Reactions of Etched, Single Crystal (111)B-Oriented InP to Produce Functionalized Surfaces with Low Electrical Defect Densities

Article abstract of DOI:10.1021/jp992290f

Synthetic routes have been developed that allow attachment of a variety of functional groups to etched, single-crystal InP surfaces. Benzyl halides, alkyl halides, silyl halides, and esters reacted readily with InP to yield covalently attached overlayers

Full text of DOI:10.1021/jp992290f

10838
J. Phys. Chem. B 1999, 103, 10838-10849
Reactions of Etched, Single Crystal (111)B-Oriented InP To Produce Functionalized
Surfaces with Low Electrical Defect Densities
Marcel Sturzenegger, Nicholas Prokopuk, C. N. Kenyon, William J. Royea, and
Nathan S. Lewis*
DiVision of Chemistry and Chemical Engineering, Noyes Laboratory, California Institute of Technology,
Pasadena, California 91125
ReceiVed: July 8, 1999
Synthetic routes have been developed that allow attachment of a variety of functional groups to etched, single-
crystal InP surfaces. Benzyl halides, alkyl halides, silyl halides, and esters reacted readily with InP to yield
covalently attached overlayers on the semiconductor surface. High-resolution X-ray photoelectron spectroscopy
(XPS) revealed that the functionalization chemistry was consistent with the reactivity of surficial hydroxyl
groups. Analysis of the XP spectra of the (111)B-oriented (P-rich) face in ultrahigh vacuum revealed signals
ascribable to a monolayer of oxidized P atoms on the etched (111)B InP surface. The lack of reactivity of the
(111)A-oriented (In-rich) face with these same functionalization reagents is therefore attributed to the difference
in the nucleophilicity and acidity of the In and P oxides that are present on the (111)A and (111)B faces,
respectively. The coverage of benzylic groups obtained through functionalization of (111)B-oriented InP with
benzyl halides was estimated to be 4 × 1014 cm2. This coverage implies that the functionalization can only
proceed at alternate surface P atom sites in this system, which is expected from molecular packing considerations
of these particular functional groups. Photoluminescence decay measurements were performed to investigate
the electrical properties of the etched and modified InP surfaces, and these data indicated that the surface
recombination velocity of the functionalized InP surface was ≈102 cm s-1. This low surface recombination
velocity implies that <1 electrically active defect is present for every 105 atoms on the modified InP surface,
indicating that high electrical quality can be maintained while introducing a variety of chemical functionalities
onto the (111)B surface of InP.
I. Introduction
properties of the InP surface prevents fabrication of the desired
device structures and interfaces. For example, improved semi-
conductor/metal Schottky barriers could be obtained if Fermi
level pinning could be prevented on InP surfaces.1,6 Additionally,
improved metal-oxide-semiconductor field-effect transistors
could be obtained if suitable chemical control were achieved
over the chemistry of InP surfaces. The InP/native oxide
interface has been shown to have a low surface recombination
velocity in contact with air ambients,7-9 so oxide/hydroxide
functionalities on this surface might be useful starting reagents
for effecting modification of this semiconductor without induc-
ing a concomitant degradation in the electrical properties of the
resulting devices.
One of the major goals in the area of semiconductor
electrochemistry is to achieve molecular-level control over the
physical, chemical, and electrical properties of semiconductor
surfaces. For some semiconducting solids, such as GaAs,
improved methods to control the surface characteristics are
crucial because the highly defective properties of the native
GaAs surface preclude fabrication of entire classes of electrical
and optical devices. For other semiconductors, such as Si and
InP, the native or deliberately oxidized surfaces have electrical
properties that are satisfactory for many device applications.1,2
Nevertheless, control over the surface chemistry of these solids
is still a notable goal because as devices become smaller, the
surface properties will play an increasingly important role in
determining the overall device behavior. In addition, procedures
to improve adhesion of metals, to limit oxidation, or to inhibit
corrosion on such surfaces would be very useful in certain device
applications. Synthetic strategies that allow the introduction of
chemical functionality onto such semiconductor surfaces, with-
out introducing significant levels of electrical defects, are
therefore highly desirable.
Previous studies of InP surface chemistry have mostly em-
phasized oxide growth and passivation reactions.10,11 Little ef-
fort has been directed toward functionalizing this surface with
molecular reagents while maintaining a high degree of electrical
perfection. Waldeck and co-workers have successfully modified
InP surfaces using a high-temperature exposure to thiols,12 but
the chemical processes responsible for this surface modification
approach have not yet been elucidated. In another approach,
benzyl bromides were found to react selectively with the P-rich,
(111)B orientation of InP.13 This procedure afforded a useful
route to molecular level control over the adhesion of conducting
polymers onto the InP surface.13 In a recent X-ray photoelectron
spectroscopy (XPS) and chemical reactivity study, the reactivity
of the P-rich (111)B InP surface with alkyl halides, benzyl
halides, and silyl halides was ascribed predominantly to surface
In this work, we describe an apparently general synthetic
strategy for the functionalization of InP surfaces. InP is a useful
target for surface modification because its direct optical band
gap and its high electron mobility make InP a material of choice
in several optoelectronic technologies.1,3-5 In many types of
device applications, however, InP yields unsatisfactory perfor-
mance because the lack of chemical control over the electrical
10.1021/jp992290f CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/19/1999

Reactions of Single-Crystal (111)B-Oriented InP
J. Phys. Chem. B, Vol. 103, No. 49, 1999 10839
hydroxyl moieties.14 In this manuscript, we provide a full
description of these surface reactions. We also present an
electrical characterization of the resulting InP surface to facilitate
comparison of the level of electrical defects of this semiconduc-
tor surface before and after the chemical functionalization steps.
was cooled to -5 °C to give white crystals of triethylammonium
chloride that were removed by filtration. The filtrate was washed
three times with water, and the product was isolated as a viscous
liquid upon evaporation of the organic phase. 1H NMR (CDCl3,
δ): 1.2-1.6 (m, 16H), 1.9 (m, 4H), 3.4 (t, 2H), 4.4 (t, 2H), 7.7
(d, 2H), 8.2 (d, 2H). n-Dodecyl-4-(trifluoromethyl)-benzoate (p-
CF3C6H4COO(CH2)11CH3) was prepared in an analogous fash-
II. Experimental Section
1
ion by reacting the acyl chloride with 1-dodecanol. H NMR
A. Chemicals. Reagent grade tetrahydrofuran (THF), metha-
nol (CH3OH), and acetonitrile (CH3CN) were obtained from
EM Science. The CH3OH was dried over magnesium turnings,
and a small amount of I2 was added before distillation to
facilitate the removal of water. Acetonitrile was first dried over
CaH2 and then dried over P2O5. The solvents were distilled from
their drying agents under a dinitrogen (N2) atmosphere and
stored under N2.
Bromine (Br2; 99.5%) was obtained from EM Science. Boron
trifluoride diethyl etherate [(C2H5)2O‚BF3], 4-(trifluoromethyl)-
benzyl bromide (p-CF3C6H4CH2Br; 98%), and 4-methylben-
zotrifluoride (p-CF3C6H4CH3; 98%) were obtained from Aldrich
Chemical Company, and 4,4,4-trifluoro-1-iodobutane [CF3-
(CH2)3I; bp, 126-127 °C] was obtained from Lancaster Inc.
These reagents were used without any further purification.
(3,3,3-Trifluoropropyl)dimethylchlorosilane [CF3(CH2)2Si(CH3)2-
Cl; PCR, 97%] and triethylamine [(C2H5)3N; Aldrich, 99+ %]
were outgassed by 3 freeze-pump-thaw cycles prior to use.
Lithium chloride (LiCl), lithium perchlorate (LiClO4), and
potassium bromide (KBr) were obtained from J. T. Baker. These
reagents were converted to their anhydrous form by heating for
8 h in quartz tubes at ∼250 °C under an active vacuum (2 ×
10-2 Torr). Sodium hexafluoroantimonate(V) (NaSbF6; 98%)
was obtained from Alfa and was used as received.
(CDCl3, δ): 0.9 (t, 3H), 1.2-1.5 (m, 18H), 1.8 (m, 2H), 4.4 (t,
2H), 7.7 (d, 2H), 8.2 (d, 2H).
General procedures to synthesize indium(III)-phosphine
complexes can be found in a review by Roundhill.18 To prepare
tribromobis(tris(4-fluorophenyl)phosphine)indium(III) (InBr3-
[(C6H4F)3P]2), a solution of tris(4-fluorophenyl)phosphine (1.0
g, 3.16 mmol) in 5 mL of anhydrous ethyl acetate was added
slowly under N2(g) to a solution of indium(III) bromide (0.545
g, 1.51 mmol) in 5 mL of the same solvent. The solution was
stirred for 1 h and then warmed to 50 °C. n-Hexane was then
added until precipitation was initiated. The precipitate was
redissolved by adding 1 mL of anhydrous ethyl acetate, and
the resulting solution was slowly cooled to room temperature.
The product was filtered and washed four times with cold
n-hexane (mp 180 °C). Anal. for InBr3[(C6H4F)3P]2 Calcd C,
43.8, H, 2.46. Found: C, 43.7; H, 2.34.
2. InP Sample Preparation. Sample preparation was per-
formed in a glovebox that was maintained under an inert N2
atmosphere. Sample transfer to the XP spectrometer was also
carried out under N2(g). The samples were etched just prior to
surface derivatization, and no sample was reused more than five
times to avoid the possibility of increasing the surface roughness
and obtaining erroneous coverage values through repeated
etching of the surface.
The initial and final surfaces were characterized by XPS.
Introduction of -CF3 groups into the reagent was very useful
because F provides the highest XP sensitivity among the
nonmetallic elements and because the fluorinated carbon (CF)
yields a C 1s peak that is well separated from that of hydro-
carbons or from peaks due to adventitious carbonaceous ma-
terial.
3. Etching Procedures. Triangular or rhomboidally-shaped
InP samples with an edge length of about 4 mm were
successively immersed for 60 s in a freshly prepared, diluted
Br2 solution (0.05% in CH3OH) and in a 4 M NH3-CH3OH
solution. After each step, the samples were rinsed thoroughly
with CH3OH. The entire procedure was then repeated and the
samples were blown dry with N2(g). This procedure was
employed for the (111)B-, (111)A-, and (110)-oriented InP
samples.
InP single crystals were obtained from CrystaComm (Moun-
tain View, CA). The (111)B-oriented wafer was n-type, whereas
the (111)-oriented wafer with both A and B sides polished and
the (110)-oriented wafers were doped p-type. The dopant
densities were 5.5 × 1015, 4.2 × 1018, and 2 × 1018 cm-3
,
respectively, as reported by the manufacturer. The surfaces were
polished by the manufacturer by applying a dilute HF etch. The
(111)A and (111)B faces of the wafer that had been polished
on both sides were distinguished by using the “AB etch”.15 The
orientation of the (110) face was verified by powder X-ray
diffraction (XRD) methods.
B. Reactions. 1. Reagents. 4-(Trifluoromethyl)benzyltri-
phenylphosphonium bromide (p-CF3C6H4CH2P(C6H5)3Br) was
prepared according to a general procedure given by Hauser et
al.16 A solution of 4-(trifluoromethyl)benzyl bromide (1.20 g,
5 mmol) in 20 mL of CH3CN was added to a solution of
triphenylphosphine (1.31 g, 5 mmol) in 30 mL of CH3CN under
N2(g). The reaction mixture was stirred at 65 °C for 8 h. The
amount of solvent was reduced to 40 mL under vacuum. While
the solution was kept at 65 °C, n-hexane was added until the
product started to precipitate (ca. 5 mL). After redissolving the
precipitate by adding 0.5 mL of CH3CN, the solution was slowly
cooled to room temperature. The resulting crystalline product
was isolated by filtration and washed four times with n-hexane.
1H NMR (CDCl3, δ): 5.8 (d, 2H), 7.3-7.9 (m, 19H).
12-Bromo-n-dodecyl-4-(trifluoromethyl)-benzoate (p-CF3C6H4-
COO(CH2)12Br) was prepared according to a procedure de-
scribed in the literature.17 A solution of 12-bromo-1-dodecanol
(1.15 g, 4.34 mmol) in 10 mL of CH2Cl2 was added to a solution
of 4-(trifluoromethyl)benzoyl chloride (1.0 g, 4.77 mmol) and
triethylamine (0.88 g, 8.7 mmol) in 10 mL of the same solvent.
After the solution was stirred for 1 h at room temperature, ∼10
mL of the solvent was removed under vacuum. The solution
4. Surface DeriVatization of (111) and (110) Oriented InP.
Surface derivatization reactions and related control experiments
were performed using nominally identical procedures. Details
of the experimental conditions used for the surface derivatization
reactions are listed in Table 1. As an example of the procedures
used in these reactions, freshly etched InP samples were
immersed for 60 min in a 0.2 M solution of p-CF3C6H4CH2Br
in CH3CN at 62 ( 3 °C. The samples were removed from the
derivatization solution, thoroughly rinsed with hot (65 °C) CH3-
CN, and blown dry with N2. Similar methods were used, with
modifications in the conditions as indicated, for the other
reactions of Table 1.
5. Anion-Exchange Reactions. Immediately following the
surface derivatization, InP samples were immersed into a 0.01
M solution of either LiClO4, NaSbF6, or KBr in CH3OH.
Samples were maintained in the solution for 120 min at room

10840 J. Phys. Chem. B, Vol. 103, No. 49, 1999
Sturzenegger et al.
TABLE 1: Parameters for Surface Derivatization Reactions of (111)- and (110)-Oriented InP Surfaces
reagent concentration
promoter concentration
immersion
time (min)b
a
derivatization reagent
(mol L-1
)
promoter
(111)B InP
(mol L-1
)
p-CF3C6H4CH2Br
0.2
0.2
0.2
0.2
0.2
0.4
0.2
0.2
0.2
0.1
a) none
b) (C2H5)3N
none
a) none
b) (C2H5)3N
none
a) none
b) (C2H5)3N
none
60
0.02
0.02
0.1
60, 120 180
60
60, 120
60, 120, 180
30
120
60, 120, 240
120
p-CF3C6H4CH3
CF3(CH2)3I
(C2H5)2O‚BF3
CF3(CH2)2Si(CH3)2Cl
p-CF3C6H4COO(CH2)11CH3
p-CF3C6H4COO(CH2)12Br
none
120
(111)A InP
(C2H5)3N
p-CF3C6H4CH2Br
0.2
0.02
120
(110) InP
p-CF3C6H4CH2Br
CF3(CH2)3I
0.2
0.2
(C2H5)3N
(C2H5)3N
0.02
0.02
60, 120
60, 120
a Acetonitrile (CH3CN) was used as the solvent for all reagents, except for (C2H5)2O‚BF3, which was dissolved in a solution of hexanes/CH2Cl2
(10:1). b Reaction temperatures were 62 ( 3 °C, except for the reaction with (C2H5)2O‚BF3, which was performed at 40 ( 3 °C.
temperature. After this period, the samples were removed from
the solution and were thoroughly rinsed with pure CH3OH.
C. X-ray Photoelectron Spectroscopy (XPS). 1. Experi-
mental Procedures. The XP spectra were recorded with an
M-Probe surface spectrometer (Surface Science Instruments).
The spectrometer was divided into two separate chambers. One
chamber was equipped with an Ar+ sputtering unit, a heating
stage, a quadrupole mass spectrometer (VG Scientific), and a
low-energy electron diffraction (LEED) system. The second
chamber contained the XPS and ultraviolet photoelectron
spectrometer (UPS) sources and a hemispherical analyzer. Each
chamber was pumped with a cryogenic pump (CTI-Cryogen-
ics). The base pressure during data collection was < 5 × 10-10
Torr. The mass spectrometer identified water vapor as the main
gaseous component in the ultrahigh vacuum (UHV) chamber.
The spectrometer was connected to a glovebox via a high-
vacuum load-lock. This arrangement allowed anaerobic transfer
of the samples from the glovebox into the spectrometer. All
spectra were recorded with focused and monochromatized Al
KR1,2 irradiation (hν ) 1486.6 eV), and the X-ray beam was
incident on the surface at an angle of 55° with respect to the
surface normal. The analyzer was also positioned at an angle
of 55° with respect to the surface normal.
background. A pass energy of 54 eV was used to study the P
2p region in detail.
The InP samples were affixed with gold springs onto an
electrically grounded sample holder. Insulating samples were
mechanically powdered in a mortar and were mounted onto
double-sided adhesive tape. A fine nickel mesh was fixed ∼2
mm above insulating samples.19 During data acquisition, the
static charge was neutralized with a low-energy-electron flood
gun. The energy of the applied electron beam was adjusted so
that the measured BEs followed the applied energy linearly and
so that the fwhm values were minimal. Usually an electron beam
of 3 eV was sufficient to fulfill these two conditions. Using
this procedure, the peaks obtained from insulating samples were
typically not >20% broader than those of conducting samples.
2. Data Analysis. To correct for charging effects, the
measured BEs were referenced to the BE of adventitious carbon,
which was taken to be 285.0 eV. Chemical compositions of
homogeneous samples and attached overlayers were derived
from normalized peak areas and from Wagner’s empirically
derived atomic sensitivity factors20 for XPS. The normalized
peak areas were calculated with the standard software provided
with the M Probe XP spectrometer after a linear background
correction had been carried out.
The Au 4f7/2 line [binding energy (BE) ) 84.00 eV] from a
sputter-cleaned gold foil was used to calibrate the energy scale
of the spectrometer. The linearity of the energy scale was
checked with a sputter-cleaned copper foil [BE(Cu 2p3/2) )
932.67 eV]. On both metals, adventitious carbon was observed
at a BE of 285.0 ( 0.1 eV. The Au 4f7/2 line also provided a
measure of the instrumental line width. When the gold foil was
irradiated with the 400 × 1000 µm X-ray spot and the pass
energy of the analyzer set at 105 eV, the full-width-at-half-
maximum (fwhm) of the Au 4f7/2 line was 1.14 ( 0.01 eV.
With a pass energy set at 54 eV, a fwhm of 0.90 ( 0.01 eV
was observed.
All samples [except the UHV cleaved (110) InP surfaces]
were irradiated with an ellipsoidal X-ray spot with dimensions
of 400 × 1000 µm. The UHV-cleaved (110) InP samples were
irradiated with an ellipsoidal X-ray spot of 200 × 750 µm.
Survey scans (BE ) 0-1000 eV) were recorded with a pass
energy of 155 eV. Binding energies and chemical compositions
reported in this work are based on high-resolution spectra taken
with a pass energy of 105 eV. Sufficient scanning was performed
to provide peak heights that were at least 5000 counts above
The surface coverage of derivatized samples can be expressed
either as the overlayer-substrate intensity ratio, measured as
I(F 1s)/I(In 3p3/2) (with I being the peak intensity in arbitrary
units), or as an absolute coverage Γ in atoms per cm2. The
former is a semiquantitative measurement and can only be
rigorously used for a comparison between samples that are
derivatized with identical molecular overlayers. The absolute
coverage was calculated by applying an overlayer-substrate
model.21 The models shown in Scheme 1 are based on the
assumption that the molecules are oriented perpendicular to the
surface, as has been observed for strongly chemisorbed mol-
ecules on metal surfaces such as n-alkane thiols or benzyl
mercatpans on Ag (111) and Pd (111).22 The C4-hydrocarbon
chain was assumed to possess an all-trans conformation by
analogy to n-alkane thiols on gold23 as well as on GaAs (100).24
The bond lengths were taken from Allen et al.25 Based on these
-
assumptions, the thickness of the p-CF3C6H4CH2 overlayer on
InP was estimated to be 8.7 Å, whereas the thickness of the
-
CF3(CH2)3 overlayer on InP was 6.4 Å.
Scheme 1 illustrates the model structures used for the
coverage computations, in which the -CF3 groups are confined

Reactions of Single-Crystal (111)B-Oriented InP
J. Phys. Chem. B, Vol. 103, No. 49, 1999 10841
SCHEME 1: Representation of p-CF3C6H4CH2- and
CF3(CH2)3-units Bound to the (111)B InP Surface
sured λ values and the theoretically derived IMFPs, corrected
for inelastic scattering, agreed to within 10% (Table 2).
Equations 1-3 were used to calculate the thickness of indium
phosphate overlayers on oxidized InP samples. The number
density of P atoms (per cm3) in the overlayer was taken to be
the same as that in crystalline InPO4; namely, 1.39 × 1022
cm-3 29
Both the escape depth of the substrate and that of the
.
overlayer were taken to be 23 Å, as derived from the IMFP of
InP.27,32 To estimate the number of oxidized P atoms on an
anaerobically etched (111)B surface, the surface P atoms and
surface-bound OH groups were considered to be the overlayer.
The overlayer thickness was taken to be 3.5 Å, and the escape
depth was assumed to be the same as that of bulk InP; namely,
23 Å.
D. Surface Recombination Velocity Measurements. Time-
resolved photoluminescence measurements were performed
using a time-correlated single-photon-counting apparatus that
has been described previously.33,34 Rhodamin 6G (Exciton) was
used in the dye laser to provide an excitation at 600 nm. The
repitition rate was set to 152 kHz to ensure that the delay
between pulses was substantially larger than the time scale for
the observed InP luminescence decay (6 ms). The beam was
directed onto the sample at an angle of ∼45° and was focused
using a 25.4-mm focal length, diffraction-limited, achromatic
lens. The incident photon flux was estimated to be 2.7 × 1013
photons cm-2 pulse-1, indicating that the concentration of
photogenerated carriers exceeded the concentration of thermally
generated electrons in the lightly doped n-InP sample by >2
orders of magnitude. Under such high-level injection conditions,
the influence of any residual electric field on the observed PL
decay was minimized.35 Emission corresponding to the band-
to-band luminescence of InP (925 nm) was selected with a
monochromator and was passed to the detection system.
The two-dimensional device simulator ToSCA36 was used
to simulate the experimentally observed PL decay curves. The
software package employs Fermi-Dirac statistics and simul-
taneously solves Poisson’s equation and the current continuity
equations; a more detailed description of ToSCA has been
published previously.35,37 The excitation was approximated by
a Gaussian-shaped pulse with a fwhm of 70 ps that corresponds
to the system response of the experimental setup. The nonra-
diative bulk lifetime (240 ns) was obtained from an exponential
fit of the tail of the PL decay curve at t > 350 ns.
Because the effect of surface-state trap saturation can become
important under high-level injection conditions,38 nonradiative
recombination via surface states was simulated using a trap
model35 that treats the occupancy of recombination centers
(traps) as a dynamic process. The trap model in the present case
contrasts with the conventional Shockley-Read-Hall treatment
that assumes steady-state conditions (i.e., no net change in the
occupancy of traps). In these simulations, the rate constants for
electron and hole capture were assumed to be equal (10-8 cm3-
s-1). This value corresponds to the product of the thermal
velocity of the carriers (Vth ) 107 cm s-1) and the capture cross
section of the traps (typically σ ) 10-15 cm2 for neutral traps).
The surface traps were positioned energetically at the intrinsic
Fermi level, where such states can interact with both the
conduction and valence band and thus provide the most efficient
centers for recombination.
to the upper part of the overlayer. To localize the origin of the
F 1s photoelectrons more precisely, the overlayer (o) was
schematically divided into two sections and the -CF3 layer was
treated as a second overlayer (o′) with a thickness of 1.9 Å. A
mathematical treatment of the two-layer-substrate model for XPS
signals has been performed by Ebel.26
The surface coverage was then calculated using the following
equations:27,28
do′
IA,o′ ) I∞A‚nA,o′
‚{
1 - e-
}
(1)
(2)
(3)
λA(EA)‚cosâ
do
IB,s ) I∞B‚nB,s‚e-
λA(EB).cosâ
nA,o′
ΓA )
do′
The index A represents the element of interest in the overlayer
o′, and B represents the element of interest in the substrate s.
In eqs 1-3, IA,o′ and IB are the measured intensities of the
particular core level peaks of A and B, respectively; I∞A and IB∞
are the sensitivity factors (SFs) of elements A and B, respec-
tively; nA and nB are the number of atoms per unit volume of
elements A and B, respectively; do is the thickness of the entire
overlayer, and do′ is the thickness of the -CF3 layer; λA(EA)
and λA(EB) are the escape depths in the overlaying material at
the energies EA and EB, respectively; the angle â between the
analyzer and the surface normal was 55° for all the measure-
ments described herein; and Γ is the surface coverage in atoms
per cm2.
Based on the crystallographic data for InP,29 the number
density of P atoms (and of In atoms) in bulk InP was calculated
to be 1.94 × 1022 cm-3. An atomically perfect (111)B InP face
is terminated by 6.7 × 1014 P atoms per cm2, and the number
of P atoms on a atomically perfect (110) InP surface was
calculated to be 4.1 × 1014 cm-2. Values for λ in the organic
overlayer were derived from the work of Laibinis et al.,30 in
which the escape depth was determined for n-alkane thiols on
group IB metals (Cu, Ag, Au) at different photoelectron
energies. A quantity that is similar to λ is the inelastic mean
free path (IMFP).31 In contrast to λ, the IMFP is based on a
theoretical approach and must be corrected to take into account
inelastic scattering.27 For polystyrene, the experimentally mea-
III. Results
A. Reactivity of (111)B-Oriented InP Surfaces with Benzyl
Bromide. Figure 1 displays the XP spectra for a (111)B-oriented
InP surface before and after exposure to p-CF3C6H4CH2Br in

10842 J. Phys. Chem. B, Vol. 103, No. 49, 1999
Sturzenegger et al.
TABLE 2: Escape Depths (λ),30 Inelastic Mean Free Paths,32 and Wagner’s Empirically Derived Sensitivity Factors (SF)48 for
Some Elements of Interest
parameter
Br 3d
1418
Si 2p
1385
P 2p
C 1s
In 3d5/2
O 1s
I 3d
867
In 3p3/2
kinetic energy (eV)
escape depth, λ (Å)a
corrected inelastic mean free
path (Å)b
sensitivity factor, SFlit
sensitivity factor, SFexp
1358
38.8
36.9
1199
35.4
33.5
1042
31.9
30
955
30.0
822
27.1
24.9
c
0.83
0.37
0.39
0.39
0.25
0.27
3.9
4.62
0.66
1.75
6.0
1.85
d
a Escape depths in n-alkane thiols on group IB metals.30 b Inelastic mean free path for polystyrene,32 corrected for inelastic scattering.27 c Wagner’s
sensitivity factors.48 d Sensitivity factors derived from closely related organic compounds and single-crystal (110) InP.
compensate the excess positive charge that is introduced onto
the InP surface. To rule out the possibility that the Br- anion
had exchanged with OH- or with another anion during the
functionalization and/or rinsing steps, (111)B InP surfaces that
-
had been modified with p-CF3C6H4CH2 groups were subse-
quently exposed to solutions that contained large excesses of
anions. Metathesis of a putative surface-bound anion with these
other anions should produce clear signatures in the XP spectra
that are characteristic of the atomic constituents of these other
anionic species.
Table 3 summarizes the XPS data obtained from such surfaces
after exposure to LiClO4 or NaSbF6, each at a concentration of
0.01 M in CH3CN. In all cases, only traces of the atoms in the
anions were detectable on the InP surface. Furthermore, rinsing
the surface with pure solvent typically removed these signals
to undetectable levels. Notably, rinsing the p-CF3C6H4CH2Br-
treated (111)B InP surface with a KBr solution (0.01 M in CH3-
CN) after surface modification also lowered the residual Br
signal. This collective lack of reactivity toward anion exchange
of the modified surface strongly implies that the initial func-
tionalization reaction does not produce free or bound Br-. These
Figure 1. The XPS survey scans of anaerobically etched (111)B InP
(a) before and (b) after exposure to p-CF3C6H4CH2Br. The inset
magnifies the C 1s region and shows the presence of fluorinated carbon
(CF) on surfaces that had been exposed to p-CF3C6H4CH2Br.
CH3CN. The survey scan of the etched surface displayed the
expected peaks for In and P (Table 3, Figure 1a). Minor traces
of Br at 69.1 eV were detectable, but the intensity of this peak
varied from sample to sample. A high-resolution scan of the P
2p region of such etched samples showed no detectable
quantities of phosphate at 133.3 eV.39
-
results therefore suggest that binding of the p-CF3C6H4CH2
group does not occur through alkylation of a P lone pair on the
InP surface.
Reaction of this surface with p-CF3C6H4CH2Br produced new
XPS signals ascribable to F atoms as well as new signals
ascribable to fluorinated C (CF; Figure 1b and inset). After
correction for the cross-section differences between C and F
atoms, the ratio of these signals was close to the 1:3 ratio
expected for a -CF3 group. No detectable changes in high
resolution scans of the In 3d or P 2p XPS regions were observed
as a result of the surface modification process. Thus, the
oxidation state of these elements remained constant in the 35
Å near-surface region that is probed by the XPS experiment at
photoelectron energies of 444.5 eV (In 3d5/2) or 128.8 eV (P
2p). After surface modification, the C signal increased relative
to its level on an etched surface; however, the presence of
adventitious C on all samples precluded a quantitative analysis
of this region of the spectrum.
C. Reactivity of Etched (111)B InP Surfaces with Alkyl
Halides, Boron Trifluoride Etherate, and Silyl Halides. To
characterize further the reactivity of the (111)B-oriented InP
surface, etched (111)B InP samples were exposed to CF3(CH2)3I
in CH3CN. As shown in Figure 2 and Table 3, CF3(CH2)3I
reacted similarly to p-CF3C6H4CH2Br and yielded persistently
bound surface alkyl groups with no evidence for concomitant
formation of surface-localized iodide anions. For example, the
XPS spectra of (111)B-oriented InP surfaces obtained after
reaction with CF3(CH2)3I revealed a cross-section-corrected
CF/F ratio of 0.38 and a corrected I/F ratio of <0.14 (Table 3).
To probe the propensity of the (111)B-oriented InP surface
functionality to act as a Lewis base, the etched (111)B surface
was exposed to (C2H5)2O‚BF3 that had been dissolved in a
solution of hexanes/CH2Cl2 (10:1). Only small amounts of F
were detectable after exposure of (111)B-oriented InP to
(C2H5)2O‚BF3. On such surfaces, the F 1s/In 3p3/2 intensity ratio
was typically <0.1. Based on this value and assuming an escape
depth of 15 Å and a thickness of 2.8 Å, the number of bound
BF3 groups is estimated to be <1 × 1014 cm-2, compared with
6.7 × 1014 cm-2 surface P atoms on an atomically flat (111)B
InP surface.
Control experiments indicated that the bromide functionality
in the p-CF3C6H4CH2Br reagent was required for surface
modification because negligible CF and F XPS signals were
observed on (111)B InP surfaces that had been exposed to
p-CF3C6H4CH3. The CF and F levels after reaction with
p-CF3C6H4CH2Br were essentially unchanged after thorough
rinsing or exposure to clean CH3CN for 60 min. The spectral
changes depicted in Figure 1 are thus consistent with expecta-
-
tions for chemisorbed p-CF3C6H4CH2 groups bound to the InP
surface.
To distinguish between the presence of P lone pairs and
P-OH functionalities on etched (111)B InP crystals, the
reactivity of the (111)B-oriented surface was investigated with
a silyl chloride reagent. Figure 3 and Table 3 present the XPS
data obtained after reaction of the (111)B InP surface with
CF3(CH2)2Si(CH3)2Cl in the presence of (C2H5)3N in CH3CN.
A monochlorosilane was chosen to prevent adventitious water
B. Anion-Exchange Reactivity of the Functionalized
(111)B InP Surface. After exposure to p-CF3C6H4CH2Br, the
Br/F ratio in the XP spectra of (111)B-oriented InP samples
was < 0.02 (Table 3). This ratio would not be expected for an
alkylation reaction in which a phosphonium ion was formed
because such a process requires the presence of an anion to

TABLE 3: XPS Peak Positionsa and Intensitiesb of Derivatized (111) and (110) InP Surfaces
P 2p
BEd
In 3d5/2
BE
In 3p3/2
BE
CF 1s
O 1s
Ie
F 1s
X
intensity ratio
POx/InPf
atomic ratioc
derivatized surface
I
I
BE
I
BE
I
BE
I
CF/F
X/F
(111)B InP
anaerobically etched InP
InP + p-CF3C6H4CH2Br
InP + p-CF3C6H4CH3
128.7
128.6
128.7
0.29
0.31
444.5
444.4
2.75
2.82
665.8
665.7
665.5
665.7
0.17
0.18
<0.05
<0.05
<0.05
292.6
292.9
292.7
292.7
292.6
0.027
688.2
688.3
688.3
0.26
0.08
0.19
Br 3d
Cl 2p
Sb 3d
Br 3d
I 3d
68.9
sg
<0.02
0.41
0.58
0.46
0.45
0.38
<0.10
<0.02
<0.02
0.07
InP + p-CF3C6H4CH2Br
then: +LiClO4-CH3OH
InP + p-CF3C6H4CH2Br
then: + NaSbF6-CH3OH
InP + CF3C6H4CH2Br
0.028
665.5
665.6
0.029
688.2
688.2
0.25
0.22
sg
0.029
68.98
619.4
0.013
<0.16
0.027
then: + KBr-CH3OH
InP + p-CF3(CH2)3I
128.8
128.5
128.6
128.5
128.5
128.9
0.29
0.25
0.30
0.25
0.24
0.28
444.5
444.4
444.4
444.4
444.5
444.5
2.64
2.65
2.73
2.89
2.86
2.68
665.7
665.5
665.8
665.8
665.7
665.8
0.018
688.2
688.5
688.4
688.3
688.5
688.2
0.19
0.1
0.25
0.25
0.23
0.39
<0.14
0.56
InP + (C2H5)2O‚BF3
<0.05
<0.05
<0.05
<0.05
<0.05
InP + CF3(CH2)2Si(CH3)2Cl/(C2H5)3N
InP + CF3C6H4COO(CH2)11CH3
InP + p-CF3C6H4COO(CH2)12Br
InP + p-CF3C6H4CH2Br/(C2H5)3N
292.7
292.8
293.1
292.7
0.015
0.028
0.026
0.033
Si 2p
Br 3d
Br 3d
Br 3d
102.0
sg
69.0
69.0
0.46
0.44
0.45
0.34
<0.03
<0.03
<0.16
<0.10
(111)A InP
anaerobically etched InP
InP + p-CF3C6H4CH2Br/(C2H5)3N
128.8
0.22
444.6
2.75
665.9
665.9
0.23
0.16
<0.02
<0.02
688.42
<0.01
(110) InP
anaerobically etched InP
InP + p-CF3C6H4CH2Br/(C2H5)3N
InP + CF3(CH2)3I/(C2H5)3N
128.6
128.6
128.7
0.22
0.24
0.21
444.5
444.6
444.6
2.63
2.56
2.48
665.8
665.9
665.8
<0.05
<0.05
<0.05
292.8
292.8
0.018
0.013
688.3
688.3
0.20
0.12
Br 3d
I 3d
69.2
619.5
<0.04
0.10
0.36
0.43
<0.24
<0.19
a Peak positions were obtained with a peak-fitting procedure and are referenced against C 1s ) 285.0 binding eV. Values are from repeated experiments and are reported as averaged binding energies (BE).
Standard deviations in all cases were <0.1 eV. b Intensities were referenced to the In 3p3/2 signal after normalization with respect to scan duration and background subtraction. c Atomic ratios were derived
from the measured XPS intensities (see e.g., Seah27). Sensitivity factors were taken from Wagner et al.48 and are listed in Table 2. d P 2p binding energy measured at maximum peak height. e Intensity of
convoluted peaks with maxima at 533.0 ( 0.3 eV. f POx represents indium phosphate with a BE of 133.3 eV. g No signal detectable above background.

10844 J. Phys. Chem. B, Vol. 103, No. 49, 1999
Sturzenegger et al.
Figure 2. The XPS spectra of a (111)B InP surface (a) before and (b)
after exposure to CF3(CH2)3I.
Figure 4. The XP spectra of the P 2p region for various InP samples:
(a) solid line, etched (111)B InP; dashed line, etched (111)A InP; (b)
solid line, etched (110) InP; dashed line, (110) InP cleaved in UHV.
The dotted lines are difference spectra for the signals in each panel, in
each case showing the presence of additional, high-binding-energy
signals for the etched (111)B and etched (110) surfaces. The peaks
were normalized with respect to their area after a linear background
subtraction.
Br ester, along with the reactivity of p-CF3C6H4COO(CH2)11-
CH3, indicates that the functionalization chemistry occurs at the
ester group as opposed to at the primary alkyl carbon center.
Cleavage of esters is a well-known quantitative analysis for
-OH functionalities,42 and this type of surface reaction is fully
consistent with the presence of hydroxyl groups as the domi-
nant chemically reactive species on the etched (111)B InP
surface.
Figure 3. The XP spectra of a (111)B InP surface after exposure to
CF3(CH2)2Si(CH3)2Cl.
E. High-Resolution XPS Data for Anaerobically Etched
(111)B-Oriented, (111)A-Oriented, (110)-Oriented, and UHV-
Cleaved (110)-Oriented InP Surfaces. To obtain spectral-based
evidence for the presence of -OH functionalities on the etched
(111)B InP surface, high-resolution XP spectra for etched (111)B
and (110) InP surfaces were carefully compared with those of
etched (111)A and UHV-cleaved (110)-oriented InP surfaces.
The latter two surfaces should have reduced levels of oxidized
surface P atoms and should serve as references for analysis of
the line shapes of the XPS data arising from the etched and
functionalized (111)B-oriented InP surfaces.
Figure 4a displays the high-resolution XPS data in the P 2p
region of an anaerobically etched (111)A-oriented InP surface.
This figure also displays the difference spectrum between the
XPS spectrum of this surface and that of the anaerobically etched
(111)B-oriented InP surface. The signal for the (111)A surface
could be well fit by a single Gaussian/Lorentzian doublet, in
accord with expectations for the instrumental line shape and
detector response function. In contrast, the XP spectrum of the
etched (111)B surface displayed a shoulder to the high binding
energy side of the main P 2p signal. The difference spectrum
indicated that this peak had a binding energy of 130.0 eV, which
is consistent with expectations for the peak position of oxidized
P atoms. A P 2p XPS peak with a similar binding energy was
also observed in the difference spectrum of anaerobically etched
(110)-oriented InP relative to the spectrum of UHV-cleaved
(110)-oriented InP surfaces (Figure 4b). The estimated coverage
of the surface species in the high-binding-energy peak was 8.8
× 1014 cm-2. The contribution of the high-binding-energy peak
to the entire P 2p peak was determined to be (13 ( 3)%, and
the estimated coverage of the surface species was 3.4 × 1014
cm-2. Because the density of P surface atoms on an ideal (111)B
surface is 6.7 × 1014 atoms cm-2, this observation indicates
that ∼50% of the surface P atoms are oxidized on the etched
(111)B-oriented InP surface, even when the etching process is
performed under nominally anaerobic conditions in a N2(g)-
purged glovebox.
from cross-linking the silane reagent and from forming an
insoluble polymer that might be confused with the formation
of persistently bound chemisorbed silyl groups. The (C2H5)3N
was required to facilitate the surface reaction, as is the case for
reaction of Si-OH groups with chlorosilanes.40 As depicted in
Figure 3, exposure of the (111)B-oriented InP surface to
CF3(CH2)2Si(CH3)2Cl produced persistently bound F, CF, and
Si signals in the ratios expected for a -Si(CH3)2(CH2)2CF3
fragment on the InP surface. Negligible signals for atoms of
the anions were detectable on the functionalized surface. The
F, CF, and Si signals were robust and were not affected by
various solvent rinses or by boiling in CH3CN or CH3OH. Si-P
bonds are very weak and unstable in alcoholic solutions, and
would not be expected to withstand the reaction conditions.41
In contrast, silyl halides react readily with -OH groups, forming
strongly bound -O-Si-R linkages that could account for the
observed functionalization.
D. Reactivity of Etched (111)B-Oriented InP Surface with
Esters. To support the hypothesis that -OH groups dominate
the reactivity of (111)B-oriented InP surfaces, reactions with
esters were investigated. Reaction with p-CF3C6H4COO(CH2)11-
CH3 yielded persistently bound -CF3 groups on the surface,
as shown by the cross-section-corrected CF/F ratio of 0.44
measured in the XP spectrum of modified surfaces (Table 3).
In addition, the F/In XP signal ratio was extremely close to
that found for the reaction of (111)B InP with p-CF3C6H4CH2-
Br. This similarity implies that the alkyl chain had been removed
by the reaction, otherwise the differences in the thickness of
the overlayers would have produced different F/In signal ratios
for the same ratio of atoms in the overlayer. Strong evidence
for cleavage of the ester was obtained from reaction of (111)B
InP surfaces with p-CF3C6H4COO(CH2)12Br. Reactions with this
ester yielded essentially identical F/In ratios in the XPS spectra
as were obtained after the reactions with p-CF3C6H4CH2Br or
with p-CF3C6H4COO(CH2)11CH3. Furthermore, all such reac-
tions produced negligible Br signals in the XPS scans. The lack
of a Br signal after reaction with the p-CF3C6H4COO(CH2)12-

Reactions of Single-Crystal (111)B-Oriented InP
J. Phys. Chem. B, Vol. 103, No. 49, 1999 10845
composition and reactivity of (111)-oriented InP surfaces,
(111)B-oriented InP surfaces that were etched under anaerobic
conditions were exposed to air for 10 min and then investigated
by XPS. The XP spectra of these surfaces displayed a distinct
signal in the P 2p region at a binding energy of 133.3 eV (Figure
6a). The binding energy of this signal is in accord with literature
data39 and with our own measurements for the peak position of
oxidized P in InPO4. A numerical analysis of the intensity ratio
of oxidized to unoxidized P signals on this surface indicates
that ∼50% of the surface P atoms were oxidized to InPO4 during
air exposure (Table 6). The intentionally oxidized samples were
then exposed to p-CF3C6H4CH2Br in CH3CN. The resulting XP
spectra (Figure 6b) revealed the presence of -CF3 groups at a
coverage that was very close to the one obtained from reaction
of p-CF3C6H4CH2Br with anaerobically etched (111)B-oriented
InP surfaces.
Figure 5. Intensity ratios F 1s/In 3p3/2 as a function of time for (111)
InP exposed to p-CF3C6H4CH2Br (O), p-CF3C6H4CH2Br/(C2H5)3N (b),
CF3(CH2)3I ()), CF3(CH2)3I/(C2H5)3N ((), and (111)A InP exposed to
p-CF3C6H4CH2Br/(C2H5)3N (B). The intensity ratios are averaged values
from at least 5 samples with the error bars representing 1σ values.
F. Coverage of p-CF3C6H4CH2 Groups on Anaerobically
Etched (111) InP Surfaces. The observation that (C2H5)3N
promotes the reaction of (111)B-oriented InP surfaces with
CF3(CH2)2Si(CH3)2Cl prompted us to investigate whether an
amine also affected the reactivity of (111)B-oriented InP surfaces
toward benzyl and alkyl halides. The XP spectra indicated that
the presence of (C2H5)3N indeed led to substantially higher
amounts of F on derivatized surfaces; for example, for surfaces
exposed to p-CF3C6H4CH2Br and reaction times of 60 min, the
calculated coverage increased by 30% (Table 3) if (C2H5)3N
was added to the solution.
Curiously, the spectra of the P 2p region of such modified
surfaces displayed essentially complete removal of the phosphate
signals (Figure 6b, right panel). Extended exposure to air before
surface modification increased the amount of phosphate initially
present on the InP surface, although the overlayer growth
eventually would presumably terminate with additional exposure
time due to the passivating nature of the oxide. Such heavily
oxidized surfaces still, however, were observed to react with
p-CF3C6H4CH2Br. In this case, though, the derivatization
reaction yielded lower coverages and phosphate removal was
incomplete (Table 6).
To quantify the differences in reactivity between the anaero-
bically etched (111)B and (111)A faces of InP, the reaction of
p-CF3C6H4CH2Br with both crystal faces was monitored as a
function of time. The intensity ratios between the measured F
1s and In 3p3/2 XPS signals are displayed in Figure 5. The
reaction of (111)B-oriented surfaces with p-CF3C6H4CH2Br
showed saturation behavior, as expected for a surface reaction.
The limiting coverage of p-CF3C6H4CH2 groups was calculated
to be 4.3 × 1014 cm-2, as compared with a calculated P atom
density of 6.7 × 1014 cm-2 on a perfect (111)B-oriented surface.
To evaluate the variation of the calculated coverage as a function
of the sensitivity factors, escape depths, and the chosen core
level in the overlayer and substrate, the limiting coverage was
calculated with two sets of SF and λ values (Table 4) and with
six pairs of core levels. Table 5 summarizes the results of this
analysis. The variation is dominated by the chosen core level
pairs. With pairs that included the CF 1s peak instead of the
corresponding F 1s peak, the calculated coverage was always
higher (Table 5). The averaged limiting coverage based on pairs
with CF 1s signals was calculated to be 0.80, whereas core level
pairs including the F 1s signal yielded a limiting coverage of
0.58. The variation within the two groups introduced by different
sensitivity factors and escape depths was ( 0.1.
To investigate the chemistry involved in the dissolution of
the phosphate overlayer, intentionally oxidized InP surfaces were
exposed to either (C2H5)3N‚HBr in CH3CN, (C2H5)3N in CH3-
CN, or pure CH3CN. Surface analysis revealed that only
(C2H5)3N‚HBr removed the phosphate, whereas (C2H5)3N and
CH3CN exposure did not produce any detectable changes in
the surface composition (Table 6). This observation provides
an explanation for the reaction of p-CF3C6H4CH2Br with oxi-
dized (111)B-oriented InP because the HBr that is liberated
during the reaction can etch the InP oxide and thereby pro-
duce the same surface that is obtained by anaerobic etching
alone.
H. Reactivity of Etched (110)-Oriented InP Surfaces. To
test the derivatization strategy for surfaces other than (111)B
InP, (110) oriented InP was exposed to p-CF3C6H4CH2Br and
CF3(CH2)3I. Both reactions yielded surfaces with persistently
bound -CF3 groups as shown by the cross-section corrected
CF/F ratios (Table 3). For (110) surfaces that were exposed to
p-CF3C6H4CH2Br, the intensity ratio F 1s/In 3p3/2 was 0.20 (
-
0.01 and the coverage with p-CF3C6H4CH2 was calculated to
be 2.3 × 1014 cm-2. Surfaces that were reacted with CF3(CH2)3I
yielded a F 1s/In 3p3/2 intensity ratio of 0.12 ( 0.02 and the
In contrast to the P-rich (111)B-oriented InP surfaces, In-
rich (111)A-oriented InP surfaces that had been exposed to
p-CF3C6H4CH2Br yielded negligible F signals in the XP
spectrum. The cross-section-corrected intensity ratio of the F
1s/In 3p3/2 signals on such surfaces was independent of the
reaction time and was always <0.01. The coverage of p-CF3C6H4-
density of CF3(CH2)3 groups was calculated to be 1.6 × 1014
-
cm-2
.
I. Recombination Velocity of Etched versus Functionalized
(111)B InP Surfaces. Time-resolved photoluminescence meth-
ods were used to compare the electrical properties of etched
InP surfaces with those of functionalized InP surfaces. Native
InP has a relatively low surface recombination velocity, S ≈
13
-
CH2 groups on this surface was estimated to be < 3.4 × 10
cm-2
.
102 cm s-1 7-9
so the decay kinetics of photogenerated electron-
,
Analysis of the coverage of (111)B-oriented InP surfaces that
had been exposed to CF3(CH2)3I also revealed saturation
behavior. For reaction times >60 min, the F 1s/In 3p3/2 intensity
hole pairs are usually dominated by the nonradiative processes
in the bulk of such samples. However, for short optical
penetration depths of the incident light, increases in S will
produce diagnostic decreases in the observed luminescence
lifetime.35 The luminescence decay kinetics therefore allow a
convenient comparison of the electrical quality of etched InP
-
ratio was 0.22 ( 0.01. The limiting coverage of CF3(CH2)3
groups was calculated to be 2.9 × 1014 cm-2
.
G. Reactivity of Aerobically Etched (111)B InP Surfaces.
To investigate the consequences of exposure to air on the

10846 J. Phys. Chem. B, Vol. 103, No. 49, 1999
Sturzenegger et al.
In 3p3/2
a
TABLE 4: Experimentally Determined Sensitivity Factors (SFexp
)
P 2p
SFexp/SFlit
C 1s
SFexp/SFlit
In 3d5/2
SFexp/SFlit
substance
Teflon (exp)
SFexp
SFexp
SFexp
SFexp
SFexp/SFlit
0.255
0.25
1.02
Teflon (lit)
1.00b
(C4H9)4PF6 (exp)
AgPF6 (lit)
p-CF3C6H4CH2P(C6H5)3Br
InBr3[P(C6H4F)3]2
InF3 (exp)
0.356
0.359
0.335
0.341
0.91
0.92b
0.86
0.87
0.273
1.09
4.62
5.03
4.68
1.19
1.29
1.20b
2.10
2.20
1.13
1.19
InF3 (lit)
(110) InP, powder
(111)B InP, single crystal
(110) InP, single crystal
a SFexp were derived from measured areas, after a linear background subtraction and correction for stoichiometric coefficients. As in Wagner’s
set of sensitivity factors (SFlit), the area of the F 1s signal serves as a reference. b These ratios were calculated with the originally measured areas
and therefore they should equal 1.
TABLE 5: Variation of the Calculated Surface Coverage, Γ, for (111)B InP Exposed to p-CF3C6H4CH2Br Derived from
Measured XPS Intensities with Two Sets of Sensitivity Factors and Escape Depths, Respectively
surface coverage,a calculated
d
e
ratio
int. ratio
w/λb & SFlit
w/λ & SFexp
w/corr IMFPc & SFlit
w/corr IMFP & SFexp
I(F 1s)/I(In 3p3/2
I(F 1s)/I(In 3d5/2
I(F 1s)/I(P 2p)
I(CF 1s)/I(In 3p3/2
I(CF 1s)/I(In 3d5/2
I(CF 1s)/I(P 2p)
)
)
0.38
0.143
1.43
0.0335
0.0126
0.126
0.64
0.55
0.60
0.89
0.77
0.84
0.61
0.65
0.60
0.78
0.84
0.77
0.57
0.50
0.54
0.96
0.71
0.78
0.53
0.59
0.54
0.70
0.78
0.72
)
)
a Number of p-CF3C6H4CH2- groups referenced to the number of surface P atoms. b Escape depths in n-alkane thiols on group IB metals.30
c Inelastic mean free path for polystyrene,49 corrected for inelastic scattering.27 d Wagner’s sensitivity factors.48 e Sensitivity factors derived from
closely related organic compounds and single-crystal (110) InP.
where Nt is the density of surface traps, σ is the cross section
for carrier capture (typically 10-15 cm2 for neutral traps), and
Vth is the thermal velocity of carriers in the solid (107 cm s-1),
a value of S ) 102 cm s-1 corresponds to Nt ≈ 1010 cm-2
.
Because the density of atoms on an InP surface is on the order
of 1015 cm-2, a value of Nt ) 1010 cm-2 implies that under
these conditions only 1 in every 105 surface atoms is an effective
electrical trapping site on the chemically modified InP surface.
IV. Discussion
The etched InP surfaces investigated in this work produced
survey scan XP spectra that were remarkably close to those
expected for a 1:1 In/P stoichiometry, with negligible XPS signal
intensity detected for other elements, except for signals ascrib-
able to adventitious carbonaceous material. InP typically has a
significant overlayer of oxidation products (often oxide and
phosphates were observed) after exposure to oxidizing etches
and/or exposure to air, but such oxidation was not observed
using the etching and handling methods described herein. The
anaerobic etching process thus offers a reproducible, near-
stoichiometric starting surface for chemical modifications of this
semiconductor.
The reactions investigated herein confirm the pronounced
differences observed previously by Spool et al.13 with regard
to the relative reactivity of the (111)B (P-rich) and (111)A (In-
rich) InP surfaces toward alkylating reagents. This differential
reactivity has been used to suggest that the reactivity of the
etched (111)B face is due to lone pairs on the P atoms that
would be exposed on an atomically perfect (111)B surface.13
Many of the reactions observed in this work, however, are not
characteristic of P lone pairs, but are instead indicative of the
presence of hydroxyl groups on the (111)B-oriented InP surface
(Scheme 2). The lack of reactivity of the (111)B-oriented InP
surface with (C2H5)2O‚BF3, the facile cleavage of esters by this
Figure 6. The XPS spectra of anaerobically etched (111)B-oriented
InP surfaces (a) after exposure to air and (b) after subsequent exposure
to p-CF3C6H4CH2Br in CH3CN.
surfaces with those that have been deliberately functionalized
with organic groups.
Figure 7 depicts the luminescence decay kinetics for etched
and modified InP surfaces. The observed luminescence decays
for InP surfaces that had been modified with p-CF3C6H4CH2Br
were very similar to those of etched InP samples. Both surfaces
had relatively low surface recombination velocities, with S ≈
102 cm s-1 for each system. These data were collected under
high-level injection conditions to minimize any influence of
changes in equilibrium band-bending arising from the surface
modification process on the measurement of the surface
recombination properties. Using the relationship S ) Nt σ Vth,1

Reactions of Single-Crystal (111)B-Oriented InP
J. Phys. Chem. B, Vol. 103, No. 49, 1999 10847
TABLE 6: XPS Data of Air-Exposed (111)B-Oriented InP Surfaces
I(InPO4)/I(InP)a
ΓInPO4 (cm-2
)
Γ
InPO /ΓP
I(F 1s)/I(In 3p3/2)
b
surface
4
anaerobically etched (111)B InP (InP)
InP, 10 min exposed to air
then: + p-CF3C6H4CH2Br/(C2H5)3N
InP, 60 min exposed to air
then: + p-CF3C6H4CH2Br/(C2H5)3N
InP, 60 min exposed to air
then: + p-CF3C6H4CH2Br
InP, 60 min exposed to air
then:
<0.02
0.15
<0.02
0.38
0.07
0.34
0.08
0.36
0.5 × 1014
3.6 × 1014
0.5 × 1014
7.9 × 1014
1.7 × 1014
7.2 × 1014
2.0 × 1014
7.6 × 1014
2.7 × 1014
7.1 × 1014
7.2 × 1014
0.19
0.53
<0.08
1.18
0.26
1.08
0.30
1.13
s
s
0.34
s
0.31
s
0.25
s
a) + (C2H5)3N‚HBr-CH3CN
0.11
0.33
0.34
0.40
1.06
1.08
s
s
s
b) + (C2H5)3N-CH3CN
c) + CH3CN
a Measured in the P 2p region at 133.3 eV (InPO4) and 129.6 eV (InP), respectively. b InPO4 units referenced to the number of surface P atoms.
lone pair reactivity route because the initially neutral P species
should produce a cationic P species after alkylation. Instead, in
the hydroxyl group pathway, the surface P atoms are initially
oxidized and no further oxidation occurs during the reaction
steps of concern. The differential reactivity of the P-rich (111)B
InP surface relative to the In-rich (111)A orientation can thus
be ascribed to the differential chemical reactivity of the acidic
-OH functionalities of P oxides relative to the basic properties
expected for -OH functionalities of In oxides.
Fadley’s overlayer-substrate model allows calculation of the
coverage of derivatized surfaces based on measured XPS
intensity data. However, the calculations depend on three
Figure 7. The PL decay curves before (O) and after (b) exposure to
parameters: SF, λ, and do, all of which must be determined
p-CF3C6H4CH2Br and a simulated PL decay curve ()) obtained using
accurately. These quantities are all difficult to obtain and can
S ) 102 cm s-1
.
contain significant errors.27 The XPS data for homogeneous
compounds (Table 4) confirmed that XPS intensity ratios by
themselves can be measured with an error of <10%. The data
also show that SF values, which were computed from a
regression procedure so that they could be applied to a wide
range of compounds, in some cases describe the composition
of particular compounds inaccurately. The data further reveal a
variation in the ratio of different core levels for the same
element, depending on the sample preparation. Powdered (110)
InP samples yielded a In 3d5/2/In 3p3/2 ratio that agrees with
ratios of other powdered samples. However, for single crystal
(111) and (110) InP, this ratio is significantly higher (Table 4).
We have thus derived sensitivity factors (SFexp) from closely
related organic compounds and single-crystal (110) InP (Table
2). An analysis of the influence of SF, λ, and the chosen core
level pair on the calculated surface coverage showed that the
use of different SF and λ changes the calculated coverage of
these systems by <(15%. However, the choice of the overlayer
core level can have a large influence on the calculated surface
coverage. The CF 1s peaks are very small and are sensitive (e.g.,
to convolution with shake up satellites). In contrast, F 1s peaks
have intensities that are comparable with the substrate peaks
and are less affected by other processes.
The results of the quantification of the coverage for (111)B
and (110) InP, each derivatized with p-CF3C6H4CH2Br and
CF3(CH2)3I, are listed in Table 7. The (111)B InP surface
appears to accommodate a significantly larger number of
molecules than does the (110) face of InP. However if the
coverage is referenced instead to the number of possible reaction
sites, expressed in surface P atoms per unit area, the coverage
on (111)B surfaces is only slightly higher than that observed
on the (110) orientation. The geometry of the (111)B face might
allow a slightly denser packed overlayer, but more precise
measurements would be required to address this question
definitively.
surface, and the reaction of this surface with chlorosilanes are
examples of such observations. The consistency between the
various observations strongly supports the notion that hydroxyl
groups are responsible for the observed surface chemistry. This
conclusion results in an apparently general synthetic strategy
for derivatization of the chemically etched (111)B-oriented InP
surface.
The hydroxyl groups are apparently introduced during the
etching and handling procedures of this surface, even though
these steps were performed under nominally anaerobic condi-
tions during the course of our study. The detailed comparison
of the P 2p XPS line shapes of etched surfaces relative to the
line shapes of surfaces that are prepared in UHV (Figure 4)
reveals spectroscopic evidence for this surface oxidation process.
X-ray standing wave (XSW) and extended X-ray absorption fine
structure (EXAFS) studies have shown that surface relaxation
causes only minor changes in the surface structure of UHV-
cleaved (110) InP surfaces.43,44 The differences observed
between P 2p signals arising from cleaved (110)-oriented InP
surfaces relative to the signals from etched (110)-oriented InP
surfaces can therefore be attributed to P atoms that are
chemically different from the surface or bulk P atoms of an
unoxidized, cleaved InP surface. The surface relaxation of
(111)B InP is less characterized, but only small changes in the
surface structure have been suggested for this surface as a result
of surface relaxation.45 Thus, the differences in the XP spectra
between etched (111)A and etched (111)B orientations also
provide evidence of oxidized P atoms on the etched (111)B InP
surface.
The presence of this oxide overlayer prior to the chemical
modification step also provides a consistent explanation of why
the XP spectra showed no significant change in the P oxidation
state during the surface modification procedure. Changes in the
P 2p region of the XP spectra would have been expected for a

10848 J. Phys. Chem. B, Vol. 103, No. 49, 1999
Sturzenegger et al.
SCHEME 2: Summary of the Reactivity of Etched (111)B InP Surfaces
TABLE 7: Coverage of (111)B and (110) InP Surfaces
a
surface
I(F 1s)/I(In 3p3/2
)
nb (cm-2
)
Γc
(111)B InP + p-CF3C6H4CH2Br/(C2H5)3N
0.38
(0.03)
0.22
(0.01)
0.20
(0.01)
0.12
(0.02)
4.3 × 1014
(0.3 × 1014)
2.9 × 1014
(0.1 × 1014)
2.3 × 1014
(0.1 × 1014)
1.6 × 1014
(0.3 × 1014)
0.64
(0.05)
0.43
(0.02)
0.55
(0.03)
0.38
(0.06)
(111)B InP + CF3(CH2)3I/(C2H5)3N
(110) InP + p-CF3C6H4CH2Br/(C2H5)3N
(110) InP + CF3(CH2)3I/(C2H5)3N
a I(F 1s)/I(In 3p3/2) are averaged values from at least 5 samples, numbers in parentheses represent 1 σ values introduced by variations in the
intensity ratios. b Number of p-CF3C6H4CH2- and CF3(CH2)3- groups, respectively, bound to the surface. c Number of p-CF3C6H4CH2- and
CF3(CH2)3- groups, respectively, referenced to the number of surface P atoms.
The results show that both faces accommodate a higher
number of benzyl units than hydrocarbon chains. This result is
consistent with expected packing differences between the
platelike benzyl units relative to the rodlike alkyl chains. The
benzyl unit has a van der Waals thickness of 3.7 Å as compared
with the van der Waals diameter of 4.0 Å for alkyl chains.46 In
addition, the covalent, directed bonding of hydroxyl groups on
the oxidized InP surface to the alkyl overlayer will restrict the
orientation angle of these overlayers much more severely than
thiols on Au and therefore imposes different constraints on the
alkyl chain density in the overlayer for InP surfaces than for
metallic systems.
transfer across the derivatized surface. The functionalization
routes described herein satisfy this criterion, and thus may allow
for chemical modification of surfaces with dielectric layers for
controlling electron-transfer reactions, forming metal-insula-
tor-semiconductor structures, and inhibiting surface-corrosion
processes without destroying the desirable electrical properties
of the InP surface. Because the functionalization route is
versatile and is chemically straightforward, the approach should
afford a promising strategy for preparing improved optical and
electronic devices from InP. Efforts to exploit these desirable
properties in actual devices are in progress and will be reported
in a separate manuscript.
An advantage of functionalizing the surface through hydroxyl
groups is that it preserves the inherently low carrier recombina-
tion velocity of the (111)B-oriented InP surface. Heller47 has
previously outlined a strategy for surface modification in which
a reduction in the surface recombination velocity of a semi-
conductor can be achieved by inducing adsorbates to enter into
strong chemical bonding with states that otherwise have energy
levels in the middle of the band gap of the semiconductor. In
the case of the (111)B InP surface, this approach is naturally
implemented through surface oxidation because this process
forms strong chemical bonds to the lone pairs on the exposed
phosphorus surface atoms. In the absence of formation of these
strong bonds, these lone pairs are expected energetically to serve
as recombination sites. Routes to surface functionalization that
do not perturb this monolayer of surface oxidation therefore
should maintain the desirable electrical properties of the InP
system while still offering opportunities for efficient charge
Acknowledgment. We acknowledge the Department of
Energy, Office of Basic Energy Sciences, for support of this
work. M.S. acknowledges the Swiss National Science Founda-
tion for a postdoctoral fellowship, and we thank A. Rice for
technical assistance with the XPS measurements.
References and Notes
(1) Sze, S. M. Physics of Semiconductor DeVices, 2nd ed.; Wiley: New
York, 1981.
(2) Heller, A. ACS Symp. Ser. 1981, 146, 57.
(3) Sze, S. M. Semiconductor Sensors; Wiley: New York, 1994.
(4) Indium Phosphide and Related Materials: Processing, Technology
and DeVices; Katz, A., Ed.; Artech House: Boston, 1992.
(5) Adachi, S. Physical Properties of III-V Semiconductor Compounds
InP, InAs, GaAs, GaP, InGaAs, and InGaAsP; John Wiley and Sons: New
York, 1992.
(6) Newman, N.; Kendelewicz, T.; Bowman, L.; Spicer, W. E. Appl.
Phys. Lett. 1985, 46, 1176.

Reactions of Single-Crystal (111)B-Oriented InP
J. Phys. Chem. B, Vol. 103, No. 49, 1999 10849
(7) Bothra, S.; Tyagi, S.; Ghandhi, S. K.; Borrego, J. M. Solid-State
Electron. 1991, 34, 47-50.
(30) Laibinis, P. E.; Bain, C. D.; Whitesides, G. M. J. Phys. Chem. 1991,
95, 7017-7021.
(8) Brillson, L. J.; Shapira, Y.; Heller, A. Appl. Phys. Lett. 1983, 43,
175.
(31) Jablonski, A. Surf. Interface Anal. 1989, 14, 659-685.
(32) Tanuma, S.; Powell, C. J.; Penn, D. R. Surf. Interface Anal. 1991,
17, 927-939.
(9) Casey, H. C., Jr.; Buehler, E. Appl. Phys. Lett. 1977, 30, 247-
249.
(33) Ryba, G. N.; Kenyon, C. N.; Lewis, N. S. J. Phys. Chem. 1993,
97, 13814-13819.
(10) Koval, C. A.; Austermann, R. L.; Turner, J. A.; Parkinson, B. A.
J. Electrochem. Soc. 1985, 132, 613-623.
(34) Kenyon, C. N.; Tan, M. X.; Kruger, O.; Lewis, N. S. J. Phys. Chem.
B 1997, 101, 2850-2860.
(11) Oskam, G.; Bart, L.; Vanmaekelbergh, D.; Kelly, J. J. J. Appl. Phys.
1993, 74, 3238-3245.
(35) Anz, S. J.; Kruger, O.; Lewis, N. S.; Gajewski, H. J. Phys. Chem.
B 1998, 102, 5625-5640.
(12) Gu, Y.; Lin, Z.; Butera, R. A.; Smentkowski, V. S.; Waldeck, D.
H. Langmuir 1995, 11, 1849-1851.
(36) Gajewski, H. GAMM (Ges. Angew. Math. Mechanik) Mitteilungen
1993, 16, 35.
(13) Spool, A. M.; Daube, K. A.; Mallouk, T. E.; Belmont, J. A.;
Wrighton, M. S. J. Am. Chem. Soc. 1986, 108, 3155-3157.
(14) Sturzenegger, M.; Lewis, N. S. J. Am. Chem. Soc. 1996, 118, 3045-
3046.
(37) Kruger, O.; Jung, C.; Gajewski, H. J. Phys. Chem. 1994, 98,
12653-12662.
(38) Kauffman, J. F.; Balko, B. A.; Richmond, G. L. J. Phys. Chem.
1992, 96, 6371-6374.
(15) Thiel, F. A.; Barns, R. L. J. Electrochem. Soc.: Solid State Science
and Technology 1979, 126, 1272-1274.
(16) Hauser, C. F.; Brooks, T. W.; Miles, M. I.; Raymond, M. A.; Butler,
G. B. J. Org. Chem. 1963, 28, 372-379.
(17) Putvinski, T. M.; Schilling, M. L.; Katz, H. E.; Chidsey, C. E. D.;
Mujsce, A. M.; Emerson, A. B. Langmuir 1990, 6, 1567-1571.
(18) Roundhill, D. M. J. Inorg. Nucl. Chem. 1971, 33, 3367-3373.
(19) Bryson, C. E., III Surf. Sci. 1987, 189/190, 50-58.
(20) Wagner, C. D.; Gale, L. H.; Raymond, R. H. Anal. Chem. 1979,
51, 466.
(21) Fadley, C. F. Progr. Solid State Chem. 1976, 11, 265-343.
(22) Gui, J. Y.; Stern, D. A.; Frank, D. G.; Lu, F.; Zapien, D. C.;
Hubbard, A. T. Langmuir 1991, 7, 955-963.
(23) Nuzzo, R. G.; Korenic, E. M.; Dubois, L. H. J. Chem. Phys. 1990,
93, 767.
(24) Nakagawa, O. S.; Ashok, S.; Sheen, C. W.; Mårtensson, J.; Allara,
D. L. Jpn. J. Appl. Phys. 1991, 30, 3759-3762.
(25) Allen, F. H.; Kennard, O.; Watson, D. G.; Brammer, L.; Orpen, A.
G.; Taylor, R. J. Chem. Soc., Perkin Trans. 2 1987, S1-S19.
(26) Ebel, M. F. J. Electron Spectrosc. Relat. Phenom. 1978, 14, 287-
322.
(39) Hollinger, G.; Bergignat, E.; Joseph, J.; Robach, Y. J. Vac. Sci.
Technol., A 1985, 3, 2082-2088.
(40) Tripp, C. P.; Veregin, R. P. N.; Hair, M. L. Langmuir 1993, 9,
3518-3522.
(41) Armitage, D. A. Organosilicon Derivatives of Phosphorus, Arsenic,
Antimony, and Bismuth. In The Silicon/Herteroatom Bond; Patai, S.,
Rappoport, Z., Eds.; John Wiley & Sons: Chichester, 1991.
(42) Cheronis, N. D.; Ma, T. S. Organic Functional Group Analysis;
Interscience: New York, 1964.
(43) Woicik, J. C.; Kendelewicz, T.; Miyano, K. E.; Richter, M.;
Boulding, C. E.; Pianetta, P.; Spicer, W. E. Phys. ReV. B 1992, 46, 9869.
(44) Woicik, J. C.; Kendelewicz, T.; Miyano, K. E.; Richter, M.; Karling,
B. A.; Bouldin, C. E.; Pianetta, P.; Spicer, W. E. J. Vac. Sci. Technol., A
1992, 10, 2041-2045.
(45) Hou, X.; Dong, G.; Ding, X.; Wang, X. Chin. Phys. Lett. 1986, 3,
545-548.
(46) Dean, J. A. Lange’s Handbook of Chemistry; McGraw-Hill: New
York, 1992.
(47) Heller, A. Chemical Control of Surface and Grain Boundary
Recombination in Semiconductors. In Photoeffects at Semiconductor-
Electrolyte Interfaces; Nozik, A. J., Ed.; American Chemical Society:
Washingtion, DC, 1981; Vol. 146; pp 57-77.
(48) Wagner, C. D.; Davis, L. E.; Zeller, M. V.; Taylor, J. A.; Raymond,
R. H.; Gale, L. H. Surf. Interface Anal. 1981, 3, 211-225.
(49) Tanuma, S.; Powell, C. J.; Penn, D. R. Surf. Interface Anal. 1993,
21, 165-176.
(27) Seah, M. P. Quantification of AES and XPS. In Practical Surface
Analysis; 2nd ed.; Briggs, D., Seah, M. P., Eds.; John Wiley & Sons:
Chichester, 1990; Vol. 1; pp 201-255.
(28) Hochella, M. F., Jr.; Carim, A. H. Surf. Sci. 1988, 197, L260-
L268.
(29) Pies, W.; Weiss, A. Crystal Structure Data of Inorganic Com-
pounds; Springer-Verlag: Berlin, 1979; Vol. 7.