Hawkes P.W., Spence J.C.H. (Eds.) Science of Microscopy. V.1 and 2

Подождите немного. Документ загружается.

940 M.P. Nikiforov and D.A. Bonnell

As discussed earlier, the details of formate adsorption on TiO

(110)

were demonstrated with atomic resolution STM and FM-AFM(2) with

a conductive silicon tip. An untreated (and probably oxidized) Si tip

was used to image the (110) surface.

13,43

The (100) surface of TiO

rather more complex than is the (110) surface, exhibiting a (1 × 3) micro-

faceted structure at high temperatures and a combination (1 × 1) and

(1 × 3) structure at intermediate temperatures. Raza and co-workers

compared the intermediate structure of the TiO

(100) surface with

STM and AFM.

Figure 14–6 illustrates the level of detail evident in

the AFM image contrast. The normalized frequency shift, γ, in this case

is 0.147 fN/m

1/2

, implying that short-range van der Waals or covalent

(Morse potential) forces were accessed. It was noted that the lateral

dimensions of the features in the AFM and STM images agree but the

apparent corrugation amplitude is much larger in AFM (0.5 nm com-

pared to 0.1 nm). This illustrates the importance of height calibration.

In this case the (1 × 3) microfacet was used as an internal standard.

2.4.2 FM-AFM of Semiconductors: Si and GaAs

The Si (111) (7 × 7) reconstruction has long been a standard in surface

science. The structure based on minimization of the number of dan-

gling bonds took 20 years to solve. The “Dimer–Adatom–Staking Fault”

Figure 14–5. SPM on (100) SrTiO

. (a) STM image (V = +0.65 V, I = 0.11 nA); (b)

NC-AFM image (V = −1.49 V); (c) line proﬁ le comparison for STM and NC-

AFM; note that STM shows the Sr adatoms on the surface while NC-AFM

resolves unit cells underneath; (d) the model for

55×

Sr adatoms on

SrTiO

. [Courtesy of Kubo and Nozoye. Reprinted with permission from Physi-

cal Review Letters, 86(9), 1801–1804, 2001.]

Chapter 14 Scanning Probe Microscopy in Materials Science 941

model proposed by Takayanagi et al.

is the currently accepted struc-

ture for the silicon (111) 7 × 7 surface. Immediately after the invention

of the STM, Binnig and co-workers imaged the 7 × 7 reconstruction on

Si (111) in real space by STM.

It took more then 10 years before the

ﬁ rst NC-AFM images of the 7 × 7 reconstruction were obtained simul-

taneously by Kitamura and Iwatsuki

and Giessibl.

Both groups used

FM-AFM(2) to obtain atomic resolution as described earlier (Figure

14–4).

A number of semiconductor surfaces have been imaged since then,

including Si (111), (100), (110); GaAs (100); and InAs (110).

Several

representative examples are presented here. The most technologically

relevant Si (100) undergoes a 2 × 1 reconstruction. Figure 14–7 shows

an atomic resolution FM-AFM(1) image by Yokoyama and co-workers.

The normalized frequency shift calculated for these images is ∼4.35 ×

−14

N/m

1/2

, a magnitude that suggests that the forces are either short-

range van der Waals or Morse forces, hence atomic resolution. Although

the surface is known to adopt the (2 × 1) reconstruction, the distance

between the “dimer-like” features in Figure 14–7 is larger (0.35 nm) that

)

Figure 14–6. NC-AFM image of 1 × 1 reconstruction on TiO

(100) (a); facetted

structure with 1 × 3 and 1 × 1 reconstruction on TiO

(100) (b). [Courtesy of

Ashino et al. Reprinted with permission from Physical Review Letters, 86(19),

4334–4337, 2001.]

942 M.P. Nikiforov and D.A. Bonnell

of the expected structure (0.23–0.20 nm). This result was compared

with an NC-AFM image of the hydrogen passivated surface on which

the lateral dimensions agree with the expected positions of hydrogen

atoms. This dilemma is resolved if the image mechanism is that of a

Si dangling bond of the tip interacting with the surface. In the case of

the clean (2 × 1) surface the contrast then relates to surface dangling

bonds rather than dimer bonds and in the case of the passivated surface

the contrast relates to the dangling bond–hydrogen interaction. These

results further emphasize the role of chemical/bonding forces in the

image mechanism.

The surface of GaAs affords an opportunity to examine differences

in tip–surface inter actions on the same surface. Uehara et al.

compare

the corrugations of As atoms and Ga atoms on a GaAs (110) surface at

different tip–surface distances. Again FM-AFM(2) was used to obtain

atomic resolution. Unfortunately the oscillation amplitude is not speci-

ﬁ ed so a quantitative analysis of the forces is not possible. In this case,

though, the apparent distance between As atoms did not change with

tip–sample distance, while the distance between Ga atoms did. This is

explained by considering that the force on the surface atom increases

with decreasing tip–sample distance. Therefore, the As feature is asso-

ciated with valence charge distribution and is not sensitive to this

range of forces, while the Ga feature is associated with a dangling

bond, which can be displaced by the interaction with the tip. This is

another case of using atomic bonding interactions for chemical speci-

ﬁ city in AFM image contrast.

Figure 14–7. (a) NC-AFM image of Si (100) 2 × 1 reconstructed surface. One 2

× 1 unit cell is outlined by a box. (b) Line proﬁ le of the NC-AFM image across

the dotted line. [Courtesy of Yokoyama et al. Reproduced with permission

from Japanese Journal of Applied Physics Part 2—Letters, 39(2A), L113–L115,

2000.]

Chapter 14 Scanning Probe Microscopy in Materials Science 943

2.4.3 Layered Materials: Graphite

Numerous layered compounds have been studied by NC-AFM, includ-

ing mica,

MoO

etc. Highly oriented pyrolytic graphite is perhaps

the most general in that it is used as a calibration standard for SPM

and as an atomically ﬂ at substrate in numerous studies. As with most

layered compounds the graphite is easily prepared as a clean atomi-

cally ﬂ at surface by cleavage. Despite the proverbial ease of imaging

graphite by STM with atomic resolution, every second atom in the

hexagonal surface unit cell is usually unresolved. Giessibl and co-

workers

demonstrated that a low-temperature FM-AFM(1) with pico-

Newton force sensitivity reveals the hidden surface atom (Figure 14–8).

This is another clear example that despite similarity in implementation

and that combined AFM/STM microscopes are manufactured, differ-

ences between the image formation mechanisms of STM and NC-AFM

are very important. In STM the tip images the density of states at the

tip bias, while in NC-AFM the tip images an equiforce surface. With

an oscillation amplitude of 300 pm, a cantilever spring constant of

1800 N/m, and an eigenfrequency of 18,076.5 Hz, the normalized fre-

quency shift, γ, is −2.7 fN/m

0.5

. The magnitude of the normalized fre-

quency shift indicates that the contrast in FM-AFM(1) is due either

to short-range van der Waals or to Morse forces for the image in

Figure 14–8.

2.4.4 Dielectrics: KBr and CaF

The ability to image insulating surfaces of high resistance has been

noted as an advantage of NC-AFM. Of course the oxides presented

in Section 2.4.1 are insulating compounds when fully oxidized and

undoped. Here we illustrate compounds that are not made conducting.

Benewitz and co-workers present an illustrative example

by imaging

the potassium chloride–bromide surface. True atomic resolution was

Figure 14–8. Low temperature STM (A) and NC-AFM (B) images on graphite. One hexagon of the

graphite structure is superimposed on STM and NC-AFM images. One atom per unit cell for STM and

two for NC-AFM. [Courtesy of Hembacher et al. Reproduced with permission from Proceedings of

the National Academy of Sciences USA, 100(22), 12539–12542, 2003.]

944 M.P. Nikiforov and D.A. Bonnell

not achieved in their analysis of K(Cl,Br) single crystals, but atomic

corrugations with spacing equal to the size of unit cell were resolved.

This difference in corrugation amplitude was attributed to anion posi-

tions on the surface, providing an additional example of chemical

speciﬁ city in NC-AFM.

Giessibl and Reichling

imaged the CaF

(111) surface with atomic

resolution in both attractive and repulsive modes (Figure 14–9). The

level of spatial resolution was facilitated by the use of a quartz tuning

fork to oscillate the tip. The normalized frequency shift was −62 fN/m

0.5

for attractive and 7.4 f N/m

0.5

for repulsive interactions, suggesting that

van der Waals forces are responsible for image formation. The quartz

tuning fork has a high spring constant (∼1000 N/m compared to ∼10 N/

m for silicon cantilever), resulting in high stability of the small-

amplitude oscillations. The quality of these images emphasizes the

importance of small oscillation amplitudes and the proximity of

these oscillations to the surface. Also, imaging in repulsive mode is

challenging but yields high spatial resolution.

3 Imaging Properties: Advanced SPM Techniques

In addition to mapping topographic and atomic structure, variants of

SPM probe local properties. The most common of these are EFM and

magnetic force microscopy (MFM). These are straightforward mea-

surements in which the tip is lifted above the surface to a distance at

which only long-range forces are detected. By expanding the mecha-

nisms of force detection and utilizing all possible image modes, the list

of properties that can be probed is extensive (Table 14–2). The cantile-

ver can be driven mechanically or electrically; this can be done in

Figure 14–9. High-resolution constant height images of CaF2 (111) in (a) the attractive and (b) the

repulsive mode of FM-AFM(1) using qPlus sensor as canteliver. (Courtesy of Giessibl and Reichling.

Reproduced with permission from Nanotechnology, 16, (2005) S118–S124, 2005.

Chapter 14 Scanning Probe Microscopy in Materials Science 945

contact or noncontact regimes; and not only amplitude, but also ﬁ rst,

second, and third harmonics of amplitude (or phase) can be detected.

The relation between these operational regimes is shown in Figure

14–10. For example, conventional AFM is a noncontact, mechanically

driven oscillation with amplitude or phase-based detection. The newer

piezoelectric force microscopy (PFM) is in contact mode, electrically

driven, and with phase-based detection. Static or periodic electric or

magnetic ﬁ elds can be applied to the sample, independent of the tip

signal. An underlying theme of the newest developments is the use of

multiple signal modulations or high order harmonics of modulated

signals. Within this framework several techniques that address trans-

port and dielectric properties will be reviewed.

3.1 Advanced SPM Techniques for Transport Properties

Perhaps the most common of the so-called “advanced” SPM techniques

is scanning surface potential microscopy (SSPM), sometimes referred

to as Kelvin probe force microscopy (KPFM), which maps the work

function of surfaces.

56,57

In SSPM, the cantilever oscillation is driven

electrically. This is a noncontact (generally 50–200 nm separation)

probe, with feedback on the ﬁ rst harmonic (Figure 14–11). An ac voltage

Table 14–2. Properties accessible with scanning probe microscopy.

Tech nique Mode Property References

AFM nc/ic, mech, phase/amp vdW interaction, topography [2–5]

EFM nc, mech, phase/amp Electrostatic force [2–5]

MFM nc, mech, phase/amp Magnetic force, current ﬂ ow [2–5]

SSPM nc, elec, 1st harmonic Potential, work function, [2–5, 56–75]

(KPM) adsorbate enthalpy/entropy

SCM c, F, cap sensor Capacitance, relative dopant density [2–5, 82, 84–87,

SCFM c, elec, 3rd harmonic dC/dV, dopant proﬁ le 89–92]

SSRM c, F, dc current Resistivity, relative dopant density [2–5, 81–83, 87]

SGM nc, elec, amp Current ﬂ ow, local band energy, [2–5, 97]

contact potential variation

SIM nc, elec, phase/amp Interface potential, capacitance, time [93, 97, 99]

constant, local band energy,

potential, current ﬂ ow (in

combination with SSPM)

NIM c, F, freq spectrum Interface potential, capacitance, time [102–104]

constant, dopant proﬁ ling

PFM c, elec, phase/amp d33 [63, 105–111, 125,

126, 128, 129]

NPFM c, elec, 2nd harmonic Switching dynamics, relaxation time [130]

and domain nucleation

SNDM c, F, 1st or 3rd harmonic dC/dV, dielectric constant [133, 134]

NFMM c, F, phase Microwave losses, d33 [135–138]

946 M.P. Nikiforov and D.A. Bonnell

is applied to the tip. V

is varied until it matches the surface potential,

surf

, at which point the ﬁ rst harmonic of the force F (with frequency

ω) is nulliﬁ ed. Thus, the applied V

is equal to the surface potential.

The tip potential is expressed as V

tip

= V

+ V

sin ωt, and the force

acting between the tip and surface F(z) = -

(∂C/∂z)[V

tip

− V

surf

]

, so the

ﬁ rst harmonic of the force is F

(z) = (∂C/∂z)[V

− V

surf

The apparent ease with which potential variations can be mapped

belies the challenges in interpreting images on electrically

58,59

and

topographically inhomogeneous surfaces. In the cases of semiconduc-

tor and dielectric surfaces the electrostatic properties of a surface are

not characterized solely by intrinsic potential and topography. SSPM

images must be interpreted in terms of effective surface potential that

includes capacitive interactions, surface and volume bound charges,

double layers, and remnant polarization.

60–63

For semiconductor sur-

faces tip-induced band bending

can offset local surface potential.

Despite these challenges the obvious need to examine variations in

Figure 14–10. Variations of SPM are separated into four operational regimes

distinguished by cantilever driving force and condition of contact.

V = V

+ V

sin(ωt)

Function

Generator

Lock-in

AFM

controller

Deflection at ω

Reference at ω

V = V

sin(ωt)

Figure 14–11. Comparison of working principles for SSPM (KFM) and SIM. In SSPM (a) the oscillating

potential is applied between the sample and tip; in SIM (b) the oscillating potential is applied across

the sample.

Chapter 14 Scanning Probe Microscopy in Materials Science 947

local potential in electronic nanodevices spurred efforts to overcome

some of the obstacles with careful analytical treatments that deter-

mined limits in quantiﬁ cation. In the late 1990s SSPM was applied to

semiconductor,

66,67

organic,

and ferroelectric

69,70

surfaces, as well as to

defects,

71,72

and photoinduced

73,74

and thermal phenomena.

It is fair

to say that at this point that absolute values of potential cannot be

quantiﬁ ed, but variations in potential can be determined with energy

resolution of 2–4 meV and spatial resolution of the order of 50–100 nm.

One interesting application is the measurements of surface potential

of self-assembled monolayers (SAM). Figure 14–12 shows a typical

potential map of two SAMs patterned by microcontact printing. In this

case one SAM is an alkanethiol, while the other is a conjugated conduc-

tive molecule. The difference in surface potential is obvious but the

basis of the difference is not entirely clear. There are, in principle, four

contributions to the potential of SAM: the substrate work function, the

dipole in the surface bond (usually S—M or SiO

—Si), the dipole in

the molecule, and the terminating end group. Eng and co-workers

showed that the potential of alkanethiols on Au increases with the

increase of chain length.

Sugimura et al. calculated dipole moments

in a sequence of siloxane-coupled molecules and the trends agreed

with the experiment.

These results provide evidence for the contribu-

tion of the molecular dipole to the measured surface potential. Alvarez

and Bonnell

showed that for the SAM of alkanes on metal, the S—M

bond dipole also affects the measured surface potential. While work

remains to quantify the relative contribution of SAMs to surface poten-

tial, SSPM easily quantiﬁ es variations and is routinely used to charac-

terize complex ﬁ lms.

On samples with morphological variations, such as grains, SSPM in

the presence of a lateral ﬁ eld can elucidate the behavior of individual

microstructural features. This is illustrated in Figure 14–13, which

shows the behavior of a polycrystalline ZnO surface under different

applied lateral biases. The topography contains features due to con-

taminants and inter- and intragranular pores. On application of 5 V

Figure 14–12. SSPM image of SAM of alkanethiol and conjugated molecules

self-assembled on Au, patterned with a microcontact printing technique.

(Courtesy of Rodolfo Alvarez.)

948 M.P. Nikiforov and D.A. Bonnell

lateral bias the potential steps at grain boundaries become evident

(Figure 14–13b), and the contrast inverts when the direction of the

applied bias is reversed (Figure 14–13c). Clearly the behavior of indi-

vidual grains and grain boundaries is distinguished and the voltage

dependence of all microstructural constituents can be examined sepa-

rately. Furthermore, taking directional derivatives allows the current

ﬂ ow to be determined at all points on the surface, as shown in Figure

14–13d.

Although not explicitly stated, much SSPM is done in ambient condi-

tions. Indeed the ease of measurement is a strong advantage for its use

as a qualitative characterization tool. As with all ambient measure-

ments, the possible interaction of the environment with the surface

must be considered. There are, in fact, situations in which these effects

dominate. Some of the early studies involved imaging ferroelectric

domains in air

and it was eventually understood that the strong ﬁ eld

due to domain polarization at a surface is almost completely screened,

presumably by adsorption. Domains are easily imaged but quantifying

the magnitude of the surface potential is impossible. In fact the sign of

the measured potential is usually opposite that of the domain poten-

tial. Ferroelectric domains represent an extreme case of local electric

ﬁ elds, but even for grain boundaries intersecting a surface the effect is

observed. Figure 14–14 compares the potential of an atomically abrupt

boundary in SrTiO

measured in air and in ultrahigh vacuum (UHV).

(c)

(a)

(d)

(b)

Figure 14–13. Transport imaging in polycrystalline ZnO. (a) Surface topogra-

phy. (b and c) SSPM images under lateral bias exhibit potential drops at grain

boundaries, indicative of grain boundary resistive behavior. Note that the

direction of potential drops inverts with bias. (d) Current maps for positive

and negative bias polarity. (Partially reproduced from Kalinin and Bonnell,

Zeitschrift fur Metallkunde, 90, 983–989, 1999.) (See color plate.)

Chapter 14 Scanning Probe Microscopy in Materials Science 949

The boundary contains an intrinsic charge, which results in a potential

variation at the intersection with the surface. In ambient conditions the

charge attracts compensating adsorbates and the measured difference

in surface potential is under 10 mV. In UHV, the uncompensated dif-

ference in potential is on the order of 90 mV and opposite in sign.

Two contact techniques, scanning spreading resistance microscopy

(SSRM)

81–83

and scanning capacitance microscopy (SCM),

84–86

have been

developed to characterize semiconductors. In SSRM, a conducting tip

is biased with respect to the sample and the dc current through the

tip/surface contact is detected under force feedback control. The amount

of current is determined by the local spreading resistance of the surface,

which is related to the local conductivity. Because the native oxide layer

on Si hinders tip/surface contact, carrier proﬁ ling in Si-based devices

usually requires a tip coated with a hard material and a high spring

constant cantilever to provide strong indentation forces (∼20 µN).

SCM, a high-frequency capacitance sensor detects the tip/sample

capacitance as the tip is scanned across the sample, which allows a

smaller indentation force than is used in SSRM. Diamond-coated tips

are used in spreading resistance measurements, but lower indentation

force makes metal-coated probes suitable for SCM.

An ac voltage

applied to the tip induces the depletion and accumulation of carriers,

resulting in a change in capacitance, ∆C. In a semiconductor, the

depletion/accumulation width is inversely related to the carrier con-

centration, so mapping ∆C/∆V yields a carrier concentration proﬁ le.

Difﬁ culties in quantiﬁ cation arise if the dopant concentration is non-

uniform, and when spatial resolution decreases due to low dopant con-

centrations. A recent modiﬁ cation incorporates an additional feedback

that adjusts ∆V to maintain a constant ∆C during the scan, maintaining

a constant depletion width.

Analysis of SCM results are mathematically

200nm

0200 400 600 800 1000

Surface potential, mV

Lateral distance, nm

0200400600 800 1000

-100

-75

-50

-25

Surface potential, mV

Lateral distance, nm

Figure 14–14. KFM of the Σ5 grain boundary in SrTiO

(100). Left, in air; right, in UHV. (Courtesy of

Rui Shao.)