Download - Introduction and Scientific Background V2
1. Introduction
Presently, the environmental aspects of many manufacturing processes are
an important issue. This is particularly true in high-end technology production such
as in microelectronics and other related areas. Some of the most important
industrial steps in this manufacturing category are considered to be wet wafer
cleaning and silicon oxide etching. Both of these procedures currently utilize HF
based solutions, materials which are hazardous both to humans and the
environment. A substitute for such procedures having the same production
capabilities as HF based solutions is needed. A less harmful and more
environmentally friendly chemical media may be found in the shape of room
temperature ionic liquids.
This relatively new solvent class is composed of ions only. The salts are
characterized by weak interactions, owing to the combination of a large cation and a
charge-delocalized anion. This results in a low tendency to crystallize due to
flexibility (anion) and dissymmetry (cation). These properties endow the fluids with
ion conductivities comparable to many organic electrolyte solutions and an absence
of decomposition or significant vapor pressure up to ~300–400°C. Ionic liquids are
basically composed of organic cations and inorganic anions that may undergo almost
unlimited structural variations because of the easy preparation of a large variety of
their components. Thus, various types of salts can be used to design the ionic liquid
that has the desired properties for a given application. Other properties of RTIL's
include thermal and chemical stability and non-flammability. In recent times the
scientific and technological importance of ionic liquids spans a wide range of
applications. New types of lubricants, seals, fluids for thermal engines and
adsorption refrigeration have been suggested on the basis of the diverse and unique
properties of ionic liquids. However, one of the most interesting applications made
possible by their unique solvent capabilities involves material dissolution. A variety
of different compounds and materials were reported to be soluble in RTIL's. Amongst
them were metals, organic materials and even oxides. [49, 8, 18, 21].
In our research we study the ability of RTIL media to serve as a substitution to
HF-based solution environments for SiO2 dissolution. Our goal is to produce a
possible IL solution candidate and explore the basic science behind its system. This
will be achieved by the examination of surface reactions and by recognizing other
aspects of the reaction's nature such as whether it is chemical or electrochemical,
identification of reaction products and the investigation of basic kinetic and
thermodynamic parameters of the dissolution process.
a b
2. Scientific Background
2.1 Silicon Dioxide.
Silicon dioxide (Silica or SiO2) is known to be one of the most abundant
minerals in the Earth's crust. Nine different possible phases of SiO2 may be found in
nature. Among these phases one can include quartz, cristobalite, tridymite, opal and
inorganic amorphous silica [132, 166]. Some of these phases are crystalline as can be
seen in quartz where others are amorphous, as in opal.
At the crystallographic level silicon dioxide is constructed as tetrahedral or
triangular polyhedra of oxygen ions with a silicon atom occupying the center of the
structure. The average bond length of Si-O is 1.62Å, and that of O-O is 2.27Å. The
polyhedras are joined one to another by a bridging oxygen atom. These oxygen
atoms are common to two different polyhedras in the SiO2 matrix. In its crystalline
form all of the oxygen atoms provide such a connection as may be seen in Figure 1.a.
In vitreous non crystalline SiO2 only some of these oxygen bridging atoms connect
two polyhedras, whereas others are part of only one polyhedra structure. This
provides the amorphous SiO2 with a less rigid structure, as may be seen in Figure 1.b.
[40].
Figure 1. (a) 3D model of crystalline SiO2. Red atoms represent oxygen, blue represents silicon. A structure based on polyhedras and bridging oxygen atoms between these polyhedras is clearly seen. (b) 2D model of amorphous SiO2. Red atoms represents silicon, blue represents oxygen. The same defined polyhedras with a less rigid structure can be seen. www.3dchem.com
Different types of impurities may be found both in the crystalline and
amorphous forms of SiO2. These impurities have a significant effect on the structure
as modifiers. The impurities can be either substitution or interstitial defects and also
water imbedded in the structure. The substitution impurities replace the silicon ions
in the polyhedra structure and in cases of having a valence different from silicon (4
for silicon and 3 or 5 for the impurities) it will also create a charge defect. These
impurities are called network formers. The interstitial impurities are in most cases
large metal ions with low positive charge that can be found between the SiO2
polyhedras. Their presence produce non bridging oxygen ions due to oxygen/metal
ion interactions. These impurities are called network modifiers. Water may also
contaminate the SiO2 structure. Water presence may be part of the oxidation
process in wet conditions or contamination in dry ones, in its presence bridging
oxygen ions will react to form pairs of non bridging hydroxyl groups. [133]
Specific types of oxides are used for microelectronic and other related
applications. SiO2 was chosen as a dielectric material for gate oxides due to it high
band gap (9eV) with good band aliment compared to Si and possible easy and
reliable growth methods on a silicon wafers [130]. The most widely applied oxide in
VLSI is the thermal oxide. This oxide growth regime is based on the oxidation of the
silicon surface at elevated temperatures (between 900C˚-1200C˚) under dry or wet
oxygen conditions [130]. The overall reactions for thermal oxidation are:
(1) Si+O2(g) → SiO2 (for dry oxidation)
(2) Si+2HO2(g) → SiO2 + 2H2(g) (for wet oxidation)
The produced oxide is amorphous in it crystallography and has an exact
stoichiometric composition of SiO2.
Thermal oxide is used in a variety of utilizations and thicknesses, from less than
10nm for a gate oxide to thicker masking and field oxides (from 100 to 1000nm and
more). The growth rate is determined by the temperature, the oxygen pressure and
silicon orientation. [130]
Other wide scale method types used with these oxides are based on chemical vapor
deposition (CVD) or low pressure chemical vapor deposition (LPCVD) of SiO2 from
silane (SiH4) or alkoxysilane reagents [133]. These reagents react with oxygen in
different pressure conditions to form a silicon dioxide layer. The reaction is:
(3) SiH4+O2(g) → SiO2 + 2H2(g)
The growth rate of the oxide is determined by the reagent's pressure in the reaction
chamber. The deposition of CVD oxide usually involves less "thermal budget", on the
other hand the control of deposition rates is not as good is in the thermal oxide
procedure. This means that usually layers only thicker then 10nm will be produced in
such a way. [130]
The next two oxides which are relatively less utilized but are still possible to use in
microelectronic production are the liquid phase deposition (LPD) oxide and the
anodic oxides. Both of these oxides are based on "wet" or liquid phase growing
procedures. The LPD oxide is produced by the deposition of SiO2 from
supersaturated hydrofluosilic acid (H2SiF6). its reaction with water is presented:
(4) H2SiF6+H2O(g) → SiO2 + 6HF(g)
Anodic oxide growth is based on applying anodic currents through the
silicon/electrolyte interface in different electrolytes. The electrolyte and the applied
currents determine the growth rate and thickness of the formed layer. For instance,
in 2M KOH solution the possible growth yield (the oxide thickness at an applied
potential) at potentials between 0-4V is 6.8 Å/V [189]
The last important type of possible silicon dioxides is the native oxide. Silicon
surfaces are spontaneously covered with ~1-2nm of a thin oxide film. This film exists
essentially on all silicon surfaces due to its possible formation by reacting with both
oxygen or water [133]. The thickness of the film is a function of the storage
conditions and the surface pretreatment. For instance after etching in 40% HF for 1
minute and a 100 minute rinse in water the total native oxide thickness will be
0.1nm [157].
Some physical properties of the described oxides are summarized in Table 1.
Oxide Type Density
(gr/cm3)
Breakdown Field
(106V/cm)
Dielectric Constant O/Si
Ratio
Thermal
[156,130]
2.27 10 3.9 2
CVD
[186]
2.3 - 5 ~2
LPD
[187]
2.19 6.3 <3.9 -
Anodic
[184,133]
2.16 6-7 4 2.2
2.2 Silicon Dioxide etching.
Silicon oxide may be etched at different rates as a function of the specific
mineral and oxide type. In general, the chemical dissolution of silicon oxide is
characterized by an etching rate larger than 10-3 Å/sec. Contrarily, the dissolution of
geological systems is characterized by an etching rate that is less than 10 -5 Å/sec. The
chemical dissolution processes are mainly applied in the microelectronics and
electronic devices industries while the geological etching rates are mainly researched
in the Earth sciences [133].
Chemical etching of SiO2 is considered to be one of the key processing steps
in VLSI (Very Large Scale Integration) technology. This process may include wafer
cleaning and pattern delineation. In most cases chemical etching will be the initial
and the last step of all silicon wafer production flows. Without doubt, the most
investigated system of SiO2 chemical etching is the one based on HF based solutions.
The simplicity and wide range of possible etching rates with low native oxide
reformation after treatment makes HF one of the most useful wet treatments in
microelectronic processing [130].
Table 1. Physical properties of different types of SiO2.
As mentioned above, in general the silicon surface is covered with a thin
oxide layer. This can be explained thermodynamically as presented in Figure 2. In
this figure, clear silicon dioxide predominance in a wide range of potentials and pH's
can be seen. Silicon metal is presented as stable only under -0.8VSHE. Although Silicon
hydride existence is possible thermodynamically it is impossible kinetically and
would immediately react to produce SiO2 accompanied by hydrogen gas evolution.
The picture is more complex when the fluoride anion is introduced to the same
water based solution. The stability of the oxide layer is reduced to narrower F -
concentration dependant areas as presented in Figure 3.
At concentrations of 1M [F-], oxide stability is possible only at pH's higher than 7 (fig.
3a). When the F- anion concentration is lower, at 0.01M, stable SiO2 regions exist
between pH's of 0-2 and higher than 4 (fig. 3b). These phenomena may be explained
by the different reactions involved at the SiO2/SiF6- boundary. At low pH values the
reaction involves HF, when at higher pH values only the F- anion participates in SiO2
dissolution. Both reactions may be written as:
Figure 2. Potential - pH equilibrium diagram of the silicon-water system, at 25C˚.(SiO2
as quartz)[131].
(5) SiO2 +6HF → SiF6-2
+ 2H++2H2O (low pH)
(6) SiO2 +F- +4H+
→ SiF6-2
+2H2O (high pH)
In general fluorine may exist (as already presented in fig. 3) in different forms
as a function of HF concentration in the solution. HF, acting a weak acid, will only
partially ionize to from the F- anion. Up to 1M fluoride, the significant fluorine
species in the solution are F-, HF and HF2-. At higher concentrations HF tends to
polymerize at orders higher than n=1 such as (HF)n and F(HF)n- species[198]. The
equilibrium conditions for low HF solution concentrations (less then 1M) and their
pH dependences are [134]:
(7) 2HF →HF2- + H+, log(HF2
-)/(HF)2 = -2.51+pH
(8) HF2- →2F- + H+, log(F-)2/(HF2
-) = -3.85+pH
(9) HF →F- + H+, log(F-)/(HF) = -3.18+pH
At higher concentrations an additional equilibrium reaction needs to be considered
[198]:
(10) HF2- + HF→ H2F3
-, K = [H2F3-]/[HF2
-][HF]
The etch rate of SiO2 varies as a function of the oxide type and also the HF
solution concentration as presented in Figure 4. From this figure a linear dependence
of the etch rate in the HF concentration is shown. The etch rates differ by several
orders of magnitude for different types of oxides with quartz being the slowest and
a b
Figure 3. Potential - pH equilibrium diagram for the system silicon-water - F, at 25C˚.(a) [Si+4]=10-3, [F-]=1, (b) [Si+4]=10-3, [F-]=10-2 [134]
anodic oxide the fastest, reflecting the large difference in the density and structure
of these oxides.
In HF solutions, the etch rate of SiO2 varies as a function of the relevant fluoride
specie. Judge [197] showed in early studies that in NH4F and NH4Cl solutions at
neutral pH, where all the fluoride species are in their unprotonated form, a zero rate
of SiO2 dissolution was recorded.
It was concluded
from these
findings that F- is
not one of relevant
species for SiO2 dissolution. According to the same work it was concluded that the
relevant fluorides for SiO2 etching are HF and HF2-, which are dominant in acidic
solutions. Some reinsurance to the nature of the dissolution process in HF solution
was provided by Kimuyama et al [199]. In this work it was shown that in KF solutions
a low etch rate was recorded and could be described by the following correlation -
etch rate=0.067*[F-]1.11. In the same work it was also shown that in KHF2 solutions the
recorded etch rate was higher by at least one order of magnitude or more, with the
same fluoride specie concentrations. Furthermore, from the relationship between
the etching rate and concentration in HF and KHF2 solutions it was found that the
etch rate differs in these solutions types even when the HF2- concentration was the
same. This may indicate that both the H+ and K+ cations have a significant effect on
the etching reaction as well as on the dissociation equilibrium (eq. 7-9).
Figure 4. Etch rate of SiO2 as a function of HF concentration [133]
Although the presence of higher order fluorohydrogenats (i.e. (HF)nF- ) are known to
exist in HF solutions as already described, their influence on the etching rate is still
uncertain and not fully understood.
The mechanism of SiO2 dissolution in HF solutions is generally based on two
different assumptions. The first assumption is based on the fact that silica etches in
both acidic and basic solutions without HF. This may suggest two possible etch
reactions, one caused by dissociated water species as reactants and the other based
on HF itself as the reactant. As already mentioned above the second assumption
relates to the structural influence of SiO2 on the etch rate. For instance, vitreous
silica films etch rapidly in HF solutions. Yet other phases such as quartz, etch only
slightly in HF. Therefore, some type of surface structure on the silica is necessary to
initiate the HF etching reaction [200]. The described assumptions lead to a number
of possible schematic pathways for SiO2 dissolution as presented in Figure 5.
Based on the same previously explained assumptions, different mechanisms
were proposed by Knotter [171]. In this work a RDS (rate determine state) for SiO2
dissolution was suggested as protonation and deprotonation of the reactive surface
sites. Due to the fact that only silanol can participate in such reactions and not Si-O-
Si bonds, the proposed RDS is based on the exchange of SiOH group with a SiF group
in the silica tetrahedra. The proposed scheme for this reaction is presented in Figure
6:
Figure 5. (a) Four possible surface exposures of the silica tetrahedron, (b) Nucleophilic attack by the hydroxyl ion on a network silicon atom, (C) Simultaneous nucleophilic and electrophilic attack by hydrofluoric acid on the silicon-oxygen network [200]
b c
For a nucleophilic substitution reaction, a nucleophile has to approach the
electrophile from the opposite side of the leaving group. The SiOH group is bound to
three oxygen atoms from the SiO2 matrix, and an approach from the rear side is
impossible. Therefore, the first reaction step is the elimination of OH- or H2O from
the surface to form the reactive intermediate D (fig. 6). After the elimination of
either OH- or H2O, D can react onwards. The reaction of D with H2F2 and HF2- will
result in the reaction product E, SiF. The measured reaction rate will be the product
of the elimination and the addition reactions.
Although the proposed mechanism can explain the reaction occurrence of
fluorohydrogenat production with an order higher than n=1 , the proposed route is
based on Si+ as a highly unstable intermediate product, a mechanism which makes all
that is discussed in the fig. 6 scheme to look problematic.
2.3 Room Temperature Ionic Liquids (RTIL's).
RTIL's (Room Temperature Ionic Liquids) or IL's for short, are considered to be
a new class of solvents. The term IL is currently used to describe a wide range of
liquid materials produced from the conjugation of relatively large organic cations
(e.g. immadazolium, tetraalkylammonium, sulfonium, piperidinium, pyridinium), with
relatively small inorganic anions (e.g. PF6- , BF4
-, AlCl4-, (CF3SO2)2N-, Et3OSO3
-). These
materials which are composed only of ions, may be compared to more classic high
temperature molten salts. The most obvious difference is that the melting point of
the solution has to be near room temperature (25 C˚) [8].
Figure 6. Reaction mechanism of the RDS of SiO2 dissolution of in HFSolutions, SiOH replacement by SiF [171]
IL's offer unique properties while serving both as solvents and as electrolytes.
They are considered to be "green" and more environmentally friendly than regular
organic solutions. Their low vapor pressure, even at considerably elevated
temperatures, provide advantages such as easy containment and recycling. The
enormous number of anion-cation combinations allow considerable variation in their
properties. Moreover, the high intrinsic ionic conductivity, non flammability, thermal
and chemical stability with considerably large electrochemical windows (in some
cases as much as 5V) makes them suitable candidates for high end electrolytes in
electrochemical applications [8,3].
2.3.1 Reference Electrodes in RTIL's.
One of the major setbacks in electrochemical research in RTIL environment is
related to the incapability of generating an accurate and reproducible potential
measurement due to the lack of proper reference electrodes. In most cases, the
electrochemical measurements are conducted with the use of "quasi reference"
electrodes [16]. The use of "quasi reference" or "pseodu reference" electrodes,
mainly Pt or Ag wires [15], is somehow problematic as these electrodes are not
strictly defined electrochemically. The potentials of the electrodes are highly
influenced by impurities in the IL itself or by different oxides at the Pt or Ag
electrodes surface, and thus, are considered as polarizible [52,108]. Other associated
problems with the use of such electrodes include a potential drift with time, even
during a single measurement.
Generally, different configurations of reference electrode do exist (primarily
in aqueous systems but also in non-aqueous systems as well [128]). The most
common electrodes are based on the so called “second type electrodes”, referring to
a metal electrode sustaining equilibrium with a coating of an insoluble salt of the
same metal immersed in an electrolyte solution containing a fixed concentration of
the conjugated anion. This electrode type is largely based on the metals Hg and Ag,
with the appropriate anion solutions [122,129]. The utilization of such electrodes in
IL's is possible and Ag/Ag+ or Ag/AgCl electrodes used in ionic liquids were reported
by several researchers [36,73,74,78,79,99]. One of the problems with such reference
electrode constructions is associated with a relatively complicated fabrication and
production procedures. In some cases macromolecules need to be used as cryptate
agents [36]. In other cases there is a need for additional solutions (in some cases
organic, in other cases aqueous solutions) to be used as a salt bridge or as an
internal phase for proper Ag based salt dissolution [73]. Moreover, possible leakage
of Cl- or other anions from the reference electrode into the IL based solution may
alter the electrochemical properties of the tested electrolyte.
Stable reference potentials may also be obtained with the use of different
Red/Ox couples, mainly Ferrocene (Fc|Fc+) and Cobaltocene (Cc+|Cc) as internal
standards [26,27,142] rather than reference electrodes introduced externally to each
electrochemical system. Although this can be easily achieved and applied in almost
any electrochemical experiment it possess major disadvantages when results
obtained from more than a single working electrode are compared [111].
2.3.2 Chemistry and Electrochemistry of Silicon and Silicon Dioxide in RTIL's.
Although RTIL media seems to be a promising electrochemical environment
for silicon and silicon dioxide studies, few works were published till now on this
specific subject. One of these studies was conducted by Raz et al. [gil 49-52]. The
work included two different parts. The first part of the study was based on
electrochemical deposition of metallic Ruthenium from RuCl3 salt dissolved in
BMImPF6 (1-buthyl-3-methyl-imidazolium cation) IL on n- type silicon wafer as a
working electrode [gil 50,52]. The second part of the study was based on the
electrochemical behavior of Silicon in EMIm(HF)2.3F (1-ethyl-3-methyl-imidazolium
cation) IL as presented in Figure 7 [gil 49,51,52]. This IL synthesize is based on the
chemical reaction of EMImCl with anhydrous HF, producing the (HF)nF- anion with n
varying from n=1 to n=2.6 as a function of reaction time [gil 53-57]. The produced IL's
chemical and physical properties have been extensively studied [gil 53-57] and is
found to posses some unique qualities. Amongst them: a high conductivity in the
order of 100 mS cm-1 which is extremely high in comparison with other IL's, low
viscosity and chemical stability in ambient conditions.
The n value, meaning the number of HF ligands around a central F atom,
tremendously changes the IL properties which can be seen as a direct function of n.
EMIm(HF)nF with n=1.0-2.0 consist of HF2- and (HF)2F- anions, while the anionic
species in EMIm(HF)nF with n=2.0-3.0 are (HF)2F- and (HF)3F-. The distribution of the
anionic compounds is the result of a dissociation reaction of fluorohydrogenates
described by:
(11) (HF)n+1F- ↔ (HF)nF- + HF
The rapid HF exchange in reaction (11) leaves no free HF moleculess in the IL. Also,
the described reaction is similar to what's already presented in chapter 2.1.1
involving equilibrium reactions in high-concentration HF solutions (eq.10). Larger n
values make for lower viscosity and higher conductivity. This can be attributed to a
decrease in ion associations as the basicity decreases at larger n values, in other
words as a function of HF ligands attached to the central fluorine atom. It could also
be associated with a decrease in the electrostatic interaction between the
fluorohydrogenat anion and the imidazolium cation as a function of possible charge
distribution around the anionic species [gil 54]. Melting temperature and
hygroscopicity also decrease as n increases [20]. The electrochemical window of
EMIm(HF)nF on Glassy Carbon (GC) varies in the range of 2.9-3.4V, but no correlation
was found between the electrochemical window and n. The difference in the
electrochemical window width is most probably a result of the difference in the
cathodic limiting potentials of the IL. At cathodic limiting potentials the reaction
includes both the reduction of cation and the reduction of the anion, accompanied
by H2 evolution [gil 54]. The anodic limit reactions are observed at similar potentials
without any dependence on n, the reaction being the anion decomposition followed
by fluorination of the cation.
The anion of the ionic liquid EMIm(HF)2.3F consists of 70% (HF)2F- and 30%
(HF)3F-. The specific conductivity of this exact stoichiometry is 100 mS cm -1 at 298 K
[gil 53], the highest of all RTIL's family solutions, comparable to that of 0.5 M KCl
aqueous solution. The viscosity is 4.85 cP, and the liquid temperature ranges from
180 to 350 K. The electrochemical window is 3.2 V on GC working electrode. Due to
the fact that the cathodic limiting reaction involves H2 evolution, the electrochemical
window narrows when using an electrode with low hydrogen overpotential, such as
Pt. It has been reported that the ionic liquid had been used for different
electrochemical processes including use as an electrolyte for a double-layer
capacitor [gil 59], H2/O2 fuel cells [gil 60] and Silicon/air batteries [gil si/air]. In the
last two applications it was shown that water does not decrease the performance of
the IL or of the tested cell [gil/water]. Other applications of EMIm(HF)2.3F also include
usage as a medium for fluorination [gil 61-64], and the synthesis of other RTIL's [gil
65-66].
EMIm(HF)2.3F was applied as a electrolyte for porous formation in both n-Si
and p-Si [gil 51,49,52]. The proposed mechanism for porous structure formation was
suggested and is related to Si-F bonds at the silicon surface. The applied anodic
potential generates Si-F bonds at local spots on the silicon surface, causing a local
concentration of surface states, an observation confirmed with XPS examination. The
presence of Si-F bonds generates increased electrical fields on these specific surface
locations. Pit formation exists at the same Si-F positions, therefore initiating spots for
porous silicon. Pore growth is possible due to successive cycles of Si-F bond
formation on the silicon surface and to competitive silicon etching. The anodic
dissolution of n-Si is assumed to be an anodic oxidation of silicon, accompanied by
the formation of gaseous SiF4 and dissolved SiF6-2 ions [gil 49]. In most cases low and
medium doped n-Si dissolution is possible only under illumination [gil 72], while in
the case of EMIm(HF)2.3F no illumination was needed.
To the best of our knowledge only one work was published relating to the
chemical behavior of SiO2 in RTIL media, published by Nockemann et al. [18]. In this
specific study a HbetTf2N (Protonated betaine bis(trifluoromethylsulfonyl)imide) was
applied. The solubility of different metal oxides in this specific IL is possible due to
oxide reactions with the carboxylic acid group of the ionic liquid to form carboxylate
complexes and water. Nevertheless, SiO2 showed insoluble or very poorly soluble
behavior in this IL.