exposure conditions, thereby enabling one to elucidate on the
probable formation mechanisms of these molecular complexes from their emission dynamics. The present paper addresses a brief review
on the MCsþ n molecular ion complexes that are emitted in the SIMS process, including various phenomenological approaches on the
subject.
2006 Elsevier B.V. All rights reserved.
Keywords: Cs+ primary ions; Electronegative elements; Secondary ion mass spectrometry; MCsþ
n molecular ions; Formation mechanisms
0168-583X/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.nimb.2006.12.172
B. Saha, P. Chakraborty / Nucl. Instr. and Meth. in Phys. Res. B 258 (2007) 218–225 219
SIMS. By far, the strongest effects of matrix effects arise as Cs+-assisted secondary ion mass spectrometry or simply
from variations in the concentrations of those reactive spe- as MCs+-SIMS [10]. However, instead of Cs+ ions, other
cies that enhance ion yields strongly, e.g. oxygen and the alkali ion bombardment and monitoring of MA+ ions
halogens for positive ions or the alkali metals for negative (where A is alkali primary ion, such as K+, Rb+, Na+,
ions. More generally, the presence of electro-negative spe- etc.) have also shown a similar small matrix effect [11]. This
cies on a surface enhances positive ion yield significantly. technique has similarly been referred to as MA+-SIMS
The enhancement is larger, the larger the electro-negativity approach.
of the species with respect to the emitted element. For In a comprehensive study on positive and negative sec-
example, fluorine gives a larger enhancement of positive ondary ion yields of atomic and molecular species sput-
ion-yield than oxygen of the same concentration. An anal- tered from elemental and compound targets under
ogous effect takes place in case of negative ion emission oxygen and cesium bombardment, Storms et al. [12] noted
when some electro-positive elements are present on the sur- that diatomic species of the type MCs+, composed of a tar-
face. These effects are, however, very important in SIMS get specific element M and Cs from the primary ion beam,
analysis and very often deliberately utilized by bombarding are sometimes emitted with surprisingly high yields, nota-
the surfaces with ions of electro-negative (or electro-posi- bly in the case of Zn. However, it was only until recently
tive) elements to get appreciable positive (or negative) sig- that this finding gathered attention when several groups
nals of the emitted species that have extremely low reported on the usefulness of MCs+ ions, either for depth
ionization probabilities. profiling of inert gases [13], which are known to exhibit a
Compensation of the matrix effect is needed for quanti- very low degree of ionization in SIMS, or for almost quan-
fication in SIMS and is possible by several methods such as titative depth profiling of matrix and impurity elements in
calibration curve [4], ion implantation standards [1] etc. All III–V semiconductors [14]. Particularly attractive was the
these approaches need ‘standards’ with matrix composition observation that the MCs+ ion yields seem to depend very
close to that of the unknown sample. Moreover, it is very little on the matrix from which the element M is emitted
difficult to make compositional analysis of the interfacial [14]. This is in contrast to the severe ‘‘matrix effect’’ known
regions using the above techniques. There is another tech- from SIMS measurements involving oxygen or cesium
nique known as ‘infinite velocity approach’ [5], which bombardment and detection of atomic secondary ions.
makes use of the high energy part of the secondary ions The application of the MCs+ technique becomes more use-
that are essentially unaffected by the instantaneous local ful when analyzing an interface of matrices of different nat-
surface chemistry. The major difficulty in this method is, ures, for instance the metal/semiconductor interface. Gao
however, the detection of high-energy secondary ions et al. [11] showed that a very simple structure of a thin
because of their extremely low count-rates compared to (0.3 lm) gold over layer deposited on a GaAs substrate
that of low-energy secondary ions. exhibited least matrix effect while monitoring the AuCs+
In order to minimize the matrix effect a technique called ions. The obtained profile did not exhibit any interface
‘‘MCs+-SIMS’’ is gradually being used nowadays [6]. In peak, which is characteristic of the typical matrix effect
this technique, the sample to be analyzed is bombarded (Fig. 1).
with Cs+ ions and secondary MCs+ ions (where M stands
for the element to be analyzed) are detected instead of M 2. MCs+ ions: formation mechanisms and characteristics
ions. Normally, bombardment by alkali ions of elements
that are electropositive with respect to the target elements Formation mechanisms of MCs+ secondary ions and
causes enhanced secondary negative ion emission from the analytical applicability of MCs+-SIMS technique have
the target owing to the lowering of metal surface work been thoroughly examined by several workers. Based on
function. As the alkali atoms get adsorbed onto the sur- some experimental observations, it was proposed that an
face, the measured work function decreases rapidly from MCs+ ion forms via a recombination between an indepen-
its initially high value and with continued adsorption, the dently sputtered neutral M0 atom and a re-sputtered Cs+
work function reaches a minimum and then rises to ion in the close proximity of the sample surface. This pro-
approximately the bulk value with completion of the first cess essentially makes the MCs+ species to suffer less from
full layer of adsorbate atoms [7–9]. This phenomenon has the matrix effect. Irrespective of the formation mechanism
opened up a wide area of research owing to the general of MCs+ ions, it is obvious that the binding energy between
interest of alkali adsorption as a way of enhancing the elec- M and Cs (either charged or neutral) plays a very impor-
tron emission properties of a surface. Moreover, this sys- tant role in the formation of MCs molecule and the polar-
tem attracts wide attention as one of the simplest izability of M plays the key role in the binding strength of
examples of chemisorption. Although the application of the two species. The fact that the emission intensity of
‘‘MCs+-SIMS’’ technique is difficult for its poor sensitivity MCs+ depends linearly with the atomic polarizability of
(MCs+ molecular species are usually less abundant than M was established from the measurements of MCs+ ion
the atomic ions), MCs+ detection scheme is more favorable yields from silicon containing various low-concentration
in cases where matrix effects are most severe, particularly elements [15]. The results showed that the ratio I(MCs+)/
for major components. This novel technique is referred to I(SiCs+) increased monotonically with the polarizability
220 B. Saha, P. Chakraborty / Nucl. Instr. and Meth. in Phys. Res. B 258 (2007) 218–225
been observed that for all specimens with the exception of ions. However, some earlier observations [20] have con-
Si, the flux of MCs+ ions scales linearly with the Cs+ inten- firmed its almost constant value for low emission velocities.
sity over a wide range of cesium concentration (several Assuming the constancy in e0, if /þ þ
i and /j are the two val-
þ þ
orders of magnitude in some cases); for silicon deviations ues of /, and bi and bj represent two values of b+, then
from this scaling are found at the highest Cs content and from Eq. (5)
are tentatively ascribed to changes of the sputtering yield
in the heavily doped Si specimens. The Csþ þ bþ þ
j =bi / exp½ð/j /i Þ=e0
2 =Cs ratio, on ð6Þ
the other hand, has been found to decrease more than lin- / exp½D/=e0 ;
early (possibly exponentially) upon reduction of the Cs
concentration, indicative of a decreasing formation proba- where D/ = /j /i; the change in surface work function.
bility of Csþ 2 dimers at low concentrations due to an Eq. (6) implies that the normalized integrated count should
increasing (average) distance between individual Cs atoms. show an exponential dependence on D/. The normalized
Theoretical studies carried out by Mootz et al. [18] show Cs+ count, plotted as a function of D/, showed a very good
that MCs+ ions have energy distributions, which are deter- agreement with the theoretical prediction for e0 = 0.29 eV.
mined by the energy distributions of the sputtered neutral That is probably the first experiment to determine e0 for
M atoms. For both distributions the mean emission energy Cs+.
is enhanced if the element’s surface binding energy is Integrated counts of AgCs+ showed an overall fall with
enhanced too. In contrast, the energy distributions of the increasing oxygen pressure. This observation apparently
re-sputtered Cs+ ions appear to be element independent. contradicts the concept for MCs+ molecular ions, as it is
The efficiency of MCs+ molecule formation monitored by known that the formation of such molecular ions remains
the MCs+ signals is maximum for low mean emission ener- almost unaffected by the change in instantaneous local sur-
gies and low surface binding energies. This shows the face chemistry. This apparent discrepancy was answered by
importance of a correlated emission of the partners form- taking into account the exponential fall of re-sputtered Cs+
ing the molecule. The formation mechanism of the MCs+ ions. Theoretically, the intensity of MCs+ molecular ions
clusters was investigated using a Monte Carlo model [19]. can be expressed as [17]
It was shown that the majority of the constituent particles
of the formed clusters are initially first or second neighbor I MCsþ / YcM ð1 bMþ bM ÞYcCs bþ fMCsþ ; ð7Þ
atoms at the surface and that the velocity distribution of
the MCs+ clusters becomes broader and peaked at higher where cM is the concentration of M, Y is the global sputter-
velocities with increasing surface binding energy of the M yield of the target, cCs is the surface cesium concentration,
atom. In addition, it was demonstrated that the interaction b+ is the ionization probability of Cs and fMCsþ is the for-
potential between the M and Cs+ particle has no influence mation probability for MCs+. bMþ and bM are the ioniza-
on the velocity distribution of the MCs+ clusters. On the tion probabilities for M+ or M secondary ions,
other hand, the cluster formation probability, defined as respectively. For most practical purposes, bMþ and bM in
the probability that a sputtered M and Cs+ particle will the case of MCs+ ion intensities are negligible and all the
form a MCs+ cluster, is extremely sensitive to this interac- factors except b+ remains constant. fMCsþ , estimated at var-
tion potential. It is also shown that the cluster formation ious oxygen exposures, has been found to be unchanged
probability decreases with increasing surface binding [21]. Hence Eq. (7) can be rewritten as
energy. Finally, a good correspondence is obtained
I MCsþ ¼ Kbþ ; ð8Þ
between the calculated and experimental velocity distribu-
tions of MCs+ clusters sputtered from different mono-
atomic materials. As a consequence, the Monte Carlo
model and the discussed results could be validated [19]. 1.0 Experimental data
Normalized AgCs intensity
þ
tributions of Cs+, Csþ +
2 , AgCs and AgCs2 ions measured
at different oxygen pressures have exhibited changing
slopes in their leading parts that hinted appreciable 0.6
changes in the local instantaneous surface work function.
The ionization probability b+ depends on the ionization +
0.4 IMCs+ = K
potential I of the sputtered atom and surface work function
/ according to the relation
0.5 0.6 0.7 0.8 0.9 1.0
bþ / exp½ðI /Þ=e0 ; ð5Þ +
Normalized Cs intensity
where the parameter e0 is considered to depend on the nor- Fig. 3. Normalized AgCs+ intensity as function of normalized Cs+
mal component of the emission velocity of the secondary intensity.
222 B. Saha, P. Chakraborty / Nucl. Instr. and Meth. in Phys. Res. B 258 (2007) 218–225
where K is a constant. In other words, the normalized tion measurements in terms of concentration, dose, decay
MCs+ intensity should show linear dependence on normal- length and detection limit. The quantification of the impu-
ized Cs+ intensity. The normalized AgCs+ intensity, plot- rities in GaN and AlGaN films by using the MCs+ and
ted against normalized Cs+ intensity, showed a linearity MCsþ 2 ions enabled the detection of both electronegative
(Fig. 3), which was in good agreement with the theoretical and electropositive elements in the same analysis run. It
prediction for all oxygen pressures. This observation, even was shown that the p-type dopant Mg and H could be mea-
under a combined influence of electropositive and electro- sured together [23]. The detection limits of HCsþ þ
2 , CNs2
+
negative elements on the surface, strengthened the accepted and MgCs were reported at the same level as the limit
concept that MCs+ formation is essentially due to the com- of M ion detection. Further, the application of the MCs+
bination of a neutral M0 with a re-sputtered Cs+ ion. and MCsþ 2 ions suppresses the matrix effect in the
AlGaN/GaN films.
Although several works have reported the mechanism of
3. MCsþ2 ions: importance, formation mechanisms and MCs+ formation [24–29], that of MCsþ n ions (where
characteristics n = 2, 3, . . .) still remains to be well understood. In view
of the fact that the kinetic energy distribution of a certain
Although MCs+-SIMS technique has found its applica- emitted species is an effective diagnostic to understand the
bility in direct quantification, it generally suffers from a low formation mechanism of the species and its dynamics, a
useful yield. In such cases, detection of MCsþ 2 molecular systematic study on the kinetic energy distributions (EDs)
ions offers a better sensitivity to that of MCs+ ion of MCsþ n (n = 0, 1, 2, . . .) of species emerging from various
(Fig. 4), as the yield of MCsþ2 ions has been observed much metal targets under Cs+ ion bombardment was reported
higher than that of MCs+ ions [11]. This is true in most of to elucidate the probable formation mechanism of these
the cases where the elements are strongly electronegative molecular complexes. [15]. Several previous studies have
with respect to cesium. Holliger et al. [22] investigated investigated the energy distributions of MCs+ ions [24–
the analytical usefulness of monitoring MCsþ 2 ions for 29] but corresponding data for MCsþ 2 appear to be lacking.
semiconductor profiling of n-type dopants such as P, Ge Several studies suggest the following three possible
or As with a primary (Cs+ beam) impact energy of routes for the formation of these ions.
1 keV. They indicated that the MCsþ 2 technique has an
advantage over the MCs+ technique of providing higher M0 þ Csþ
2 ! MCs2
þ
ðiÞ
useful yields for electronegative elements by up to 102 M þ Csþ þ Csþ ! MCsþ ðiiÞ
2
and can therefore be applied to perform more precise junc-
MCs0 þ Csþ ! MCsþ
2 ðiiiÞ
(eV)
(EDs) of a certain species emitted in the SIMS process is an -0.6
effective diagnostic to understand the formation mecha- +
Al
nism of the species and its kinematics, energy distributions -0.8 +
AlCs
(EDs) of MCs+ and MCsþ 2 molecular species have been +
-1.0 AlCs2
made for various target materials under varying Cs envi-
ronments [31–33]. Instantaneous local surface work func-
1 2 3 4 5
tions estimated from the leading parts of the kinetic
Impact Energy (keV)
energy distributions were found to play a major role in
the formation of these ion complexes. In the energy distri- Fig. 6. Surface work function change (D/) versus impact energy for Al+,
butions of M, MCs+ and MCsþ AlCs+ and AlCsþ 2 ions [33].
2 ions for the elements Al,
Cu and Ag for a typical bombarding energy of 2 keV, it
was observed that for a given element the peak position the study, the peak shift of the EDs towards lower energies
of the ED shifted to lower values in the order M+, MCs+ was due to the increase of surface work function (due to
and MCsþ lowering of Cs concentration with the increase of impact
2 ions (Fig. 5). This was found to be true for
any value of impact energy. From the bombarding energy energy). The lower the work function, the greater is the
variations of EDs for M+ or MCs+ or MCsþ shift of the energy distribution towards higher energies
2 , it was
observed that the peak of the ED for each species shifted [34]. Any change in the instantaneous surface work func-
towards lower energy as the impact energy was increased tion should reflect the variation of the leading edge profile
[32]. The peak position of the energy distribution for any of the energy distributions. By fitting tangents to the lead-
emitted species depends essentially on the surface binding ing parts of the EDs and subsequently measuring the inter-
energy (U) of the species as well as the surface work func- cepts of the tangents with the energy axis one does the
tion (/), as understood by the simple expression estimation of the work function changes. Fig. 6 shows
the variation of the measured surface work function
E ðI /Þ changes (D/) with impact energies for M+, MCs+ and
Ion Yield / exp : ð9Þ
ðE þ U Þ3 e0 MCsþ 2 species, thus evidencing the role of surface work
function changes on the energy distributions. The effect is
The sublimation energy of pure Cs being much lower more pronounced when the surface work function changes
(0.83 eV) than that of the three metals, decrease of Cs become more significant and this was demonstrated for sil-
concentration (i.e. increase of impact energy) should tend ver by changing the impact angle of the incident beam [33].
to increase the magnitude of the surface binding energy. The impact angle variation on the work function change
Since this was contrary to what was actually observed in was found to be more sensitive than the impact energy var-
0.8
1.0 +
AlCs
Normalized Intensitay
0.4 0.8
0.0 0.6
Nomalized intensity
Cu b 0.4
0.8
0.2
0.4 0.0
+
1.0 AlCs 2
0.0
Ag c 0.8
0.8 0.6
0.4 0.4
0.2
0.0 0.0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 16 18 20
Energy (eV) θ (degree)
Fig. 5. Energy distributions of MCs+ and MCsþ 2 molecular ions for Au, Fig. 7. Energy distributions of AlCs+ and AlCsþ
2 ions for different impact
Cu and Ni targets at 2 keV impact energy [32]. angles [34].
224 B. Saha, P. Chakraborty / Nucl. Instr. and Meth. in Phys. Res. B 258 (2007) 218–225
iation on the same. Fig. 7 shows the energy distributions of To explore the enormous quantitative potential of the
AlCs+ and AlCsþ 2 ions for different impact angles [33]. MCsþ n , in recent times, several experiments have been
So far, a handful of works has been done to reveal the reported where dual beam have been successfully used to
underlying formation mechanism of the MCsþ n molecular improve the MCsþ n signal. In a ‘‘Cesium-Xenon’’ co-sput-
ions. The study of energy distributions of the MCsþ 2 mole- tering experiment, performed on pure silicon, silicon oxide
cules ions and its proposed constituents has been the key and aluminum oxide, MCsþ n clusters’ signals were reported
probe to understand its formation mechanism. The energy to enhance significantly for a Cs beam concentration, with
distribution of the MCsþ 2 molecules (where M = Zn, Cu, maxima that varied from 10% to 30% [35] and the data rel-
Ge, Ni and Mo) has been found to depend on the surface ative to the silicon wafer were successfully interpreted by
binding energy, which provides evidence that the energy models derived from the tunneling model. To optimize
distribution of the neutrals is also responsible for the the advantageous quantification technique consisting in
MCsþ 2 distributions [18]. Furthermore, the narrower energy analyzing MCsþ n clusters, Writz et al. have developed the
distribution of the trimers compared to respective dimers’ Cation Mass Spectrometer (CMS), a new instrument espe-
energy distribution supported a recombination of the con- cially dedicated to perform this kind of analysis [36]. In this
stituents of the trimers subsequent to their sputter process. novel technique, collimated and adjustable stream of neu-
Comparison of the energy distributions of the components tral Cs atoms are deposited on the surface of the sample
involved in the dimer formation process with the energy while this one is being bombarded with Xy+ (where ‘X’
distributions of components, which might be responsible stands for any element excepting Cs). This experimental
for the trimer formation, strengthened the belief that the technique permits a successful decoupling of the sputtering
formation of an MCsþ 2 molecular ion was due to the recom- and Cs introduction processes by avoiding the constraints
bination of MCs0 with Cs+ (process (iii)). The MCs0 occur- imposed by an energetic Cs+ ion bombardment. As a con-
ring in this formation process results from the sequence, it becomes possible to optimize simultaneously
neutralization of an MCs+ ion by electronic capture [27]. the sensitivity of the analysis, by carefully adjusting the
In another study of MCsþ n molecular ions formation from Cs concentration to its optimum value, and the depth res-
silver surface under a varying oxygen environment, a linear olution of the analysis, by choosing adequate primary
dependence of AgCsþ 2 intensity on the ionization probabil- bombardment conditions. For three samples (Al, Si and
ity of Cs+ (Fig. 8) suggested that an AgCsþ 2 molecular ion Ni) covering an important range on the work function scale
forms via the recombination of AgCs0 and Cs+ [21]. Fur- (U = 4.28–5.15 eV), it has shown that the behavior of the
thermore, the estimated mean emission energies of various useful yields of the detected MCsþ n signals only depends
species also indicated that the formation of MCsþ 2 via pro- on the ratio between the erosion and deposition rates,
cess (iii) seemed to be the most probable one in that partic- but not on the individual values of these two rates. This
ular case. A quantitative evaluation of the respective mass characteristic becomes very important within the perspec-
signals, with respect to the spatial and temporal correlation tive of a low-energy primary bombardment with a view
of the components forming the molecule, corroborated the to an improvement of the depth resolution. Furthermore,
fact that in the case of the MCsþ 2 formation the spatial and the deposition of neutral cesium has been reported to
temporal correlation is one order of magnitude higher than reduce the work function of the sample that increases the
that in the MCs+ formation process [18]. Besides, it was useful yields significantly and helps avoiding the matrix
observed that in both the cases of dimer and trimer forma- effect for elements with high electron affinity [37].
tion, the spatial and temporal correlation decreases with
increasing surface binding energy. 4. Conclusions