4.

10 Chemical Vapour
Deposition (CVD)
The principle of CVD was known in the nineteenth century, but it was not
1940s and 1950s that interest in it as a untilthe
method of
coating materials commenced. Initial studies at the producing
hard and refractory
the United States Battelle Memorial Institute n
demonstrated that CVD could be used to
variety of metals and non-metals, synthesise a broad
nitrides and oxides including their compounds like borides, carbides
(Campbell et al,
2013). Such studies were conducted1949; Powell, 1966; Pierson, 1992; Kafizas et a
on a
passing appropriate gases over very small scale and merely
i
heated wires inductively heated carbon substrates or resisia
(usually tungsten). Such work
reduction of metal halides (centring on the endothermic e
advances. A list of some of by hydrogen) fundamentally underpinned later
the CVD
reactions with useful na ntials
given in Table 4.10, whilst engineering Po this way
are crystalline and
compared to those made via PVD amorphous deposits produced
CVD reaction and (Section 4.111) in Table 4.11.
tion deposition rates increase reac
temperature, obeying exponentially witn
always) two basic reactions: Arrhenius-type relation. There are no hutno
an
creasing
tnot

secondary substrate reaction. (1) a primary main gas phase somead )3

on, and (2)
The following rea
relationships
ships are typical:
typica
Fm AmeXp (Equation43)

Fs Asexp. (Equation4 4

where Fm and F, are

the
coating deposit mass
reactions respectively: Qm s u b s t r a l e

substrate reactions and Q, are increases due to tne

ses the maii
and
m a i na n d

the
for the main respectively; and Am and activation
and substrate A, are
rgies for the
energies m a s si n c r e a s

reactions. Further, T initial constaCuon emperature

is the
absolute reac tem
 Surface Engineering with Deposition Technologies 155

f Coatings deposited by CVD (without plasma assistance). After Yee 1978; Pierson,
et al, 2013.
1997; Kafizas
Table4.10

B e s m a n
eettal,
n
.

1992;

Typical deposition Typical deposition

temperature (°C) Coating Precursor system temperature (°C)
Coating Precursor system

Borides
Carbides 930-1030 TiB2 TiCl4-BCl-H2 830-1030
TICl-CH4-H2

930-1030 MoB MoCly-BBr-H2 1430-1630

TIC HfCk-CH-H2

930-1030 WB WCl-BBr-H2 1430-1630

HfC ZrCly-CH4-H2

930
ZrC 930-1230
ZrBr-CH4-H2

930-1230 NbB2 NbCl5-BCl3-H2

CHSiCl-H TaBrs-BBr3-H2 1230-1630
SiC
1230-1430 TaB2
BCly-CH4-H2 ZrCl-BCIl-H2 1030-1530
B,C
460-500 ZrB2
WF-CH-H2
HfCl-BClH2 1030-1630
WC
CC CrClh-CH4-H
1030-1280 HfB
1030-1280
Cr(CO)-CH4-H2
CrC2 1030-1280 Silicides
TaCls-CH4-H2 TiCl4-SiCl4-H2 1230-1530
996
TaC 1030-1280 TiSi
VCl-CH4-H2 MoCl5-SiCl-H2 >930
VC 1530-1930 MoSi2
NbCl-CH4-H2
NbC Sulphides
180-350
Nitrides 930-1030 MoS2 Mo(CO)HS
TIC4-N2-2 (or NH,) WCO)%-H,S 300
TIN 930-1030 WS2
HfCl, N-H2
HAN 1030-1430 Oxides
SiCl-NH3-H2
SigN4 330-430 700-1200 (see text)
SiH4-N-H2 a-Al2Os AICl-CO2
1030-1430
BN BCl3-NH-H2 K-AlLOs or
430-730
B3N H-Ar Y Al2O3 AICl-H,O
1030-1330
BaH-NH-H2 [HS may be
added as a

catalyst]
900-1100
ZrCly-H,O
1130-1230 monoclinic
ZrN ZrCl-Na-H2 ZrO2
>830
ZrBr-NH-H2
830-15300
TaN TaCly-N2-H2
830-1230
AIN AICly-NH-H2
930-11300
Al(CH3)-NH3-H2
VN VClL-N-H 930-1230
NDN NbCl5-Na-H2 930-13300

QRT 1S
whilst the quotient
and R is the gas constant (8.314 J/K), growth of
to the
d eaction may positively
contribute

ionless. The substrate of TiC coatings

the growth
the of carbon to
(Sectio
coating, as in the substrate supply substrate
reactions. For example,

Section 4.10.1). However, there

one of
can be malign
W (tungsten)
is (Bryant,
1977):
the used to prepare
possible main reactions (Equation 4.5)
6HF
WF%+3H2 W + react with the
can in principle
c o r r o s i v e and
Here,t reaction product HFis very reaction is (Bryant,
1977):
[Link] Case
substrate, a possible
ofa Ta
adverse

(Equation 4.6)
+24H2
SHF+ Ta+ TaFs
158 Introduction to Surface Engineering

TaFs is a weak solid with a low melting point (-97°C). Fortunately,

energy change for the Eq. 4.5 reaction is higher than for the reaction in
Hence, there is insufficient driving force for the formation of TaF. M 6
there are no adverse kinetic mechanisms that promote Eq, 4.6 in favour of Ea.4
On the other hand, when depositing W2C onto steel via a reaction analo
Eq. 4.5. the reaction product HF reacts vigorously with the substrates, rec
them to be prior protected with electrolytic or electroles Ni platings
requirir
Yee, 1978).
(Archer &
CVD reactions can be carried out at atmospheric (100kPa) or sub-atmosnhe
pressure. often in the range of 0.1 to 6.6kPa (Kafizas et al, 2013). To be of m
use. the kinetics and thermodynamics of the main CVD reactions should haomem
at temperatures capable of producing coatings of useful thickness (-5-15um n
reasonable time (-5-6 hours). Although the temperatures initially investigated for
many CVD systems exceeded 1000°C, later development demonstrated that this
was not always necessary (Table 4.10). Hence, much effort was made from the
19805 onwards to discover lower temperature reaction systems (Sections 4.10.4 and
4.10.5) and to deploy various types of plasma assistance (Section 4.10.6); also see
Haubner. 2013.
A schematic representation of a commercial CVD unit, used for the high
volume production of coated cemented carbide cutting tools, is shown in Fig 424
Such CVD systems utilise a hot wall design. On the other hand, cold wall reactors
appear more suited for niche coating production, as for the deposition of metal
borides such as TiB2 (Voigt & Westphal, 1981). Process cycle times (including
heating and cooling steps) can be as long as 12 hours. Hence, it is common to use

H2 CH N2 Ha, etc.

External
furnace

Substrates

Liquid
ring
pump Water
Supply

TICI
evaporator HCI
scrubber

Figure 4.24 Set-up for conventional elevated temperature CVD at sub-atmospher

 Surface Engineering wth Deposition Technologies 159

than one reaction vessel for better efficiency; i.e., one vessel can be occupied
more
whilst another is cooling. It is also sensible to reserve at least one
deposition wh
CVD
for
ction vessel for each coating compositional design (e.g., one for TicTiN and
CIAlO3 and so on) to avoid possible cross-contamination. With
for TiC/
a n o t h e r

tinual use, significant

use, significant build-up of coating material takes place on the internal

surfa of CVDvessels, which necessitates periodical disassembly and cleaning (via

blasting and the like); this avoids unwanted transfer of fine particulate debris to
beadblasting
surfaces iuring the processing cycle.
substrate
the
A virtue of CVD is that it has excellent "throwing power'; i.e., it is a non-line-of.
sieht process (Bryant, 197E Garg et al, 1988). This makes it possible to apply coat
inside surfaces of small diameter (<Imm) holes, for example. However, it
on the
ings in practice to determine the limit of the 'coatable' hole length to
isoften necessary
diameter ratio for the range of operating CVD conditions being used, which will vary
with spatial position within a given CVD reactor design. As with other coating
methods, it is important that substrate surfaces are free from oils and greases
from
residual machining or grinding operations (Chapter 2), since these can
prior
adversely affect the coating adherence, purity and uniformity. The charge is then
loaded onto graphite or nickel alloy trays, which are stacked vertically (Fig 4.24).
The balance of the process gas is typically made up of hydrogen, which acts as both a
reductant and a diluent. Dry hydrogen is normally flushed through the system from
the commencement of the CVD cycle, preventing oxidation of the vessel and the
and removing
support 'furniture' during heating, whilst chemically reducing
con-

taminant oxides on the surfaces of the components. Once the desired temperature is
reached (Table 4.10), the reactants are introduced into the chamber. Many CVD
processes generate residual corrosive gases like HF or HCI (Eqs. 4.7 to 415). which
are disposed of by passing through appropriate chemical or cryogenic traps. For
SOme sub-atmospheric pressure CVD processes, a liquid ring pump is commonly
used to attain the desired processing pressure; this also aids the dissolution o
Lanted process gases prior to final neutralisation and disposal (Fig. 424).

4.10.1 CVD of TiC

1950s, was that of titanium carbide
(C:the first applications to be tried, during the
1953). carried out at -1000°C in
lCe uon on steel (Münster &Ruppert, CVD-coated cemented carbide
nydrogen. The first commercially produced
metal 1969. Although initially used by
itsele ng tools appeared on the market in about for multi-phased,
ula 4,25a), TiC was later used as a foundation' coating and TiN/
Alayered CVD coatines like TiNTi(C,N)/TiC (Fig. 4.25b), Al,0/TiCMT-CVD
Fig. 4,26). This role was subsequently (circa 1995)) replaced by
OyTiC (Fig
TiC
TiCN)- see Section 4.10.4.
The main
TiC deposition reaction is:
TiC4 +CHa(+H2)> TIC+4HCi (+H;) (Equation 4.7)
160 Introduction to Surface Engineering

(a)
TIC coating (a)

Cementeed
carbide
substrate

10um

(b) TiN coating

-Ti(C,N) coating
TiC coating

Cemented
carbide
substrate

20um

(c)

-TiN coating
Ti(C,N) coating
-TIC coating

Cemented
carbide
substrate
10um
.

Figure 4.25 Light-optical

CVD-coated micrographs of polished and etched micro-section taken throuph
cemented carbides.
(a) First-generation TiC-coated carbide
(b) Second-generation TiN/Ti(C.N)/TiC-coated carbide
(c) As (b) but deeply etched in an HF/HC/HNO/H,O solution to reveal the columna
morphology of the exterior TiN layer

avail-
whilst the accompanying substrate reaction
able) is:
(provided that excess carbo
(Equation 48)
TiCl4+2H2 +C TiC+ 4HCI
For the main reaction
(Eqs. 4.3 & 4.7), the activation energy for the
mined (Hara et al, 1977)
to be about 370kJ/mol compared with 110kJ/mioni ction

substrate reaction (Eqs. 4.4 &

4.8), indicating
that the latter
 Surface Engineering with Deposition Technologies 161

AlLOg coating
AlOg/TiC coating interface

Coarse equiaxed TiC coating

Fine equiaxed grained TiC coating

Coating/substrate interface
2um
Cemented carbide
substrate

TiN Coating

C-Al2O3 Coating

TiC coating

10um

figure 4.26 Micro-sections of CVD coatings:

racture surface of first generation Al,O3/TiC-coated carbide, SEM (Dearnley, 1980)
Polished and etched late generation multi-phase TiN/AlLO/TiC (multi-layer) coating
gradated cemented carbide substrate, LOM (Courtesy of K J A Brookes,
international Carbide Data, 1996.)

getically casier. At the commencement of the process, the sum of both reac
tions
s n a deposition rate of -1-2m per hour for CVD carried out in the
Tange and 1060*'C. The substrate reaction is rate-controlling during nuclea-
tion and U
ion the diffusion path for
initial grow
owth, but as the coating becomes thicker,
carbonsugh the TiC lengthens and the substrate carbon supply diminishes.
Hence
ence the action (Eq. 4.7) gradually becomes predominant and the growth
main react
approaches a lowe
lower limit of about 1um per hour. In practice, the carbon
Content of the substrate must
substrate mus be exactly known and the substrate carbon loss
carburisation via Eq. 4.8) allowed for. For cemented carbide substrates,
162 Introduction to Surface Engineering9

Figure 4.27 Cross-section through a Ti.

ce
carbide cutting insert revealing eta-phase ented
at the coating/substrate interface. Light-oaTowe
graph, etched in 10% potassiunm ferricyanide s micro
ution.

20um

excess decarburisation (refer to the vertical section taken through the ternary
wCCo phase diagram shown in Appendix A) can result in the formation of
n-phase (Co3W,C) - a very brittle phase at the coat/substrate interface (Fig. 42
This can, when of sufficient thickness, aid exfoliation or spalling of the adjacent TiC
coating during use in its intended application - as for example in metal cutting

operations like milling or interrupted turning, which is further worsened under cuting
conditions that produce built-up-edge formation (see Section 7.6.1 in Chapter 1,
Detrimental n-phase formation due to decarburisation can be avoided by carburising
the substrate prior to the commencement of CVD or by using substrates with opti-
mised uncombined carbon contents (adjusted during the carburising of W powder
substrate reaction
in WC manufacture). Both routes supply sufficient carbon for the
(Eq. 4.8) without n-phase formation taking place.

4.10.2 CVD of TiN

10n
The deposition of TiN proceeds in an analogous way to the main deposition reau
for TiC:
(Equation 49)
2TiCl4 + 4H2 + N2 2TiN +8HCI
4.3)
The kinetics of TiN deposition also obey an Arrhenius relationsnptio deposition

and, at about 1000°C, with optimised gaseous flow rates, results in a s i t i o

Faster depOS
per hour in many commercial
perations.
rate of about 1um d h e r e n t

y
rates are possible, but this eventually leads to micro-porosity and poo 2013
al,
coatings. Sometimes, ammonia (NH,) is used in place of N2 (Kanzaserte v e rt e m

allowing an efficient deposition reaction to take place at the siig former

o ft h e
y
activityo*
perature of -850"C,made possible by
the greater chemical
compound.
 Surface Engineering with Deposition Technologies 163

can applied to substrates pre-coated with TiC, Ti(C, N) or Al2O3

be applied
TiIN
TIN
4 & 4.26). Here, coatings of Ti(C, N) are simply produced by
(Figs. 4.25
Na and CH4 in the presence of H2. In cases where TiN is reacting
TICL4 with
TiCl deposited
directly
onto carbon-bearing substrates, substrate reaction,
a
analogous to Eq. 4.8,
in the formation of a discrete layer of Ti(C,N), usually <0.5um
formation
results

the
substrate via: thick, directly
adjacent
ked
ed) 2TiCl4+N2 +2C+ 4H2 2Ti(C, N) +8HCI
TO (Equation 4.10)
1on. The compounds T1C, T1(C, N) and l1N are isostructural and
han oe in lattice being observed between the mutually soluble, a
parameter
ence in the case of
triplex CVD coating material extreme composi-
rtN (Fig. 4.25b), a gradual "blending' ot nitrogen and designs, like TiC/Ti(C, N)/
dhrouieh the thickness of the layer, resulting in a smooth carbon content is possible
microstructure and properties like microhardness. progression in composition,

rnary
on 4.10.3 CVD of AlLO.
 Surface Engineering with Deposition Technologies 173
44 Physical Vapour Deposition (PVD)
4.11.1 Early Background

grown in ttechnological importance since its early beginnings in the nineteenth

PVD has
u to its mature status in the twenty-first. It was known for many years that metal
century
coatings could be applied to glass substrates by placing them in a vacuum system and
tino a metal source beneath them. This process is termed 'vacuum evapora-
tion
nd was the means of making mirrors, initially by vaporising silver and latterly
minium. This was probably the first type of PVD process used on a large industrial
ale The process required the objects to be held under high vacuum (-10Pa) whilst
scale.

aporation took place. Vacuum evaporation is still carried out in the twenty-first
century and may be used for creating a wide range of metal and/or ceramic coatings.
The evaporation process is highly line of sight' (poor throwing power), as depicted in
Fig, 4.35a, this means it is not possible to completely cover a surface with the evapo-
rated material, without rotating it with respect to the source position.
Another early type of PVD coating technique was sputter coating. The origins of
the process are not entirely clear. The physical effect of sputtering was first reported
by Grove (1852), who observed that a prior oxidised metal surface could be 'beauti
faully cleared off'- that is, sputter cleaned -by a discharge created in hydrogen at low
pressure (approximately 10'Pa). This was before the invention of vapour or diffusion
pumps. Subsequently, Wright (1877), observing phenomena inside sealed glass
vacuum tubes containing a much lower pressure (-4Pa) of air (obtained with the
then recently invented mercury vapour pump - refer to the review of Redhead,

(a) (b)
Air pressure -10*Pa Argon pressure -0.4Pa

Negative
Substrate DC bias
Coating Substrate

1ttt Direction of
vapour
flux

Evaporant
Heat Heat Positive
ground

Dlac
place during:Schematic depiction comparing the difference in the vapour pathways that take
(a) Simple evaporation
(6) lon plating
The glow discharge
ge plasma surrounding the substrate in (b) is not shown for better clarity. The
wing power of ionplating is superior, providing greater coating coverage.
ased on a
drawing by Talivaldis Spalvins (Spalvins, 1978).
 6

woy6 Aysuep 6upeoo

 Surface Engineering with Deposition Technologies 175

Company (USA auSA), which named it Ivadising' (Muehlberger, 1986). Typically such
partsreq
ire a final chromating treatment to provide additional corrosion resistance
andt oa i d
adherence of any subsequently applied organic paints.
Based [Link] of ion plating, it was discovered that sputter-deposited
coatingm a t e r i a .
aaterials could also be made more adherent and could be delivered with
fewer roblems when a glow discharge plasma is made to surround the
of-sight prob
sputter
As with ion plating, this is achieved by applying a
parts
Huring deposition.
current or radio frequency-induced negative bias voltage to the
negative d i r e

during the «coating cycle. Both ion plating and the latter form of sputter
bstrates
subst

bias sputtering) can be regarded as being plasma-assisted PVD or

deposition

PAPV processes (Fancey & Matthews, 1991). Nowadays, the term PVD' is gen-
1sed, whether or not plasma assistance is involved in the process concerned.
erally used,
Eor both categories of PVD process, the substrates must be thoroughly cleaned
aCombination of degreasing and particulate removal methods (Chapter 2)
rior to mounting within the vacuum chamber. For the best coating adhesion, the
aihstrates should always be subjected to an initial intensive ionic bombardment
-

see Fig. 3.19 in Chapter 3) by surrounding them with a

(sputter cleaning/etching
hieh-power density glow discharge plasma, for 20 to 60 minutes, prior to commen
The latter step is carried out using a much lower substrate
cing coating deposition.
to avoid unnecessary coating loss through ionic bombardment.
plasma power density
The purpose of the former stage (sometimes supplemented with radiant heating) is
the removal of surface-bonded (adsorbed) water vapour and any thin (nanometre
scale) passive oxide films (Fig. 2.2). Once cleaned in this way, excellent chemical
bonding is made between the substrate surface and the arriving coating material fux
(delivered from the sputter or evaporation sources), even at low (-200°C) substrate
temperatures. Various methods of assessing coating adhesion exist (Table 4.12). Of

Table 4.12 Coating adhesion (adherence) test methods.

Test method Suitable coating/substrate Reference/standard

indentation with a 'Rockwell C diamond PVD/CVD coatings, VDI (1992)

thermal spray coatings
Pull-off, peel, spiral bend test, tensile or Thermal spray coatings on metals Dearnley (1999);
shear and alloys Dennis & Such
(1993)
ratch tests Burnett& Rickerby
using macro-, micro- and/or Electroplatings, PVD/CVD (1988); Bull &
nano-indentation with diamond coatings
Indenters Rickerby (1990)
3-or Thermal spray and electroplated Tsui & Clyne (1996)
4-point bend tests
coatings on metals and alloys
Fracture initiation and Weiss et al (1996)
coatinal on and propagation
prc along Thermal spray and electroplated
Coating/substrate interface coatings on metals and alloys
Fracture initiation by 10-indentation Electroplatings, PVD/CVD Saunders (1993)
1Olowed by
fracture
propagation along coatings
Coating/substrate interface
 S

6x/ (peo) sseuw yoejos jpouO

 Surface Engineering with Deposition Technologies 177

Figure 4.38 Reactive sputter

deposited TiN coating on a
cemented carbide metal
cut
ting tool substrate. Light-opti-
cal
micrograph. Polished
section etched in
potassium
ferricyanide solution. A thin
bond layer of unreacted Ti is
visible at the
coating/substrate
interface (arrowed) - no
phase is present. eta
10um

-800 volts

Ti target

Plasma
N+ Cathode Figure 4.39 Schematic diagram showingg
the principle of
O TO simple reactive sputter
TiN molecule deposition of TiN (without any magnetic
field assistance). Note that this
requires a
higher applied voltage than for magne-
TiN coating tron sputtering.
Based on an
original drawing by John
Substrate Thornton.

these
materials, coatings produced in this way can only be produced at low deposi-
Tion rates.
Fortunately, much higher deposition rates are possible if such coatings are
synthesised reactively. In the
example of TiN deposition, this means evaporating or
puttering metallic Ti into a plasma containing Ar and N2 (the reactive element).
ureatly simplified, this can be
expressed by the equation:
2Ti
(vapour or sputter flux) +N22TiN (Equation 4.16)
The
nese PVD methods were termed reactive ion plating and reactive sputtering
Typicale PVD processes). A schematic of the latter method is shown in Fig. 4.39.
deposition rates using simple reactive sputter deposition are only fractions of
tim tres per hour compared to CVD (-1um/hour); these demand process cycle
mes >12
was the case for producing thecoated cemented carbide shown
hours. This was
178 Introduction to Surface Engineering

in Fig. 4.38. The earliest attempts to prepare TiN and TiC coatings via t
PVD route of reactive ion plating were similarly frustrated. Here alternative
[Link]. (-1660°C) and vaporisation temperature of Ti (compared t e high

of Ti could
be vaporised
[Link]. 660°C) meant only small quantities aluminium:
electrical resistance heating (small diameter Ti wires/fla
(duris ion
plating) using
used as the source). Both limitations were overcome (circa 197 nts were
1970 to 1980) bytWo
technological advances:
1. High-energy vapour sources for reactive ion plating

2. Magnetron cathode sources for reactive sputtering

These were rapidly implemented into large-scale reactive PVD nrod
uction
equipment.
Magnetron Sputter Sources
From the early work of Kay (1963) and Gill and Kay (1965), it was known th
ions within the visible negative glow and Crooke's dark space of a glow diseha
plasma (Fig. 3.18) are created by collisions between electrons and neutral a
sharge
or molecules. Importantly, Kay (1963) had also shown that an order of magnitude
increase in ionisation and ion current density (to -2-3mA/cm) was possible y
confining the electrons (within a glow discharge) by an externally applied mag
netic field. Such confinement increased the number of electron-atom collision
events, and it enhanced the ion bombardment (sputter erosion) of the cathode
(the sputter target). Consequently, the deposition of sputtered (coating) material
on an adjacent substrate was increased -compared to identicalexperimentsmade
without the use of an external magnetic field. These observations led to the
development of the magnetron cathode, one early design being described by
hapin, 1974, which was subsequently shown to attain sputter coating rates
almost an order of magnitude higher than those possible using non-magnetron
not-
sputtering techniques (Table 4.13). Whilst many hours (>8) are required for
magnetron reactive sputter deposition of TiN (Rickerby & Newbury, 198) and
other nitride compounds, only -1-2 hours are necessary for magnetron-assiste
sputter deposition of the same materials (Sproul& Rothsetin, 1985; Sprouleta
1989).

Table 4.13 Some approximate comparisons ofcoating deposition rates

for non-magnetron (conventional (non-magnetron) and magnetron
sputtering technologies all data approximate). Compounds prepared
by reactive deposition.

Coating Type Sputter deposition rates (um/hour)

Conventional Magnetron
TiN 0.9 12.6
TiC 1.0 14.4
Cr 0.5 28.8
 Surface Engineering with Deposition Technologies 179

Evaporation SourceS

t h e efficient high-rate evaporation of high melting point materials like

To enand c
chromium, tu
h r o m i u m , two main devices were developed (Matthews, 1985):
and
titanium
electron
ron ibeam (EB) source (Chambers & Carmichael, 1971)
The
a.
electric a r c
(EA) source (Sablev, 1974)
The
b.
Although a thiro hird alternative vapour source the hollow cathode (Vossen & Kern,
1991) - can also be used to meet the same demand, presently
Vossen & Kern, 199
1978;
it
(circa 2016) it is quite unpopular, despite its simplicity of construction and ease of
operation a n d m a i n t e n a n c e .

A popular design of EB source used for ion plating is shown in Fig. 4.40. To
operateefficiently, this type of EB must be placed in its own separate (differentially
chamber (devised by Chambers & Carmichael, 1971) and
umped) low-pressure

DC
power
supply
0.5-5kV

Cathode
Gas inlet
Glow
discharge Electron emitter
Primary
pumps
(tungsten filament)
Auxiliary Components
anode
Dark space
AC power
supply
MtMt M+
Vapour
Low Dividing
Voltage wall
C
Supply
Evaporant
Electron
beam gun

Secondary pumps

Figure 4.40 Simplified schema natic diagram showing the principle of triode reactive ion plating.
Electrons
metal iorom the emitter are dragged across the rising vapour stream, causing an increasein the
The glow discharge plasma surrounding the cathodically charged parts
Based enhan
becomesonConcentration.
a ed or *supported', requiring less potential to sustain a given current density.

carlier an original drawing by Allan Matthews (Matthews & Teer, 1981), following an
ideaby Wan et al (1974).
180 Introduction to Surface Engineering

Electron beam
(deflected through 270)

Vapour flux
Pressure
Evaporant (0.4Pa

Pressure
Dividing wall 104Pa
EBgun Aperture

Electromagnet
Figure 4.41 Detailed schematic showing the main features of a conventional electron beam
evaporation source deployed in PVD ion plating.

Evaporant
target plate
High-power Vapour flux Figure 4.42 Detailed schematic
electric arcs showing the main features ofa
conventional high power density
30 to -100 electric arc evaporation soure
volts
deployed in PVD ion plating.
Pressure
0.4Pa
Isolation
Conical anode

maintained at -10 Pa, compared to -0.4Pa in the main process chamber.

chamber. Otherwise
the EB filament (made from
tungsten) will become poisoned by water aole of
reactive process gases,
renderingit inoperable. The EB is steered throughe the solid
a
270° via a small aperture,
c h a m b e r
using complex electromagnet system, ou
evaporant material placed in a
crucible in the adjacent ion tech
mare
(Fig. 4.41). This design has obvious limitations, not the least beingpla
tha
s y s t e m containing
(Pig.442)
cally difficult (though not impossible) to build a
vacuum syster
than one EB source. The
same pressure
alternative device, the electric arc (EA)the sourc
does not suffer this limitation, as it can be safely operated at tne Casin
as that to around the substrates
requiredsustain a glow ced in any

-
discharge plasma
recent
Fig. 4.40) i.e., -0.2 to 0.4Pa. Here,
mor
several EA sources can De a f
um vessel. The early and blev, 1974
orientation around the internal walls of a
vacuum
developments Martin
of the EA sources
have been well
Matthews, 1985; Bergman, 1986; Martin documentea &Mar
& Bendavid, 2001; Benda
 Surface Engineering with Deposition Technologies 181

10um

Figure 4.43 SEM image showing

local erosion of a Ti arc-source tar-
get plate. Solidified Ti droplets
(arrows) are visible surrounding
the rim of the
vaporised crater.
After Bergman, 1986; ©. SM
International, 1986.

Figure 4.44 Commercial line-of-sight

arc-source synthesised PVD TiN coat-
ing on a bio-medical stainless steel
substrate.
Note: Macro-particles (arrows) and
shallow scars (S) remaining after
macro-particles have been removed.

10um

2014). In a simple EA design, a circular cathodically charged target material (e.g. Ti

or Cr) is arranged such that it intercepts the path of high-power density arcs (-10° to
10°A/cm; Matthews, 1985) originating from a surrounding conically placed anode
eg 442). Once initiated by a short circuiting device, like a Mo wire trigger, the arcs
are self-sustaining and continuously "bombard' the target surface. In a time period of

y afew milliseconds, the tip of the arc melts and evaporates a small volume of
arget material required for the reactive process (e.g.,as in Eq. 4.16). A smal
T of liquid droplets are also produced at the same time; these can be seen
n g arc melted craters on the target surface (Fig. 4.43). Commonly, how-

the number of liquid droplets have sufficient kinetic energy to strike

thegniicant or
Macrate, resulting in near-spherical coating defects termed 'macro-particles'
are a concern, as
strate. hese so
Moreover, Some they increase the surface roughness of the coated
are quite loosely bound to the surface and once removed,
lcave
scars (Figs. 4.44 & 4.45). Other macros, especially those produced
sing Jasoinall
vaporation, are hollow; following impact with a substrate surface, these
Tesemble ring dou -
1987. To minimise
4.45) Dearnley and Anderson,
has been made to prevent particle/substrate impacts. The
such "defects', much effort(Fig.
simplest s to pass the evaporated flux through a 90-degree elbow/conduit