Department of Chemical Engineering Materials Environment

Electrical, Materials and Nanotechnology Engineering (EMNE)

(XXX cycle)

Ph.D. Thesis

Modified aluminide coatings

with reactive element
for turbine blades protection

Virgilio Genova

Supervisor: Prof. Cecilia Bartuli

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ……….. 7

Chapter 1 – Gas turbojet…………………………………………………………………8

1.1 Principle of operation
1.2 Operating conditions
1.3 Desing of a turbojet
1.4 Ni-based super alloy.
1.5 Degradation phenomena
References…………………………………………………………………………………25

Chapter 2 – High temperature coatings………………………………………………..26

2.1 Coating requirements
2.2 Protection systems for harsh environments
2.2.1 Thermal barrier coating (TBCs)
2.2.2 Bond coat: state of art
2.3 Diffusion aluminide coatings
2.4 Platinum modified aluminide coatings
2.5 Modified aluminide coatings with reactive elements
2.5.1 State of the art
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . ..41

Chapter 3 – Processes for production of diffusion coatings…………………………..43

3.1 Coatings technique
3.2 Growing mechanisms of diffusion coatings
3.2.1 High temperature low activity process
3.2.2 Low temperature high activity process
3.3 Pack cementation
3.3.1 Above the pack cementation
3.4 Slurry aluminization
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Chapter 4 – Design of experimental activities…………………………………………54
4.1 Phase 1: oxidation and hot corrosion resistance
4.1.1 Standard aluminide caotings
4.1.2 Isothermal oxidation
4.1.3 Hot corrosion tests
4.1.4 Modified aluminide coatings with Zr
4.2 Phase 2: Electroless nickel plating
4.2.1 Principle of deposition
4.2.1 Plating solution compositions
4.2.3 Plating solution parameters
4.2.4 Selected plating solution parameters
4.2.5 Nanocomposite (Ni-nAl2O3)
4.3 Phase 3:Slurry aluminization of electroless pure nickel coatings
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78

Chapter 5 – Experimental results………………………………………………………..80

5.1 Phase 1: Oxidation and hot corrosion resistance

5.1.1 Samples obtained with NH4F as activator salt

5.1.2 Samples obtained with AlF3 as activator salt

5.1.3 Isothermal oxidation
5.1.4 Oxidation kinetic
5.1.5 Hot corrosion
5.1.6 Modified aluminide coating with Zr
5.1.7 Isothermal oxidation
5.1.8 Oxidation kinetic
5.1.9 Hot corrosion
5.2 Phase 2: Electroless pure nickel plating
5.2.1 Solution A
5.2.2 Solution B

5.2.3 Nanocomposite (Ni-nAl2O3)

5.3 Phase 3: Slurry aluminization of electroless pure nickel samples
5.2.1 Plating solution A
 5.2.2 Plating solution B
5.2.3 Surface characterization
5.2.3 Wettability

5.4 Electroless pure nickel plating with n-Al2O3

5.5 Slurry aluminization
5.4.1 Surface characterization
5.4.2 Preliminary oxidation tests
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .156
Lyst of acronyms VPA= Vapor Phase Aluminizing
XRD = X-Ray Diffraction
ATR-FTIR = Attenuated Total Reflectance
Fourier Transform Infrared Spectroscopy List of symbols
CVD = Chemical Vapor Deposition
DST = Dynamic Segregation Theory PR = plating ratio
René 108DS = Directionally Solidified wt % = weight percentage
EDS = Energy Dispersive Spectroscopy T = temperature
EDTA = Ethylendiaminotethracetic acid h = hours
ENL = External Nickel Layer x = mass or thickness of oxide
EPNP = Electroless Pure Nickel Plating kP = parabolic contant
FE-SEM = Field Emission gun Scanning (g) = gas phase
Electron Microscopy (l) = liquid phase
GDOES = Glow Discharge Optical Emission (s) = solid phase
Spectroscopy ∆M = mass variation
HPC = High Pressure Combustor
HPT = High Pressure Turbine
HTHC = High Temperature Hot Corrosion
HTLA = High Temperature Low Activity
HRTEM = High Resolution Transmission
Electron Microscopy
IDZ = Interdiffusion Zone
IPT = Intermediate Presurre Turbine
LPC = Low Pressure Combustor
LPT = Low Pressure Turbine
LTHA = Low Temperature High Activity
LTHC = Low Temperature Hot Corrosion
PN_A_ = Pure Nickel_solution A_
PN_Al_ = Pure Nickel_Aluminum_
RE = Reactive Element
SEAD = Selected Area Diffraction
TBC = Thermal Barrier Coating
TGO = Thermally Grown Oxide
INTRODUCTION
Increasing the maximum temperature of the gases introduced into the turbine increases both the
efficiency and performance of a turbo-gas system; Switching from an input temperature of 900 ° C
to 1250 ° C can result in a 30% increase in output power, leaving unchanged consumption. In order
to ensure the structural integrity of components working in the hottest areas of the engine, research
over the last decades has focused on the development of increasingly innovative technologies that
aim to increase the efficiency of the system while at the same time increasing its operating life.

Turbine blades are one of the most critical components of this system because they not only work in
a highly aggressive environment, but are also subject to strong mechanical stresses related to the
rotation of turbochargers and aerodynamic loads. In addition to the presence of oxidation and hot
corrosion mechanisms, high temperatures contribute to the degradation of the structural properties of
the component and result in thermo-mechanical stress related to the temperature gradients and to the
different coefficients of thermal expansion of the used materials.

In those applications where operating temperatures are compatible with the thermomechanical
properties of the materials used, the turbine blades are coated to provide protection against high
temperature oxidation and hot corrosion phenomena. If the operating temperatures are too high,
multiple solutions are taken to reduce the temperature experienced by the component; the most
important are: internal and external cooling techniques and the use of special coating systems (TBCs
- Thermal Barrier Coating).
TBC coatings are a multilayer system consisting of a top ceramic layer (top coat) and an intermediate
layer (bond coat) designed to protect the component from high operating temperatures and oxidation,
respectively.

The aim of this work is to increase the performance of turbojet engines used in the aeronautical field,
by appropriately modifying the bond coat formation processes. In order to achieve this, at the first
part of this work, the change of processes already used in the industrial field was proceeded, while in
the second part of the work an innovative and alternative method was proposed and studied to
improve these processes.

7
CHAPTER 1

Gas Turbojet

Gas turbines have been used since 1791[1] in a wide variety of applications including aircraft
propulsion, marine propulsion and power generation. Components of these engines experience very
high stresses and temperatures in highly corrosive and oxidizing gases, and are required to maintain
their mechanical properties for thousands of hours of service.

1.1 Principle of operation

The turbojet is a turbogas-based aeronautical propulsion. Figure 1.1 shows a section of a simple
turbojet, where the various stages of the engine are illustrated.

Figure 1.1: Cross-section of a simple turbojet

The dynamic intake moves air from outside and drives it into a compressor, where it slows down
and compresses. The compressed air then enters in the combustion chamber at a pressure of about
30 ÷ 35 bar and at a temperature around 550 ÷ 625 ˚C. Combustion takes place here, with fuel
injection in the compressed air stream. At this point, hot combustion gases are forced into the
turbine, expanding them and having the role of providing the power needed to move the
compressor. At the exit of this stage the gas is still at a higher pressure than the ambient
temperature, and can therefore be accelerated in a nozzle. In this way, the propulsion fluid is ejected
at a higher speed than the one with which it enters the engine, thus generating the thrust.

The excellent properties in terms of thrust, specific consumption, thrust / weight ratio, have
imposed this family of propulsion as the most widely used in civil aviation as well as military.
8
The jet engine based on the gas turbine cycle was first patented by Frank Whittle in Great Britain in
1930, but only in 1939 was the first aircraft propelled by a turbojet. After the end of the Second
World War it was established thanks to its high specific thrust (per unit weight of the engine) and
the low-power specific at high speeds.

The simple turbojet is extremely efficient at high subsonic flight speeds. To extend the field of
application of the turbojets and to improve their efficiency throughout the subsonic field, the
turbofan was introduced, as shown in Figure 1.2

Figure 1.2: Turbofan Pratt & Whitney PW4000

1.2 Operating condition

Figure 1.3 shows a section of the Rolls-Royce Trent 800 reaction engine (Boeing 777 aircraft engine),
and illustrates the various phases and pressure and temperature profiles along the engine, while Table
1.1 summarizes the operating conditions of each section:

9
Figure 1.3: Section of the Rolls-Royce Trent 800 engine with temperature and pressure profile for the different
stages[2]

Table 1.1: Operating parameters for Rolls-Royce Trent 800

Pressure Temperature Rotation

Components Material
[bar] [°C] speed [rpm]

FAN Titanium 1 810 3500

LPC Titanium 8 290 6800

HPC Titanium 37 600 10200

COMBUSTOR Ni-alloy 35 1500 -

HPT Ni-alloy 35 1400 10200

LPT Ni-alloy 14 900 6800

EXHAUST Ni-alloy 6 860 3500

10
As can be seen from Table 1.1, the operating temperatures to which the various components are
subjected are several, and this directly affects the choice of materials in the different engine sections
(Figure 1.4)

Figure 1.4: Materials used in a Rolls-Royce Trent 800

Finally, Table 1.1 also shows how high pressure turbine blades (HPTs) are the components
operating in the most severe conditions in terms of temperature and pressure. They also experience
high tension due to centrifugal forces, constant over time and rapid thermal transients during the
flight cycle.

1.3 Design of a turbojet

The criterion on which the choice of materials is based is, once resistance specifications have been
achieved, the performance improvement in terms of internal speed and temperature reached by the
gas, while maintaining the entire lightweight system.
As mentioned in Section 1.2, the turbine is the most critical part of an aeronautical engine; For this
reason, over the years, research has focused particularly on the study of materials that meet certain
thermal and mechanical resistance requirements at high temperatures. From here on, attention will be
focused on turbine blades, as it is the resistance to the thermal and mechanical stresses of these
components that discriminates the maximum permissible temperature in the combustion chamber and
consequently the performance of the entire engine.
From a mechanical point of view, the main factors to be taken into account during design are creep
and thermal and mechanical fatigue, as well as the resistance requirement to the tested loads. The first

11
two phenomena mainly affect the first stage of turbines, while for the second stage the limiting factor
is mainly creep.
In addition, the hot gases that invest the blades are highly oxidizing, and contain contaminants such
as chlorides and sulphates, which can cause hot corrosion of the component; these gases may also
contain erosive elements [3]. Erosion can be caused by the sand during take-off and landing.
Figure 1.5 shows the temperature profile of a HPT blade. Considering the gas temperature increases
at the turbine entry these values are now considerably higher, although the temperature profile
essentially remains the same.

Figure 1.5: Temperature profile for a blade of High Pressure Turbine (HPT)[4]

Then, a turbine blade must have the following characteristics:

1. High mechanical resistance at high temperatures;
2. Resistance to creep, thermal fatigue and mechanical fatigue;
3. High microstructural stability;
4. Resistance to oxidation and hot corrosion;
5. Resistance to erosion.
In practice, it is not easy to collect all of these properties in a single material, and that is why
superficial coatings that can meet the characteristics of the substrate are used in this kind of system.

12
1.4 Ni-based superalloy.

The word superalloy means a group of alloys developed for high performance applications at high
temperatures, such as aircraft turbine engines [5].
As reported by Sims and Hegel [6], superalloy is defined as "an alloy developed for high temperature
applications, generally based on elements of Group VIII-A, which have high mechanical strength and
surface stability."
The superalloys can generally be divided into three classes, according to the main alloy element:
1. Nickel-based (Ni)
2. Cobalt-based (Co)
3. Iron-based (Fe)
Ni-based superalloy is the most commonly used for turbine blades manufacturing.
Figure 1.6 describes the turbine working temperature trend as a function of the materials that have
been used for the realization of the turbines over the last sixty years; in particular, the monocrystalline
Ni-based superalloys have been greatly developed over the last few years due to the growing demand
for materials capable of delivering high mechanical performance at high temperatures. These alloys
have a very high mechanical strength at high temperatures, allowing theoretical achievement of
temperatures up to 1300°C without any damage. The high thermomechanical resistance is mainly due
to the absence of discontinuities originating from grain boundaries and precipitate concentration sites,
which in the polycrystalline superalloys represent the triggering and propagation zones of the cracks.
Table 1.2 shows the characteristic temperatures at the compressor output and at the turbine intake
from 1955 to the present.

Figure 1.6: Evolution of high temperature superalloys capacity for a period of 60 years after their appearance in 1940.

13
Table 1.2: Operating temperatures in the last 60 years

Temperature [°C] Temperature [°C]

Service year
Outlet compressor Inlet turbine

1955 379 771

1965 427 938

1975 593 1343

1995 693 1427

2008 >700 >1500

2015 766 1760

Thanks to the availability of materials to withstand the increasing turbine gas entry temperatures,
aircraft engines quadrupled (from the first Henkel turbine in 1939) their push / weight ratio with a
specific consumption of approximately halved. Figure 1.7 shows the three types of superalloys used
for the production of turbine blades and the improvements recorded according to the life time of
each type of material as previously stated, having conventionally associated polycrystalline alloys
with a value of 1.

Polycristalline Directionally Monocristalline

solidified

Polycristalline
Directionally solidified
Monocristalline
Cycles

Creep Thermomechanical Corrosion

Resistance fatigue resistance
Resistance

Figure 1.7: Features of three different superalloy

14
Ni-based super-alloys offer high performance:
1. High melting temperature
2. High resistance, which includes tensile strenght, creep and fatigue resistance
3. Ductility
4. Toughness
5. High temperature corrosion resistance
However, nickel does not have these properties, since pure metals, such as nickel, are generally too
ductile and do not have sufficient resistance to the harsh environments, both mechanically and
chemically. Therefore, these are rarely used without being reinforced by various mechanisms.
Nickel-based superalloys are generally reinforced by precipitation hardening or solid-solution
strengthening, both of which are effective as reinforcing methods at high temperatures.

Precipitation hardening
In the precipitation hardening, soluble atoms participate in the creation of a second fine phase and
highly dispersed. Traditionally, two thermal treatments are used for Ni-based superalloys. The first
is the thermal treatment to homogenize the microstructure and reduce the effects of elemental
segregation. The second is the aging to develop phase precipitates g'- [Ni3(Al, Ti)]. These precipitates,
consistent with the matrix, and extremely stable in temperature, act as a barrier to the movement of
the dislocations, greatly increasing the resistance to creep. Elements such as Al and Ti promote the
creation of phase g', by alloying with the phase g-Ni. The size of this phase can be precisely controlled
by means of thermal treatments. Generally, this type of hardening is used in turbine pallets.

Solid solution strenghtening

In this type of hardening, solute atoms (such as Al) randomly substitute solvent atoms (Ni, in this
case) without altering the crystalline structure.
The continuous phase, characterized by a center-faced cubic lattice (cfc), forms the matrix of Ni-
based alloys; this is hardening by the solid solution with several randomly distributed atomic species.
Various metallic elements (Co, Cr, Mo, Fe, Ta, W, Re) are dissolved in the matrix by randomly
replacing Ni atoms. Due to their different sizes than Ni, the solid solution causes local stresses that
prevent the movement of the dislocations, interacting with the stress field caused by the dislocations
themselves. In addition, solute atoms may also go to the interstices between the atoms of the matrix
by anchoring the grain boundaries and thereby improving the sliding and migration resistance that
would occur during the creep diffusion.
15
It is preferred to use this type of hardening in applications where high mechanical strength is required.

1.5 Degradation phenomena. Hot corrosion and oxidation at high temperature

Turbine blades, especially those in high pressure conditions (HPT), are subjected to high stresses and
temperatures; according to the operating conditions, can trigger creep, hot corrosion, high
temperature oxidation and surface erosion as mentioned in Section 1.4. All these aspects must
necessarily be taken into consideration during the design process to ensure proper operation and
useful life of the component. Figure 1.8 shows the complexity of the elemental composition of a Ni-
based superalloy for aeronautical applications [7]. The complexity, in terms of chemical
compositions, don’t allow to obtain a material that can have some high mechanical properties with a
high resistance to harsh environments.

Figure 1.8: Alloy elements present in Ni-based superalloys. The elements with lesser benefits are marked with a cross-
hatch, while the harmful elements for the performance of the material are marked with a horizontal hatch [8].

For this reason, it is simpler and more functional to use surface coatings. The idea of applying
protective layers on the surface of turbine blades was practiced already in 1960 and found extensive
applications. Since then, protective coatings are conventionally used for aeronautical engines and
ground turbines. Corrosion phenomena can be divided into two types: uniform corrosion and
localized corrosion.
Localized corrosion is particularly present at the root of the blade, exposed to temperatures around
620 to 760 ° C. These areas of damage can then expand and reach to the center of the component
causing the failure. There is a close relationship between the damage and the characteristics of
stresses and the environment: the higher the velocity of the flow that invests the blade, the less the
16
corrosion damage; the lower the quality of the fuel used and the more severe will be the corrosion;
the longer the operating time, the more penetration depth of the corrosion products will be.
The uniform corrosion can be classified according to three different phenomena (Figure 1.11): high
temperature oxidation (T> 1000 ° C); hot corrosion at high temperatures (Type I: T = 850 to 950 °
C); hot corrosion at low temperatures (Type II: T = 650 ÷ 800 ° C) [9].

Figure 1.9: Corrosion rate vs Temperature [10]

High temperature oxidation is a phenomenon where metals and alloys exposed to oxygen or a gas
containing oxygen, at high temperatures, convert a part or all of the metals into their corresponding
oxide.
The oxide, if it is attached to the substrate, can give rise to a degree of protection and slow down the
oxidation kinetics, or may become detached, exposing the underlying substrate to oxidation; In
addition, for long exposure times, the phase responsible for selective oxidation can be exhausted in
the substrate and thus initiate the oxidation of a different element / phase with formation of a
smaller scale oxide, resulting in a sensible acceleration of the oxidative phenomenon.

The theory of Pilling-Bedworth (1923) tried to explain why some metals form a protective scale and
others do not; this theory argues that the ability to protect the scale depends on the ratio between the
molar volumes of the metal and its oxide. A general indicator of the protection of the oxide scale is
given by the Pilling-Bedworth ratio (PBR) defined as:
For:

17
• PBR <1: the volume of oxide formed is less than the volume of metal consumed; porous, cracked
and non-protective scale are formed;
• PBR> 1: layers of protective oxides are formed;
• PBR >> 2: the volume of oxide is much greater than the volume of the metal; so, the scale is very
compressed.
However, this theory is not always verified, but there are some exceptions due to approximations
that are adopted; these relate to the mechanism of oxide growth, geometry, the volatility of the
oxide and the chemical effects of alloy elements. However, in general, it is a qualitative indication.

A metal is oxidized since the free energy variation ∆G0, associated with the reaction, is negative.
∆G is influenced by a number of parameters such as:
• Temperature
• Partial oxygen pressure
• Composition
The effect on the ∆G0 of the temperature and partial pressure of O2 at the variation of the oxidized
element is given in the Ellingham diagram (Figure 1.10):

18
Figure 1.10: Ellingham’s diagram

Figure 1.10 shows that noble metals are at the top of the diagram, indicating that they are less reactive.
The Ellingham diagram allows:
1. to order oxides according to their greater or less metal reduction reaction,
[Link] determine the partial pressure of oxygen in equilibrium with a metal at a certain temperature
3. to determine the degree of CO / CO2 or H2 / H2O that cause the reduction of a metal oxide at a
given temperature.

19
However, Ellingham's diagram does not allow predicting the rate of oxidation reactions and therefore
does not account for kinetic aspects.

It has been experimentally observed that the growth of the oxide layer follows one of the following
laws:
• Linear Growth: 𝑥 = 𝑘% 𝑡
()
• Parabolic Growth: 𝑥 ' = 𝑡
'

• Logarithmic Growth: 𝑥 = 𝑘* log(𝑎𝑡 + 1)

where x is the thickness of the oxide layer after exposure time t, kL, kP and ke are growth rate indices,
temperature functions T, and a is a constant.

Linear growth
If the metal surface is not protected by an oxide scale, the oxidation rate remains constant over time.
In this growth mechanism is the reaction between metal and oxygen to control the degree of oxidation
at the interface between oxide and metal scales, while the species transport process is not influential.
The oxidation kinetics is described by:

𝑑𝑥
= 𝑘%
𝑑𝑡

𝑥 = 𝑘% 𝑡

with x mass or thickness of oxide, considered zero at the initial moment (Figure 1.11).

Figure 1.11: Linear oxidation kinetic

20
Oxidation never slows down. For sufficiently long exposure times the metal is consumed
completely.
As mentioned, this phenomenon occurs when the RPB ratio is less than 1, or if the oxide is volatile
or liquid.

Parabolic Growth
Many systems oxidize by following a parabolic pattern. Oxidation in these cases is thermally
activated and the rate of growth is controlled by ion diffusion through the oxide layer, with driving
force caused by a chemical potential gradient. The growth rate is inversely proportional to the
thickness, and kinetics are described by:
𝑑𝑥 1
= 𝑘4
𝑑𝑡 𝑥
𝑘4
𝑥' = 𝑡
2
with x mass or thickness of oxide, considered zero at the initial moment (Figure 1.12).

Figure 1.12: Parabolic oxidation kinetic

As the thickness of oxide increases, the diffusion distance increases and the kinetics thus slows
down.
Logarithmic growth
The oxidation rate generally follows a logarithmic or reverse logarithmic law at low temperatures
(below 400 ° C) when only a thin film (<200 nm) of oxide has formed [12]. The driving force that
allow the growth is the formation of localized electric fields in the vicinity of the metal surface;
absorbed oxygen atoms acquire electrons from metal atoms, creating large electric fields along the
thin layer between positive metal ions and negative oxygen ions. Metal atoms are driven by the
electric field through the oxide layer. The oxidation kinetics in the case of logarithmic kinetics is
described by:

21
 𝑥 = 𝑘* log(𝑎𝑡 + 1)
in the case, instead, of inverse logarithmic kinetics we have:
1
= 𝑏 − 𝑘8 log 𝑡
𝑥
with x mass or thickness of oxide, considered zero at the initial moment (Figure 1.13).

Figure 1.13: Logarithmic and inverse logarithmic oxidation kinetic

The metals that oxidize with this kinetics reach a thickness of oxide which oxidation apparently
stops.

When metals and alloys, in a high temperature oxidizing environment, have the surface covered
with a thin film of salts, accelerated oxidation can occur. This phenomenon is referred to as hot
corrosion, and affects the ability to protect the substrate by the oxide scale. This type of attack is
sensitive to a different number of variables such as composition and amount of deposit, gas
composition, temperature, surface erosion, composition and alloy microstructure.
It has been shown that the aggressiveness of this type of attack, which can be catastrophic, consists
of two phases:
1. Initial Stage: The alloy and the deposit are subject to alteration, such as the reduction of the
element favoring the formation of the protective oxide scale, the incorporation of a deposit
component into the alloy, the dissolution of oxides in the salts, the formation of cracks and / or
channels in the scale. During this phase the corrosion rate is slow and similar to that in the absence
of the deposit;
2. Propagation Phase: The salts enter through the damaged scale and propagate to the scale-alloy
interface. When the deposit reaches low oxygen sites and meets the depleted zone of element alloy
that create a protective oxide scale (such as Cr and Al), propagation occurs quickly.

22
Generally, corrosion that occur above the melting point of the salts is called type I hot-corrosion,
while the one that occurs below the melting point of the individual constituents of the salts, unless
they form eutectic, is called type II hot corrosion.

High Temperature Hot Corrosion (HTHC) Type II

The HTHC begins with the condensation of alkaline metal salts melted on the surface of the
component. A number of subsequent chemical reactions cause the attack of the protective oxide scale
with an acceleration of the oxidation phenomenon and the formation of a porous oxide scale.
Mainly, during the combustion, the salt formed is Na2SO4 starting from Na and S present in the
atmosphere and in the fuel; this salt has a high thermodynamic stability. Other impurities contained
in the fuel or atmosphere such as vanadium, phosphorus, lead and chlorides can be combined with
sodium sulphate and form a mixture of salts with low melting temperature, thus expanding the area
undergoing corrosive attack.
Potassium sulphate K2SO4 behaves in a similar way to sodium sulphate.
Vanadium is an inevitable contaminant in some fuels, especially in marine ones. In addition to its low
melting point, vanadium compounds greatly increase the solubility of the oxide when mixed with
Na2SO4.

This type of corrosion can generally be divided into four progressive stages, from initial start, up to
component failure:
1. At the initial stage, a slight increase in surface roughness is observed, due to the growth and
localized fracture of the oxide layer. However, neither the chromium impoverishment in the substrate
nor the mechanical integrity loss of the component is observed;
2. At this stage, the roughness of the surface increases due to continuous breaks of the oxide layer.
Chromium in the substrate begins to run out, but mechanical integrity is not affected yet;
3. The oxidation of the material reaches a considerable depth; Mechanical integrity is compromised
and blades must be replaced. Also, at this stage, corrosion is able to proceed independently of the
presence of sodium;
4. At this stage, the catastrophic attack of the material occurs. The failure of the component becomes
very likely due to the loss of structural material.
From a macroscopic point of view, this type of corrosion is characterized by the material's peeling
and significant changes in color (greenish tone resulting from the formation of NiO). Microscopically,
morphology is characterized by the presence of sulphates and depleted regions under the porous, non-
protective scale.
23
Figure 1.14 shows the components used in aeronautical field coated with alumina (Figure 1.14a)),
subject to HTHC, which reveals the characteristics of the attack. In Figure 1.14b), sulfur attack is
evident, which play an important role in the degradation process, while Figure 1.14 (c) is shown the
penetration of the coating and exposure of the substrate to the attachment.

Figure 1.14: Micrographs of the microstructural evolution during hot corrosio – a) Ni-base aluminized HTHC; b-c)
cross-section microstructure, respectively, for the coating and the substrate [11]

Low Temperature Hot Corrosion (LTHC) Type I

LTHC generally occurs below the melting point of pure salts, but may also occur above the melting
temperature of the salt itself if the deposits form a eutectic with a melting point far lower than that
of the individual constituents. This is the case of the Na2SO4 and V2O5 compound, characterized by
a melting temperature significantly lower than the individual compounds (Figure 1.25).

Figure 1.15: Phase diagram of Na2SO4 and V2O5

24
Similarly, nickel-based superalloys form a eutectic Na2SO4-NiSO4. A high partial pressure of SO3 in
the gas phase is required to proceed with the reactions, as opposed to what happens in the HTHC. In
addition, compared to Type I hot corrosion, no incubation period is observed in this type, nor is the
sulphate presence and the depletion of reactive elements (Cr and Al) is generally observed. The LTHC
typically causes pitting of the material (Figure 1.16).

Figure 1.16: Type II hot corrosion for a CoCrAlY after 4200h of test

References

[1] A.M.Y. Razak, Industrial gas turbines: Performance and operability, CRC Press.
[2] Micheal Cervenka, Rolls-Royce Corporation.
[3] Driver, D.; Hall, D.W.; Meetham, G.W, ed. G.W. Meetham, Applied Science Publishers,
1981, pp. 1-30.
[4] C. Duret-Thual, R. Morbioli, and P. Steinmetz, Luxembourg, 1986.
[5] Bradley, E.F.; ‘Superalloys, ASM International, 1988.
[6] Sims, C.T.; Hagel, W.C.; Preface, John Wiley & Sons, 1972.
[7] T. R. Pollock and S. Tin, J. Propuls. power, vol. 22, no. 2, pp. 361–374, 2006.
[8] Matthew.J. Donachie, Stephen.J. Donachie: Superalloys-A tecnical guide, 2nd Edition.
Copyright by the Materials Information Society, 2002.
[9] Felix, P.C., High Temperature Alloys for Gas Turbines 1978, ed. D Coutsouradis et al,
Applied Science Publishers, 1978, pp. 69-80.
[10] N. Eliaz et al., Engineering Failure Analysis, 9, 31-43, 2002.
[11] R. Streiff, J. De Physique IV.,Vol.3, 17-41, 1993.
[12] Briks N., G.H. Meier, F.S., 2006, Cambridge University Press.

25
CHAPTER 2

High temperature coatings

2.1 Coating requirements

Because of the extreme conditions experienced by the turbine blades, this is in the aeronautical field
or in a power generation plant (heavy duty), protection against the harsh environmental effects is
required in order to maintain the integrity of the components and to increase overall performance and
useful life of the system. Surface engineering is intended to combine the mechanical properties of the
substrate with the required surface protection.

Any coating deposited on a turbine blade must necessarily ensure protection within a specified period
of useful life from the destructive attacks due to high-temperature corrosion, oxidation and erosion;
a coating to be protective must meet the following requirements [1]:
1. Must resist corrosion, oxidation and erosion.
2. Must safely withstand the applied static and cyclic stresses on the surface of the turbine.
3. It must exhibit good microstructural stability.
4. Do not degrade the mechanical properties of the substrate.
5. Must be as compatible as possible with substrate alloy in terms of chemistry and thermal expansion
coefficient (CTE).
The choice of a suitable coating for a given application depends on the complex interaction between
surface and coating and the substrate properties.

2.2 Protection systems for harsh environments

The primary objective of a coating or surface treatment for high temperature working components is
the ability to form a thermodynamically stable, slow growth superficial oxide capable of protecting
the underlying alloy from the environment [3]. For this reason, most of the coatings contain the three
key elements: aluminum, chrome and silicon. These, in part, are added between the alloy elements in
small quantities because an excessive concentration of these elements causes the embrittlement of the
superalloy.
The main limitation to improving the efficiency of aeronautical engines is the materials with which
the turbine blades are made: in fact, the high inlet temperatures in the turbine are not compatible with

26
the thermo-mechanical properties of the materials used for blades. The superalloys also show a
dramatic decay of their mechanical properties.
In practice, therefore, it is aimed at maximizing the operating temperature of the first stages of the
turbine without affecting its mechanical properties; it is sought with appropriate methods to lower the
temperature experienced by the turbine blades without affecting the temperature of the gas leaving
the combustion chamber. This can be achieved in two different ways; the blades design includes [4]:
1. Installation of integrated cooling systems, in which the coolant is conveyed into some suitable
holes or directly into the blade surface (Figure 2.1).

Figure 2.1: Examples of cooling systems

2. Protection with antioxidant coatings and thermal barrier coatings (Thermal Barrier Coatings - TBC)
An integrated cooling system has the advantage of not causing any weight gain and not affecting the
useful life of the component; however, it has a negative effect on the overall engine yield, as it draws
fluid from the cycle to use it for non-propulsion purposes; Moreover, from a technological point of
view, the realization of this type of system entails greater complexity and cost of production.
Protective coatings have the advantages to be less expensive in terms of production costs. On the
contrary, the use of thermal barrier coating implies a weight gain and a greater risk of sudden coating
delamination caused by centrifugal forces.
However, it should be pointed out that the adoption of one and / or the other method becomes
necessary when designing more powerful engines is required.

27
2.2.1 Thermal Barrier Coatings (TBCs)

Thermal barrier coating (TBCs) are able to break down the temperature experienced by the underlying
alloy from high operating temperatures, but also to protect the turbine blades from the aggressive
environment in which it operates.
TBCs are a multilayer system made by:
1. An insulating ceramic coating often about 100 ÷ 400 µm [2], called top coat, typically yttrium-
partially stabilized zirconia (Y-PSZ)
2. A metallic bond coat between the substrate and the top coat, resistant to oxidation and corrosion,
and with the task of promoting adhesion between the thermal barrier and the substrate, while also
complying with the thermal expansion coefficients (CTE) between the substrate itself and the Y-PSZ
coating.
3. Thermally Grown Oxide (TGO) is formed at the time of deposition of the ceramic layer which is
interposed between bond coat and external ceramic coating. This layer continues to increase during
exercise, due to the permeability of oxygen through zirconium oxide.

Figure 2.2 illustrates a schematic of the three elements that make up a TBC, while Figure 2.3 shows
a section of a turbine blade covered by a thermal barrier and the temperature profile that the three
layers experience up to the substrate.

Figure 2.2: TBC systems [5]

28
Figure 2.3: Multilayer coatings schema

The top coat is the most external layer of the system and, as seen in Figure 2.3, has a thermal
shielding power of a few hundred degrees, which results in an elongation of the useful life of the
blades; Furthermore, it allows less use of cooling systems, thus reducing the design difficulties and
the realization of the cooling systems. Such coating allows to operate with higher temperature gas,
resulting in improved efficiency and performance. It also provides protection from the erosion
phenomena.

In this system, the oxidation protection is ensured by the bond coat, which is oxidized, giving a
slow-growing oxide scale (TGO) which is opposed to the further diffusion of oxygen. This scale
becomes thicker as the time of exposure to high temperatures becomes, as oxygen can easily diffuse
through the very porous structure of the overlying ceramic coating (Figure 2.4).

Oxygen

Figure 2.4: TBC microstructure

29
By their nature, top coat and substrate have a very different thermal expansion coefficient, which
results in considerable thermal stresses during service and can cause coating failure, leading to
exposure to the surface of the blade. The metallic bond coat has an intermediate thermal expansion
coefficient between the two layers, ensuring continuity between alloy and ceramic, thus reducing this
phenomenon.

2.2.2 Bond coat: state of the art

The bond coats are generally divided into two groups:

• Diffusion coatings: coatings made by diffusion of one or more elements in the surface of the metal
to be protected. These are applied by thermochemical processes, in particular Pack Cementation and
Chemical Vapor Deposition (CVD).
• Overlay coatings: coatings with a specific composition applied to the surface to be protected by
thermal spray or physical vapor deposition (PVD). This category belongs to the MCrAlY systems,
where M is usually nickel and / or cobalt.

Both coating classes are based on Al as oxidizing form a-Al2O3, characterized by excellent adhesion
and a very low permeability to oxygen.

In the first type, the surface is enriched with elements capable of forming the protective scale, while
in the second it overlaps the substrate with a predefined composition material capable of forming a
protective scale. In diffusion coatings, therefore there is a chemical reaction, which changes the
composition of the substrate; with overlay coatings simply goes to deposit a layer of a new material
on it (Figure 2.5).

Figure 2.5: Differences, in terms of microstructure, for a diffusion aluminide coating and an overlay [3]

30
The main difference between the two protection systems is the adhesion of the coating; this aspect is
fundamental: in a rotating turbine running at thousands of revolutions per minute, a leak of the coating
will cause a damage to the system. In power generation applications, this results in an implant stop,
and restoration of the naked component exposed to surface degradation phenomena. As far as
aeronautical applications are concerned, having high-speed projected elements inside the engine can
cause the engine to lock itself. Figure 2.6 describes properties and compositions of the typically used
antioxidant coatings. It is to be noted that Al and / or Cr coatings have excellent oxidation behavior
as they rapidly form a slow-growth protective oxide (Al2O3 and Cr2O3) scale; when Cr content
increases, an improvement in corrosion resistance is observed;

Figure 2.6:Compositions and properties, in terms of oxidation and corrosion resistance, for several coatings [2]

2.3 Diffusion aluminide coatings

In this type of protective coating, the surface is enriched with aluminum which, as said, is able to
form a slow-growth protective oxide scale. Depending on the surface, Al reacts chemically with the
substrate in Ni, forming a new phase called b-NiAl; it will be the aluminum present in this
intermetallic to undergo oxidation by protecting the underlying substrate. Figure 2.7 shows the result
of 2500 hours flight at low altitudes at sea level on a turbine blade. The macroscopic difference
between a non-coated blade and a protected b-NiAl layer is evident.

31
Figure 2.7: Aluminized and not coated turbine blades

In general, diffusive coatings can easily cover complex shapes and inner surfaces, which are generally
complicated to coat; there is no substrate-coating interface, so there are no adhesion problems since
the coating is contained in the component, there are no elements that can be detached. The process is
very reproducible, as it is driven by thermodynamics, and finally the apparatus is economical.
Conversely, the diffusing coating that is formed cannot be different from the substrate composition,
since from a chemical point of view the reaction takes place between the substrate and the gas
containing the element that will diffuse inside. Another negative aspect is that the substrate to be
coated should be exposed to high temperatures: superalloys undergo very complicated high
temperature thermal treatments, and the risk is that by enriching the surface at temperatures
comparable to those with which these treatments are performed, the resulting hardening may be lost
and the substrate properties may be compromised.

Stages of equilibrium of Ni-Al alloy

Ni crystallizes in a cubic face centered structure (CFC). If Ni is allied with Al, there are many changes.
Up to an approximately 4 wt% Al, there is no change in the Ni atomic structure, except for occasional
random replacement of Al atoms against Ni; This phase is called g phase and is characterized by CFC
structure (Figure 2.8a).
By increasing the content of Al, it selectively begins to replace Ni atoms at the angles of the cell,
while the atoms on the cube faces remain of Ni (Figure 2.8b). The ratio of the atoms to each cell is 1

every 3 Ni, giving the Ni3Al composition, known as g' phase.

32
If the alloying is increased up to 25 wt% Al, the crystalline structure changes, giving rise to an ordered
biatomic cubic structure. The ratio between Al and Ni atoms is 1: 1, providing the NiAl composition,
known as phase b, (Figure 2.8c).

Figure 2.8:Allotropic phases of NiAl

The Ni-Al binary phase diagram, shown in Figure 2.9, may provide useful information on the phase
that is characterized aluminide coating obtained by diffusion over Ni-based superalloys.

Figure 2.9: Ni-Al diagram phase

The field of the b phase is very wide, indicating that at this stage the concentration of Al can be
very far from the stoichiometric NiAl composition. In addition, the melting point of the b phase is
much higher than the pure Ni. These characteristics make the b phase very favorable for use in high
temperature coatings.

33
2.4 Platinum modified aluminide coatings

The deposition of an electrolytic platinum layer before the aluminization, followed by a diffusion
thermal treatment (at 1000 ÷ 1100 ° C, up to 5 hours at 10-7 torr) was initially designated to create a
barrier against the diffusion of aluminum towards the substrate, delaying the decomposition of NiAl
to [Link] analyzes have shown that Pt does not act as a diffusion barrier, and a number
of other mechanisms have been proposed to explain the effective role of Pt in increasing the
oxidation and corrosion resistance of aluminide, because this element does not participate directly
to the formation of the protective oxide scale. Figure 2.10 shows how the presence of Pt increase
the performance of an aluminide in terms of hot corrosion resistance:

Figure 2.10: Comparison of hot corrosion resistance for a simple aluminide coating, a platinum modified aluminide
coating and a CoNiCrAlY overlay coating

As mentioned, the mechanisms through which platinum acts on the bond coat properties are not
clear. For example, it has been shown that the formation of the phase PtAl2 increase the quantity of
the aluminum inside the bond coat, with the consequent increasing of the selective oxidation of the
Al [4]. This is essentially due to two factors:
1. No platinum oxide is formed at operating temperatures;
2. Al is very mobile in rich platinum phases.

34
This results in an alumina scale with increased purity and reduced growth rate. The efficacy with
which platinum performs this action is related to an optimum concentration of Pt, which in
excessive amounts would brittle the superalloy, and greatly increase the process costs.

In addition, it suppresses the precipitation of Cr rich phases inside the outer layer of the coating. In
fact, Cr in solid solution increases the activity of the Al near the surface; this benefit is less in the
presence of precipitates, which therefore results in a reduction in the oxidation resistance of the
coating. Pt also suppresses the diffusion of refractory elements such as Mo, V and W from the alloy
to the outer layer and improves the stability of the coatings against the inter-diffusion between the
substrate and coating itself, preventing the phase transformation from b to Ni-richer phases [5].
Finally, it suppresses the formation of voids at the interface, thus increasing the adhesion of the
scale to the coating [6].

2.5 Modified diffusion aluminide coating with reactive elements

It has been known for more than 70 years that the addition of small quantities of reactive elements
(RE), generally transition elements such as Hf, Zr, Y, Si, La or Ce, has beneficial effects on the
oxidation behavior of aluminide. They take the name of reactive elements because they themselves
react in the coating during oxidation.
Pfeil in 1937 [7] was the first to report the beneficial effects of the RE; Since then, the scientific
literature agrees on the best adhesion of the oxide and the reduction of the rate of growth of the
scale in Ni-based superalloys [8].
Figure 2.11 clearly shows the benefits introduced by RE on oxidation rate. The kP of a b-NiAl
coating is compared with a Y, Y2O3 and Zr doped b-NiAl.

35
Figure 2.11: RE effects on weight gain of b-NiAl doped

The actual mechanisms that take place for the RE are a very controversial topic from a scientific point
of view. In general, the mechanisms proposed were:
• REs segregate as ions along the polycrystalline oxide scale grain boundaries, significantly limiting
the transport to the outside of aluminum and thus reducing the rate of growth of the oxide. From a
combined contribution of both aluminum and oxygen in the absence of reactive elements, oxidation
is mainly governed by oxygen transport. It has been experimentally observed that the rate of growth
of the oxide scale is influenced differently depending on the element forming it: in Al2O3 there is a
reduction in the speed of about 2-4 × while in Cr2O3 of a factor 10-100 ×.

• REs segregate as ions, but are not static, but spread outward during oxidation, driven by the
Dyanamic Segregation Theory [10].

• REs occupy reticular vacations, causing decreasing of Kirkendall voids formation at bond coat /
oxide scale interface: Al and Ni interdiffuse with different speeds, causing the formation and
coalescence of voids, which reduces adhesion of the scale of oxide to the substrate [10],[11].

• REs cause the formation of oxide pegs in the substrate, which improve adhesion to the interface by
means of a real mechanical anchorage: the reactive element oxides grow by internal oxidation below
the scale. The distribution of these pegs is a function of solubility: reactive elements with higher
solubility result in the formation of more thin oxide pegs than those with limited solubility [12].

36
• The REs bind to S, inhibiting its segregation to the oxide-substrate scale interface: in fact, the
presence of S at the interface lowers the cavity formation energy, which would affect the adhesion of
the oxide scale [13].

• The REs determine strong interactions with the oxide interface: Density Functional Theory (DFT)
principles based on theoretical calculations indicate that the first transition elements, characterized
by open electronic shells, are able to determine, added to bond coat, bond-to-bond interactions to the
strongest interface.
In practice, it is believed that co-sharing of these mechanisms causes a beneficial effect on oxidation
and hot corrosion resistance [14].

2.5.1 State of the art

It is preferred to use reactive elements with a high atomic radius such as Zr and Hf, because they can
react more easily thanks to their atomic dimensions.
Guo et al [15] compared the behavior of some coatings after 100h of cyclic oxidation at 1200 ° C,
showing that among all the compositions studied, indeed, doped aluminide with Hf, Zr and an Hf-Zr
blend are those which presented a greater oxidation resistance (Figure 2.12) with a lower kP.

37
Figure 2.12: Square weight gain (sample + spalling) for NiAl with different amount of RE during 100h cycles at 1200°C
in air: (a) NiAl–0.05%Dy–0.05%Hf, NiAl–0.05% Dy e NiAl–0.05 %Hf ; (b) NiAl–0.05%Hf–0.05% Zr, NiAl–0.05% Hf e
NiAl–0.09% Zr ; (c) NiAl–0.05% Hf–0.05%La, NiAl–0.05% Hf e NiAl–0.09% La ; (d) NiAl–0.05% Y–0.05%La, NiAl–
0.06% Y e NiAl–0.09% La

Guo suggested that Hf and Zr cations segregate to the oxide grain boundaries by interacting with
oxygen (Figure 2.13); Moreover, if deposited together, they form an ionic group that synergistically
produces stronger interactions than the two individual ions.

Figure 2.13: Zr and Hf doping of a modified aluminide coating

38
Hamadi et al. [17] characterized a doped NiAl coating with Zr; Figure 2.14 shows the concentration
profile of the Zr obtained by GDMS, in which it is noted that the RE moves to the interface between
the interdiffusion zone and b-NiAl.

Figure 2.14: Cross-section of the coating with the Zr profile

The cyclic oxidation resistance of the AM1 / NiAl (Zr) system was compared to that of the AM1 /
NiAl system to obtain the effect of Zr; In addition, the AM1 / (Ni, Pt) Al system has also been taken
as an industrial reference. From the curves shown in Figure 2.15 it is shown that the behavior of the
doped lining with Zr is absolutely comparable to Pt aluminum.

Figure 2.15: Mass variation vs time for cyclic oxidation tests at 1100°C

SEM (Figure 2.16) analysis also showed that after 250 cycles, spallation and oxidation phenomena
are present on the coating surface of NiAl, while appearing on the surface of the modified coating
only after 500 cycles. Moreover, in the case of NiAl, the spallation zones reveal the presence of
39
numerous cavities at the intermetallic-oxide interface; these cavities have not been observed in the
doped coating

Figure 2.16: Cross-section SEM micrographs of oxidized sample AM1/NiAl (on the left) and AM1/NiAl (Zr – on the
right) after: a) 50 cycles; b) 250 cycles; c) 500 cycles of oxidation tests at 1100°C. In the figure b, the arrows
indicate the cavities[18].

40
Figure 2.17 compares the isothermal oxidation behavior of modified aluminide and simple aluminide
coatings.

Figure 2.17: Change in weight after 100h of isothermal oxidation at 1100 ° C of doped and standard NiAl coating.

In the first 3h the gain by weight is significant, and corresponds to the transient oxidation regime; at
this stage, oxides of all the constituents of the alloy are formed, provided that under such conditions
the relative free energy variation is negative. By the progress of the oxidative process, it is expected
that the thermodynamically stable oxide predominates with respect to the others. The metastable
phases of the alumina (g, d, q), do not have the same protective action as it guarantees the a-Al2O3.
Therefore, avoiding or minimizing transient oxidation processes improves the protective action of the
oxide scale.
Subsequently, oxide growth reaches a steady state. Between 20 and 100h the curves for both coatings
have the same pattern and thus the same steady oxidation regime.
The data show that with Zr the transient oxidation phase is reduced by a factor of 2. This may be due
to or to inhibition of the transport to the outside of the Al, which thus promotes anionic oxidation and
the growth of a-Al2O3, or to the growth of q-Al2O3 nucleation.

References

[1] Y. Tamarin, ASM International, Ohio, 2002.

[2] J.R. Nicholls, Journal of metals, 52 (1), pp. 28-35, 2000.
[3] N.P. Padture, M. Gell, and E.H. Jordan, Science, 296 pp. 280-284, 2002.
[4] J. R. Nicholls, “Designing Oxidation-Resistant Coatings,” no. January, 2000.
[5] A. G. Evans, D. R. Mumm, J. W. Hutchinson, G. H. Meier, and F. S. Pettit, ,” Prog. Mater.
Sci., vol. 46, no. 5, pp. 505–553, 2001.
[6] G. Evans et al., Progress in Materials Science ,46, 505-553, (2001).

41
[7] Tawancy H. M.; Sridhar N.; Abbas N. M., “Journal of Materials Science, 35 (2000), p. 3615-
3629.
[8] G.W. Meetham, Mater. Sci. Eng. 2 (1986) 290.
[9] G.W. Goward, Surf. Coat. Technol. 108-109 (1998) 73-79.
[10] D. P. Whittle, J. Stringer, Phil. Trans. R. Soc. Lond. A, (1980) 295
[11] J.D. Kuenzly and D.L. Douglass: Oxid. Met. Vol. 8 (1974), p. 139
[12] R. Bianco and R. A. Rapp, J. Electrochem. Soc. (1993) vol. 140, issue 4, 1181-1190
[13] ee, W.Y., Wright, I.G., Pint, B.A. et al. Metall and Mat Trans A (1998) 29: 833
[14] B. A. Pint, Oxid. Met. 45 (1996) 1-37.
[15] B.A. Pint, Proc. of the J. Stringer Symposium, November 2001.
[16] M.W. Brumm, H.J. Grabke, Corros. Sci. 34, 547, 1993.
[17] H. Guo et al., Corrosion Science 88 (2014) 197–208.
[18] S. Hamadi et al., Surface & Coatings Technology 204 (2009) 756–760.

42
CHAPTER 3
Processes for production of diffusion aluminide coatings

3.1 Coating technique

Diffusion coatings are obtained from an alloy that is superficially enriched with elements capable of
forming oxide scale such as Al, Cr, Si, or combinations [1], up to a depth of 10 ÷ 100 µm. These
elements bind to the primary elements of the alloy, resulting in intermetallics. This type of coating
can be obtained by various techniques, as shown in Figure 3.1:

Figure 3.1: Diffusion coating technique

Figure 3.2 describes the process steps generally common to all techniques, which are:
• Generation of vapors, generally halide, containing the element that will form the protective oxide
scale (Al, Cr or Si);
• Gas transport to the surface of the component, by a partial pressure gradient;
• Gas reaction with substrate followed by solid state diffusion processes within the alloy;
• Additional thermal treatments, if needed, to obtain the desired composition and coating properties.

Figure 3.2: Deposition mechanism of diffusion coating procedure by vapour phase process

43
From a thermodynamic point of view, the most important chemical reactions are those related to the
formation of volatile halide and its decomposition, and those related to solid state diffusion. There
are therefore two crucial temperatures for the correct deposition of an element: the temperature that
allows the formation of the gaseous halide and the one that activates the diffusion.
The microstructure, the activity of the Al (or Cr, Si) within the coating and the thickness of the coating
depend on the alloy used as substrate, on the process parameters and on the following heat treatment
selected.

3.2 Growing mechanism of diffusion coatings

There are two mechanisms for a diffusive coating formation, depending if the main diffusing species
is aluminum inside the substrate or is the Ni diffusing outwardly. The two mechanisms lead to the
achievement of different types of coatings, known as inward and outward coatings.

There are, therefore, two distinct process, which differ in the term of the Al activity in the gaseous
phase and process temperatures:
• High Temperature, Low Activity HTLA
• Low Temperature, High Activity LHTA
The activity of a gaseous phase can be approximated, at low pressures, with the ratio between the
partial pressure of the gas and the atmospheric pressure. High activity involves high partial pressure,
which in the specific case indicates the availability of Al-gaseous halide.

3.2.1 High Temperature Low Activity Process

This method applies a high temperature (> 1000 ° C), for a time ranging from 3 to 4 hours, in the
presence of a low activity of the Al (HTLA).
It is a "single-stage" method with the outward diffusion of the Ni and the subsequent reaction with
Al to form the b-NiAl coating: the growth of this phase is therefore going outward. In practice, Al
also diffuses into the substrate, resulting in coating growth in that direction, but this contribution is
significantly lower than that due to the transport of the Ni (DNi> DAl). Conversely, the diffusion of Ni
to the outside causes the formation of a poor Ni area just below the original surface of the substrate;
such a zone can be a phase precipitate rich in metal elements and is commonly referred to as
interdiffusion zone (IDZ).
44
Figure 3.3 shows schematically the progressive stages that characterize the deposition. It is noted
that, as compared to the surface of the substrate, the coating has grown outwardly.

Figure 3.3: Schema for the progressive stages of aluminization by HTLA pack cementation – a) Pure Ni (e1 is formed
’
firstly, then e2,e3 and so on); b) Ni-based super alloy (e1 and e1 are formed firstly, then the others) [2]

As mentioned, the outside diffusion of the Ni is much faster than the diffusion of aluminum. This
difference between the two speeds involves the voids, known as Kirkendall voids near the original
interface (Figure 3.3a). The advancement of the process causes, in a Ni-based superalloy, a depletion
of Ni in its respective areas within the alloy (Figure 3.3b). Thus, when a layer e1 is formed externally,
a corresponding layer, depleted in Ni, forms inside the alloy, and so on. This technique therefore
allows to obtain a b-NiAl coating, which is however accompanied by the presence of precipitates
(such as TiCs) that are not soluble in the intermetallic matrix (Figure 3.4).

Figure 3.4: Final microstructure of a Ni-based super alloy coated by HTLA pack cementation
45
Figure 3.5 shows in section a sample aluminized by HTLA pack cemetation. It is noted that this
coating is characterized by two areas, which are roughly the same thickness:
• External β-NiAl layer
• Intermediate area, poor in Ni with heavier element precipitates

Figure 3.5: Ni-based superalloy aluminized by HTLA pack cementation [3]

3.2.2 Low Temperature High Activity Process

In this method the deposition takes place at low temperature (LTHA) and involves the execution of
two steps: a low temperature deposition (700 - 850 ° C) followed by diffusion annealing at
temperatures above 1000 ° C . Downstream of the first step, the coating is made up of an excessively
rich phase in aluminum (d-Ni2Al3 or d-Ni2Al3 + b-NiAl rich in aluminum), carbides and precipitates
of the various constituent elements. This microstructure is due to the diffusion into the interior of the
inward diffusion, which causes an increase in the interior of the coating (DAl> DNi). During the second
stage (diffusion treatment), the Ni moves outwards leading to b-NiAl formation, quite similar to
HTLA mode. This step is crucial, as the d-Ni2Al3 phase is a fragile phase with mechanical properties
much lower than b-NiAl. Again, in this case, the formation of an interdiffusion zone is provided at
the interface between the coating and the substrate.

Figure 3.6 shows the first stage of deposition, the aluminization. It is noted that, as compared to the
surface of the substrate, the coating is increased inward.

46
Figure 3.6: Schema for the progressive stages of aluminization by LTHA pack cementation – a) Pure Ni (e1 is formed
firstly, then e2,e3 and so on); b) Ni-based super alloy (e1 is formed firstly, then the others)

Figure 3.7 shows the final microstructure; this consists of:

an outer layer of b-NiAl with carbides and precipitates of substrate elements that were formerly
bonded to Ni, an intermediate layer of b-NiAl free of carbides and precipitates, and finally an
interdiffusion area. In addition, it is noted that during the thermal treatment, following the
aluminization, the coating is characterized by a portion of growth towards the outside, due to the
activation of the Ni diffusion outwardly.

47
Figure 3.7: Final structure of a Ni-based super alloy coated by LTHA pack cementation

Figure 3.8 shows a section of an aluminized sample by pack cemetation LTHA. It is noted that the b-
NiAl thickness is substantially larger than that of the interdiffusion zone.

Figure 3.8: Ni-based superalloy aluminized by LTHA pack cementation

3.3 Pack Cementation

In this process, the components to be coated are first cleaned, generally by means of alumina blasting
to remove oxides and contaminants from the surface. If necessary, the area of the component to be
protected from deposition of the coating is masked with a material that can withstand the process
temperatures without going against degradation. At this point, as shown in Figure 3.9, the components
are buried in a mixture of powders (packs) inside a sealed (or semisealed) reactor. The reactor is then

48
inserted into a furnace and heated in an inert atmosphere (Ar or H2), in order to avoid the oxidation
of the elements that will produce the coating in the pack.

Figure 3.9: Pack cementation process

The reactor is composed of materials that do not degrade at process temperatures such as Inconel
alloys and stainless steels 310, 321 and 347.

Powders contain:
• The aluminum source (or Cr, Si), generally called master alloy.
• An "activator" salt, usually an ammonium or sodium halide, which, by reaction with the source, will
form the gaseous species containing aluminum and transport it to the surface to be coated.
• An inert filler, such as powdered alumina, designed to create interconnected porosity for gas
transport and prevent sintering, which would occur at process temperatures.
The inert filler is not always indispensable; the choice depends on what type of Al source is used.
Generally, in LTHA processes, it is preferred to use Al pure as a master alloy, which allows to cast
more gas in the gas phase, resulting in a greater availability of the deposition elements. However,
choosing a low-core element involves the need to adopt an inert filler that prevents the sintering of
the pack. If this is not used, there will be difficulties in recovering the components.
In HTLA processes Al (Al-Cr, Al-Co) alloys are generally used as the source; The filler is not
necessary, in this case, because of high melting point of the master alloy

A salt, to be suitable for activating a pack cementation, must be necessarily unstable, to react at the
process temperatures. Salt selection guidelines are the following:

49
The anion is generally a halide. Concernign the cation, ammonium salts are preferred because they
decompose to high T, producing H2 that increase the reducting atmosphere.
The reaction that the activating salt has with the component depends on the type of salt used. In the
case of NH4Cl, NH4Br and NH4I (Figure 3.10a), the halide stays in the gas phase even after reaction
with the substrate and deposits only solid Al: there are no co-products that can be deposited on the
surface of the component. The F is still used, despite the disadvantage of ist utilization, since it is
characterized by high partial pressure, therefore forms AlF very easily. Finally, using Na-halides
(Figure 3.10c), instead of solid precipitates, there are liquids on the surface.

Figure 3.10: Mechanism of reaction and diffusion for different activator salts

The reactions between NH4X ammonium halide and a substrate in Ni-base super alloy are as follows:

• Decomposition of NH4X

NH4 X(s) =NH3 g +HX(g) (3.1)

• Volatile Al halide formation

6HX(g) +2Alpack =2AlX3 g +3H2(g) (3.2)

2AlX3 g +2Alpack =3AlX(g) (3.3)

• Al deposition on top of the substrate

2AlX g +3Nisub =2NiAlalloy +NiX2(g) (3.4)

3NiX2 g +5Alpack =3NiAlalloy +2AlX3(g) (3.5)

50
 3AlX g +2Nisub =2NiAlalloy +AlX3(g) (3.6)

2AlX g +2Nisub =2NiAlalloy +X2(g) (3.7)

2AlX g +H2(g) +2Nisub =2NiAlalloy +2HX(g) (3.8)

where reactions (3.4) and (3.5) are displacement reactions, disproportionation (3.6), decomposition
(3.7), and finally reduction reaction (3.8).
The advantages of this technology are:
1. the exercise at atmospheric pressure;
2. the good reproducibility of the coating and
3. low plant costs, since the latter is essentially a furnace.
The main disadvantages concern the limited flexibility of the process in terms of coating composition
and the length of the process for high thicknesses. In addition, pack particles may be trapped in the
outer layer of the coating: this phenomenon is generally negative since such inclusions will prevent
the formation of a continuous oxide scale.

3.1.1 Process above the pack

This process works very similar to that discussed in Section 3.3, with the exception that the
components to be coated are placed above the powders containing Al.
Al transport from the pack to the substrate occurs by gas phase diffusion while the substrate surface
inside for solid phase diffusion. The first type of diffusion increases the surface concentration of Al
in the coating while the second decreases it to a steady state. In the vapor phase, the Al transport rate
is much faster than solid phase diffusion within the substrate, which is therefore the deciding step for
the deposition rate. Finally, there is the diffusion of reaction products from the substrate to the outside
of the reactor, which controls the purity of the coating.
The diagram of an above the pack system is shown in Figure 3.11.

51
Figure 3.11: Above the pack process

A typical temperature profile of the above the pack aluminization process is shown in Figure 3.12:

Figure 3.12: Profile temperature in a out-of-pack process

This method, as compared to standard pack cementation, simplifies the process of unpacking
components from dust; Moreover, it allows for a much cleaner and homogenous coating, even for
complex geometry components, without embedding of particles during the process.

Another variant, more technological than concept, is VPA (Vapour Phase Aluminizing); the only
difference lies in the fact that no more dusts but pellets are used. The advantage is the simplification
of the charge and discharge processes of Al source, since no volatile metallic dust is formed: this in
fact provides for the most in-worker protection systems and under a certain dimension can become
dangerous (pyrophoric phenomenon). Finally, it is easier to ignore the waste of the process: once
depletion of Al under some concentration occurs, the source is exhausted and can be replaced and
reloaded more easily.

52
3.4 Slurry aluminization

The coatings obtained by this technique offer a very similar chemical composition to that found in
alluminide products by pack cementation. In this process, powders and activators are mixed with an
organic binder (or an aqueous emulsion) to obtain a slurry. The coating is applied to the substrate by
immersion or spraying at room temperature; then undergoes a low temperature treatment treatment,
typically around 200 ° C. After this, the diffusion process of the Al is activated: the components are
brought to temperature (650 ÷ 1100 ° C), which varies depending on the substrate to be coated and
its final use.

With this technique, it is easy to restore if necessary parts of the coating, since it is possible to apply
the slurry locally. However, it is less clean than above the pack as it introduces an interface between
a solid precursor and the component; all those reactions between the gas and the surface become
reactions between a solution and the component, which will then be covered with that part of the
slurry that has not reacted. At the end of the treatment, you will need to clean up the residues present,
which is not trivial when you want to cover the cavities.

References

[1] Z.D. Xiang, P.K. Datta, Mat. Sc. and Eng. A356 (2003) 136/144
[2] R. Pichoir, Corrosion, D.R. Holmes and A. Rahmel, Ed.,(1978) Applied Science Publishers
Ltd., London p 271.
[3] G.W. Goward and D.H. Boone, Oxid. Metals, (1971),Vol 3, p 475

53
CHAPTER 4

Design of experimental activities

The substrate

The substrate used is a directionally solidified nickel-based superalloy, commercially known

as René 108ds, whose composition is shown in Table 4.1. The alloy was supplied as bars with a
diameter of 20 mm. These bars were cut in order to obtain the 5 ÷ 6 mm thick disks for the phase 1
and 1 ÷ 2 mm thick disk for the phase 2

Table 4.1: Chemical composition of René 108ds, the substrate used in this work

Compositions (wt%)

Ni Cr Co Al Ti Mo W Hf Ta

Renè 108DS Bal. 8,4 9,5 5,5 0,7 0,5 9,5 1,5 3

Coating requirements
Superalloys used in high-temperature applications are generally developed with optimized
mechanical properties such as tensile, creep, and fatigue strength while maintaining microstructural
stability over a wide temperature range. These properties come from the chemical modification of the
superalloy. Despite of this mechanical resistance, the environmental protection has to be provided by
coatings, which are not suitable with the thermal stress exposition during operating time. So, to design
a performing material, some requirements are to be considered [1]:

Oxidation and corrosion resistance:

• Study of the formation of the protective coating
• Selective permeability of the oxygen to the oxide scale
• Tailored growth rate of protective surface scale
• Controlled doping of the protective coating by reactive elements
54
The coatings adopted within the turbine differ according to the considered rotor stage. From an
industrial point of view the blades of the first stage at a higher temperature (HPT) are coated with an
aluminide modified with Pt, which ensures high performance with very high production costs, both
for the cost of the raw material and the cost of the electrodeposition process, despite the deposited
layer being ≤ 10 µm. The third stage blades (LPTs) are coated with standard aluminide because of
the low temperatures to which they are subjected. For IPT, industrial modified Pt aluminide coating
is used: the performance and cost of this coating are oversized to the needs of the blades, but there
are no other solutions that guarantee the same safety against a lower cost. For this reason, the first
phase of this work was to find a good candidate, as doping element, to replace the Pt and reduce cost
without losing the protective properties for the IPT.
The selected process for the coating formation is the aluminization by pack cementation which is an
industrially well consolidated procedure. This first step of the work was performed in order to obtain
a know-how to increase the coating characteristics. Then, as the best parameters for the standard
process was evaluated, the doping with reactive element was introduced and studied by comparison
with the standard coatings. This was an important step to understand deeply the mechanism of the
oxidation and corrosion resistance increased by the introduction of the reactive element.

As it was said before, a protective coating can increase the environmental corrosion resistance
despite the decreasing of mechanical properties of the component. This variation in mechanical
properties can involve a damage to the component with the consequent failure which can have
disastrous consequences in some applications. So, the performing material, with oxidation/corrosion
resistance increased, has to respect some other requirements in terms of stability of the component:

Stability
• No undesired phase formation within the coating
• Compositional stability across interface
• Minimized brittle phase formation

The final microstructure of the coating is the result of the diffusion of chemical elements between the
coating and the substrate during the aluminization. Consequently, as the interdiffusion zone is formed
into the superalloy, the system can be described as a complex multilayer material (Figure 4.1). The
microstructure of each layer evolves under service conditions due to the combined effect of high
temperature and mechanical stresses. As a consequence of the high temperature, the materials are
55
subjected to a thermal expansion without any exception. Each phase has his thermal expansion with
his coefficient; even if the properties of IDZ are unknown [2], it is evident that the thermal expansion
of the three different phases can’t be considered independent since the IDZ expansion, for example,
is prohibited by the coating zone. This kind of stress can generate thermo-mechanical fatigue during
the complex thermal and mechanical history for a typical cycle of operation.

Figure 4.1: Image of the multilayer system composed by: substrate, IDZ, protective coating. The arrows indicate the
sense of thermal expansion

According to the principle of outward diffusion in aluminization process, the Ni diffuses and reacts
with Al to form the β-NiAl coating: the growth of this phase is therefore external. In practice, Al also
diffuses into the substrate, resulting in coating growth in that direction, but this contribution is
significantly lower than that due to the transport of the Ni (DNi> DAl). As a consequence of this
diffusion, the IDZ can be considered as a depleted zone of Ni in which the phases are rich in heavier
elements of the superalloy. So, theoretically, the addition of an external nickel layer (ENL) onto the
substrate could have several advantages:
• Control of chemical composition of the β-NiAl phase
• No IDZ formation
• Substrate-independent process
First of all, the possibilities to tailor the composition of the ENL could open the way to embed
particles inside the diffusion coating with a consequent increasing of corrosion resistance. Then, the
deposition of a nickel layer onto the substrate before the aluminization could play the role of a Ni
source for the aluminization process, without modifying the composition of the superalloy. And
finally, the possibilities to have a Ni reservoir for the diffusion coating formation could positively
influence the choice of the chemical composition of the superalloy by freeing the presence of the Ni
as based element instead of other elements that can influence the mechanical properties of the alloy.
About the possibilities to embed particles inside the ENL, it was already studied that the
doping of the bond coat with some oxides particles could increase the oxidation resistance: Peng and
56
co-workers [3] added La2O3 particles into aluminide coatings using two steps: the first step is to
electroplate a Ni-La2O3 composite film on Ni, and the second one is to aluminize on the as-deposited
composite film. The oxidation results showed that the aluminide coatings with La2O3 particles
exhibited better oxidation resistance compared to La2O3-free aluminide coatings. Similar technique
was applied to manufacture aluminide coatings with CeO2 or Y2O3 dispersion [4], [5].
Furthermore, it was reported that during the low activity aluminization, the filler Al2O3 [6] could be
entrapped into the outer layer of aluminide coating, thus affecting the grain growth and oxidation
behavior of the aluminide coatings [7]. The author works exhibited that the Al2O3-dispersed
chromizing coating produced by chromizing an as-electrodeposited Ni-Al2O3 nanocomposite film
exhibited an increased oxidation resistance. In this field, nanoparticles show some great properties
because of their high reactivity but the dispersibility in a matrix is a fundamental parameter to have
a positive influence on the bulk properties. The idea, developed in the phase 2 of this work, is to
deposit a nickel-matrix n-Al2O3 nanocomposite onto the substrate in order to use the nanoparticles as
reactive element and as nucleation grain to promote the rapid transition of the alumina through the
various metastable phases up to phase α so as to form a protective and stable TGO. In order to
incorporate nanoparticles into the coating, it was necessary to develop a production process capable
of transporting the nanoparticles within the bond coat: if this is impossible in the case of vapor-phase
processes such as CVD and pack cementation, it is possible in case of liquid deposition such as
electrodeposition. In fact, particles could be conveyed by platinum electrodeposition, but this also
requires a cost that is not indifferent even for the stack of blades that would not require a modified
aluminide with platinum: it is therefore necessary to use a coating that is economical, both from the
point of view of the materials, both from the point of view of the synthesis process.
In addition, the high complexity of blades geometry makes it difficult to handle electrolytic
deposition, since the coatings produced with this technique are not homogeneous when complex
geometries are present.
In summary, the coating has to meet these requirements:
1. Must be compatible with the CVD process used to form β-NiAl
2. Must be able to convey nanoparticles
3. It must have homogeneous thickness even on complex surfaces
4. Must be economical
The technique that could respond to these needs is that called autocatalytic or electroless deposition,
which is, a deposition that operates in the absence of electrical current and which produces more
homogeneous coatings than those obtainable by electrolytic depositions. As will be discussed later,
however, this process has undergone modifications to produce a pure nickel layer.
57
Operatively, at first, the optimization of the solution formulation was achieved to reach the required
thicknesses; Subsequently, the nanoparticles were introduced and the concentrations were optimized.
The third phase of this work has been about the aluminization process of the new electroless
nickel plating samples. The introduction of an ENL was, theoretically, positive for the aluminide
coating. However, there is no publications in literature about the application of an electroless nickel
layer with the aluminization. So, it was necessary to select a technique that was easy to apply,
economical and clean from a chemical point of view. Furthermore, the toxicity of the compounds
used and the residues and process risks associated with the use of halogen activators make the pack
cementation a process to replace. Slurry aluminization was the perfect candidate because the main
benefit of the slurry compared to phase vapour deposition is the easier manufacturing and the lower
cost especially for large parts. In collaboration with University of La Rochelle and the group of
Professor F. Pedraza, the EPNP samples were aluminized by the slurry aluminization. In summary,
in Figure 4.2, the scheme collects the idea of the phase 2 and 3.

Figure 4.2: Schema of the aluminization for the new electroless nickel plating substrate.

58
4.1 Phase 1: Oxidation and hot corrosion resistance

4.1.1 Standard aluminide coatings

Standard diffusion aluminide coatings were obtained by a semi-sealed HTLA (High Temperature
Low Activity) above-the-pack cementation in VPA configuration at Centro Sviluppo Materiali
(CSM) S.p.A. (Figure 43).

Figure 4.3: Deposition plant for above the pack cementation (thanks to CSM S.p.a.); Heat treatment profile (below)

The heat treatment profile, in Figure 43, have two plateaux, one at about 750°C and the other at about
1100°C. The heat treatment temperatures were chosen in order to follow the single steps of
aluminization described in the paragraph 3.2.1 and respecting the temperature of the last heat
treatment of the super alloy.

59
The samples were introduced in the furnace (Figure 4.4) and heated with a temperature profile shown
in Figure 43. In order to avoid the atmosphere contamination, the aluminization was conducted under
continuous flux of Ar. The activator salts powder (NH4F pur. ≥99.99% and AlF3 pur. ≥99.99%) and
the master alloy (Al-Cr 60 wt%) were purchased from SIGMA-Aldrich and dried before the
deposition at 80°C.

Figure 4.4: Sample before (a) and after (b) aluminization; samples placed in the basket before (c) and after
aluminization (d)

The Ni-based super alloy rod was cut in cylinders (19.2mm of diameter and 6 mm of thickness),
cleaned with acetone and polished with a P400 mesh SiC paper.
To determine the effect of activator salts in the coating, two different concentrations were chosen for
each salt. In the

Table 4.3 concentrations in terms of wt% of the pack alloy and mole of fluoride are reported for NH4F

and AlF3. In all the depositions, it was respected the equi molarity of fluoride.
60
Table 4.3

Concentrations of activator salts, respect to Al master alloy, for the series NF and AlF

wt% of activator mol of activator

Deposition
salt salt

NF1 0.050 0.0135

NF2 0.075 0.0202

AlF1 0.100 0.0135

AlF2 0.170 0.0202

4.1.2 Isothermal oxidation tests

High temperature oxidation tests were carried out in air into a Lenton Thermal Design Ltd
tubular furnace, shown in Figure 4. .

Figure 4.5: Tubular furnace Lenton Thermal Design Ltd, with quartz reactor.

The coated samples were placed on a substrate holder made with aluminum oxide, in order to avoid
some interactions with the substrate. For each test, the samples were introduced in the furnace by the
61
heating zone, then slowly move close the hot zone (1050°C) and, after the test time, removed from
the furnace with a slow cooling rate (Figure 4.). This operation was performed for alternating cycles
of 7/14 hours for a total of 105 hours.

Figure 4.6: Isothermal oxidation test: a) Sample holder; b) Example of a test at 1050°C. Below is represented the
schema of the different zones inside the furnace

It is noteworthy that the heating and cooling zone is a fundamental step of the oxidation tests: in fact,
a sudden decreasing or increasing of temperature, mostly for the samples after the first hours of
oxidation, could damage the oxide scale by thermal shock and compromise the test. Oxidation curves
were obtained by measuring the weight gain for each cycle with an analytical balance.

4.1.3 Hot corrosion tests

Hot corrosion tests were performed in a high temperature furnace Lindberg. The alkaline salts chosen
for the corrosion were Na2SO4 and NaCl in a ratio of 1/3. The procedure to perform the test was
selected to be reproducible and significant for the operating conditions (temperature and quantity of
corrosive agent). For this reason, the corrosive agent was added by mixing the salts with a minimum
quantity of deionized water (20 wt%) in order to obtain a slurry that can easily be spread over the
sample’s surfaces (Figure 4.-left). Once the slurry was prepared, spread and weighed, water has been
removed by heating in hot plate in order to obtain a film of salts mixture all over the surface of the
samples.

62
Figure 4.8: Hot corrosion test procedure: slurry application on a sample (left, red circle) and samples prepared for the
heat treatment (right)

Hot corrosion cycle tests were performed at 900°C with cycle of 10 hours for a total of 200 hours.
Every cycle, the samples were washed in hot deionized water, in order to remove the un-reacted salts
and the soluble corrosion products before measuring the mass variations. Once the mass variation is
measured, the slurry was refilled with the same step described before. This can ensure the
reproducibility of the process.

4.1.4 Modified aluminide coatings with Zr

Once the results of the standard aluminization have shown the best parameters in terms of yield of
the process (thickness and quality of the coating) and in terms of oxidation and hot corrosion
resistance, the modified aluminization was performed by adding Zr in the coating.
The aim of this operation is to obtain a modified coating with higher performance in terms of
oxidation and corrosion resistance without modifying the plating process studied in the first part of
this work. The reactive element chosen for the doping of the coating is the zirconium, that has an
atomic dimension suitable for the interaction with oxygen anions and a relative low cost of
production.
Modified coatings were obtained by adding ZrF4 in the activator salts mixture. The concentration of
the standard fluoride has been kept constant while the concentration of ZrF4 was studied and the
values selected for the aluminization were reported in Table 4.4. The presence of Zr, after the
aluminization, has been studied by GDOES in collaboration with the group of Prof. Fedrizzi at
University of Udine, Polytechnic Department of Engineering and Architecture.
Isothermal oxidations and hot corrosion tests were performed and compared with the results obtained
for the standard coatings.
63
Table 4.4: Concentrations of activator salt, respect to the Al master alloy, for the series ZrF

wt% of activator mol of activator

Deposition
salt salt

ZrF1 0.121 0.0072

ZrF2 0.145 0.0087

ZrF3 0.165 0.0100

ZrF4 0.335 0.0201

ZrF5 0.670 0.0402

4.2 Phase 2: Electroless nickel plating

4.2.1 Principle of deposition

The term electroless plating refers to an autocatalytic deposition technique that is purely chemical, in
the absence of electrical current. The first publications on this technique are by Brenner and Riddell
[8] who first coined the term “electroless plating”. The deposition occurs by selective reduction of
metallic ions on the surface of the substrate, creating a catalytic layer, which is why the electroless
deposition is also called autocatalytic [9].

Figure 4.9: Comparison between electrolytic and electroless plating: inhomogeneities for the coating electrolytic

Operating in the absence of an electric power source, there is a need of a reducting agent within the
solution which, when oxidized, is made available the electrons for the reduction of the metal ions.
This results in a very uniform thickness coating, when compared to those obtained by electrolytic
64
deposition, which does not produce thickening near the edges, i.e. where the electric field is
concentrated. In general, the process can be considered as a classical electrochemical reaction in
which an anode reaction and a cathodic reaction occur [10] [11] [12]:

Anodic reaction (oxidation):

𝑅𝑒 → 𝑂𝑥 + 𝑛𝑒 >

Cathodic reaction (reduction):

𝑚𝑀𝑒 AB + 𝑧𝑚𝑒 > → 𝑚𝑀𝑒 D

Where Re is the reducting agent and Ox is the oxidized form. Specifically, electroless deposition also
provides a number of collateral reactions that differ in function of the reducting agent and the
alkalinity or acidity of the solution. All the reactions that produce the coating can be universally
described by a series of anodic and cathodic elemental reactions as claimed by Van Den Meerakker
[13].
For alkali solutions the anode reactions are:
1. Dehydrogenation of the reducting agent:
2. Oxidation:
3. Recombination:
4. Oxidation
The cathode reactions are:
5. Metal Deposition / reduction:
6. Hydrogen evolution:
These reactions proceed under a thermodynamic push that is a function of the temperature, activity
and the used species. This is driven by Gibbs' free energy which is related to the electrode potential.
The electrode potential quantifies the tendency of a given chemical species to gain electrons, so to
reduce itself: it corresponds to the electromotive force supplied by a galvanic cell consisting of a
standard hydrogen electrode and an electrode of the material to be measured, and is expressed in Volt.
Since it is virtually impossible to measure the potential of a single cell electrode, it is always referred
to a standard configuration in which an electrode is hydrogen and the potential of the entire cell is
measured at 298 ° K and the pressure of 1 atm. If redox occurs in a concentration and temperature

65
other than standard, the reduction potential can be calculated according to what is known as the Nernst
equation. For a generic reduction reaction:
𝑀𝑒 EB + 𝑛𝑒 - → 𝑀𝑒 D

Nernst equation can be expressed as follows:

𝑅𝑇 𝑂𝑥 𝑅𝑇 𝑀𝑒 'B
𝐸 = 𝐸D + ∙ ln = 𝐸D + ∙ ln
𝑛𝐹 𝑅𝑒𝑑 𝑛𝐹 𝑀𝑒 D

Where:
𝑂𝑥 : Molar concentration of the species in oxidized form
𝑅𝑒𝑑 : Molar concentration of the species in reduced form
𝐸D : Reduction standard potential
R: gas constant with value 8.314
T: absolute temperature in Kelvin
F: Faraday constant value 96485 C
n: the number of electrons transferred to the reaction

This equation is related to the free energy of Gibbs, which takes into account the thermodynamics of
the reaction. For the generic oxidation reaction, the free energy variation is:
𝑎N*
∆𝐺 = ∆𝐺 D + 𝑅𝑇 ln
𝑎N* OP

Where ∆𝐺 D is the standard molar energy of formation, that is, the energy associated with the
formation of a chemical species in the state of aggregation more stable under standard conditions.
The relationship between free energy and reduction potential is:

∆𝐺 = -𝑛𝐹𝐸

Where n represents the number of electrons per product moles and F is the faraday constant.
explaining:
∆𝐺 D 𝑅𝑇 𝑎N*
𝐸=- - ln
𝑛𝐹 𝑛𝐹 𝑎N* OP

From the sign of drift the spontaneity of the reaction:

•∆𝐺 = 0: The system is in balance
66
•∆𝐺 < 0: The reaction happens spontaneously
•∆𝐺 > 0: The reaction to progress requires an energy supply

The electroless plating technique requires the presence of a catalytic surface, specifically the almost
all metals of the group VIII periodic table, i.e. Fe, Co, Rh, Pd, Pt and Ni. Ni, Co, Rh and Pd are
considered as catalytic metals; instead, if the substrate metal is less electronegative than the metal to
be deposited, a catalytic nickel layer is formed by the substrate according to the following reactions:
Example nickel on Iron [9]:
D
𝐹𝑒 D + 𝑁𝑖 'B → 𝐹𝑒 'B + 𝑁𝑖VWX
Example nickel on aluminum:
D
2𝐴𝑙 D + 3𝑁𝑖 'B → 2𝐴𝑙\B + 3𝑁𝑖VWX

Overall, the coating process can be divided into five steps:

1. Reagent transport on the surface
2. Adsorption of reagents on the surface
3. Redox reactions
4. Desorption of Redox reaction products from the surface
5. Redox reaction products transport away from the surface
The chemical reactions that occur are dependent on the composition of the solution, the nature of the
reducing agent and the pH. As hypothesized by Brenner and Riddell [8], a fundamental role plays the
ions 𝐻B and 𝑂𝐻> actually oxidises the reducting agent by providing electrons to the metallic ions.
As the reaction proceeds, nucleation zones are formed that continue to grow in size until a continuous
film is produced.

4.2.2 Plating solution compositions

Electroless nickel coatings are obtained by a reduction in aqueous solution of nickel ions on a catalytic
surface. The deposit itself is catalytic for reduction, so the reaction continues until the surface is
immersed. The solution consists of a Ni-soluble salt which acts as a Ni2+ reaction reservoir, a
reducting agent that provides electrons for nickel-reducing, complexing agents that control the
available nickel for the reaction, buffer solutions that inhibit pH variations, accelerators that increase
reaction rate and stabilizers. The nature and quantity of these components determine the
characteristics of the plating solutions and, therefore, of the deposit.
Below are discussed in detail all the components of the solution as well as the process parameters.

67
Reducting agents:

The most industrially used reducting agents are mainly four. These four reducting agents are
structurally similar as they all contain two or more reactive hydrogens: in fact, the reduction of nickel
occurs by catalytic dehydrogenation [12].

Figure 4.10: Structural formulae for the reductng agents used in electroless plating

The most used reducting agent is the sodium hypophosphite and the detailed electroless deposition
mechanism can be summarized by four reactions:
𝐻' 𝑃𝑂'> + 𝐻' 𝑂 → 𝐻B + 𝐻𝑃𝑂_'> + 2𝐻W`a (1)

𝑁𝑖'B + 2𝐻W`a → 𝑁𝑖 B + 2𝐻B (2)

𝐻' 𝑃𝑂'> + 𝐻W`a → 𝐻' 𝑂 + 𝑂𝐻> + 𝑃 (3)

𝐻' 𝑃𝑂'> + 𝐻' 𝑂 → 𝐻B + 𝐻𝑃𝑂_'> + 𝐻' ( 4)

In reaction (1), hypophosphite ions oxidize into orthophosphite ions in the presence of sufficient
energy and a catalytic surface. Part of the atomic hydrogen product is adsorbed by the substrate. The
adsorbed hydrogen reduces Ni2+ ions in nickel metal (2) and reacts with hypophosphite ions
producing water, hydroxyl ions and phosphorus (3). Also, part of the hypophosphite ions is oxidized
into orthophosphite ions with molecular hydrogen formation (4). Below is the complete reaction (5)
68
 𝑁𝑖 'B + 4 𝐻' 𝑃𝑂'> + 𝐻' 𝑂 → 𝑁𝑖 D + 3𝐻' 𝑃𝑂_ + 𝐻B + 𝑃 + 3 2 2𝐻' (5)

The ions in solution migrate to the substrate by creating a diffusion flow from the solution to the
surface of the sample to be coated. The same surface works as a catalyst by activating the electronic
exchange and allowing the oxidation of the two-species present in the solution. The H+ ions formed
during the redox reaction combine to form molecular H2 that bubble on the surface of the solution.
Figure 4.11 shows the diffusion layer with the species present for a Ni-P deposition.

Figure 4.11: Deposition Mechanism. Diffusion layer where ions migrate from the solution to the substrate, then reduce
or oxidize

Even if the properties of the co-deposited coatings (Ni-P and Ni-B) allow to use them in a wide range
of applications such as fouling and wear resistance, the presence of the phosphorous and the boron in
the alloys does not allow the utilization of this systems in applications involving high temperature
conditions because of the high reactivity of the hetero-atoms. In order to avoid the co-deposition of
the elements coming from the chemistry of the plating solution, hydrazine (N2H2) was already
selected as reductive agent for the production of metallic nano-particles [14],[15],[16],[17],
functional coating onto nanostructured systems[18],[19] and membrane[20],[21],[22],[23]. It is
known [17] that hydrazine can oxidize, in basic conditions, by the following reaction:

The cathodic semi-reaction of Ni2+ reduction is the following:

2𝑁𝑖 'B + 4𝑒 > → 2𝑁𝑖 D 𝐸D = −0.25 𝑉

By combining the two reactions, the reaction that produces the catalytic reduction of nickel on the
substrate is the following:

69
 2𝑁𝑖 'B + 𝑁' 𝐻e + 4𝑂𝐻> → 2𝑁𝑖 D + 4𝐻' 𝑂 + 𝑁' 𝐸D = 0.91 𝑉

This is the reaction hypothesized for the reduction of nickel by hydrazine in the presence of a catalytic
surface. Thus, considering an efficiency of 100% hydrazine oxidation, a mole of hydrazine is
sufficient to reduce two moles of nickel. However, this reaction does not justify the formation of
hydrogen which has been confirmed experimentally and above all the formation of precipitates in the
absence of catalytic surface; such formation, in alkaline solutions, is justified by the reduction of
hydrolyzed nickel directly from hydrazine according to the reactions:

𝑁𝑖 'B + 2𝑂𝐻> → 𝑁𝑖(𝑂𝐻)'B

'

'B B
𝑁𝑖 𝑂𝐻 ' + 𝑁' 𝐻e → 𝑁𝑖 𝑂𝐻 W` + 𝑁' 𝐻_ 𝑂𝐻 + 𝐻W`a

B
𝑁𝑖 𝑂𝐻 W` + 𝑁' 𝐻_ 𝑂𝐻 → 𝑁𝑖 + 𝑁' 𝐻' 𝑂𝐻 ' + 𝐻W`a

2𝐻 → 𝐻'
the overall reaction can therefore be written as follows:

𝑁𝑖 'B + 𝑁' 𝐻e + 2𝑂𝐻> → 𝑁𝑖 D + 𝑁' + 2𝐻' 𝑂 + 𝐻'

In this case a molar amount of hydrazine is required to reduce a mole of nickel, hence a greater
quantity than that described in the previous mechanism that does not envisage hydrogen evolution.
They should then be considered some collateral reactions that see hydrazine decompose or react with
other elements, such as oxygen [17]: in fact, hydrazine may decompose in both alkaline and acidic
solutions according to the reaction:

3𝑁' 𝐻e → 𝑁' + 4𝑁𝐻_

Otherwise, hydrazine can react with oxygen by the following reaction:

𝑁' 𝐻e + 𝑂' → 𝑁' + 2𝐻' 𝑂

70
As a result, the real efficacy of reducing nickel with hydrazine is to be considered less than 100%.
Nickel salts

The cation Ni2+ comes from the dissolution of a metal salt. Typically, the most used salts are sulfates,
chlorides or acetates. Nickel sulphate is preferred for its higher solubility and lower cost. For
chlorides, it has been found that chloride ion can disadvantage deposition on aluminum [1].
Complexing agents

Complexing agents are additives used to prevent the decomposition of the bath. These additives play
three roles [10],[12]:
1. Reduce the concentration of free ions
2. Prevent the precipitation of nickel hydroxide
3. Help maintain pH stable
When nickel ions are in aqueous solution, they are bound to a precise number of water molecules:
this number is called a coordinate number. The complexing agents, once introduced into the solution,
are to replace the water molecules, by changing the properties of metal ions: in particular, the
solubility and reduction potential are modified.
The binding agent that is put into solution replaces the water molecules so much more efficiently as
the bond is stronger with the metal cation. A measure of this force is given by the constant of stability
(or stability constants) of the complex, which is the function of all the sequential reactions that
originate from the substitution of the individual molecules of water by the binding agents.
A n example of binding agent is ethylenediaminotetraacetic acid - EDTA, the structure of which is
shown in Figure 4., which can form 6 coordinate bonds: 2 sites are represented by nitrogen atoms and
4 by the carboxylic groups -COOH.

Figure 4.12: Structural formula of ethylenediaminotetraacetic acid (EDTA)

Although polydentate binding agents such as EDTA are able to make multiple bonds, they do not
necessarily form all available bonds but are strongly influenced by the pH of the solution. The most
common and stable Ni - EDTA complex is 1: 1 stechiometry obtained with EDTA in completely
deprotonated form EDTA4-. This form is present at high pHs, ranging from 10 to 14. By varying the
71
concentration and type of complexes within the solution, the rate of deposition, stability of the
solution and coating quality can be changed.
Table 4.5 presents some of the most commonly used complexes in the electroless depositions, with
relative equilibrium constants with nickel.
Table 4.5: Binding agents for electroless nickel plating with stability constants for the complex

Chelating
Binding agent Structure pKstab = - log K
groups

Acetic acid CH3COOH 1 1,5

Lactic acid CH3CHOHCOOH 1 2,2

Succinic acid HOOCCH2CH2COO- 1 2,2

-
Malonic acid OOCHCH2COO- 2 4,2

ethylendiammine H2NCH2CH2NH2 2 13,5

Citric acid HOOCCH2(OH)C(COOH)COOH 4 6,9

EDTA ((HOOCCH2)2NCH2)2 6 18,6

4.2.3 Plating solution parameters

The two fundamental parameters, in addition to the reagent concentrations, are the process
temperature and the pH of the solution. As far as temperature is concerned, it has been found
experimentally that the solutions with hydrazine begin deposition at 60°C, increasing the deposition
rate by increasing the temperature until degradation to 95°C.

Figure 4.13: Plating rate vs pH (left) and temperature (right) for Ni-P depositions.

72
In fact, evaporation of the solution occurs at temperatures close to boiling temperature of the solvent,
generating a variation of reagent concentrations within the solution, and thereby modifying the
reaction stoichiometry and coating quality. This variation can also lead to the precipitation of nickel
hydroxide Ni(OH)2 following the increase in Ni2+ and OH- ion concentrations beyond the hydroxide
solubility product.
Such increase in deposition rate corresponds to a change in the morphology of the coating and its
adhesion. Typically, increasing the temperature will result in porous deposits, while deposits with
metallic appearance will be favored by low deposition rates. Adhesion varies because the density of
nucleation sites on the substrate is changed, varying the stress condition of the coating.
As for the pH of the solution, solutions based on hydrazine requires a pH range of between 9 and 12.
As in the case of temperature, changing the pH of the solution results in a modification of the
morphology and adhesion of the coating. Since the hydrazine reduction reaction consumes ions, and
since hydrogen production is a side-effect reaction, with the advancement of deposition there will be
a decrease in pH: if this decrease is important it is necessary to use a pH correction or a buffer solution.
A buffer solution is a substance capable of opposing a pH modification following addition of an acid
or base: in general, it is a solution consisting of a weak base and its conjugated acid, or a weak acid
and its conjugated base. The compound in the buffer varies the degree of dissociation depending on
the pH variations that are produced in the solution. This allows to maintain bath reactions within an
optimum pH window for redox reactions and complex formation during deposition.

4.2.4 Selected plating solution compositions and parameters

In the selection of plating solution chemical compositions, the analysis started from some publications
relating to the production of coated membranes for gas separation and other publications related to
the production of battery electrodes. After analyzing these publications, a composition was discarded
due to the low declared deposition rates (20 µm in 8 hours) and the poor homogeneity of the coating
as reported in Figure 4.

73
Figure 4.14: Examples of coatings obtained by varying the reagent concentrations reported in the publications
[24],[25]

The two starting compositions selected for the phase 2 of this work are reported in (from now solution
A and solution B)

Table 4.6: Chemical compositions and parameters for the solution A and B

Solution A Solution B
g/l (mol/l) g/l (mol/l)

Nickel acetate Ni(OCOCH3)2 · 4H2O 21,2 (0,08) -

Nickel chloride NiCl2·6H2O - 23,76 (0,1)

Hydrazine N2H4·H2O 20,02 (0,4) 32,56 (0,65)

Na2EDTA Na2EDTA·2H20 5,62 (0,016) -

Lactic acid C 3H 6O 3 13,5 (0,15) -

Potassium carbonate K2CO3 - 69,1 (0,5)

Potassium hydrogen
KHCO3 - 45 (0,45)
carbonate
Potassium hydrogen
K3HPO4 - 87,1 (0,45)
phosphate

Sodium hydroxide NaOH 9,87 (0,247) -

Potassium hydroxide KOH - 9,78 (0,3)

pH 9,2 10,9

T 75°C 85°C

74
Solution A was initially formulated by Wen et al. to coat zirconium oxide powders and subsequently
adopted by Haag et al. for functionalizing aluminum oxide membranes. Two different binders are
used: EDTA forming a very high stability constant Ni-EDTA complex (pK=18.6) and lactic acid
forming a complex with a modest stability constant (pK=2.2): experimentally it has been observed
that coatings obtained using both complexes exhibit a higher deposition rate than the case of EDTA
alone because of the two different stability constants. The plating solutions were prepared at room
temperature, then placed on a heating plate equipped with a magnetic stirrer and heat to the deposition
temperature (75 or 85 ° C). Only when the process T is reached the sample to be coated has been
inserted.
During experimentation of solution A, the pH was increased to achieve industrially desirable
deposition rates, bringing the pH to process T around the value of 9.2 and velocity in the order of 10.
Subsequently, the concentrations of the two complexes were varied so that the coatings produced
were adherent to the substrate and porous-free. Despite this, the coatings still have little adhesion,
stress and porosity.
During experimentation of solution B, it was studied the complex named tris(hydrazine carboxylato-
N’,O) nickelate(1-) (called Ni complex from this moment) formation by varying the preparation of
plating solution; briefly the nickel complex is formed by adding a solution of Ni2+ obtained by
dissolving 7mmol of NiCl2 hexahydrate in 100 ml of DW to the hydrazine carboxylate solution and
stirred until complete mixing. Then a buffer solution of K2HPO4 was added and the pH was adjusted
with KOH to a value of 11.5.
For both solutions (A and B), the plating solution was heated in a glass reactor with a magnetic stirrer
IKA RET control-visc (sensitivity ±0,5°C) until the temperature was reached. During the deposition
temperature and stirring was controlled by the magnetic stirrer and always stable while pH was
constantly measured with a METTLER TOLEDO pHmeter S400 (Figure 4.).

75
 Figure 4.15: Experimental set-up for plating deposition

4.2.5 Nanocomposite coatings (Ni-nAl 2 O 3 )

The choice to use electroless deposition as a vehicle for nanoparticles dispersion, resides essentially
in the fact that the metal deposition mechanism itself allows the incorporation of finely dispersed
particles into solution. The ion diffusion flow in solution to the substrate allows the capture of
particles that are embedded within the metal matrix (Figure 4.).

Figure 4.16: Deposition Mechanism. Diffusion layer can transport the nanoparticles to the substrate and then
embedded them into the matrix
76
 The nanoparticles, when introduced into aqueous solution, are able to charge superficially by creating
an electric-double layer with the ions in solution. The intensity of this electrostatic field is defined by
the Z potential. Depending on the pH of the solution in which this double layer of ions is found, the
repulsion of two adjacent particles can be favored: if repulsion is favored, it will be said to be a stable
suspension. If the Z potential is not sufficiently high, the particles will tend to agglomerate and
superficial modifications will be required to ensure the stability of the suspension.
Figure 4. shows the trend of the Z potential of α-Al2O3 particles in aqueous solution. Given the high
potential in the range of pH at which the depositions take place (pH between 10.9 and 10.3), it was
decided not to functionalize the nanoparticles.

Figure 4.17: Z potential vs pH for α-Al2O3 in water solution [26]

The 50 nm diameter nanoparticles were purchased by Io-Li-tec and presented in anhydrous form. The
powder was then put into an aqueous solution at concentrations ranging from 25 to 37.5 g / l and
sonicated with a Fisher Scientific 505 tip ultrasonic sonicator with 20% intensity for 10 minutes
before introduction into the plating solution. In this way, by exploiting the cavitation generated by
the ultrasound, the agglomerates are broken and a homogeneous suspension is formed. Once the
dispersions were prepared, 20 ml of this dispersion per liter of solution was added to the plating

77
solution so that the starting solution was not excessively diluted. In order to obtain the nanocomposite
onto the superficial part of the coating, the nanoparticles suspension was added after a certain time
of pure nickel deposition.

4.3 Slurry aluminization

The reactivity of the nickel deposited by electroless plating during phase 2 was tested using the slurry
aluminization technique. Before the heat treatment, Al slurry was air-sprayed on top of the substrate
and the quantity, in terms of mg/cm2, was optimized in order to obtain a thickness of bond coat
comparable to the coatings obtain during phase 1 (20 µm of b-NiAl).
Aluminum microparticles (Hermillion, France - 5 µm of diameter; Figure 4.) were mixed with a
binder before to being air-sprayed.
The water-based binder is composed of 90 wt.% of ultra-pure water (R=18.2 M .cm-1) and the
remaining 10 wt.% of polyvinyl alcohol (PVA) produced by MERCK (CAS n°: 9002-89-5). The
water soluble PVA powder was added to water with some drops of EtOH in order to increase the
solubility. Then the solution was heated (85°C, close to the glass-transition temperature of PVA)
under magnetic stirring to allowing the polymer to dissolve faster. Thereafter, the Al particles (load
agent) was added to the cold binder, in a pill box with the proportions of 50:50, and finally shaken
mechanically for obtaining the homogeneity.
The slurry was, then, sprayed on top of the substrates by the use of an aerograph and the slurry was
dried leaving the samples exposed to air (Figure 4.).

Figure 4.18: Aluminum microparticles purchased from Hermillion (France) on the left; aerograph on the right

In order to achieve a high reproducibility for the quantity of slurry coated on top of the substrates, the
sprayings were conducted by the use of a rotator disc: the pressure (1 bar), the rotation speed and the
78
diameter of the nozzle were optimized to obtain 5 ± 1 mg/cm2 of aluminum microparticles on top of
the substrates.
This treatment is composed of three temperature steps, which allow the removal of the binder
(400°C), the Al inward diffusion (700°C) and finally, the topcoat consolidation and the
homogenization of the intermetallic phases of the substrate enriched in Al (1050°C).
Before the aluminization of the electroless nickel samples, the heat treatment procedure was studied
for pure nickel samples. This has allowed to obtain a temperature profile that would allow the
formation of an optimal bond coat without lengthening the diffusion times of aluminum.
The complete heat treatment is carried out in a quartz tubular furnace under flowing Ar(g) atmosphere
(300 ml/min), which contains enough oxidizing species (2 ppm H2O) to oxidize the aluminum
material. In addition, evaporation of the binder shall lead to H2O vapour and CO2 that may also
oxidize the Al particles. Before treating the samples, a pump allows to realize a partial vacuum (2.10-
1 mbar) before filling the chamber with Ar(g).

References

[1] S. Bose, et al.: High temperature Coatings, Rensselaer Polytecnic Hartford, CT, 2007
[2] M. Okazaki, Sci. Technol. Adv. Mater., vol. 2, no. 2, pp. 357–366, 2001.
[3] X. Peng, T. Li, and W. P. Pan, vol. 44, pp. 1033–1038, 2001.
[4] [Link], Y. Guan, Z. Dong, C. Xu, and F. Wang, Corros. Sci., vol. 53, no. 5, pp. 1954–1959,
2011.
[5] X. Tan, X. Peng, and F. Wang, Surf. Coatings Technol., vol. 224, pp. 62–70, 2013.
[6] Z. D. Xiang and P. K. Datta, J. Mater. Sci., vol. 38, no. 18, pp. 3721–3728, 2003.
[7] Y. bo Zhou, H. yu Chen, H. jun Zhang, and Y. dong Wang, Trans. Nonferrous Met. Soc.
China (English Ed., vol. 18, no. 3, pp. 598–602, 2008.
[8] A. Brenner and G. E. Riddell, J. Res. Natl. Bur. Stand. (1934)., vol. 37, no. 1, p. 31, 1946.
[9] G. O. Mallory, J. B. Hajdu, A. Electroplaters, S. F. Society, and K. (Firm), Int. Bus., 2000.
[10] G. O. Mallory and J. B. Hajdu. (1990), AESF.
[11] C. R. K. Rao and D. C. Trivedi, Coord. Chem. Rev., (2005) vol. 249, no. 5–6, pp. 613–631.
[12] M. Schlesinger and M. Paunovic, (2010) vol. 4, no. 3. wiley.
[13] J. E. A. M. Van Den Meerakker, J. Appl. Electrochem., (1981), vol. 11, no. 3, pp. 395–400,.
[14] Chen, H., Xu, C., Chen, C., Zhao, G. & Liu, Y. Mater. Res. Bull. (2012), 47, 1839–1844.
[15] Park, J. W. et al.. Mater. Chem. Phys (2006). 97, 371–378.
[16] Djokić, S. S. & Djokić, N. S., J. Electrochem. Soc., (2011) 158, D204–D209.

79
[17] Djokic., J. Electrochem. Soc, (1997), 144, 6–11
[18] Genova, V., Marini, D., Valente, M., Marra, F. & Pulci, G, Chem Eng Trans. (2017), 60, 73–
78.
[19] Arai, S., Kobayashi, M., Yamamoto, T. & Endo, Solid-State Lett. (2010) 13, D94.
[20] Haag, S., Burgard, M. & Ernst, B. Surf. Coatings Technol. (2006), 201, 2166–2173.
[21] Cheng, Y. S. & Yeung, K. L, J. Memb. Sci. (2001), 182, 195–203.
[22] Bulasara, V. K., Thakuria, H., Uppaluri, R. & Purkait, M. K. Desalination, (2011), 268, 195–
203.
[23] Bulasara, V. K., Thakuria, H., Uppaluri, R. & Purkait, M. K. Desalination, (2011), 275, 243–
251.
[24] S. N. Jenq, H. W. Yang, Y. Y. Wang, and C. C. Wan, Mater. Chem. Phys., (1997), vol. 48, no.
96, pp. 10–16,.
[25] S. N. Jenq, H. W. Yang, Y. Y. Wang, and C. C. Wan, J. Power Sources, (1995), vol. 57, no.
1–2, pp. 111–118,.
[26] F. Tang, T. Uchikoshi, K. Ozawa, and Y. Sakka, Mater. Res. bullettin, (2002),vol. 37, pp. 653–
660,

80
CHAPTER 5

Experimental results

5.1 Oxidation and hot corrosion resistance

5.1.1 Samples obtained with NH4F as activator salt

In Figure 5.1 is reported the weight gain, normalized for each sample surface, for the specimens
obtained by VPA with NH4F as activator salt. Contrary to what was expected, in the deposition called
NF2, weight gain is lower than the NF1 deposition despite a higher quantity of activator salt.

Figure 5.1: Weight gain, normalized for the samples surfaces, for specimens aluminized by VPA with NH4F as
activator salt

Cross-section SEM analysis have confirmed the different weight gain: the thickness, measured
directly by SEM images, is different for the two samples and, as expected, is higher in the case of
NF1 sample, as shown in Figure 5.2. With “thickness” is considered the total of β-NiAl+IDZ

81
Figure 5.2: Cross-section SEM images of samples obtained by VPA with NH4F as activator salt. a) NF1; b) NF2

In Tabe 5.1 are collected the features, in terms of thickness, for the samples obtained by aluminization
with NH4F as activator salt.

Table 5.1: Features obtained from depositions with NH4F as activator salt

β-NiAl thickness IDZ thickness

Deposition
[µm] [µm]

NF1 23 (±1.0) 21 (±0.9)

NF2 20 (±0.6) 19 (±0.7)

Top-view SEM analysis has revealed the typical polyhedral structure of β-NiAl lattice (Figure 5.3).
As it is possible to observe in the pictures, there are no great differences between NF1 sample and
NF2 sample (except for the different grain size dimension).

Figure 5.3: Top-view SEM images of samples obtained by VPA with NH4F as activator salt. a) NF1; b) NF2

82
XRD analysis has confirmed the presence of β-NiAl phase as the only crystalline structure onto the
coating surface. The only two peaks in the pattern can be indexed as the (1 1 0) and (1 1 1) planes of
the cubic phase NiAl with lattice constant a = 2,8770 Å, which is comparable with the literature data
(JCPDS No. 20-0019, a = 2,8870). Furthermore, it is clear that the relative intensities of the face (1
1 1) to the face (2 0 0) is higher than the conventional value. This is probably due to the epitaxial
growth of the (1 1 1) phase onto a substrate which already had a high orientation (the superalloy was
directionally solidified)

Figure 5.4: XRD pattern for the sample NF1

5.1.2 Samples obtained with AlF3 as activator salt

In Figure 5.5 is reported the weight gain, normalized for each sample surface, for the specimens
obtained by VPA with AlF3 as activator salt. It is notable that the weight gain is higher than the NF
samples (about 5 mg/cm2 for the AlF1, the sample with lower quantity of activator salt).

83
Figure 5.5: Weight gain, normalized for the samples surfaces, for specimens aluminized by VPA with AlF3 as activator
salt

The different weight gain is confirmed by thickness measurement performed with cross-section SEM
analysis for the AlF samples, reported inFigure 5.6. The coating thickness, in terms of β-NiAl+IDZ,
is higher for the sample AlF2.
This result can be explained by considering the different quantities of aluminum available on the
super-alloy surface as diffusion occurs. Ammonium fluoride sublimes at 100°C and form ammonia
and fluoridric acid. Then the HF can react with aluminum to form the fluoride. In the case of
aluminum fluoride, the reactions are different: Menz et al.1 has shown that AlF3 loses water traces up
to 600°C. As a result of the hydrolysis not only HF, but also gaseous HF-AlF3 complexes are formed.
Therefore, aluminum on the super-alloy surface can derive from packs and from activator salts as
well. This can explain why, in the case of AlF3, the b-NiAl and the IDZ layers are thicker respect to
the NH4F diffusion coatings despite of the same deposition procedure and parameters.

Figure 5.6: Cross-section SEM images of samples obtained by VPA with AlF3 as activator salt. a) AlF1; b) AlF2
84
In Table 5.2 are collected the features, in terms of thickness, for the samples obtained by
aluminization with NH4F as activator salt.

Table5.2: Features obtained from depositions with AlF3 as activator salt

β-NiAl thickness IDZ thickness

Deposition
[µm] [µm]

AlF1 27 (±1.2) 24 (±0.9)

AlF2 30 (±0.9) 25 (±0.8)

Top-view SEM analysis, also in this case, has revealed the the typical polyhedral structure of β-NiAl
lattice (Figure 5.7). The difference with the NF samples lies in the presence of homogenously
dispersed particles onto the coating surface. These products were characterized by EDS analysis
which has identified that these residuals are actually Cr and AlF3 crystallities.

Figure 5.7: Top-view SEM images of the samples AlF1: a) homogeneous dispersed deposited particles b) detail at
higher magnification (5000X) for the deposited particles

5.1.3 Isothermal oxidation

Isothermal oxidation test at 1050 ° C was carried out on the samples of the NF and AlF series, as
described in Section 4.4.1. Besides the four coatings produced was also included in the test a sample
of the substrate, to verify the difference in terms of useful life of the guaranteed by aluminides
component.
Figure 5.8 shows the result of the test, which shows that the substrate is characterized by a linear
oxidation kinetics that, after 25 hours, results in catastrophic oxidation. After 50 hours, there is a loss
of structural material of the component. The aluminized samples instead have a parabolic kinetics of
oxidation, thus forming a protective oxide scale on the surface.
85
Excluding the substrate from the graph, in Figure 5.9 is observed in detail the different behavior of
the standard aluminide coatings in oxidation.
In both deposition families, where a higher activating salt concentration is used, a better oxidation
behavior is observed. Moreover, it is noticed that the NF samples behave considerably better than the
AlF, exhibiting a slower oxidation kinetics.

Figure 5.8: Isothermal oxidation (1050°C) for NF and AlF series. “Not coated” refers to a simple substrate.

Figure 5.9: Isothermal oxidation (1050°C) for NF and AlF series.

86
The reasons for the oxidation behavior of the coatings are to be found in the features of the Al2O3
scale formed superficially. SEM micrographies shown in Figure 5.10 show the differences between
the various protective oxides scales.

Figure 5.10: Cross-section SEM micrographs for the sample after 105h of isothermal oxidation (1050°C): a) NF1; b)
NF2; c) AlF1; d) AlF2

All coatings have a thick oxide layer of about 2 µm. The NF1 and AlF2 coatings show a continuous
oxide scale, but it is not adherent on the entire surface of the sample; furthermore, in AlF2 there is
also the presence of a series of zones where oxygen is percolated through the scale, causing the
internal oxidation of the b-NiAl and consequent depletion of such zones in Al.
The NF2 and AlF1 coatings exhibit the formation of an adherent and continuous oxide scale; is the
clear difference in thickness to determine the different oxidation behavior of the two coatings.
Therefore, it is apparent that the Al2O3 layer formed by the AlF1 coating has a higher permeability
by the O2.
In any case, the oxide scale formed by the NF1 coating, the best in terms of gain weight in oxidation,
shows undesirable characteristics. Oxide, in fact, is adherent, cohesive and continuous, but on its
surface has small aggregates of Al2O3, probably detached during cooling or residues of transitional

87
oxides. Finally, Figure 5.10 also shows how IDZ changes as a result of the heat treatment due to the
diffusion between Ni and Al.
EDS analysis of the cross-section of oxidized coatings was performed on all samples. Figure 5.11
shows the map performed on the NF2 coating after 105 hours of oxidation at 1050 ° C; the elemental
distribution is also representative for the other analyzed compositions.

Figure 5.11: EDS maps of the sample NF2 after 105h of oxidation at 1050°C; a) cross-section micrograph of the
investigated area; b) Ni; c) Al; d) O; e) Ti; f) Hf

88
At the top of the sample, at the oxide scale, a rich area of Al and O is observed (Figure5.11-d),
confirming the presence of Al2O3. The EDS also reveals the presence of a rich Ti area and, in much
smaller amounts, of Hf at the bottom of the Al2O3 scale (Figure 5.11e-f): this zone is the result of the
outward diffusion of these two elements of alloy from the substrate to β-NiAl.

Figure 5.12: Surface structure evolution during oxidation for NF2 sample: after 25h (red spectrum), 50 h (green
spectrum), 75h (blue spectrum) and 105h (purple spectrum)

Figure 5.12 shows the surface structure evolution during oxidation for NF2 sample. XRD spectra
confirms the results obtained by EDS with the presence of the stable phase of a-Al2O3. Thanks to
this analysis, it was possible to understand the kinetic of diffusion for the heavier element in the
superalloy. It is observed, in fact, that the Ti and Hf diffuse into the surface already after 50h of high
temperature treatment. In Figure 5.13 is presented the zoom for the most intense peak of NiAl (refers

89
to (1 1 1) phase. It is evident that the formation of the oxide scale causes the NiAl peak to shift to
higher degrees

Figure 5.13: Zoom of the NiAl most intense peak for the XRD spectrum of NF2 sample.

Surface micrographies of oxidized coatings show the presence of cavities on the oxide layer that
covers the samples. These are generated during cooling, as the thermomechanical stresses occurring
at the ceramic oxide interface and intermetallic coatings are such as to cause the release of protective
oxide zones. Figure 5.14 shows the presence of a cracked and discontinuous oxide.

Figure 5.14: SEM micrograph of a cavity on the NF2 coating

The EDS map in Figure 5.15 confirms that these zones are oxide cavities that expose the underlying
b-NiAl oxidation.

90
Figure 5.15: EDS maps of the NF2 coating after 25 hours of oxidation at 1050 ° C: a) SEM micrograph of the
investigated area; b) Ni; c) Al; d) O

With the progress of oxidation these cavities are closed thanks to the formation of new oxide and
thickening of the layer. Figure 5.16 shows the surface appearance of the NF2 coating at 25-hour
oxidation intervals.

91
Figure 5.17: Surface micrographs of the NF2 coating after oxidation at 1050 ° C: a) appearance after 25 hours; b)
after 50 hours; c) after 75 hours; d) after 105 hours.

Coatings obtained with NH4F as an activators salt exhibit better oxidation behavior; in particular,
NF2 has an increase in weight below NF1, with a more protective oxide scale. It was then chosen to
carry out a further coating with a higher activator salt concentration in order to see if an increase in
this process parameter would also improve the oxidation resistance. In Table 5.3 is reported the
concentration of activator salt present in depositions NF2 and NF3.

Table 5.3: Concentrations of activator salt for the deposition NF2 and NF3

wt% of activator mol of activator

Deposition
salt salt

NF2 0.75 0.0201

NF3 1.00 0.0273

The NF3 sample was subjected to an isotermal oxidation test at 1050 ° C. Figure 5.18 shows the
comparison between NF3 and NF2.

92
Figure 5.18: Isothermal oxidation at 1050°C for the sample NF2 and NF3

It is noted that the increase in concentration of activator salt has no benefit in terms of oxidation
performance; it is concluded that there is a concentration threshold of the activating salt beyond which
the benefits of the high temperature oxidation coating reach a plateau.

5.1.4 Oxidation kinetic

The thermally grown alumina is a dense, protective oxide layer, whose growth is controlled by the
diffusion of anions and cations through the layer itself. Its growth kinetics is therefore parabolic, ie
the oxidation time is proportional to the square of the sample weight gain.
Generally, the model used to describe the oxidation kinetics of these samples is the Wagner model.
Wagner's theory describes the growth of an oxide scale on a metal, where ion and electron diffusion
controls the speed of the whole process, with the equation:

𝑑𝑥 𝑘 ′
=
𝑑𝑡 𝑥
where x is the thickness of the oxide or the weight of the material and k 'is the parabolic constant of
growth (cm2/s or mg2/s), correlated with the diffusion rate of species in the oxide layer. The integrated
form of the equation is:
𝑥 ' 𝑡 − 𝑥 ' 𝑡D = 2𝑘 ′ 𝑡 − 𝑡D
where t0 is the time at which diffusion control begins.

93
However, this model hypothesizes the oxidation of a purely metallic sample without the presence of
a surface oxide. In practice, there is always a layer on the surface of the -NiAl, which is formed by
passivation of the material at the end of the deposition process or by thermal growth during oxidation.
A kinetic model that takes into account all of these aspects is that proposed by Monceau [5.2], in
which the independent variable becomes the weight gain of the material (or thickness of the oxide).
This model is described by the equation:

1 '
𝑡 = 𝐴 + 𝐵𝑥 + 𝑥
𝑘4
where parameters A and B contain information about the initially oxide layer (t = 0), while the
coefficient kP is the parabolic growth constant (cm2/s or mg2/s). A regression of experimental data
has been made with both models in order to assess which of the two follows them more accurately
(Figure 5.19-Figure 5.28).

Figure 5.19: Sample NF1 – oxidation kinetic, regression with Wagner’s parabolic model

94
Figure 5.20: Sample NF2 – oxidation kinetic, regression with Wagner’s parabolic model

Figure 5.21: Sample NF3 – oxidation kinetic, regression with Wagner’s parabolic model

95
Figure 5.22: Sample AlF1 – oxidation kinetic, regression with Wagner’s parabolic model

Figure 5.23: Sample AlF2 – oxidation kinetic, regression with Wagner’s parabolic model

96
Figure 5.24: Sample NF1 – oxidation kinetic, regression with Monceau’s parabolic model

Figure 5.25: Sample NF2 – oxidation kinetic, regression with Monceau’s parabolic model

97
Figure 5.26: Sample NF3 – oxidation kinetic, regression with Monceau’s parabolic model

Figure 5.27: Sample AlF1 – oxidation kinetic, regression with Monceau’s parabolic model

98
Figure5.28: Sample AlF2 – oxidation kinetic, regression with Monceau’s parabolic model

To effectively compare the two models, Figures 5.29-Figure 5.33 show the regressions with the
Monceau parabolic model with inverted axes.

Figure 5.29: Sample NF1 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

99
Figure 5.30: Sample NF2 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

Figure 5.31: Sample NF3 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

100
 Figure 5.32: Sample AlF1 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

Figure 5.33: Sample AlF2 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

The regression curves obtained with the Monceau parabolic model have R correlation coefficients
greater than 0.9 and therefore seem more suitable for the description of the experimental data. Table
5.5 shows the parabolic constraints extracted from this model.

101
Table 5.4: parabolic constants growth for the samples of the series NF and AlF

kP
Deposition R2
[mg2/s]

NF1 3.55´10-8 0.81

NF2 3.69´10-8 0.91

AlF1 1.30´10-6 0.88

AlF2 3.52´10-7 0.87

The kP calculated with this model roughly approximate the oxidation kinetics; it is noted, in fact,
that the NF coatings exhibit a parabolic constant growth an order of magnitude less than the AlF
coatings.

5.1.5 Hot corrosion

Samples were subjected to hot corrosion tests at 900 ° C, as described in Section 4.4.1. Besides the
four coatings produced was also included in the test a sample of the substrate, to verify the difference
in terms of useful life of the guaranteed by aluminides component.
Figure 5.34shows the result of the test in which the substrate is observed after 200 hours undergoing
a loss of structural material. Aluminized samples, on the other hand, exhibit significantly greater
corrosion resistance due to the formation of a protective oxide scale on the surface.

Figure 5.34: Hot corrosion test at 900°C for 100h for the series NF and AlF. “Not coated” refers to a substrate not
aluminized

102
Figure 5.35 shows the first 100 hours of the test, in which the substrate was excluded from the graph.
For the first 50 hours, a slower oxidation kinetics is observed for the NF2 coating; however, all
specimens exhibit an alternation between increases and losses in weight, due to the localized attack
of corrosion salts. In fact, they insist on a very small area of the sample, greatly accelerating the
oxidation phenomenon. This causes a very high increase in the protective scale, which is unstable
after a few hours of testing.

Figure 5.35: Hot corrosion test at 900°C for 105h for the series NF and AlF

Figure 5.36 shows the XRD analysis performed after 100 hours of testing at 900 ° C

Figure 5.36: XRD spectra after 100h of hot corrosion test for the samples NF1(red) and AlF2 (blue)

103
From the spectra, it can be noticed the absence of peaks corresponding to the initial composition of
the corrosion salts, meaning that the salt mixture is fully reacted according to a very fast reaction
kinetics. The presence of some peaks due to corrosion products, in particular Ni2S, is observed, which
results in a high surface deterioration. The presence of Ni3Al denote the depletion of Al of the phase
b and subsequent phase transformation: therefore, corrosion, after 100 hours, is already in advanced
stage. Figure 5.37 shows the first 150 hours of the test. At 100 hours, the only coating that has not
suffered material losses is AlF2. However, in the next cycle there is a sharp loss of material, as well
as the NF2. From 110 and 150 hours onwards all samples maintain a roughly stable weight

Figure 5.37: Hot corrosion test at 900°C for 150h for the series NF and AlF

Figure 5.38 shows the full test run. The NF1 and ALF2 samples suffer a loss of structural material, a
symptom of the end of the coating life. The NF2 and AlF1 coatings have a stable weight, but below
the starting weight, meaning that the coating has been subjected to substantial oxidation and
subsequent release of oxide, but it is still protective.

Figure 5.38: Hot corrosion test at 900°C for 100h for the series NF and AlF

104
To investigate the consequences of the lack of homogeneity of the distribution of corrosion salts on
the surface of the sample, superficial EDS analysis were performed to the center and to the external
zone of the sample. Figure 5.39 shows the center of the sample, the area where the coating has
undergone a more intense attack.

Figure 5.39: EDS maps in the attacked zone (centre of the sample) after 100h at 900°C; a) SEM micrograph of the
investigated area; b) Ni; c) Al; d) O; e) Cr; f) Co

Figure 5.39b-d denotes the presence of a cavity at the center of the sample, where the Ni substrate
has been exposed to the attack and is bonded to the O: the cubic NiO formations appear in the center
of the sample. Al2O3 is present only in the external zone. Alloy Co and Cr are widespread on the
surface and reacted with oxygen (Figure 5.39e-f).

Figure 5.40shows EDS analysis of the sample external zone, the area where the coating has suffered
a blander attack from the corrosion salts.

105
Figure 5.40: EDS maps of the external zone for the sample NF2 after 100 h of hot corrosion at 900 °C: a) SEM
micrograph of the investigated area; b) Ni; c) Al; d) O.

Figure 5.40b-d shows a continuous layer of a-Al2O3, with the absence of Ni on the surface. This
evidence confirms what it was assumed seen the trend of the weights of the samples during the test:
the samples undergo an inhomogeneous attack of the surface, accelerating to an appreciable extent
the detachment of the protective oxide scale (Figure 5.40e-f).
In conclusion, the diffusion coatings obtained with NH4F as the activator salt show a better oxidation
behavior than those obtained with AlF3; in particular, NF2 has an increase in weight after 105 hours
of oxidation lower than all other samples, characterized by a more protective oxide scale. Through
the Monceau model, the parabolic constants of kP growth have been calculated: NF coatings exhibit
a parabolic growth constant of an order of magnitude less than the AlF coatings. There is also an
activator salt concentration threshold beyond which the oxidation benefits of the coating do not
change.
During the cooling following the oxidation test, the release of protective oxide zones is due to the
different thermal expansion coefficient of the oxide and the undercoat.

106
The hot corrosion test has generated a very accelerated and inhomogeneous oxidation on the surface
of the sample. This is due to the restriction area on which the corrosion salts insist, causing a very
high increase in thickness of the protective scale, which is unstable after a few hours of testing; such
loss of surface material is not found in the external zone of the samples.

5.1.6 Modified aluminide coating with Zr

Modified coatings were obtained using a NH4F / ZrF4 mixed activator salt, in which the concentration
of NH4F (0.75 wt%) was maintained constant while the concentration of ZrF4 was varied in each
deposition, as reported in Table 5.5.

Table 5.5: Concentrations of activator salt for the deposition of ZrF series

wt% of activator mol of activator

Deposition
salt salt

ZrF1 0.121 0.0072

ZrF2 0.145 0.0087

ZrF3 0.165 0.0100

ZrF4 0.335 0.0201

ZrF5 0.670 0.0402

Table 5.6 summarizes the features of the product coatings. From this table, it is possible to see how
modified coatings are considerably thicker than conventional ones, which implies a direct
involvement of ZrF4 in the deposition process.

Table 5.6: Features obtained from depositions with NH4F/ZrF4 mixture as activator salt

β-NiAl thckness IDZ thickness

Deposition
[µm] [µm]

ZrF1 35 (±0.8) 24 (±0.9)

ZrF2 39 (±1.2) 27 (±1)

ZrF3 27 (±1.1) 19 (±0.9)

107
 ZrF4 32 (±1.8) 24 (±1.2)

ZrF5 32 (±0.8) 24 (±0.5)

Beyond the thickness, the coatings are very similar to each other, as shown in Figure 5.41. In addition,
there are no particular differences compared to the conventional coatings obtained during the first
phase of the experiment.

Figure 5.41: Cross-section SEM micrographies for coatings obtained with the NH4F/ZrF4 mixture as activator salt - a)
ZrF1; b) ZrF2; c) ZrF3; d) ZrF4; e) ZrF5.

The quantity and distribution of Zr present within the coating was studied by GDOES technique. The
concentration profiles in weight of the ZrF3 coating are shown in Figure 5.42
; the elemental distribution is also representative of the other analyzed compositions.

108
 Figure 5.42: Concentration profile obtained through GDOES of the main elements in the ZrF3 coating

Figure 5.42 shows the presence of about 25 m of b-NiAl, followed by 20 m of IDZ, as found in SEM
micrographies. In this area, the intensity of W is too high compared to alloy concentration; this
experimental error is due, probably, to the fact that there are some optical interferences with other
elements ( such as Hf and Ta, which possess a very similar atomic weight). Thus, a concentration is
obtained that is the overlap of the three elements. Considering the low concentration of Zr, it is
necessary to modify the y-axis scale to find the peak of the reactive element taken into account.

109
Figure 5.43: Concentration profile obtained through GDOES of the Zr in the ZrF3 coating

The concentration profile of Zr (Figure 5.43) highlights how it is located at the interface between b-
NiAl and IDZ, finding results with literature [1].
Top-view SEM analysis did not show any difference from the coatings obtained during the first phase
of the experimentation: in fact, all the specimens show the polyhedral structure of the b-NiAl. Except
for the ZrF5 deposition, where it is possible to observe the presence of macroscopic crystals on top
of the surface of some specimens of the same batch (Figure 5.44).

Figure 5.44: Surface optical microscope image of a sample from the ZrF5 deposition.

EDS analysis were performed to investigate the chemical compositions of those crystals (Figure 5.45
– Figure 5.46)

110
Figure 5.45: EDS maps of the coating ZrF5, top-view – a) SEM micrograph of investigated area; b) Ni; c) Al; d) F

Figure 5.46: EDS maps of the coating ZrF5, cross-section–a) SEM micrograph of the investigated area; b) Ni; c) Al; d)
F

111
It was hypothesized that these deposits were linked to excessive production of AlF vapors during the
process, which accumulate near the substrate. In fact, the stage determining the deposition rate is the
solid state diffusion of the Al into the substrate; consequently, an excess of halydes of Al causes the
deposition of solid AlF3 near the surface, which at the end of the process remains mechanically
anchored to the surface. The same situation was observed for the sample aluminized with AlF3 as
activator salt. In the case of ZrF4, however, the concentration of gaseous Al halydes is high enough
to prevent the completely diffusion of Al within the coating with a consequent re-crystalization of
AlF3on top of the substrate. It is concluded that for an efficient deposition of the coating the gas phase
produced must be accurately proportioned to the diffusion velocity within the superalloy.

5.1.7 Isothermal oxidation

Isothermal oxidation test at 1050 ° C was carried out on the samples of the ZrF series, as described
in Section 4.4.1. Besides the four coatings produced was also included in the test a sample of the
substrate, to verify the difference in terms of service life of the aluminides component.
Figure 5.47 shows the result of modified coatings compared to the best specimen of standard diffusion
coatings. From the Figure 5.46 is possible to see that all modified coatings have a parabolic oxidation
kinetics and after 105 hours they show a better oxidation resistance then the standard coating.

0,6

0,5

NF1
Weight gain [mg/cm2]

0,4 ZrF1
ZrF2

0,3 ZrF3
ZrF4
ZrF5
0,2

0,1

0,0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time [h]

Figure 5.47: Isothermal oxidation (1050°C) for ZrF series.

From the previous figure, it is shown how the ZrF1 and ZrF5 coatings, which have respectively the
minimum and maximum amount of starting activators, are those that are a higher weight amount
112
between the modified diffusing coatings. However, it is noted that their oxidation kinetics is in any
case is comparable to that of NF1. Figure 5.48 shows ZrF coatings which show a higher oxidation
resistance than NF1. It is noticed that adding a reactive element in the right amount leads to a decrease
in weight gain by almost 20%.

Figure 5.48: Isothermal oxidation (1050°C) for ZrF2, ZrF4 in comparison with NF1

Figure 5.49 shows SEM micrographs after 100 hours of isothermal oxidation of ZrF3, ZrF4 and ZrF5
coatings. There is a clear note of the presence of a protective oxide scale of a-Al2O3 and the
differences between the various protective oxides formed can be observe.

113
Figure 5.49: Cross-section SEM micrographs for the sample after 105h of isothermal oxidation (1050°C): a) ZrF3
2000X; b) ZrF3 5000X; c) ZrF4 2000X; d) ZrF4 5000X; e) ZrF5 2000X; f) ZrF5 5000X

The difference between the three coatings is in the thickness of oxide formed: the ZrF 4 (Figure 5.49c-
d), which shows the best oxidation resistance, after 105 hours is covered with an oxide layer of about
1.5 microns. The ZrF5 (Figure 5.49e-f), on the other hand, is characterized by a 2.7 µm protective
oxide, which makes it the coating that gain more weight at the end of the test. Finally, the ZrF3
(Figure 5.49a-b), which exhibits an intermediate behavior between the two, has a protective oxide
scale of about 2.3 µm.
EDS analysis of the cross-section of oxidized coatings was performed on all samples. Figure 5.50
shows the maps for the oxidized ZrF5 coating for 105 hours at 1050 ° C; in particular, the presence
114
of heavy elements around the scale is observed, diffused by the IDZ trough the bond coat and up to
the oxide scale. Zr seems to have the same role of the RE already present in the chemical composition
of the super alloy. As is known2, the reactive elements increase the resistance to oxidation by
increasing the adhesion of the protective scale. The dynamic segregation theory states that the
diffusion mechanism of the reactive elements takes place through the grain boundaries of the oxide
scale up to the interface with the gaseous phase. In this case, it was seen that Zr can play a different
role in terms of oxidation resistance: in fact, before to reach the interface between the bond coat and
the oxide scale, the Zr has blocked the oxygen path through the scale, by avoiding the penetration
inside the bond coat.

115
Figure 5.50: EDS maps of the coating ZrF5, top-view, after 105h of oxidation test – a) SEM micrograph of investigated
area; b) Ni; c) Al; d) O; e) Hf; f) Zr

Zr increases the oxidation resistance of modified coatings. The hypothesis is that Zr limits the
release of protective oxide during cooling, suppressing the formation of Kirkendall's voids at the
interface. According to the Kirkendall mechanism, the selective oxidation of Al causes a depleted
area of this element to the oxide-metal interface, which induces opposite atomic flows: the Ni
present in this area diffuse to the superalloy and while the Al diffuse to this depleted area. Since Ni
is characterized by a higher diffusion velocity than Al, reticulate vacations are generated which, by
coalescence, form void in the oxide-metal interface, which reduce the adhesion of the scale. Figure
5.51-Figure 5.52 shows EDS spectra of NF2 and ZrF4 coatings; it is highlighted that in both cases
116
there is the presence of heavy element (reactive element) aggregates on the oxide-coating interface.
In the case of NF2, the presence of Hf is observed, while ZrF4 also has Zr, confirming the
interaction between this element and the diffusion of Al, Ni and O.

Figure 5.51: EDS spot analysis on NF1 sample after 100h of oxidation. SEM micrograph of the investigated area

117
Figure 5.52: EDS spot analysis on AlF1 sample after 100h of oxidation. SEM micrograph of the investigated area

Figure 5.53 shows the surface structure evolution during oxidation for ZrF5 sample. XRD spectra
confirm the results of the EDS maps. As it was seen for the standard coatings, after 50 hours of
oxidation at high temperature, Hf and Ti diffuse trough the bond coat to the surface.

118
Figure 5.53: XRD Surface structure evolution during oxidation for ZrF5 sample: after 25h (red spectrum), 50 h (green
spectrum), 75h (blue spectrum) and 105h (purple spectrum)

As it was seen for standard diffusion coatings previously tested, surface micrographs show the
presence of cavities on the oxide layer generated during cooling. Figure 5.54 shows the presence of
cracked and discontinuous oxide.

119
Figure 5.54: SEM micrograph of a cavity on the ZrF5 coating

Also in the case of Zr-doped coatings, the oxidation process allows these cavities to close. Figure
5.55 shows the change in the surface appearance of the ZrF coating at 25-hour oxidation intervals.

Figure 5.55: Surface micrographs of the ZrF5 coating after oxidation at 1050 ° C: a) appearance after 25 hours; b)
after 50 hours; c) after 75 hours; d) after 105 hours.

During oxidation tests, modified coatings show the oxidation resistance depending by the activator
salts concentrations (Figure 5.56).
120
Figure 5.56: Weight gain in oxidation tests as a function of Zr content for modified aluminide coatings

Thus, a range of concentrations has been identified within which the addition of Zr improves the
performance of the standard coating. For concentrations lower than the optimum values, the effects
generated by the presence of the reactive element are invalid or minimal; in fact, the ZrF1 coating
has the same performance as NF2. For addition above the optimum values, the oxidation rate
increases as the reactive element reactor formed oxides allow a faster diffusion of the oxygen into the
bulk and / or determine heterogeneous interfaces within the oxide scale where the diffusion velocity
of the oxygen can be higher. The ZrF2, ZrF3 and ZrF4 coatings have a kP parabolic constants of
magnitude smaller than ZrF1 and ZrF5.

5.1.8 Oxidation kinetic

Also for Zr-doped coatings a regression of experimental data was made with the models of Wagner
and Monceau. The regressions obtained are shown in Figure 5.57-Figure 5.66.

121
Figure 5.57: Sample ZrF1 – oxidation kinetic, regression with Wagner’s parabolic model

Figure 5.58: Sample ZrF2 – oxidation kinetic, regression with Wagner’s parabolic model

122
Figure 5.59: Sample ZrF3 – oxidation kinetic, regression with Wagner’s parabolic model

Figure 5.60: Sample ZrF4 – oxidation kinetic, regression with Wagner’s parabolic model

123
Figure 5.61: Sample ZrF5 – oxidation kinetic, regression with Wagner’s parabolic model

Figure 5.62: Sample ZrF1 – oxidation kinetic, regression with Monceau’s parabolic model

124
Figure 5.63: Sample ZrF2 – oxidation kinetic, regression with Monceau’s parabolic model

Figure 5.64: Sample ZrF3 – oxidation kinetic, regression with Monceau’s parabolic model

125
Figure 5.65: Sample ZrF4 – oxidation kinetic, regression with Monceau’s parabolic model

Figure 5.66: Sample ZrF5 – oxidation kinetic, regression with Monceau’s parabolic model

126
To effectively compare the two models, Figure 5.67- Figure 5.71 show regressions with Monceau
parabolic model with inverted axes.

Figure 5.67: Sample ZrF1 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

Figure 5.68: Sample ZrF2 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

127
Figure 5.69: Sample ZrF3 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

Figure 5.70: Sample ZrF4 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

128
Figure 5.71: Sample ZrF5 – oxidation kinetic, regression with Monceau’s parabolic model with inverted axes

The regression curves obtained with both models have R correlation coefficients greater than 0.9;
however, for each data set, Monceau's parabolic model is characterized by greater R than the Wagner
model, and therefore it is more appropriate to describe the experimental data. Table 5.6 shows the
parabolic constants extrapolated from the Monceau’s model. For completeness, kP is compared with
the NF2 sample.
Table 5.6: Parabolic constants growth for the samples of the series ZrF compared with the NF2 sample

kP
Deposition R2
[mg2/s]

NF2 3.69´10-8 0.91

ZrF1 1.26´10-7 0.95

ZrF2 9.35´10-8 0.98

ZrF3 9.77´10-8 0.97

ZrF4 4.79´10-8 0.89

ZrF5 3.29´10-7 0.99

The kP calculated with this model roughly approximate the oxidation kinetics; it is noted that ZrF2,
ZrF3 and ZrF4 coatings have a parabolic constant growth rate of 10-8 mg2 / s, while ZrF1 and ZrF5
129
coatings, which have shown lower oxidation performance than these three, are characterized by a kP
of the order of 10-7 mg2 / s. The growth constants of the three modified coatings with the best
oxidation performance are of the same order of magnitude, meaning that oxygen permeates the oxide
scale at the same speed.

5.1.9 Hot corrosion

Samples were subjected to hot corrosion tests at 900 ° C, as described in Section 4.4.1. Besides the
five coatings produced was also included in the test a sample of the substrate, to verify the difference
in terms of useful life of the guaranteed by aluminides component as done for the standard coatings.
Even in the case of doped coatings, significant weight loss is observed after 100 hours of hot
corrosion. This is definitely due to the fact that the amount of salt is added as a refill every 14-hour
cycle, as described in the paragraph (reference). This creates a very corrosive environment that allows
to make predictions on the durability of coatings in less extended time.
Figure 5.72 shows the result of the test for the first 100h. The weight gain, as expected, is similar to
the standard coatings because reactive element doping doesn’t influence positively but neither
negatively the corrosion resistance.
15,0
NF 1
ZrF 1
ZrF 2
ZrF 3
ZrF 4
Weight change [mg/cm2]

10,0
ZrF 5

5,0

0,0
0 10 20 30 40 50 60 70 80 90 100
Time [h]

Figure 5.72: Hot corrosion test for sample of the ZrF series; to compare the results with phase 1, the results for NF1
were added

130
5.2 Phase 2: Electroless pure nickel plating

The results of the phase 2 are presented in this section. The aim was to obtain an external pure nickel
layer by electroless nickel plating technique. SEM analysis were carried out by a FEG-SEM ZEISS
(Carl Zeiss Microscopy, Oberkochen, Germany) Auriga 405 Scanning Electron Microscope coupled
with its energy dispersive spectrometer. Samples were mounted in epoxy resin, cut and polished with
SiC papers and diamond suspension up to 1 µm.
X-Ray diffractometry (XRD) was performed with a Philips X’Pert device (PANalytical B.V., Almelo,
The Netherlands) on the samples in order to analyze the phase structure and the purity of the coatings
after deposition. The XRD device operated at 40 KV and 40 mA with CuKa1 radiation (lKa1 =
1.540598 Å, l Ka2= 1.544426 Å) with a scan range of 30-80° (2q), step size of 0.02° and counting
time of 2 sec.
The formation of the nickel complex was studied by Fourier-transform infrared (FTIR) analyses and
carried out with a Bruker Vertex 70 spectrometer (Bruker Optik GmbH) equipped with a single
reflection Diamond ATR cell. The ATR-FTIR spectra were recorded with 256 scans in the mid
infrared range (400–4000 cm−1) at a resolution of 4 cm−1.

5.2.1 Solution A

Chemical composition of solution A is reported in Table 5.7

Table 5.7: Chemical composition for solution A

g/l (mol(l)

Nickel acetate Ni(OCOCH3)2 21,2 (0,08)

Hydrazine N2 H4 20,02 (0,4)

Na2EDTA Na2EDTA 5,62 (0,016)

Lactic acid C3H6O3 13,5 (0,15)

Sodium hydroxide NaOH 9,87 (0,247)

Starting from the solution A, a highly stressed coating was produced that tends to detach, especially
near the edges of the sample, as can be seen in Figure 5.73 and as reported by the SEM micrograph
in Figure 4.2 .

131
Figure 5.73: As-coated sample plated with the solution A

Following the analysis of the cross-section, we measured the thickness of the coating, in the order of
5 µm, obtained with 1 h of deposition.

Figure 5.74: Top-view SEM micrographs of a sample plated with the solution A

Also, as shown in Figure 5.74, there are some cracks that pass through the thickness of the sample
and porosity to the metal-coating interface, a sign of poor adhesion and detachment following the
polishing process.

132
Figure 5.75: Cross-section micrograph for a sample obtained with the solution A

Considering the poor adherence of the coating, reagent concentrations were varied to obtain a more
adherent and less stressed coating: at first the amount of NaOH was increased so as to increase the
pH of the solution to a value of 9.2. In contrast to what was reported in the literature, the increase in
pH did not increase deposition rate but rather a decrease in this. Thus, the degree of Ni2+ ions
complexation was varied by acting on the concentrations of the complexes so as to decrease the nickel
reduction rate with the aim of decreasing the internal stress. In fact, considering the complexes present
in excess of nickel, the free Ni2+ ions concentration can be obtained from the nickel complexation
equilibrium:
𝑁𝑖 'B + 𝐸𝐷𝑇𝐴 ↔ 𝑁𝑖𝐸𝐷𝑇𝐴
with a stability constant given by:
𝑁𝑖 − 𝐸𝐷𝑇𝐴
𝐾k8>lmno =
𝑁𝑖 'B 𝐸𝐷𝑇𝐴
Thus, the concentrations of Ni2+ ions are:
𝑁𝑖𝐸𝐷𝑇𝐴 𝑁𝑖 'B D
𝑁𝑖 'B = =
𝐾k8lmno 𝐸𝐷𝑇𝐴 1 + 𝐾k8lmno 𝐸𝐷𝑇𝐴
Where [Ni2+]0 is the initial concentration of nickel ions, which can be considered as the concentration
of nickel salt dissolved in the solution. Having in the solution two distinct complexes, four different
trials of 2.5 h were performed, each with a molar concentration of the complexes increased by 15
wt% or 30 wt%. Table 4.2 shows all changes to the starting solution. Variations in the concentration
of the reagents of the solution did not produce appreciable results from the point of view of the

133
internal stresses: the coatings continue to exhibit detached areas as well as poor adhesion as can be
seen in Figure 5.76.

Table 5.8: Experimental matrix for the solution A. “PN_A” stands for “Pure Nickel_solution A”

Lactic
Nichel acetate Hydrazina Na2EDTA NaOH Deposition
Samples acid
[mol/l] [mol/l] [mol/l] [mol/L] rate [𝝁𝒎 𝒉]
[mol/l]

PN_A_03 0,12 0,4 0,016 0,15 0,18 5 ±0,43

PN_A_10 - - - - 0,24 4 ±0,47

0,172
PN_A_13 - - - 0,24 4.54 ±0.51
(+15%)
0,195
PN_A_14 - - - 0,24 -
(+30%)
0,018
PN_A_15 - - - 0,24 2.9 ±0.54
(+15%)
0,021
PN_A_16 - - - 0,24 -
(+30%)
PN_A_03 0,12 0,4 0,016 0,15 0,18 5 ±0,43

The depositions carried out with a higher complexing increases did not produce coatings suitable: the
solution with + 30% of lactic acid has produced a very porous coating, while the prepared solution
with + 30% Na2EDTA is precipitated after 2 hours.

Figure 5.76: Optical microscopy photographs for the sample of the series PN_A. All the variation has no lead to a
good coating in terms of adhesion – a) PN_A_10; b) PN_A_13; c) PN_A_03

The modifications made to the composition of the solution did not produce consistent improvements
in coating quality. The production of stressed coatings, which is not homogeneous and porous with
this particular composition, was also found in other papers available in literature [2]. It seems that the

134
production of these stressed coatings is due to the strategy of using a very strong complexing agent
(EDTA) with a weaker one (lactic acid). Always according to Muench better results are obtained by
adopting a single complexing agent from the constant of intermediate stability between EDTA and
lactic acid.

5.2.2 Solution B

Solution B provides the use of only one complexing agent which at the same time also acts as a
reducing agent: the bath composition is reported in Table 5.9. The coatings obtained with solution B
showed excellent adhesion and absence of internal stresses and higher deposition rate than in case of
solution A. It was possible to produce coatings up to 20 µm with 2 hours of deposition with no
impurities.

Table 5.9: Chemical composition for solution A

g/l (mol(l)

Nickel chloride NiCl2 23,76 (0,1)

Hydrazine N2 H4 32,56 (0,65)

Potassium carbonate K2CO3 69,1 (0,5)

Potassium hydrogen
KHCO3 45 (0,45)
carbonate
Potassium phosphate K3HPO4 87,1 (0,45)

Sodium hydroxide KOH 9,78 (0,3)

According to the principle of electroless plating, Ni ions reacts in stoichiometric quantities with the
reducting agent (hydrazine, in this case) by giving metallic nickel onto the autocatalytic surface. As
already known[2], under alkaline conditions, hydrazine is a stronger reducting agent with a potential
of 1.16 V (referring to the semiconducting cell reaction). This could lead, if the reactivity is not
controlled, to the precipitation of Ni(OH)2 and the consequent decomposition of the plating solution.
Theoretically, the formation of the nickel complex allows to create an “in-situ redox reactor” where
the Ni2+ ions and the hydrazine lie in the same molecular structure. This could be the way to obtain
a controlled reduction of the nickel and to avoid the formation of products that can decompose the

135
plating solution. The crystal structure of the complex reveals that the Ni2+ ions are coordinated by
three ligands of hydrazine carboxylate[3]. The hydrazine carboxylate coordinates the nickel ions in a
chelating system with the atom of N and O represented by the bold curves in Figure 5.77 (the arcs
structure is N' H2 -NH-COO). The presence of this high spin complex is confirmed by the peculiar
blue color of the plating solution.

Figure 5.77: Structure of tris(hydrazine carboxylato-N’,O) nickelate(1-)[3]

The formation of the nickel complex was confirmed by FTIR analysis. It is notable in Figure 5.78
that the presence of the water (dashed red line) does not allow to observe the characteristic peaks of
the nickel complex, preventing its interpretation. For this reason, concentrated plating solution,
prepared as described in the previous paragraph, was left in vacuum drying in order to obtain a solid.
In Figure 5.78 is reported the FTIR spectrum (solid line) for the solid nickel complex salt.

136
 Figure 5.78: FTIR spectrum of tris(hydrazine carboxylato-N’,O) nickelate(1-) salt. Comparison between the
water solution (dashed red line) and the precipitated salt (solid blue line)

The stretching vibrational mode (N—N) of hydrazine is shifted to 1000 cm-1 because the hydrazine
radical H2N’-NH- is attached to a conjugate system[3]. The presence of characteristic peaks, such as
the Ni-N stretching (415 cm-1), the rocking vibration (600 cm-1) and the symmetric distortion of the
ligands in the complex (region between 1000 and 1600 cm-1) are consistent with the values reported
in earlier literature[4],[5],[6],[7]. The characteristic peaks for the nickel complex are listed in Table
5.10

Table 5.10: Characteristic peaks for the nickel complex salt

Assignment Wavenumber (cm-1)

Stretching (N-H) 3335

Stretching (COO-) 1385

Stretching (N2H3-) 1580

Rocking (N2H3-) 1220

137
 Stretching (N-N) 1000

Rocking (N-N) 600

Figure 5.79: Evaluation of thickness for different parameters: pH (a); temperature (b); plating ratio (c). Deposition
rate for the best parameters (d)

The formation of the complex nickel has allowed to control the deposition and to study the deposition
rate. Three different plating parameters were investigated to optimize the plating rate (temperature,
pH and plating ratio). pH was controlled by varying the quantity of KOH in the solution preparation
and the chosen range was 9-12. Higher temperature value for the selected range (70-90°C) was chosen
in order to avoid the concentration of the solution by the solvent evaporation. Plating ratio (PR) is
referred to the quantity of solution per sample surface area (ml/cm2). All the depositions were
conducted with the same composition for the plating solution (except for pH investigations) and for
the same time (60 min). As expected hydrazine reactivity is strongly influenced by the pH. Despite
this, the results presented in Figure 5.79a show that after a pH of 11 the formation of Ni(OH)2 is
relevant and it prevents the reduction of nickel. Temperature influences the reduction of the nickel

138
not only in terms of deposition rate but also on the surface microstructure. In fact, after 85 °C the
coating is thicker but the surface morphology is negatively affected. The fast and not controlled
reduction of the nickel ions involves the formation of a porous coating. Finally, the quantity of nickel
ions in terms of volume solution seems to have a negative influence just for small quantity of volume.
This was predictable but the investigation of PR for a large quantity of plating solution shows that
the surface has a catalytic role for the deposition as expected from the theory [8]. It was seen that
after a threshold (60 ml/cm2 for one sample) the plating solution is more efficient in terms of coating
thickness by increasing the surface available for the deposition. The optimized parameters selected
for the characterization analysis are presented in Table 5.11.

Table 5.11: Optimized parameters for plating solution

Parameters Best value

pH 11

Temperature (°C) 85

Plating ratio (ml/cm2) 50

Deposition rate (µm/60min) 15

The purity and the structure of the as-coated samples were analyzed by XRD (Figure 5.80). The only
three peaks in the pattern can be indexed as the (1 1 1), (2 0 0) and (2 2 0) planes of the fcc cubic
phase Ni with lattice constant a = 3.525 Å, which is comparable with the literature data (ICDD No.
04-0850, a = 3.524 Å). Furthermore, the relative intensity ratio for the face (1 1 1) to the face (2 0 0)
is higher than the conventional value. This was already seen [9] as the consequence of a preferential
orientation of the (1 1 1) crystallographic plane. XRD pattern does not reveal the presence of
impurities such as nickel oxides or hydroxides and confirms that the deposition correctly allows to a
metallic pure nickel coating.

139
Figure 5.80: XRD pattern of the electroless nickel coating.

The surface microstructure was investigated by SEM analysis. Cross-section analysis has allowed to
obtain the direct measurement of the thickness for each sample. Furthermore, as it is shown in Figure
5.81, the cross-section micrographies has revealed a good interface between the substrate and the
coating. Typical top-view morphologies of nickel coating obtained by electroless plating is shown in
Figure 5.82. The surface presents a hierarchical microstructure that goes from the cauliflower-like
structure (Figure 5.82a,b), typical for the electroless coating [10], to a nanoarchitecture made by pine
cone-like shape (Figure 5.82c,d) with an average diameters of 2 µm. Further SEM analysis at higher
magnifications has revealed that this nickel flowers were made by nickel nano-spike with the average
size of 150-200 nm.

Figure 5.81: Cross-section SEM micrograph of electroless pure nickel plating in 2 hours

140
Figure 5.82: Surface microstructure of as-coated electroless nickel with increasing magnification (from a to d).

Specific tests were conducted in order to understand if the formation of this peculiar nano structure
came from the first step of the reaction. Three different plating time (1 min, 5 min and 10 min) were
studied to understand the growing mechanism of the coating. SEM micrographs have shown that for
the first minute of reaction, nickel grows as domain onto the surface (Figure 5.83). It has been
demonstrated [9] by combined HRTEM and SEAD analysis that the single nano-spike of nickel grows
along the preferential orientation (1 1 1) of the nickel phase structure. So, theoretically, this should
mean that the domains, formed in the first minute of the plating deposition, must have time to coalesce
in a layer.

141
Figure 5.83: SEM micrograph for a sample plating in 1 minute

The microstructure for 2 and 5 minutes of plating deposition have confirmed this theory. As it is
possible to see in Figure 5.84 the nano-spike are growing in the sense of the arrows to form the layer
of the coating. The results obtained by the SEM analysis have allowed to confirm the theory of the
growing mechanism as “nickel domain” already formulate in a previous work [11].

Figure 5.84: SEM micrographs for the samples coating in 5(a) and 10(b) minutes. In the figure is possible to see the
coalescence of the micro-flower in a single layer.

The peculiar nanoarchitecture obtained with electroless plating has shown a super hydrophilic
behavior for the sample as-coated. In order to understand the roughness influence on the wettability
of the coating, photographs of water droplet on coating electroless plating with and without polishing
are compared in Figure 5.85. The as-coated electroless plated sample shows a super-hydrophilicity
142
with a water CA <5° (Figure 5.85a). The Ni coating could be described by the Wenzel’s model [12],
according to which the surface roughness should increase the hydrophilicity if the smooth material is
hydrophilic itself. To valuate this, a sample was polished with a 4000 SiC paper to eliminate the
nanostructure obtained with the plating. The sample has shown a water CA of 58,27°, confirming the
hypothesis.

Figure 5.85: Photographs of the contact angle for the samples: a) electroless pure nickel plated as-coated; b)
electroless pure nickel plated polished with P4000 SiC paper

The Ni-coatings became super hydrophobic after being exposed in ambient air for two weeks. The
wetting transition phenomenon was studied by measuring the water contact angle as a function of the
air exposition time. In Figure 5.86 it can be clearly noticed that the evolution of contact angles at
different air exposition time can be divided into three different regions: for the first 24 hours, the
samples are substantially superhydrophilic with contact angle <5° and can be described by the
Wenzel’s model [12]. The water contact angle drastically increase to higher values and reach the
values of about 100° after 100 hours; this means a transition from the super-hydrophilic state (for the
freshly coated) to a hydrophobic state that can be described by the Cassie-Baxter’s model [13]. Then,
after more than 200 hours, the surface of the nickel coating became super-hydrophobic (with contact
angle = 150°) with an excellent water repellent properties. This transition was already seen in
literature for different systems: Ni and Ni-Co electroplated surface [14], [15] that grow with the same
superficial structure and exhibit the same wetting transition after approximately the same time (about
200h). This phenomenon, then, could be theoretically attributed not only to the surface roughness
nanostructure but also to the nature of coating. However, the same transition was observed for
different chemical nature systems: boron nitride nanotubes (BNNTs) [16], aluminum surface [17]
and ZnO nanorod [18], for example.

143
Figure 5.86: Time-dependent evolution of water contact angle for sample coated with electroless pure Ni and exposed
to ambient air

The nature of wetting transition can be attributed to a surface morphological modification or chemical
variation. As described for similar surface roughness1, the simple exposition to the air can’t afflict
the surface morphology in any way. This was confirmed by comparison surface roughness obtained
with AFM analysis (Figure 5.87). Surface roughness can be compared (220 nm for the super-
hydrophilic sample and 234 nm for the super-hydrophobic sample) and there is no evidence of
morphological modifications.

Figure 5.87: AFM images of the as-coated electroless super-hydrophilic sample (a) and the same sample after 200h of
air exposition(b)

144
The chemical modifications of the surface seem to be the only phenomenon that can modify the
wettability. Although it was reported that the surface layer of nickel film went through a process of
oxidation and nickel oxide is formed by exposing the film in air [19], the hydrophobicity of NiO can’t
be attributed as the cause of wetting transition because the presence of oxides onto the surface wasn’t
observe during surface characterizations. Another reason for the hydrophobic state of nano-structured
coating has been related to adsorption of organic molecules [16]. This phenomenon is in accordance
with thermodynamic because the adsorption of molecules onto the surface can decrease the total free
energy of the system by diminishing the surface energy part. So, if the electroless nickel surface has
a great surface energy (given by the super-hydrophilic state), a wetting time dependent transition can
be well explained as a consequence of decreasing surface energy by chemisorption or physisorption
of adsorbates. The super-hydrophobic stability tests have partially demonstrated this theory. In order
to remove chemically adsorbed contaminants onto the surface, all the samples were washed into
several organic solvents. To demonstrate the presence of the organic airborne hydrocarbons as
adsorbates, the solvents were chosen with different and increasing polarity. All the samples have kept
the hydrophobic state without any drastically change for all the solvents except for the ethanol. The
coating washed in ethanol returns to the super-hydrophilic state after 10 min of washing treatment.
The water CA measurements as a function of time performed before and after washing in ethanol are
presented inFigure 5.88. As expected, after the physico-chemically removal of the hydrophobic
adsorbates, the coating state returns to the originally super-hydrophilic and the wetting transition
occurs with the same time-conditions before the washing.

Figure 5.88: Time-dependent evolution of sample water contact angle before (dashed line) and after (solid line)
washing in ethanol at room temperature for 10 minutes.

145
5.2.3 Nanocomposite Ni-nAl2O3

Once the parameters of solution B were optimized, the Ni-aAl2O3 nanocomposite was studied by
introducing a water dispersion of nanoparticles into the plating solution. In Table 5.12 are
summarized the different concentrations studied in this phase of the work

Table 5.12: a-Al2O3 concentrations, in terms of water solution

a-Al2O3 a-Al2O3
Samples
Water solution [g/l] Volume in solution [g/l]

PN_Al_B_09 25 0,5

PN_Al_B_10 37,5 0,75

PN_Al_B_16 30 0,6

The deposition time for all samples containing alumina was 2 hours, and the nanoparticles were added
after an hour of pure Ni deposition. In all cases, the coatings produced have a thickness similar to that
obtained without nanoparticles: this is an indication that the deposition rate is not changed.
In low concentrations (0.5 g / l), a good dispersion was observed, although the number of particles
that can be detected is very low. In fact, as can be seen from Figure 5.89 a,b the particles embedded
in the coating are few, not sufficient to achieve the desired effects. On the contrary, increasing the
concentration up to a value of 0.75 g / l the particles are agglomerated, with dimensions reaching 200
nm. This agglomeration can be seen in Figure 5.89 c,d.
It was then chosen to work with an intermediate concentration of 0.6 g / l of Al2O3 per liter of solution,
resulting in a good compromise between dispersion and the amount of particles embedded in the
coating, as can be seen in Figure 5.89 e,f.
The presence of Al2O3 was confirmed by EDS analysis.

146
Figure 5.89: Cross-section SEM micrographs of the Ni-aAl2O3 nanocomposite –a,b) PN_Al_B_09; c,d) PN_Al_B_10;
e,f) PN_Al_B_16

147
5.3 Phase 3: Slurry aluminization of Electroless pure nickel samples

Firstly, heat treatment parameters were optimized by performing slurry aluminization on nickel
(purchased by GoodFellows inc.) samples. In Table 5.13 are reported the temperature profile selected
for the aluminization.

Table 5.13: Heat treatment parameters selected for the pure nickel (Good Fellows)

Samples dT/t (RT-650°C) t1 (650°C) dT/t (650-1050°C)

Ni_1 5 K/min 1h 10 K/min

Ni_2 5 K/min 1.5h 10 K/min

Ni_3 5 K/min 1h 5 K/min

The fixed parameters selected for the aluminization were the first plateau at 650°C to allow the
melting of aluminum and its consequent diffusion within the substrate and the plateau at 1050°C to
homogenize the b-NiAl phase on top of the substrate.
The SEM micrographs (Figure 5.90, Figure 5.91, Figure 5.92) show the cross-section for the sample
of the Ni series.
On the sample 2 and 3 it is possible to see the formation of three different zones:
One, on top of the coating, zone of Al-rich b-NiAl; the second one, less thick and more bright at BSE,
is compositional Ni-rich b-NiAl. The third one is the “interdiffusion zone” Ni3Al phase, formed by
the inward diffusion of the Al and the outward diffusion of Ni. Of course, the is no precipitation of
heavier element because we are considering a Ni pure sample.
In the sample Ni_1 seems that the Ni-rich zone is less pronounced but, above all, there is a sharp
decrease in the amount of Al at the b-NiAl phase and Ni3Al interface. For this reason, this heat
treatment was selected for the aluminization on electroless pure nickel plating. Aluminization on pure
Ni samples has allowed to optimize the heat treatment parameters depending on the amount of slurry
deposited on the surface of the sample (5mg / cm2). It should be remembered, however, that the
objective is to obtain a coating with the same thickness as the coatings obtained in phase 1 (20 µm of
b-NiAl). For this reason, heat treatment 1 has been preferred to others as it allows to obtain a b-NiAl
layer of the required thickness and with a rather satisfactory composition.

148
 Ni_1
100

90
Zone 2

Zone 3
80

70
Zone 1
Concentration (at. %)

60

50

40

30

20

10

0
0 5 10 15 20 25 30 35
Depth (µm)

Figure 5.90: Cross-section SEM micrograph for the samples Ni_1 with the relative diffusion profile obtain by EDS

149
Figure 5.91: Cross-section SEM micrograph for the samples Ni_2 with the relative diffusion profile obtain by EDS

150
Figure 5.92: Cross-section SEM micrograph for the samples Ni_3 with the relative diffusion profile obtain by EDS

The aluminization were carried out on the electroless pure nickel sample with the heat treatment
selected and reported in Table 5.14.

151
Table 5.14: Heat treatment selected for the slurry aluminization of the electroless nickel samples

Samples dT/t (RT-650°C) t1 (650°C) dT/t (650-1050°C)

Ni_1 5 K/min 1h 10 K/min

In order to study the influence of the ENL on the aluminide coating oxidation resistance, the same
aluminization was carried out on a superalloy not coated. In Figure 5.93 the cross-section SEM
micrographs show the influence of the ENL on the aluminization: there is no difference, in thickness
and microstructure, on the edges of the sample. This is due to the homogeneity of the ENL but also
to its wettability: the super-hydrophilicity of the surface, given by the nano-roughness, allow the
water-based slurry to cover all the sample surface.

Figure 5.93: Cross-section SEM micrographs for slurry aluminization on René 108DS(a) and electroless pure
nickel plating sample (b)

The thickness of the ENL was 16 µm before the aluminization. In Figure 5.94 the cross-section SEM
micrograph of the electroless pure nickel sample is shown. The bond coat formed is uniform, has a
good thickness but, despite the expectations, there is the formation of a small IDZ at the interface
between the bond coat and the substrate. This is due, probably, to a diffusion of the heavier elements
inside the ENL during the heat treatment.
152
Figure 5.94: Electroless nickel sample cross-section SEM micrograph aluminized with slurry with the relative diffusion
profile obtained by EDS

For comparison, a sample of René 108DS was aluminized by slurry with the same heat treatment.
The cross-section micrograph is shown in Figure 5.95. The microstructure resulted from the
aluminization is the typical Low Temperature High Activity with the precipitation included in the
bond coat.

153
Figure 5.95: René 108DS cross-section SEM micrograph aluminized by slurry

Preliminary isothermal oxidation was carried out on the electroless sample aluminized by slurry in
the same conditions of phase 1. The results are shown inFigure 5.96: the curve relative to the
electroless nickel sample show a not expected mass gain with a consequent loss after 30 hours of heat
treatment. Evidently the aluminization has not brought to equilibrium intermetallic phase which is,
therefore, unstable and continues its diffusion during oxidation at high temperature. The oxidation
of the bond coat seems not to be selective; this is probably because the aluminum diffuses to the
substrate and thus subtracts itself from the oxidation phenomena. On the contrary, Ni, tends to diffuse
outwardly making it more available to the oxidation. This would explain, in part, the total absence of
formation of a protective oxide scale with a spalling phenomenon that occurs only after 30 hours of
high temperature testing.

154
Figure 5.96: Isothermal oxidation (1050°C)

XRD spectrum (Figure 5.97) has confirmed this theory. As it was said, these are the preliminary test
on this sample. It is evident that the b-NiAl phase formed after aluminization needs a further heat
treatment to reach the final equilibrium and allows to the Al to oxidize in a protective way.

Figure 5.97: XRD spectrum of the electroless nickel plating sample aluminized by slurry after 25h of isothermal
oxidation test at 1050°C.

155
Although the isothermal oxidation test did not yield a satisfactory result, the aluminization obtained
by slurry deposition showed an interesting result. As expected, adding an external layer of nickel
involves a variation of the aluminization mechanism: the substrate does not participate in the
deposition mechanism. The only result is the formation of a small IDZ that can be considered as a
reinforcing mechanism in adhesion between the substrate and the external nickel layer. Thus, the
typical microstructure of a deposited HTLA coating has been obtained with the times and especially
the most contained parameters of a LTHA aluminization.

References

[1] S. Hamadi et al., Surface & Coatings Technology 204, 756–760, 2009.
[2] Mallory, G. O., Hajdu, J. B., Electroplaters, A., Society, S. F. & (Firm), Int. Bus. (2000).
[3] Cheng, Y. S. & Yeung, K. L., Memb. Sci. (2001), 182, 195–203 (2001).
[4] Braibanti, A., Dallavalle, F., Nghelli, M. A. P. & Leporati, E., Inorg. Chem. (1968), 7, 1430–
1433.
[5] Haag, S., Burgard, M. & Ernst, B. Surf. Coatings Technol., (2006), 201, 2166–2173.
[6] Patil, K. C. & Rattan, T. M., (2014).
[7] Park, J. W. et al., Mater. Chem. Phys. (2006), 97, 371–378
[8] Mallory, G. O., Fundam. Appl., (2009), 69; 71; 72
[9] Chen, H., Xu, C., Chen, C., Zhao, G. & Liu, Y. Mater. Res. Bull. (2012), 47, 1839–1844.
[10] Baskaran, I., Narayanan, T. S. N. S. & Stephen, A. Mater. Chem. Phys. (2006), 99, 117–126.
[11] Genova, V., Marini, D., Valente, M., Marra, F. & Pulci, G., Ch. Eng. Trans., (2017), 60, 73–
78.
[12] Wenzel, R. N., (1936), Ind. Eng. Chem. 28, 988–994.
[13] Cassie, B. D. Of porous surfaces, (1944), 546–551.
[14] Khorsand, S., Raeissi, K., Ashrafizadeh, F. & Arenas, M. A. Chem. Eng. J. (2015), 273, 638–
646.
[15] Khorsand, S., Raeissi, K. & Ashrafizadeh, F. Appl. Surf. Sci. 305, (2014), 498–505.
[16] Boinovich, L. B. et al. Langmuir 28, (2012), 1206–1216.
[17] Zheng, S. et al. Surf. Coatings Technol. (2015), 276, 341–348.
[18] Guo, M., Diao, P. & Cai, S. Thin Solid Films, (2007), 515, 7162–7166.
[19] Geng, W., Hu, A. & Li, M. Appl. Surf. Sci., (2012), 263, 821–824.

156
CONCLUSIONS
Diffusion aluminide coatings are used on turbine blades, both as a single coating and as a bond coat
in a Thermal Barrier Coating System (TBC), to ensure its protection against oxidation and hot
corrosion. Their protective function is achieved through the selective oxidation of the aluminum
present in the coating; this process leads to the formation of a Thermally Grown Oxide (TGO).
In this work, the diffusion coatings corrosion resistance was studied and the modification of the
standard aluminide coating with the aim of improving its high temperature performance was carried
out by studying different techniques.

In the first part of the work, the process of VPA (Vapour Phase Aluminizing) was optimized, varying
the concentration of the two deposition salts, NH4F and AlF3. The diffusion coatings obtained with
NH4F show better oxidation resistance than those obtained with AlF3. In particular, the coating with
a higher concentration of ammonium fluoride has a weight gain after 100 hours of oxidation lower
than all others.
For the description of experimental oxidation data, Monceau's parabolic model seems more suitable
than the Wagner's model.
During the hot corrosion test all coatings exhibit an alternation between increases and losses by
weight, as a high amount of corrosion salt is applied on a very narrow area of the surface. This greatly
accelerates the oxidation phenomenon: there is a very high localization of the protective scale, which
becomes unstable after a few hours of testing.
Once the optimal NH4F concentration has been selected, five coatings in -NiAl doped with Zr were
produced using a mixed NH4F / ZrF4 activator salt. The Zr is deposited on the interface between the
superalloy and the interdiffusion zone, and during oxidation diffuse to the surface to the oxide scale.
A range of concentrations was identified within which the addition of Zr improves the performance
of the conventional coating. For concentrations lower than the optimum values, the effects generated
by the presence of the reactive element are zero or minimal; for higher concentrations, the internal
oxidation of the reactive elements is observed, which results in an increase in oxygen permeability
and hence the rate of oxidation.
Experimental evidence shows that zirconium has the effect of increasing the adhesion of the oxide
scale to the substrate. This may be essentially due to two reasons, both of which are explained in the
segregation of Zr at the interface between coating and oxide during oxidation. The first hypothesis is
that Zr reduces the rate of transient oxidation: -Al2O3 is characterized by a rapid growth driven by the
cationic diffusion of the Al towards the outside. Zr would thus slow the diffusion of Al, favoring the
157
nucleation of -Al2O3. The second hypothesis is that Zr limits the release of protective oxide during
cooling, suppressing the formation of Kirkendall's voids at the interface.
The experimental data produced in the second phase of the experiment also confirm that Monceau's
parabolic model seems more suitable for the description of the oxidation kinetics than the Wagner
model.
From preliminary considerations on the hot corrosion test of the doped samples, it is assumed that
modification of the coatings with Zr does not give rise to substantial benefits in the resistance to this
degradation phenomenon. Samples with conventional and doped coatings, in fact, have a comparable
increase in weight.

In the second part of the work, electroless pure nickel plating was selected as technique to deposit an
ENL on top of the substrate before the aluminization process. This modification of the standard
coating was conducted in order to reduce the formation of the IDZ that can influence the mechanical
properties of the turbine blade. Furthermore, the deposition technique allows to introduce Al2O3
nanoparticles inside the coating in order to increase the oxidation resistance by promoting the
nucleation of the protective oxide scale on top of the bond coat.
The study of several plating parameters (pH, T, PR etc.) has allowed to obtain a thicker (20 µm),
homogeneous, continuous pure nickel layer on top of the superalloy. Well dispersed nanoparticles
were embedded inside the Ni matrix and the quantity of a-Al2O3 nanoparticles inside the plating
solution was studied and optimized.

The electroless pure nickel plating sample were aluminized by slurry deposition, in order to study the
reaction of the ENL, plated directly onto the Ni-based super alloy, to the aluminization heat treatment.
Once the aluminization heat treatment was optimized, isothermal oxidation was carried out for the
samples that showed the best microstructure after slurry aluminization. Preliminary results have
shown that the heat treatment selected for the slurry aluminization doesn’t allow the formation of a
stable b-NiAl on the top of the substrate with a consequent diffusion to equilibrium that occur during
oxidation heat treatment. Future test of annealing after aluminization will carried out in order to
rebalance the bond-coat superficial structure.
Despite this, the most important result is that the typical microstructure of a deposited HTLA coating
has been obtained with the times and especially the most contained parameters of a LTHA
aluminization.

158
159