Investigations of the Kinetics of Surface Treatments

by Advanced Methods

____________________________

TABLE OF CONTENTS

Summary of Principal Results

Part 1 Introduction

1. General Description; Methods Developed in this Project

2. Detailed Description of ICP Spectroelectrochemistry

5

7

9

13

Part II The Alkaline Stability of Phosphate Coatings

3. Alkaline Resistance as a Characteristic Measurement for Conversion Coatings

4. Synthesis and characterization of phosphate layers

5. Leaching studies using inductively coupled plasma spectroelectrochemistry

6. Leaching studies using in-situ Raman Spectroscopy

19

21

23

27

33

Part III Kinetics of Surface Treatment

7. The Effect of Accelerators on the Phosphating Speed

8. Zinc aluminum stability in a degreasing bath

9. The Use of the Quartz Crystal Microbalance for the Study of Surface Treatments

41

43

47

49

Part IV Nucleation and growth of phosphate layers

10. Nucleation and Growth by Environmental SEM

11. Germination, Growth and Destruction of a Phosphate Film on Zinc : An Atomic Force

Microscopy Investigation

57

59

63

Part V Conclusions and perspectives

12. Conclusions and perspectives

65

67

References 71

List of FIGURES & TABLES 75

Figures 79

Tables 91

3

SUMMARY OF PRINCIPAL RESULTS

The Development of New Experimental Methods for Following Surface Reaction

The goal of this project was to develop new techniques for the study of surface treatment

reactions during the reaction. Surface treatment reactions involving simultaneous dissolution of

substrate metal and film growth therefore a number of complementary techniques were

required to fully characterize the reaction. ICP spectroelectrochemistry gives a direct measure of

the metal and alloy dissolution rate by monitoring the elemental composition of the electrolyte.

Quartz crystal gravimetry measures the mass changes of the sample that reflect the difference

between mass loss by dissolution and mass gain by film formation. When coupled with the ICP

technique, the rate of film formation can be determined. In situ Raman yields information on

solid state changes in the surface film. All of these techniques are combined with standard

electrochemical measurements to yield a thorough analysis of the reaction process. In addition,

environmental scanning electron microscopy and atomic force microscopy have been used to

observe nucleation and growth of phosphate films.

Alkaline Stability of Phosphate Conversion Coatings

ICP spectroelectrochemistry, In situ Raman spectroscopy, quartz crystal microgravimetry, and

atomic force microscopy were used to investigate the nature of the reaction between phosphate

conversion coatings and alkaline solutions. The ICP method was used to quantitatively analysis

the residual layers remaining after complete reaction. It was determined that the alkaline

stability of a phosphate layer was improved in proportion to the quantity of Mn+2 and Ni+2

incorporated into the layer. The mechanism of alkaline leaching involves ion exchange of

hydroxide and phosphate with the formation of slightly soluble hydroxide crystals. It is

proposed that the absorption of the hydroxide ion in this manner, without destruction of the

crystals themselves, is an important factor in improving the cosmetic corrosion resistance during

cathodic delamination. A variety of commercial post treatments were also investigated and they

were found effective in reducing the alkaline reaction rate. During conditions of anodic

delamination, it was found that the presence of Ni+2 was more important than Mn+2 in promoting

corrosion resistance. This is probably due to the presence of metallic nickel on the zinc surface

surrounding the crystals.

Kinetics of Phosphating

ICP spectroelectrochemistry and the quartz crystal microbalance were used to study the

reactivity of galvanized steel with phosphate solutions. A particular emphasis was placed on the

role of different accelerators including nitrate, hydrogen peroxide, hydroxylamine, and

nitrobenzoic acid. It was found that the more effective accelerators are strong oxidants, but lead

to a stabilization of the anodic dissolution reaction probably by reducing hydrogen evolution.

For the all the accelerators except hydroxylamine, there was a net reduction in the dissolution

rate when in the presence of phosphate. For nitrate, clear evidence of film formation was

observed. This suggests that nitrate is unique in that, in addition to its accelerating role, it

passivates the surface between the crystals. This may be a partial explanation of why nitrate is

found to be such an effective component of phosphating solutions. The role of Al dissolution in

the phosphate bath was investigated but seems to be complex.

5

Zn-Al Alloy Stability in Alkaline Solution (Degreasing)

ICP spectroelectrochemistry was used to measure the partial currents of Al and Zn during the

open circuit reaction and during a polarization curve for HDG and Galfan (5% Al). A complex

interplay between Al and Zn dissolution was observed. It was demonstrated that majority of Al

dissolution in alkaline solution is due to the oxidation of Al metal rather than previously formed

oxides.

Phosphate Layer Germination by Environmental Scanning Electron Microscopy

Observations were realized on cold rolled steel, electrogalvanized, galvannealed, Aluzinc

(composition of the bath in weight percent: 55%Al-43.5%Zn-1.5%Si) and Alusi (88%Al-9%Si-

3%Fe) surfaces. Growth kinetics, covering, size and morphology of phosphate crystals were

compared. Surface roughness was found to play an important role on the phosphatability of the

metal substrate: on the one hand the roughness due to skin-pass (seen on hot dip galvanized and

galvannealed), on the other hand the roughness of the coating (well observed on

electrogalvanized and also galvannealed). The chemical effect with an important Aluminum

content in the coating (like Aluzinc and Alusi) is also noticeable in the phosphating behavior of

the metal substrate.

6

Part 1

Introduction

7

CHAPTER 1

GENERAL DESCRIPTION; METHODS DEVELOPED IN THIS PROJECT

1.1 Scientific Objectives and Philosophy of this Study

The objective of this project is to develop advanced methods for the study of surface

treatment. For practical reasons, we felt the best way to demonstrate these methods would be to

study the mechanism and kinetics of the phosphating process and the chemical properties of the

phosphate layer. Specifically we would address the role played by heavy metal ions and

accelerators in the phosphating reaction, so as eventually to propose new environmentally

friendly solutions.

Although the proposition was general, we chose to focus on phosphating for a number

of reasons. First of all, phosphating is still the most common conversion layer in the automotive

industry, and it does not appear likely to be replaced in the foreseeable future. Second, there are

a number of unanswered questions concerning phosphate conversion coatings which limit the

development of steel products - how can we speed up the process so as to perform it on a

galvanization line without the use of toxic additives as accelerators, how do we eliminate the

use of Ni+2 in the solution? More generally, how does the phosphate layer work? The third

reason for our choice of phosphating purely pragmatic in view of our experimental methods -

is that reaction is sufficiently slow and the layer sufficiently thick that the methods developed in

this project should work very well. Finally we cite the large body of scientific data on

phosphating kinetics and phosphate layers, arguably far more than is available for any other

type of surface treatment, and certainly more than for any surface treatment without chromate.

Two bibliographic reviews of this literature have been published by members of this project

[1,2]. Our study differs in two respects from most previous studies for two important reasons :

1 --- The use of advanced methods yields far more information than has usually been

available for previous studies. The methods developed for this work include ICP

spectroelectrochemistry, Raman spectroscopy and quartz crystal microbalance, coupled

with electrochemical methods, and environmental scanning electron microscopy,

2 --- The use of synthetic phosphating baths of composition consistent with modern practice

has allowed us to vary the composition in a coherent manner so as to study the influence

of different bath components. Almost all previous studies have either used industrial

baths of essentially unknown composition, or model laboratory solutions having little

or no relevance to practice.

Nevertheless, the study of phosphating was only a preliminary goal to allow us to develop our

laboratory methods and general scientific approach. In the present work we begin an intial study

of the degreasing reaction which is fundamental to all surface treatment, and nonchromate post

treatments. It is our intention that the experimental methods developed in this project will, in the

future, make many fundamental contributions to an understanding of surface treatment and

anticorrosion behavior .

1.2 Inductively Coupled Plasma ICP Spectroelectrochemistry

We have developed an inductively coupled plasma atomic emission

spectroelectrochemical system (AESEC) by which we can continuously monitor the solution

composition downstream from an electrochemical flow cell, as described in several recent

publications [3, 4, 5, 6] and the cell has been patented [7]. A schematic diagram of the system

and the electrochemical flow cell is shown in Fig. I.1 and I.2. The AESEC method directly

measures the partial elemental dissolution reactions either under spontaneous reaction

conditions or with electrochemical polarization. A major advantage of this technique is that it

can measure a large number of elements simultaneously. As this is a new technique, we present

9

a detailed description in the next chapter including an analysis of the characteristics of the flow

cell.

The inductively coupled plasma source is particularly well suited for this application for

several reasons : (a) it is sensitive to most metallic elements as well as many nonmetals, with

detection limits in the ppb range and below; (b) it has a linear dynamic range for most

wavelengths extending over five to six orders of magnitude; (c) it is subject to relatively few

matrix effects due to the high temperature of the plasma; and (d) in practice almost any liquid

can be aspirated into the plasma for analysis including electrolytes with very high salt

concentration. The later point is very important for the study of industrial surface treatments.

1.3 Electrochemical quartz crystal microbalance

Although the AESEC method does measure the rate of the dissolution reactions, it does

not directly give information on the rate of film formation. We have therefore coupled this

technique with a very sensitive gravimetric method, the electrochemical quartz crystal

microbalance (EQCM). This allows us to simultaneously measure the rate of metal dissolution

and (under proper conditions) the mass changes at the electrode surface. The principle of the

microgravimetric technique is shown in Fig. I.3. Fig. I.4 shows a schematic diagram of the ICP

flow cell used in this work. We have used this combination of EQCM -AESEC to study anodic

dissolution of Zn in alkaline medium and preliminary results have been published. Although we

have not used the combination of techniques to study phosphating reactions, the authors of this

proposition have independently published work on the use of EQCM alone for phosphating

[8,9] and coupled with the ICP for degreasing [5] and chromating [6] of zinc.

Earlier work with the EQCM was limited by the fact that, by itself, only a measure of

the total mass change can be performed. Most surface treatments involve simultaneous

dissolution and precipitation. Further, the nature of the surface films formed during surface

treatment is often incompatible with the assumptions of the Sauerbrey equation relating

frequency changes to mass changes. To this end, the coupling of the EQCM with AESEC is

particularly interesting in that : (a) a total mass balance can be performed since the dissolution

rate is measured independently ; (b) the frequency -mass relationship can be confirmed under

steady state conditions, when the mass change is dominated by metal dissolution.

For mass measurement with the EQCM the piezoelectric properties of small quartz

plates are used to force vibration by connecting the quartz plate to a suitable oscillatory unit.

For this purpose the quartz plate has to be covered with conductive layers (usually Au or Pt) on

both sides, which work as electrodes. Depending on the crystalline direction of the cut of the

quartz plate, the piezoelectric effect gives rise to a certain vibrational mode. The so-called AT-

cut has proven to be the most suitable for use in electrochemistry. The forced vibration mode is

the shear deformation as shown in Fig. I.3.

The mass attached to the quartz surface determines the frequency of its vibration. If the

mass is increasing, the resonant frequency decreases and vice versa. The relationship of mass

and frequency is described by the Sauerbrey equation.

f = - Cf .f2/A. m

The mass sensitivity depends only on the dimensions of the quartz (area A), the

fundamental frequency f0. (without any additional mass on the quartz) and a proportional factor

Cf, which considers mechanical and physical properties of the quartz like density and elasticity

module. Deviations of this relationship may be caused when using the EQCM in an electrolyte.

Generally, the electrolyte dampens the vibration of the quartz, because of an energy loss due to

the formation of a shear wave inside a thin liquid layer near the quartz surface. The magnitude

of this effect is determined by the viscosity of the electrolyte, which in turn depends on the

temperature.

10

A film of material deposited on the QCM contributes directly to mass change. However,

when the growing film is rough, parts of the electrolyte may be trapped within depressions in

the layer and contribute to the apparent mass change. A rigid junction of the deposited film is

also crucial for the fulfilment of the theoretical relationship. Loosely precipitated layers may

cause deviations from the apparent mass as determined by the measured frequency.

1.4 In situ Raman Spectroscopy

Neither of the above techniques directly yield information concerning the molecular

composition of the conversion coating. It is of interest therefore to combine these techniques

with other methods such as vibrational spectroscopy. In particular, Raman spectroscopy is a

very sensitive form of vibrational spectroscopy, and like infrared spectroscopy, yields molecular

bonding information of the sample. It is particularly well suited for in situ measurements at the

metal/electrolyte interface because bulk water interferes very little with the measurement*.

Further the use of a laser source permits one to focus the beam directly onto the desired part of

the surface. Numerous studies involving in situ Raman spectroscopy coupled with

electrochemistry and occasionally with the quartz crystal microbalance, have appeared in the

literature with considerable impact on questions related to catalysis, corrosion and

electrochemistry. For instance, the technique has proven particularly interesting for the

investigation of oxide scale pickling [10] to cite one example within the context of surface

treatment.

Modern Raman spectrometer require recording times below 1s for one spectrum. When

used in an in-situ cell with electrolyte covering the sample, the Raman light is dampened and

longer recording times are necessary. Therefore we did no investigations according the

monitoring of the film growth during the rather fast phosphating reaction. Raman spectroscopy

was exclusively applied to the investigation of the alkaline stability of final phosphate layers.

To our knowledge, no in situ Raman studies have been performed concerning the

chemical stability of phosphate layers in alkaline media. This is surprising as Raman is very

sensitive to phosphate, and the spectra are highly characteristic of the exact nature of the

phosphate species including the crystalline form and the level of hydration. Hopeite,

Zn3(PO4)2.4H2O which is one of the mores common phosphate species on zinc surfaces, has a

strong symmetrical stretching vibrations at 996 cm-1 [8, 11,12]. Therefore, we should be able to

directly follow in situ the leaching of phosphate on the sample and to identify new phases or

intermediate products as they form. This should yield an unprecedented insight into the

dissolution of phosphate layers in alkaline media.

A block diagram of the in situ Raman spectroscopy system is shown in Fig. I.5. A

commercial Raman spectrometer from Dilor, LABRAM, was used. This is a confocal Raman

microscope with a He-Ne laser excitation at 633 nm. The spectrometer was equipped with a

CCD-detector and an 1800 grating. Spectra were collected in a range of approximately 1000 cm-

1 at a resolving power of 0.9 cm-1. The laser power incident on the sample surface was ~2.4

mW. For the in-situ measurements in the flow cell a 50x objective (Olympus) with a numerical

aperture (NA) of 0.5 and a working distance of 8 mm was used. The confocal pinhole diameter

was 1000 m, and the slit width was adjusted to 400 m.

A flow cell has been constructed from teflon as shown in Fig. I.6. For the work

described here, Ag/AgCl reference electrode could be placed in the compartment. Since Raman

spectroscopy was only applied to the investigation of the alkaline stability, a detailed description

of the experimental procedure will be found there (Part II).

* The strength of the Raman effect is determined by the derivative of the molecular polarizability of the sample with respect to the vibrational coordinates. For comparison, the IR effect is determined by the derivative of the dipole moment. It is interesting that molecules with a large dipole moment like H2O give very strong IR signals, but much weaker Raman signals.

11

1.5 Environmental scanning electron microscopy

In addition to following the course of the surface treatment reaction as a function of

time, it is of interest to measure crystal germination. The conventional method of performing

this measurement is by microscopic observation of the surface after phosphating for a very short

period, on the order of several seconds. In this work the environmental scanning electron

microscope is used to directly observe the germination of phosphate crystals at the metal/

electrolyte interface after short exposure to the electrolyte. Environmental scanning electron

microscopy is used to image a surface as in regular SEM, but under atmospheric conditions, and

with the unique advantage that nonconducting surfaces can be studied directly, without the

introduction of a conducting deposit such as gold or carbon.

The principle of the technique involves amplifying the number of secondary electrons

produce by the interaction between a primary electron beam and the sample by the mean of

collisions of gaseous molecules in an electric field. The collisions provide new electrons called

environmental secondary electrons and positives particles. The electrons are collected by the

ESE detector and the positives particles are attracted by the sample, which neutralizes the

negatives charges of the sample and eliminates the problems of charging often associated with

the analysis of nonconducting conversion layers by regular SEM.

12

CHAPTER 2

DETAILED DESCRIPTION OF ICP SPECTROELECTROCHEMISTRY

2.1 Spectrometer and flow cell description

A block diagram of the atomic emission spectroelectrochemistry system was shown in

Fig. I.1 and I.2. The ICP spectrometer is used to measure the composition of the electrolyte

downstream from the dissolution cell permitting a qualitative identification of the soluble

reaction products and a quantitative measure of the simultaneous dissolution rates of the

products. A commercial ICP atomic emission spectrometer from Jobin Yvon, Inc (JY 74) is

used in this work. The plasma source consists of a 40 MHz, 1 kW inductively coupled Ar

plasma, into which the electrolyte sample is continuously aspirated. The spectrometer consists

of a polychromator for the simultaneous detection of 17 predetermined elements, and a

monochormator for the detection of an additional element of choice. The polychromator is

based on the Paschen-Runge configuration and was equipped with a holographic grating of

3600 groves/mm and 22 photomultiplier tubes. The number of photomultipliers is larger than

the number of elements because several elements are measured at more than one wavelength to

avoid interelement interference. The theoretical resolution of the polychromator is 0.028 nm

covering a range of 165 nm to 425 nm. The monochromator is based on a Czerny-Turner

configuration with a focal length of 64 cm and a holographic grating of 2400 groves /mm,

yielding a practical resolution of 0.010 nm covering a spectral range of 165 nm - 800 nm. Both

the polychromator and the monochromator are nitrogen purged.

With the system as configured, the concentration of 18 different elements could be

measured simultaneously. This number could be increased by adding additional

photomultipliers, limited by the spatial requirements in the spectrometer focal plane. The

commercial spectrometer was modified for the purposes of this work by the addition of a rapid

data acquisition system, which monitors the entire array of photomultipliers and displays the

signals in real time, and permits software control of the photomultiplier voltage. With this

system, the photomultiplier voltages are sampled at a rate of 0.1 kHz with a 12 bit A/D

converter, and the results are averaged over an arbitrary integration period, which in this work,

was normally 0.5 s or 1 s. Because of the averaging period, the dynamic range of the system

was significantly greater than that of the 12 bit A/D converter. The spectrometer data

acquisition electronics and software were further modified so that the potential and current

signals from the analog output of the EG&G M273A potentiostat/galvanostat could be recorded

simultaneously with the spectroscopic emission data using the same A/D converter and

integration/collection algorithm. This ensures that all simultaneous data are on exactly the same

time base.

A cyclonic aspiration chamber, chosen for its rapid response time, and a concentric

glass nebulizer, chosen for its good performance with electrolytes of high salt concentration,

were used in this work.

A two compartment, three electrode flow cell was constructed from teflon as illustrated

in Fig. I.7. The surface of the working electrode (0.52 cm2) is brought in contact with the

flowing electrolyte (2 - 12 ml/min) in a small volume compartment (0.2 ml), separated from a counter electrode compartment by a porous membrane allowing passage of ionic currents while

preventing bulk mixing of the two electrolytes. For the work described here, a standard

reference electrode was placed in the counter electrode compartment. In an electrolyte of 0.1 M

ionic strength, the ohmic drop across the membrane was estimated at approximately 5.0 from

high frequency impedance measurements. The geometrical surface area exposed to the

electrolyte was measured by dissolving a 5 m zinc coating from an electrogalvanized steel

sample and measuring the area of the exposed steel. The value was found to be 0.53 0.02 cm2.

13

An EG&G PAR273A potentiostat / galvanostat, programmed from the front panel was

used for control of all electrochemical experiments. The analog signals from the potentiostat

were amplified and offset to fall within a 0 to 10 V range so that they would resemble the

photomultiplier signals. In this way the potential and current of the potentiostat were monitored

simultaneously with the photomultipliers using the same data acquisition card and software.

This insures that the spectroscopic and electrochemical data are on the same time scale.

2.2 Quantitative Relationship between Rate and Intensity

This method was developed in our laboratory, we present a detailed description of the

basic calibration of the system. ICP emission spectroscopy is a well-established method of

elemental trace analysis for liquid samples and a large number of commercial systems are

available. The methodology is described in numerous texts and review articles, and the treatise

of Bowmans [13] can be considered as a standard reference. For the work described here it is

important to understand that the spectrometer system can be divided into three modules: (1) the

nebulization / aspiration system into which the liquid sample is transformed into a fog of very

small droplets, a fraction of which are aspirated into the (2) inductively coupled plasma. The

high temperature of the plasma ( 8000 K) leads to a rapid desolvation of dissolved species, which are usually reduced to the atomic state. The spectral emission of the resulting atoms in

the plasma then passes into (3) the optical detection system consisting of (a) a polychromator

with an array of photomultipliers preset for a specific wavelength, and (b) a monochromator

with a single photomultiplier for an adjustable wavelength. The signal produced by the

photomultipliers is monitored as a function of time.

The basic principle of analytical ICP atomic emission spectrometry is that the emission

intensity (I ) of a given element, M, in the plasma at a wavelength specific to M, is proportional to the concentration of that element (CM) in the electrolyte stream. Assuming that

only the element M emits radiation at wavelength, , the standard quantitative relationship is :

CM = (I-I) 1

where is the constant of proportionality and I is the background intensity. Note that the concentration C, will be expressed in g/cm3 (or ppm) as is customary in analytical applications

of ICP spectroscopy. With appropriate choice of this proportionality is normally valid over four to six orders of magnitude, limited at low concentrations by the fluctuations of I due to

background emission from the plasma and scattered radiation, and at high concentrations by

self-absorption [13] . At intermediate concentrations, where I >> I, the noise level is

determined by fluctuations in the nebulization system and is a constant percentage of the

analytical signal.

Under steady state conditions, the elemental dissolution rate of component M of the

sample working electrode within the flow cell, (RM (t) expressed in g/ sec), is equal to its

concentration downstream from the cell multiplied by the flow rate of the electrolyte ( expressed in cm3/s), :

RM (t) = CM(t) = k ((t) - I) 2

In order to compare spectroscopic dissolution rates with the Faradaic current measured by the

potentiostat, it is convenient to express the elemental dissolution rate in terms of a partial

elemental current, in A :

iM(t) = nFRM(t)/MA 3

Where MA is the atomic mass of element M. In this manner, the external current iex (defined as

the current measured by the electrometer of the potentiostat, in contrast to the partial elemental

current which is calculated from the spectroscopic transient) would be identical to iM if the only

reaction were M M+n + ne- with 100% Faradaic efficiency. In general however, this cannot be assured: either the n value will be unknown, metal ion release may be due to non

14

electrochemical dissolution of oxides or other corrosion products, or the formation of insoluble

surface films will compete with dissolution. The expression of metal ion release in terms of an

equivalent current should be considered only as a convenience to facilitate comparison of two

fundamentally different data sets (spectroscopic and electrochemical) and does not imply that

the release mechanism is known or even that it is electrochemical in nature.

The quantitative relationship between emission intensity and dissolution rate has been

verified by measuring the dissolution rates of pure metals under imposed current conditions. An

example for pure copper in 1.2 M HCl is shown in Fig. I.8. The experiment on the left shows

the signal transient resulting from the galvanostatic application of 300 A to the copper

electrode in the flowing stream of 1.2 M HCl. The solid line gives the total current transient,

which consists of a step from zero current to 300 A for 5 minutes, followed by a return step to

zero current. Both the current and spectroscopic intensities are expressed in as equivalent

concentrations, which are calculated by Equation 1 for the spectroscopic intensities; and by Eqn.

4 for the total current :

CM, eq = MA iex / n F 4

where n is assumed to be 1 for copper dissolution in 1.2 M HCl. Again, note that the units are

chosen to give the concentration in the conventional analytical units of g/cm3. Following the

imposition of a constant current, there is a lag period, t, of approximately 10 seconds, during

which no copper emission is detected. This time is associated with the transport of the ions to

the spectrometer and depends on the flow rate and the length of tubing between the cell and the

nebulizer. This is followed by a signal increase to a steady state value. Approximately 43

seconds are required to reach 95% of the steady state value. Quantitatively, the steady state

value is 10% higher than that predicted from the imposed current using Equation 3. After five

minutes, the programmed current returns to zero. Following the period t, the measured

concentration of copper drops to 5% of the steady state value after approximately 50 s.

Figure I.9 illustrates how this technique may be used to directly measure the number of

electrons, n, transferred during the anodic dissolution of pure metals. The rate of dissolution of

copper was measured as a function of applied current in 1.2 M HCl and 0.60 M H2SO4. Values

of n=0.86 in 1.2 M HCl and n= 1.94 in 0.6 M H2SO4 were determined, in good agreement with

the well known Cu(I) and Cu(II) species formed respectively in these two electrolytes [14].

Reactions occurring at a solid / liquid interface involve diffusion processes in the

electrolyte. In industrial surface treatment, the hydrodynamics differ according to the method of

application: immersion, spray, roll coat, Therefore, in comparing experiments between

laboratories, or in extrapolating from the laboratory scale to the industrial scale, it is necessary

to understand the different hydrodynamic regimes which might be encountered. Also, for the

ICP flow analysis method, the hydrodynamic conditions may influence the measurement of the

rate, which relies on downstream concentration measurements. Therefore, we have proposed a

number of typical experiments which can be performed for the flow cells used in this work.

2.3 Temperature Calibration

Temperature control is an important consideration in the simulation of surface

treatments. This is performed for the ICP cell by circulating water from a thermostated water

bath through a hollow copper plaque as shown in the schematic diagram. To calibrate the

temperature, a thermocouple was spot welded onto the surface of a steel sample of comparable

thickness to those used in this study. Fig. I.10 gives the difference of temperature between the

steel plaque and the water bath in the presence and absence of recirculating water. This result

demonstrates that approximately 5 C to 6 C are lost in the cell, even though the electrolyte is

not heated.

15

2.4 Thickness of the Diffusion Layer

A solution of 10 mM K3Fe(CN)6 in a 1 M NaOH solution is passed through the flow

cell with a Pt electrode positioned as the working electrode sample. A potential is imposed such

that the Fe(III) is reduced to Fe(II) at a diffusion limited rate. In the absence of flowing

electrolyte, the current decreases with time, and is directly proportional to the 1/t1/2 as described

by the Cottrell equation. The diffusion coefficient for Fe(III) can be determined from the slope

of the resulting line of i vs. 1/t1/2.

In the presence of electrolyte flow the current decreases until it reaches a limiting value

which is given by ilimit = nFAD/ . The diffusion layer measured in this way is of course, only an average value, as it is expected that the thickness will vary across the sample, being very large

near the o-ring and greatly reduced near the center of the channel. Fig. I.11 gives the variation

of the diffusion length with flow rate for the ICP-SEC flow cell.

2.5 Residence Time Distributions

Fig. I.8 and I.9 demonstrate the quantitative nature of steady state measurements.

However, a significant response time was observed associated with diffusion and mixing in the

flow cell and to a lesser degree, mixing in the nebulization chamber. The limited time

resolution is a problem for many applications, and therefore, it is of interest to treat the problem

in more detail. The measured concentration transient, CM(t), is determined by the convolution of

the dissolution rate, RM(t) with the distribution of residence times, H(t) [15]. Because of the

complexity of the system, H(t) must be determined experimentally. This was done by measuring

the concentration-time response to a delta function of dissolution : A copper electrode was

placed in contact with the flowing 1.2 M HCl for five minutes at the rest potential, followed by

a 0.5 second galvanostatic pulse, and then a return to the rest potential. The transitory emission

intensities were measured with a time resolution of 0.5 seconds. A typical result is shown in

Fig. I.12, with an applied current of 10 mA and at a flow rate of 2.3 ml/min. Three different

experiments are superimposed to demonstrate the reproducibility. The 0.5 second galvanostatic

pulse is considered to be sufficiently short, on the time scale of these experiments, that it

approximates a delta function.

The pulsed anodic dissolution of copper gives rise to an asymmetric peak, which

increases quickly to a maximum and then returns slowly to the background signal, I.

Although, copper dissolution is known to proceed via the formation of a CuCl film followed by

chemical dissolution of the film [16][17][18], the conditions of these experiments were chosen

so that the initial film would be formed very rapidly, and its dissolution would be practically

instantaneous by reaction with Cl- in the electrolyte.

It is of interest to define several time parameters in Fig. I.12. The previously mentioned

lag time, t, is the time between the anodic pulse and the first point of data which rises above

background. The inset to Fig. I.10 shows the signal variation around t, and demonstrates that

this point is readily recognizable to within the error of the data acquisition rate (0.5 seconds).

This time is associated with the time necessary for the copper ions to travel between the

between the dissolution cell and the nebulization system.

Considerable progress has been made in modeling transients obtained by flow-injection

analysis from first principles [19], and it is tempting to try a similar approach here.

Unfortunately, a number of complex physical processes contribute to the broadening of H(t).

These processes include diffusion from the surface into the flowing electrolyte stream, mixing

in the channel flow cell, spreading out during the laminar flow in the capillaries between the cell

and the spectrometer, and the complicated nebulization system itself. Further, the dissolution

cell is not ideal and the diffusion distances probably differ from the edge to the

16

center of the sample and from top to bottom of the channel flow cell. Finally, there is probably a

significant dead volume near the o-ring. The situation could be simplified by using a small

electrode mounted in a nonconducter and centered in the flow cell, however, this approach

would lower the surface area and the sensitivity accordingly, and would further complicate

sample preparation. Nevertheless the use of geometrically limited electrodes represents a

particularly promising method of improving the time resolution of the system for certain

applications.

Despite the complexity of the system, the curve of Fig. I.12 can be simulated by an

empiric function in the form of a log-normal distribution :

2)(

tQ

IItH

=

=

*

2ln

t

e

5

where and * are characteristic time constants for the log normal distribution and t = t(experimental time scale)-t. Q is the integral of the transient in arbitrary units. The advantage

of the log-normal distribution is that it predicts the parabolic form of the rising part of the

residence time distribution. For the treatment of experimental data, it is convenient to define a

simplified version of the log-normal distribution in which the preexponential factor is constant.

2)(

=

=

Q

IItH

4

1

e

t

e

2ln

6

Fig. I.13 displays the data of Fig. I.12 as a function of t on a logarithmic scale. A

symmetrical peak is observed with a maximum at 14.8 seconds. The solid curves show non

linear least square curve fits to Equation 5 and 6 superimposed. Obviously, the two models

cannot be distinguished from one another. For Eqn. 6, is located at the peak maximum, in this case 14.8 s, and for Eqn. 5, * is located to the right of the peak maximum, at 25.7 seconds. By taking the derivative of Eqn. 5, it can be shown that = *e-1/(2) and that is the same for the two models.

2.6 Characterization of the ICP- QCM flow cell.

Figures I.14 I.16 show data from a single experiment involving a 4 s galvanostaic

pulse of Cu to the QCM ICP flow cell (Fig. I.4). Figure I.14 compares the electrical current

and the QCM response. At the beginning of the galvanostatic pulse there is a small mass

increase, followed by a rapid decrease. After the pulse is over, the decreasing mass continues for

about 20 to 30 seconds. These results demonstrate that Cu dissolution in the QCM cell proceeds

by the formation of a CUCl film at the surface, and the dissolution of this film continues after

the galvanostic pulse has been completed. In order to use this as a reliable method of measuring

the time constant distribution for the cell, another electrolyte should probably be chosen so as to

avoid this problem. Nevertheless, the amount of Cu dissolving during the secondary time

period is small and probably does not have a major effect on the measured distribution.

Fig. I.15 gives the time constant distribution determined for this system. By the analysis

presented in the previous section, this data gives a time constant of 13 seconds in good

agreement with that obtained for the standard ICP cell.

17

Fig. I.16 gives the surface concentration transient calculated from the variation of the

open circuit potential following the galvanostatic pulse. The combination of these two data sets

should allow us to separate the time constant distribution into that due to mixing in the cell and

nebulizier of the ICP, and that due exclusively to diffusional processes. However the formation

of CuCl films would complicate the interpretation of the open circuit potential data and further

analysis is pending.

18

Part II

The Alkaline Stability of

Phosphate Coatings

19

CHAPTER 3.

ALKALINE RESISTANCE AS A CHARACTERISTIC MEASUREMENT FOR

CONVERSION COATINGS

Phosphate conversion coatings are commonly used on steel, zinc, and aluminum alloys,

particularly in association with cataphoretic electropainting in the automotive industry. The

enhanced corrosion resistance associated with the use of a phosphate on steel and zinc, has been

well documented [20,21,22]. The primary role of the phosphate film for this application is to

anchor the electrodeposited paint into the porosity, increasing adherence in a lock and key type

of mechanism [23]. Therefore, the quality control of phosphate coatings is for the most part

based upon static features such as crystal morphology, porosity and uniformity on the surface

[24,25,26]. The integrity of the metal / polymer interface however, will also depend upon how

these static features will evolve with time. This will in turn depend upon the evolution of the

local chemical environment in the very confined zone of the metal / paint interface.

The delamination mechanism of paint during the atmospheric corrosion of zinc-coated steel may

be divided into two broad categories depending upon whether the surface under the paint is

anodic or cathodic [27]. The anodic mechanism involves an undermining of the paint by the

oxidation and physical removal of the zinc coating. It is a very common for galvanized steel

especially when steel is exposed in a defect area, which is sacrificially protected by the

surrounding zinc coating.

The cathodic delamination mechanism is common for the cosmetic corrosion of painted

steel, and under certain circumstances, galvanized steel [20,28,29,30]. In this mechanism,

oxygen reduction occurs under the paint giving rise to a significant increase in the pH in the

interfacial region between the paint and the metal substrate. pH values greater than 14 have been

measured under certain circumstances [31,32,33]. The exact mechanisms of disbonding during

cathodic delamination are poorly understood. One important factor is thought to be the base

catalyzed dissolution of the interfacial conversion layer and / or degradation of the polymer

[34]. The reaction of the phosphate layer with hydroxide ion may result in significant

modifications of the phosphate crystals, with complete leaching of the phosphate layer

[35,36,37,38,39,40]. If the reaction is complete, the conversion layer may be destroyed leading

to immediate disbonding of the polymer layer. However, if the phosphate layer is not destroyed,

the reaction may actually slow the progression of cathodic delamination by neutralizing the

hydroxide ions thereby maintaining a less aggressive environment under the paint.

The reactivity of the phosphate layer with an alkaline medium is therefore an important

property of a conversion layer at the metal / paint interface. It is generally accepted that the

composition of the phosphate layer has a strong influence on its chemical stability. The primary

phase formed on the steel surface is phosphophyllite which contains Fe+2. It has been shown to

be more resistive than hopeite containing only Zn+2 cations [36,41,42]. For zinc, secondary

elements such as Ni+2 and Mn+2 are used to increase the alkaline stability [43,44]. Nevertheless,

there are few published fundamental studies concerning the alkaline stability of the phosphate

layer. This may be partly explained by the fact that rate measurements for this type of process

are particularly difficult. The reaction is not electrochemical in nature so that neither current nor

potential can be used as reliable measures of the extent of reaction. Further, gravimetric

techniques are ambiguous because weight losses and weight gains occur simultaneously during

the reaction. Therefore, there is a clear interest in applying more sophisticated techniques of in

situ analysis to this problem. To this end, we have applied a number of complementary in situ

techniques to yield a more detailed reaction mechanism than has been possible in previous

studies. The techniques include ICP atomic emission spectroelectrochemistry, in situ Raman

spectroscopy, and the quartz crystal microbalance. In some recent publications, we have

demonstrated the usefulness of using time resolved dissolution analysis with ICP atomic

21

emission spectroscopy to quantify and understand the leaching of different species into solution

during the electrochemical dissolution of stainless steel [45,46].

The zinc-phosphating of zinc-coated steel leads to a layer of hopeite (Zn3(PO4)2.4H2O)

crystals, whereas on steel the crystals mainly consist of phosphophyllite, which is known to

have a higher stability against alkaline media [47,48]. Various efforts were made in the recent

decades to improve the alkaline stability of hopeite layers on zinc-coated steel [49,50] leading to

the modern trication phosphating bath which we may consider to be the state of the art. These

phosphating solutions contain nickel and manganese in addition to zinc (trication refers to the

three metal cations present: Zn,Mn and Ni) and have a greatly improved corrosion resistance

compared to pure hopeite layers. Manganese is known to substitute zinc atoms on the octahedral

and on the tetrahedral positions in the hopeite crystal lattice leading to a finer crystal size of the

layers and to a higher alkaline and thermal stability. Nickel only occupies the tetrahedral

positions in the crystal, which gives the same improvements as manganese, but to a lower extent

[51]. Nickel is more noble than zinc and will deposit as metal on the surface of the zinc coated

steel, which accelerates the overall phosphating reaction by an additional cathodic reaction and

ensures that zinc metal surface between the phosphate crystals is protected by metallic nickel

[52].

22

CHAPTER 4.

SYNTHESIS AND CHARACTERIZATION OF PHOSPHATE LAYERS

4.1 Metal Substrates

The metal substrates were supplied by Voest-Alpine and typical of the products used in the

automotive industry. The galvanized steel and the cold rolled steel had a thickness of 1.0 mm,

and the galvanized steel had a 7.5 m zinc coating electrodeposited by a sulfate electrolyte

process. The phosphate layers were synthesized directly onto these surfaces in an as received

condition.

4.2 Synthesis of the Phosphate Layers

The laboratory phosphate layer synthesis consisted of three different steps - alkaline

degreasing, activation, and phosphating. Standard commercial products were used for the

degreasing and activation steps. Degreasing was carried out in a spray chamber with an

aqueous solution of 10 g/l Ridoline 1372 (Henkel KGaA) at 50 C for 15 s. Activation was

carried out as described previously [53] by dipping the substrates in an aqueous dispersion of 5

g/l Fixodine 50 CF (Henkel KGaA) at room temperature for 10 s. The samples were then

phosphated immediately after the activation step. The phosphate-treatment was done by

spraying the metal surfaces with one of the phosphating solutions of Table I at 55 C for 20 s.

The final samples were rinsed in deionized water, dried under flowing air and stored in a

dessicator.

Two series of phosphate layers were synthesized on the electrogalvanized steel surface

as indicated in Table I. The base of all these solutions was 0.16 M H3PO4, with 9 g/l NO3- added

as an accelerator, and 0.36 g/l F- as a common additive in industrial practice. In addition, tartaric

acid was added to one trication Zn/Ni/Mn solution. Bath preparation was performed by adding

ZnO, Ni(NO3)2, Mn(NO3)2, NaNO3-, and NaF, to an initial solution of 0.16 M H3PO4 so as to

give the final composition of metal ions and nitrate ions indicated in Table 1. The free acid,

defined as the number of ml of 0.1 M NaOH required to reach the first endpoint, was adjusted to

2 by the addition of sodium carbonate. The pH of the final solution was around 3.0. Reagent

grade chemicals and distilled water were used for all solution preparations.

The Series I samples were prepared especially for the Raman experiments because the

Raman technique is very sensitive to coating weight. Other factors being constant, manganese

and nickel lower the crystal size and hence decrease the total coating weight [54]. In order to

counter this tendency, the activation products Fixodine 50CF and Fixodine 950 (trademarks by

Henkel) were used in varying concentrations to get comparable coating weights of

approximately 2 g/m2. Scanning electron micrographs were taken without Au-sputtering of the

samples using a LEO 1530 field emission electron microscope.

Table II gives the bath formulation for the synthesis of phosphate layers on cold rolled

steel samples. In this case, the free acid was adjusted to 1.0, hydroxylamine (NH2OH) was used

in place of the NO3- as an accelerator, Mn+2, Ni+2 and Zn+2 were added as M(HPO4), and the F

-

concentration was 1.5 g/l. Cu+2 was added to one of the trication ion bath solutions. Degreasing

was carried out as described above. Activation was performed by dipping the sample into a 1 g/l

Fixodine 50 CF solution for 10 s at room temperature. Phosphating was performed by spray for

30 s at 50C.

Table IIIA and IIIB also give the chemical and gravimetric analysis of the phosphate

conversion layers. This was performed by etching the phosphate coating with a 0.5% CrO3

solution for 5 min. The phosphate coating weight was determined by the mass loss during this

etching period. The elemental composition of the layers was determined by analysis of the etch

solutions with ICP atomic emission spectrometry and photometry using standard techniques.

23

Scanning electron microscopy showed typical phosphate crystals associated with zinc-coated

steel. Slight differences in morphology were observed as a function of the metal ion

composition and consistent with phosphating practice. X-ray diffraction revealed only hopeite

for all layers on zinc-coated steel as expected for this type of sample. It is known that Mn+2 and

Ni+2 substitute for Zn+2 to form pseudophosphophyllite", Zn(3-x-y-z)FexMnyNiz(PO4)2.4H2O

leading to partial amorphization of the zinc phosphate layer. For cold rolled steel, hopeite and

phosphophyllite were detected.

4.3 Morphology Studies

By changing the product for activation it was possible to obtain similar coating weights

for the Series I sample, even when Mn and Ni are present, elements which would otherwise

cause a decrease in coating weight. Mn was found in a concentration of 5% in the crystals, Ni of

1.5%. Fig. II.1 shows scanning electron micrographs of the four phosphate layers. A clear

influence of Mn and Ni in changing crystal morphology was observed, indicated by the cubic

habitus of the crystals in the Zn/Mn-phosphate layer and the plate shaped habitus in the Zn/Ni-

phosphate layer. The habitus on the trication layer (Fig. II.1D) can be interpreted as a mixture of

the different influences of Mn and Ni on the morphology as seen on Fig. II.1B and Fig. II.1C.

Average crystal size was around 2 m on all samples.

Fig. II.2 shows the SEM pictures of the samples after 225 s of attack by the electrolyte

(0.1M NaOH) in the flow cell. Traces of attack can be seen on all samples, but no conclusion

about the remaining phosphate concentration in the layer can be made by just investigating the

morphology. This is especially clear for the bication Zn/Ni sample (Fig. II.2C), which retained

the very nearly the original crystal morphology after the attack, but showed a clear decrease in

the P:Zn ratio (measured by EDX) compared to the unattacked surface. The percentual decrease

of the P:Zn was 92% on the monocation Zn-layer, 49% on bication Zn/Mn-layer, 89% on the

bication Zn/Ni-layer and 55% on the trication Zn/Mn/Ni-layer.

4.4 Corrosion Testing

Scab Corrosion Test Samples were painted with a lead-free ED-paint and tested according to

VDA 621-145 in a scab corrosion test. Fig. II.3 shows delamination of Clemen cut and Fig. II.4

shows the rating of the stone chipped area (VDA 621-427) after 10 cycles. The influence of

nickel in increasing the corrosion resistance is clearly seen. Although the other experiments

show that Mmanganese raises alkaline stability, its influence in this corrosion test is rather

small. The covering of the free zinc pores in the phosphate layer by Nickel cementation

therefore seems to be more important in the performance of a phosphate layer in corrosion

protection.

Salt Spray Test The phosphated samples (trication phosphating) were ED-painted and

artificially damaged by a Van Laar cut (5cm). The samples were exposed in the salt spray

chamber for 120h. The spray solution consists of 5% NaCl at 35C according to DIN 50021SS.

After the exposure and rinsing of the residual salt the delaminated paint was removed with a

sharp knife. The delamination was not uniform, but appeared in semicircles (blisters) with about

5 to 10 mm distance from each other. (Fig. II.5) This is typical for zinc-coated steel substrates.

The scratch itself was covered with white corrosion products.

Two different regions could be seen on the scratch in various distances (5 to 20mm)

characterized as a clear white area, and a brown spot. Fig. II.5 shows a part of the scratch,

where both regions can be seen. The according Raman spectra of these spots can be seen in Fig.

II.5A and Fig. II.5B. The product on the brown coloured part shows peaks at 330, 437, 570 and

a broad band at 1100cm-1. This could be identified as ZnO [55]. The spectrum of the white part

shows peaks at 386, 738, 1070, 1370 and 1550cm-1, typical bands for zinc hydroxycarbonate.

24

The EDX-spectra (Fig. II.5C and Fig. II.5D) of both spots show no clear distinction. The

sensitivity for carbon in EDX seems to be too low to distinguish ZnO and zinc

hydroxycarbonate. Chlorine is found in a slightly higher amount on the brown spots.

Fig. II.6 shows an analysis of the blister region after removing the paint. No typical

phosphate crystals are visible, and the EDX-spectrum (Fig. II.6B) shows no phosphorus signal,

but a distinct chlorine signal. The Raman spectrum (Fig. II.6C) shows peaks at 210, 250, 397,

731, 910, 3460 and at 3488cm-1 and could be clearly identified as the spectrum of

4Zn(OH)2.ZnCl2. This was seen all over the blister. No trace of hopeite or phosphorus was

detected.

After removal from the sample, the paint was examined on the side faced to the

substrate and phosphate layer respectively. Fig. II.7 shows an overview picture in the SEM, Fig.

7A and 7B show two regions which could be visually distinguished. First we find again crystals

which were identified as 4Zn(OH)2.ZnCl2 in the EDX (Fig II.7C). This crystals could be

removed from the sample during the cut with the knife and apparently sticked to the paint.

Another part showed an EDX-spectrum (Fig. II.7D) with a clear phosphorus signal, but it does

not show the typical crystal structure of hopeite crystals. Both spectra also show elements from

the components of the paint. No distinct signal was detected in Raman spectroscopy.

The experiment shows that the phosphate layer is completely removed from the sample even

in neutral media. The essential delamination happens between the phosphate layer and the zinc

substrate, which is confirmed by the phosphate remainders found on the backside of the

removed paint. These results are consistent with the anodic delamination mechanism.

25

CHAPTER 5.

LEACHING STUDIES USING INDUCTIVELY COUPLED PLASMA

SPECTROELECTROCHEMISTRY

5.1 Overview of the Experiment

The ICP experiment has been described in great detail in Chapters 1 and 2. The

experiments in this work were performed with a flow rate of 2.5 ml/min. The elemental

concentration of P, Zn, Mn, and Ni were determined from the emission intensity at 178.225 nm

(P), 213.856 nm (Zn), 257.610 nm (Mn), 216.555 nm (Ni). In addition, Fe, Al, and Ti were

followed but not detected in these experiments unless otherwise noted. The open circuit

potential of the phosphated sample was measured relative to a saturated calomel electrode

placed in the external compartment of the cell as shown in Fig. II.8. This compartment was

separated from the main cell by a porous membrane that allows passage of ionic currents, but no

bulk mixing of the two electrolytes.

Two types of leaching experiments were performed in this work. First, the reactivity of

the phosphate layer with a 0.1 M NaOH solution was measured. The exposure to the alkaline

solution continued until phosphate and zinc leaching had decreased to undetectable levels.

Following the alkaline exposure, the surface was exposed to 0.01 M HCl so as to dissolve and

analysis the composition of residual films following the reaction of the phosphate layer with

hydroxide.

Initially the sample was placed on the empty cell, with the 0.1 M NaOH electrolyte flow

bypassing the cell. This permits a precise measurement of background emission signal for the

electrolyte. To begin the exposure, the two valves indicated in Fig. II.8 were adjusted so that the

electrolyte flow passed through the cell. Next, the alkaline solution was substituted for an acid

solution 0.01 M HCl solution containing 1 ppm Y as an internal standard. The pump was

allowed to aspirate air for several seconds during transfer of the electrolyte so as to avoid

mixing in the tube itself. At the end of the experiment, the valves were returned to the bypass

position so as to measure the background signal and the emission intensity of one or more

standard solutions. Calibration of the emission intensity during the exposure was performed by

comparison with the emission intensity of a standard solution, usually 5 ppm of the element in

question.

5.2 General Features of the Leaching Transients

Fig. II.9 shows typical leaching transients obtained when the low Mn trication

phosphate layer (Series II) is exposed to 0.1 M NaOH at ambient temperature in aerated (upper)

and desaerated (lower) solution. The rate of PO4-3, Zn+2 and Mn+2 leaching are shown as a

function of time during the exposure to the aggressive electrolyte. (Note that the atomic

emission technique gives only elemental composition, we assume the common oxidation states

for these elements.) The open circuit potential, E, vs. a saturated calomel electrode in the

reference compartment is also shown. Contact with the electrolyte begins at the vertical dashed

line (here positioned at t=0). Prior to t=0 the surface was dry in the empty cell and the pure

electrolyte was aspirated into the plasma by the bypass tube shown in Fig. II.8.

These results demonstrate that PO4-3 and Zn+2 dissolution occur simultaneously during

the early stages of the alkaline attack. After approximately 300 s, the PO4-3 dissolution rate

drops to undetectable levels while the Zn+2 dissolution rate drops down to a steady state value.

The steady state zinc dissolution rate is significantly lower under nitrogen (9A) as compared to

air (9B) and therefore is probably due to the corrosion of the Zn substrate.

27

The open circuit potential was measured simultaneously with the leaching rate

transients and is shown on the secondary axis to the right. For all experiments in this work, the

initial open circuit potential was around -0.85 V vs. SCE, and dropped off progressively during

the exposure to the alkaline solution. During the early stages of this experiment, the potential

drops off slowly until approximately -1.1 to -1.2 V is obtained. At that point, there is an abrupt

drop in the potential to about -1.39 V, consistent with the Zn/Zn(OH)4-2 couple.

The rate of phosphate leaching is observed to be about 30% higher in the presence of

air. This is seen both as the absolute value of the rate, and as the time at which the value has

dropped to half the maximum value. By contrast, the OC potential plateau is more than doubled

in the presence of air, as compared with the situation under nitrogen. This demonstrates that the

time of the abrupt change in the potential cannot be used as a reliable measurement of the

phosphate layer stability.

5.3 Kinetics of Alkaline Leaching of Phosphate Coatings on Zinc

Figure II.10A shows the reaction occurring during the leaching of a monocation

phosphate layer. The rates of Zn+2 and PO4-3 dissolution rise rapidly to steady state values and

remain relatively constant for about 100 seconds. During this steady state period, P dissolution

is slightly faster than Zn+2 dissolution. The rate of Zn+2 and PO4-3 leaching decrease

progressively, but PO4-3 leaching decreases more rapidly than Zn+2 resulting in a change in the

dissolution stoichiometry in the later stages of the reaction. Integration of the respective peak

areas gives a total Zn / P atomic ratio of 1.36, as compared with 1.58 by chemical analysis

(Table III) of the layer. The theoretical value is 1.5. The low value for the Zn / P ratio is

probably due to the presence of zinc hydroxide on the surface after complete leaching of the

phosphate ion. Both the P and Zn signals decrease to the background level at the end of the

reaction.

A dissolution transient for a bication Zn / Mn phosphate layer is shown in Fig. II.10B.

For this case, the stoichiometry between Zn+2 and PO4-3 remains constant (within experimental

error) throughout the profile. The Mn dissolution rate has been multiplied by 1000 so as to

appear on the same scale as the Zn+2 and PO4-3 dissolution rates. Some Mn dissolution is

observed, but as shown in Table III, only a small fraction ( 0.2%) of the total Mn appears to be soluble in the alkaline electrolyte. The majority of the Mn is observed during post dissolution in

HCl solution, as described in the next section. The rate of phosphate and zinc dissolution of the

Zn/Mn bication layer has been reduced by a factor of almost 3 with respect to the monocation

situation. The final potential is again around -1.38 V consistent with the zincate equilibrium in

this electrolyte.

Figure II.10C shows the dissolution profile obtained for the Ni/Zn bication sample. No

Ni dissolution is observed during the alkaline portion of this experiment. The dissolution rate of

P is not significantly different from that observed for the monocation Zn layer. This may reflect

the fact that Ni+2 is incorporated into the phosphate crystals to a smaller extent than Mn+2.

The dissolution curve for the trication layer is shown in Fig. II.10D. The behavior is

very similar to the bication Zn/Mn and demonstrates that the chemical stability of the phosphate

crystals is primarily determined by the Mn+2 ion. In the series II samples, several different

trication formulations were used, with consistent results. One interesting difference however is

in the open circuit potential. Although the two layers show similar leaching behavior for all

elements, the open circuit potential drops to lower values immediately upon contact with the

alkaline solution, and reaches the zincate potential well before the rate of phosphate dissolution

begins to decrease. For the Ni+2 containing phosphate layers, it was observed that the length of

time for the potential jump was very sensitive to oxygen in the electrolyte. This result is

illustrated in the data of Fig. II.9. When the solution was aerated, the potential jump was

observed only at very long times (880 s in Fig. II.9B) and sometimes not at all. The other

systems were much less sensitive to air.

28

5.4 Composition of the Residual Layers

After the measurement of the leaching profile, residual P, Ni, and Mn were measured by

exposing the surface to 0.01 M HCl. The resulting series of dissolution profiles and peak

integrals are given in Fig. II.11 and Table III respectively. Residual Zn could not be measured

because of the dissolution of the metallic zinc coating by the acid solution. No significant

phosphate dissolution was detected in any of the profiles, which demonstrates that the hydroxide

/ phosphate exchange occurs with nearly 100% efficiency. The Ni and Mn quantities determined

by chemical analysis (Table I) were about 30% higher on average than the values given in Table

III. This discrepancy is probably due to systematic error due to the estimation of the effective

surface area in these experiments. A large dispersion is indicated probably due to the

significantly smaller surface area in the ICP experiments.

Ytrium was added to the HCl electrolyte as an internal standard. The Y concentration

was 2 ppm and the Y signal is shown in arbitrary units in Fig. II.11. This is used to identify the

time at which detection of the HCl electrolyte begins. This is important because, prior to this

experiment, the cell is filled with NaOH electrolyte. The open circuit potential begins to rise at

the same time as the Y signal increases, however, the dissolution of the zinc substrate and

surface layers is delayed for approximately 30 seconds. This period is probably due to the time

required to neutralize the 0.1 M NaOH solution with 0.01 M HCl.

In Fig. II.11C, some small peaks are observed before the onset of the Y signal. These

artifacts were common in this type of experiment and are explained by the procedure for

changing electrolytes. At the end of the NaOH experiment the capillary was removed from the

alkaline electrolyte, maintained in the air for approximately 5 seconds, and then placed in the

0.01 M HCl solution. In this way, the two electrolytes are separated in the capillary by an air

bubble. The bubble will prevent mixing of the two electrolytes, but will tend to push through

any stagnant electrolyte on the capillary walls. This cleaning effect in the tube leads to peaks

such as seen here.

5.5 Alkaline Leaching of Phosphate Layers on Cold Rolled Steel

To determine whether or not residual zinc remains on the surface, we have performed a

similar analysis of phosphate layers on cold rolled steel. The alkaline dissolution of the

phosphate layer on cold rolled steel followed by the acid dissolution of the residual layer are

shown in Fig. II.12. The alkaline dissolution profiles of these layers closely resemble those

obtained for the trication galvanized steel in Fig. II.10 and II.11. This is true even for the

phosphate layer formed from a monocation bath suggesting that Fe+2 serves a similar role as

Mn+2 or Ni+2 when incorporated into the phosphate crystal lattice.

The 0.01 M HCl experiments demonstrate that a residual film is obtained on the surface

following the alkaline treatments. Zn is detected in the acid experiment at a level of

approximately 10 to 30 % of the total zinc. The sum of the integrated transients gives a +/-

charge ratio < 1 because of undetected Fe+2 dissolution.

Finally, the rate of steel corrosion appears to be increased by more than a factor of 10

for the trication layers as compared to the monocation layer. The steady state open circuit

potential was -0.52 V for the trication layers as compared with -0.64 V for the monocation

layers. One possible explanation for these differences would be the presence of metallic nickel

(and copper for III.3) on the metal substrate increasing the likelihood of galvanic corrosion in

the acid medium.

29

5.6 Kinetics of Alkaline Leaching on Hot Dip Galvanized Steel

Finally, it was of interest to demonstrate that the alkaline stability observed was a

function of the phosphate crystal composition and not the substrate material. To this end, the

trication and monocation phosphating treatment was applied to a galvanized steel substrate

described in the experimental section. This coating is quite different from the electrogalvanized

in that it could contain Al in the range 0.1 0.3%. Nevertheless, a similar behavior was

observed as for the electrogalvanized steel surface, the dissolution rate being only slightly

higher. It should be noted that the galvanized steel surface had not undergone the usual skin

pass, which is a process treatment in which a certain roughness is "stamped" onto the coated

steel product. It is generally accepted that this process increases the quality of phosphate

coatings by breaking up oxide islands.

5.7 Analysis of Results - Rate Variation with Composition

Figure II.13 summarizes the rate data for phosphate leaching as a function of the metal

ion composition for the Series and II samples and the cold rolled steel samples. The rate is

observed to decrease in proportion to the fraction of Mn and Ni incorporated into the layer. For

most samples, the quantity of Ni is well below the quantity of Mn, so the inclusion of Ni in the

sum is not obvious. However the Series II bication Zn/Ni sample contained significantly more

Ni than the series I, and the fact that the rate was reduced proportionally indicates that Ni and

Mn both contribute to the alkaline resistance of the layer.

The phosphate layers on the cold rolled steel sample are also shown on this curve. These

points are interesting in that they represent a higher secondary metal ion contentl. The good

agreement between the monocation steel sample and the other phosphate layers indicates that

presence of Fe+2 in the phosphate layer acts in a similar manner to Mn+2 or Ni+2 in reducing the

rate of alkaline attack. The Fe+2 comes from the acid attack of the steel substrate and thus it is

difficult to produce a true monocation phosphate layer on steel. The trication phosphate layers

on steel allow us to go to much higher %M values than was possible with the zinc surface.

5.8 Mechanistic Considerations

The results presented here demonstrate that time resolved dissolution using the ICP

method is a powerful tool for measuring the stability of conversion layers with an aggressive

electrolyte. With this method we have demonstrated that the elemental composition of the

phosphate layer plays an important role in determining the alkaline stability of the layer. In

particular, it was found that the addition of Mn+2 and/or Ni+2 to a zinc phosphate layer results in

a significant decrease in the leaching rate of the phosphate layer. The decrease is proportional to

the percentage of the substitution cations incorporated into the layer.

The fact that the Mn and Ni remain on the surface after the alkaline leaching experiment

strongly suggests a chemical origin for the increased stability of the conversion layer.

Morphological changes are apparent between the different conversion coatings, and we cannot

completely rule out the idea that the morphology of the crystal is a determining factor, however

no clear correlation with either these differences, or with the coating weight was observed.

Therefore, within the framework of the chemical hypothesis, it is of interest to discuss possible

reaction mechanisms for the alkaline attack of the phosphate layer.

In all cases an initial dissolution stoichiometry of Zn to P close to 1:1 was measured,

despite the fact that the ratio of Zn / P ratio should be 1.5 according to the stoichiometry of the

phosphate layer. This result suggests that the first step of phosphate dissolution involves the

incorporation of a single hydroxide ion into the phosphate layer, with the resulting loss of a

single Zn+2 and PO4-3, as shown in reaction 1 below. The result would be a residual layer of

30

Zn(OH)2. To explain the kinetic results, it may be assumed that this layer dissolves more slowly.

In addition, the electrochemical potential transients suggest that zinc hydroxide can be formed

by the simultaneous oxidation of the zinc coupled with oxygen reduction or hydrogen

formation. Therefore, for the monocation phosphate layer, it is reasonable to write the leaching

reactions as follows:

Zn3(PO4)2 + OH- Zn+2 + PO4

-3 + Zn(OH)2 (solid)

1

Zn (solid) + OH- Zn(OH)2 (solid) + e-

2

Zn(OH)2 (solid) + OH- Zn(OH)4

-2

3

Within the framework of this model, it is reasonable to assume that the substitution of Mn+2 into

the phosphate layer, gives rise to a mixed hydroxide film. The resulting Mn hydroxide is less

soluble than Zn hydroxide, and the film may protect the underlying crystal, slowing the reaction

considerably. This model will be further elaborated in the second part of this paper dealing with

in situ Raman spectroscopy.

The basic phosphate crystalline structure is that of hopeite, Zn3(PO4)2.2H2O. When

present in the phosphating solution, Ni+2 and Mn+2 will replace Zn+2 in the hopeite structure.

Ni+2 and Mn+2 each reduce the leaching rate by a factor proportional to their incorporation in the

crystals. However, Ni+2 replaces Zn+2 in the hopeite crystal structure to a much lesser extent

than Mn+2, despite their nearly equivalent concentration in the original bath. Therefore, in

practical terms, the addition of Ni+2 to conventional phosphate baths has only a minor influence

on the alkaline stability of the phosphate layer. According to previous workers, Ni+2 will only

replace Zn+2 in the octahedral coordination centers of hopeite, whereas Mn+2 replaces both

tetrahedral and octahedral sites.[56] Also, the Ni+2 concentration in the phosphate crystals is

probably much less than that measured in the analysis of Table II. It is well known that during

phosphating, Ni+2 deposits on the Zn surface as Ni metal, by cementation.[57,58]

We have also observed a slight decrease in the phosphate dissolution rate when oxygen

is removed from the electrolyte by bubbling with nitrogen. An example is shown in Fig. II.9A,

but this effect was observed systematically for the other samples. This would suggest that the

corrosion of the underlying metal substrate might contribute to phosphate dissolution, perhaps

through a mechanism of anodic undermining.

The variation of the open circuit potential is also an important clue to the phenomena

that occur. The final potential of -1.38 to -1.42 V corresponds to approximately to the

equilibrium value obtained for the Zn/Zn(OH)4-2 couple of -1.457 V vs. SCE [59]. It may be

assumed that the abrupt potential drop occurs when the zinc metal is first exposed to the

electrolyte and the predominate reaction becomes Zn + 4OH- Zn(OH)4-2 + 2e-. This seems

to be the case for the experiment of Fig. II.9A where the abrupt potential drop under aerated

conditions coincides with an increase in the rate of zinc dissolution.

For the monocation Zn layer and the bication Zn/Ni layer, the potential drop occurs

before complete dissolution of the phosphate layer. This would suggest the residual film

forming under these circumstances has little protective effect on the metal substrate. For the

bication Zn- Mn layer, the potential drop occurs at times considerable longer than that required

for the dissolution of the phosphate film. This suggests that the final steps of the phosphate film

dissolution involve a slowly dissolving, protective, residual layer, and indeed, the presence of a

Mn rich hydroxide film is clearly observed by acid dissolution.

31

When Ni is present in the phosphate layer, the potential in aerated solutions seems to

stabilize at higher values and the potential drop is observed only after the passage of

considerable time, regardless of the rate of phosphate leaching. It is reasonable to assume that

the higher potential values are determined by the presence of metallic nickel, which will form a

galvanic couple with zinc. In a similar way, there is a large increase in potential during the

phosphating of zinc with Ni containing baths. The Ni seals the phosphate layer, by depositing

on the zinc surface exposed at the bottom of the phosphate crystal pores.

From the results of this study, we are now in a better position to appreciate the value of

Mn and Ni in the phosphate layer as far as increasing the chemical resistance of the conversion

coating at the metal / polymer interface. Mn enhances the alkaline resistance of the phosphate

layer by forming a less soluble, protective, Mn hydroxide film when the phosphate layer is

exposed to the hydroxide ion. Ni by contrast, has only an indirect influence on the phosphate

film. It deposits on the zinc surface surrounding the crystals, thereby protecting the crystals

from anodic undermining. Which of these two aspects will be important for determining the

corrosion resistance of a specific painted steel product will depend upon the specifics of the

corrosive environment. Other things being equal, the alkaline resistance afforded by Mn might

be most helpful for a mechanism of cathodic delamination such as is commonly observed on

painted steel, while the resistance offered by Ni might be more important in the situation of

anodic delamination such as is observed for painted zinc coated steel in a salt spray test.

It is interesting to compare the function of these two species with the role of the

chromate post treatments for sealing, which were commonly used after phosphating in the

automotive industry until they were recently abandoned because of environmental problems

associated with the use of Cr(VI). It has been observed that the chromate reacted with the

phosphate crystals to form a thin, alkaline resistant layer on the outside of the crystals, while

simultaneously passivating the exposed zinc (or steel) surface around the crystals by forming a

Cr(III) oxide film.[60,61]

5.9 Conclusions

In this chapter, we have demonstrated that the ICP method may be used to measure the

dissolution rate of conversion coatings. The major advantage of this technique is the capacity to

measure each element independently. We have measured the rate of phosphate, zinc,

manganese, and nickel leaching during exposure to an alkaline electrolyte. The composition of

the residual film has been analyzed by a subsequent dissolution in an acidic solution. These

results demonstrate the importance of Mn+2 in increasing the alkaline resistance of the

phosphate layer. By contrast, the effect of Ni+2 on the alkaline resistance is significantly less,

reflecting its lower incorporation into the phosphate crystal lattice. It is proposed that the Ni+2 in

the phosphate bath, increases alkaline resistance of the final layer by depositing as Ni metal on

the zinc surface exposed in the porosity of the phosphate layer, and thereby inhibiting anodic

undermining. The two mechanisms taken together present an interesting synergy, and are

analogous to the mechanisms proposed for chromate post treatment.

32

CHAPTER 6.

LEACHING STUDIES USING IN SITU RAMAN SPECTROSCOPY

6 .1 Introduction

In this study we used Raman spectroscopy to follow the attack of alkaline media on the

phosphate layers on zinc-coated steel with the object of estimating reaction rates for different

layers and to identify the final products of the attack. The on-line ICP technique of the previous

chapter measures the elements released from the surface during the reaction of the surface layer

with an aggressive electrolyte. It was therefore of interest to investigate this reaction with a

surface specific technique which would permit a complementary in situ analysis of the chemical

changes occurring on the surface. Raman spectroscopy is an ideal candidate for this

measurement. It has been frequently used to characterize phosphate conversion coatings. The

symmetrical stretching vibration of the tetrahedral phosphate anion gives rise to a strong

emission intensity at 996 cm-1.[62] Raman spectroscopy has been used to characterize

phosphate layers by Sato et al.[63] Sommer and Leidheiser[64] used Raman spectroscopy to

examine the dissolution behavior of precipitated zinc phosphate crystals after long time

exposure in various alkali metal hydroxides.

6.2 Experimental

A description of the instrument and the experimental setup was already given in Part I.

A simple flow cell was constructed from Teflon as illustrated in Part I. The upper part of the

cell accomplished the electrolyte flow and is provided with the cell window. This part was

mounted with screws onto the lower part, where the sample is fixed. Samples up to 5 cm x 5 cm

in size can be used with this cell. However, turbulent flow conditions at the inlet into the sample

compartment leads to non uniform attack of the surface which may influence the open circuit

potential by quicker reaction rates and earlier exposing the of the base metal substrate. For the

work described here small samples of 2.5 x 2.5 cm size were mounted into the flow cell with a

special adhesive tape leaving free a circular area with 2 cm diameter on the sample exposed to

the electrolyte. Pickling experiments with dilute hydrochloric acid on zinc coated steel revealed

a uniform dissolution on the exposed sample surface that is left blank in the adhesive tape (area:

3.14 0.08 cm)

The diameter of the circular sample compartment was 4 cm giving a volume of

approximately 1.8 ml with the sample mounted in the cell. The thickness of the electrolyte layer

above the sample surface was 1 mm. Under these conditions the velocity of the solution was 20

cm/min in the center of the cell at a turnover time of around 11s in the whole compartment. The

center of the sample compartment is covered with a 1mm thick window made of flat glass

giving a transparent area of 20mm in diameter. A capillary introduced into the cell via a small

hole in the side of the upper part allowed the measurement of the open circuit potential versus a

Ag/AgCl-electrode (Schott) during the reaction.

The flow cell was mounted on the x,y motorized table of the Raman microscope, focusing

of the laser light was done by moving the table mechanically in the z-direction. The electrolyte

(0.1M NaOH, deaerated with N2) was pumped through the cell with a peristaltic pump at a rate

of 10 ml/min.

For in-situ experiments a suitable sample spot was adjusted in the dry state of the cell to be

sure the chosen spot exhibited no fluorescence, which would render the detection of the Raman

spectra impossible. With the electrolyte in the cell, a new adjustment of the laser focus had to be

done due to the high refraction of the laser light at the phase boundaries air-glass and glass-

electrolyte [65]. Subsequent spectra were recorded by the software coming with the

33

spectrometer at a collection time of 45 s, which was found to give reasonable intensities of the

Raman signal for further quantification.

6.3 General Features of the Raman Spectra

Fig. II.14 shows the Raman spectra of the zinc phosphate layer mounted into the flow

cell in dry conditions (14A) and under water flow (14B) collected with the same parameters (45

s, 50x objective). The intensity under water flow was lowered to less than a half compared to

the dry sample rendering some bands hardly detectable anymore. The spectrum of the sample

not covered with water (14A) clearly showed all the features of zinc phosphate (hopeite) in this

frequency region.[62] The main band at 996 cm-1 indicates the symmetrical stretching vibration

of the phosphate (PO4-3) anion. It is escorted by three smaller (~one quarter of intensity) peaks

at 940, 1056 and 1149 cm1, which originate in the P-O stretching vibration of the phosphate

anion. For a free PO4-3 ion (e.g. in liquid phase) these three vibrational features are degenerated

and give rise to only one Raman band as it is observed in phosphoric acid. However, distorted

symmetry in the phosphate crystal lattice unties and splits the degener