Martin P.M. Handbook of Deposition Technologies for Films and Coatings, Third Edition: Science, Applications and Technology

Подождите немного. Документ загружается.

Unﬁltered and Filtered Cathodic Arc Deposition 501

As with closed ﬁlters, open ﬁlters can have a variety of shapes and sizes. Each of the ﬁlters has

advantages and disadvantages, as discussed in greater detail in dedicated reviews ([138] and

Chapter 7 of [3]), where other geometries such as the Venetian blind ﬁlter or large area ﬁlters

for elongated cathodes are also considered.

10.5 Energetic Deposition

10.5.1 Energy Considerations

Film growth mechanisms are dealt with in Chapter 2 of this handbook and many other

publications [139–141]. Therefore, it is sufﬁcient here to point out some special effects and

properties associated with the high degree of ionization and relatively high energy of the

species in the condensing plasma.

Energetic deposition may be deﬁned as a process in which most ﬁlm-forming ions or atoms

exceed the displacement energy [142], E

, allowing them to penetrate the surface and come to

rest under the surface. Film growth can therefore occur under the surface rather than on the

surface. Such a process can be considered as ultra-shallow implantation or subplantation

[143–145]. As a result, ﬁlms are often highly defective and hard, and show high intrinsic

stress. Yet, owing to ﬁlm–substrate intermixing and with the formation of chemical bonds, the

ﬁlms can be well adherent unless the intrinsic stress is excessive, which leads to catastrophic

failure by delamination. Therefore, special attention must be given to stress control, as

discussed below.

At the low-energy end of energetic deposition, when the kinetic energy of condensing particles

is in the range 5–25 eV and therefore not quite enough for subplantation, defects are greatly

reduced and growth occurs on the surface with high adatom mobility, leading to dense, often

well-textured ﬁlms. The kinetic and potential energies of the condensing particles assist in the

growth process and, to some limited degree, those particle energies are equivalent to substrate

heating. This allows us to obtain ﬁlms structures and properties at relatively low substrate

temperature otherwise associated with higher temperatures. This is very important especially

when a substrate is used that cannot tolerate higher temperature because it would anneal,

decompose or otherwise be damaged.

Let us have a closer look at the energy contributions brought to the surface by the particles.

First, there is the kinetic energy associated with the ﬂow velocity. Cathodic arc ions have a

considerable natural kinetic energy gained at the cathode spot, as mentioned in Section 10.2.2,

= m

/2 (10.29)

where v

can be approximately determined by the expression (10.25). The surface of a

substrate is rarely exactly at the plasma potential, and therefore a sheath will form at the

502 Chapter 10

surface accommodating the potential difference V

. In most cases, the surface will be

negative with respect to the plasma potential, and the ions gain kinetic energy according to

E

i,kin

= QeV

(10.30)

where Q is the charge state number (see Table 10.2). This point is very important because

biasing of the substrate can be utilized to control the kinetic energy of arriving ions.

In one extreme case, when the negative bias is high, perhaps reaching the kV level, ions can be

used to sputter the substrate and to produce buried layers and mixed interfaces. These effects

are considered in the ﬁeld of plasma immersion ion implantation and deposition [146].

In another extreme case, when positive bias is applied to the substrate, large electron currents

are taken from the plasma, which leads to heating of the surface. This can be intentionally

utilized or may be considered detrimental in cases where the objective was only to decelerate

ions.

When ions arrive at the surface, they carry not only kinetic energy but also potential energy.

The latter is usually neglected in the treatment of conventional PVD or CVD methods because

the condensing species are not ionized. In cathodic arc deposition, however, the fraction of

ions is very large, and so is the effect of potential energy. The largest contribution to potential

energy is the ionization energy, E

ion

. Strictly speaking, since we generally deal with multiply

charged ions, it is the cumulative ionization energy,

ion

= E

sum

Q+

Q−1





(10.31)

where the ionization energy E

is deﬁned as the energy needed to remove a bound electron

from an ion of charge state Q, forming an ion of charge state Q +1[147, 148]. Therefore, the

greatest effect is from the more highly charged ions. The ionization energy should be reduced

by the work function times the charge state number to account for the uptake of electrons from

the solid.

There are other forms of potential energy, too. For example, the arriving ion may have a bound

electron in an excited state, or if molecules are present, rotational and vibrational energy could

be considered. Furthermore, as the condensing particle becomes bonded to the substrate, the

cohesive energy is released. Therefore, the total energy brought to the surface by an ion may

be written as

total

(

)

= E

kin,0

+ QeV

sheath

+ E

ion

− Qeφ + E

+ E

else

(10.32)

where E

else

summarizes all other small contributions including the energy gained by image

charge acceleration.

Unﬁltered and Filtered Cathodic Arc Deposition 503

A numerical evaluation of (10.32) shows that the kinetic energy readily exceeds the

displacement energy in most cases, and that the contribution of potential energy is substantial,

too, especially for multiply charged ions. Such large energy input could be described as atomic

scale heating (ASH) [149, 150]. Considering ﬁlm growth based on the arrival of large ﬂuxes of

condensable (ﬁlm-forming) ions, each atom of the ﬁlm was subject to ASH several times,

namely, once when it arrived, and again when neighboring atoms arrived. Therefore, ASH is

equivalent – to some degree – to conventional heating, resulting in dense ﬁlms via enhanced

surface mobility at generally low bulk temperature [149]. ASH eventually gives rise to

temperature elevation of the substrate and the growing ﬁlm as a whole [141].

10.5.2 Stress Generation and Relief

The effect of the high-energy input is signiﬁcant for the properties of the resulting ﬁlms. At

high kinetic energy, facilitated by high negative bias, many processes occur, including the

sputtering of surface atoms (ion etching), formation of short collision cascades as they are

usually known from ion implantation [146, 151], and related rearranging of atoms and their

bonds. Depending on the energy level and other factors, this allows us to inﬂuence the stress of

the material, and in particular to produce thicker coatings at relatively low stress.

Following Bilek and co-workers [152–154], let us consider subsurface processes at high and

moderate energies (those terms should always be seen in relation to the material’s

displacement energy, which lies generally between 20 and 40 eV). As the energy of arriving

ions exceeds the displacement energy, they are inserted into the subsurface region where they

cause densiﬁcation and compressive stress. If the material is a metal it will yield to high stress

by plastic deformation. However, if the material is covalently bonded it will longer resist

plastic deformation. The material is metastable and tends to prefer a high coordination of

bonds if more than one phase exists. For example, carbon will preferentially establish sp

bonds, diamond bonds, as opposed to the sp

bonds of the less dense graphitic material.

The evolution of stress can be understood by a competition of stress generation by

subplantation (insertion) and atomic-scale annealing in a thermal spike volume [155]. For a

thin ﬁlm the stress σ is related to the strain ε by [156]

σ =

1 − ν

ε (10.33)

where Y and ν are the elastic (Young’s) modulus and Poisson’s ratio of the coating material.

According to Davis [155],

σ(E

) ∝=

1 − ν

1/2



J/j +κE

5/3



(10.34)

504 Chapter 10

where E

is the energy of arriving ions, J is the net condensing ﬂux, j is the ion bombarding

ﬂux, and κ is a material-dependent parameter. The more recent model by Bilek and McKenzie

[152, 153] gives an improved ﬁt to experimental data. Stress is determined by whether or not

atoms have time to rearrange their bonding structure within the thermal spike volume. At low

energy (e.g. ∼100 eV), the energy is removed quickly from the spike volume, i.e. the quench

time is very short and the material densiﬁed, hence the compressive stress increases. At higher

energies (e.g. ∼1 keV or greater), the atoms in the heated volume can rearrange, the volume

expands toward the surface, and hence stress is relieved.

This concept can be quantiﬁed, and because of its practical importance to cathodic arc

deposition, it is brieﬂy summarized here. One assumes that the region of the thermal spike is

spherical, with radius R

. Then the quench time is proportional to the square of the thermal

spike radius or to the 2/3-power of the ion energy E

[157],

quench

∝ R

∝ E

2/3

(10.35)

For E

∼ 100 eV, the quench time is less than one picosecond, which is too short for

signiﬁcant rearrangement of atoms. The atoms will use a high bond coordination with the

available neighbors to minimize the total system energy under constraints. The volume of the

thermal spike is proportional to κ

(

− E

)

(

2/3

)

πR

, where E

is the energy barrier for

the ion to be inserted into the surface. The volume of the light gray region (Figure 10.19) can

Figure 10.19: Geometry for a model of stress generation and relaxation: the thermal spike volume

is assumed to be spherical with the inserted atom in the center. The stress is accumulated in the

light gray region, the size of which is proportional to E

2/3

. (Adapted from [152].)

Unﬁltered and Filtered Cathodic Arc Deposition 505

be calculated as

stressed

= 2 πR

dτ = 2π



2π



2/3

dτ (10.36)

where dτ is the thickness of the region where atom movement is restrained by bonds to

unaffected atoms in the bulk. The thickness dτ is about one bond length regardless of the

thermal spike volume or ion energy. The number of stressed sites is proportional to the volume

stressed

and

σ = C(E

− E

)

2/3

(10.37)

where C and E

are material constants.

When the kinetic ion energy clearly exceeds 100 eV, the effect on stress is quite different

because the impacting ion creates a much larger volume of a thermal spike. The dark gray

zone of Figure 10.19 is a transient melt where atoms are able to access many possible bonding

conﬁgurations. Those atoms remain in a thermodynamically preferred conﬁguration, i.e. one

that minimizes bond and strain energy. The thermal spike volume can expand and raise the

level of the free surface. The possibility of raising the surface, thereby minimizing the strain

energy without constraint, is the basis for ion-induced stress relief by energetic ion

bombardment.

Experimentally it has been observed that only a small fraction of high energy ions is needed to

relieve stress, and that saturation of such stress relief occurs when the percentage of the

energetic ions is still small. Saturation of stress relief occurs when the volume regions of

unconstrained rearrangement start to overlap. The volume V still untreated by a thermal spike

is reduced exponentially [152] as the surface is bombarded with sufﬁciently energetic ions,

= exp



−

spike



(10.38)

where V

is the volume of a ﬁlm slab under consideration, V

spike

= E

(

)

is the volume of

each thermal spike, E

is the average energy given to atoms of the thermal spike volume,

N = n

Vδ is the number of high-energy impacts before the slab is covered by more

condensing atoms, n

is the atom density of the slab, δ = wf is the duty cycle of high-energy

ions, produced by pulsed biasing, for example, with w being the pulse width of the bias and f

the bias pulse repetition frequency. Alternatively, δ could be the fraction of energetic ions

produced by other means such as an external ion source. While these considerations were

developed with the plasma immersion technique [146] in mind, the underlying physics and

derivations are rather general. Using the above expressions, Eq. (10.38) becomes

= exp



−



(10.39)

506 Chapter 10

The stress can be assumed to vary with the volume fraction treated by high-energy ions, from

the as-deposited value, σ

, to the relaxed, residual value after ion treatment, σ

res

, leading to

σ = σ

+ σ

res

(10.40)

Using (10.39) gives

σ = σ

exp



−



+ σ

res

(10.41)

We can use Eq. (10.37) to express the stress generated by ions with energy E

, inserted into or

just below the surface, and replace the stress-relieving ion with energy E

= E

, one obtains

σ = C

(

− E

)

exp



−



+ σ

res

(10.42)

where the exponent χ could be 2/3, as suggested by (10.37), but other exponents such as χ =1

or χ = 1/2 also give a good ﬁt [152] to experimental data for a number of materials such as TiN

[158], AlN [155],BN[159], a-SiH [160], SiBCN [161], and hard amorphous carbon [162].

The different exponents can be interpreted to deviations from the idealized spherical shape of

the thermal spike; for example, one could consider a cylindrical geometry of the thermal spike

volume [163].

10.5.3 Preferred Orientation and Adhesion

The energy ﬂux brought by the condensing ions can have a profound effect on nucleation,

growth, and the resulting ﬁlm microstructure. In 1983, Krohn et al. [164] found epitaxial

relations of ultrathin gold islands deposited by pulsed cathodic arc on NaCl at 150

◦

C, namely

(001)Au||(001) NaCl and [100]Au||[100]NaCl. The islands were smaller than those obtained

by evaporation and coalescence occurred at smaller nominal ﬁlm thickness.

In situ conductivity measurements of cathodic arc silver ﬁlms showed that the contribution of

kinetic energy promotes the formation of smaller, more frequent islands, as compared to

sputtering [165]. Also here, the coalescence occurred at smaller nominal ﬁlm thickness, which

can be readily detected by a sharp increase in electrical conductivity. Investigating ultrathin

titanium ﬁlms made by a pulsed cathodic arc using in situ ellipsometry, Oates et al. [166]

determined the percolation threshold for titanium on silicon to be 2.7–3.1 (±0.1 nm) using a

method by Arwin and Aspnes [167] which is based on ﬁnding the moment when the real part

of the dielectric function becomes negative.

Unﬁltered and Filtered Cathodic Arc Deposition 507

Figure 10.20: High-resolution TEM image showing the epitaxial growth of TiN on MgO(100); the

titanium arc was operated in nitrogen at 7 ×10

−2

Pa with the substrate at room temperature.

(Courtesy of J. Gerlach (experiment) and T. H

oche (TEM), both IOM Leipzig.)

Films made by energetic deposition tend to grow with a preferred orientation, where the

speciﬁcs depend on the material and ion bombardment parameters; the ﬁlm tends to prefer an

orientation that is thermodynamically stable at low levels of biaxial stress [153]. The

microstructure is directly linked to the ion-generated stress, which is a thermodynamic driving

force toward a preferred orientation thereby minimizing the system’s energy [158]. In some

cases, the preferred orientation matches the substrate’s crystal orientation and one obtains

epitaxial growth (Figure 10.20).

The adhesion of a coating is greatly affected by the chemistry of the ﬁlm and substrate, as well

as by stress. A critical factor is whether or not the ﬁlm material forms chemical bonds with the

substrate material. For example, an amorphous carbon ﬁlm made by a cathodic arc tends to

adhere well to carbide-forming substrates such as silicon or titanium. The other important

factor is stress. The related strain energy limits the thickness of the coating because it may be

greater than the interface energy: the ﬁlm cracks and delaminates. The energy related to strain

grows as the ﬁlm grows in thickness, which explains why in some cases one is able to grow a

thin ﬁlm (e.g. tens of nm) but completely fails to make thick ﬁlms (e.g. ∼1 ␮m or thicker).

10.5.4 Metal Ion Etching

A special case of processing occurs when a high negative bias is applied in either DC or pulsed

mode with a high duty cycle because the result can be a net removal of material. Sometimes

this is called metal ion etching – a sputtering process caused by the impact of energetic metal

ions. This can indeed be an important ﬁrst process step in the formation of a well-adherent

coating. Since metal ions are deposited on or under the surface, we deal with a combined

deposition and etching situation where the etching rate exceeds the deposition rate, hence the

508 Chapter 10

net removal by sputtering. For the process to work, the energy of ions needs to be about 1 keV

or larger in order to obtain a sputtering yield that exceeds unity. Energy-dependent sputtering

yields can be readily calculated for arbitrary material combinations using a Monte Carlo code

such as TRIM or SRIM [168].

Metal ion etching by sputtering with typical voltages of −1000 V or slightly greater can be

beneﬁcial compared to sputter etching with argon ions because some argon will remain below

the surface and cause a defective, weak interface to the coating that is subsequently deposited.

In contrast, the interface can be beneﬁcially engineered when metal ions remain in the

thus-sputtered cleaned surface. For example, cleaning of stainless steel with chromium ions

results in a surface layer with enhanced chromium content. Chromium can become part of the

substrate because, according to the equilibrium phase diagram [169], it has unlimited

solubility in Fe below 512

◦

C. Towards the coating, chromium can readily form bonds to other

metals or to nitrogen or oxygen.

In the 1970s, Aksenov et al. [170] used this cleaning effect obtained by metal ion etching. The

cleaning of parts via energetic metal ions was used in the early 1980s, for example Bergman

applied Ti ions etching with a bias of −1000 V prior to the cathodic arc deposition of TiN

[171]. One should keep in mind that a bias of −1000 V translates into energies of at least

1 keV, 2 keV, and 3 keV, as the ions have a CSD including Q = 1, 2 and 3 (see Eq. 10.30).

The idea of replacing argon etching by metal ion etching prior to sputter deposition was later

utilized in a patented process called arc bond sputtering (ABS), by M

unz et al. [172]. This

process was optimized using steered arcs for etching and unbalanced magnetrons for sputter

deposition of coatings with very high critical loads [173–175].

10.6 Deposition of Coatings Using Unﬁltered Cathodic Arc Plasma

10.6.1 Hard Coatings

10.6.1.1 Binary Coatings

Perhaps the most important coating made by (unﬁltered) cathodic arcs is TiN because it

combines decorative gold color with excellent wear and corrosion properties. The original

widespread use of arc-deposited TiN coatings was in the former USSR and later, since the

1980s, also in Western Europe and the USA [171, 176–180]. Large batch coaters were

developed to mainly deposit TiN at elevated temperatures (typically 450

◦

C) on cutting tools

(drills, cutting inserts, etc.).

Other binary nitrides such as CrN, ZrN, and HfN [181, 182] have advantages over TiN in some

applications. For example, CrN exhibits a lower coefﬁcient of friction than TiN [181], and ZrN

is superior to TiN for cutting of titanium alloys [183].

Unﬁltered and Filtered Cathodic Arc Deposition 509

For high-speed cutting, where the coating of the cutting edge is subject to very high

temperature (∼1000

◦

C) in an oxidizing environment, TiN fails owing to oxidation. There was

(and still is) a need for more advanced materials.

10.6.1.2 Ternary Coatings

By the late 1980s, Ti

1−x

N, sometimes labeled (TiAl)N, and often simply abbreviated as

TiAlN, emerged as a superior tool coating because it has a better oxidation resistance and

hardness at high temperature compared to TiN [184]. Even though its wear properties at high

temperature are superior, TiAlN has not completely replaced TiN because of its grayish color:

the decorative character of the coating remains important, especially in the marketing phase of

a product.

As with other ternary compounds, the ratio of the constituents can be tuned to affect the

desired microstructure and properties. In the case of TiAlN, Ti and Al can be obtained from

one TiAl alloy cathode, which is convenient and usually done. However, one may opt to utilize

two separate cathodes to ﬁne-tune the Ti–Al ratio in the ﬁlm. In case of the alloy cathode, little

can be done to adjust the Ti–Al ratio in the ﬁlm, which is usually not exactly the same as the

cathode owing to preferential sputtering from the substrate in the energetic deposition process.

The substrates are slightly biased in the ﬁlm growth phase, and especially the multiply charged

ions have enough energy to cause sputtering of the growing ﬁlm. The yields of sputtering Al

and Ti are not exactly equal, and hence this preferential sputtering shifts the composition of

the ﬁlm [185].

The as-deposited ﬁlms show a dense and columnar microstructure for various Ti

1−x

cathodes. When the aluminum content is x ≤0.66 one ﬁnds a metastable cubic phase but for

higher Al content, x = 0.74, a second hexagonal phase appears [186]. Higher Al content

promoted the (200) orientation and had a large inﬂuence on the hardness of the as-deposited

coatings with up to 32 GPa. The TiAlN ﬁlms were stable with respect to phase composition

and grain size when annealed at 900

◦

C; however, when the temperature was increased to

1100

◦

C, ﬁlms deposited from the 67 at.% Al cathodes showed phase separation forming

c-TiN and h-AlN via spinodal decomposition [187] to c-TiN and c-AlN [186, 188]. The

formation of two or more phases increases the strength of the material by impeding the motion

of dislocations (spinodal hardening). Therefore, the nanostructure that forms at high

temperature increases the hardness of the coating and contributes to the superior performance

of such a coating under load [189].H

orling et al. [186] argued that softening by residual stress

relaxation through lattice defect annihilation is approximately balanced by hardening by the

formation of a coherent nanocomposite structure of c-TiN and c-AlN domains by spinodal

decomposition, and therefore the hardness is approximately constant upon annealing.

The above conclusion was realized using a systematic set of single cathodes containing

different ratios of the two metals. More ﬂexibility in terms of research and development is

510 Chapter 10

given by systems with different cathodes such as Ti, Al, and Cr cathodes, leading to coatings

in the Ti–Al–N, Ti–Cr–N, and Cr–Al–N range, and eventually even in the quaternary system

where all three metals are used in the compound [190]. The Ti–Cr–N system showed only one

phase, c-(TiCr)N, as long as the Cr content was low (10 at.%) [191]. Two phases, cubic

(TiCr)N and hexagonal β-(CrTi)

N, appeared with higher Cr content (17–58 at.%). The Cr

content can be adjusted not only by the Cr arc current but also by biasing: the Cr content

decreased with increasing substrate bias due to preferential sputtering of Cr. The two-phase

coatings were nanocrystalline and showed high hardness with a maximum in the range

3700–3900 Vickers at a chromium contents of about 25–30 at.% and a load of 0.5 N [191].

Other examples of arc-deposited ternary nitrides are Ti

0.94

0.06

N and Ti

0.92

0.08

N, which

have been deposited on cemented carbide inserts using random and steered arc sources [192].

These coatings showed less ﬂank and crater wear than conventional TiN and CVD-deposited

TiC–TiN coatings.

Ternary compounds do not necessarily need two metals; rather, the process gas can be the

source of two components, and the cathode delivers only one. Carbo-nitrides are prominent

examples where changes in the gas composition between N

and CH

can be used to obtain

TiC

1−x

, for example, with x ranging from zero (TiN) to unity (TiC) [193, 194]. The

composition is determined not only by the gas supply variations but also by substrate bias. One

should keep in mind that bias is also a means of supplying energy, and hence the substrate

temperature increases. In this example, bias of −400 V led to a temperature of 550

◦

Cofthe

cemented carbide inserts (WC-6 wt.% Co). The resulting ﬁlms were dense and single-phased

with a NaCl-type structure, highly stressed (up to −5.9 GPa) and superhard (up to 45 GPa).

10.6.1.3 Quaternary Coatings

The idea of using the gas as a source of two components can readily be applied to ternary

coating systems, like TiCrN, to obtain quaternary compounds like TiCr(C,N). It should be

emphasized that those multielement coatings tend to decompose into different phases, and in

some cases one should speak of nanocomposites. The Ti–Cr–C–N system produces a

nanocomposite of nanocrystalline TiCr(C,N) in an amorphous carbon (a-C) matrix when the

carbon content is in the range 9–27 at.% [195]. This is not surprising because the carbon

concentration exceeds the solubility of carbon in TiCrN. The nanocomposite exhibits higher

hardness (about 30 GPa) compared to monolithic TiCrN (about 26 GPa).

Quaternary coatings can also be obtained from several cathodes or corresponding alloy

cathodes. For example, the Ti–Al–N system can be expanded by adding Cr with the idea that

protective chromium oxide might form on the surface (similar to stainless steel). As a

side-effect, it was found that cathodes with several elements tend to produce fewer

macroparticles [196]. The orientation of ﬁlms of (Ti

)N, with x:y ≈1:2, z = 0.63–0.73,

x + y + z = 1, changed from hexagonal to cubic when the negative substrate bias was increased