**3.3 Pulse generator chamber**

The pulse CD supply (see figure 5) consists of a three phase full wave rectifying circuit (D1- D6) coupled to a Variac which enables the user to select the CD output voltage level. Later

PIII for Aluminium Surface Modification 181

output is obtained, due to the SW1 switch, between 0.2 and 20Hz and between 20 and 2000 Hz. The voltage is applied to the double probe in the order of ±150V by two STK4050V high voltage operational amplifiers connected in a differential way. Each amplifier was individually configured as an inverter with a gain factor of 47, given the relationships R17/R16 and R27/R26. Consequentially, the output peak value is ±75 V when 1.6Vp are

applied at the amplifier input.

Fig. 6. Longitudinal positioning mechanism of the probe

Fig. 7. General electronics circuit diagram

on, the voltage is filtered so to supply the high voltage pulse transformer with a CD signal bearing the least possible ripple. The selected commutation device is an IGBT SKM200GB125D by SEMIKRON™ that is driven by an M57959L module. This is a high speed component endowed with a voltage logic level input and insulated by a high speed opto-coupler protected against the event of a short circuit. A commuted converter in a flyback configuration was chosen to build it, due to its relative simplicity and low cost. The control stage has been implemented by a SG3527A pulse wave modulator (PWM) which imposes the width and repetition rate of the pulses applied to the M57959L module. According to the circuit configuration, the PWM work cycle can vary between 0 and 49% at a 55-2600 Hz rate. Such characteristics can be set up by the P1 and P2 potentiometers.

Fig. 5. High voltage pulse generator diagram

### **3.4 Electric probe diagnostics**

In order to measure plasma parameters such as electron temperature ( *Te* ) and plasma density ( *n* ) in a simple way, a double Langmuir probe is used. In order to increase the lifetime of the probe, a mechanical system (guard) was designed and constructed. This guard protects the electric probe within the chamber when the diagnostics is not being carried out. The probe is exposed to the plasma discharge for short intervals and only when the diagnostic system is activated. A probe was built out of an alumina rod with two perforations as the insulating element between two tungsten conductors and the metallic capsule that gives support to the probe (figure 6). The tungsten filaments are 0.195 mm in diameter and 4.3 mm long. The probe, inside the guard, is made of a stainless steel pipe, 0.95 cm in diameter and 25 cm long, intended to couple to the engine system that shifts the guard on and off.

This double probe was biased by means of a specifically designed and constructed triangular and sawtooth waveform generator operating in either modality thanks to the SW2 switch (see figure 7). An XR2206 function generator, and associated electronic components, was also used at the low voltage (±15V) stage so that a two scale frequency

on, the voltage is filtered so to supply the high voltage pulse transformer with a CD signal bearing the least possible ripple. The selected commutation device is an IGBT SKM200GB125D by SEMIKRON™ that is driven by an M57959L module. This is a high speed component endowed with a voltage logic level input and insulated by a high speed opto-coupler protected against the event of a short circuit. A commuted converter in a flyback configuration was chosen to build it, due to its relative simplicity and low cost. The control stage has been implemented by a SG3527A pulse wave modulator (PWM) which imposes the width and repetition rate of the pulses applied to the M57959L module. According to the circuit configuration, the PWM work cycle can vary between 0 and 49% at a 55-2600 Hz rate. Such characteristics can be set up by the P1 and P2 potentiometers.

0-600 V

8 Ω

DIODE

2.5μF 200Ω

RL

CURRENT MEASUREMENT

PULSE

HIGH VOLTAGE MEASUREMET

SKM200GB125D IGBT

ROGOWSKI COIL

CLAMPER NEGATIVE

10ETF12FP

PULSE TRANSFORMER 1:50

6 5

0.1μF

In order to measure plasma parameters such as electron temperature ( *Te* ) and plasma density ( *n* ) in a simple way, a double Langmuir probe is used. In order to increase the lifetime of the probe, a mechanical system (guard) was designed and constructed. This guard protects the electric probe within the chamber when the diagnostics is not being carried out. The probe is exposed to the plasma discharge for short intervals and only when the diagnostic system is activated. A probe was built out of an alumina rod with two perforations as the insulating element between two tungsten conductors and the metallic capsule that gives support to the probe (figure 6). The tungsten filaments are 0.195 mm in diameter and 4.3 mm long. The probe, inside the guard, is made of a stainless steel pipe, 0.95 cm in diameter and 25 cm long, intended to couple to the engine system that shifts the guard

This double probe was biased by means of a specifically designed and constructed triangular and sawtooth waveform generator operating in either modality thanks to the SW2 switch (see figure 7). An XR2206 function generator, and associated electronic components, was also used at the low voltage (±15V) stage so that a two scale frequency

MODULATOR SG3527A

DRIVER M57959L

D1 D3 D5

D2 D4 D6

2 16

P2 10 kΩ

Fig. 5. High voltage pulse generator diagram

**3.4 Electric probe diagnostics** 

on and off.

P1 100 kΩ

15

VARIAC STEP-UP TRANSFORMER

F1 F2 F3


output is obtained, due to the SW1 switch, between 0.2 and 20Hz and between 20 and 2000 Hz. The voltage is applied to the double probe in the order of ±150V by two STK4050V high voltage operational amplifiers connected in a differential way. Each amplifier was individually configured as an inverter with a gain factor of 47, given the relationships R17/R16 and R27/R26. Consequentially, the output peak value is ±75 V when 1.6Vp are applied at the amplifier input.

Fig. 6. Longitudinal positioning mechanism of the probe

Fig. 7. General electronics circuit diagram

PIII for Aluminium Surface Modification 183

where *AS* is the area of sheath [m2], *e* is the electron charge [C], *mi* is the mass of ion [kg], *k* is Boltzmann´s constant [J/K], *Ii* is the ion current [A], *Vp* is the probe bias voltage [V], and *AP*

Instrumentation hardware and software have been calibrated on DC plasmas from argon, nitrogen and gas mixtures. Figure 8 displays some results obtained when the probe reached the centre of the reactor filled with nitrogen. The DC current supply went from 500mA to

In a typical experiment under the previously specified plasma parameters, the following steps are conducted by means of the electric probe. First, the guard and electrical probe array are positioned inside the vacuum vessel, and, as a second step, the probe is moved outside the guard, putting it in contact with the plasma. When the electric probe stops, it is biased by one cycle of the sawtooth signal. With the electric probe system in position, it is possible to measure the electron density and temperature at different locations inside the vacuum chamber. The graphics shown in figure 8 correspond to the centre of the chamber.

Plasmas are ionised gases and, therefore, electrically conductive to some extent. In this way, plasmas are capable to shield regions of a scale estimated by the Debye length (λDe) provided that the number of charged particles within the Debye sphere is far greater than one and that their motion obeys forces essentially electromagnetic. Other assumptions of the

• Electrons are massless so that they respond instantly to the applied potentials given that the implantation time scale is much greater than the plasma cyclotron frequency

(a) Density (b) Temperature

1.5

2.0

2.5

3.0

Temperature [eV]

3.5

4.0

4.5

*OML*

*n*

is the exposed probe electrode surface area [m2].

0.0 3.0x10-2 6.0x10-2 9.0x10-2

3000mA, both cases being a function of the work pressure.

• The ion flow is not collisional which apply to low gas pressures.

Pressure [Torr]

 3.0 A, 2.5 A 2.0 A, 1.5 A 1.0 A, 0.5 A

Fig. 8. Nitrogen plasma parameters in DC

**3.5 The non collisional ion sheath model** 

model are [Anders, 2000]:

0.00E+000

2.50E+016

5.00E+016

Density [m-3

]

7.50E+016

1.00E+017

1.25E+017

2

3 2 [ ( )/ ] 0.2

*e A*

*I Vm*

*i p i*

−Δ Δ <sup>=</sup> (4)

0.0 3.0x10-2 6.0x10-2 9.0x10-2

ω*e* .

Pressure [Torr]

 3.0 A, 2.5 A 2.0 A, 1.5 A 1.0 A, 0.5 A

*p*

The data acquisition of the voltage and current signals from the signal conditioning stage was carried out with a DAQ PCI-6023E National Instruments board. The measurement of the current circulating through the plasma immerse probe was achieved by the use of an RM, resistor. Thus, when the current varies between ±IM, a ±VM voltage will be obtained in a directly proportional way. This voltage is located at the amplification stage input with a gain factor GAMP=200. Then, assuming a maximal voltage limit at the board VOA(MAX), a sensor resistance RM and an attenuation factor FA, the expression of the maximal current interval IM(MAX) can be given by:

$$I\_{M(MAX)} = \frac{V\_{OA(MAX)}}{R\_M F\_A G\_{AMP}}\tag{1}$$

The attenuation network (R47 and R48) of the current sensor where the low pass filter (R49 and C34) output signal is connected to the AD624 instrumentation amplifier. In this way, the IM proportional VM voltage signal is increased to a 200 gain. Once conditioned, the signal complies with the voltage specifications of the DAQ PCI-6023E acquisition board.

The probe displacement has been achieved by means of a longitudinal positioning mechanism allowing the guard, containing the double probe, to be introduced to the reactor as much as 25 cm while allowing exposing or retracting a 3 cm long tip of the probe with respect to the guard (figure 6).

A LabVIEW™ compatible program was specifically designed to operate the system. The software was applied to process, visualise and storage: the applied voltage, response current and probe positioning. The latter is set at the graphic interface (figure 8.a) which transmits the advance, stop and retreat signals, through an 8 bit terminal provided by the DAQ PCI-6023E board, to the power electronics associated to the mechanism. Then, a fraction of the collected current is selected in order to be plotted against the applied voltage and, from this characteristic curve, determining the main plasma parameters (figure 8b).

The saturation current and the electron temperature provide valuable information in determining the plasma parameters. Two values are calculated from the locus of the V-I plot, both from its positive and negative parts. The V-I double symmetrical cylindrical probe characteristic curve can be approached by the nonlinear function (Equation 1). The Levenberg-Marquardt fit method was implemented so to determine the coefficients of it [Herman and Gallimore, 2008]:

$$I(V) = I\_{isat} \tanh\left(\frac{V}{2T\_{eV}}\right) + A\_1V + A\_2\tag{2}$$

here, *TeV* is the electron temperature [eV], *Iisat* is the saturation ion current [A] of each one of the probe ends, *A1* is an account of the expanded ion saturation current sheath depth, whereas *A2* refers to the reflection and displacement currents resulting from stray capacitances. The density calculation was performed on the basis of two types of data analysis: Bohm Approximation and Orbital Motion Limit (OML), given respectively by [Herman and Gallimore, 2008]:

$$m\_i = \frac{I\_{isat}}{0.61A\_Se} \sqrt{\frac{m\_i}{kT\_{eV}}}\tag{3}$$

The data acquisition of the voltage and current signals from the signal conditioning stage was carried out with a DAQ PCI-6023E National Instruments board. The measurement of the current circulating through the plasma immerse probe was achieved by the use of an RM, resistor. Thus, when the current varies between ±IM, a ±VM voltage will be obtained in a directly proportional way. This voltage is located at the amplification stage input with a gain factor GAMP=200. Then, assuming a maximal voltage limit at the board VOA(MAX), a sensor resistance RM and an attenuation factor FA, the expression of the maximal current

( )

complies with the voltage specifications of the DAQ PCI-6023E acquisition board.

characteristic curve, determining the main plasma parameters (figure 8b).

*M MAX*

*I*

( )

*R FG* <sup>=</sup> (1)

(2)

*OA MAX*

*V*

The attenuation network (R47 and R48) of the current sensor where the low pass filter (R49 and C34) output signal is connected to the AD624 instrumentation amplifier. In this way, the IM proportional VM voltage signal is increased to a 200 gain. Once conditioned, the signal

The probe displacement has been achieved by means of a longitudinal positioning mechanism allowing the guard, containing the double probe, to be introduced to the reactor as much as 25 cm while allowing exposing or retracting a 3 cm long tip of the probe with

A LabVIEW™ compatible program was specifically designed to operate the system. The software was applied to process, visualise and storage: the applied voltage, response current and probe positioning. The latter is set at the graphic interface (figure 8.a) which transmits the advance, stop and retreat signals, through an 8 bit terminal provided by the DAQ PCI-6023E board, to the power electronics associated to the mechanism. Then, a fraction of the collected current is selected in order to be plotted against the applied voltage and, from this

The saturation current and the electron temperature provide valuable information in determining the plasma parameters. Two values are calculated from the locus of the V-I plot, both from its positive and negative parts. The V-I double symmetrical cylindrical probe characteristic curve can be approached by the nonlinear function (Equation 1). The Levenberg-Marquardt fit method was implemented so to determine the coefficients of it

> 1 2 ( ) tanh <sup>2</sup> *isat eV*

here, *TeV* is the electron temperature [eV], *Iisat* is the saturation ion current [A] of each one of the probe ends, *A1* is an account of the expanded ion saturation current sheath depth, whereas *A2* refers to the reflection and displacement currents resulting from stray capacitances. The density calculation was performed on the basis of two types of data analysis: Bohm Approximation and Orbital Motion Limit (OML), given respectively by

0.61

*i*

*n*

*isat i*

*I m*

*S eV*

*A e kT* <sup>=</sup> (3)

*<sup>V</sup> IV I AV A T* ⎛ ⎞ <sup>=</sup> ⎜ ⎟ <sup>+</sup> <sup>+</sup> ⎝ ⎠

*M A AMP*

interval IM(MAX) can be given by:

respect to the guard (figure 6).

[Herman and Gallimore, 2008]:

[Herman and Gallimore, 2008]:

$$m\_{\rm OML} = \sqrt{\frac{[-\Delta(I\_i^2)/\Delta V\_p]m\_i}{0.2e^3A\_p^2}}\tag{4}$$

where *AS* is the area of sheath [m2], *e* is the electron charge [C], *mi* is the mass of ion [kg], *k* is Boltzmann´s constant [J/K], *Ii* is the ion current [A], *Vp* is the probe bias voltage [V], and *AP* is the exposed probe electrode surface area [m2].

Fig. 8. Nitrogen plasma parameters in DC

Instrumentation hardware and software have been calibrated on DC plasmas from argon, nitrogen and gas mixtures. Figure 8 displays some results obtained when the probe reached the centre of the reactor filled with nitrogen. The DC current supply went from 500mA to 3000mA, both cases being a function of the work pressure.

In a typical experiment under the previously specified plasma parameters, the following steps are conducted by means of the electric probe. First, the guard and electrical probe array are positioned inside the vacuum vessel, and, as a second step, the probe is moved outside the guard, putting it in contact with the plasma. When the electric probe stops, it is biased by one cycle of the sawtooth signal. With the electric probe system in position, it is possible to measure the electron density and temperature at different locations inside the vacuum chamber. The graphics shown in figure 8 correspond to the centre of the chamber.

### **3.5 The non collisional ion sheath model**

Plasmas are ionised gases and, therefore, electrically conductive to some extent. In this way, plasmas are capable to shield regions of a scale estimated by the Debye length (λDe) provided that the number of charged particles within the Debye sphere is far greater than one and that their motion obeys forces essentially electromagnetic. Other assumptions of the model are [Anders, 2000]:


PIII for Aluminium Surface Modification 185

0.0 50.0µ 100.0µ 150.0µ 200.0µ 250.0µ 300.0µ

 1000V, 2000V 3000V, 4000V 5000V, 6000V 7000V, 8000V

Time [s]

Fig. 9. Ion sheath evolution at a 9×1016 m-3 plasma density and biasing between -1 and -8 kV polished and ultrasonically cleansed in acetone. A base pressure ~10-6 Torr was achieved with a turbo-molecular vacuum pump, and then the work pressure was established at 3×10-2 Torr (see figure 8.a) by admitting nitrogen of a 99.998% purity and its mixtures. Each sample receives a previous 30 min cleansing stay in Argon plasma to be finally implanted for 1-1.5 h periods. The PIII process was conducted with -2 to -5.5 kV bias squared pulses, with ion doses in the order of 1.7×1016 ions/cm2 per pulse, while the electron plasma density is kept at about 9x1016 m-3. The gas admission mixture was calibrated by using flow regulators. All the specimens were treated at 400°C. The samples were treated in four separate groups under previously optimised controlled conditions: the first one in 99.998%pure nitrogen (N), the second one in 70%Nitrogen and 30%Argon (70N-30A), the third one in 50%Nitrogen-50%Argon (50N-50Ar) and the last one in 30%Nitrogen-70%Argon (30N-

The first treatment was applied during 1h periods at a 500 Hz repetition rate. The resulting micrographs are shown in figure 10 and the corresponding EDX spectra in figure 11. O, N

The micrograph in figure 10.d (N30-Ar70) presents a smoothed surface due to the intense bombardment with Ar, which is corroborated by the respective rugosity plot (figure 12). Likewise, the smoothing of the surface in micrograph 10.a follows from the N treatment. In the case of the N50-Ar50 mixture (figure 10.c) the grain size appears particularly inhomogeneous, with an average magnitude of 0.35 μm. Spiked grains of different sizes confirm the highest rugosity (Ra) values occurring in this lot of specimens. As the nitrogen bombardment intensity decreases, the hardness declines, except in the case of N50-Ar50 which displays the highest hardness (figure 13). This result could be explained from the

0.5

70Ar) mixtures.

**4.1 Aluminium treated at 2kV and 150 µs width pulse** 

and Al are always present in this specimen.

1.0

1.5

2.0

Sheat [cm]

2.5

3.0

3.5

4.0

4.5


In this manner, the Child's law current density at a voltage *V*<sup>0</sup> , through a sheath of thickness *s,* can be expressed as

$$j\_c = \frac{4}{9} \frac{\varepsilon\_0 \left(V\_0\right)^{3/2}}{s^2} \sqrt{\frac{2e}{M}}\tag{5}$$

where 0 ε is the vacuum permittivity, *e* is the electron charge and *M* the ion mass. By equating *cj* with the amount of charge per unit of time and per unit of area that crosses the sheath border, *en ds dt* <sup>0</sup> ( ) , one can find the expansion speed of this border:

$$\frac{ds}{dt} = \frac{2}{9} \frac{s\_0^{-2} u\_0}{s^2} \tag{6}$$

where *s V en* 0 00 0 = ( ) 2ε is the ion matriz thickness and *u eV M* 0 0 = (2 ) is the characteristic ion speed. Equation (6) becomes, after integration,

$$s(t) = s\_0 \left(1 + \frac{2}{3} a o\_i t\right)^{1/3} \tag{7}$$

where <sup>2</sup> *i* 0 0 00 ω ε = *en M u s* = is the ion frequency of the plasma. Estimating the ion matrix thickness during a PIII process is crucial as its size must not approach the reactor dimensions (see section 3.1) in order to have enough plasma to collect and implant ions from. Figure 9 illustrates the dynamic evolution of the ion sheath in the case of nitrogen ( *M* =28) when the bias potential −*V*0 ranges from 1kV to 8 kV, provided that the plasma density is 9×1016 m-3 (cf. figure 8.a).

By integrating Eq. (5), the ion fluence *F* impinging on the aluminium piece can be calculated. In a planar geometry with a maximum sheath width *s t*( ) and a voltage *V*<sup>0</sup> during a pulse of length *pt* , the fluence is:

$$F = n\left(s(t) + t\_p \sqrt{\frac{kT\_e}{M}}\right) \tag{8}$$

Typically, in a 50 μs pulse, a plasma density of 9×1016 m-3, *Te* =3eV and voltage of 5 kV, the fluence density can reach up to 1.7×1016 ions per m2. With a 500 Hz repetition rate, it is possible to implant doses in the order of 1021 ions/m2 in ~1 hour.

### **4. Results and discussion**

The experimentation was carried out on a commercially pure aluminium rod (6061-T6) sectioned into cylindrical pieces, 10 mm in diameter and 5 mm thick. The samples were mirror

• A bias −*V*0 is applied to the piece at *t*=0, where *V0* >> *Te* (with *Te* clearly expressed in

• The transit time through the matrix is null, i.e., the implantation current is identical to

In this manner, the Child's law current density at a voltage *V*<sup>0</sup> , through a sheath of

( )3 2 0 0 2 4 2

*s M*

is the vacuum permittivity, *e* is the electron charge and *M* the ion mass. By

= (5)

*dt <sup>s</sup>* <sup>=</sup> (6)

⎝ ⎠ (7)

(8)

is the ion matriz thickness and *u eV M* 0 0 = (2 ) is the characteristic

1 3

ω

⎛ ⎞ = + ⎜ ⎟

Estimating the ion matrix thickness during a PIII process is crucial as its size must not approach the reactor dimensions (see section 3.1) in order to have enough plasma to collect and implant ions from. Figure 9 illustrates the dynamic evolution of the ion sheath in the case of nitrogen ( *M* =28) when the bias potential −*V*0 ranges from 1kV to 8 kV, provided

By integrating Eq. (5), the ion fluence *F* impinging on the aluminium piece can be calculated. In a planar geometry with a maximum sheath width *s t*( ) and a voltage *V*<sup>0</sup>

> ( ) *<sup>e</sup> <sup>p</sup> kT F n st t*

fluence density can reach up to 1.7×1016 ions per m2. With a 500 Hz repetition rate, it is

The experimentation was carried out on a commercially pure aluminium rod (6061-T6) sectioned into cylindrical pieces, 10 mm in diameter and 5 mm thick. The samples were mirror

⎛ ⎞ = + ⎜ ⎟ ⎝ ⎠

*M*

s pulse, a plasma density of 9×1016 m-3, *Te* =3eV and voltage of 5 kV, the

*<sup>V</sup> <sup>e</sup> <sup>j</sup>*

equating *cj* with the amount of charge per unit of time and per unit of area that crosses the

2 9 *ds s u*

ε

9 *<sup>c</sup>*

sheath border, *en ds dt* <sup>0</sup> ( ) , one can find the expansion speed of this border:

( )

= *en M u s* = is the ion frequency of the plasma.

0 <sup>2</sup> <sup>1</sup> <sup>3</sup> *<sup>i</sup> st s t*

*De* << <sup>0</sup>*s* the latter being the initial sheath depth. • A quasi-static matrix is formed instantly demanding a current which satisfies Child's

volts) whereby

λ

• All charged particles are singly ionised.

ion speed. Equation (6) becomes, after integration,

that the plasma density is 9×1016 m-3 (cf. figure 8.a).

possible to implant doses in the order of 1021 ions/m2 in ~1 hour.

during a pulse of length *pt* , the fluence is:

μ

**4. Results and discussion** 

*i* 0 0 00

 ε

thickness *s,* can be expressed as

where *s V en* 0 00 0 = ( ) 2ε

where <sup>2</sup>

ω

Typically, in a 50

where 0 ε

law and is provided by the uncovered ions nearby.

the amount of ion charge uncovered per second.

Fig. 9. Ion sheath evolution at a 9×1016 m-3 plasma density and biasing between -1 and -8 kV

polished and ultrasonically cleansed in acetone. A base pressure ~10-6 Torr was achieved with a turbo-molecular vacuum pump, and then the work pressure was established at 3×10-2 Torr (see figure 8.a) by admitting nitrogen of a 99.998% purity and its mixtures. Each sample receives a previous 30 min cleansing stay in Argon plasma to be finally implanted for 1-1.5 h periods. The PIII process was conducted with -2 to -5.5 kV bias squared pulses, with ion doses in the order of 1.7×1016 ions/cm2 per pulse, while the electron plasma density is kept at about 9x1016 m-3. The gas admission mixture was calibrated by using flow regulators. All the specimens were treated at 400°C. The samples were treated in four separate groups under previously optimised controlled conditions: the first one in 99.998%pure nitrogen (N), the second one in 70%Nitrogen and 30%Argon (70N-30A), the third one in 50%Nitrogen-50%Argon (50N-50Ar) and the last one in 30%Nitrogen-70%Argon (30N-70Ar) mixtures.

#### **4.1 Aluminium treated at 2kV and 150 µs width pulse**

The first treatment was applied during 1h periods at a 500 Hz repetition rate. The resulting micrographs are shown in figure 10 and the corresponding EDX spectra in figure 11. O, N and Al are always present in this specimen.

The micrograph in figure 10.d (N30-Ar70) presents a smoothed surface due to the intense bombardment with Ar, which is corroborated by the respective rugosity plot (figure 12). Likewise, the smoothing of the surface in micrograph 10.a follows from the N treatment. In the case of the N50-Ar50 mixture (figure 10.c) the grain size appears particularly inhomogeneous, with an average magnitude of 0.35 μm. Spiked grains of different sizes confirm the highest rugosity (Ra) values occurring in this lot of specimens. As the nitrogen bombardment intensity decreases, the hardness declines, except in the case of N50-Ar50 which displays the highest hardness (figure 13). This result could be explained from the

PIII for Aluminium Surface Modification 187

(a)

(b)

(c)

absence of aluminium oxide in the N70-Ar30 and N50-Ar50 treated samples with respect the pure nitrogen and N30-Ar70 cases. All the treated samples improve their hardness with respect to that of the untreated one, with the exception of the N30-Ar70 case which exhibits only the cubic crystalline phase, while the rest do both the cubic and hexagonal phases.

(a) (b)

Fig. 10. Treatment micrographs: (a) Ar(N), (b) Ar/(N70-Ar30), (c) Ar/(N50-Ar50), (d) Ar/(N30-Ar70)

(c) (d)

Diffractogramme 14.a, corresponding to N, shows the highest presence of AlN, both in cubic and hexagonal phases. Figure 14.c identifies the highest content of AlN in the cubic phase, as the presence of Ar seems to promote this phase and to inhibit the hexagonal one. This fact is due, perhaps, to the catalytic potential of argon, even though there is a competition between sputtering and implantation. In figure 14.b and 14.d, one observes only a small peak of the cubic phase of AlN. Thus, the high treatment temperature (400°C) may have changed the surface microhardness unfavorably, compared with the untreated case. The insignificant content of phases may be due to the relatively high sputtering produced by Ar in contrast with the nitrogen implantation.

absence of aluminium oxide in the N70-Ar30 and N50-Ar50 treated samples with respect the pure nitrogen and N30-Ar70 cases. All the treated samples improve their hardness with respect to that of the untreated one, with the exception of the N30-Ar70 case which exhibits only the cubic crystalline phase, while the rest do both the cubic and hexagonal

(a) (b)

(c) (d)

Fig. 10. Treatment micrographs: (a) Ar(N), (b) Ar/(N70-Ar30), (c) Ar/(N50-Ar50),

Diffractogramme 14.a, corresponding to N, shows the highest presence of AlN, both in cubic and hexagonal phases. Figure 14.c identifies the highest content of AlN in the cubic phase, as the presence of Ar seems to promote this phase and to inhibit the hexagonal one. This fact is due, perhaps, to the catalytic potential of argon, even though there is a competition between sputtering and implantation. In figure 14.b and 14.d, one observes only a small peak of the cubic phase of AlN. Thus, the high treatment temperature (400°C) may have changed the surface microhardness unfavorably, compared with the untreated case. The insignificant content of phases may be due to the relatively high sputtering produced by Ar

phases.

(d) Ar/(N30-Ar70)

in contrast with the nitrogen implantation.

(b)

PIII for Aluminium Surface Modification 189

 Alpattern N N70Ar30 N50Ar50 N30Ar70

0 50 100 150 200 250 300

4-AlN-h 3-AlN-c 2-Al2 O3 1-Al

Load [gf]

30 40 50 60 70 80 90

1

1 1,3

1

1

1,3

1,3

1,3

1

1

1 4

2θ [degrees]

Fig. 14. Difractogrammes from: (a)Ar/(N), (b) Ar/(N70–Ar30), (c) Ar/(N50–Ar50) and

(a)

3

2

1 1

1

1

3

3

2

3 2

1

2

2

1

2

1

2

2

2

(b)

(c)

(d)

Internsity [a.u]

Fig. 13. Hardness

(d) Ar/(N30–Ar70)

Hardness [HV]

Fig. 11. Aluminium spectra: (a) Ar(N), (b) Ar/(N70-Ar30), (c) Ar/(N50-Ar50), (d) Ar/(N30- Ar70)

(d)

Fig. 12. Rugosity

#### **4.2 Aluminium treated at 3.5kV and 75 µs width pulse**

These tests were aimed at attracting the ions with greater energy by increasing the bias voltage while reducing the pulse width. The resulting micrographs are presented in figure 15. The maximal rugosity is identified in the sample treated with the Ar/(N30-Ar70) mixture (figure 15.c) given the respective rugosity tests (see figure 17). The changes in the morphology of the Ar/(N70-Ar30) and Ar/(N50-Ar50) treated samples, given that of the

(d) Fig. 11. Aluminium spectra: (a) Ar(N), (b) Ar/(N70-Ar30), (c) Ar/(N50-Ar50), (d) Ar/(N30-

Al pattern N N70-Ar30 N50-Ar50 N30-Ar70

Mixtures

These tests were aimed at attracting the ions with greater energy by increasing the bias voltage while reducing the pulse width. The resulting micrographs are presented in figure 15. The maximal rugosity is identified in the sample treated with the Ar/(N30-Ar70) mixture (figure 15.c) given the respective rugosity tests (see figure 17). The changes in the morphology of the Ar/(N70-Ar30) and Ar/(N50-Ar50) treated samples, given that of the

0

**4.2 Aluminium treated at 3.5kV and 75 µs width pulse** 

50

100

150

Roughness [nm]

Fig. 12. Rugosity

200

250

300

350

Ar70)

Fig. 13. Hardness

Fig. 14. Difractogrammes from: (a)Ar/(N), (b) Ar/(N70–Ar30), (c) Ar/(N50–Ar50) and (d) Ar/(N30–Ar70)

PIII for Aluminium Surface Modification 191

(a)

(b)

(c) Fig. 16. Aluminium spectra from: a) Ar/(N70-Ar30), b) Ar/(N50-Ar50), c) Ar/(N30-Ar70)

control one (figure 1) are evident. Figure 16 displays the EDS outcome suggesting the presence of N and an increase in O.

(a)

(b)

(c)

Fig. 15. Micrographs from: a) Ar/(N70-Ar30), b) Ar/(N50-Ar50), c) Ar/(N30-Ar70)

control one (figure 1) are evident. Figure 16 displays the EDS outcome suggesting the

(a)

(b)

(c)

Fig. 15. Micrographs from: a) Ar/(N70-Ar30), b) Ar/(N50-Ar50), c) Ar/(N30-Ar70)

presence of N and an increase in O.

Fig. 16. Aluminium spectra from: a) Ar/(N70-Ar30), b) Ar/(N50-Ar50), c) Ar/(N30-Ar70)

(c)

PIII for Aluminium Surface Modification 193

As follows from the hardness plot in figure 18, the maximal enhancement, up to 62HV0.1 at a 100 g load, was achieved with the highest nitrogen concentration mixture Ar/(N70-Ar30). By contrast, the Ar/(N30-Ar70) and Ar/(N50–Ar50) treated samples measured 60HV0.1 and

A comparative view of the diffractogrames shown in figure 19 indicates that the relative intensity of the Al peak at 2*θ* = 38.47° attained with the Ar/(N50-Ar50) mixture, decreases when the Ar/(N70-Ar30) and Ar/(N30-Ar70) ones are used (figures 19.a and 19.c). Quite the opposite with respect to the peak at 78.23° (figure 19.b) whose intensity is greater than those seen in figure 19.a y 19.c. The latter may be due to a low concentration of compounds like AlN and Al2O3 and to Al peaks either intrisically moderate or missing (such as the one expected at 2*θ =* 65.13°). Likewise, the Ar/(N50-Ar50) diffractogramme seen in figure 19.b, does not show the 2θ = 82.43° Al peak while the main Al peak is particularly reduced wich is ultimately attributable to the implantation process itself. The diffractogramme of figure 19.c points to the fact that a low N concentration favours the cubic phase of AlN at 2*θ*=41.80°

> 30 40 50 60 70 80 90 2θ [degrees]

3

<sup>4</sup> <sup>1</sup>

3 1

<sup>1</sup> 1,3 <sup>1</sup>

2-Al 2 O3

3-AlN-c 4-AlN-h

1

1

1

1,3

3 2

Fig. 19. Diffractogrammes a) Ar/(N70-Ar30), b)Ar/(N50-Ar50) c)Ar/(N30-Ar70)

32

1

1 1

2

2

(c)

(b)

Intensity [a.u.]

(87-1053 JCPDS standard), 78.41° and 82.62° (46-1200 JCPDS standard).

(a) 1-Al

3 2

1

2

2

35HV0.1, respectively.

Figure 17 shows that the greatest rugosity is achieved by the Ar/(N50-Ar50) sample which, at the same time, presents the lowest (290 nm) hardness (figure 18). The Ar/(N70-Ar30) sample reached an average value of 267 nm despite the maximal N concentration in the mixture and the consequent ion impact on the piece. The Ar/(N30-Ar70) sample obtained an *Ra* value of 195 nm: the lowest in the present experiment.

Fig. 17. Rugosity

Fig. 18. Hardness

Figure 17 shows that the greatest rugosity is achieved by the Ar/(N50-Ar50) sample which, at the same time, presents the lowest (290 nm) hardness (figure 18). The Ar/(N70-Ar30) sample reached an average value of 267 nm despite the maximal N concentration in the mixture and the consequent ion impact on the piece. The Ar/(N30-Ar70) sample obtained

Al-Pattern N70-Ar30 N50-Ar50 N30-Ar70

Mixtures

 Alpattern N70Ar30 N50Ar50 N30Ar70

0 50 100 150 200 250 300

Load [gf]

an *Ra* value of 195 nm: the lowest in the present experiment.

0

Hardness [HV]

50

100

150

Roughness [nm]

Fig. 17. Rugosity

Fig. 18. Hardness

200

250

300

As follows from the hardness plot in figure 18, the maximal enhancement, up to 62HV0.1 at a 100 g load, was achieved with the highest nitrogen concentration mixture Ar/(N70-Ar30). By contrast, the Ar/(N30-Ar70) and Ar/(N50–Ar50) treated samples measured 60HV0.1 and 35HV0.1, respectively.

A comparative view of the diffractogrames shown in figure 19 indicates that the relative intensity of the Al peak at 2*θ* = 38.47° attained with the Ar/(N50-Ar50) mixture, decreases when the Ar/(N70-Ar30) and Ar/(N30-Ar70) ones are used (figures 19.a and 19.c). Quite the opposite with respect to the peak at 78.23° (figure 19.b) whose intensity is greater than those seen in figure 19.a y 19.c. The latter may be due to a low concentration of compounds like AlN and Al2O3 and to Al peaks either intrisically moderate or missing (such as the one expected at 2*θ =* 65.13°). Likewise, the Ar/(N50-Ar50) diffractogramme seen in figure 19.b, does not show the 2θ = 82.43° Al peak while the main Al peak is particularly reduced wich is ultimately attributable to the implantation process itself. The diffractogramme of figure 19.c points to the fact that a low N concentration favours the cubic phase of AlN at 2*θ*=41.80° (87-1053 JCPDS standard), 78.41° and 82.62° (46-1200 JCPDS standard).

Fig. 19. Diffractogrammes a) Ar/(N70-Ar30), b)Ar/(N50-Ar50) c)Ar/(N30-Ar70)

PIII for Aluminium Surface Modification 195

treated samples indicates a more electrochemically noble surface and enhanced tolerance to corrosion. The treated sample response is quite similar, irrespective of the temperature and

The samples treated with a mixture of nitrogen-argon were analyzed by means of Raman Spectroscopy in order to validate the results obtained from XRD (He-Ne laser at a 632.8 nm wavelength). Figure 21 shows the resulting Raman spectra for AlN. The 514 cm-1 peak, typical of AlN, is always visible, which coincides with the information provided by XRD diffractogrammes. Likewise, the spectra displayed in figure 4.a., where the absence of aluminium nitride is evident, are confirmed. The results of XRD and Raman spectroscopy show that the general sample improvement depends on the amount of N in the Ar/N

The mechanical improvement of aluminium alloys by conventional nitriding techniques is considerably complex whereas plasma immersion ion implantation provides a simple and effective way to enhance the wear resistance and corrosion tolerance of these alloys. The present study has contributed to the knowledge of the AlN structure in 6061-T6 aluminia following a low energy (2-6 KeV) PIII at a 75-150 µs pulse width. Such a process approaches the conditions of a glow discharge where the anode is a 15 cm long stainless steel rod, 3.5 cm in diameter, placed horizontally at the top of the cylindrical 304 stainless steel vacuum chamber 60 cm high and 30 cm in diameter which, in turn, plays the role of cathode of the discharge. The latter is fed by a DC power supply with a maximum output power of 1200 W, specifically designed and constructed from a current-source converter operating in a resonant mode. Several previously optimized work gas compositions were used: pure nitrogen, 30% argon/70% nitrogen, 50% argon/50% nitrogen, 70% argon/30% nitrogen providing ion doses in the order of 1021 ions/cm2. Samples were implanted at ~400°C for ~1 h periods. X-ray diffractometry, scanning electron microscopy, Vickers microhardness tests, profilometry, corrosion and Raman spectroscopy methods were applied to evaluate the treatment outcome. The highest microhardness values were achieved with the equal part gas mixture and a voltage bias. The greatest roughness was obtained by increasing the implantation pulse width up to 150 µs with the same mixture. The roughness seems to remain invariant when pure nitrogen is used provided that longer time implantation periods are completed. Increasing the surface microhardness of aluminium without jeopardizing its average rugosity depends critically on selecting the correct Ar and N

An ostensible improvement on the AlN microhardness results from the presence of the hexagonal crystalline phase and the elimination of the cubic one, the characteristic peaks of AIN in the near surface having been confirmed by Raman spectroscopy. A compromise is to be established between rugosity and microhardness through the several variables in the

The optimal conditions characterised in the present study for the PIII treatment of 6061T6 aluminium samples can be summarised as plasmas made out of mixtures of argon with, at least, 50% nitrogen, applied once the sample temperatures reach around 450°C. Thus, the implantation of the aluminium samples results in the surface formation of nitrides. X ray

duration of the treatment.

**4.4 Raman analysis** 

mixture concentration.

proportions to be used in a very low voltage PIII.

**5. Conclusions** 

process.

Fig. 20. Potentiodynamic polarization curves

Fig. 21. Raman spectra (a) Nitrogen, (b) 50%N/50%Ar mixture and (c) 70%N/30%Ar

### **4.3 Corrosion analysis**

Electrochemical potentiodynamic polarization tests were carried out within a cell containing 1 l of de-aerated 1.0N solution of H2SO4 as electrolyte. The measured results for untreated as well as treated samples are shown in figure 20. The more positive corrosion potential of the treated samples indicates a more electrochemically noble surface and enhanced tolerance to corrosion. The treated sample response is quite similar, irrespective of the temperature and duration of the treatment.

### **4.4 Raman analysis**

194 Recent Trends in Processing and Degradation of Aluminium Alloys

N50-Ar50

1 10 100

Log current density [A/cm2

200 400 600 800 1000

]

Wavenumber [cm-1

Electrochemical potentiodynamic polarization tests were carried out within a cell containing 1 l of de-aerated 1.0N solution of H2SO4 as electrolyte. The measured results for untreated as well as treated samples are shown in figure 20. The more positive corrosion potential of the

Fig. 21. Raman spectra (a) Nitrogen, (b) 50%N/50%Ar mixture and (c) 70%N/30%Ar

Untreated

(b)

(c)

(a)

Nitrogen

]


Intensity [a.u.]

**4.3 Corrosion analysis** 

Fig. 20. Potentiodynamic polarization curves


0

500

Potential [mV]

1000

1500

2000

The samples treated with a mixture of nitrogen-argon were analyzed by means of Raman Spectroscopy in order to validate the results obtained from XRD (He-Ne laser at a 632.8 nm wavelength). Figure 21 shows the resulting Raman spectra for AlN. The 514 cm-1 peak, typical of AlN, is always visible, which coincides with the information provided by XRD diffractogrammes. Likewise, the spectra displayed in figure 4.a., where the absence of aluminium nitride is evident, are confirmed. The results of XRD and Raman spectroscopy show that the general sample improvement depends on the amount of N in the Ar/N mixture concentration.

### **5. Conclusions**

The mechanical improvement of aluminium alloys by conventional nitriding techniques is considerably complex whereas plasma immersion ion implantation provides a simple and effective way to enhance the wear resistance and corrosion tolerance of these alloys. The present study has contributed to the knowledge of the AlN structure in 6061-T6 aluminia following a low energy (2-6 KeV) PIII at a 75-150 µs pulse width. Such a process approaches the conditions of a glow discharge where the anode is a 15 cm long stainless steel rod, 3.5 cm in diameter, placed horizontally at the top of the cylindrical 304 stainless steel vacuum chamber 60 cm high and 30 cm in diameter which, in turn, plays the role of cathode of the discharge. The latter is fed by a DC power supply with a maximum output power of 1200 W, specifically designed and constructed from a current-source converter operating in a resonant mode. Several previously optimized work gas compositions were used: pure nitrogen, 30% argon/70% nitrogen, 50% argon/50% nitrogen, 70% argon/30% nitrogen providing ion doses in the order of 1021 ions/cm2. Samples were implanted at ~400°C for ~1 h periods. X-ray diffractometry, scanning electron microscopy, Vickers microhardness tests, profilometry, corrosion and Raman spectroscopy methods were applied to evaluate the treatment outcome. The highest microhardness values were achieved with the equal part gas mixture and a voltage bias. The greatest roughness was obtained by increasing the implantation pulse width up to 150 µs with the same mixture. The roughness seems to remain invariant when pure nitrogen is used provided that longer time implantation periods are completed. Increasing the surface microhardness of aluminium without jeopardizing its average rugosity depends critically on selecting the correct Ar and N proportions to be used in a very low voltage PIII.

An ostensible improvement on the AlN microhardness results from the presence of the hexagonal crystalline phase and the elimination of the cubic one, the characteristic peaks of AIN in the near surface having been confirmed by Raman spectroscopy. A compromise is to be established between rugosity and microhardness through the several variables in the process.

The optimal conditions characterised in the present study for the PIII treatment of 6061T6 aluminium samples can be summarised as plasmas made out of mixtures of argon with, at least, 50% nitrogen, applied once the sample temperatures reach around 450°C. Thus, the implantation of the aluminium samples results in the surface formation of nitrides. X ray

**9** 

*Italy* 

**Optimizing the Heat Treatment** 

Andrea Manente1 and Giulio Timelli2

*1Cestaro Fonderie Spa* 

**Process of Cast Aluminium Alloys** 

*2University of Padova, Department of Management and Engineering* 

The unfailing increased use of light alloys in the automotive industry is, above all, due to the need of decreasing vehicle's weight. The same need has to be taken into account in order to face up also both energetic and environmental requirements (Valentini, 2002). In terms of application rates, Al and its alloys have an advantage over other light materials. The reduced prices, the recyclability, the development of new improved alloys and casting processes, the increased understanding of design criteria and life prediction for stressed components and an excellent compromise between mechanical performances and lightness are the key factors for the increasing demand of Al alloys. A consolidated example of aluminium alloy employment regards the production of wheels, which, together with an improved aesthetic appearance, guarantees an improvement of driving, like directed consequence of the inertia reduction. These critical safety components are somewhat unique as they must meet, or exceed, a combination of requirements, from high quality surface finish, as wheels are one of the prominent cosmetic features of cars, to impact and fatigue performance. Due to their excellent castability and good compromise between mechanical properties and lightness, AlSiMg alloys are the most important and widely used casting alloys in wheel production (Conserva et al., 2004). Further, the increasing application of these alloys has been driven by the possibility to improve the mechanical properties of cast components through the use of heat treatments. Various heat treatments, e.g. different combinations of temperatures and times, have been standardized by Aluminium Associations and they are used in Al foundry depending on the casting process, the alloy type and the casting requirements (ASM Handbook, 1990). Standard T6 heat treatment is generally applied in wheel production. This heat treatment provides two beneficial effects for cast aluminium alloy wheels: an improved ductility and fracture toughness through spheroidization of the eutectic silicon particles in the microstructure and a higher alloy yield strength through the formation of a large number of fine precipitates which strengthen the soft aluminium matrix (Zhang et al., 2002). The T6 heat treatment comprises three stages

(ASM Handbook, 1991): solution heat-treating, quenching and artificial aging.

*Solution heat-treating* at relatively high temperature is required to activate diffusion mechanisms, first, to dissolve Mg-rich phases formed during solidification and, then, to homogenize the alloying elements, such as Mg and Si, so as to achieve an elevated yield stress subsequent ageing (ASM Handbook, 1991). Further, the solution heat treatment

**1. Introduction** 

diffraction of the implanted pieces reveals the presence of AlN in the cp and hcp crystalline phases where the peak intensities increase along with the nitrogen content. The presence of the hexagonal phase has been detected when either pure nitrogen or a 50% mixture have been used, suggesting a correlation between the h phase and the enhanced microhardness. Raman spectroscopy has confirmed the signature peak of AlN and, in addition to XRD, shows that the general surface improvement is enhanced with the N proportion in the Ar/N mixture concentration.

A compromise between high hardness and low roughness in pure nitrogen is observed due to a competition between sputtering and nitriding after, at least, 1 hour of treatment. In particular, maximal microhardness values were found in samples treated with the equal part mixture. The best roughness was achieved with this gas mixture in all cases, although increasing along with the implantation pulse width up to a 300 nm peak at 150 μs. Such a performance can be maintained in a pure nitrogen plasma, provided that longer 1 hour implantation periods are performed.

### **6. Acknowledgment**

The authors are grateful to the technical collaboration received from Israel Alejandro Rojas Olmedo, Hannalí Millán Flores, Everardo Efrén Granda Gutiérrez, María Teresa Torres Martínez, Pedro Angeles Espinoza and Isaías Contreras Villa,.

### **7. References**

