Vij D.R. Handbook of Applied Solid State Spectroscopy

Подождите немного. Документ загружается.

of the molecular system of interest. This latter strength is sometimes a

weakness, since molecular systems having large band gaps (> 5 eV) have

OMTS bands that can only be observed at very large applied voltages. At

large bias voltage the elastic tunneling current and the potential for current-

induced instabilities in the tip and substrate can effectively mask the OMTS.

Fortunately, a very large percentage of all molecules have either occupied or

unoccupied states within a few volts of the Fermi energy of a typical metal

substrate, and many have both occupied and unoccupied states within this

range. The large background elastic currents at high bias can be reduced in



/IV (vide

infra), thereby canceling out much of the exponential dependence of T(E, eV,

r) on V. We will place our primary focus on STM-OMTS in the examples that

follow.

One view of the OMT process is that the molecule M is reduced, M

–

, or

oxidized, M

, during the tunneling process [92, 93, 96, 101, 109–112, 115]. In

this picture a fully relaxed ion is formed in the barrier. The absorption of a

phonon (the creation of a vibrational excitation) then induces the ion to decay

completes tunneling through the barrier. For simplicity, the reduction case

will be discussed in detail; but the oxidation arguments are similar. A

transition of the type M + e

–

 M

–

is conventionally described as formation

of an electron affinity level. The most commonly used measure of condensed

phase electron affinity is the half-wave reduction potential measured in non-

aqueous solvents, E

1/2

. Often these values are tabulated relative to the

saturated calomel electrode (sce). In order to correlate OMTS data with

electrochemical potentials, we need the data referenced to an electron in the

vacuum state. That is, we need the potential for the half-reaction:

M(solution) + e

–

(vac)  M

–

(solution)

These values can be closely approximated from those referenced to sce by

adding 4.70 V to E

1/2

(sce) [111–114]. That is, E

1/2

(vacuum) = E

1/2

(sce) +

4.71 V. This connection between solution phase electrochemical potentials

and vacuum level-based spectroscopic values such as OMTS and UPS is

extremely useful, but the derivation is rather complex. For example, the

difference in electron affinity in the gas phase and in solution is primarily due

to solvation stabilizing the reduced form. The reader wishing to better

understand its origins is encouraged to consult [113] and [114]. The various

energy conventions are depicted in Figure 7.11, where the connection

between OMTS bands and electron affinities is made within the context of the

7. Scanning Tunneling Spectroscopy (STS)

significance by reporting the normalized OMTS,

dI//dV

and Electrochemistry

7.5.1 STM-Based Orbital-Mediated Tunneling Spectra

back to the neutral molecule with emission of the electron––which then

328

model. This diagram is based on one presented by Loutfy et al. [113] and

expanded to OMTS by Mazur and Hipps [7, 92, 93, 96, 101, 115–118]. By

using the measured value of E

(from UPS), the OMTS bands can be located

both relative to the vacuum level and also to electrochemical potentials. The

details of the procedure for locating the vacuum level will be presented in a

later section.

Figure 7.11 Electrochemical energy level model for orbital-mediated tunneling. A

and A

are

the gas and crystalline phase electron affinities, E

1/2

(sce) is the electrochemical potential

referenced to the saturated calomel electrode, and provides the solution phase electron affinity.

is the Fermi level of the substrate (Au here). The corresponding positions in the OMT

spectrum are shown by '

and '

and correspond to the electron affinity and ionization

potential of the adsorbate film modified by interaction with the supporting metal, A

. The

spectrum is that of nickel (II) tetraphenylporphyrin on Au(111). Reprinted with permission

from [85]. Copyright (2005) The Journal of Chemical Education.

Because the dielectric constant of most organic solids is less than that of

common solvents, redox potentials in the solid state are expected to differ

from those in solution as shown (qualitatively) in Figure 7.11 [113]. There

will also be shifts associated with intermolecular interaction that are very

difficult to predict and that vary considerably in different types of molecules

that are not depicted. Moreover, there will be shifts in the potentials of a thin

film relative to that of a solid due to interactions with the metal support and

counter electrode, including image charge effects. These all tend to stabilize

ion formation. Thus, they act to return the ionization and reduction potentials

for the species adsorbed on a metal surface to those for the species in solution,

(as shown in Figure 7.11). There also may be an opposite signed shift due to

the absence of a covering layer of solvent or adsorbed molecules in the case

of a monolayer (or less) in UHV [116]. Another complication is the fact that

electrochemical potentials are equilibrium values and therefore reflect the

energy associated with the formation of an ion in its equilibrium state. OMTS

transitions, as discussed in the next section, may occur so rapidly that the ion

is formed in an excited state––a vertical transition in the Frank-Condon sense.

7.5 Practical Considerations 329

For a wide range of materials and film thickness (sub––monolayer to about

0.5 nm) studied to date, a fortuitous cancellation of polarization terms and

differences between vertical and equilibrium affinities has resulted in many

OMTS bands lieing close to the positions predicted from electrochemistry

(see Figure 7.12). This correlation is especially good for unoccupied orbitals.

Unfortunately, OMTS bands associated with occupied orbitals generally lie

deeper than predicted by solution phase electrochemical oxidation potentials.

The transferability of electrochemical values to thin film band positions for

affinity levels but not ionization levels indicates that the polarization energy

terms differ for these processes. This is a failure in the simple model used to

generate Figure 7.11, where it was assumed that only the sign of the

polarization energy changed [101]. This failure is particularly large for

porphyrins. Given the trends in stabilization of ion energies by the

surrounding molecules and image charges induced in the metal substrate, we

would expect the ionization potential of thin film NiOEP to be about 0.5 to

1.0 eV less than for the gas phase. Instead, the ionization energies measured

from a thin film are nearly identical to those reported from the gas phase [101,

119]. As we shall see in the next section, the STM-OMTS band positions are

consistent with UPS observations, suggesting that the problem is in the model,

not the technique.

While the first electrochemical reduction potential provides an estimate for

(assuming it is a reversible nonreaction process), the second and higher

reduction potentials do not provide the spectrum of single electron affinity

levels. Rather, they provide information about 2-electron, 3-electron, and

higher electron reduction processes and, therefore, depend on electron pairing

energy. Thus, the utility of solution phase reduction potentials for estimating

solid state affinity levels is limited to the lowest affinity level. The same

argument applies to oxidation potentials beyond the first. OMTS, on the other

hand, probes the single electron reduction energies for the spectrum of states

of the negative ion and the single electron ionization energies for the spectrum

of states of the positive ion. Thus, OMTS can be used to determine ionization

spectra and affinity levels beyond the first transitions of each type [7, 115].

Molecular ball and stick models of a metal tetraphenylporphyrin (MTPP)

and of a metal phthalocyanine (MPc) are presented in Figure 7.13. Metal ions

are in the +2 oxidation state while the rings are in the –2 state. Note that there

are 4 nitrogen atoms coordinating each metal. The Pc ring has a total of eight

nitrogen atoms while the TPP ring has only four.

7. Scanning Tunneling Spectroscopy (STS) 330

Figure 7.12 Correlation between electrochemical potentials and OMTS bands for more than

10 compounds including polyacenes, phthalocyanines, and porphyrins. OMTS data were

acquired both from tunnel junctions and STM measurements, and individual points have been

reported in [92–94, 96, 101, 115,117, 118]. The potential on the left is that associated with the

half-reaction M(solution) + e

–

(vac)

 M

–

(solution).

Figure 7.13 Molecular models of MPc and MTPP.

7.5 Practical Considerations 331

Spectroscopy

An alternative mechanism for OMTS is one in which the electron residence

time is long enough to cause electronic excitation of the molecule, but not so

long as to allow vibrational relaxation to occur. Thus, the electron capture or

emission is a vertical process in the Frank-Condon sense. For the reverse bias

region of the OMTS, this is essentially the condition for ultraviolet

photoemission. In general, ultraviolet photoemission spectroscopy (UPS) is

described by the expression

+ hQ! M

+e

–

The ionization energy to produce state j of the positive ion M

is expressed

by: H

= hQ – KE

, where H

is the ionization energy measured relative to the

Fermi energy (E

), KE is the kinetic energy of the ejected photoelectron, and

hQ is the energy of the photon. KE and Q are the directly measured quantities.

is the ground state molecule of interest and M

is the ionized molecule in

its j

th excited electronic and vibrational state reflecting a vertical Frank-

Condon transition.

To acquire a UPS spectrum, one irradiates the sample with UV light,

causing electrons to be ejected from the higher bands. The kinetic energy

of these ejected electrons is then measured. UPS is an extremely

Because of the photoionization cross-section differences at the relatively low

energies typically used for UPS, the technique is much more sensitive to p-

type valence orbitals than d-type. Thus, valence shell UPS is the technique of

choice for studying the highest energy S orbitals of a molecular system.

Figure 7.14 shows schematic band diagrams for a metal, and for a molecular

film coated on a metal, that illustrate the important energetic parameters

derivable from UPS data [17, 120–122]. The energy of the vacuum level over

the molecular coating relative to that of the clean metal is given by the

quantity ' (shown for the case where ' is a negative quantity: This is not '

or '

). The work function of the clean metal is given by )

, and can be

calculated from the relationship hQ = W

+ )

, where W

is the width of the

photoelectron spectrum of the clean metal expressed in eV. In order to

compare ionization energies to other quantities, one needs the ionization

energy of an occupied orbital relative to the vacuum level. This is given by

H

= H

+ )

+ '

where H

is the energy of the jth occupied state relative to the vacuum level

and H

is the energy of the jth occupied state relative to the Fermi level of the

metal contact.

Both the UPS and the STM-OMTS obtained from a nickel(II)

octaethlyporphyrin (Ni OEP) film on Au (111) are depicted in Figure 7.15.

7. Scanning Tunneling Spectroscopy (STS)

surface-sensitive tool, probing only about 1nm into the surface of the sample.

The UPS sample was about 1 monolayer (~10

332

7.5.2 STM-Based OMTS and Ultraviolet Photoemission

molecules) while the STM-OMTS

same scale, one correction and one assumption was required. The assumption

was that the potential at the molecule was essentially the same potential as the

substrate (gold electrode). We will address this assumption in more detail

later.

The correction required first that the vacuum level over the NiOEP film on

gold be correctly located. This was done using equation (7.3), where )

and

' were obtained form clean gold and NiOEP-covered gold spectra. The

OMTS band energies were also referenced to the vacuum level through the

measured work function. The vacuum level-referenced OMTS energy H

OMTS

then calculated from the bias voltage of the transition peak V

peak

using

equation (7.4):

Figure 7.14 Schematic of significant parameters in the UPS of a clean metal (left) and a

molecular coating on a metal (right). The work function of the metal is given by )

and can

be calculated from the relationship hQ = W

+ )

, where W

is the width of the

photoelectron spectrum in eV. Ionization energies of the highest occupied and second

highest occupied molecular orbitals relative to the vacuum level are given by H

HOMO

and

SOMO

. The corresponding quantities measured relative to the Fermi level are given with a

superscript F. ' is a measure of the interface dipole moment and is shown in the case where

it has a negative sign. E

and E

vac

are the energies of the Fermi and vacuum levels of the

clean metal substrate, respectively. E

vac

is the vacuum level energy over the adsorbate-

covered metal surface. Note that these are cartoons. In real cases, one sometimes sees the

Fermi edge of the metal superimposed on that of the adlayer (high KE end of the spectrum).

Reprinted with permission from [85]. Copyright (2005) the Journal of Chemical Education.

was taken from a single molecule. In order to place both spectra on the

7.5 Practical Considerations 333

OMTS

= –e (V

peak

) + )

+ ' .

(7.4)

The values of )

and ' were the same as used for the UPS spectrum. The

derived values of energy relative to the vacuum level are shown in the right

panel of Figure 7.15 as the top scale.

In some cases the UPS spectrum is taken with the adsorbate on a different

crystal face or metal than is used for the OMTS. In these cases caution is

required. Different values of )

ҏ must be used for UPS and OMTS, and the

values of ' may also vary.

Thus, it is possible to make a direct comparison of occupied orbital

energies as derived from UPS and those observed by OMTS. We have made

these measurements for a number of phthalocyanines and porphyrins [85, 96,

101, 118] and find good agreement between the UPS positions of the

occupied S-type orbitals and the band positions seen in OMTS. In cases

where there are occupied d orbitals near the Fermi surface, the OMTS

spectrum is richer than that of the UPS. Because of the difference in cross-

section for electron emission by tunneling and electron ejection by photon

absorption, d-type orbitals are seen with much greater intensity in OMTS than

in UPS. For example, the UPS spectra of nickel (II) tetraphenylporphyrine

(NiTPP) and cobalt (II) tetraphenylporphyrine (CoTPP) near E

are almost the

same, but the STM-OMTS of CoTPP has a well-defined band due to

the partially occupied d

orbital that is absent in the OMTS of NiTPP (see

Figure 7.9) [96].

Figure 7.15 Constant current STM image, OMTS, and UPS of a monolayer of nickel (II)

octaethylporphyrin (NiOEP) on Au (111). The absolute energy scale was derived by taking

)m = 5.20 V and ' = – 0.2 V. The STM image was acquired at a sample bias of – 0.6 V and

a set-point voltage of 0.30 nA. The STM image is reprinted with permission from [101].

7. Scanning Tunneling Spectroscopy (STS) 334

Because of the relatively relaxed selection rules associated with near-

resonant elastic electron scattering, OMTS provides more information about

occupied orbitals near E

than does UPS. But OMTS offers much more than

that. The ability to observe unoccupied orbitals (affinity levels) gives OMTS

the capabilities normally associated with inverse photoemission without the

problems created when organic molecules are subjected to the intense electron

beams essential to the inverse photoemission (IPS) process. The practical

resolution of IPS when studying organic adlayers is generally of the order of

0.5 volt [123]. OMTS, on the other hand, has significantly greater resolution

capabilities. Even at room temperature, an instrumental resolution of less than

0.1 volt is easily obtained. By cooling the sample to 100 K, the resolution can

be improved below 0.03 volt (240 cm

–1

). While this has little effect on the

overall band intensity (as discussed above), it does provide sufficient

resolving power to make possible the observation of vibronic structuring on

the OMTS bands.

In the previous section we stated that the agreement between occupied

orbital positions as predicted by oxidation potentials did not compare well

with the OMTS band positions for the ring oxidations of metal porphyrins.

Armstrong and coworkers have also noticed this difficulty in reconciling

solution phase oxidation potentials with thin film UPS data [97, 124]. Rather

than using the simple formula that results from Figure 7.11 (I

= 4.71 eV + E

(sce)

1/2

) they find that multiplication of the oxidation potential by a factor of

about 1.7 is necessary to bring UPS HOMO peaks and solution phase

electrochemical first oxidation potentials into agreement. The close agreement

between UPS and OMTS positions, and the disagreement with

electrochemical predictions, suggests that the OMT process is a vertical

process in the Frank-Condon sense. As shown in Figure 7.16, the center of

the UPS band is the vertical transition labeled 'E

. The width of the band

geometry inherent in M M

transition. The electrochemical potential, on

the other hand, is given by 'E

. The larger the difference in geometry

between the parent and the ion, the greater the difference between 'E

and

. If the OMTS in the porphyrins case is a vertical excitation process, then

the OMTS energy should agree well with that from UPS. Based on our

results, it would appear that the two processes are very similar.



arises from vibrational (molecular and lattice) vibrations associated with the change in

7.5 Practical Considerations 335

Figure 7.16 Schematic potential energy surfaces for a parent molecule M and its most stable

positive ion M

This conclusion should be drawn with care. First, the bandwidths seen in

both UPS and OMTS of thin films are of the order of 0.5 eV, or 4000 cm

–1

Since a typical C-C stretch or C-H bend is less than half this value, there

could be some significant vibronic differences in the transition types that are

hidden within the room temperature peak shapes. Low temperature

measurements are essential in resolving this issue. The second caution is that

each molecular type will interact in a complex way with the tunneling

environment. There is no single mechanism that describes OMTS for all

molecules.

Why, one might ask, is it only the porphyrins that show the large disparity?

One likely answer within the context of Figure 7.16 is that the porphyrin ring

is smaller and less rigid than that of phthalocyanine. Moreover, all of the

porphyrins studied have peripheral substituents that are subject to low

frequency vibrational modes. Thus, one would expect a larger geometry

change upon removing an electron from a porphyrin relative to a

phthalocyanine. A larger geometry change translates into a greater difference

between 'E

and 'E

. If this argument is accepted, one must then ask why the

other OMTS bands of the porphyrins agree with the electrochemical model.

The answer here is more complex and ultimately will require detailed

quantum mechanical calculations. Consider the cobalt ion. The molecular

geometry change most likely to play a role is the change in force constant

between cobalt and nitrogen—not a change in position since the entire ring

tends to restrict the possibilities. A force constant change with no change in

equilibrium geometry leads to no difference in 'E

and 'E

Tsiper, Soos, Gao, and Kahn have performed STM-OMTS, UPS, and IPS

measurements of PTCDA (perylenetetracarboxylic acid dianhydride) films as

a function of thickness [125]. They also find good agreement between the

7. Scanning Tunneling Spectroscopy (STS) 336

three techniques. In this context it is interesting to note that PTCDA is a large

rigid planar molecule and that the first ionization and affinity levels involve

electrons delocalized over much of the molecule.

7.5.3 OMTS as a Chemical Analysis Tool: Direct Spectral

Characterization

Based upon its ability to determine the location of HOMOs and LUMOs,

OMTS offers at least as much chemical selectivity as does electrochemistry.

The limiting factor in the selectivity of OMTS is the thermal line width—of

the order of 3 kT or about 0.09 eV at room temperature. Because typical UV-

Vis bands observed in molecular species are of the order of 0.25 eV, the

thermal broadening is not significant until about 700qC. Thus the selectivity

of OMTS is better than, or of the order of that afforded by, UV-Vis spectra.

To quantify these ideas, consider Figure 7.17.

The room temperature STM-OMTS obtained from single molecules of four

chemically similar species are shown in Figure 7.17. All have a central +2

transition metal ion, all have four coordinating nitrogens, and all have large S

systems. Nevertheless, each is clearly distinguishable from the others by its

OMTS spectrum in a 4-eV window centered about the Fermi energy of gold

(the substrate used). In the cases where the S systems are different, the

positions of the S HOMO and LUMO distinguish the compounds. For those

complexes having the same ligand but different metals, the metal d-orbital

occupancy provides clear selectivity.

What about other types of molecules? Are the porphyrin and phthalo-

cyanine cases really representative, or do they constitute some kind of special

(best) case? To answer that question we have prepared Table 7.1. This table

gives a selection of electrochemical potentials for the first reduction or

oxidation of a wide range of molecules. Polyacenes, hetero-organics,

metalorganics, and transition metal complexes all appear on the list. We have

then used the correlation between first reduction and oxidation potentials

discussed in a previous section, and exemplified in Figure 7.12, to estimate

the position of the first ionization and affinity levels in OMTS. We used the

formula

OMTS

| E

1/2

(sce) + 4.71 V

7.5 Practical Considerations 337