Gupta D. (Ed.). Diffusion Processes in Advanced Technological Materials

Подождите немного. Документ загружается.

carbon, boron, and/or oxygen leads to the so-called stuffing of the fast dif-

fusion paths, such as grain boundaries and dislocations, and concomi-

tantly to the barrier effectiveness. Note that the presence of oxygen in

these materials is related to extremely high affinity of the refractory met-

als that make these compounds. Invariably the oxygen, trapped in such

films from contamination in the deposition chamber, has been detected

but ignored from discussions. Such small amounts of oxygen can stuff the

grain boundaries and/or form the oxide on the surface or on the grain

boundary surfaces. For example, a Ti film deposited on air-exposed Al

was stable during a 86-hour, 350°C anneal.

[6]

When both layers were

deposited during one pump-down of the evaporator, a reaction between Ti

and Al occurred readily.

[43]

The effect of nitrogen as an impurity in Mo

used as a barrier between Ti and Au and in the case of Ti-W as the effec-

tive barrier has been clearly demonstrated.

[44–46]

In many such triple-layer

barriers, an exposure to air was found to be the key to the success of these

barriers.

[6, 47]

Similarly, all carbon-rich carbides were effective barriers,

whereas titanium-rich carbides were not.

[48]

Presently, all technologies using Al as the interconnect metal use a

TiTiN bilayer or a TiTiNTi trilayer as the APDB layer. Ti is used to

provide enhanced adhesion and protection from nitrogen exposure of the

underlying material. These layers are also now used as antireflective coat-

ing to facilitate lithography on Al.

5.4.2 Barriers Between the ILD and Cu

Tables 5.2 and 5.3 list resistivities and reported effectiveness (with

Cu) of various electrically conducting barrier films. Table 5.2 lists the

compounds, and Table 5.3 lists the elemental metals. The resistivity and

thickness of APDB material will determine its applicability in schemes of

the ULSIGSI interconnection technologies using Cu. Although the

resistivities of these materials in pure bulk form could be low, the resis-

tivities of the same materials in thin-film form are considerably higher,

depending on the techniques used for deposition and purity. Both the

resistivity and thickness of APDB material should be as low as possible.

A simple computation of the total interconnect resistance, as illustrated

in Fig. 5.4, shows the need for a barrier thinner than 10 nm

[49]

if the resis-

tivity advantage of copper over aluminum or its alloys is to be fully

exploited. The higher the resistivity of the barrier film, the lower the

thickness required. As the metal width is reduced, the barrier thickness

must also be reduced to maintain the resistance advantage of a given

scheme of metallization. Simple calculations of the line resistance of

copper with and without barriers of 150 and 5 mΩ cm resistivities,

262 DIFFUSION PROCESSES IN ADVANCED TECHNOLOGICAL MATERIALS

Ch_05.qxd 11/29/04 6:20 PM Page 262

DIFFUSION BARRIERS IN SEMICONDUCTOR DEVICES, MURARKA 263

Table 5.2. Resistivities and Reported Barrier Effectiveness (with Cu) of

Various Barrier Films (Refractory Compounds)

Reported

Barrier Resistivity Film Thickness Stable to

Material (mΩ cm) (nm) (°C) References

TiN 20

to1000

100 to 110 500 a

40 450 b

10 400 to 450 c

TiCN 2000 10 400 d

600 e

Nitrogen-

treated 200 10 400 f

TiCN

660/410

***

∼100 850 g

TaN 200 ∼8 700 h

200 50 700 a

N 200 50 500 a

500 40 700 g, i

WN/W – 25 650 j, k

2000 100 700 l

Wsi

0.6

N– ∼30 600 m

Reported approximate value for the pure and bulk material.

Thin films of reported resistivity as high as this number.

***

As-deposited/annealed l hour at 700°C.

a. N. Awaya, H. Imokawa, E. Yamamoto, Y. Okazaki, M. Miyake, Y. Arita, and T. Kobayashi, in

Conf. Proc. VMIC, Cat. No. 951ISMIC-104, VMIC, Tampa, FL (1995), p. 17

b. A. Berti and S. P. Murarka, Mat. Res. Soc. Symp. Proc., 318:451 (1994)

c. S. Kumar, M. S. thesis, Rensselaer Polytechnic Institute, Troy, NY (1995)

d. N. Agrawal, M. S. thesis, Rensselaer Polytechnic Institute, Troy, NY (1998)

e. M. Eizenberg, Mat. Res. Soc. Symp. Proc., 427:325 (1996)

f. D. Smith, Applied Materials, Santa Clara, CA, private communication (1993)

g. X. Sun, J. S. Reid, F. Kolawa, and M.-A. Nicolet, in Conf. Proc. VLSI XI, MRS, Pittsburgh,

PA (1996), p. 401

h. T. Oku, M. Uekubo, E. Kawakami, K. Nii, T. Nakano, T. Ohta, and M. Mrakami, in Conf.

Proc. VMIC, Cat. No. 951ISMIC-104, VMIC, Tampa, FL (1995), p. 182; see also J. O.

Olowolafe, C. J. Mogab, R. B. Gregory, and M. Kettke, J. Appl. Phys., 72:4099 (1992);

K. Holloway, P. M. Fryer, C. Cabral, Jr., J. M. E. Harper, and K. H. Kelleher, J. Appl. Phys.,

71:5433 (1992); and L. A. Clevenger, N. A. Bojarczak, K. Holloway, J. M. E. Harper,

C. Cabral, Jr., R. G. Schad, F. Cardone, and L. Stolt, J. Appl. Phys., 73:300 (1993)

i. E. Kolawa, J. S. Chen, J. S. Reid, P. J. Pokala, and M.-A. Nicolet, J. Appl. Phys., 70:1369 (1991)

j. T. Nakano, H. Ono, T. Okta, T. Oku, and M. Murakami, in Conf. Proc. VMIC, VMIC, Tampa,

FL (1994) ; see also Kailasam et al.

[69]

k. J. G. Fleming, E. Roherty-Osman, J. Custer, P. MartinSmith, J. S. Reid, and M.-A. Nicolet, in

Conf. Proc. VLSI XI, MRS, Pittsburgh, PA(1996), p. 369

l. T. Iijima, Y. Shimooka, and K. Suguro, in Conf. Proc. VLSI XI, MRS, Pittsburgh, PA(1996), p. 325

m. S.-Q. Wang, in Conf. Proc. VLSI IX, MRS, Pittsburgh, PA (1994), p. 31

Ch_05.qxd 11/29/04 6:20 PM Page 263

264 DIFFUSION PROCESSES IN ADVANCED TECHNOLOGICAL MATERIALS

Table 5.3. Resistivities and Reported Barrier Effectiveness (with Cu) of

Various Barrier Films (Elemental Metals)

Barrier Resistivity Reported Film Stable to

Material (µΩ cm) Thickness (nm) (°C) References

Cr ∼15 20 to 60 200 to 650 a

Ta ∼15 to 200 20 to 60 200 to 650 a, b

W ∼7 to 10 50 to 60 200 to 600 a

Nb ∼15 to 200 60 500 a

Mo ∼6 to 10 60 500 a

Co ∼10 – 250 to 450 a

Ni ∼10 – 250 to 450 a

Pd ∼11 to 12 100 200 a

Pt ∼11 to 12 – 250 to 450 a

α-C High 6.5 to 13.5 500 b, c, d

35 to 7 350 e, f

10 350 e, f

Original references cited in Wang

Barrier effectiveness determined by electrical tests of p-Si/SiO

/Al/Cu capacitors. Reaction does occur in

Al/Cu sandwich leading to higher resistivity if Al concentration in Cu exceeds 0.3 at.%.

a. S.-Q. Wang, in Conf. Proc. VLSI IX, MRS, Pittsburgh, PA (1994), p. 31

b. N. Awaya, H. Imokawa, E. Yamamoto, Y. Okazaki, M. Miyake, Y. Arita, and T. Kobayashi, in Conf. Proc.

VMIC, Cat. No. 951ISMIC-104, VMIC, Tampa, FL (1995), p. 17

c. R. G. Purser, J. W. Strane, and J. W. Mayer, Mat. Res. Soc. Symp. Proc., 309:481 (1993)

d. Jan M. Neirynck, thesis, Rensselaer Polytechnic Institute, Troy, NY(1996)

e. E. Kirchner, S. P. Murarka, E. Eisenbraun, and A. Kaloyeros, Mat. Res. Soc. Symp. Proc., 318:319 (1994)

f. E. Kirchner, Ph. D. thesis, Rensselaer Polytecnic Institute, Troy, NY (1996)

Figure 5.4 The resistance ratio of the various interconnection schemes as a func-

tion of the TiN barrier thickness.

Ch_05.qxd 11/29/04 6:20 PM Page 264

DIFFUSION BARRIERS IN SEMICONDUCTOR DEVICES, MURARKA 265

respectively, are shown in Fig. 5.5

[50]

to show the need for lower resis-

tivity APDB materials. Conventional deposited barrier materials may not

be useful in devices with minimum feature sizes of 50 to 120 nm unless

a technique is found to make these films robust at thicknesses 5 nm.

Note that “stuffing” the fast-diffusion paths, dislocations, and grain

boundaries (see Sec. 5.4.1) does not help because stuffing leads, in gen-

eral, to an increase in resistivity and is the primary cause of highly vari-

able resistivities reported in the literature.

The use of pure metal as an APDB can be incorporated using a dif-

ferent approach.

[51]

Part of the APDB is allowed to interact with the dielectric

Figure 5.5 Total resistance of a 1-mm-thick interconnect line as a function of the

interconnect linewidth.Comparison of pure Cu with a Cu with 20-nm-thick diffusion

barrier/adhesion promoter of 150 mΩ cm (a) and 5 mΩ cm (b) resistivity. The ratio

of Cu plus barrier resistance to pure Cu resistance is also shown.

(a) 150 µΩ cm resistivity

(b) 5 µΩ cm resistivity

Ch_05.qxd 11/29/04 6:20 PM Page 265

and the top metal in such a way that the reaction products stabilize the

dielectric-APDB-metal stacked layers against further interactions during

thermal or bias-temperature excursions. This approach, which can also be

taken using Cu alloys, is discussed at the end of Sec. 5.5.1.

Wang (73) has reviewed the effectiveness of these barriers for Cu

technologies. In all cases, the barrier effectiveness was determined using

the barrier between silicon and copper (that is, Si/barrier/Cu) and exam-

ining the interactions between Si and Cu, Si and the barrier, and the barrier

and Cu. The barrier thicknesses used were larger than 20 nm (generally

much larger). A variety of techniques were used to evaluate the interac-

tions and performance of barriers as a function of annealing in different

conditions. Both adhesion promoter and diffusion barrier are needed

between an ILD and Cu; the material interactions may be different and the

methodology required to determine barrier performance (such as leakage

current) may be application-dependent. However, Wang’s conclusions

regarding various barriers are very instructive, and we reproduce these in

the following paragraphs.

[52]

(Note that the references cited in the quota-

tion have been renumbered, to follow those in this chapter.)

“The reported stable temperatures (T

) for transition metal barriers

between thin Cu films and Si substrates range from 200 (Pd, Cr and Ti) to

650°C (Ta). Contradictory results were reported for W barrier (T

varies

from 200 to 500°C) and for Ta barrier (T

varies from 200 to 650°C). Cu

reacts with near noble metals such as Cr, Co, Ni, Pd, Pt in the temperature

range of 250 to 450°C but is non-reactive and immiscible with refractory

metals such as Mo, Ta and W(T

 750°C)

[53]

. It should also be noted that

the metals from the first group react with Si to form silicides at lower tem-

peratures (100–450°C) than those in the second group (525–650°C).

[54]

Therefore, it is not surprising that the metals from the first group fail as

diffusion barriers at lower temperatures than those in the second group.

The failure mechanism is usually due to the high reactivities of barrier

metal-Cu or Si-barrier metal, followed by ultimate Cu-Si reaction for the

metals from the first group and diffusion of Cu through grain boundaries

of the polycrystalline barrier films at relatively low temperatures for the

metals from the second group. In general, transition metals are not stable

diffusion barriers between Cu and Si.

“Transition metal alloy barriers studied usually consist of one near-

noble metal (reactive with Cu to form compounds) and one refractory metal

(non-reactive and immiscible with Cu) element. They can be deposited in

amorphous states, hence free of grain boundaries. However, the crystal-

lization temperatures of these amorphous alloy films are not as high as

those of some other binary or ternary amorphous systems which consist of

at least one non-metal element (see below). Grain boundaries form in these

films at elevated temperatures. Nevertheless, T

of transition metal alloy

266 DIFFUSION PROCESSES IN ADVANCED TECHNOLOGICAL MATERIALS

Ch_05.qxd 11/29/04 6:20 PM Page 266

DIFFUSION BARRIERS IN SEMICONDUCTOR DEVICES, MURARKA 267

diffusion barriers are all above 500°C, higher than those of polycrystalline

transition metal diffusion barriers. Metal alloy-Cu reaction is suppressed

due to the presence of the non-reactive and immiscible refractory metal

element. The temperature of reaction at the Si/metal alloy interface

depends on the composition and crystal structure of the metal alloys. It is

between the reaction onset temperature of the system of Si/near noble

metal component of the metal alloy, and that of the system of Si/refractory

metal component of the metal alloy. Failure occurs as a result of Si-metal

alloy reaction or Cu diffusion through grain boundaries of polycrystalline

metal alloy films at elevated temperatures.

“The effectiveness of silicide or transition metal-silicon systems as

diffusion barriers follows the similar trend of their corresponding transi-

tion metal diffusion barriers. The diffusion barriers formed by refractory

metal (Ta and W)-Si are more stable (T

ranges from 600 to 700°C) than

near noble metal (Cr and Co-Si) barriers (T

ranges from 200 to 300°C).

Adding Si to W to form an amorphous W-Si diffusion barrier improves

barrier performance significantly (T

 600–700°C) as compared to poly-

crystalline W barrier (T

 250–500°C). The barriers fail by the reaction

of Cu with transition metal silicide to form Cu-silicide, by Cu diffusion

through polycrystalline transition metal silicides or by Cu-induced pre-

mature crystallization of the amorphous metal-silicon barrier films. Ti-N,

Ta-N and W-N are three transition metal-nitrogen systems studied for dif-

fusion barrier applications. High stability of these barriers is achieved due

to non-reactivity of Cu with N, Ta and W. Barrier failure is caused by dif-

fusion of Cu along grain boundaries or through weak spots generated at

elevated temperatures in barrier films, which are relatively intact, or by

the reactions between barrier films and Si to form metal-rich silicides. The

conducting metal-oxygen diffusion barrier (Ru

O and Mo-O) can survive

thermal anneals up to the temperature range of 500 to 600°C. Cu-oxide

tends to form at the metal-oxygen/Cu interface at higher temperatures.

The amorphous TiB

barrier breaks down above 750°C.

“Some amorphous ternary systems consisting of either one or two

nonmetal components such as Ta

and TiPN

, exhibit highly sta-

ble barrier properties due to their high crystallization temperatures.

Although their crystallization process is somewhat accelerated by the

presence of Cu overlayer, T

are still among the highest. The barrier

property of amorphous C is marginal, which shows a best T

of no more

than 500°C.

“Polycrystalline compound TiN and polycrystalline TiW (usually

W-rich) solid solution are the two most commonly used diffusion barrier

materials for Al metallization on Si substrates and have been successfully

integrated into current VLSI devices. TiW can prevent interaction

between Cu and Si at temperatures up to 775°C in an RTP anneal time of

Ch_05.qxd 11/29/04 6:20 PM Page 267

30 sec, as reported by Wang et al.

[55]

for TiN barriers range from 400 to

700°C (30 min anneal) or to 900°C (30 sec RTPanneal) The failure mech-

anism for these two diffusion barriers between Cu and Si depends on the

reactivity of Cu and Si with each constituent of the barriers, or interdiffu-

sion of Cu and Si through grain boundaries of the barriers. Cu does not

react with either W or N. Ti is bonded with N in the TiN compound and

tied up with Win the TiW solid solution Therefore, free Ti is not available

and reactions between Cu and either the TiN or TiW do not occur up to

the lower temperature between the Si/diffusion barrier reaction tempera-

ture and the interdiffusion temperature of Cu and/or Si through grain

boundaries of the barriers. TiN is stable in contact with Si to high tem-

peratures. However, TiW reacts with Si to form binary and ternary sili-

cides at 800°C.

[55]

Up to this temperature, no significant Cu diffusion

through TiW grain boundaries occurs. Therefore, the failure mechanism

for TiN in the Si/TiN/C structure is not barrier dissociation by chemical

reaction but rather by diffusion of Cu in TiN films along grain boundaries

or through weak spots generated at elevated temperatures.

[56]

TiW, on the

other hand fails as a diffusion barrier between Cu and Si by dissociation

reaction with Si substrates.

[55]

“The studies conducted by Kolawa et al.

[57–60]

using polycrystalline

Ta, amorphous Ta

and amorphous Ta

as diffusion barriers

between thin Cu films and Si substrates can be used here to illustrate the

process of choosing a stable diffusion barrier. Ta is non-reactive immisci-

ble with Cu, and reacts with Si to form TaSi

around 650°C.

[54]

Therefore,

Ta is a good diffusion barrier candidate compared to other polycrystalline

transition metal films which usually react with both Cu and Si at relatively

low temperatures. However, Ta fails to prevent Cu from diffusing through

its grain boundaries to react with Si at 500°C in the Si/Ta/Cu structure, as

detected by leakage current measurement. To prevent Cu grain boundary

diffusion, Si is added to Ta to create a grain-boundary-free amorphous

diffusion barrier with a crystallization temperature of about

850°C. Unfortunately, the crystallization temperature of Ta

film

drops to 650°C when in contact with Cu. As a result, Cu diffuses through

grain boundaries of the crystallized Ta

film at 650°C and reacts with

Si, resulting in barrier failure. In order to improve the performance of

Ta-Si barrier Ta

barrier is made by reactively sputtering Ta

. This film is amorphous and has higher crystallization temperature

(1100°C) than that of Ta

. Consequently, the Ta

barrier does

not fail until 950°C. By using amorphous Ta

and Ta

barriers,

the stable barrier temperature is improved by 150°C and 450°C respec-

tively, over that of polycrystalline Ta barrier. As a matter of fact

has been reported to be a stable diffusion barrier between A1

and Si at 700°C (above the Al melting temperature) and is also the most

268 DIFFUSION PROCESSES IN ADVANCED TECHNOLOGICAL MATERIALS

Ch_05.qxd 11/29/04 6:20 PM Page 268

DIFFUSION BARRIERS IN SEMICONDUCTOR DEVICES, MURARKA 269

stable diffusion barrier ever reported for Cu metallization. Further work

needs to concentrate on integration issues such as patterning, electromi-

gration and stress voiding behavior of the metallization on this barrier

material.”

Most of the published work on this topic has paid attention to the bar-

rier needed between Cu and SiO

as the dielectric. Polymers offer lower

dielectric constants and are being researched for their application as the

interlayer dielectric material, replacing deposited SiO

-based materials.

An APDB will be needed for the Cu-polymer interconnection scheme. A

variety of polymers could find application as an ILD, with one APDB

probably not optimum for all polymers of interest. Each polymer presents

a different chemical composition and surface structure to which adhesion

must be achieved and into which diffusion must be inhibited. Since all

polymer surfaces are carbon-rich, bonding to carbon may lead to the

desired adhesion. Most transition metals bond well with C; for example,

the diatomic bond strengths of C-Ti, C-Nb, and C-Ir are 101, 136, and

151 kcal/mol, respectively.

[61]

In this regard, a carbide like TiC may be a

good choice.

[48]

Such barriers might be a universal APDB material for

polymers.

As an alternative to conducting barriers, we can use insulating dielec-

tric materials like Si

or SiO

as APDB materials. Cho et al.

[62]

com-

pared 4% phosphosilicate glass, plasma-enhanced CVD silicon dioxide,

low-temperature CVD silicon dioxide, plasma-enhanced silicon oxyni-

tride, and CVD silicon nitride subjected to bias temperature aging at

250°C with a copper electrode on these dielectrics. Capacitors with sili-

con nitride showed no shifts, indicating that this material is an excellent

barrier. However, use of thicker Si

layers is undesirable because of the

high dielectric constant (approximately 7), which contributes significantly

to the effective dielectric constant of the ILD. Also, as mentioned earlier,

such APDB material will require a clean etch process to remove it from

the metal surfaces in contact windows.

5.5 New Concepts in Affecting APDB Behavior

at the Interfaces

5.5.1 Alloying of Cu to Form an APDB at

Interfaces/Surfaces

Two approaches, different from the conventional ones, have been

used recently to address the APDB materials needs. The first approach

explores alloying Cu with small amounts of another element.

[8, 63, 64]

The

Ch_05.qxd 11/29/04 6:20 PM Page 269

criteria for selecting this alloying element include (1) low solubility and

low ∆r∆c in Cu, (2) the capability to form chemical bonds through a

thermodynamically favorable self-limiting interaction with the ILD, and

(3) high diffusivity in Cu and low diffusivity in the ILD and in the inter-

facial reaction product on the ILD surface. Here ∆r∆c refers to the resis-

tivity change ∆r for a concentration change (∆c) of the alloying element

in Cu. The role of the solid solubility and the interfacial reactivity, as

determined by thermodyanamic factors, were discussed in Sec. 5.3.3. A

large variety of alloying elements, selected arbitrarily or using the criteria

given, have been examined: Al, B, Be, Ga, In, Sn, Mg, Ti, Nb, Ni, Pd, Ti,

V, W, Zn, Ca, Cd, Zr, Ag, and Au. Some of them (Ag, Au, Ca, Cd, Zn, Pd,

Be, Ga, and In) do not satisfy one or more of the criteria. For example, Ag,

Au, and Pd offer low ∆r∆c but will not provide enhanced adhesion as

defined by the second criterion. There are others, such as Sn, Zr, and Ti,

that have a very high ∆r∆c ratio. We had carefully examined these cri-

teria and possible alloying elements for Cu in the late 1980s before select-

ing Mg, Al, and B. Once ∆r∆c was established by experiments, Ta was

included in the list. However, it was found that for Ta-Cu alloys, very high

temperatures (higher than 500°C; usually 800°C) and long anneals are

needed to bring the alloy resistivity down to an acceptable range of 2 to

2.5 mΩ cm. Similar but unreported results were obtained with Nb addi-

tions to Cu.

[65]

Such high temperatures are impractical in modern chip

fabrication.

Interactions of Mg or Al with Cu and the effectiveness of’Mg and Al

alloying additions in Cu in concentrations below the solid solubility limit,

including the evaluation of the electrical behavior of metal-oxide semi-

conductor (MOS) capacitors made with these alloys, have been thor-

oughly investigated and reviewed.

[1, 63–67]

Copper alloys containing either

2 at.% Mg or approximately 0.5 ∼ 2 at.% Al with electrical resistivities

of about 2 or 2 to 5 mΩ cm, respectively, seem to provide all the necessary

properties for use as APDB materials on the SiO

used as an ILD. In the

case of Mg, high-resolution microscopy has demonstrated that Mg does

react with SiO

to form interfacial bonding or a barrier layer a few mono-

layers thick.

[66, 67]

5.5.2 Self-Assembled Molecular Monolayers

The second approach is radically different: self-assembled molecular

(SAM) monolayers are used as an APDB interfacial layer between Cu and

SiO

[9]

Because the thicknesses of SAMs are near zero, they will occupy

an insignificant fraction of the total via/hole volume, thereby maximizing

the room for filling in low-resistivity Cu metal. Because of their high

270 DIFFUSION PROCESSES IN ADVANCED TECHNOLOGICAL MATERIALS

Ch_05.qxd 11/29/04 6:20 PM Page 270

DIFFUSION BARRIERS IN SEMICONDUCTOR DEVICES, MURARKA 271

sticking probability on the substrate and low probability of depositing on

themselves, they offer very good step coverage, which is necessary for

effective APDB behavior. Krishnamoorthy et al.

[68]

investigated SAMs

where, characteristically, trimethoxysilane (Si with three OCH

groups) is

tethered to the SiO

substrate and the fourth Si bond is attached to a tail

group that consists of aliphatic and/or aromatic groups, for example, a

3-[2-(trimethoxysilyl)ethyl] pyridine and n-propyl trimethoxysilane. MoS

devices of Cu/SiO

/Si (001), with and without SAMs, were subjected to

biased thermal stressing (BTS) at 200°C with a nominal electrical field of

2 MV/cm. The current-voltage and capacitance-voltage measurements

showed that SAMs with an aromatic terminal group were effective adhe-

sion promoters and barriers to Cu diffusion into SiO

Figure 5.6 shows typical normalized capacitance C/C

max

versus volt-

age plots, one each in the as-deposited state, that is, t

BTS

 0, and one

Figure 5.6 Typical normalized capacitance C/C

max

vs.voltage plots of MOS capac-

itors with and without the self-assembled molecular (SAM) monolayer between Cu

and SiO

, bias annealed at 200°C for duration t

BTA

with an electric field of 2 MV cm

1

Open symbols without SAM; closed symbols with SAM.

Ch_05.qxd 11/29/04 6:20 PM Page 271