Quick viewing(Text Mode)

The Wear and Corrosive-Wear Response of Tungsten Carbide-Cobalt Hardmetals Under Woodcutting and Three Body Abrasion Conditions

The Wear and Corrosive-Wear Response of Tungsten Carbide-Cobalt Hardmetals Under Woodcutting and Three Body Abrasion Conditions

The wear and corrosive-wear response of - hardmetals under woodcutting and three body abrasion conditions

Submitted to

the Faculty of Engineering of the University of Erlangen-Nürnberg

in fulfillment of the requirements for the degree of

DOKTOR-INGENIEUR

by

Natasha Sacks

Erlangen 2002

Das Verschleiß- und Korrosions-Verschleißverhalten von Wolframkarbid-Kobalt-Hartmetallen bei der Holzzerspanung und unter Drei-Körper-Abrasivverschleiß

Der Technischen Fakultät der Universität Erlangen-Nürnberg zur Erlangung des Grades

DOKTOR-INGENIEUR

vorgelegt von

Natasha Sacks

Erlangen 2002

Als Dissertation genehmigt von

der Technischen Fakultät der

Universität Erlangen-Nürnberg

Tag der Einreichung: 19. Dezember 2002

Tag der Promotion: 31. März 2003

Dekan: Prof. Dr. Winnacker

Berichterstatter: Prof. Dr. H.-G. Sockel Prof. Dr. P. Schmuki

Table of Contents i

Table of Contents

1 Introduction (English) ...... 1 2 Introduction (German)...... 3 3 Objectives ...... 6 4 Literature Review...... 7 4.1 Woodcutting tool wear...... 7 4.1.1 Wood...... 7 4.1.2 Tool wear ...... 8 4.1.3 Methods used to assess the blunting of the cutting edges...... 8 4.1.4 Influence of force, temperature and speed...... 10 4.1.5 Tool wear relating to the cutting of natural wood...... 13 4.1.6 Tool wear relating to the cutting of synthetic wood products...... 15 4.2. Woodcutting tools...... 17 4.3 Tungsten carbide hardmetals ...... 18 4.3.1 Hardmetal manufacture ...... 18 4.3.2 Hardmetal constituents ...... 20 4.3.3 Porosity ...... 26 4.3.4 ...... 26 4.4 ...... 28 4.4.1 and Stern’s rule...... 28 4.4.2 Corrosion of cobalt...... 29 4.4.3 Corrosion of WC ...... 30 4.4.4 Corrosion of WC-Co ...... 31 4.5 Wear...... 32 4.5.1 wear ...... 33 4.5.2 Lubrication ...... 36 4.5.3 Abrasive wear of WC-Co alloys...... 37 4.5.4 Corrosive-wear...... 41 4.5.5 Corrosive-wear of WC-Co alloys ...... 43 5 Experimental Methods ...... 45 5.1 Material Characterization...... 45 5.1.1 Investigated hardmetals...... 45 5.1.2 Hardmetal quality evaluation ...... 46 5.1.3 ...... 47 5.1.4. Magnetic saturation ...... 48 5.1.5 Coercivity ...... 48 5.1.6 Hardness and fracture toughness ...... 48 5.1.7 Surface roughness...... 49 5.1.8 Quantitative microstructure analysis...... 50 5.2 Woodcutting tests ...... 50 5.3 Corrosion tests...... 53 5.4 Three body abrasive wear tests...... 57

Table of Contents ii

6 Results...... 63 6.1 Material characterization ...... 63 6.2 Woodcutting tests ...... 64 6.2.1 Cutting edge widths and wear rates in the incipient wear stage ...... 64 6.2.2 Cutting edge widths and wear rates in the steady wear stage ...... 70 6.2.3 Microstructural examination...... 73 6.2.4 Summary of the main findings ...... 82 6.3 Corrosion tests...... 83 6.3.1 Microstructural examination...... 92 6.3.2 Summary of the main findings ...... 96 6.4 Abrasive wear and corrosive abrasive-wear tests ...... 96 6.4.1 Comparison of the wear rates under dry, water and tannic conditions...... 97 6.4.2 Comparison of the wear rates under varied tannic acid concentration ...... 99 6.4.3 Comparison of the wear rates under varied applied normal force...... 101 6.4.4 Comparison of the wear rates under varying abrading wheel speed ...... 106 6.4.5 Microstructural examination...... 109 6.4.6 Summary of the main findings ...... 116 7 Discussion ...... 118 7.1 Evaluation of the wear response during the woodcutting tests ...... 118 7.1.1 Influence of the wood-tool set-up on the wear process...... 118 7.1.2 Influence of the wood type on the wear process ...... 119 7.1.3 Comparison between the wear response of the two hardmetal woodcutting knives...... 121 7.1.4 Comparison of the initial and the steady state wear stages ...... 123 7.2 Corrosion response of the hardmetals ...... 123 7.3 Evaluation of the wear response during the three body abrasion tests ...... 125 7.3.1 Evaluation of the general wear process for the investigated system ...... 125 7.3.2 Influence of fluids on the wear process ...... 128 7.3.3 Influence of corrosion on abrasive wear...... 129 7.3.4 Influence of applied normal force and abrading wheel speed on abrasive wear ...... 132 7.3.5 Comparison between the wear response of the investigated hardmetals...... 133 8 Conclusions (English)...... 136 9 Conclusions (German) ...... 138 10 References ...... 140

Inhaltsverzeichnis iii

Inhaltsverzeichnis

1 Einleitung (Englisch)...... 1 2 Einleitung (Deutsch)...... 3 3 Zielsetzung...... 6 4 Grundlagen und Kenntnisstand ...... 7 4.1 Holzschneidenverschleiß ...... 7 4.1.1 Holz...... 7 4.1.2 Schneidenverschleiß ...... 8 4.1.3 Verwendete Methoden das Abstumpfen der Schneiden festzusetzen...... 8 4.1.4 Einfluss der Kraft, der Temperatur und der Geschwindigkeit...... 10 4.1.5 Werkzeugverschleiß in Bezug auf Schnitt von natürlichem Holz ...... 13 4.1.6 Werkzeugverschleiß in Bezug auf Schnitt von künstlichen Holzprodukten...... 15 4.2. Holzwerkzeugmaterialien ...... 17 4.3 Wolframkarbid Hartmetalle...... 18 4.3.1 Hartmetallherstellung...... 18 4.3.2 Hartmetallbestandteile ...... 20 4.3.3 Porosität...... 26 4.3.4 Härte ...... 26 4.4 Korrosion ...... 28 4.4.1 Galvanische Korrosion und Stern’sche Regel...... 28 4.4.2 Korrosionsverhalten von Kobalt...... 29 4.4.3 Korrosionsverhalten von WC...... 30 4.4.4 Korrosionsverhalten von WC-Co ...... 31 4.5 Verschleiß ...... 32 4.5.1 Abrasivverschleiß ...... 33 4.5.2 Schmierung...... 36 4.5.3 Abrasivverschleißverhalten von WC-Co Hartmetalle ...... 37 4.5.4 Korrosiver-Verschleiß ...... 41 4.5.5 Korrosiver-Verschleiß von WC-Co Hartmetalle ...... 43 5 Experimentelle Methoden ...... 45 5.1 Charakterisierung der untersuchten Werkstoffe...... 45 5.1.1 Untersuchte Hartmetalle ...... 45 5.1.2 Qualitätsbewertung...... 46 5.1.3 Dichte...... 47 5.1.4.Magnetische Sättigung ...... 48 5.1.5 Koerzitivität ...... 48 5.1.6 Härte und Zähigkeit ...... 48 5.1.7 Oberflächenrauheit ...... 49 5.1.8 Gefügeanalyse...... 50 5.2 Experimente zum Schneiden von Holz...... 50 5.3 Korrosionsexperimente ...... 53 5.4 Experimente zum Drei-Körper-Verschleiß ...... 57

Inhaltsverzeichnis iv

6 Ergebnisse ...... 63 6.1 Charakterisierung der untersuchten Werkstoffe...... 63 6.2 Experimente zum Schneiden von Holz...... 64 6.2.1 Schneidenabsatz und Verschleißrate im anfänglichen Verschleißstadium...... 64 6.2.2 Schneidenabsatz und Verschleißrate im stationären Verschleißstadium ...... 70 6.2.3 Mikrostrukturelle Untersuchung...... 73 6.2.4 Zusammfassung der wichtigen Punkte...... 82 6.3 Korrosionsexperimente ...... 83 6.3.1 Mikrostrukturelle Untersuchung...... 92 6.3.2 Zusammfassung der wichtigen Punkte...... 96 6.4 Abrasive wear and corrosive abrasive-wear tests ...... 96 6.4.1 Vergleich des Verschleißverhaltens unter trockenen und normalen Bedingungen...... 97 6.4.2 Vergleich des Verschleißverhaltens unter variiert Tanninsäure Konzentrationen...... 99 6.4.3 Vergleich des Verschleißverhaltens unter variiert aufgebrachter Normalkraft ...... 101 6.4.4 Vergleich des Verschleißverhaltens unter variiert Radgeschwindigkeit...... 106 6.4.5 Mikrostrukturelle Untersuchung...... 109 6.4.6 Zusammfassung der wichtigen Punkte...... 116 7 Diskussion ...... 118 7.1 Auswertung der Verschleißverhaltens der Holzzerspanungs Versuche ...... 118 7.1.1 Einfluss des Versuchsaufbaus auf den Verschleißprozess...... 118 7.1.2 Einfluss der Holzart auf den Verschleißprozess...... 119 7.1.3 Vergleich des Verschleißverhaltens der zwei ultrfeinkörnigen Hartmetallmesser...... 121 7.1.4 Vergleich der anfänglichen und stationären Verschleißstadien ...... 123 7.2 Korrosionsverhalten ...... 123 7.3 Auswertung der Verschleißverhalten der Drei-Körper-Verschleißversuche ...... 125 7.3.1 Auswertung der Allgemeine Verschleißverhalten...... 125 7.3.2 Einfluss von Flussigkeiten auf den Verschleißprozess ...... 128 7.3.3 Einfluss der Korrosion auf den Verschleißprozess...... 129 7.3.4 Einfluss der aufgebrachter Normalkraft und Radgeschwindigkeit auf den Verschleißprozess ...... 132 7.3.5 Vergleich der Verschleißverhalten der untersuchten Hartmetalle ...... 133 8 Zusammenfassung (Englisch)...... 136 9 Zusammenfassung (Deutsch) ...... 138 10 Literaturverzeichnis...... 140

1 Introduction 1

1 Introduction

Tungsten carbide hardmetals consist predominantly of tungsten carbide which is then cemented together with a binder phase of which cobalt is the most commonly used. The basic tungsten carbide-cobalt has been modified over the years for use in a wide range of applications including mining, cutting, rock drilling, woodcutting tools, structural parts and wear parts. The main application has been in the field of metal cutting as the hardmetals can retain sharp edges and resist chipping to a greater extent compared to other [1]. Tungsten carbide alloys were introduced into the woodcutting industry as cutting tools due to their mutually exclusive properties of high wear resistance and moderate toughness [2].

The lifetime of cutting tools may be limited by a variety of wear processes which are caused by factors such as the loading, the cutting speeds and the movement of the chip over the tool face. The cutting tool environment may feature high localized temperatures which may to thermal cracking and high stresses. The tool may also be subjected to cyclic impact loading during interrupted cutting. Figure 1.1 shows some of the different types of wear which may be experienced during the cutting process [3]. Along the rake face, the chip motion and high normal stresses can cause crater wear. Along the clearance surface, the tool motion and normal stresses increase the contact area between the tool and the work-piece and produce flank wear. Cutting edge wear is caused by the combined effect of the above two wear types and is the place where the dominant damage mechanisms occur in woodcutting. The response of the tool in this region generally dictates its performance and lifetime.

Chip

Crater wear Built-up edge

Cutting edge wear Fracture

Flank wear New tool outline

Figure 1.1. Wear types commonly found on cutting tools [3].

Wear of the cutting edges has many disadvantages such as a reduction in the surface quality finish, increase in jaggedness of cutting tool edges, loss of dimensional accuracy of the wood and an increase in power consumption [2]. Studying tool wear however, can be complicated for several reasons; • the wear process cannot be directly observed during cutting, 1 Introduction 2

• quantitative data for individual wear mechanisms are not available, • more than one mechanism may operate under similar conditions, • synergistic interaction between different mechanisms may exist.

Although there is an enormous amount of tool life data for metal cutting which can be used to design and improve cutting tools, there is a lack of comprehensive, systematic tool life data in the field of woodcutting. There also appears to be little detailed understanding concerning the mechanisms of woodcutting tool wear. Unfortunately, the data from metal cutting cannot be directly applied to woodcutting since the environments and tool sensitivities are different.

Tool wear during woodcutting is complex due to the multi-component nature of the wood and the nature of the cutting process itself. The tungsten carbide tools, which are generally used to cut wood, are known to be one of the most wear resistant tool materials. However, their exceptional wear resistance is adversely affected due to the combined presence of corrosive wood extractives and the abrasive wear experienced during machining [4-9]. The degradation mechanisms which develop during the cutting process have been shown to be system dependent, i.e. on in the type of wood that is being cut and in the cutting tool and machine conditions being used.

Wood can be classified into three main types namely, wood in its natural state, cured wood and wood-based composite materials. The mechanical, physical, compositional and structural variables of wood differ not only between types but also within each category. This means that each wood type would exhibit its own specific wear characteristics. With respect to the contribution to blunting from the tool piece and machine conditions, two of the important factors are the cutting speed and the loading [4]. They affect factors such as the machine vibrations, the type of chip formed, frictional forces and the surface finish of the wood. Depending on the cutting and wood conditions, some of the wear mechanisms may become rate controlling. For example, it has been shown that the chemical attack by the wood accelerate the degradation process, whereby the cobalt binder phase is preferentially removed from the hardmetal matrix thus subjecting the exposed carbide grains to increased mechanical wear [5-9].

Although considerable research has taken place, showing that cobalt removal occurs, it does not elucidate the mechanisms by which the cobalt is removed from the microstructure nor the factors which play a role during this removal process. Identification of these factors is often difficult due to the complexity of the woodcutting procedure and the possible synergistic interactions which may be present between mechanisms. Furthermore, previous research has concentrated on hardmetals with carbide grain sizes greater than one micron, whilst recently finer-grained alloys have been employed where limited research into the wear and corrosive-wear interactions of the woodcutting tools has been conducted. There is thus a need for more systematic studies of the wear and the corrosion-wear synergism of the cutting tools used in woodcutting. It is necessary to understand these mechanisms in order to minimize and possibly control wear such that either the original tool edge geometry is retained or the regions of breakage and wear are minimized. In addition, the establishment of more fundamental quantities for describing tool wear may make comparison of separate tool wear studies easier. 2 Einleitung 3

2 Einleitung

Hartmetalle bestehen überwiegend aus Wolframkarbid im Verbund mit einer metallischen Binderphase, für die in der Regel Kobalt verwendet wird. Durch Modifikation der grundlegende Wolframkarbid-Kobalt Legierung kommen Hartmetalle im einem großen und sich noch ausweitendem Anwendungsbereich zum Einsatz. Sie finden Anwendung im Bergbau, bei der Zerspanung, als Konstruktionsteile und im Verschleißschutzsektor. Ihre Hauptanwendung liegt auf dem Gebiet der spanenden Bearbeitung, da Sie die Erhaltung einer scharfen Schneidkante ermöglichen und gleichzeitig eine hohe Resistenz gegen Ausbrüche aufweisen [1]. Wolframkarbid- Hartmetalle wurden in die Holzindustrie eingeführt, da sie die gegenläufigen Eigenschaften - hohe Verschleißfestigkeit bei angemessener Zähigkeit – in sich vereinen [2].

Im Folgenden soll speziell auf die Anwendung von Hartmetall für die Holzzerspanung eingegangen werden. Die Lebensdauer eines Werkzeuges kann durch verschiedene Verschleißprozesse begrenzt werden, die von Parametern wie der Belastung, der Schnittgeschwindigkeit und der Bewegung des Spanes über die Werkzeugoberfläche verursacht werden. Einsatz eines Schneidwerkzeuges kann durch die Entstehung - hoher örtlich begrenzter Temperaturen gekennzeichnet sein, die zu thermischen Risse und hohen mechanischen Spannungen führen können. Die Werkzeug stehen beim Einsatz im unterbrochenen Schnitt unter zyklisch wechselnder Belastung. Abbildung 1.1 zeigt einige der unterschiedlichen Verschleißarten, welche während des Schneidens auftreten können [3]. Entlang der Spanfläche können die Spanbewegung und hohe Normalspannungen Kolkverschleiß verursachen. Entlang der Freifläche erhöhen die Werkzeugbewegung und die Normalspannungen die Kontaktfläche zwischen dem Werkzeug und dem Werkstück und bewirken Freiflächenverschleiß. Der Schneidenverschleiß wird durch den kombinierten Effekt der oben genannten zwei Verschleißarten verursacht und ist das Gebiet, in dem die dominierenden Schädigungsmechanismen beim Holzschneiden auftreten. Das Verhalten des Werkzeugs in diesem Gebiet diktiert gewöhnlich seine Leistung und Lebensdauer.

Span

Kolkverschleiß

Aufbauschneide

Schneidkantenverschleiß Bruch

Freiflächenverschleiß Ursprüngliche Schneidengeometrie

Abb. 1.1. Typische Verschleißarten an Schneidwerkzeugen [3]. 2 Einleitung 4

Der Schneidenverschleiß hat viele negative Auswirkungen, wie eine Verringerung der Oberflächengüte, die Zunahme der Schneidkantenrauhigkeit, des Verlustes der Maßgenauigkeit des Holzwerkstücks und einer Zunahme der Leistungsaufnahme [2]. Die Untersuchung des Werkzeugverschleißes kann aber aus mehreren Gründen schwierig sein; • Mann kann den Verschleiß-Prozess während des Schneidens nicht direkt beobachten, • quantitative Daten für einzelne Verschleißmechanismen sind nicht vorhanden, • unter ähnlichen Bedingungen kann mehr als ein Mechanismus aktiv sein, • zwischen unterschiedlichen Mechanismen können synergistische Wechselwirkungen bestehen.

Obgleich es eine enorme Menge von Werkzeuglebensdauer-Kennwerten für die Metallzerspanung gibt, die verwendet werden können, um Schneidwerkzeuge zu entwerfen und zu verbessern, gibt es einen Mangel an kompletten, systematischen Werkzeuglebensdauer-Kennwerten auf dem Gebiet der Holzzerspanung. Es scheint auch, daß sehr wenig tiefergehendes Verständnis der Verschleißmechanismen vorhanden ist. Die Daten aus der Metallzerspanung können nicht direkt auf die Holzzerspanung übertragen werden, da die Umgebung und die Anforderungen an die Werkzeug unterschiedlich sind.

Der Werkzeugverschleiß während der Holzzerspanung ist komplex auf Grund der mehrkomponentigen Natur des Holzes und der Art des Scheidprozesses selbst. Wolframkarbid- Hartmetall, das am häufigsten benutzt werden, um Holz zu zerspanen, ist dafür bekannt um einer der verschleißbeständigsten Werkstoffe zu sein. Jedoch wird die hohe Verschleißbeständigkeit nachteilig beeinflusst durch die Wechselwirkung zwischen dem Vorhandensein korrosiver Hölzextrakte und dem Abrasiververschleiß, die während des Schneidprozesses auftritt [4-9]. Es wurde gezeigt, dass die Schädigungsmechanismen, die sich während des Schneidprozesses entwickeln, systemabhängig sind. D.h. sie hängen ab von der Art des Holzes, das geschnitten wird und den Schnittparametern des Schneidwerkzeuges und der Maschine, die verwendet werden.

Holz kann in drei Hauptsorten eingestuft werden, in seinen Naturzustand, im trocknes Holz und in künstliche Holz. Die mechanischen, physikalischen, und strukturellen Eigenschaften des Holzes und seine Zusammensetzung unterscheiden sich nicht nur je nach Hauptsorten, sondern auch innerhalb jeder Kategorie. Dies heißt, daß jede Holzart ihre eigenen spezifischen Verschleißeigenschaften hat. In Bezug auf den Beitrag der Schnittparameter des Werkzeuges und der Maschine zum Schneidkantenverschleiß, ist eine der wichtigen Kenngrößen die Schnittgeschwindigkeit [4]. Sie beeinflußt Faktoren wie Maschinenschwingungen, die Art der Spanbildung, Reibkräfte und die Endoberfläche des Holzes. Abhängig von den Schnittparametern und der Art des Holzes können einige der Verschleißmechanismen bestimmend für die Verschleißrate. Zum Beispiel ist es gezeigt worden, daß der chemische Angriff durch die im Holz enthaltenen Säuren den Abbauprozess beschleunigt. Die Kobaltphase wird vorzugsweise aus der Hartmetall Matrix entfernt, wodurch die hervorstehenden Karbidkörner einem erhöhten mechanischen Verschleiß unterworfen werden [5-9].

Obgleich die bisherige Forschung zeigt, dass der bevorzugte Abtrag von Kobalt auftritt, erklärt sie nicht die Mechanismen, durch die das Kobalt entfernt wird, noch bestimmt sie alle Parameter, die während dieses Prozesses eine Rolle spielen. Die Bestimmung dieser Faktoren ist schwierig aufgrund der Kompliziertheit des Zerspanungsprozesses und der möglichen synergistischen Wechselwirkungen zwischen Mechanismen. Außerdem hat die bisherige Forschung sich auf Hartmetalle mit Karbidkorngrößen größer als ein Mikrometer konzentriert, während seit kurzem feinkörnige Legierungen eingesetzt werden, für die bisher nur wenig Forschung über die 2 Einleitung 5

Wechselwirkung von Korrosion und Verschleiß geleistet worden ist. Folglich gibt es eine Bedürfnis für systematischere Studien des Verschleißes und der Korrosion-Verschleiß Wechselwirkung an Schneidwerkzeugen, die bei der Holzzerspanung benutzt werden. Es ist notwendig, diesen Mechanismen zu verstehen, um der Verschleiß zu verringern und gegebenenfalls so zu steuern, dass entweder die ursprüngliche Werkzeuggeometrie erhalten bleibt, oder die Bereiche der Ausbrüche und des Verschleißes minimiert werden. Zusätzlich kann eine grundlegendere Beschreibung der Werkzeugverschleiß-Parameter den Vergleich verschiedener Verschleiß- Studien erleichtern. 3 Objectives 6

3 Objectives

The main objective of this study was to conduct a systematic investigation of the abrasive wear and corrosive-abrasive wear performance of tungsten carbide hardmetals used in woodcutting in order to quantify some of the microstructural, environmental and operational parameters which control the overall wear process. This objective has been explored through two different perspectives, namely: • An investigation of the wear response of ultrafine-grained tungsten carbide hardmetal woodcutting knives during commercial woodcutting operations, i.e. field testing. • An investigation of the wear response of tungsten carbide hardmetals in closely controlled laboratory conditions in order to examine in detail some of the parameters which are believed to control the wear process.

The field tests had the following objectives; • to quantify the wear response of the hardmetals with respect to the type of wood cut, • to correlate the microstructural properties of the hardmetal to the wear process, • to investigate the incipient wear stage, • to assess the influence of corrosion on the wear process and • to quantify the effect of cutting speed on the wear response. The laboratory tests had the following objectives; • to investigate the mechanisms of wear taking place under dry and corrosive wear conditions in a purpose built three body abrasive wear rig, • to evaluate the influence of acid concentration on the corrosive wear behaviour, • to quantify the effect of normal applied loading and abrading wheel speed on the wear response, • to investigate the corrosion response of the hardmetals by conducting polarization curves for each hardmetal using the electrolytes used in the wear tests, • to determine the mechanical properties of the hardmetals, such as hardness and fracture toughness, which may influence wear, • to characterize the microstructural parameters of the hardmetals, • to correlate the microstructure with the wear response of the alloys and • to compare the wear response of the ultrafine-grained hardmetals to that of a fine-grained hardmetal.

To the author’s knowledge this type of approach to study the wear of woodcutting tools using laboratory tests in addition to in-situ experiments has never been carried out or reported previously. This is probably due to the fact that parameters such as temperature and machine vibrations which influence the wear process in practice are significantly reduced and possibly eliminated during laboratory wear testing. Nevertheless, laboratory testing allows the possibility of a closer investigation of some of the variables which may contribute to the wear and corrosive wear behaviour of the hardmetals. As such the laboratory tests should be seen as complimentary tests which can be used to gain new insights into hardmetal behaviour under abrasive wear and corrosive-abrasive wear conditions. 4 Literature Review 7

4 Literature Review

4.1 Woodcutting tool wear

The blunting and eventual wear of woodcutting tools depends on the mechanical, physical, structural and compositional variables of the wood being cut and the cutting tool machinery being employed. The basic parameters which are used in industry to supervise cutting tool wear include [2]; • assessing the overall product quality, • monitoring power consumption of the machinery, • monitoring the cutting edge profile, • working with defined service tool lifetimes, • monitoring deviations from desired cutting lines, • observing changes in the surface roughness of the wood.

Woodcutting tool research done thus far primarily focuses on either the cutting tool, the type of wood or a combination of the tool-wood system. A review of some of the general trends in woodcutting tool wear will be given in the following sections, beginning with a basic definition of wood itself.

4.1.1 Wood

Wood is a natural product with a complex structure, making it an inhomogeneous material. It is made up of three polymers, namely, cellulose, hemicellulose and lignin [10]. In addition, small amounts of substances such as tannins, resins and fatty acids can also be found in wood. Although these substances are found in small amounts, they play a huge role in the properties of the trees. Within a single tree the chemical composition of the wood can vary from the root to the leaf, from the center to the bark, from young sappling to mature growth, thus making generalizations of wood behaviour difficult [10]. A number of organic acids have been found to occur in wood, for example, tannic acid, citric acid, acetic acid and formic acid. The type of acid is related to the tree species and identification of these acids is normally done by steam distillation of the wood [11]. Kivimaa [12] was one of the first researchers to show that these wood acids, which depending on the water content of the wood, cause electrochemical processes during woodcutting which lead to cutting tool wear.

Wood is also known to be an anisotropic material which displays varying mechanical properties in different directions which is often influenced by its moisture content [13]. Below the fibre saturation point(FSP), which is about 30%, mechanical properties such as hardness, fracture toughness and modulus of elasticity increase with decreasing moisture content. Above the FSP, the mechanical properties are less sensitive to increasing moisture content [14]. The equilibrium moisture content of cured wood is determined largely by the relative humidity of the surrounding air [15]. Green wood which has a higher moisture content, is weaker than dry wood. When it dries, thereby losing its moisture, it shrinks, becomes denser and therefore stronger. It has been postulated that the denser a wood becomes the more rapid the blunting rate of the cutting tools will be [10]. 4 Literature Review 8

Different wood species can produce, under the same working conditions, different types of tool wear resulting in different service lifetimes. For example, it has been shown that when tungsten carbide are used to cut Douglas-fir, the tool life times are 40 hours. However, they blunt in half this time when cutting Cedar [9]. The fundamental causes of tool wear, when cutting natural wood, are not clear since they are not easily identifiable. Thus, research normally attempts to relate wear to a quantity which describes an important characteristic of the wood being machined, such as specific gravity, density, hardness or moisture content [16-18]. In addition to wood in its natural state there is also a sub-category of wood, referred to as wood-based composites. These are synthetic materials, for example, chipboard and fibreboard, and since their composition and structure are known, the isolation of potential wear causes is easier than for natural woods.

4.1.2 Tool wear

It has been reported that the blunting of cutting edges can take place by one of two ways [4]. Firstly by large scale fracturing of the tool edge which is usually due to poor tool design or operating procedures and is easily avoidable. The second type of wear which is unavoidable, occurs on a microscopic scale where particles are gradually worn away. Wear makes cutting tools unsuitable for continued use and the cost of replacement of worn tools can be minimized if the wear can be controlled. The characterization of tool wear requires a study of the interaction between the tool, the wood and the machining environment in an attempt to identify the primary wear mechanisms. Once these are known, then further research can be more clearly focused such that the influence of changing cutting parameters can be more accurately assessed. Since several parameters can be changed, the direct effect of the basic factor influencing wear due to a parameter change is not always clear. For example, decreasing the speed is expected to generally decrease the wear [4]. However, it is uncertain if it is due to the change in temperature, forces, type of chip formed or to a combination of these factors. Every wood-tool-machine system is unique, and a small change of system parameters may lead to radical changes in both the wear rate and the wear mechanisms.

Several wear mechanisms can occur either individually or simultaneously, including gross fracture or chipping, abrasion, erosion, microfracture, chemical or electrochemical corrosion and oxidation. For example, Nordström and Bergström [14] conducted case studies of teeth used in saw mills used to cut Spruce and Pine and found that several mechanisms were simultaneously interacting, namely, abrasion, chipping, tribochemical, delamination and cracking. Depending on the cutting conditions and the work piece conditions, some of these mechanisms may play a dominant role and may become rate controlling. Several researchers have investigated the wear characteristics of woodcutting tool wear under various conditions [see for example 16-28]. The research includes methods used to describe blunting of tools, the parameters which influence wear and the major mechanisms which have thus far been identified. These will be highlighted in the following two sections.

4.1.3 Methods used to assess the blunting of the cutting edges

The cutting tool performance depends on the cutting edge profile since the energy is transmitted into the wood through the edge. A sharper edge is more efficient in focusing the energy, but is conducive to rapid edge deterioration. Typical cutting edge wear, when cutting wood, is progressive tool nose rounding, which is different from that found when metals are machined in which crater and flank wear often predominate. This typical wear profile consists of a rounded edge joining the rake and clearance faces such that negative rake and clearance angles are produced 4 Literature Review 9 by wear [19,20]. The profile has been found to be independent of the tool used or the wood being cut [20].

Tool nose rounding has generally been studied by giving geometric descriptions of the worn edges. Figure 4.1 shows the three typical wear forms a cutting edge can have which depends on the working environment and lifetime. They may also occur simultaneously. The three types are ‘rounding’ wear, ‘free surface’ wear and erosive wear [2]. Rounding wear defines the rounding off of the cutting edge so that it has a constant radius. Free surface wear defines a combination of rounding wear and in addition flattening of the edge occurs. Erosive wear occurs in the area close to the cutting edge. There are many ways in which to describe the changing wear profile and in figure 4.1 three dimensional descriptions are given by; the edge recession from the original edge parallel to the rake face(RR), the cutting edge offset(CEO) and the nose width(NW).

Rounding wear

CEO

Free surface wear NW

RR

CEO

Erosive Wear

RR

CEO

Figure 4.1. Types of cutting edge wear [2].

Since the cutting edge changes during use, the strength and wear resistance of the tool and the force system changes with time. Therefore when cutting edges reach a critical condition, they need to be re-sharpened or changed. Due to the various wood-tool combinations, the critical conditions need to be assessed for each combination separately. In the woodcutting industry, a RR of 0,1 – 0,2mm and a NW = 0,2mm are often deemed critical [2]. Cleaning the tools at regular intervals limits wear. Uncleaned knives lead to bad quality in wood products, shortens the cutting length and increases the cutting forces. Dirt particles can also act as abrasive wear particles during further cutting. 4 Literature Review 10

Frequently cutting edges do not wear evenly and due to a local wear ‘difference’ have to be changed even though they are still operational. This unevenness of the edge profile is shown in figure 4.2 which is taken from research done by Sheik-Ahmad and Bailey [21] who used a range of hardmetals to cut particleboard. The amount of edge recession was found to vary with board density variation, giving a jagged profile.

Figure 4.2. Schematic showing the variation of wear along the cutting edge as caused by density gradients across the thickness of the particleboard [21].

Methods to give geometric descriptions of the edge wear profiles include a measure of the edge recession from its original position which is often quoted as an average value. The average edge recession value is obtained by calculating the average of a limited number of measurements along the cutting edge. This is sometimes done by performing equi-spaced measurements from the original edge to the worn edge. As the edge recession is generally non-uniform along the cutting edge, sometimes the projected area representing the recession is used as a measure instead. The edge blunting can also be measured in terms of the volume of the tool worn away. Specific comparisons of worn tool geometries for different tool materials are available. For hardmetals the edge recession has been found to generally decrease linearly with an increase in hardness of the tool and decreasing binder content.

4.1.4 Influence of force, temperature and speed

Measurement of tool wear using force, temperature and speed is more representative of the cutting environment as it takes the entire cutting region into account, while geometric measurement is more useful to study wear mechanisms since it can be localized. Much research has characterized tool blunting in terms of measured cutting forces and temperature, and to a lesser extent with regards to cutting speed.

Forces which act during woodcutting consist generally of [2]; • direct cutting forces, • frictional forces, • centrifugal forces associated with rotating cutters, • oscillating forces associated with bi-directional movement cutters, • thermal forces, • impact forces, • forces due to attachment of the tool in the cutterhead and 4 Literature Review 11

• forces due to machine vibrations.

In cutting tools, the rake, clearance and wedge angles describe the distribution of material in the cutter and to a large extent determine the magnitudes and lines of actions of the cutting forces. Hence, they would affect tool wear and several studies have been done which show this effect, with the wedge angle having the most effect [22-24]. A rapid increase in the cutting force occurs during the initial stages of cutting after which the force tends to level off [20,25]. Machine vibrations have been found to influence not only chip formation but also the friction forces and coefficients [4,26].

The inclination of the knife in the tool holder also influences the magnitude of the forces applied. The benefit of inclining the knife to the work piece is to control the amount of energy input into the system, which results in a lower transfer of stress from tool face to wood, hence causing smaller deformation zones in the contact zone. Sheik-Ahmad and Bailey [21] measured the normal cutting force for a range of cemented when cutting particleboard and found that the edge recession increased with increasing cutting force for the range of materials tested. The cutting forces measured were dependent on the binder content of the hardmetal, with higher forces being recorded with increasing binder content. This relationship is shown in figure 4.3.

Figure 4.3. The relationship between the normal force and the average edge recession for a variety of tungsten carbide grades after a cutting distance of 2200m [21].

The cutting process involves intense contact between tool and wood, which can result in high temperatures and stresses. Cutting temperature is important since material properties such as hardness, toughness and chemical stability change with increasing temperature. The problem of determining tool temperature has been the subject of many experimental investigations and a summary of some of the work is given in Table 4.1 [27]. The temperatures shown in the table were recorded close to but not directly on the cutting edge. A few researchers have attempted to determine the temperature theoretically, where the accuracy of the calculation depends on the assumptions made concerning boundary conditions and the proportion of the cutting energy which is converted into heat at the tool/wood interface [28]. The cutting tool temperature is dependent on factors such as speed, feed rate, depth of cut and wood type, with speed being the most important. Kivimaa [12] showed that the cutting speed had an influence on the blunting of the cutting knives 4 Literature Review 12 and proposed that for each woodcutting case an optimum speed exists where blunting can be kept to a minimum and cutting length would therefore be at a maximum.

Table 4.1. Previous experimental work on wood cutting tool temperature [27].

The temperatures experienced during woodcutting may influence the type of wear mechanisms which occur and it has been postulated that high temperature oxidation of cobalt takes place during the cutting of manufactured wood products. The validity of this statement has been subject to debate since some researchers claim that the temperatures measured during woodcutting are not high enough to cause oxidation of the tungsten carbide grains and the oxidation of cobalt occurs in air over a wide range of temperatures above ambient temperature and that the oxidation rate is very low, below 700°C [29]. The differing opinions on this mechanism is mainly due to the inability to directly measure the temperature at the cutting edge and possibly also due to the type of micro-analytical methods used to examine the worn surfaces [27]. X-ray photon spectroscopy (XPS) is a good technique which can be used in the determination of oxides, since the emitted 4 Literature Review 13 photoelectrons are sensitive to the bonding of the atoms. However, this method has rarely been used and thus the type of reactions and mechanisms caused predominantly by the cutting temperatures remain unclear.

4.1.5 Tool wear relating to the cutting of natural wood

In natural wood, the quantity and chemical nature of the extractives are important factors which affect tool wear. In section 4.1.1 it was mentioned that acids such as tannic and acetic occur in wood and depending on the moisture content of the wood, these acids can lead to corrosion of the tool during cutting, adding to the overall tool degradation. It is thus expected that the wear rates for cutting green woods would be higher than those for cured woods, since the moisture content is higher in the green woods.

It has been proposed that the tool metal forms chelation reactions with the extractives present in the wood [9]. Chelation is defined as the formation of a chemical complex between a metal and two or more polar groupings of a singe molecule, by which the metal ion is firmly bound with an organic molecule to form a ring structure [30]. Most metals form chelation reactions, thus the probability of a metal forming chelates with wood must be high. The chemical properties of iron and cobalt are known to be similar, so it is expected that cobalt will undergo chelation with the extractives as readily as iron, leading to increasing wear rates. The chelation of tungsten carbide is difficult to predict.

When cutting wet Western Red Cedar, it was found that tungsten carbide tools blunt within 15 minutes of cutting. This was indicated by a rapid increase in power consumption and surface roughness [9]. Kirbach and Chow [9] investigated this effect by studying the reactivity of cobalt and tungsten carbide in solutions of Western Red Cedar extractives at temperatures of 20°C and 60°C. They showed that the cobalt binder is preferentially attacked while the carbide grains remained relatively unaffected since they retained their sharp edges. The corrosive nature was found to be temperature dependent where the metals lost 35% more weight at 60°C than at 20°C. This indicates that the increase in temperature accelerated the dissolution process and may control the rate of chelation. Both worn and unworn tool inserts were investigated and the unworn carbide inserts showed very little dissolution of the carbides, while the worn inserts showed a higher amount of carbide dissolution. This indicates that lattice damage due to wear facilitates chemical access resulting in higher wear rates. They further proposed that carbide grains are held in position by the cobalt and when this cobalt is removed then the carbide grains either fall out or are removed by mechanical means.

Bailey et al. [6] found a similar type of degradation when examining the mechanisms of wear using tungsten carbide inserts to machine green Oak. The cobalt binder is removed from the interstices between the carbide grains, while no attack occurred between the extractives and tungsten carbide grains. They proposed that the depth to which the cobalt binder is removed, is of the order of the tungsten carbide grain size at any stage in the life of a tool. They further proposed that wear occurs primarily by preferential dissolution of the binder phase though chemical attack by the extractives present in the Oak which is then followed by loss of individual carbide grains when the strength of the remaining WC-Co-bond is insufficient to resist action of the shear forces caused by motion of chip and wood over the tool face. This motion would, in addition, be responsible for removing any corrosion reaction products which may form during cutting. Such a mechanism is fundamentally different to that which usually occurs in metal cutting.

4 Literature Review 14

Tsai and Klamecki [31] formulated a simple model, which represents a special case of the Taylor tool life equation, which could be used to determine the relative contributions to the total tool wear due to abrasion and due to electrochemical effects in high speed steel cutting tools. They also showed that when cutting wood which had a low pH, the wear rate was greater than cutting wood with a higher pH value.

Bayoumi and Bailey [5] developed a wear model, shown in equation 4.1, to predict the progressive wear of cemented carbides in the presence of dilute organic acids. The expression gives the wear rate in terms of the interparticle spacing(p), the average carbide grain size(d), the first order corrosion reaction rate constant(k), the shear force(F) exerted on each carbide grain during testing and on the ultimate shear strength(τ) of the binder particle interface.

−1 • d + p   144p(d + p)2 ( p + d cosθ )  E4.1 W = k ⋅ ln 0  3 2 2  πd (3p + d)(8 + 9cosθ0 − cos3θ0 ) − 48d( p + d) ( p + d cosθ0 ) where −1  2F  θ 0 = cos 1−   πτd 2 

The model was found to be in good agreement with experimental data obtained using two tools to cut fibreboard in wet conditions. The model was not tested on green woods. However, the model was found to be invalid for small binder volume fractions due to the assumption made that the carbide particles are arranged in a bcc array. Later research done by Bayoumi and Bailey [7], showed that the model was in good agreement with experimental results for a range of cemented carbides. From both the model and experimental data it was found that the wear rate increased with increasing binder volume fraction, an increase in the normal rubbing force and an increase in the carbide particle size.

Tsai and Klamecki [31] found a similar influence of binder content on the width of the wear surface when testing various tungsten carbide grades against fibreboard in wet and dry conditions. There was a significant difference in the wear behaviour of the grades which depended on the amount of binder as is shown in figure 4.4. This was attributed to the fact that more binder, meant greater binder mean free paths, which allowed easier removal of the cobalt through electrochemical corrosion. This was confirmed by experimental work done by Bayoumi et al. [8].

4 Literature Review 15

Wear rate

Figure 4.4. Effect of binder content on the wear rate for a range of hardmetals tested under wet sliding conditions using a tannic acid concentration of 0.1N [8].

The same authors [31] also found that the tool lifetimes were shorter when cutting water-saturated wood than when cutting dry wood of the same species. This was attributed to increase in the wear rates due to the corrosive attack on the metal by the extractives present in the wet wood. The mechanism of wear proposed when cutting dry wood was that removal of the binder occurs mechanically by rubbing or chemically by oxidation due to high temperature, which is then followed by attrition of the oxide. However, this mechanism was not confirmed.

The chemical processes which occur when cutting fresh wood has been found to be sensitive to electrical fields and by applying a negative potential to the tool, the electrochemical activity in the tool-work-chip region is suppressed and thus wear is reduced [11,31,32]. Kivimaa [12] applied both positive and negative potentials and found that the wear rate was significantly greater with positive potentials. Tsai and Klamecki [31] found the same effect when cutting Ponderosa Pine and Incense Cedar. It has also been speculated that wood chips possess a negative electrostatic charge and they may cause pitting as they leave the tool face [12].

An interesting fact is that some tannin extracts have been shown to form passive films on irons and due to reactions between the carboxyl groups and the metal. These films then give protection against atmospheric corrosion [33]. However, in the cutting procedure, film formation is prevented due to the rapid motion of chip, wood and tool. This motion removes any reaction products which may accumulate, preventing film formation. Thus, essentially a clean surface is repeatedly exposed to corrosive attack.

4.1.6 Tool wear relating to the cutting of synthetic wood products

Research which focused on the cutting of synthetic wood products such as particleboard, showed that factors such as the board density, resins and glues as well as dioxide particles which may be present, influence the tool wear. Silicon dioxide sand is normally used to smooth the surfaces of the particleboard and since a large amount of the powder is usually ground against the wood surface, some of the particles may become embedded in the surface which subsequently may cause increasing wear rates. The effect of silicon dioxide particles was investigated by Huber 4 Literature Review 16

[34], who found that the wear rate increased linearly with increasing silica particle content. Tool wear was shown to be more closely related to the quartz content of wood than to the total silacaceous residue content [4].

Pahlitzsch and Jostmeier [35,36] showed that when cutting particle board with tungsten carbide tools the wear showed a maximum wear tool area which corresponded to the region in contact with the board surface and a wear minimum near the center of the cutting region. This profile has been mentioned in section 4.1.3 and is attributed to differences in board density and silicate content. Huber [34] suggested that particleboard which can exhibit low tool wear in both the surface and middle layers can be manufactured depending on it’s structure and the composition of the materials used.

Bridges [37] investigated the dulling of steel cutting edges when machining three layered Southern Pine particleboard. The blunting was attributed to a combination of the effects of cured urea- formaldehyde resin and the higher board density. A similar effect of the resin glue content on wear was also shown by Rackwitz [38] and Pahlitzsch and Dziobek [22]. Bridges [37] also found linear increases in the abrasiveness of the board with increasing board density, silicates and resin content. Similar linear relationships were found by Huber [34]. Bridges [37] found that increasing the resin content from 5 to 8% had no effect on the tool wear, but when increasing the resin content from 9-11%, the abrasiveness almost doubled. He also observed during the cutting of wood of the same species dulling does not occur.

Sheik-Ahmad and Bailey [21] found that during the machining of manufactured wood products, most of the wear is found on the clearance face of tool, due to the rubbing action between the tool face and the wood. Less wear is found on the rake face, since no continuous chip is formed in machining synthetic wood products, as they machine mostly by fracture. When machining such materials which can have abrasive fragments, loose micro-fragments can penetrate the tool under cutting pressure and preferentially remove the cobalt.

Sheik-Ahmad and Bailey [21] attempted to elucidate the removal mechanism of hardmetal tools used in the machining of particleboard. They found fracturing of the tungsten carbide grains, but no evidence of oxidation nor pitting. Prakash [39] found similar features when machining fibreboard. Sheik-Ahmad and Bailey [21] proposed that the mechanism of cobalt removal takes place via a ‘soft abrasion’ mechanism as defined by Larsen-Basse et al. [40] and Jia and Fischer [41] when hardmetals are abraded by which are softer than the hardmetal. In this mechanism, wear proceeds by removal of the cobalt binder from between the carbide grains, which is followed by the fragmentation and removal of the carbide grains. When relating it to woodcutting tool wear, Sheik- Ahmad and Bailey [21] suggested that the cobalt binder is extruded from between the carbide grains due to oscillation of the carbide grains which is caused by fluctuating forces during the cutting of the particleboard. This type of mechanism is then dependent on the hardmetal binder mean free path and the carbide grain size. It would not be valid for very small mean free paths as the carbide grains would not have any room to oscillate. For hardmetals with a high content this mechanism may also not be valid since the binder is harder. In this latter case it is more likely that brittle fracture in the binder would occur leading to large fall-out of tungsten carbide agglomerates, resulting in depressions and pitting on the tool surface.

4 Literature Review 17

4.2. Woodcutting tools

The most important quality for a is its wear resistance and as a result hardness and toughness are priorities. Cutting edges need to retain their sharp edges in order to ensure long service lifetimes. In general, according to Maier [2], the metals for woodcutting need to be; • compact: they need to resist mechanical destruction. Due to the cutting forces, the edges can experience large internal stresses; • tough: with increasing brittleness, the probability of chunking and uncontrollable damage increases; • wear resistant: ensures increased lifetimes; • hard: resistance due to wearing increases with increasing hardness; • fine-grained: smaller chunks fall out and the durability increases; • easily machinable: production and re-sharpening costs are kept low.

It has been reported that the surface condition of the knives also play a role in determining useful tool lifetimes. Galli [42] reported on the importance of correct carbide grade selection, grinding and mounting techniques in the woodcutting industry. He discussed the importance of the cutting edge preparation, since this is what reflects the performance of the tool. For example, in soft wood applications, improper edge preparation can cause surface defects such as ‘fuzzing’ of the grain. Tool life can also be compromised by lack of precision and care of grinding operations. For example, improper grinding can cause overheating of carbide tools resulting in thermal cracks and oxidation of the binder.

Prokes [18] showed that for a surface roughness greater than 2µm, wear increased with increasing surface roughness and that carbide cutting tools showed greater wear rates than tool steels. This may have been due to the effect of corrosion on the carbide tools. Pahlitzsch and Dziobek [43] showed that measured cutting forces and temperatures were lower for smoother surfaces. Surface treatments such as coatings or can increase the wear resistance. However, they can be sensitive to temperature changes during cutting. Reynolds et al. [44] showed that the power requirements for chain saws decreases when the burrs due to sharpening are removed.

In order to rationally design cutting tools, it is necessary to identify and understand the mechanisms which cause wear. Such studies can be time consuming and a simple way which has often been used, is to simply evaluate the wear performance of different tool materials and assess their suitability for woodcutting. Comparative studies of the wear resistance of various metals are available, see for instance Bayoumi and Bailey [7]. It has however been shown that abrasion is one of the dominating degradation mechanisms, hence the use of hard, abrasion-resistant materials such as tool steels, high speed steels and cemented carbides [4]. Cutting tools made from polycrystalline diamond have very recently been introduced, but are expensive compared to hardmetals. Though abrasion is a dominant degradation mechanism in woodcutting, there is often more than one mechanism acting as was discussed in the previous section. Thus, in order to properly select tool materials for woodcutting, it is important to know the different types of degradation mechanisms, their relative magnitudes, their interaction and the parameters which influence them. There is also a need to consider tool-wood pairs individually since different combinations may give different results. For this study tungsten carbide hardmetals were chosen and thus the constituents and their respective properties will be reviewed in the next section. 4 Literature Review 18

4.3 Tungsten carbide hardmetals

Tungsten carbide was discovered more than 100 years ago by Henri Moissan in 1896 [45]. He first melted buttons of tungsten together with a great excess of tungstic oxide and obtained a bright and very hard metal with a density of 18.7g/cm3. He then heated a mixture of and sugar coal which resulted in a melted tungsten button assaying 99.87%W, without any carbon. Later by melting the buttons with excess carbon he prepared the compound W2C, a gray, very hard material, with a carbon content of 3.05 to 3.22% and a density of 16.06g/cm3.

By 1926, the Krupp Company commercially produced “WIDIA-N” (WC-6%Co), the first hard metal tool, where the name meant “diamond-like”. This hard metal was based on a patent developed by Karl Schroter [46]. The tools were initially developed in response to demands for a material that was sufficiently wear resistant for wire drawing dies. It was soon discovered that the new material could be used for machining and that cemented carbides have a higher hardness and wear resistance, including hardness at elevated temperatures, than high-speed tool steels. Their great technical and industrial importance has led to numerous experimental investigations aimed at developing the best hard material for a particular application.

4.3.1 Hardmetal manufacture

The most important factors affecting the properties of hardmetals in service are the fabrication process and the microstructure generated during manufacture. This means optimization of the material selection, design and manufacturing process for a particular service application. Hard metals, such as the cemented carbides are more sensitive to the manufacturing process than many other materials due to their brittle nature.

The manufacturing process consists predominantly of; • production of powders, • milling, • pressing, • pre- and • sintering.

Various techniques have been developed to produce tungsten carbide powders. The traditional method of tungsten carbide powder manufacture is based on the production of tungsten powder by hydrogen reduction of tungsten oxide and subsequent carburisation. Alternatively, direct carburisation of tungsten oxides can also be carried out [69]. In 1992, a newly developed spray conversion process (SCP) made it possible to synthesize WC-Co composite powders “in situ” compared with the making of tungsten carbide and cobalt powders separately and then blending them together [47].

Once the powders are produced, they are milled together. The main aim in hardmetal milling is to obtain a homogeneous dispersion of tungsten carbide in the cobalt. During the course of milling, stresses are induced in the tungsten carbide grains and often a decrease in the tungsten carbide particle size also occurs. These two factors facilitate the sintering process [48-51]. The cobalt 4 Literature Review 19 phase may change from a predominantly cubic to a hexagonal close packed structure during milling [49]. Carol [52] found that milling time influences the microstructure of the sintered component. When short milling times, i.e., 12 to 24 hours, were employed, cobalt pooling and extensive residual porosity were present in the sintered structure. Shorter milling times can also result in inhomogeneous mixing which to discontinuous grain growth in the sintered metal.

After milling the powder, a lubricant (e.g. paraffin wax) dissolved in a volatile solvent (e.g. xylol or carbon tetrachloride) is blended into the powder. The main function of the lubricant, besides reducing friction between the powder mixture and the surfaces of the tools, is to minimize the tendency to form cracks. Sintering without such a lubricant may result in interlocking, bridging, intermittent voids, irregular shrinkage, variable density or severe distortion [53].

The milled powders are pressed into shape using rigid steel or carbide dies under pressures of 150-900MPa [53]. Compacting pressures from 50 to 150MPa are usually applied via single or double-acting hydraulic presses. Components may be pressed directly into specified shapes or they may be pressed into a large block and subsequently shaped. During this stage the powder compacts maintain their shape by virtue of cold-welding of the powder grains within the mass. The compacts must be sufficiently strong to withstand ejection from the die and subsequent handling before sintering. The level and uniformity of the as-pressed density determine the final shape and influence the mechanical properties.

After pressing, the metals are pre-sintered in hydrogen at a temperature gradually rising from room temperature to 800°C. The hydrogen reduces the amount of adsorbed and oxides present on the surfaces of the powder particles. On cooling, the material is sufficiently coherent to allow further shaping and is also less susceptible to damage than in the compacted form.

Sintering of cemented carbides is referred to as ‘liquid phase sintering’ because when the pressed compact is heated, a liquid phase is formed by the reaction between the tungsten carbide and the cobalt. Sintering is usually carried out in a vacuum at temperatures between approximately 1350°C (high cobalt compositions) and 1550°C (low cobalt compositions). These temperatures are chosen with reference to the Co-W-C ternary system and the pseudo-eutectic WC-Co diagram(figure 4.5) which is most commonly used in describing the reactions that occur under commercial sintering conditions [54].

Figure 4.5. Psuedo-eutectic WC-Co [54]. 4 Literature Review 20

According to the calculated phase diagram the equilibrium contents of tungsten and carbon in the liquid are as much as 30wt%W and 2wt%C in a stoichiometric WC-Co alloy at 1400°C. This corresponds to a mole fraction of liquid of about 20% for an alloy containing 10wt%Co. The WC- carbide has a very narrow compositional range at temperatures below 2000°C and does not dissolve cobalt [54].

The result of the sintering process is dependent on the inherent properties of the alloy system, but also to a large extent on the production techniques used. Factors that influence sintering also include [54,55]; • the carbon content of the powders, which should be sufficient to maintain stoichiometric WC and avoid eta-phase formation; • the purity of the starting materials, since any oxide particles present are mainly responsible for residual porosity; • the particle size, distribution and morphology, since these parameters strongly influence the packing density and uniformity of packing in the cold compact; • the use of grain growth inhibitors which effectively mitigate tungsten carbide particle growth during liquid phase sintering; • the amount of binder present and the of the carbide in the binder.

During sintering, the compact acquires the strength needed to fulfill the intended role as an engineering component. The rounded shape of the tungsten carbide in the early stages of sintering gives way to a faceted morphology and finally the grains will have the shape of flat trigonal prisms [56].

Hot isostatic pressing(HIP) was invented by Saller et al. [57] in 1955 and first commercially used as a post-sintering operation for cemented carbides some twelve years later [53]. It involves the application of a high pressure (typically 100-150MPa) by an inert gas medium at temperatures 50°C above the solidus temperature of the binder phase. The treatment results in a reduction in porosity. However, HIP does not remove inclusion containing or surface connected pores.

4.3.2 Hardmetal constituents

The cemented carbides investigated in this work are comprised of tungsten (W), carbon (C), cobalt (Co), chromium(Cr) and (V). Tungsten and carbon form the carbide, WC that constitutes approximately 85-97% of the phases present in the sintered hard metal. Cobalt is added to cement the carbides together and provide ductility to the final component. Chromium and vanadium are used as grain refiners during sintering. A brief review is given of the different constituents and their properties.

4.3.2.1 Tungsten

Tungsten exists in nature as , a complex iron-manganese-tungsten oxide, and scheelite, a calcium tungsten oxide. It has excellent resistance to attack by most of the common acids even at high temperatures. The best known use of tungsten is in combination with carbon as sintered carbide for cutting tools, where excellent abrasion resistance is needed. The ability of carbide 4 Literature Review 21 grains to accommodate substantial plastic deformation without the occurrence of brittle fracture is one of the most important properties which has led to the success of cemented carbides.

Tungsten forms two carbides, W2C and WC, but only the latter is used in the manufacture of cemented carbides. Tungsten monocarbide, WC has a simple hexagonal structure, with lattice parameters a = 0.291nm and c = 0.284nm. The ratio of c/a = 0.976 [56]. The is shown in figure 4.6. Carbon atoms are either located at 1/3, 2/3, and 1/2 or all at 2/3, 1/3, or 1/2, void positions.

Figure 4.6. Crystal structure of tungsten carbide [58].

In liquid phase sintered alloys, the carbide grains tend to assume the shapes of flat trigonal prisms impinging upon each other but in insufficiently sintered alloys the crystal faces are not fully developed. Tungsten carbide crystals are anisotropic in most of their properties [56]. No anisotropy is present in sintered alloys because of the random orientation of the carbide particles. The strength of individual tungsten carbide particle decreases with increasing size and that is probably one of the reasons for the different fracture mechanisms in fine-and coarse-grained alloys [47].

Numerous studies have been conducted into the influence of carbide grain size on the mechanical properties of the alloys [59-61]. Figure 4.7 shows that both the hardness and compressive strength decrease with increasing particle size. In figure 4.8 the transverse rupture strength shows a pronounced maximum which lies at a particle size of approximately 3µm for a 12wt%Co alloy. This maximum has been explained in terms of crack propagation. In alloys with a small carbide size, fracture occurs mainly in the cobalt phase, but proceeds preferentially through the carbide particles in coarse alloys [61,62]. The maximum strength is reached when both phases have equal strength.

4 Literature Review 22

Figure 4.7. Effect of WC grain size on hardness and compressive strength [63].

Figure 4.8. Influence of WC grain size on transverse rupture strength for 6wt%Co alloys [63].

4.3.2.2 Carbon

One of the important factors in the sintering of cemented carbides is the carbon balance. Correct control of the carbon is required in order to avoid the formation of the eta phase when there is a carbon deficiency, while if there is an excess of carbon, free carbon can form [1,56,64]. It has been stated that the best combination of mechanical properties can be achieved at the lowest possible carbon balance close to the appearance of the eta phase i.e., the highest fracture toughness for a given hardness [65]. It is necessary therefore to control the carbon content of the system in order to allow the composition to fall inside the two-phase region in which Co(W,C) phase is in equilibrium with WC.

During sintering studies, Gurland [62] determined that the eta phase will form during sintering and will be present at room temperature if the tungsten carbide contains less than 6.00% carbon. The tungsten and carbon taking part in the reaction are the tungsten and carbon in solid solution in the cobalt. In the course of the reaction, tungsten and carbon are removed from the solution, hence the carbide grains adjacent to the growing eta phase partially dissolve [64]. This leads to embrittlement of the structure by the replacement of the binder with an eta-phase skeleton.

4 Literature Review 23

An excess of carbon results in the appearance of clusters, which form weak points in the material. The graphite clusters are small and relatively isolated and are known to moderately decrease the strength and hardness. French and Thomas [58] found that such defects are located in a narrow zone at the surface of the material. According to them, localized excess carbon defects approximate small holes and result in a strength reduction.

The effect of carbon content on the hardness and strength is shown in figure 4.9. The Vickers hardness decreases linearly with increasing carbon content [66] while the transverse rupture strength drops rapidly with increasing decarburisation [62]. This is considered to be due to the brittleness of the eta-phase and to the fact that the eta-phase removes cobalt from the binder [62]. Some researchers [60], have reported only a slight decrease in the strength, in the presence of free carbon, whereas other workers [58] show a pronounced influence, comparable to that of the eta phase. The transverse rupture strength has been shown to increase with increasing carbon in solution in fine-grained alloys, but decreases in coarse-grained alloys [60,62]. It is also believed that the composition of the binder phase will not be affected by the presence of free carbon, due to the limited solubility of carbon in the solid cobalt phase [67].

Figure 4.9. Effect of C content on hardness and strength [63].

4.3.2.3 Cobalt

Although other materials have been used as the binder for tungsten carbide hard metals since 1914, cobalt has dominated this market as the ultimate binder. This is due to cobalt’s good wetting of tungsten carbide in the solid state and its adequate mechanical properties which result in a tougher final product [56]. Cobalt’s good wettability with WC together with the existence of a two- phase region(WC + liquid) in the Co-W-C ternary diagram ensures that high density carbides that do not contain eta-phase or graphite are possible with cobalt [68].

Cemented carbides for technical applications contain from 5 to 25 wt% Co and its effect on properties has been extensively studied. Hardness decreases with increasing cobalt content, and the compressive strength reaches a maximum at 5wt% and then drops sharply when the cobalt content is increased. The transverse rupture strength improves with increasing cobalt content up to a maximum at approximately 20±5wt%Co [59,61,62]. According to Suzuki and Kubota [60] the high ductility of the cobalt phase results in lower strength in coarse-grained alloys and greater 4 Literature Review 24 toughness in fine-grained alloys. Figure 4.10 shows the general trend of these properties for a tungsten carbide particle size of approximately 2µm.

Figure 4.10. Effect of cobalt content on strength [63].

There are two allotropic forms of cobalt namely, the hexagonal centered packed form that is stable at temperatures below 417°C and a face centered cube form that is stable at higher temperatures, up to the (1495°C) [56]. The most accurate lattice parameters of cobalt are the following: hcp: a = 2.5071A and c = 4.0686A with c/a = 1.623; fcc: a = 3.5441A.

The cobalt present in sintered tungsten carbide-cobalt is mainly the metastable face centered cubic phase but can also contain lamellae of hexagonal cobalt [47,69]. The free energy change associated with the fcc⇔hcp transformation is very low and is usually suppressed during cooling by mechanical constraints developing as a result of differential thermal contraction [70].

Other factors which determine the transformation are the amount of tungsten carbide dissolved in the binder and the binder mean free path (i.e. the distance between two carbide particles). A high concentration of tungsten and carbon in the binder has been shown to increase the martensite transforming temperature (FCC to HCP) to about 750°C from 417°C [47]. This avoids formation and expansion of the brittle hcp-phase at low temperatures. Finer-grained hardmetals have been found to have a higher FCC/HCP ratio than conventional metals, due to the higher solution of tungsten in the binder of finer-grained hardmetals, which favors the solubility of tungsten [47].

The amount of tungsten in solution in the binder varies with the total carbon content of the hard metal. At high total carbon contents, the tungsten content is low and vice versa [69]. At room temperature, the cobalt binder contains about 2-10wt% tungsten carbide dissolved in solid solution. The tungsten atoms substitute for cobalt atoms while the carbon atoms fill the octahedral interstices in the cobalt matrix [56]. Cobalt does not dissolve in the tungsten carbide phase.

4 Literature Review 25

The binder mean free path is a measure of the thickness of the cobalt layers and depends on both the cobalt content and the tungsten carbide particle size. The mean free path generally describes the distance a dislocation can move provided the binder region is free of precipitates [70]. Information on the state of the cobalt binder can be obtained from the magnetic coercive force. The coercive force is a measure of the saturation magnetization of the cobalt. It is sensitive to the mean free path due to residual lattice strains and it is also very sensitive to the binder composition since saturation is limited by the electron-to-atom ratio. An increase in coercivity implies that smaller values of mean free path should exist [71]. A slight decrease in coercivity with increasing cobalt content indicates that the mean free path in the cobalt phase increases slightly with cobalt content, which would lead to an expected increase in toughness with increasing cobalt [71].

The thickness of the binder phase layers is the primary variable controlling the fracture toughness, a relationship which has been studied by many researchers [61,72-76]. Gurland and Bardzil [61] showed that above a critical value of the mean free path, the strength follows a dispersion- hardening mechanism [76] and is proportional to the volume fraction of tungsten carbide, but inversely proportional to the tungsten carbide particle size. Below this critical value, a Griffith type [74] relation is followed and strength is controlled by the ease of crack propagation. The compressive strength decreases with increasing mean free path as does the hardness. The transverse rupture strength exhibits a maximum when plotted against increasing mean free path as shown in figure 4.11 [61,63].

Figure 4.11. The influence of the binder mean free path on strength [63].

4.3.2.4 Grain growth inhibitors

Grain growth inhibitors are used in order to produce a finer carbide grain size which increases the hardness and hence the wear resistance of the hardmetals. The most effective grain growth inhibitors have been found to be VC, Cr3C2 NbC and TaC, while the restricting effect is controlled by the MxCy/Co ratio [1]. For example, the optimal VC content is near 5% of the binder content [77]. The incorporation of different carbides such as VC and Cr3C2 in the tungsten carbide grades has proven to be useful in improving the quality of tools employed for specific applications and inhibiting and controlling the grain growth [78,79]. They are usually added in small amounts, less than 1wt%, which predominantly dissolve in the cobalt binder. However, it has been found that VC precipitates of (V,W)C occur while Cr3C2 completely dissolves in the binder and diffuses toward the WC/Co boundaries [80,81]. 4 Literature Review 26

The effectiveness of the grain growth inhibitors is dependent on factors such as the sintering temperature. For example VC is more effective during sintering of micron-sized powder compared to sub-micron and nano-grained powders [53]. The reason for this is that vanadium carbide is most active as a grain growth inhibitor at 1300°C where sintering of micron sized powder specimen is carried out. The sintering of submicron grades is achieved at lower temperatures where vanadium carbide is not so effective as a grain growth inhibitor.

Sintering studies by Egami et al. [80], suggest that one mechanism of grain growth inhibition of tungsten carbide by vanadium is the decrease of the motive force of the precipitation of tungsten carbide from cobalt by the prior precipitation of vanadium compounds at the WC/Co interface during cooling. Luyckx and Alli [81] conducted comparison tests on the effectiveness of VC and Cr3C2 additions and found that VC gave higher hardness for equal inhibitor content, while Cr3C2 gave higher hardness for equal carbide grain size. Cr3C2 also gave higher crack resistance values, which was attributed to the complete dissolution of Cr3C2 in the binder phase which also strengthens it.

4.3.3 Porosity

Studies of the total porosity on the strength of WC-Co alloys have been carried out by various researchers [66,82]. Their results show that material strength varies with the size and distribution of pores. However, hardness is not affected by the variation in porosity, within limits [83]. Thus attainment of a low level of porosity not exceeding 0.1%, is a key requirement during hard metal production. A uniformly distributed residual porosity is usually present and is commonly accepted as not being harmful. Large pores often co-exist with other coarse flaws such as graphite inclusions and uneven cobalt distribution [83]. The porosity level can also be used to judge the success of the densification process [71].

4.3.4 Hardness

The hardness of a metal is usually used as a measure of it’s wear resistance – the harder a metal, the higher the resistance to wear is expected to be. This trend is shown in figure 4.12 for a range of cemented carbides. Hardness is normally defined as a measure of the material’s ability to resist plastic deformation under a load applied to the material by a hard indentor. It is determined by the mean indentation force resisted by the material when the deformation is fully plastic. Since it is a force or a pressure it can be related to the flow stress of the material.

4 Literature Review 27

Wear resistance 3 (m/mm )

Figure 4.12. The relationship between wear resistance and harness for various cemented carbide grades [84].

Thus hardness is a material property that depends on the microscopic structural details of the material such as the amount of cobalt present in the material and the size of the carbide grains. A study of cemented carbides over a limited range of grain sizes has shown that the relationship can be expressed adequately by a Hall-Petch type relation shown in equation 4.2 where a and b are constants [85].

-1/2 2 HWC = a + bd (kg/mm ) E4.2

Hardness has been shown to increase with a decrease in the binder mean free path. The ascending rate of the hardness with decreasing mean free path in ultrafine hard metals has been shown to be much faster than the rate in the coarser alloys. Gurland and Bardzil [61] suggested that the hardness of sintered cemented carbides varies with the mean free path according to an exponential relationship.

Fracture toughness and hardness of cemented carbide hard metals vary in different ways when the composition and/or microstructure of the material are varied. However, not all published results agree on the type of relationship that exists between these two properties [72,86,87]. Schubert et al. [88] found that a higher binder content at a given hardness, combined with a lower average carbide grain size, did not necessarily mean a higher toughness. In particular, at lower hardness values this relationship does not exist. At higher hardness values, high binder contents can reduce the carbide contiguity and thus improve the hardness-toughness relationship.

It has been shown that the increase of the hardness in the ultrafine hardmetals does not always decrease their bulk fracture toughness [89]; behavior opposite to what is found in the coarser alloys. This implies that different toughening mechanisms may exist in the conventional and ultrafine composites. Jia and Fischer [47] explained it in terms of the plastic deformation concept in conventional composites whereas for ultrafine hardmetals, the bridging ligament mechanisms play a significant role in the alloy’s toughness. This explanation was based on the fact that the bulk fracture toughness is related to crack propagation through the phases of the material.

4 Literature Review 28

4.4 Corrosion

When a metal specimen comes into contact with a corrosive environment, the metal typically oxidizes and the environment is reduced. The corrosion response of tungsten carbide hardmetals has been studied in various aqueous solutions, mainly by weight loss measurement from immersion tests, with electrochemical procedures only being used in recent years. Although many different hardmetals have been tested, there remain open questions regarding the exact mechanisms which occur during corrosion and the influence of microstructural parameters such as interfaces and binder properties.

Tungsten carbide cobalt alloys can be considered to be a galvanic coupling whose corrosion behaviour has been modeled according to a modified version of Stern’s Rule for heterogeneous alloys [90]. The corrosion characteristics of the hardmetal has been shown to be dependent on the corrosion response of the alloy constituents and some of the results will be highlighted in the following sections.

4.4.1 Galvanic corrosion and Stern’s rule

The coupling of two or more metals in a corrosive environment leads to galvanic corrosion when the rate of dissolution of the less is high, while the more noble metal remains protected. This is illustrated in figure 4.13 where two metals P and Q are electrically coupled P Q together. Each metal has a corrosion potential of E corr and E corr. After coupling, the potential of the system moves to the galvanic potential, Eg which lies between the corrosion potentials of the two metals. At this point when no net current flows, the oxidation rate equals the reduction rate and the system is at rest. Metal P oxidizes while metal Q is reduced.

Figure 4.13. Schematic current vs. potential curves for metals P and Q illustrating galvanic effects [90].

The current density for the system is then given by equation 4.3, when the surface areas exposed to corrosion are equal.

P Q iT = IT = Ig + Ig E4.3

4 Literature Review 29

where: iT = total current density of system

IT = total corrosion current of system P Q Ig , Ig = corrosion currents of each metal after coupling

When the surface areas exposed to corrosion are not equal, the current density is then given by equation 4.4 which is known as Stern’s rule, where AP & AQ represent the respective exposed surfaces areas of metals P and Q.

P P Q Q iT = A i + A i E4.4

Stern was the first researcher to examine the relationship between the polarization characteristics of individual metals and a macroscopic couple [91]. As shown by equation 4.4, he found that the corrosion response of an alloy could be predicted by the corrosion response of the alloy constituents. In general, two phase alloy systems have been found to follow Stern’s rule, with deviations being attributed to compositional changes which may occur during coupling.

4.4.2 Corrosion of cobalt

The main oxidation states for cobalt are Co2+ which is the stable oxidation state and Co3+, which is rare. The oxidation reactions are shown in equations 4.5 and 4.6 where the corrosion potential, E° is quoted versus the standard calomel electrode(SCE).

Co = Co2+ + 2e-, E° = -518mv vs. SCE E4.5 Co2+ = Co3+ + e-, E° = 1567mV vs. SCE E4.6

The Pourbaix diagram for cobalt, shown in figure 4.14, indicates that cobalt corrodes actively in acid media and forms oxides at pH levels above 7 only in oxygenated environments [92]. Cobalt oxidation is influenced by the current density, temperature, type and concentration of the electrolyte used. For example, pure cobalt has been shown to have a corrosion potential of – 370mV in a 95-97% H2SO4 solution [90] and a potential of –460mV in a 0.01M H2SO4 + 0.99M Na2SO4 solution [93]. It has been found that pure cobalt exhibits no distinct potential or passive region [90,93]. The crystal structure of cobalt has also seen to affect the corrosion resistance in that, hcp cobalt was found to oxidize at a faster rate than fcc cobalt at high temperatures [94], but at room temperature the curves have been found to be the same [95].

4 Literature Review 30

Figure 4.14. Potential-pH diagram for the cobalt-water system at 25°C [92].

Cobalt-base alloys, most of which are chromium bearing, are resistant to galvanic corrosion because of their noble position in the galvanic series. However, in environments in which their passive film is not stable, they occupy a more active position and can be adversely affected by more noble materials. The fact that the alloys polarize readily tends to reduce their galvanic effects on less noble materials [96].

4.4.3 Corrosion of WC

Tungsten carbide is known to be extremely corrosion resistant due to its noble position in the electrochemical series. The corrosive attack of tungsten carbide has been studied by Voorhies [97] in a 2N H2SO4 solution and by Ghandehari [98] in a 1.2M H3PO4 solution. According to Voorhies [97], tungsten carbide oxidizes according to equation 4.7 and according to Ghandehari [98] tungsten oxidizes according to equation 4.8.

+ - WC + 5H2O = WO3 + Co2 + 10H + 10e E4.7 + - W + H2O = WO4 + 8H + 6e E° = 200mV vs. SCE E4.8

Voorhies [97] also theorized that carbon has a significant effect on the corrosion rate of tungsten carbide since he found appreciable differences between corrosion rates of tungsten metal and tungsten carbide. It has been shown by various researchers that the oxidation of tungsten carbide becomes more significant at potentials greater than 500mV [93,98].

4 Literature Review 31

4.4.4 Corrosion of WC-Co

The corrosion resistance of cemented carbides is principally determined by the corrosion resistance of the binder, which is predominantly cobalt. Tungsten carbide is more noble than cobalt in acidic media and as such corrosion generally progresses by oxidation of the binder phase which is expected to be followed by fall-out of the tungsten carbide grains. A number of studies have shown this effect and the influence of hardmetal corrosion on woodcutting tool wear was reviewed in section 4.1.

The corrosion response of hardmetals has been shown to follow Stern’s rule provided the influence of dissolved tungsten is taken into account [90]. It has been shown that decreasing the amount of tungsten in solution decreases the corrosion resistance of Co-W-C alloys [77,90]. The amount of tungsten in the binder depends on the grain size, cobalt content, carbon content, sintering and cooling conditions. For the reaction WC = W + C there exists an equilibrium reaction (W)(C) = constant [56]. Thus, a low concentration of carbon in the binder will mean the presence of a high concentration of tungsten in the binder and therefore a higher degree of resistance to corrosive attack.

The exact distribution of dissolved tungsten in the binder is an open question and due to this, there is uncertainty regarding the exact mechanism of corrosion taking place in the binder phase. For example Hellsing [99] found a 0.05µm layer of cobalt with a lower tungsten concentration at the carbide-matrix interface. Hence, he postulated that the corrosion rate of the interfaces would be higher than the bulk binder phase. If this tungsten depleted zone exists for all hardmetals, it would imply that hardmetals with smaller grain sizes would have higher corrosion rates since they have more interfaces, but this has never been seen. Tomlinson and Linzell [93] found that hardmetals with a lower concentration of carbon in the binder had a lower critical corrosion current and higher corrosion potential value than those with high carbon in the binder. This was confirmed by Human in a later study [90].

Ghadahari [98] postulated that the corrosion rate of the binder phase would depend on it’s thickness, which is determined by the carbide grain size and the amount of cobalt. Hence, smaller- grained hardmetals with a thinner binder thickness would have lower dissolution rates. It is well established that the corrosion rate increases with increasing cobalt content [1]. This relationship is shown in figure 4.15 taken from work done by Human [90]. Increasing the binder content raises the corrosion current but, it has not been seen to affect the corrosion and passivation potentials. It is not known if grades with the same cobalt content, but different binder mean free paths, hence different carbide grain sizes, would have different corrosion rates.

4 Literature Review 32

Figure 4.15. Critical current density of the binder phase area versus the binder content [90].

There are inconsistent results found in literature, for example, increasing corrosion resistance with increasing grain size [100], decreasing corrosion resistance with increasing grain size [77] and no effect by varying the carbide grain size [90]. Fernandes et al. [101] tried to establish if the mean free path would play a role in the corrosion resistance of hardmetals by using Stern’s rule to hypothesize the corrosion behaviour of the hardmetals due to corrosion at the interfaces and in the bulk Co-binder. The tungsten carbide phase was assumed to be cathodic. Inconclusive results were obtained and they commented that the inconsistent results in literature regarding trends are probably due to differences in the amount of dissolved tungsten, different amounts of dislodged carbide grains during corrosion and varying extrapolations of the icorr value which is used to calculate the corrosion rate.

There have been attempts to increase the corrosion resistance of the binder by additions of more corrosion resistant elements such as chromium, , and . Chromium has been found to be the most successful [100,102]. The corrosion rates is lowered since passive layer formation is promoted. The corrosion potentials were found to shift to more noble values in the presence of chromium, as additions of Cr3C2 were found to significantly decrease the current by several orders of magnitude [103]. Tomlinson and Ayerst [100] found that vanadium has an indifferent effect or reduced the beneficial effect of chromium additions. They concluded that the vanadium reduces the corrosion resistance by possibly being incorporated into the passive film and interfering in the formation of chromium-oxide. Mori et al. [103] found that small additions, >0,5% of and had no influence on the corrosion resistance but additions between 4-8% did have a beneficial effect, although the exact mechanism is not yet fully understood. Enqvist [104] compared the corrosion response of a nickel-cobalt alloy to a cobalt- alloy in alkaline solutions and found that the passive films formed by the nickel are dense, while those formed by the cobalt were porous and thus unable to give protection to the underlying metal surface. Overall, improvements were noted more in metals with a higher binder content, indicating once again that the binder is preferentially attacked.

4.5 Wear

According to DIN 50320 wear is defined as “the progressive loss of material from the surface of a solid body due to mechanical action, i.e. the contact and relative motion against a solid, liquid or 4 Literature Review 33 gaseous conterbody.” It forms part of the wider field known as tribology which encompasses friction, wear and lubrication. Friction leads to the dissipation of energy, wear to the dissipation of surface material and lubrication to the dissipation of load between two surfaces moving relative to each other [105].

Wear processes can often be described in terms of a tribosystem which is shown in figure 4.16. A tribosystem usually consists of a solid body, a counterbody, an interfacial element and the environment. Action and/or interaction between the different parameters in the tribosystem will lead to different types of wear taking place, for example, sliding, rolling, and impact wear.

Figure 4.16. Schematic representation of a tribosystem [105].

In order to differentiate between the types of wear, the mechanisms due to wear are usually described. Four main basic wear mechanisms have been defined and are used in all tribosystems [DIN50320,106]. These are: • adhesion • abrasion • fatigue • corrosive wear

Adhesion involves the interaction of asperities on two opposed surfaces in motion. Fusion of the metals may occur followed by fracture of the asperities. Abrasion is the removal of material from one surface by the harder asperities of another surface. Fatigue occurs due to a cyclic stress state on the metal surface causing mechanical damage to surface and sub-surface regions. Damage accumulation eventually leads to failure via fracture. Corrosive wear is the synergistic interaction between chemical reactions and wear processes occurring simultaneously. Abrasive wear and the influence of corrosive action on abrasive wear are the major themes for this project and will thus be described in further detail.

4.5.1 Abrasive wear

Abrasive wear has been defined as the displacement of material which is caused by the presence of hard particles between or embedded in one or both of two surfaces which are in relative motion, 4 Literature Review 34 or by hard protuberances(asperities) on one or both of the surfaces [107]. Figure 4.17 shows the two main classifications of abrasive wear, namely two-body and three-body abrasive wear. In two- body abrasive wear the abrading surfaces move freely over each other and wear is caused by hard protuberances on one or both surfaces. In three-body wear, the abrading particles are free to roll and slide between two surfaces, acting as interfacial elements [108]. In both cases, three types of material stress situations, namely goughing, high and low stress abrasion can be identified. Goughing occurs when coarse particles impact and cut into the material displacing sizeable amounts of material. In high stress abrasion the crushing strength of the abrasive particles are exceeded, so that they are broken up leading to high localized stresses. Low stress abrasion occurs when particles slide freely over a surface without fracturing.

Figure 4.17. Schematic showing two-body and three-body abrasive wear [105].

The physical interactions which occur between the abrasive particles and the surfaces of the materials are known as micro-mechanisms and are shown in figure 4.18. These are described as microploughing, microcutting, microfatigue and microcracking. In microploughing, a prow is formed ahead of the abrading particle and material is continually displaced sideways forming ridges adjacent to the groove produced. In microcutting, the material loss is equal to the volume of the wear groove which is produced. Microfatigue occurs when the repeated passing of particles caused by simultaneous or successive abrading particles cause ‘ridged’ material to break off. Microcracking occurs when highly concentrated stresses are imposed by the abrasive particles in the material’s surface. Microploughing and microcutting are dominant features of ductile materials while microcutting is more dominant in brittle materials. The transition from one mechanism to another depends on the properties of the material being worn and the environmental conditions. For example, an increase in hardness can result in microploughing changing to microcutting which with a further increase in hardness may change to microcracking [105].

4 Literature Review 35

Figure 4.18. Micro mechanisms due to physical interactions between abrasive particles and metal surfaces [105].

The factors which influence abrasive wear are shown in figure 4.19. The physical and microstructural properties of the alloy are among the most important factors which effect the wear response of an alloy. These properties include hardness, fracture toughness, work hardening, ductility, strain distribution, mechanical instability, crystal anisotropy, inclusions, second phases for hardening like precipitates, the matrix structure, crystal defects and internal notches [105]. Hardness and fracture toughness are the two most commonly used properties to assess an alloy’s ability to withstand wear, since mechanisms of abrasive wear involve both plastic flow and brittle fracture [108]. For plastic flow, the hardness of the binder is more important, while for brittle fracture, fracture toughness is more important. However, at low fracture toughness values, hardness becomes more important.

Figure 4.19. Factors influencing the abrasive wear in a tribosystem [105]. 4 Literature Review 36

Hardness is related to the yield and flow properties of the metal. Abrasive wear resistance increases with increasing hardness of a material to a first approximation. However, this is not always the case. For example, as the hardness of grey increases, the wear resistance increases up to a critical hardness value after which the wear resistance decreases. It has been suggested that it may be better to correlate wear to the hardness of worn samples instead of unworn samples for two reasons. Firstly, during abrasion, the surface tends to plastically deform, which can lead to high work hardening of the surface and secondly materials may have the same bulk hardness, but exhibit different wear values. Work-hardening depends on the mode of deformation and the microstructure of the metal. For example it has been shown that amorphous metals have lower work-hardening rates [105].

Fracture toughness describes a metal’s resistance to crack propagation which is dependent on the energy of plastic deformation and to a crude approximation it decreases with increasing hardness. The influence of fracture toughness can be seen when the mechanism of microcracking starts to occur. Cracking depends on the ductility of the alloy which influences the transition from ductile wear processes such as ploughing and cutting to cracking. This in turn is tribo-system dependent and sensitive to the type of loading applied.

The hardness, shape and size of the abrading particle are known to influence wear rates. Generally the abrasives are divided into hard and soft abrasives. Hard abrasives are defined as having a hardness which is approximately 50% greater than the hardness of the unworn material, whereas soft abrasives are not able to indent a surface. The wear rate has been seen to be sensitive to the ratio of abrasive hardness to alloy surface hardness, where higher wear rates are measured when the particle hardness is more than 1.2 times that of the alloy surface [108]. Wear has also been shown to increase with increasing particle angularity and smaller particles cause less damage than larger ones.

4.5.2 Lubrication

In most wear environments, a certain amount of moisture is always present. Atmospheric composition such as oxygen content and humidity can lead to the formation of oxides on the metal surfaces, which, although having low shear strength, reduces the contact between abrasive particle and metal. Lubricants are generally used in industry to reduce wear and friction, however, they can also accelerate wear processes. Hence, the influence of a liquid in a wear system is also significant. Friction is generally defined as the resistance encountered by one body in moving over another [108]. Friction has been shown to be proportional to the normal load, independent of contact area and independent of sliding velocity. The latter being broken down into static and dynamic friction since a certain amount of force is required to initiate movement. In the absence of lubricants, high frictional forces can occur which may lead to high energy losses.

The Stribeck curve, shown schematically in figure 4.20, can be used to describe the different types of friction-lubrication patterns experienced by two surfaces in contact [109]. It shows the dependency of this relationship on the surface roughness, the viscosity of the lubricant, the speed at which the two surfaces move in relation to each other, the pressure and the lubricant film thickness. The viscosity of the film is important and measures the resistance of the fluid to shearing flow and can be defined as being the shear stress on a plane within a fluid, per unit velocity gradient normal to that plane [108]. Four different types of friction can be identified, namely boundary friction(1), mixed friction(2), elastrohydrodynamic friction(3) and hydrodynamic friction(4).

4 Literature Review 37

Film thickness - h

Friction coefficient - µ

Viscosity - η x speed - v Pressure - p

Figure 4.20. Stribeck curve [109].

In boundary friction, the surfaces are separated by a molecular thick film. Appreciable asperity contact occurs since the bulk pressure is carried by the asperities of the two surfaces making contact. In mixed friction, the pressure is shared between the film and the surfaces in contact. In elastohydrodynamic friction, elastic deformation of surfaces occur and due to the increasing pressure, the viscosity increases. In this case the film is very thin. In hydrodynamic lubrication, the surfaces are separated by a thick film where the thickness is greater than the combined surface roughness and there is no mutual asperity contact. A small amount of surface deformation is caused due to the hydrostatic pressure in the film since the film can be thought of as being rigid. Wear is minimized in this case.

4.5.3 Abrasive wear of WC-Co alloys

The abrasion resistance of cemented carbides depends on the material’s ability to resist penetration by abrasive particles and the level of difficulty in removal of material by fracture or plastic flow. In section 4.5.1, these properties have been shown to be determined mainly by the hardness and fracture toughness of the alloys. The relative wear ranking of hardmetals is tribosystem dependent and is influenced by factors such as size and nature of the abrasive used, the abrading wheel speed, the applied loading and to a large extent on the varying material properties. For example ductile materials have been shown to have a good correlation between hardness and abrasion resistance, while for more brittle materials local fracture toughness is more important [110]. The microstructure of hardmetals has been shown to play a decisive role in determining the fracture mode and wear resistance [59,84,89,111].

Cemented carbides have a good wear resistance especially to repeated localized deformation. Some of the main contributing factors are that with finer carbide grain sizes the microfracture strength is increased due to the strong interphase strength which is achieved [112]. The cobalt binder is also known to have excellent deformation characteristics and it bonds with tungsten carbide with optimal strength which is better than, for example, bonding tungsten carbide with iron or nickel [111]. During deformation, the cobalt often undergoes the transition from fcc to hcp leading to a higher stacking fault energy state which is more wear resistant. Gant and Gee [113] found that hardmetals with high cobalt contents(20-25%Co) have faster wear rates than those with low cobalt contents(6-9%Co). The wear rate of high cobalt alloys is governed by ploughing through 4 Literature Review 38 the binder phase together with fracture and possible re-embedment of carbide grains. In low cobalt alloys, wear is governed by wear of the hard phase and extrusion of the binder.

The wear resistance of hardmetals has been modeled by Enqvist [104], where he showed that the wear resistance of a composite can be determined from the load distribution on its phases and their individual wear resistances. The load distribution depends on the tribosystem characteristics such as thermal effects, loading conditions and properties of the phases. This leads to two extremes of wear that can occur, namely, the optimum wear achieved when phases are worn down in parallel(OL) and the second extreme is when each phase wears independently of the other at the minimum load(ML). The model is shown in figure 4.21. In the OL mode, the hard phase carries the maximum share of load and thus contributes optimally to wear resistance. In the ML mode, the hard phase has a minimal contribution and the phases carry load in relation to their area fractions only. This model is valid only when the wear resistance of the bulk composite is preserved in the composite.

1

Figure 4.21. Schematic illustration of the wear resistance which shows the different types of wear mechanisms which may occur in a composite [106].

It has been shown that the wear resistance of cemented carbides increase with decreasing carbide grain size, decreasing cobalt content, higher carbide volume fractions and smaller binder mean free paths [63,84,85,104,114,115]. Trying to find the optimum combination of these factors to provide excellent wear resistance is the topic of on-going research.

The wear response of fine-grained hardmetals often differs to that shown by coarser grained hardmetals. A fine-grained microstructure offers a higher resistance to microcracking than a coarse-grained structure of the same hardness. At a constant cobalt content, finer carbides have a higher hardness and microfracture strength [71,89,116,117]. As carbide grains become smaller, their individual crack resistance becomes greater [89]. O’Quigley et al. [71] studied a range of hardmetals and found a grain size dependency on the wear between finer-grained and coarser- grained hardmetals. Up to 1000HV no dependency of abrasion on grain size was found. All grades showed the conventional linear relationship between abrasion and hardness. Between 1000 and 1600HV the coarser grades had better abrasion resistance, and greater than 1600HV the finer grained hardmetals had a better abrasion resistance. The better resistance between 1000-1600HV 4 Literature Review 39 was ascribed to coarser grades being tougher than finer grades up to a critical hardness which is grain size dependent. Above this critical hardness the finer grades are tougher. In contrast, Cuddon and Allen [118] found that as the hardness of carbides decreases from around 1600HV the wear rate remains approximately constant until a hardness of 1300HV where it increases rapidly with a continued decrease in hardness. This shows that this relationship is also dependent on the tribosystem environment and cannot easily be generalized.

Jia [89] found that the abrasion resistance of nano-grained hardmetals are less dependent on hardness than conventional samples. They may be better related to the binder mean free path as has been shown by Herr [84]. During the study of ultrafine grained hardmetals, Herr [84] found a better correlation between binder mean free path and wear resistance than with hardness. A low to high wear correlation was found as depicted in figure 4.22. A binder mean free path less than 40nm would be in the low wear region. This implies that by a further reduction of the mean free path either by smaller carbide grains or more carbide volume, no further significant increase in wear resistance can be achieved.

Low wear Transition zone

Wear resistance [m/mm3]

High wear

Figure 4.22. Dependence of wear resistance on the binder mean free path [84].

Various researchers have shown that small binder mean free paths give higher abrasion resistance and when combined with smaller carbide grains, give higher hardness values [71,84,104]. Increasing the mean free path or the carbide grain size causes increasing wear loss, caused by plastic deformation and uprooting and microcracking of interfaces or carbides [105]. Abrasion resistance is more sensitive to changes in the grain size at low mean free paths than at higher ones [71]. Dependence of the wear loss on the mean free path or the carbide grain size decreases with increasing abrasive hardness [105].

Binderless carbides have been finding more applications in wear resistant parts due to the excellent hardness achievable and, as already mentioned, an increase in the wear resistance can be achieved by increasing the carbide volume fraction. Binderless carbides have a higher strength since a reduced mean free path results in an increase in the plastic constraint at the grain boundary interfaces thereby strengthening the alloy [118]. The abrasion resistance of binderless carbides has been studied by Enqvist [104] using diamond scratch tests. He found that due to the anisotropic behaviour of single tungsten carbide crystals, different wear mechanisms can be achieved depending on the crystallographic orientation, from brittle on the basal plane to ductile on the prismatic plane. Good wear resistance can thus be achieved by correct orientation of carbide 4 Literature Review 40 planes to the surface. Increasing the carbide volume fraction however, means decreasing the binder phase which can lead to a reduction in the toughness. It has also been suggested that strengthening of the binder phase may be a better solution since it would be able to maintain durability characteristics, such as plasticity, toughness and strength. Iron and nickel have often been used to strengthen the cobalt phase and in some cases a better wear resistance is achieved [84,117]. Increasing the strength of the binder phase by means of increasing the dissolved tungsten content in the binder can also lead to an increased wear resistance [119].

The wear mechanisms in hardmetals vary with service conditions, composition and microstructure of the alloys and with the hardness of the abrasives [117]. The main mechanisms are brittle fracture, fatigue and plastic deformation. Each major mechanism may act at a different level; • macroscopic level: affecting a major part of the component, • microscopic level: affecting a substantial number of grains, • finer level: effecting either individual carbide grains or sections of binder.

The wear rate is strongly influenced by the hardness of the abrasive relative to the hardness of the metal alloy and this leads to wear being characterized in terms of low to high wear [71,105,117]. This transitional relationship has been shown to be carbide grain size dependent with finer grades exhibiting this change at a higher hardness than the coarser grades [71]. Relatively hard abrasives cause material loss by grooving and crater formation by gross plastic deformation. In addition, surface layers are weakened due to the gradual extrusion of the binder with resulting fracturing of carbide grains.

Relatively soft abrasives, i.e. when the hardness of the abrasive is lower than the hardness of the surface being abraded, cause more localized wear. They cannot easily indent the surface and thus slide over the surface with considerable frictional forces. This causes cyclic loading of the carbide grains which leads to gradual extrusion of the binder phase. Surface layers need to be exposed to several attacks before a significant amount of material is removed. In this mechanism of binder removal, the residual compressive stresses in the carbide grains, due to cooling, are gradually relaxed causing the grains to fragment. Extruded cobalt also smears across the surface and can reduce the frictional forces. Very soft abrasion can also occur where the carbide grains have a polished finish. Carbide grains tend to be supported by three or four of their six sides and are eventually torn loose by frictional forces.

The size, shape and hardness of the abrasive particles also influence the effect which the microstructural parameters have on the abrasive wear loss [105]. For example, the rate of abrasive wear loss increases more rapidly for softer abrasives than harder abrasives as the cobalt volume increases [105]. Abrasion resistance has been shown to be more dependent on decreasing carbide size than lowering the cobalt content due to the higher fracture strength of smaller grains when using diamond as an abrasive [89]. Abrasion, using , showed less dependence on carbide grain size and a better hardness dependence than diamond abrasives. Larsen-Base [120] compared silicon carbide to quartz and found that the wear by quartz was approximately one tenth that of the silicon carbide.

In diamond scratch testing, Jia [89] found that in conventional carbides the wear mechanism proceeded by extrusion of the material, both carbide and binder, fragmentation of the carbide grains and then fracture within the binder phase or at the interface between carbide and binder. For ultrafine-grained hardmetals, Herr [84] found differing dominant wear mechanisms which were 4 Literature Review 41 dependent on the amount of binder mean free path and speculated that in the high wear region(figure 4.22) microploughing and microfatigue are the dominant mechanisms. In the transition zone(figure 4.22) from high to low wear, wear is controlled by the plastic deformation ability of the surface. In the low wear region(figure 4.22), wear proceeds by microcracking due to the increasing brittleness of the surfaces.

4.5.4 Corrosive-wear

Corrosive wear occurs when two rubbing surfaces interact within a corrosive environment. The relative importance of the wear-corrosion synergism depends on the electrolyte, the alloy and the tribological conditions. Due to tribochemistry, surface films may form and these can either improve or impair the tribo-properties of the alloy. They can provide protection against wear or when they adhere loosely to the surface, they can be easily removed [108]. In non-passivating alloys, the synergism may be high due to surface roughening by corrosion, abrasion by corrosion products and corrosion assisted removal of hard phases [121]. According to Zum Gahr [105], oxidation of metallic wear debris can occur even in the absence of a corrosive environment. Non-metallic wear debris may also form. These mechanisms can occur repeatedly, causing extensive material loss.

In order to develop methods to reduce corrosive-wear, a fundamental understanding of the mechanisms involved, their interaction and their significance in a particular environment is needed. An effective way to achieve this is to study the effects of wear and corrosion separately and then the degradation process due to the combined effect can be more easily and clearly identified. Many researchers have studied the corrosive-wear synergism although most of the research has focused on two-body abrasive-corrosive wear or erosive-wear of ferrous metals. To the author’s knowledge, extremely little research has been done in the field of corrosive-wear of tungsten carbide hardmetals. A few highlights from ferrous metal corrosive-wear research will be given below followed by the corrosive-wear research in hardmetals in the next section.

Nöel and [122] and later Barker and Ball [123] studied the synergistic effect of a variety of steels in a wear-corrosion set-up to simulate a mining environment and found that the abrasion accelerates material degradation by removing corrosion products or surface films. The effects were studied as a function of abrasive load, corrosion time and frequency of abrasive and corrosive treatments. An increase in load gave only a slight increase in the volume loss due to corrosion, but due to abrasion the increase was far more. The test results imply that when abrasive wear occurs continuously under high loads in a corrosive medium, the percentage of the corrosion effect is negligible. For low loading and infrequent abrasion the corrosion effect can be high. They found that the greater the frequency of corrosion-abrasion for a set period of corrosion time and abrasion distance, the greater the volume loss. An initial high rate of corrosion occurs due to increased internal energy from high dislocation densities in the worn surface region. The corrosion rate decreases as the deformed layer is removed by further wearing. High volume losses were recorded initially and were ascribed to a high corrosive attack of freshly worn surfaces.

Batchelor and Stachowiak [121] found that the corrosion-wear synergism can be four times the corrosion rate when the abrasion rate is negligible. Kim et al. [124] showed that wear shifted the corrosion potential in the more active direction thereby increasing the corrosive effect. In corrosive- wear experiments they found that the corrosion current density increases by about two orders of magnitude from a non-wear to a wear condition. Kotlyar et al. [125] when studying steels, found a decrease in the total wear rate for increasing ratios of the hardness of the sample to the hardness of the abrasive. The magnitude of the synergistic wear-corrosion component decreased with 4 Literature Review 42 increasing ratios. The total abrasive-corrosive wear increased with applied load, increasing hardness of abrasive and decreasing pH.

Madsen [126] developed a set of penetration equations which can be used to quantify the corrosion-wear data in slurry and sliding wear. The equations can be used to calculate the standard penetration rates due to corrosion only, wear only and due to the combined effect. Tests need to be run on specified test apparatus which are able to accommodate the type of measurements needed. The wear-corrosion degradation(T), wear alone (W) and corrosion alone(C) degradations are given by the equations 4.9 to 4.11 The total wear-corrosion synergism is then defined by equation 4.12 which represents the difference between total degradation and the individual wear and corrosion components.

∆mwocathTF T = E4.9 SAwocath D∆hwocath

where: ∆mwocath = the mass loss without

SAwocath = the surface area without cathodic protection D = density

∆hwocath = the experiment time duration without cathodic protection TF = the time factor to convert hours to years

∆mcathTF E4.10 Wo = SAcath D∆hcath

where: ∆mwocath = the mass loss with cathodic protection

SAwocath = the surface area with cathodic protection D = density

∆hwocath = the experiment time duration with cathodic protection TF = the time factor to convert hours to years

i CRCF(EW ) C = corr E4.11 o D

where: icorr = corrosion current density CRCF = corrosion rate conversion factor EW = equivalent mass D= density

E4.12 S = T −Wo − Co

Using these equations. three dimensionless factors were then defined to assess the degree of damage due to each process, namely a wear-corrosion factor, a wear factor and a corrosion factor. 4 Literature Review 43

4.5.5 Corrosive-wear of WC-Co alloys

The corrosive-wear research carried out to date on hardmetals has focused on erosive wear and sliding wear in corrosive environments. Some research has been conducted on hardmetal coatings, but extremely little on the hardmetal alloys themselves. The available literature has been reviewed and is presented in this section.

Early loss of the carbide phase due to corrosive attack of the binder phase leading to more rapid wear was found by Fink et al. [127] when studying the electrochemical aspects of WC-Co wear. They showed that due to cobalt loss, ‘grain plucking’ or gross tungsten carbide matrix fracture occurred. A decrease in the hardness on the sides of the bits after drilling into limestone was found and this was attributed to preferential oxidation and dissolution of cobalt. This removal of the cobalt was confirmed by EDX analysis which showed that it is surface effect.

Tomlinson and Molyneux [128] investigated the erosion-corrosion response of tungsten carbide alloys in a pH 1 solution of sulphuric acid and used alumina balls to erode the surface. They found that the corrosion rate increased by 16 to 29 times in this set-up, compared to standard corrosion tests without the wear component. The surface layer was denuded of cobalt to a depth of about 10µm and the flexural strength of the hardmetals was decreased by 0.54 to 0.76 times. The strength of pre-corroded specimen was not affected.

Wentzel and Allen [129] investigated the erosion-corrosion response of a variety of hardmetals using tapwater and saltwater and found that although passivation behaviour occurs during pure corrosion testing, these same metals show a lower erosion resistance in saltwater than tapwater, since the films are broken down during erosion. The erodent caused a rise in the corrosion potential increasing the corrosion rate and hence they postulated that corrosion of the binder phase is rate controlling during the erosion process and thus the more binder present, the longer it will take to be removed to the extent that tungsten carbide grain pullout is possible. They further found that under corrosive-wear conditions hardness, deformation characteristics and phase transformations of the binder phase are more important.

In the polishing of tungsten using chemo-mechanical polishing in a low pH slurry, Larsen-Basse and Liang [130] found that the metal passivated, thus slowing down the corrosion rate. Limited penetration of the film occurred by the abrading particles present. When the film was penetrated, it occurred at very small points, after which the film would either reform or small pits were formed.

Cuddon and Allen [118] investigated the abrasion resistance of a range of carbide alloys(11- 31vol% cobalt binder and 1.5-8µm grain size) in coal ash conditioners. Microstructural investigations showed preferential removal of the binder for all grades which was attributed to the synergistic effect of corrosion and abrasion when the wet abrasive coal ash moves along the blades. Smearing of the surfaces during abrasion limits corrosive attack since the carbide grains are unexposed and these are needed to sustain the electrochemical reaction. With low binder contents and small grain sizes the wear resistance is high and controlled by the corrosion rate of the binder phase. As the mean free path increases with increasing grain size and cobalt content, abrasive wear plays a dominant role by increasing the binder phase wear rate too. At higher binder contents the role of corrosion on the overall wear rate is minimal and wear resistance is controlled by the work-hardening capacity of the cobalt during abrasion. The rate of corrosive attack was found to depend on pH, electrolyte temperature and constitution. Temperatures of 70°C were 4 Literature Review 44 reached in the experiments and the pH was as low as 1.7. The electrolyte contained sulphates and chlorides which are known to attack cobalt.

Enqvist [104] investigated the sliding wear of various hardmetal alloys (WC-Co, WC-NiCo, binderless WC) in lubricated conditions where the pH was varied. For Co-alloys, the wear rate increased with increasing pH, while the NiCo-alloys were more stable. The binderless alloys showed the strongest dependence on pH, where the wear rate increased by 20 times at higher pH values. The reduced ability of the binderless alloys to form small fragments of tungsten carbide at higher pH values led to higher wear rates and higher temperatures due to continuous contact. At low pH, the alloys have different corrosion responses but similar wear rates. This is due to the high load which was used, which reduces the influence of corrosion on the wear rate. The Co-alloys form varying oxides at higher pH values whereas the NiCo-alloys do not form any. Friction coefficients were found to be independent of pH except for the Co-alloys where it dropped at higher pH values. 5 Experimental Methods 45

5 Experimental Methods

In this chapter, details of the experimental methods used, in investigating the wear behaviour of three tungsten carbide hard metals during woodcutting in which the synergistic action of corrosion plays an important role, are given. In addition, various experimental techniques have been used to determine the microstructure properties that not only quantify each alloy but also influence their wear and corrosion response. This chapter is divided into four main sections namely; • material characterization, • woodcutting tests, • corrosion tests and • three body abrasive wear tests.

5.1 Material Characterization

This section describes the hardmetals used in the project, the quality control methods used to ascertain their ‘fitness for use’, the microstructural analysis used and the mechanical properties measured which have a direct influence on the wear and corrosion of each hardmetal grade.

5.1.1 Investigated hardmetals

The three cemented carbides used in this project were produced using conventional powder metallurgical techniques, which are described elsewhere [84]. The nominal compositions of the grades are given in table 5.1 and typical microstructures for each grade are shown in figure 5.1. The specimen designation shown in column one of table 5.1 was chosen to be descriptive of the grain size distribution and the cobalt content, namely, UFW – ultrafine WC grain distribution FW - fine WC grain distribution The numbers in each description represent the cobalt content.

For example, UFW3 represents an hardmetal with an ultrafine tungsten carbide grain size distribution having a cobalt content of 3wt%.

Table 5.1. Hardmetals investigated in this study.

Hardmetal grade Cobalt content – wt% Inhibitors used UFW3 3 UFW10 10 Cr2C3 and VC FW15 15

5 Experimental Methods 46

UFW3 UFW10

FW15

Figure 5.1. Scanning electron micrographs showing the microstructure distribution of the hardmetal grades.

5.1.2 Hardmetal quality evaluation

Samples were selected from each grade to assess their ‘fitness for use’. The specimen were ground and polished to a 1µm finish, ultrasonically cleaned in alcohol, dried and examined using optical microscopya according to the appropriate DIN standards. The features examined are described in the following sections.

5.1.2.1 Porosity and free carbon

Pores can act as fracture initiation points and stress raisers. Free carbon, also known as carbon porosity or graphite, forms in the presence of excess carbon. DIN ISO 4505 was used to assess the porosity and free carbon with respect to presence, type and distribution. Optical electron examination of the polished specimen surfaces showed the free carbon levels of the grades to be

5.1.2.2 Cobalt pooling

Cobalt pooling arises due to insufficient or inhomogeneous mixing during milling. This causes a poor cobalt/grain growth inhibitor distribution, as well as partially de-agglomerated tungsten carbide powder in the formulated composition. Long milling times can also lead to the formation of cobalt flecks. Cobalt pooling when present in large amounts can lower the hardness of a hardmetal, giving rise to a lower wear resistance. It may also increase the corrosion rate, by providing increased cobalt surface areas for corrosive attack. The polished specimens were examined using optical light microscopy at a various magnifications to detect the presence and extent of cobalt pooling.

5.1.2.3 Eta Phase

Eta phase is formed during sintering, when there is a carbon deficiency in the metal. Due to the shortage of carbon, tungsten will combine with some of the cobalt during the sintering cycle and will form this eta phase where it’s formation is a nucleation and growth process. Depending on the morphology of the eta phase present it may cause embrittlement of the material and can lower its fracture toughness, which results in a reduction of the effective contribution of the tungsten carbide to the strength of the composite.

The polished samples were etched for 30 seconds using Murakami’s etchant described in table 5.2. The time is sufficient for the eta phase to be clearly visible, if present, in the form of black patches. Optical light microscopy at a magnification of 200 times was used to detect the possibility of its presence.

Table 5.2. Composition of Murakami’s etchant.

100ml Distilled water 5-10g Potassium hydroxide 5-10g Potassiumferrocyanide

5.1.2.4 Discontinuous grain growth

Discontinuous grain growth is also known as non-uniform grain growth which manifests itself as exaggerated carbide crystal growth in comparison to the rest of the microstructure. This can lead to a lowering of the hardmetal hardness. The distribution of the types of grain growth can be described as either isolated, single crystals or clusters of many crystals.

5.1.3 Density

Density is a measure of the mass per unit volume with the unit of density being grams per centimeter cubed (g/cm3). The determination was carried out on each specimen using the Archimedes submersion method outlined in DIN ISO 3369. Density can also be determined theoretically using the densities of the alloy constituents. One of the essential differences between theoretical and measured values is that the theoretical determination does not take into account the presence of porosity or free carbon, which decreases the density, nor the presence of eta phase which causes the density to increase.

The actual cobalt content of each hardmetal can be calculated from these density measurements. This value for the cobalt content usually differs from the nominal composition due to factors such as the sintering conditions. It is assumed that the volume of the hardmetal sample(VS) consists of the volume of the cobalt binder(VCo) and the volume of the tungsten carbide(VWC) phase according to the relation shown in equation 5.1. 5 Experimental Methods 48

VS = VCo +VWC = 1 5.1

Using the relation that density equals the sample mass divided by the sample volume, the VCo can be determined from equation 5.2, which is then used to determine the wt%Co according to equation 5.3 developed by Underwood [131].

ρ − ρ E5.2 V = S WC Co ρ − ρ Co WC

100 E5.3 X = V ⋅8.9 ⋅ Co Co ρ S

5.1.4. Magnetic saturation

Magnetic saturation was measured according to DIN ISO 3326 using a saturation magnet systemb. Magnetic saturation is a measure of the magnetic permeability of the metal, which depends on the ferromagnetic cobalt phase. Magnetic saturation increases with increasing cobalt volume content. Eta phase results in a reduction of the measured magnetic saturation value. The units of measurement are Gauss centimeter cubed per gram (Gcm3/g). The magnetic saturation values can be used to determine the amount of dissolved tungsten in the binder phase. Dissolved tungsten has been shown to influence the corrosion response of hardmetals [90].

5.1.5 Coercivity

Coercivity is a measure of the force required to completely demagnetize a magnetically saturated material. The thinner the binder mean free path, the greater the coercive force needs to be in order to influence the magnetic domains [133]. Coercivity is also used as a measure of the stress experienced by the cobalt following sintering. The coercivity was measured according to DIN ISO 3326 using a Koerzimatc. The unit of measure is the Oesterds (Oe) while the SI unit commonly used is the kA/m which is calculated from the relationship shown in equation 5.4.

1Oe = 4πkA/ m E5.4

5.1.6 Hardness and fracture toughness

5.1.6.1 Preparation of specimen surface for hardness and toughness measurements

It has been shown that on cooling down from the sintering temperature, microscopic residual stresses are present in the hard metal [134]. This occurs as a result of the differences in the co- efficient of between the cobalt and the tungsten carbide phases. The thermal co-efficient of expansion of cobalt is three times that of tungsten carbide. This causes the cobalt binder to be in a state of tension and the tungsten carbide regions go into compression, thus setting up stress gradients in the hard metal. Exner [134] found that the residual tensile stresses present in the cobalt after being cooled down from the sintering temperature added to the effective hardness indenting load and facilitated crack b Koerzimat CS 1.096, Förster GmbH c Koerzimat CS 1.096, Förster GmbH 5 Experimental Methods 49 propagation. He found that polishing and annealing the hardmetals restored their stress pattern to the original one produced exclusively by the residual thermal stresses. It was shown that the stress state of an annealed or polished surface, from x-ray measurements, is similar to the stress state as it would be in fracture, which is believed to be the best approximation of an undeformed surface.

The main requirement for the surface preparation for hardness and fracture toughness measurements is that it should be prepared in such a way to obtain the maximum crack length at the corners of the Vickers hardness indent. This is determined exclusively by the indenting load and the properties of the alloy such as the residual thermal stresses, alloy composition and phase structure. Based on Exner’s theory a polishing procedure was composed and the hardness and toughness measurements were determined as described below.

5.1.6.2 Hardness

The Vickers hardness of the hardmetals was measured according to DIN ISO 3878 using a macrohardness indentord. The diagonals produced by pyramidal indentation of the hardmetal surface was used to calculate the Vickers hardness according to equation 5.5, where F is the indenting load, 30kg in this case, and dD is the arithmetic mean diagonal length.

The indentation diagonal measurements were performed using an optical light microscope at a magnification of 200X using a digital camerae and image analysis softwaref. An average of three indentations were measured on each sample.

2F.sin(136° / 2) E5.5 HV = 0.102 ⋅ d 2 D 5.1.6.3 Fracture toughness

The fracture toughness was determined using equation 5.6 which was developed by Shetty et al. [135] and based on the Palmqvist method [136], which uses the lengths of the cracks produced during Vickers hardness indentation as a measure of the hardmetal’s toughness.

1 (HV.F)1/ 2 K = . E5.6 Ic 2 1/ 2 1/ 2 1/ 3 1/ 2 3(1− v )π (2 π tanψ ) (4ar )

In the equation ν is Poisson’s ratio = 0,22 for hardmetals, ψ = 68° half the indenting diamond angle, HV – Vickers hardness, F – indenting load and ar is the average crack length given by the sum of the crack lengths from the corners of the indentation divided by 4. An average of three fracture toughness measurements were calculated for each sample.

5.1.7 Surface roughness

The surface roughness of the hardmetals was determined before and after wear testing. It is known, that as a metal is worn the surface roughness increases, thereby increasing the contact area between the abrasive and the metal surface.

The surface roughness of the hardmetals was determined using a Perthometer,g according to DIN 4768, and was characterized using the RZ value. RZ is the average value obtained from five d V100-A, Leco e Prog/res/3008, Kontron Elektronik f Image C, version 2.50, Imtronic GmbH g SP6, Perthen 5 Experimental Methods 50 successive roughness measurements made across a specimen as depicted in figure 5.2. Five measurements, with Im = 5mm per worn surface were made in the direction perpendicular to the wear direction. Roughness measurements were also made on unworn surfaces.

Figure 5.2. Determination of surface roughness value RZ.

5.1.8 Quantitative microstructure analysis

Quantitative analysis was used to determine the tungsten carbide grain size and the cobalt binder mean free path. The microstructural analysis was carried out using a linear intercept method. Micrographs, representative of the hardmetal microstructure were taken at a magnification of 14 000X using field emission scanning electron microscopyh. A computer program and a digitized boardi were used to carry out the stereological analysis. For each hardmetal, approximately 500 grains were measured to obtain a statistical distribution. This method was used to measure intercept lengths for the carbide(lwc) and binder(lb) phases. These values were then placed in a histogram according to representative classes. Using this histogram the following parameters were determined: • distribution of the carbide grain size and the arithmetic mean value Lwc • distribution of the binder mean free path length and the arithmetic mean Lb.

5.2 Woodcutting tests

Woodcutting tests were conducted in order to compare the wear response together with the dominant damage mechanisms and possible corrosion-wear interactions of the two ultrafine- grained tungsten carbide-cobalt hardmetal woodcutting knives with respect to the type of wood cut. Another objective was to investigate the incipient wear stage after cutting a few meters of wood. This wear response can then be compared to the steady wear state, which occurs during the cutting of large volumes of wood.

h Leica Stereoscan 360FE i Digitizer II, Wacom 5 Experimental Methods 51

The hardmetals were supplied in the form of industrial woodcutting knives. Their geometry is shown in figure 5.3 where the knife is 30mm long, 12mm wide and 1.5mm thick with an included angle of 35°.

cutting edge

12mm

1.5mm 30mm

Figure 5.3. Schematic of woodcutting knife geometry.

Four different types of wood were cut, namely, fresh Spruce, fresh Oak, cured Spruce and chipboard. These four wood types are representative of the three wood classifications namely, fresh wood, cured wood and synthetic wood products. Oak is a harder and a more acidic wood species than Spruce so theoretically the two wood types should display different wear characteristics. Fresh and cured woods were selected in order to observe the effect of the wood extractives on the wear rates and mechanisms. Chipboard was included to note the influence of cutting a completely dry wood and to assess if the wood composition would have any corrosive influence at all.

The woodcutting tests were carried out in the woodworking department at the Friedrich Alexander University in Erlangen, using a milling machine in which the wood feed rate and the depth of cut were kept constant at 12m/min and 1.2mm respectively. For each test, two knives are clamped into a 125mm diameter cutterhead as shown in figure 5.4. During the test, the cutterhead rotates in a anti-clockwise direction, while the wood is fed through the machine from right to left.

knife

cutterhead

wood direction of wood feed

Figure 5.4. Schematic showing woodcutting procedure.

Two different cutting procedures were chosen in line with the objectives mentioned in chapter 3. One procedure involved altering the length of wood cut and the second procedure varied the cutting speed. Thus, on one set of knives three different lengths of wood, 1, 2 and 6m, were cut using a constant cutterhead speed of 58.9m/s. On a second set of knives three different cutterhead speeds, 29.5, 39.3, and 58.9m/s, were used to cut a constant wood length of 4m. The cutting speeds represent those used daily by the woodcutting practice. The experimental matrix is summarized in Table 5.3.

5 Experimental Methods 52

Table 5.3. Summary of the woodcutting experiments conducted.

Types of wood cut Fresh Oak Fresh Spruce Cured Spruce Chipboard Lengths of wood cut using a 1m constant speed of 2m 58.9m/s 6m Speeds used to cut 4m of 29.5m/s wood 39.3m/s 59.9m/s

After the cutting tests, the knives were cleaned ultrasonically and then placed in a specially designed electron microscope sample holder which enabled direct examination of the worn cutting edge as shown in figure 5.5(a). Figure 5.5(b) is a schematic showing the typical form of the knife after the cutting test. The scanning electron micrographs in figures 5.5(c) and 5.5(d) show the cutting edge of a UFW10 grade before and after cutting 6m of chipboard. The worn cutting zone is clearly identifiable. The previously sharp cutting edge has blunted after cutting only 6m of chipboard. From micrographs such as these, the blunting of the cutting edge was determined by measuring the width, wW, shown in figure 5.5(d).

5.5 (a)

5.5 (b)

Worn cutting edge

Original cutting edge

wW

(c) New UFW10 knife in unworn state (d) UFW10 knife after cutting 6m of chipboard

Figure 5.5. Examination of worn cutting edges using scanning electron microscopy. 5 Experimental Methods 53

For each knife, under each cutting condition, approximately 250 measurements of wW was made from which an average blunting value was determined. Using this value, the average volume loss of each knife under each cutting condition was determined using equation 5.7 which was calculated using the geometry of the knife.

E5.7 w + w 1 O W   VL = t ⋅  ⋅ ()wW − wO  2 tan 35°  where: t = thickness of wood being cut: 8mm for this study wO= width of unworn knife wW= width of worn knife 35° represents the included angle of the knife

A wear rate for each knife was defined as the volume loss per meter of wood cut. The tests described above represent the incipient wear stage.

The steady wear stage was evaluated by examining knives which were subjected to woodcutting tests in which 3048m of cured oak was cut using a speed of 30.52m/s. These tests were conducted by the hardmetal company who supplied the hardmetals for this project. The four knives used and their compositions are given in table 5.4. The wear rates of the knives were evaluated according to the procedure described above.

Table 5.4. Hardmetal grades used to cut 3048m of cured Oak.

Grade Binder composition Vickers Hardness Fracture Toughness wt.% HV30 MNm-3/2 UFW10 10 (Co) 1729 9.87 #46j 10 (Co) 2038 8.63 #52 10 (FeCoNi) 1760 8.28 #59 17,5 (FeCoNi) 1650 8.22

Microscopic examination was performed on the worn cutting edges to ascertain the degradation mechanisms and to classify the general wear profiles.

5.3 Corrosion tests

In order to characterize the influence of corrosion on the wear behaviour of the hardmetals it is important to evaluate their corrosion response as a separate process. The understanding gained from corrosion testing can then be used to interpret the corrosive-wear results more intensively.

A standardized potentiodynamic polarization procedure, based on DIN ISO 50 918, was used in which a three electrode corrosion cell is connected to a potentiostat as shown schematically in figure 5.6.

j different tungsten carbide powder manufacturer to the UFW10 grade 5 Experimental Methods 54

A Potentiostat PC V

reference electrode

salt bridge

counter electrode working electrode

Farady cage

Figure 5.6. Schematic of the experimental set-up used for the corrosion testing.

The reference electrode was a -silver chloride electrode connected via a luggin capillary, and the counter electrode was a (Pt) wire gauze. The open beaker cell contained 350ml of solution, which was replaced for each test. Immediately after immersion, the specimen was allowed to stabilize for 10 minutes before commencing testing.

The polarization scan was started at a potential of 500mV more negative to the corrosion potential with a scan rate of 10mV/s. Measurement was performed until +8000mV. Current and potential data were recorded using a potentiostatk and an electronic data measurement system connected to a computer. From the current and potential measurements a polarization scan of logIcurrent densityI versus potential, shown in figure 5.7, is generated.

1,E-02

1,E-03

1,E-04

1,E-05

1,E-06

1,E-07 -2000 0 2000 4000 6000 8000 10000 Potential - mV

Figure 5.7. Polarization scan for a UFW10 hardmetal in an tannic acid concentration of 0.00588M.

k EG&G Princeton Applied Research 273A 5 Experimental Methods 55

By extrapolation of the linear parts of the anodic and cathodic Tafel slopes from a polarization scan, the value of icorr is determined. This is shown in figure 5.8. Icorr is the point of intersection of the anodic and cathodic Tafel plots where iox + ired = 0; the point at which the corrosion system is in equilibrium. The value of icorr is related to a potential of Ecorr.

Icurrent densityI Ecorr (log scale) cathodic anodic reaction reaction icorr

potential

Figure 5.8. Schematic showing the Tafel plots from which the value of icorr is determined

The value of icorr is then used to calculate the corrosion rate of the metal in units of millimeters per year (mmpy), according to equation 5.8. Equation 5.8 is derived from Faraday’s Law, shown in equation 5.9, which assumes 100% current efficiency regarding material dissolution.

i ⋅ M CR(mmpy) = 3270 ⋅ corr n ⋅ ρ E5.8

Where: M = molecular weight (g/mol) n = no of electrons in the oxidation reaction ρ = density of sample (g/cm3)

MIt ∆m = nF E5.9

Where: ∆m= mass loss (g) M = molecular weight (g/mol) I = current (A) t = time (s) F = Faraday’s constant (96487 C mol-1)

The electrolytes used during testing are given in table 5.5. Pure tannic acid solutions were used initially in order to provide reference corrosion rates. Tannic acid represents one of the acids found in wood. Thereafter, solutions consisting of tannic acid and SiO2 sand were employed. These tests were performed as they are more representative of the corrosive environment experienced during the wear testing in which Quartz sand is used as the abrasive. This test is described in section 5.4.

The corroding potential of the tapwater used during wear testing was also evaluated. The acid- sand solutions were made up as follows: • the acid and sand was mixed in a ratio of 2:3, which corresponds to the ratio used in the wear tests • the solution was stirred for 15 minutes, which is the time required for a single complete polarization scan to be recorded 5 Experimental Methods 56

• the solution was drained from the sand and then used for the corrosion tests • the sand was removed from the solution in order to avoid impingement by the sand particles on the corroding metal surface during testing.

Table 5.5. Electrolytes used for the potentiodynamic tests.

Tannic acid concentration g/l M 5 0.00294 With and without SiO sand 10 0.00588 2 20 0.0118 Tapwater

In order to account for any IR-drop effects that may be present the conductivity of all the solutions were recorded using a conductivity meterl so that the polarization curves could be adjusted accordingly if required.

5.3.1 Reproducibility

The reproducibility and accuracy of the polarization curves are shown for hardmetal grade UFW10 in figure 5.9. The hardmetals were tested three times to confirm repeatability. There is good repeatability with a maximum deviation of a factor of 1.5. The maximum corrosion current density of 1.6e-4A/cm2 is reached at ±1100mV with good repeatability. Thus reproducibility was found to be in agreement with required corrosion standards. Any deviations measured can be attributed to the electrochemical behaviour of the specimen and not due to the testing apparatus or procedure. Specimens were initially polished to a 1 micron surface finish to ensure that the results would be comparable.

1,E-02

1,E-03

1,E-04

1,E-05

1,E-06

1,E-07

1,E-08 -2000 0 2000 4000 6000 8000 10000 Potential - mV

Figure 5.9. Two polarization scans for a UFW10 grade showing the reproducibility of the corrosion tests.

l WTW, Microprocessor Conducting Meter LF96 5 Experimental Methods 57

5.3.2 Microstructural investigation

In order to determine the corrosion mechanisms taking place, simple immersion tests were also carried out during which no voltage was applied. The solutions chosen are shown in table 5.6. In order to quantitatively compare the corroded surfaces, each test was conducted for a time period of 60 minutes and prior to testing the specimen surfaces were polished to a 1 micron finish. The samples were placed in a 500ml solution, in a open beaker and was stirred magnetically to eliminate concentration gradients during testing. The corroded specimen surfaces were examined using x-ray diffraction(XRD), field emission scanning electron microscopy(FESEM) and energy dispersion spectroscopy(EDS).

Table 5.6. Solutions used for immersion corrosion testing.

0.00588M Tannic acid With and without SiO2 sand

5.4 Three body abrasive wear tests

Three-body abrasive wear testing was selected since the tribological system which encompasses woodcutting comprises of three bodies, namely, the cutting tool, the wood, and the chips and particles which are by-products of the cutting process. The three body abrasive wear tests, using quartz sand as the abrasive, were carried out using an apparatus which was designed and built at the Friedrich Alexander University in Erlangen. A full description of the design of the apparatus including the software used to determine the volume loss can be found in reference Herr [84]. A schematic of the apparatus is shown in figure 5.10.

Holding bin for abrasive sand Flow meter for liquid medium

Abrasive flow regulator

Normal load

Wear sample

Abrading wheel

Figure 5.10. Schematic of wear apparatus. 5 Experimental Methods 58

The basic operation of the system consists of a constant normal applied force holding a hardmetal sample against a rotating steel wheel, on which a film of abrasive sand and liquid medium have formed. The normal applied force(FN) is applied perpendicularly onto the hardmetal sample such that it forms a right angle to the steel wheel. The steel wheel speed is controlled via an electric motor connected to a voltmeter. The standard fluctuation of the voltmeter is 0.01m/s. The liquid medium, used to wet the abrasive sand, is pumped onto the steel wheel from a container by means of an electric motor. The medium flow rate is controlled by a flow meter. The dry abrasive sand flows down from a holding bin, through a flow regulator and onto the wheel. The flow regulator is driven by an electric motor which is controlled by means of a voltmeter. The water and sand form a homogenous film on the rotating steel wheel. Once this film is formed on the wheel, the sample is pressed against the rotating wheel. The resulting wear scar is shown in figure 5.11 along with the specimen geometry used.

29mm

2mm 19.5mm

Figure 5.11. Wear sample geometry: schematic (left), as-received state (middle), worn state (right).

During a wear test, the friction force, the applied normal force, the wear depth and the liquid medium flow rate are measured on-line by making use of a analog digital converter – amplifier(ADCA)m. The data from the ADCA was transferred via measurement softwaren into the computer. An example of an online measurement graph is shown in figure 5.12.

120 700

FN 600 100 500 80 µ 400 60 L 300 40 200 FF 20 100 WD 0 0 0 500 1000 1500 2000 2500 3000 3500 Time [s]

Figure 5.12. Online measurement graph for a grade FW15 sample showing the normal force (FN), liquid medium (L), friction force (FF) and wear depth (WD) plots. m Spider8, Hottinger Baldwin Meßtechnik n CATMAN version 2.1, Hottinger Baldwin Meßtechnik 5 Experimental Methods 59

The friction and normal force are measured using load cells built into the apparatus, while the wear depth is measured using a displacement sensor. The liquid medium flow regulator is connected to the ADCA which made on-line observation of the medium flow rate possible. A more detailed description of the measurement set-up can be found in Herr [84].

The wear depth is then used to calculate the wear rate for each hardmetal. This is done by converting the wear depth into a volume loss value according to equation 5.10 where A is the area shown in figure 5.13 and β is the wheel width. The area A can be calculated in terms of the wear depth and the radius of the wheel as shown in equation 5.10.

2 r  D    D   E5.10 V = A⋅ β = ⋅ 2arccos1−  − sin2arccos1−  ⋅ β L      2  r    r  

Sample

FF

100 rpm FN A

r = 97,5mm

V D

Figure 5.13. Schematic of the force vectors(left) and schematic of the contact area between abrading wheel and hardmetal specimen(right).

The volume loss values from equation 5.10 are plotted against the wear distance for a wear test duration as shown in figure 5.14. The wear distance(WD) is calculated using the wheel diameter(2r - mm), the wheel speed(s - rpm) and the test duration(t - s), according to equation 5.11.

π ⋅ 2r ⋅ n ⋅t WD = E5.11 1000 ⋅ 60

From the gradient of the graph shown in figure 4.14 – dVL/dWD the wear rate - W is determined. This wear rate, which is used in this thesis, is quoted in units of m3/m, which measures the volume loss per meter.

5 Experimental Methods 60

1,8E-09

1,6E-09

1,4E-09

1,2E-09

1,0E-09

8,0E-10 dVL

6,0E-10

4,0E-10

2,0E-10 dWD 0,0E+00 0 500 1000 1500 2000 2500 3000 Wear distance / m

Figure 5.14. Determination of the wear rate using alloy UFW10 as an example.

The wear tests conducted were divided into three groups; • dry wear tests using no liquid medium, • standard abrasion wear tests using water as the liquid medium and • corrosive-abrasive wear tests using tannic acid as the liquid medium.

Dry wear testing was done in order to observe the influence of the abrasive sand alone, in which no liquid is present. A similar situation is found in the cutting of manufactured wood products such as chipboard, where the wood is completely dry. The hardmetal grades used in this project are also often used to cut such wood products, so it was of interest to assess the wear properties of the hardmetal under dry abrasive conditions, and then to compare it to wet abrasive conditions. The parameters used for the dry testing are given in table 5.7.

Table 5.7. System parameters used in wear test.

Applied normal force 100N Wheel speed 1.02m/s

SiO2 sand flow 60g/min

The choice of SiO2 sand as the abrasive was two-fold: • wood is softer than the hardmetal cutting tools used, so the abrasive needed to be softer than the hardmetals. SiO2 has a hardness of approximately 975HV according to Hutchings [108]. • SiO2 is often found in manufactured wood products such as chipboard and as was discussed in chapter 4.1.6, the silicon content was seen to influence the wear of the hardmetal cutting tools. Thus SiO2 seemed to be the most practical choice. Sieve analysis of the SiO2 abradent done by Collenz [137] showed that the grain size of 85wt% of the sand is 212 µm.

The three body abrasive wear tests performed using water are referred to as standard wear in this thesis. This is to differentiate between tests carried out using water and tests carried out using the corrosive fluid, where the wear tests done are referred to as corrosive wear. Standard wear testing was done to provide a comparison basis for the corrosive-wear tests.

The parameters for the standard and corrosive wear testing are the same as those used for dry testing shown in table 5.7. In addition, the fluid flow rate was kept constant at 40ml/min. The choice of the abrasive to fluid ratio, 60g sand / 40ml water, was based on the fact that abrasion has been 5 Experimental Methods 61 shown to be a dominant wear mechanism in woodcutting and thus more sand than liquid was used.

The corrosive wear tests were carried out using tannic acid as the corroding medium. The choice of tannic acid was based on a literature survey which examined the corrosive nature of organic wood elements such as tannic acid. Tannins are found throughout the plant kingdom and may be divided into two groups, namely condensed tannins that are derivatives of flavanols and hydrolyzable tannins that are esters of a sugar, usually glucose. They have an extremely complex chemistry with the molecular structure being depicted in figure 5.15 [11].

Figure 5.15. Tannic acid molecule structure [11].

The exact concentration of tannic acid differs amongst wood species, but has been shown to exist in very weak concentrations. In order to determine a suitable concentration of tannic acid for the wear tests, Pine and Oak wood chips were placed in distilled water for 24 hours and the pH periodically measured. The pH was found to vary between 3-4. Four different concentrations of tannic acid, with a pH lying between 3-4 were used during the wear testing in order to assess the influence on tannic acid concentration on the wear of the hardmetals. The concentrations of tannic acid and their respective pH values are given in Table 5.8.

Table 5.8. Concentration and pH of tannic acids used in wear tests.

Concentration of tannic acid - M pH 0.00118 3.58 0.00294 3.43 0.00588 3.20 0.0118 2.96

It was decided to assess the influence of varying tribosystem parameters on the corrosive wear of the hardmetals. Within the limits of the wear apparatus and based on the review of cutting tool wear in chapter 4.1, it was decided to vary the applied normal force and abrading wheel speed. Although, the wear apparatus does not allow measurement of the high speeds usually encountered during woodcutting, it does allow a variation of the speed by small increments. For the wear tests evaluating the influence of applied load, the system parameters were kept constant as shown in table 5.7; with the force being varied between 20-120N. During wear tests evaluating the influence of abrading wheel speed, the system parameters were also kept constant as shown in Table 5.7; with the speed being varied between 0.51-1.02m/s. These low speeds minimize thermal effects. These tests were done under standard and corrosive conditions to provide comparisons and to assess the influence of the tannic acid on the wear rates. For the determination of the corrosive wear rates, a tannic acid concentration of 0.00588M was used.

A complete summary of all three body abrasive wear testing done is given in table 5.9. 5 Experimental Methods 62

Table 5.9. Summary of wear tests conducted.

Wear test type Parameters used Dry abrasive testing applied normal force: 100N wheel speed: 1.02m/s SiO2 flow rate: 60g/min

Standard abrasive testing applied normal force: 100N wheel speed: 1.02m/s SiO2 flow rate: 60g/min water flow rate: 40ml/min

Variation of tannic acid concentration acid concentration – M applied normal force: 100N 0.00118 wheel speed: 1.02m/s 0.00294 SiO2 flow rate: 60g/min 0.00588 acid flow rate: 40ml/min 0.0118 Variation of applied force force - N 20 wheel speed: 1.02m/s 40 SiO flow rate: 60g/min 60 2 acid & water flow rate: 40ml/min 80 100 120 Variation of wheel speed speed – m/s 0.51 applied normal force: 100N 0.61 SiO flow rate: 60g/min 0.71 2 acid & water flow rate: 40ml/min 0.81 0.91 1.02

A microstructural examination was carried out on the worn surfaces following testing in order to ascertain degradation mechanisms. Techniques used included, light microscopy, scanning, field emission and atomic force microscopy, x-ray diffraction, and energy dispersion spectroscopy. 6 Results 63

6 Results

In this chapter, the results of the experimental work carried out to assess the wear response of the tungsten carbide cobalt hardmetals are described. The more important relationships are displayed graphically and some microstructural features are shown in the form of optical and electron micrographs. Where applicable, the influence of corrosion has been highlighted.

This chapter is divided into four main sections; • material characterization, • woodcutting tests, • corrosion tests and • abrasive wear and corrosive-abrasive wear tests.

6.1 Material characterization

In this section the hardmetals are described with respect to both material and mechanical properties which are relevant for understanding the wear and corrosion response of the alloys. These properties were determined by the techniques described in the previous chapter and are listed in table 6.1. From the table it is noted that the cobalt content values, as calculated from the density measurements, are in good agreement with those given by the manufacturer. The coercivity and magnetic saturation values are quality control parameters and were found to be acceptable, as they lie within the specifications used in industry.

Table 6.1. Material properties of hardmetals UFW3, UFW10 and FW15.

Property Units UFW3 UFW10 FW15 wt%Co (given) 3 10 15 Density g/cm3 15.28±0.96 14.57±0.40 14.02±0.36 wt%Co(calculated) 3.60 10.15 15.68 Vickers Hardness HV30 2148±20 1729±21 1362±25 Fracture toughness MNm-3/2 8.64±0.16 9.87±0.01 13.82±0.23 Coercivity kA/m 44.30 25.63 15.24 Magnetic Saturation Gcm3/g 4.38 13.30 21.63 WC grain size µm 0.33±0.28 0.46±0.17 0.84±0.11 Binder mean free path µm 0.13 0.22 0.38

The ultrafine hardmetals are 30-60% harder than the fine-grained FW15 alloy, which in turn is approximately 40-60% tougher than the ultrafine alloys. Furthermore, when the binder is increased by a factor of 3.33, the carbide grain size increases by 0.1micron which leads to a decrease in the hardness by 24% while the fracture toughness increases by 14%. It can also be observed from

6 Results 64

Table 6.1 that as the cobalt content and carbide grain size increases, the binder mean free path increases.

6.2 Woodcutting tests

This section describes the results of the incipient wear stage during woodcutting in which the hardmetals UFW3 and UFW10 were subjected to testing under various conditions. The wear response of the two alloys are described with respect to the variation of wood type, length of wood cut and cutting speed. The dominant damage mechanisms including possible corrosion-wear interaction are described. Steady state wear phase results, in which extensive lengths of wood were cut using the UFW10 knives, are also presented and compared to the results obtained during the incipient wear stage.

6.2.1 Cutting edge widths and wear rates in the incipient wear stage

Table 6.2 shows the data obtained for the cutting edge width for hardmetals UFW3 and UFW10 after cutting successive lengths of wood at a cutting speed of 58.9m/s. The values shown in the table represent the average of at least 250 measurements made per grade per length of wood cut. The value shown at zero meters of wood cut represents the cutting edge width of an unused knife. This value was found to be the same for both hardmetals.

Table 6.2. Cutting edge width values for the UFW3 and UFW10 knives with respect to the length of wood cut.

Length of wood cut Chipboard Fresh Spruce Cured Spruce Fresh Oak m µm µm µm µm UFW3 0 1.7±0.11 1.7±0.11 1.7±0.11 1.7±0.11 1 4.56±1.54 4.65±1.46 4.72±1.60 4.41±1.78 2 4.98±0.77 5.21±1.22 5.18±1.59 5.76±1.25 4 5.38±0.98 5.60±0.97 5.49±1.23 6.52±0.76 6 5.57±0.43 5.94±0.65 5.78±0.86 7.13±0.53 UFW10 0 1.7±0.11 1.7±0.11 1.7±0.11 1.7±0.11 1 5.76±2.61 2.91±0.54 2.62±0.57 3.32±0.87 2 6.47±2.16 3.21±0.68 3.29±0.43 3.85±0.56 4 6.68±1.56 3.56±0.62 3.54±0.70 4.31±0.64 6 6.91±0.92 3.80±0.52 3.80±0.56 4.92±0.55

6 Results 65

In all cases, the sharp edge wears off rapidly after cutting only one meter of wood. The highest cutting edge width value was recorded for cutting chipboard using a UFW10 knife, where an increase of 340% from the original cutting edge width was measured. The cutting edge width due to cutting the three other wood types show an increase between 150-200% when using the UFW10 knives, while for the UFW3 knives an increase in 250-280% of the cutting edge width was measured for all four wood types cut.

After the initial rapid blunting, the rate at which the cutting edge width increases appears to become constant. This is illustrated in figure 6.1. After cutting one meter of wood an increase in the cutting width area of between 10-35% for each alloy was found for each wood type, with the exception of cutting fresh oak where higher increases are found. An increase of 15-50% was recorded for the hardmetal UFW10 and 30-60% for the hardmetal UFW3 when cutting fresh Oak.

10 10 µ µ 8 8

6 6

4 4 UFW10 2 UFW10 UFW3 2 UFW3 0 0 02468 02468 Length of wood cut / m Length of wood cut / m (a) Chipboard (b) Fresh Oak

10 10 µ µ 8 8

6 6

4 4 UFW10 UFW10 2 2 UFW3 UFW3 0 0 02468 02468 Length of wood cut / m Length of wood cut / m (c) Cured Spruce (d) Fresh Spruce

Figure 6.1. Cutting edge width values for the woodcutting knives with respect to the length of wood cut, for each type of wood, using the hardmetals UFW3 and UFW10 at a constant cutting speed of 58.9m/s.

With respect to the type of wood cut, the UFW3 knives show the greatest increase in cutting edge widths when cutting fresh Oak(figure 6.1(b)). The cutting edge width values for UFW3 when used to cut the three remaining wood types are similar within experimental error. The UFW10 knives have the greatest increase in cutting edge widths when cutting chipboard and the lowest when used to cut the two Spruce species. With respect to the type of hardmetal used, the UFW10 knives showed larger cutting edge width values than the UFW3 knives in the cutting of chipboard, while the reverse is true for the remaining wood types. These responses can be related to the properties of the wood, the hardmetals and the cutting conditions and will be discussed in chapter 7.

6 Results 66

Generally, after cutting 6m of wood, the knife edge has a width between 2-5microns depending on the type of wood cut and the cobalt content of the knife.

Despite the large number of cutting edge width measurements made, a reasonable scatter was recorded for each case as shown in figure 6.1. As more wood is cut, the scatter becomes less. For UFW10 the scatter recorded after 1m of wood is cut decreases from 15-40% from 1-6m of wood cut, while for UFW3 there is a 20-50% decrease in the scatter as more wood is cut. This trend in the scatter may be related to the development of the cutting edge profile as the cutting process progresses. During the initial cutting stages, the cutting edge is in a ‘settling down’ phase, resulting in a slightly jagged appearance along some parts of the cutting edge. As more wood is cut, this jaggedness decreases and the edge profile becomes more consistent as shown in the micrographs(figure 6.2) which compares the cutting edge after successive lengths of wood have been cut.

1m 2m 6m

Figure 6.2. The degree of jaggedness of a UFW3 cutting edge after cutting successive lengths of cured Spruce.

The degree of jaggedness is dependent on the type of wood being cut and the type of knife used, as reflected in the scatter shown in figure 6.1. For example, the UFW10 knives shows higher scatter values when cutting chipboard compared to the Spruce woods. However, the UFW3 knives display higher scatter than the UFW10 knives when cutting Spruce. These differences in the amount of scatter measured may be attributed to the properties of the woods and to the mechanisms of material removal which occur during cutting.

The wear rates, with respect to the length of wood cut, are shown table 6.3 and illustrated in figure 6.3. The two wear stages are clearly seen. The initial high wear rate, due to the rapid wearing off of the cutting edge within cutting one meter of wood, becomes slower as more wood is cut and appears to reach a constant value. It also appears that the wear rates of the two hardmetals become similar with increasing length of wood cut. For both hardmetal grades and all wood types cut, it can be deduced that after cutting 6m of wood the wear rate has decreased by a factor of at least 10.

6 Results 67

Table 6.3. Wear rates for the UFW3 and the UFW10 knives with respect to the length of wood cut. Values are in units of 10-15m3/m. Length of wood cut Chipboard Fresh Spruce Cured Spruce Fresh Oak m m3/m m3/m m3/m m3/m UFW3 1 151 158 164 140 2 17 23 19 58 4 9 9 7 20 6 3 6 5 12 UFW10 1 256 47 34 69 2 37 8 17 16 4 6 5 4 8 6 4 2 3 8

The relatively small differences between the cutting edge widths and the wear rates noted between the cured and fresh Spruce is possibly linked to the small lengths of wood cut. In literature, for instance, important differences in wear rates between cured and green wood of the same species has been seen after cutting numerous lengths of wood [31].

1000 1000 UFW10 UFW10

100 UFW3 100 UFW3

10 10

1 1 02468 02468 Length of wood cut / m Length of wood cut / m

(a) Chipboard (b) fresh Oak

1000 1000 UFW10 UFW10

100 UFW3 100 UFW3

10 10

1 1 02468 02468 Length of wood cut / m Length of wood cut / m

(c) cured Spruce (d) fresh Spruce Figure 6.3. Wear rates of the woodcutting knives with respect to the length of wood cut for each type of wood using hardmetals UFW3 and UFW10.

6 Results 68

The cutting edge width values with respect to the variation in cutting speed are given in table 6.4 and the corresponding wear rates in table 6.5. The results are illustrated graphically in figure 6.4. For each cutting speed and wood type, 4m of wood was cut.

Table 6.4. Cutting edge width values for the UFW3 and the UFW10 knives with respect to cutting speed variation.

Length of wood cut Chipboard Fresh Spruce Cured Spruce Fresh Oak m µm µm µm µm UFW3 0 1.7±0.11 1.7±0.11 1.7±0.11 1.7±0.11 29.5 5.07±2.03 5.21±1.66 5.04±2.57 6.21±1.83 39.3 5.20±1.46 5.03±1.32 5.17±1.90 6.59±1.56 58.9 5.38±0.98 5.60±0.97 5.49±1.23 6.52±0.76 UFW10 0 1.7±0.11 1.7±0.11 1.7±0.11 1.7±0.11 29.5 5.81±1.88 3.29±1.33 3.44±1.42 4.01±1.94 39.3 6.30±1.90 3.36±1.09 3.48±0.61 4.19±1.32 58.9 6.68±1.56 3.56±0.62 3.54±0.70 4.31±0.64

Table 6.5. Wear rates for the UFW3 and the UFW10 knives with respect to cutting speed variation. Values are in units of 10-15m3/m.

Length of wood cut Chipboard Fresh Spruce Cured Spruce Fresh Oak m m3/m m3/m m3/m m3/m UFW3 29.5 48 51 48 75 39.3 51 47 50 86 58.9 55 60 58 84 UFW10 29.5 65 17 19 28 39.3 78 18 19 31 58.9 88 21 20 33

6 Results 69

Chipboard

10 1000

µ UFW10 8 UFW3 100 6

4 10 UFW10 2 UFW3 0 1 020406080 20 30 40 50 60 70 Cutting speed / m/s Cutting speed / m/s Fresh Oak

10 1000

µ UFW10 8 UFW3 100 6

4 10 UFW10 2 UFW3 0 1 0 20406080 20 30 40 50 60 70 Cutting speed / m/s Cutting speed / m/s Cured Spruce

10 1000

µ UFW10 8 UFW3 100 6

4 10 UFW10 2 UFW3 1 0 20 30 40 50 60 70 0 20406080 Cutting speed / m/s Cutting speed / m/s Fresh Spruce

10 1000 µ UFW10 8 UFW3 100 6

4 10 UFW10 2 UFW3 0 1 0 20406080 20 30 40 50 60 70 Cutting speed / m/s Cutting speed / m/s Cutting edge widths Wear rates

Figure 6.4. Cutting edge widths and wear rates for hardmetals UFW3 and UFW10 with respect to cutting speed variation and wood type. In all cases 4m of wood was cut.

6 Results 70

From the graphs in figure 6.4 there appears to be a small but definite increase in the cutting edge widths and the wear rates of the two hardmetals as the speed increases from 30-60m/s. A possible reason why the cutting speed role appears to be diminutive could be that the cutting speeds used are very high and the amount of wood cut very small. In literature it has been shown that after cutting hundreds of meters of wood the influence of cutting speed becomes more obvious [31].

It was also found that the cutting edge widths of the UFW10 knives are higher than those of the UFW3 knives when used to cut chipboard, with the reverse being true for the three remaining wood types. Amongst the wood types, both Spruce species showed the lowest wear rates at all cutting speeds tested. No differences are visible between the two Spruce species. This can once again be linked to the small lengths of wood cut, but in addition possibly to the high speeds used. At high speeds, material removal would tend to occur at a fast rate, and in order for corrosion to play a role, contact time between the wood and knife is needed. At much lower cutting speeds any differences are expected to become more noticeable.

The only notable differences lie in the scatter, which appears to be highest at the lowest speed and decreases by approximately 20% as the speed increases for all cases, except chipboard. In the case of chipboard, the scatter remained constant at all speeds. This may be linked to the type of chip formation during woodcutting which will be discussed in chapter 7.

6.2.2 Cutting edge widths and wear rates in the steady wear stage

The cutting edge widths and the wear rates for the steady wear stage during which 3048m of cured Oak was cut are listed in table 6.6 and illustrated in figure 6.5. These values represent the average of approximately 250 measurements made per grade. For these tests the hardmetal UFW10 was used. The three remaining grades listed in the table were included for comparison purposes.

Table 6.6. Cutting edge widths and wear rates for various WC-alloy knives after cutting 3048m of cured Oak using a cutting speed of 30.52m/s.

WC Alloy Binder content Cutting edge width Wear rate wt% µm 10-18 m3/m UFW10 10(Co) 19.6±0.28 1323 #46 10(Co) 20.5±0.37 1442 #52 10(FeNiCo) 18.1±0.26 1127 #59 17.5(FeNiCo) 16.9±0.32 973

6 Results 71

24

20 m

16

12

8

Edge blunting(w) / 4

0 UFW10 #46 #59 #52 WC Grade Figure 6.5. Blunting for various WC-alloy knives after cutting 3048m of cured oak using a cutting speed of 30.52m/s.

From figure 6.5 it can be seen that the cutting edge width values of the FeNiCo-binder knives is between 8-20% lower than that of the cobalt-binder knives. For the FeNiCo-alloys, as the binder content increases, the cutting edge width increases. A 75% increase in binder content brings about a 7% increase in the edge width of these knives. The two cobalt-binder binder knives show a difference of approximately 5%, even though they have the same cobalt content and carbide grain size. They are however produced from different starting powders. The scatter for all grades is very low and lies between 0.25-0.35µm. These low values are indicative of the wear profile of the hardmetals after cutting longer lengths of wood, which is much smoother and less jagged compared to the incipient wear stage. A comparison between the two wear stages is shown in figure 6.6 for a UFW10 grade.

(a) (b) Figure 6.6. Scanning electron micrographs comparing the (a)incipient [after cutting 6m of fresh Oak] and (b)steady [after cutting 3048m of cured Oak] wear stages for UFW10 knives.

From the wear rates shown in table 6.6, it can be seen that the cobalt grades have a 20-50% higher wear rate than the FeNiCo-grades. Between grades UFW10 and #46, which are the straight 10wt% cobalt grades there is a 9% wear rate difference, with that of the UFW10 being lower. A 75% increase in the binder content for the FeNiCo-grades show a 16% increase in the wear rate. From these results it can be seen that the addition of iron and nickel to the tungsten carbides provide better wear resistance than the pure cobalt binder during the cutting of the cured Oak. This result is important since it may represent an alternative binder system for hardmetals in the field of woodcutting.

6 Results 72

A comparison of the incipient wear stage to the steady one, can be made for the UFW10 grade used to cut fresh Oak, which represents the incipient stage, and cured Oak which represents the steady stage. Unfortunately no other testing was possible in the steady wear stage, and as such only this one type of wood was cut. This is illustrated in figure 6.7 where the wear rates are plotted on a log-log graph. The cutting speeds used are similar and lie within 3% of each other. From this graph it can be postulated that if the curve of the fresh Oak wear rates were extrapolated, then the wear rate at 3048m for fresh Oak would lie above that of the cured Oak value. This result could imply that the corrosion processes which occur when cutting fresh wood increase the wear rates. However, more tests would need to be done to validate this theory.

1000000

100000 6m Fresh Oak cut @ 29,5m/s

/m 10000 3 m -18 3048m Cured Oak cut @ 30,52m/s 1000

100 Wear rate / 10

10

1 1 10 100 1000 10000 100000 1000000 Length of w ood cut / m

Figure 6.7. Comparison of the wear rates for the UFW10 hardmetal when cutting fresh and cured oak.

A comparison of the cutting edge width values, albeit for different wood types, show a factor of 4 difference between the cutting edge widths after 6m and after 3048m of wood has been cut. This indicates that after the initial rapid wearing off of the sharp edge, the knives then wear at slower rates which remain approximately constant even after larger volumes of wood have been cut.

6 Results 73

6.2.3 Microstructural examination

The worn edges of the knives were examined using scanning electron microscopy in order to obtain microstructural information. From these micrographs it is possible to gain a general impression of the effect the different wood types have on the two hardmetals. From this study, holes are observed in the worn zone where sections of the hardmetal have been dislodged due to either the presence of wood nodes or due to the general wear process. Features such as pits and uneven worn regions are easily recognized.

Wear profiles of the knives can be constructed from a number of electron micrographs to illustrate these effects better. One section of such a profile is shown in figure 6.8 for a UFW10 knife used to cut 1m of chipboard at a speed of 58.9m/s. The micrographs suggest that, in general, abrasion is the main wear process, since the edge of the worn region follows the profile of the ground surface, i.e. the groove geometry on the rake face. The width of the cutting edge remains approximately constant for this region, with very few areas showing enhanced knife wear. In the area marked A, for example, enhanced wear occurred possibly due to a high density region in the chipboard coming into contact with the knife. Board density is known to increase the wear on the tools [37].

Features marked B represent wear debris, which is formed during the cutting process and consists of chipboard fragments and sections of hardmetal. When cutting chipboard most of the wear took place on the clearance face of the knives. This may be due to the rubbing action between the chipboard and the tool face. Less wear occurs on the rake face since chipboard machines by fracture and no continuous chip is formed, which would otherwise cause wear on the rake face. For the Oak and Spruce woods more wear occurred on the rake faces and this could be seen by the amount of wear debris present and by the presence of abrasion scratches running perpendicular to the grinding lines on the rake faces of the knives. The abrasion scratches are indicative of the chip flow over the rake face. The scratches were found to be, in general, more intense close to the cutting edge and gradually decrease further away from the cutting edge, further confirming that chip motion may be the primary cause for the scratches.

6 Results 74

Rake face

Clearance face

B

B

A

Figure 6.8. Wear profile of the cutting edge for a UFW10 knife used to cut 1m of chipboard employing a cutting speed of 58.9m/s

It could not be distinguished whether Oak or Spruce wood caused more wear on the rake face. This may be attributed to the small lengths of wood cut. In literature, it has been shown that different natural woods result in different degrees of wear. However, it is clear that chipboard caused more wear on the clearance face of the hardmetal cutting knives compared to the other three wood types.

As the length of the wood cut increased, the wear on the rake and clearance faces of the knives increased for both alloys. More wear debris was found and the number of scratches increased. However, as more wood was cut the wear profiles showed more even-wear profiles in the direct cutting zone, i.e. fewer regions of enhanced wear are seen. This is an indication that the wear process is beginning to become more consistent as more wood is cut. This was illustrated in figure 6.2.

Regions of smearing commencing from the cutting zone were found on the UFW10 knives used to cut cured Spruce, while no smeared regions were found on similar knives used to cut the fresh

6 Results 75

Spruce. This is shown in figures 6.9(a)-(d). The smeared regions are approximately 1 to 2 microns in width and were found to occur along at least 60% of the knife edge. No smearing was found along the clearance face, which indicates, in part, that the smearing may be due to the chip motion over the rake face. In a few sections along the cutting edge, the cutting zones themselves were covered by these smeared regions, as depicted in figure 6.9b.

Smearing

Notches

(a) (b)

Holes

(c) (d)

(e) (f) Figure 6.9. Smearing found on the rake face of UFW10 knives used to cut cured Spruce(a-c; e) but not on the rake face of similar knives used to cut fresh Spruce(d; f). Circled areas represent areas of ductile cobalt films which indicate that carbide grains have been removed.

It can also be seen in figure 6.9(b) that part of the rake face has no smeared region, even though the adjoining cutting edge is covered in this smeared metal. The reason for this type of smearing is not entirely clear. Although chip motion can be responsible, it is interesting that these smeared regions are absent when fresh Spruce is cut, since continuous chips are formed when cutting both woods. Some likely contributing factors may include the orientation of the wood to the rake face

6 Results 76 and the forces exerted on the tool face by the wood during cutting. It is also interesting that along the clearance face of the knives used to cut the cured Spruce, small notches, approximately 1micron in length were found, as shown in figure 6.9(a). These type of notches were notably absent from the clearance faces of the knives used to cut the fresh Spruce.

There is no obvious difference in the worn surfaces in the direct cutting zone for both hardmetals in cutting fresh and cured Spruce, other than more holes being found in the direct cutting zone of the UFW10. This may be due to the corrosive influence of the wood acids which leach the cobalt binder out of the material and result in carbide grain fall-out. The wear features appear to be similar as depicted in figure 6.9(e) and 6.9(f) for alloy UFW10. A few broken carbide grains and some pits can be seen in both micrographs. Thin films of ductile cobalt, the regions marked with circles, also show where carbide grains have been removed from the matrix during cutting.

(a) (b)

Transgranular cracking of WC grains

(c) Figure 6.10. Wear features on cutting edges of hardmetal UFW10 used to cut chipboard. Worn surfaces display a very rough appearance.

Figure 6.10 shows a closer view of the general effect of cutting chipboard with the hardmetal knives. Fragmentation of the carbide grains appears to be a dominant feature. The presence of smaller than nominal grain size fragments of tungsten carbide indicates that larger grains have been fractured into smaller ones. There is a large range of sizes of fractured carbide grains indicating that the chipboard can inflict a variety of different levels of severe wear on the hardmetals. Very few carbide grains in the worn zones were found to be in the order of the nominal grain size. Transgranular cracking of a few of the carbide grains, such as that circled in figure 6.10©, can also be clearly seen. In addition, most of the grain edges have lost their angularity, becoming rounded due to the wear process. Fragments of the chipboard components were also found in the wear zones of the knives. Some regions depict clusters of grains which appear to be

6 Results 77 secured onto the hardmetal on their undersides, with the cobalt binder having been removed from between them. This is an indication of the wear mechanism which may take place during the machining of chipboard.

Holes such as those depicted in figure 6.10(b), were found along the wear zones which indicate areas where cluster grain removal took place. Further examples of holes are shown in figure 6.11 for the various wood types. The holes could be attributed to various reasons which depend on the wood type, hardmetal type and the machining conditions. For example some of the contributing factors could include impact from a sand particle present in the chipboard and holes linked to the corrosive nature of the tannins present in the fresh woods which remove the binder and lead to grain fall-out or grain pull-out.

(a) (b)

(c) (d) Figure 6.11. Types of hardmetal material holes due to factors such as notches in the wood, sand particles in chipboard and corrosive attack due to extractives in fresh woods.

Smearing in the cutting zone was commonly found on the knives used to cut chipboard. This type of smearing, shown in figure 6.12, is different to that depicted in figure 6.9. Three kinds of smearing were found in the cutting zones and these are shown in figure 6.12. On the smeared zone in figures 6.12(a) and 6.12(b), there are lines running at an acute angle in the cutting zone. This may be due to the grain orientation of the chipboard particles in this region being different to that in the rest of the chipboard and hence smear the hardmetal in the indicated direction. The wood grains in this region could also have been slightly more abrasive since a section of the cutting zone below this smeared zoned has been removed causing a hole along the clearance face, as shown in figure 6.12(a).

6 Results 78

(a) (b)

(c) (d)

(e) (f) Figure 6.12. Comparison of cutting edges of chipboard which show different types of smeared regions which arise from various factors.

The smeared zone depicted in figures 6.12(c) and 6.12(d) has a film-like shape which covers the entire cutting zone. This picture is very different to that shown in figure 6.10 which is the normal worn appearance of the knives when cutting chipboard. This type of smearing which is approximately 15microns in length could be attributed to the glue-line of the chipboard. The glue- line is smooth and would therefore not be expected to cause rough abrasive wear. Instead under the cutting pressures, the glue-line may deform and smear both the glue and hardmetal fragments across the cutting zone. The type of smearing shown in figures 6.12(e) and 6.12(f) were found at random intervals along the cutting zone, making their existence more difficult to classify. However,

6 Results 79 in figure 6.12(f), fragments of carbide grains can be seen in the smeared zone which has the appearance of a film of diffused metal. It is possible that hardmetal wear debris became trapped between the chipboard and the cutting tool and due to further machining this diffused layer is formed which then adheres to the surface, by possibly bonding with the underlying hardmetal.

Evidence of the corrosive effect of the wood extractives present in the natural woods was found on the hardmetal knives which were used to cut the fresh Oak. Micrographs showing the worn cutting zones for hardmetal UFW10 used to cut the fresh Oak are illustrated in figure 6.13. The worn appearance is markedly different to that obtained when cutting chipboard or Spruce. The carbide grains can be seen to have retained their original angular shape. Very few grains have rounded edges and few fragmented carbide grains are seen. From EDS analysis it was found that the hardmetal surface layer was practically denuded of cobalt since smaller cobalt peaks were detected. This indicates that the cobalt was selectively removed from the cutting zone, most likely by the combined action of leaching out by the wood extractives and the general abrasion process. Transgranular cracking of the carbide grains was found and this is probably due to abrasion as the cutting proceeds. The general worn appearance when cutting fresh Oak, was found to be a relatively smooth surface, with random holes and a few smeared regions, such as those depicted in figure 6.12(b).

Figure 6.13. Worn cutting edges of the UFW10 knives used to cut fresh Oak.

Micrographs taken of the worn zones of the knives used to cut 3048m of cured Oak revealed that the FeNiCo-binder knives had relatively smooth wear zones with hardly any holes or depressions in the cutting zone. When present, the holes were about 1micron in diameter. The UFW10 knife showed some signs of holes which are approximately 1-3microns in diameter and some small chipping fractures were found along the rake face. The hardmetal #46 showed extreme damage in the cutting zone, with cracks, holes and chipping fractures visible throughout the cutting zone. Examples of the four alloys are shown in figure 6.14.

In all four cases, more wear occurred on the rake faces, than the clearance faces, probably due to chip flow. It was also noted that the grinding lines on the rake face were absent and had been abraded away most likely due to chip motion. The connecting zone between the rake face and the cutting zone was found to be smooth, unlike the rough edge appearance found in the figures 6.8 to 6.11 where only a few meters of wood were cut. Some smearing was prevalent in all the grades. A few regions of hardmetal discoloration were detected at periodic intervals along the cutting zone. This is shown in figure 6.15(a), where the darker area represents the discoloration. This may be attributed to the wood staining the knife during cutting. An example of a chipping fracture is shown

6 Results 80 in figure 6.15(b) which could have been caused by a node in the wood hitting the metal or by a sudden interruption in the machining process.

(a) UFW10 (b) #46

(c) #52 (d) #59 Figure 6.14. General worn appearance of the hardmetal knives used to cut 3048m of cured Oak.

(a) (b) Figure 6.15. (a) Discoloration of the cutting edge possibly due to wood stains. (b) Chipping fracture in alloy UFW10.

A noteworthy difference between the hardmetal knives containing cobalt and FeNiCo-binder is the presence of transgranular cracks which were found in the cutting zones of the cobalt binder knives. The cracks were between 40 to 60microns in length and two examples of such cracks are shown in figure 6.16. The initiation point of the cracks is difficult to define, but from the crack profile, it is

6 Results 81 evident that there has been proceeding growth. In figure 6.16(d) it can be seen that whole-grain removal has taken place. It was noted that crack growth was predominantly intergranular. The origin of these cracks may be two-fold. The initial crack may be the result of a wood knot hitting the cutting zone at high speed and the crack progresses as the cutting process continues. The cracks may also be the result of fatigue processes. The cutting procedure is such that the knives repeatedly strike the wood in a circular motion, setting up cyclic stresses in the wear zones, which may eventually cause crack initiation.

(a) (b)

(c) (d) Figure 6.16. Transgranular cracks found in the cutting zones of hardmetal UFW10.

The majority of these cracks were found in hardmetal #46. Alloy UFW10 had very few of these cracks along the cutting zone. The extreme damage found in the wear zone of alloy #46 is further illustrated in figure 6.17. In the middle of the cutting zone it appears as though severe fracture and plastic deformation has taken place, which is evident by the ductile tears visible. Cracking of the metal can also be seen along the rake face in the transverse direction. This type of severe damage was found at random intervals along the cutting zone. The severity of the damage on this knife, which has the same composition as the UFW10 knife and a similar carbide grain size is not easily explainable.

6 Results 82

Figure 6.17. Severe wear regions found on hardmetal #46.

6.2.4 Summary of the main findings

• During initial contact between the knife and the wood, the sharp edge wears off rapidly. Then as more wood is cut, the knives wear at slower rates. • The extent to which the cutting edge recedes initially, is found to be dependent on the type of wood cut and the cobalt content of the hardmetal knife used. • As the length of the wood cut increases, the wear rates of the two alloys appear to become similar. • The 3wt% cobalt knives show lower wear rates than the 10wt% cobalt knives when used to cut chipboard. The reverse is found to be true when cutting the three remaining wood types. • The 3wt% cobalt knives have the highest wear rates when used to cut fresh Oak and the lowest when used to cut chipboard. • The 10wt% cobalt knives have the highest wear rates when used to cut chipboard and the lowest when used to cut both the Spruce species. • Little difference is found between the wear rates of the knives when used to cut the fresh and cured Spruce wood types. • The wear rates of the knives decrease by a factor of 10 after 6m of wood is cut. This includes all wood types cut. • A small, but definite increase in the wear rates is found as the cutting speed was increased from 30 to 60m/s. • In the steady state wear tests, the FeNiCo-binder knives show lower wear rates than the Co- binder knives. • In the steady state wear tests, the two Co-binder knives, which have the same amount of cobalt content and similar tungsten carbide grain sizes have different wear rates. • During the initial cutting process, the wear profile of the cutting zone is extremely jagged. It becomes more consistent as the length of wood cut increases. The degree of inconsistency of the wear profiles, is found to be dependent on the type of wood cut and the binder content of the hardmetal knife used. • Abrasion is found to be the dominant wear process. • The knives showed more wear on the clearance faces when used to cut chipboard. However, when used to cut the remaining three wood types, more wear was found on the rake faces. • Wear features located on the worn surfaces include:

6 Results 83

- various types of hardmetal smearing - holes and depressions of varying sizes and types - fragmented tungsten carbide grains - transgranular cracking of carbide grains - preferential removal of the cobalt - carbide grain pull-out - loss of angularity of carbide grains - chipping fractures • The type and the extent of wear features found in the cutting zones is dependent on the wood type and binder composition of the hardmetal knife. • Possible evidence of corrosive influence is found when cutting the fresh Oak. • In the steady state worn cutting zones, transgranular cracks between 40-60microns were found in the Co-binder knives, but none were found in the FeNiCo-binder knives. • In the steady state worn cutting zones, excessive damage was found on one of the 10wt% cobalt knives.

6.3 Corrosion tests

This section describes the results of the electrochemical investigation of the hardmetals, in which the grades UFW3, UFW10 and FW15 were subjected to corrosion tests under various conditions. The corrosion response of the alloys are described with respect to variation of tannic acid concentration, and the influence of quartz sand addition to the electrolyte. The potentiodynamic results are displayed with the potential plotted along the abscissa and the current along the ordinate. This was done since the current density is measured as a function of the applied potential. In table 6.7 the significant corrosion parameters are listed for testing conducted in tannic acid electrolytes. The corrosion currents recorded are low in this system. However, in corrosion systems in which stronger acids, such as H2SO4 were used, low currents were also recorded [90]. In addition, from the conductivity measurements carried out in the solutions, in order to determine and IR-effects, the values displayed in table 6.7 can be taken as true system values.

6 Results 84

Table 6.7. Electrochemical corrosion parameters.

Tannic acid concentration icorr Ecorr Corrosion rate M 10-7A/cm2 mV 10-5mmpy UFW3 0.00294 106±7 -321±11 62±4 0.00588 121±3 -355±13 71±2 0.0118 130±10 -329±11 77±6 UFW10 0.00294 141±2 -353±4 100±1 0.00588 171±5 -373±12 121±4 0.0118 233±3 -391±16 166±2 FW15 0.00294 161±5 -354±9 129±4 0.00588 201±3 -369±7 162±2 0.0118 269±4 -388±7 216±3

In figure 6.18 the potentiodynamic polarization scan for the composite UFW10 recorded in a 0.00294M tannic acid concentration is shown on a log-linear plot. The polarization characteristics of all the hardmetal grades in all electrolytes tested are similar and this curve is used to describe the general behaviour. The cathodic current density decreases exponentially with increasing potential indicating activation control. The hardmetals exhibit a corrosion potential between –320 to –390mV, which is about 150mV positive of the single potential of cobalt (E° = -518mV vs SCE). The anodic current density initially increases exponentially as the potential is increased above Ecorr. At around –100mV the rate of increase starts to slow down and all the alloys reach a maximum current density after which a distinct drop in current density occurs. Beyond this potential the current density gradually increases again with further increasing potential. This increase may be due to the onset and increase of tungsten carbide oxidation which has been shown to occur at high potentials(human) or it can be related to oxygen evolution. This system is not truly passivating due to the high current densities recorded, even though the observed behaviour shows passivating characteristics. Consequently such behaviour is termed ‘pseudo-passivity, a term developed by Human [90] for similar systems.

6 Results 85

1,E-01

1,E-02

1,E-03

1,E-04

1,E-05

1,E-06

1,E-07 -2000 0 2000 4000 6000 8000 10000 Potential / mV Figure 6.18. Potentiodynamic polarization curve of the UFW10 composite recorded in a 0.00294M tannic acid concentration.

In figure 6.19, log-linear polarization scans are shown for each hardmetal grade, recorded in a 0.00294M tannic acid concentration. The anodic slopes are similar for all grades, but the maximum current density is reached at higher potentials with increasing binder content. This trend is in agreement with that found in literature. The maximum current density increases with increasing cobalt content, a trend which is observed in all electrolytes tested and agrees with trends observed in literature. The pseudo-passive current values tend to increase with increasing cobalt content and are reached at higher potentials with increasing binder content.

1,E-01

1,E-02

1,E-03

1,E-04

1,E-05

1,E-06 FW15 UFW10 UFW3 1,E-07 -2000 0 2000 4000 6000 8000 10000 Potential / mV

Figure 6.19. Potentiodynamic polarization curves for all three hardmetal grades recorded in a 0.00294M tannic acid concentration.

6 Results 86

From figure 6.19 it can be seen that the UFW3 grade shows a steeper increase in current density as the potential increases after the pseudo-passivation current is reached, compared to the other two grades. This may be a confirmation that the increase in current density is associated with tungsten carbide oxidation, since the UFW3 grade has about 7-15% more tungsten carbide than the other two hardmetals. This trend was observed in all the electrolytes tested.

The corrosion rates, as determined by the Faraday equation, are shown in figure 6.20 for the three hardmetals corroding in a 0.00294M tannic acid concentration. As expected from the polarization curves, the corrosion rates increase with increasing cobalt content and binder mean free path. When comparing the two ultrafine hardmetals, UFW3 and UFW10, it can be seen that the corrosion rate has increased by 60% when the cobalt content increases by a factor of 3.3. The corrosion rate of the FW15 grade is 30-200% greater than the ultrafine corrosion rates. The scatter on the values are similar and lie within a 5mmpy band. The increasing corrosion rates with increasing cobalt content can be related to the fact that since the corrosion of hardmetals is governed by cobalt oxidation, the more cobalt present the more susceptible the metal is to attack, hence leading to higher corrosion rates.

140

120

100

80

60

40

20

0 UFW3 UFW10 FW15

Figure 6.20. Corrosion rates for the alloys in a 0.00294M tannic acid concentration.

The corrosion rates of the hardmetals with respect to increasing tannic acid concentration are illustrated in figure 6.21. As the concentration of the acid increases, associated with decreasing pH(see table 5.8 in chapter 5), the corrosion rates increase at a linear rate. The UFW3 grade corrosion rate shows the least sensitivity to increasing acid concentration, where an increase of 24% was measured as the tannic acid was increased from 25 to 120e10-4M. In contrast the UFW10 and FW15 grades were more susceptible to the decreasing pH and the corrosion rates increased by 65% for both grades.

6 Results 87

300 FW15 UFW10 250 UFW3

200

150

100

50

0 0 20 40 60 80 100 120 140 Tannic acid concentration / 10-4 M Figure 6.21. Corrosion rates of the hardmetals with respect to increasing tannic acid concentration.

The significant corrosion parameters for the polarization scans recorded in the tannic acid-quartz sand electrolytes are listed in table 6.8. The current densities are much higher than those recorded in pure tannic acid solutions, while the corrosion potentials are lower by 20-40mV than those observed in pure tannic acid. Figure 6.22 illustrates two polarization scans for a FW15 hardmetal recorded in a 0.00294M tannic acid solution, with and without quartz sand. The higher current densities in the acid-sand solution are clearly illustrated. The two profiles are similar, except that no drop in current density is recorded in the acid-sand solutions. This is true for all solutions tested using all hardmetals. Thus, no passivation occurs in these solutions for the hardmetals tested indicating that the hardmetal simply continues to corrode.

6 Results 88

Table 6.8. Electrochemical corrosion parameters for tests conducted in acid-sand electrolytes.

Tannic acid concentration icorr Ecorr Corrosion rate M 10-4A/cm2 mV 10-2mmpy UFW3 0.00294 39±4 -324±7 23±2 0.00588 43±2 -306±15 25±1 0.0118 48±3 -318±11 28±2 UFW10 0.00294 48±5 -339±9 34±4 0.00588 54±2 -332±12 38±1 0.0118 62±2 -318±11 44±1 FW15 0.00294 61±2 -366±5 49±2 0.00588 72±4 -350±12 58±3 0.0118 83±1 -344±8 67±1

1,E+01

1,E+00

1,E-01

1,E-02

1,E-03

1,E-04

1,E-05

1,E-06 FW15-with sand FW15 1,E-07 -1000 -500 0 500 1000 1500 2000 2500 3000 3500 4000 Potential / mV

Figure 6.22. Potentiodynamic polarization scans of the FW15 composite recorded in a 0.00294M tannic acid concentration, with and without quartz sand.

The higher current densities in the acid-sand solutions are an indication of the aggressiveness of the solution. This most likely arises due to dissolution of impurities, found in the quartz sand, into the acid solution since the quartz sand does not dissolve in tannic acid. The exact nature and composition of the impurities has proven difficult to identify. EDX and XRD measurements made on the quartz sand did not reveal any significant measurable amounts of other periodic table elements. Spectral analysis of the acid-sand solution was conducted, from which small amounts

6 Results 89

(1-3mg/l) of , potassium and iron were detected. The complete set of elements found by spectral analysis is shown in table 6.9.

Table 6.9. Elements found in acid-sand solution by spectral analysis in units of mg/l.

Li B Na Mg Al K Ca Cr Mn Fe 0.007 0.029 0.338 0.01* 1.698 2.563 0.484* 0.007 0.182 2.248 Co Ni Cu Zn Sr Ag Cd Ba Pb 0.006 0.004 0.104 0.003* 0.026 0.005 0.001* 0.561 0.053* *Units of ppm

The impurities may also be of an organic nature due to the sand preparation techniques, and are more difficult to identify. In order to clarify that it was indeed impurities present in the sand which increased the current densities during corrosion testing corrosion testing was carried out using sand which was repeatedly ‘washed’ prior to testing. Figure 6.23 illustrates the corrosion rates determined for hardmetal FW15 in the three different solutions. As can be seen in figure 6.23, the corrosion rate of the acid-cleaned sand solution is lower than that calculated using the as-received sand. The corrosion rate has decreased by 40%. However, both acid-sand solutions are still far more aggressive than the pure tannic acid solutions. The acid-sand corrosion rates are 140-350% greater than the corrosion rates measured in pure tannic acid for the FW15 grade.

100000

10000

1000

100

10 acid acid-sand acid-cleaned sand

Figure 6.23. Corrosion rates for hardmetal FW15 in solutions of pure tannic acid, tannic acid to which sand has been added and tannic acid to which ‘washed’ sand has been added. A tannic acid concentration of 0.00588M was used for all tests.

The corrosion rates for all three hardmetals in the acid-sand solutions are illustrated in figure 6.24. The sand used was in the as-received condition. Once again, the corrosion rates increase with increasing tannic acid concentration and with increasing cobalt binder content.

6 Results 90

80 FW15 UFW10 70 UFW3

60

50

40

30

20

10

0 0 20 40 60 80 100 120 140 Tannic acid concentration / 10-4 M

Figure 6.24. Corrosion rates of the hardmetals with respect to increasing tannic acid concentration.

The UFW3 grade displays a similar sensitivity to increasing tannic-acid solution here as before in pure tannic acid solutions and an increase in the corrosion rate of 23-25% was recorded in both solutions. The sensitivity of the UFW10 and FW15 grades is lower in the acid-sand solutions than in the pure acid solutions. In the acid-sand solutions the corrosion rates increase by 30-35% with increasing tannic acid concentration, while a 65% increase was found in the pure acid solutions. This is interesting and indicates that although the sand increases the conductivity of the acid solution, it’s presence does not necessarily imply an increase in the solution aggressiveness with increasing tannic acid concentration. pH measurements made of the acid-sand solutions were similar to those found in pure acid solutions. Figure 6.25 summarizes the corrosion results for all three hardmetals in all acidic electrolytes tested. The corrosion rates measured in acid-sand electrolytes are approximately one and a half decades higher than the corrosion rates measured in pure tannic acid.

6 Results 91

1000000

FW15-with sand UFW10-with sand UFW3-with sand FW15 UFW10 UFW3

100000

10000

1000

100

10

1 0 20 40 60 80 100 120 140 Tannic acid concentration / 10-4 M

Figure 6.25. Summary of corrosion results as a function of tannic acid concentration for all three hardmetals.

The significant corrosion parameters for the polarization scans recorded in the tapwater and tapwater-quartz sand electrolytes are listed in table 6.10. The current densities recorded in the water-sand electrolytes are much higher than those recorded in the water solutions, while the corrosion potentials are slightly higher than those observed in the tapwater. The IR-drop effect has been accounted for in these measurements. Figure 6.26 illustrates four polarization scans for a FW15 hardmetal recorded in acid and tapwater solutions, with and without quartz sand. The higher current densities in the acid-sand solution are clearly illustrated. The separation between the acid and tapwater curves, with and without sand is similar, which indicates that the sharp increase in currents recorded is due to the sand addition. No drop in current density is recorded in the tapwater and tapwater-sand solutions. This is true for all hardmetals tested.

6 Results 92

Table 6.10. Electrochemical corrosion parameters for tests conducted in tapwater and tapwater- sand electrolytes.

Alloy icorr Ecorr Corrosion rate 10-4A/cm2 mV 10-2mmpy Tapwater UFW3 0.039±0.002 -380±2 0.023±0.001 UFW10 0.054±0.002 -387±2 0.038±0.001 UFW15 0.054±0.001 -395±3 0.043±0.001 Tapwater + Quartz sand UFW3 19±3 -421±2 11±2 UFW10 20±1 -434±2 14±4 UFW15 16±1 -398±3 13±2

1,E+01

tannic acid + sand 1,E+00

tapwater + sand 1,E-01

1,E-02 tapwater

1,E-03 tannic acid

1,E-04

1,E-05

1,E-06

1,E-07 -1000 -500 0 500 1000 1500 2000 2500 3000 3500 4000 Potential / mV Figure 6.26. Comparison of the polarization curves for hardmetal FW15 in a tannic acid concentration of 0.00588M and tapwater, with and without the addition of quartz sand.

6.3.1 Microstructural examination

Microstructural studies of the corroded specimen were done in an attempt to identify the corrosion mechanisms which take place. In order to compare the studies quantitatively, separate immersion studies were conducted solely for this purpose and the specimen were all polished to a 1 micron surface finish prior to testing.

6 Results 93

XRD analysis conducted on the hardmetals after corrosion detected no oxides on the corroded surfaces. XRD analysis was also conducted on the specimen corroded via the polarization tests with the same end result. This may validate the fact that no passivation occurs on these hardmetals in the electrolytes tested. Any corrosion products which may form either do not adhere to the surface at all or they may be too low or too thin to detect.

From the microstructural studies the following important observations were made for all the hardmetals tested in all the electrolytes; • the cobalt binder is preferentially removed from the corroding surface, • the tungsten carbide grains retain their sharp edges and are not attacked, • corrosion was found to be localized across the exposed surface, • in order of greatest to least attacked: FW15; UFW10; UFW3. • the tannic acid-sand electrolytes showed more aggressive microstructural attack than the tannic acid electrolytes.

The preferential removal of the cobalt binder is illustrated in figures 6.27 and 6.28 for alloy FW15 which was immersed in a tannic acid concentration of 0.00588M for 90 minutes. Figure 6.27 shows two optical light microscope micrographs of the cross-section of the hardmetal in the un-corroded and corroded state. The cobalt binder phase are the regions which appear to be white areas on the metal surface. In figure 6.27(b) it is clearly evident that these white areas are missing in great numbers from the edge of the alloy and have been replaced by black areas which represent the sites from which the cobalt binder has been removed from the hardmetal surface. The depth to which the binder is corroded in this picture is around 4 microns. The carbide grains appear to have been unaffected by the corrosion process.

Figure 6.27. Optical light micrographs showing the preferential removal of the cobalt binder from the surface region of the FW15 alloy. (a) uncorroded condition. (b) corroded condition.

Figure 6.28 shows two scanning electron micrographs of the specimen surface in the un-corroded and corroded state. As both holes and cobalt binder regions appear black in the these micrographs, it is hard to distinguish between the two. However, regions can be detected where the cobalt has been removed and exposes the underlying tungsten carbide grains. These regions of exposed carbide grains have been circled in figures 6.28(a) and 6.28(b) to make the comparison easier.

6 Results 94

Figure 6.28. Scanning electron micrographs showing the preferential and localized removal of cobalt from alloy FW15. (a) un-corroded condition. (b) corroded condition.

Figure 6.29 shows two scanning electron micrographs of a FW15 specimen surface in the un- corroded and corroded state after being immersed for two hours in a tannic acid concentration of 0.00588M to which silicon dioxide sand had been added. The edges of the tungsten carbide grains and, in some cases, whole grains appear white in the micrograph, which is an indication that the cobalt has been removed from these areas. The two micrographs were taken under the same conditions, which implies that the contrast seen in figure 6.29(b) is due to corrosion and not to the microscope settings. From the observations made it was noted that the corrosive attack is not uniform across the surfaces.

Figure 6.29. Comparison of (a) uncorroded FW15 grade and (b) FW15 grade in corroded state after two hour immersion test.

In order to investigate further this localized effect, the immersion times were shortened. Figure 6.30 shows field emission electron micrographs for alloys UFW10 and FW15 in the un-corroded and corroded states. The metals were immersed for 1 hour in a tannic acid concentration of 0.00588M to which silicon dioxide sand had been added. Localized attack is clearly evident. In figure 6.30 the white edges of the carbide grains from which the cobalt binder had been removed is visible. The carbide grains themselves do not appear to have been attacked at all, as they retain their sharp edges.

6 Results 95

It appears that regions of the hardmetal which contain larger binder islands are attacked first. The reason for the localized attack is not clear and may be linked to various aspects such as the amount of tungsten in solution in the binder phase. It has been shown in literature [99] that a tungsten depleted zone can exist at the binder/carbide phase boundaries and a decrease in tungsten implies a greater sensitivity to corrosive attack. It is possible therefore that certain areas of the corroded hardmetals have zones of depleted tungsten at the interfaces and are thus more susceptible to corrosive attack. It could also be related to the thickness of the binder phase, which may be thinner in some areas, and as a result these regions would be depleted of cobalt easily than those which have thicker cobalt layers.

(a) (b)

(c) (d)

(e) (f) Figure 6.30. Corroded (a) UFW10 uncorroded; (b)-(c) UFW10 corroded; (d) FW15 uncorroded; (e)- (f) FW15 corroded.

6 Results 96

6.3.2 Summary of the main findings

• the corrosion rates increase with increasing binder content and binder mean free path. • the corrosion rates increase linearly with increasing tannic acid concentration. • no passivation occurs in any system tested. • the critical current density increases with increasing binder mean free path and is reached at higher potentials as the binder mean free path increases. • at higher potentials tungsten carbide oxidation may occur when the surface layer is partially denuded of cobalt. • the corrosion rates for the tannic acid-sand and the tapwater-sand electrolytes are approximately one and a half decades higher than those for the tannic acid and the tapwater solutions. • impurities present on the sand increase the conductivity of the liquid-sand solutions. • the rate of increase in the acid-only corrosion rates are more sensitive to increasing tannic acid concentration than that of the acid-sand rates for alloys UFW10 and FW15. The rate of increase for alloy UFW3 was similar in both solutions • microstructural studies revealed that - the cobalt binder is preferentially removed from the corroded surface - the tungsten carbide grains retain their sharp edges and are not attacked - corrosion was found to be localized across the exposed surface - the surfaces exposed to the tannic acid-sand electrolytes showed more aggressive microstructural attack than those exposed to the tannic acid electrolytes.

6.4 Abrasive wear and corrosive abrasive-wear tests

This section describes the results of the abrasive and corrosive-abrasive wear behaviour of the hardmetal grades UFW3, UFW10 and FW15. The wear response of the alloys are presented with respect to the varied test parameters in the following order; • comparison of wear rates under dry, water and tannic acid conditions, • comparison of wear rates under varied tannic acid concentration, • comparison of wear rates under varied applied normal force and • comparison of wear rates under varied abrading wheel speed.

In this section use will be made of the following terms, ‘standard wear rate’ and ‘corrosive wear rate’. The term ‘standard wear rate’ refers to the wear tests conducted using water as the liquid medium. The term ‘corrosive wear rate’ refers to the wear tests conducted using tannic acid as the liquid medium.

6 Results 97

6.4.1 Comparison of the wear rates under dry, water and tannic acid conditions

The wear rates of the three hardmetals under dry, water and tannic acid conditions are listed in table 6.11, along with the corresponding friction force, surface roughness and friction co-efficient values. These tests were conducted using an applied normal force of 100N and an abrading wheel speed of 1.02m/s. Each value represents the average of at least 3 tests.

Table 6.11. Wear data for the hardmetals UFW3, UFW10 and FW15.

Wear test condition Wear rate Friction co- Friction force Surface roughness: efficient: µ RZ 10-15m3/m N µm UFW3 Dry 623±12 0.58±0.04 57.74±4 1.78±0.23 Water 178±19 0.31±0.02 30.78±2 0.84±0.19 0.00588M Tannic acid 333±17 0.32±0.02 32.13±2 1.11±0.02 UFW10 Dry 1440±22 0.50±0.03 49.53±3 1.41±0.04 Water 229±27 0.31±0.01 31.47±1 1.01±0.11 0.00588M Tannic acid 345±34 0.32±0.02 32.17±2 1.19±0.12 FW15 Dry 14600±32 0.50±0.03 49.49±3 6.53±1.02 Water 1260±53 0.34±0.01 34.04±1 3.52±0.14 0.00588M Tannic acid 1530±51 0.35±0.01 34.50±1 4.08±0.28

From the table it can be seen that the wear rates measured under dry conditions are far higher than those measured under the wet1 conditions for all the hardmetals. In addition, the wear rates measured in the tannic acid are greater than those measured in the water for all three alloys. The values obtained for the corresponding surface roughness, co-efficient of friction and the friction forces are highest in the dry condition, but similar values are obtained for the water and tannic acid conditions for each hardmetal case. The decrease in the wear rates from dry to wet conditions are an indication that the liquid acts a ‘lubricant’ during the wear test. This is confirmed by the lower friction forces measured during the wet conditions. A reduction in friction implies a reduction in surface contact between the two surfaces and hence less wear occurs. In addition, the liquid increases the removal of wear debris, which would otherwise act as abrasive particles and possibly add to the wear process. The difference in the wear rates between those in water and those in acid may be primarily attributed to the corrosive influence of the acid.

1 Wet conditions refers to wear testing conducted using water and/or tannic acid

6 Results 98

The wear rates for the alloys are displayed graphically in figure 6.31. Alloy FW15 has higher wear rates than the two ultrafine-grained alloys under each condition tested. The UFW10 grade has a higher dry wear rate than the UFW3 alloy. However, the wear rates for these two alloys determined under wet conditions are similar. The general trend of increasing wear rates with decreasing hardness and increasing cobalt content, shown in this figure, is similar to that shown by previous authors [chapter 4.6].

100000 Dry Water 0.00588M Tannic acid

10000

1000

100

10

1 UFW3 UFW10 FW15 Decreasing hardness and increasing cobalt content

Figure 6.31. Comparison of wear rates for the alloys under dry, water and tannic acid conditions.

The dry wear rate of the FW15 grade is 23 times greater than that of the UFW3 grade and 10 times higher than that of the UFW10 grade. The UFW10 grade’s dry wear rate is approximately 2.3 times higher than that of UFW3. The standard wear rate of FW15 is approximately 5-7 times greater than that of the ultrafine alloys. A comparison of the corrosive wear rates show that the FW15 wear rate is between 4-5 times greater than that of the ultrafine alloys. The difference for the corrosive wear rates is lower than for the standard wear rates. In the ultrafine alloys, an increase in the binder content by a factor of 3, results in a 30% increase in the standard wear rate while a corrosive wear rate increase of only 4% is measured.

The difference between dry and wet wear rates is about 10% for alloy FW15, for hardmetal UFW10 the difference is between 15-25%, while UFW3 showed the greatest difference between 30-50%. This can be an indication of the good wear resistance achievable by harder, finer-grained metals under lubricated conditions. It can also be deduced that the difference between standard and corrosive wear rates for each hardmetal increases with decreasing binder mean free path and decreasing cobalt content. Changing from water to tannic acid produced an increase in the wear rates of 20, 50 and 80% respectively for alloys FW15, UFW10 and UFW3 respectively.

6 Results 99

6.4.2 Comparison of the wear rates under varied tannic acid concentration

The corrosive wear rates in which the tannic acid concentration was varied are listed in table 6.12 and illustrated in figure 6.32 for the alloys UFW10 and FW15. Testing on hardmetal UFW3 was not carried out as it was decided that evaluation of the influence of corrosion on wear could be done in more detail when the samples used were limited to two alloys. It was thus decided to compare the wear response of the ultrafine-grained alloy(UFW10) to that of the fine-grained alloy(FW15). The tests were done using an applied normal force of 100N and an abrading wheel speed of 1.02m/s.

Table 6.12. Corrosive wear data for the hardmetals UFW10 and FW15 with respect to variation of the tannic acid concentration.

Tannic acid Wear rate Co-efficient of Friction force Surface roughness: concentration friction µ RZ M 10-15m3/m N µm UFW10 0.00118 269±23 0.32±0.01 31.67±1 1.23±0.17 0.00294 306±27 0.32±0.02 32.35±2 1.22±0.16 0.00588 345±34 0.32±0.02 32.17±2 1.19±0.12 0.0118 376±18 0.32±0.01 32.26±1 1.22±0.11 FW15 0.00118 1390±39 0.34±0.01 34.25±1 3.87±0.11 0.00294 1440±43 0.34±0.02 34.44±2 4.04±0.14 0.00588 1530±51 0.35±0.01 34.50±1 4.08±0.28 0.0118 1805±37 0.35±0.01 34.63±1 4.40±0.12

From the wear data it is observed that the corrosive wear rates increase as the tannic acid concentration increases for both alloys. The friction force and co-efficient of friction remained constant at all acid concentrations tested. This may imply that the viscosity of the acid does not increase such that it influences the friction forces and the contact zones. The surface roughness values for alloy UFW10 remain constant, while a slight increase in the roughness values is noted for alloy FW15 between the first two and the last two tannic acid concentrations. However, this may be due to experimental scatter.

6 Results 100

2000

y = 4x + 1326 FW15 1600

1200

800

y = x + 272 400 UFW10

0 0 20 40 60 80 100 120 140 Tannic acid concentration / 10-4 M

Figure 6.32. Corrosive wear rates as a function of increasing tannic acid concentration for hardmetals UFW10 and FW15.

In figure 6.32, the trend of increasing wear rate as the tannic acid concentration increases can be clearly seen. The aggressive nature of the sand-acid mixture was shown in the corrosion tests, and this would most likely influence the wear rates of the alloys. The corrosive wear rates obtained for the FW15 alloys are approximately 4-5 times greater than those obtained for the UFW10 alloys at all the acid concentrations tested. The scatter in the FW15 wear data is almost double that obtained for the UFW10 grades.

From the graphs it appears that the FW15 grade is more sensitive to increasing tannic acid concentration than the UFW10 grade. However, an examination of the relative wear rates with respect to increasing acid concentration, show that the opposite is true. This is illustrated in figure 6.33, where the corrosive wear rates are expressed as normalized values with respect to the value measured at the lowest acid concentration. This normalized graph reveals that the rate at which the corrosive wear rate increases as the tannic acid concentration increases is higher for the finer- grained UFW10 grade than the coarser-grained FW15 alloy. An increase in the corrosive wear rate of about 45% is measured for the UFW10 alloy while an increase of approximately 30% is calculated for alloy FW15.

6 Results 101

2,0

1,5 UFW10

FW15 1,0

0,5

0,0 0 20 40 60 80 100 120 140 Tannic acid concentration / 10-4 M

Figure 6.33. Normalized corrosive wear rates for hardmetals UFW10 and FW15 with respect to increasing tannic acid concentration.

6.4.3 Comparison of the wear rates under varied applied normal force

The standard and corrosive abrasive wear rates with respect to increasing applied normal force are listed in table 6.13 for the hardmetals UFW10 and FW15. Testing was done using an abrading wheel speed of 1.02m/s and for the corrosive wear tests a 0.00588M concentration of tannic acid was used. The corresponding friction force and surface roughness values are also given. The stress exerted on the specimen due to the applied force was calculated and is given in table 6.13. From the data it can be seen that the wear rates, friction forces, surface roughness and applied stress all increase with increasing normal force.

6 Results 102

Table 6.13. Standard and corrosive abrasive wear data for UFW10 and FW15 with respect to applied normal force variation.

Applied normal Wear rate Applied stress: σ Friction force Surface roughness: RZ force N 10-15m3/m MPa N µm UFW10: Water 20 142±23 0.15±0.002 6.03±3 0.58±0.12 40 153±19 0.31±0.005 12.60±1 0.67±0.02 60 175±22 0.46±0.007 18.94±2 0.74±0.13 80 190±22 0.62±0.009 25.85±1 0.83±0.04 100 229±27 0.77±0.012 31.47±1 1.01±0.11 120 246±33 0.92±0.014 36.19±1 1.05±0.12 UFW10: 0.00588M Tannic acid 20 211±24 0.15±0.002 6.66±3 0.68±0.09 40 233±17 0.31±0.005 13.50±1 0.83±0.05 60 286±21 0.46±0.007 19.65±1 0.98±0.12 80 328±38 0.62±0.009 26.53±2 1.01±0.15 100 345±34 0.77±0.012 32.17±2 1.19±0.12 120 414±51 0.92±0.014 36.29±1 1.35±0.16 FW15: Water 20 320±47 0.15±0.002 6.14±2 1.38±0.07 40 445±58 0.31±0.005 14.40±1 1.85±0.13 60 706±43 0.46±0.007 20.93±1 2.28±0.16 80 951±47 0.62±0.009 27.19±2 2.92±0.05 100 1260±53 0.77±0.012 34.04±1 3.52±0.14 120 1500±31 0.92±0.014 43.21±2 4.21±0.28 FW15: 0.00588M Tannic acid 20 373±53 0.15±0.002 6.94±1 1.60±0.06 40 650±41 0.31±0.005 14.25±3 2.08±0.18 60 922±48 0.46±0.007 20.68±1 2.76±0.25 80 1240±56 0.62±0.009 28.27±2 3.31±0.38 100 1530±51 0.77±0.012 34.50±1 4.08±0.28 120 1895±38 0.92±0.014 42.03±1 4.59±0.25

In figure 6.34(a) the effect of doubling the applied normal force, from 60 to 120N, on the standard abrasive wear rate is illustrated. It was observed that a two-fold increase in the force led to a 41%

6 Results 103 increase in the standard wear rate for UFW10 and a 210% increase for FW15. At 60N, the FW15 wear rate is about 4 times higher than that of UFW10 and at 120N, it is 6 times higher.

1800 2500 60N - water 1600 120N - water 2000 1400 tannic acid

1200 1500 1000 120 N 800 60 N water 1000 600

400 500

200

0 0 UFW10 FW15 UFW10 FW15 (a) (b) Figure 6.34. The effect of doubling the applied normal force on the standard and corrosive wear rates of the alloys.

The more force applied to a surface the greater the resulting damage will be and as a result the wear rate increases. In addition, at low loads, the amount of each abrading particle making real contact with the surface is small. Asperity contact takes place with very little crushing or grinding of the particles against the sample surface. At higher loads the contact area increases, thus creating more wear on the surface. This effect was confirmed by studying the used abrasive particles after wear testing, using optical light microscopy and it is shown in figure 6.35. At higher loads the abrasive particles were crushed and fragmented compared to the original particles.

(a) (b) Figure 6.35. Silicon dioxide abrasive sand after wear test using (a) 60N applied normal force and (b) 120N applied normal force.

When tannic acid is used in the wear system, it is once again shown to have an increasing effect on the wear rate as is displayed in figure 6.34(b). A two-fold increase in the force led to a 45% increase in the corrosive wear rate for UFW10 and a 206% increase for FW15. At 60N, the FW15 corrosive wear rate is about 3 times higher than that of UFW10 and at 120N, it is 5 times higher. With an increase in the load the effect of the acid also increases. These increases are most likely due to the greater amount of hardmetal surface deformation at the higher loads, which not only increases the surface area and dislocation density, which facilitate corrosion, but also leads to more cracks and fissures being formed, which provide additional access routes into the material for the acid.

6 Results 104

In figure 6.36, the results for both hardmetals are shown at all loads tested, in both water and tannic acid. A linear relationship is found between the wear rate and the applied load, which is in good agreement with trends found in literature. The FW15 wear rates increase by a factor of 5 while those of the UFW10 grade increase by a factor of 2. The scatter for the FW15 grade is almost double that for grade UFW10.

2000 FW15 - tannic acid

FW15 - water

1600 UFW10 - tannic acid UFW10 - water

1200

800

400

0 0 20 40 60 80 100 120 140 Applied normal load / N

Figure 6.36. Standard and corrosive wear rates as a function of increasing normal applied force for the hardmetals UFW10 and FW15. Corrosive wear tests were conducted in a tannic acid concentration of 0.00588M.

The tannic acid has an increasing effect on the wear rates across the range of loads tested. As the load tends towards zero, one would expect a zero wear rate, but this is not the case, because even when no force is applied corrosion still occurs. Due to the tannic acid being a weak organic acid, the difference between the two curves may tend to become negligible the closer the loading tends towards zero.

At low loads the standard and the corrosive wear rates are similar. As the load increases, the curves deviate significantly. This implies that there is a possible interaction between abrasion and corrosion and the degree of this interaction increases as the load increases. In order to assess this synergistic effect, the wear rates for each hardmetal under each condition were normalized with respect to the wear rate measured at the lowest load for that condition. The results of this normalization are plotted in figure 6.37(a). From these curves it is clear that the corrosive wear rates are sensitive to increasing applied loads, hence showing that the interaction between corrosion and wear increases as the load increases, for the range of loads tested.

The range of reasons for this synergistic effect include the surface roughness contribution, the results of which are shown in figure 6.37(b). The surface roughness increases as the load increases. This implies an increase in the contact area between abrasive and hardmetal, which

6 Results 105 leads to increased wear rates. In addition, according to Faraday’s Law for corrosion, the corrosion rate is proportional to the surface area. Therefore an increase in the contact area through a roughness increase, facilitates corrosion, leading to an increased corrosion sensitivity as the load increases. For alloy UFW10 the roughness increased by a factor of 2, while it increased by a factor of 3 for FW15.

6 6 FW15 - tannic acid 5 5 FW15 - water UFW10 - tannic acid 4 4 FW15 tannic acid UFW10 - water FW15 water 3 3

UFW10 tannic acid 2 2

UFW10 water 1 1

0 0 0 20406080100120140 0 20 40 60 80 100 120 140 Applied normal load / N Applied normal load / N (a) (b) Figure 6.37. (a) Normalized wear rates for alloys UFW10 and FW15. (b) Surface roughness for the alloys UFW10 and FW15.

The role of the friction force during the wear tests is also important and it was found that the friction force increased as the applied load increased. An interesting result was found when comparing the friction forces measured in the tannic acid and those measured in the water. It was found that the friction forces were more sensitive to increasing force under the standard wear conditions than under the corrosive wear conditions. This result is displayed in figure 6.38 in the form of normalized curves. As the applied force increases, the rate at which the friction force increases is higher under the standard wear conditions, than under the corrosive wear conditions. This was found to be true for both alloys. This implies a possible lubrication effect from the acid which decreases the friction between the abrasive and the hardmetal surface. The lubrication effect may be linked to the acid being more viscous than the water. This theory may be validated with the use of the Stribeck curve, which shows that an increase in liquid viscosity, within limits, lessens the friction. The effect may also be linked to the organic nature of the acid, which provides better lubrication than the water molecules.

8 8

7 7 6 Water 6 5 5 Water 4 4

3 3 Tannic acid Tannic acid 2 2

1 1

0 0 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Applied normal load / N Applied normal load / N (a) (b) Figure 6.38. Normalized friction forces for hardmetals (a) FW15 and (b) UFW10.

6 Results 106

6.4.4 Comparison of the wear rates under varied abrading wheel speed

The standard and corrosive-abrasive-wear rates with respect to increasing abrading wheel speed are listed in table 6.14 for the hardmetals UFW10 and FW15. Testing was done using an applied normal force of 100N and for the corrosive wear tests a 0.00588M concentration of tannic acid was used. The corresponding friction force, co-efficient of friction and surface roughness values are also given. From the data it can be seen that the wear rates increase with increasing wheel speed, while the friction force and co-efficient of friction remain constant. The surface roughness increases slightly as the speed increases.

Table 6.14. Standard and corrosive abrasive wear data for UFW10 and FW15 with respect to abrading wheel speed variation.

Abrading wheel Wear rate Friction force Co-efficient of Surface roughness: RZ speed friction: µ m/s m3/m N µm UFW10: Water 0.51 145±16 31.69±1 0.32±0.01 0.61±0.04 0.61 160±18 31.44±1 0.31±0.01 0.70±0.14 0.71 169±30 31.60±1 0.32±0.01 0.81±0.08 0.82 182±22 31.57±2 0.32±0.02 1.00±0.13 0.92 208±23 31.92±2 0.32±0.02 0.99±0.26 1.02 229±27 31.47±1 0.31±0.01 1.01±0.11 UFW10: 0.00588M Tannic acid 0.51 204±25 32.55±1 0.33±0.01 0.72±0.03 0.61 239±23 31.90±1 0.32±0.01 0.84±0.06 0.71 266±19 32.01±3 0.32±0.03 0.97±0.25 0.82 281±31 32.76±1 0.33±0.01 1.11±0.12 0.92 302±27 32.15±1 0.32±0.01 1.11±0.11 1.02 345±34 32.17±2 0.32±0.02 1.19±0.12 FW15: Water 0.51 473±51 33.70±1 0.34±0.01 2.87±0.15 0.61 564±37 33.80±2 0.34±0.02 2.95±0.08 0.71 645±46 33.36±1 0.33±0.01 3.18±0.13 0.82 836±43 33.93±1 0.34±0.01 3.30±0.05 0.92 1070±47 34.23±1 0.34±0.01 3.41±0.13 1.02 1260±53 34.04±1 0.34±0.01 3.52±0.14 FW15: 0.00588M Tannic acid 0.51 557±47 34.34±2 0.34±0.02 3.06±0.17 0.61 692±43 33.38±1 0.33±0.01 3.19±0.21 0.71 880±46 34.42±2 0.34±0.02 3.53±0.07 0.82 1100±39 34.35±1 0.34±0.01 3.82±0.20 0.92 1335±43 33.94±1 0.34±0.01 3.97±0.10 1.02 1530±51 34.50±1 0.35±0.01 4.08±0.28

6 Results 107

In figure 6.39(a), the effect of doubling the abrading wheel speed, from 0.51 to 1.02m/s, on the standard abrasive wear rate is illustrated. It is observed that a two-fold increase in the speed led to a 41% increase in the standard wear rate for UFW10 and a 266% increase for FW15. The FW15 wear rate is about 3 times higher than that of UFW10 at 0.51m/s and at 1.02m/s, it is 6 times higher.

1400 1800 0.51m/s - water 1600 1200 1.02m/s - water tannic acid 1400 1000 1.02m/s 1200 water 800 1000 0.51m/s 600 800 600 400 400

200 200

0 0 UFW10 FW15 UFW10 FW15 (a) (b) Figure 6.39. Effect of doubling the abrading wheel speed on the standard and the corrosive wear rates for each alloy.

When tannic acid is used in the wear system, it is once again shown to have an increasing effect on the wear rate as is displayed in figure 6.39(b). This increase is most likely due to the increase in surface deformation as the speed increases, which not only increases the surface area and dislocation density, which facilitate corrosion, but also leads to more cracks and fissures being formed, thus providing more access routes into the material for the acid. However, this cannot be the only factor since the surface roughness did not increase as much as it did under the increasing load wear tests.

A two-fold increase in the speed led to a 51% increase in the corrosive wear rate for UFW10 and a 275% increase for FW15. At 0.51m/s, the FW15 corrosive wear rate is about 2 times higher than that of UFW10 and at 1.02m/s, it is 4 times higher. For hardmetal UFW10, the difference between the standard and corrosive wear rates is approximately 58% at 0.51m/s and 70% at 1.02m/s. For FW15 the difference was found to be 17 and 21% respectively. Therefore this implies that as the speed increases, the synergistic effect between acid and abrasion increases.

In figure 6.40, the results for both hardmetals are shown at all speeds tested in both water and tannic acid. An exponential trend is found for the speed-wear rate relationship. In literature, there is no general agreement concerning the speed-wear rate relationships which are tribosystem dependent. In majority of the literature reviewed however, a trend of increasing wear rates with increasing speeds has been seen. The FW15 wear rates increase by a factor of 2.7 while that of the UFW10 grade increase by a factor of 1.6.

6 Results 108

1800 FW15 - tannic acid

FW15 - water 1500 UFW10 - tannic acid y = 202e2x UFW10 - water 1200

y = 167e2x

900

600

y = 130ex

300

y = 92ex 0 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 1,2 Abrading wheel speed / m/s

Figure 6.40. Standard and corrosive wear rates as a function of increasing abrading wheel speed for hardmetals UFW10 and FW15. Corrosive wear tests were conducted in a tannic acid concentration of 0.00588M.

A similar trend showing that the standard and the corrosive wear rates are closer at lower speeds and separate as the speed increases is observed. By plotting normalized curves, the effect of the tannic acid can be identified more clearly. This is illustrated in figure 6.41(a). The rate at which the wear rates increase as the speed increases is higher under corrosive wear conditions than under standard wear conditions for both hardmetals.

3,0 6 FW15 - tannic acid FW15 - tannic acid 5 2,5 FW15 - water UFW10 - tannic acid 4 FW15 - water UFW10 - water 2,0 UFW10 - tannic acid 3 1,5 2 UFW10 - water 1,0

Surface roughness(Rz) / um 1

0,5 0 0,4 0,6 0,8 1,0 1,2 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 Abrading wheel speed / m/s Abrading wheel speed / m/s (a) (b) Figure 6.41. (a) Normalized wear rates for alloys UFW10 and FW15. (b) Surface roughness for the alloys UFW10 and FW15.

The surface roughness, which is depicted in figure 6.41(b), shows a small increase as the speed increases. Thus, the contact surface increases and the corrosion process is favored. However, since the roughness increase is low, it cannot be the single dominant driving factor behind the corrosive wear rate sensitivity. In addition, a rather interesting result is once again found, this time with respect to the roughness values. From figure 6.41(b) it appears as though the roughness of

6 Results 109 the FW15 grade is more sensitive to increasing speed since it has the steeper gradient. However, the rate at which the surface roughness increases as the speed increases is higher for the finer- grained UFW10 alloy, where an increase of approximately 60% is measured, while the roughness of the FW15 grade increases by only 20%.

6.4.5 Microstructural examination

Microstructural studies of the worn specimen were done to assist in identifying the wear mechanisms which take place. Comparisons between the standard and the corrosive wear conditions are made in addition to comparisons between the different hardmetal grades. In general, the wear features are found to be similar, with a few notable exceptions. The microstructures from varying tannic acid concentration, varying applied normal load and varying abrading wheel speed are similar, with only the severity of the damage differing.

The as-received condition for the hardmetal FW15 is shown in figure 6.42 at three different resolutions. The surface appearance is similar for alloys UFW3 and UFW10. The surface has been ground with the main grinding lines visible, giving it a rough appearance. A few pits and some small furrows are randomly found across the surface. These features are expected to facilitate the damage initiation process during the wear tests.

pits

Grinding lines

Figure 6.42. Hardmetal FW15 in the as-received condition.

6 Results 110

Worn surfaces in the dry condition are shown in figure 6.43 for alloy FW15. By comparing figures 6.42(a) to 6.43(a), 6.42(b) to 6.43(b) and 6.42(c) to 6.43(c) and 6.43(d), it is seen that the roughness of the surface in the dry state has increased significantly due to the abrasive action of the silicon dioxide particles. Holes and depressions, in the order of 1 to 2 microns, can be found throughout the worn surface, indicating regions where whole grain removal took place. ‘Pockets’ of wear debris, consisting of fragmented carbide grains, smeared cobalt and silicon sand particles, were found at random sites. Wear furrows are also seen which indicate the path taken by the silicon dioxide particles during the wear test. The profile of the furrows give the impression that the sand particles cut through the material. From EDS analysis it was confirmed that single sand particles were often found lodged in holes and in between carbide grains. This can be attributed to the sand particles being pushed into the hardmetal by the applied force via the abrading wheel.

holes cracks

(a) (b)

(c) (d) Figure 6.43. Worn FW15 surfaces in the dry condition. Circled areas in (c) represent wear debris, while the circled region in (d) shows a cracked tungsten carbide grain. Line in (d) shows the direction of wear, in which a particle cut through the hardmetal.

The majority of the carbide grains are smaller than the nominal carbide grain size indicating that the original grains have been broken down into smaller pieces. Carbide grains are also seen with transgranular cracks running along their lengths. Fragmented and chipped carbide grains give the impression that most of the cracking was transgranular. The size range of the broken grains are between 0.1 to 1micron indicating the severity of the damage incurred by the abrasive sand under dry conditions. A number of carbide grains were observed protruding from the surface and remain fixed to the bulk hardmetal on their underside. This gives the impression that the cobalt binder has been preferentially removed from between the carbide grains during the wear process.

6 Results 111

It was found that the worn FW15 surface was rougher than that of the UFW10 grade which was in turn rougher than that of the UFW3 grade (see table 6.11). The extent of grooving is greater in the FW15 which may be attributed to its lower hardness and the fact that the FW15 hardmetal hardness is more comparable to that of the abrasive sand. As a result, the abrasive is able to inflict more damage onto this alloy than the two ultrafine alloys.

The worn surfaces in the standard(water) and corrosive(tannic acid) conditions show similar features to those found in the dry condition, with a few exceptions. The surfaces in the dry conditions are much rougher than those in the wet conditions, with more grooving and holes. There is a greater amount of wear debris and lodging of silicon sand particles in the holes on the surfaces in the dry state. This is to be expected since the liquids in the wet states help to remove the debris from the worn surfaces during the test. Figure 6.44 gives the general impression of the worn regions in the wet states. The micrographs represent features common to all the hardmetals.

crack

(a) (b)

crack

holes depressions crack (c) (d) Figure 6.44. Comparison of the worn surfaces in various conditions. (a) as received FW15; (b) UFW10 in the corrosive wear state; (c) UFW3 in the standard wear state; (d) FW15 in the corrosive wear state. The cracks run perpendicular to the arrowheads in (b) and (d).

By comparing figure 6.44(a), the unworn state, with figure 6.44(b), the worn state, it is evident that the grinding lines have been removed and that the worn surface has a less rougher appearance. Practically no holes are visible in figure 6.44(b). However, In figures 6.44(c) and 6.44(d), the holes and depressions are clearly visible. In figure 6.44(c), the regions which appear dark grey in color are shallow depressions, i.e. regions which vary between 1 to 10microns in diameter and are approximately 0.3 to 1micron deep. This type of depression was not found in the dry state. Figure 6.44(d) shows a significant number of holes where grain removal occurred either by fall-out or rip- out. Faint groove lines can be seen in figure 6.44(d) which indicate the direction of the sand

6 Results 112 particle flow. These three types of regions depicted in figures 6.44(b)-(d) were found to be typical of the worn surfaces. Hardmetal FW15 had more holes and grooves than the ultrafine grades, which may once again be attributed to the differences in metal and abrasive hardness and the type of wear mechanisms which take place. Cracks in the order of 5 to 10 microns can be seen on the worn surfaces.

Figures 6.45 and 6.46 depict the wear features mentioned previously in more detail. Figure 6.45 illustrates the shallow depression regions, while figure 6.46 gives a more general impression of the worn surfaces. By closer examination of the depressions it can be seen that majority of the darker regions are what can be described as cobalt ‘tribofilms’. The remainder of the dark regions are holes which, due to their depth, appear darker than the rest of the surface. The tribofilms are transparent such that the underlying microstructure is visible, which is clearly seen in figure 6.45(e). Confirmation that the films comprise only cobalt is not easy since tungsten peaks also show up in the EDS spectrums. However, the tungsten peaks can be attributed to the cobalt film being thin thus allowing tungsten peaks to be detected from the carbide grains underneath. This conclusion was drawn since the strength of the tungsten peaks from the EDS spectrum results varied across the films. The use of the word ‘tribofilm’ is done with caution, since the actual cause of these films is not clear. They may simply be binder material from which the carbide grains have been removed. However, if this were the case then the films would be expected to have a rougher appearance. These films however, have a very smooth appearance and their size is larger than the nominal carbide grain size. The film may result from cobalt being removed from between the carbide grains and smeared across these regions. This smearing is an indication of the ductile nature of the cobalt binder.

6 Results 113

(a) (b)

(c) (d)

tribofilm

(e)

Figure 6.45. Worn surfaces of the hardmetal FW15 in the standard wear state showing typical wear features. Line in (c) shows wear direction as abrasive cuts through hardmetal. Circled areas in (c) show chipped tungsten carbide grains. Cracks run perpendicular to the arrowheads in (b). The circled areas in (a) show tansgranular cracking of tungsten carbide grains.

6 Results 114

(a) (b)

hole

(c) (d)

(d) (e)

Figure 6.46. Worn surfaces of the hardmetal FW15 in the corrosive wear state showing typical wear features. Line in (d) shows wear direction as abrasive cuts through hardmetal. The circled areas in (e) show tansgranular cracking of tungsten carbide grains.

6 Results 115

A closer examination of the carbide grains in the micrographs reveal that on some of the grains, ductile cobalt lips are present. These represent regions where the carbide grains were either pulled out from the matrix or intergranular fracture took place leading to grain fall-out. A depression on the bottom-right corner of figure 6.45(e) shows where a cluster of grains were removed from the matrix. Small cracks between 1 to 5 microns in length can be seen in figures 6.45(a)-(d). The cracks are both intergranular and transgranular in form. A distinct wear furrow can be seen in figure 6.45(c) indicating the wear direction.

The similarity between the worn surfaces of the hardmetals is further depicted in figure 6.46, which represents the corrosive-worn surfaces. Figure 6.45 represented the standard worn surfaces and by comparing all the micrographs from both figures, it is seen that similar wear features are present in both conditions. By careful examination and comparison however, more holes and depressions were found in the corrosive-worn surfaces. Transgranular cracking of the carbide grains can be clearly seen in figure 6.46(e) for alloy FW15, which showed in general more cracked carbide grains than the ultrafine-grained hardmetals. A hole in the order of 1 micron in diameter is shown in figure 6.46(d). Such holes are a common feature on the worn surfaces.

Figure 6.47 shows two interesting features which were discovered on the worn surfaces. Figure 6.47(a) shows an enlarged tungsten carbide grain from which the top part has been chipped off due to abrasive action. This feature gives some insight into the type of mechanisms taking place in this region. The cobalt binder is first clearly removed from around the grain, which is then only being held in place on the underside. The abrading particles continue to wear the grain until the carbide grain shear forces are no longer able to withstand the abrasive action, which then cracks and breaks off. Figure 6.47(b) shows a type of smeared region which is a rare feature on the worn surfaces. This type of smearing is distinctly different to the general worn appearance shown in previous figures. The carbide grains appear to have been crushed into smaller fragments and have been pressed against the hardmetal surface together with cobalt binder and sand particles. This could likely be a large piece of wear debris which was caught between the sample and the abrasive particles.

(a) (b) Figure 6.47. (a) Enlarged carbide grain which has been chipped. (b) Smeared region consisting of crushed carbide grains, cobalt and abrasive sand particles.

6 Results 116

6.4.6 Summary of the main findings

• The wear rates of the hardmetals are highest under the dry conditions. • The corrosive wear rates of the hardmetals are higher than the standard wear rates. • The fine-grained 15wt% cobalt hardmetal exhibits higher wear rates than the ultrafine-grained 10wt% cobalt hardmetal which in turn has higher wear rates than the ultrafine-grained 3wt% cobalt hardmetal. This is found in all systems tested. • The wear rates increase with the varied system parameters as depicted in figure 6.48.

Acid concentration Load Speed

Figure 6.48: General effect of system parameter changes on the wear rates.

• When changing from dry to corrosive to standard wear conditions, the change in the wear rates is greatest for the UFW3 alloy and smallest for the FW15 alloy. • When changing from alloy UFW3 to alloy UFW10 to alloy FW15, the change in the wear rates is greater under the standard wear conditions than under the corrosive wear conditions.

• As the system parameters listed in table 6.15 increased, the friction force(FF), co-efficient of friction(µ), surface roughness(RZ) and applied stress(σ) changed as indicated in the table.

Table 6.15 System parameter Acid concentration Normal applied load Abrading wheel speed

FF constant increased constant µ constant constant constant

RZ constant increased increased σ constant increased constant

• The friction forces, the co-efficient of friction and the surface roughness are highest under the dry conditions for all the hardmetals. • The surface roughness is higher under the corrosive wear conditions compared to the standard wear conditions. • Friction force is more sensitive to increasing load under standard conditions than under corrosive conditions.

6 Results 117

• The ultrafine alloys(UFW) appear to be more sensitive to increasing tannic acid concentration than the fine-grained alloy(FW). • There is a synergistic effect between corrosion and wear which increases as the normal applied load and the abrading wheel speed is increased. • Wear features located on the worn surfaces include: - holes and depressions of varying sizes and types - abrasive sand particles lodged in worn surface - fragmented tungsten carbide grains - transgranular cracking of carbide grains - intergranular fracture - preferential removal of the cobalt - carbide grain pull-out - cracking of worn surface - presence of cobalt tribofilms - various types of metal smearing The type and the extent of wear features found on the worn surfaces is dependent on the hardmetal type and the tribosystem conditions.

7 Discussion 118

7 Discussion

In this chapter the abrasive wear response of the tungsten carbide hardmetals is discussed. The influence of the testing procedures used and the testing environment is evaluated. The resulting wear mechanisms due to the tribological systems are discussed. The wear response of the hardmetals are firstly evaluated under the woodcutting conditions, followed by a discussion of their corrosion behaviour and finally their wear response in the three-body abrasive wear system is discussed.

7.1 Evaluation of the wear response during the woodcutting tests

7.1.1 Influence of the wood-tool set-up on the wear process

The study of woodcutting tool wear requires an assessment of the wood-tool-machine system employed. Due to the varying combinations which exist for this system, generalizations between different tool wear studies can only be made with caution. Even though the types of tool wear encountered when cutting specific wood types may appear to be similar, the factors causing and facilitating the wear is both machine and working environment dependent. For this study, a milling machine with a jointing and rabbeting cutterhead was used for the woodcutting tests. The cutterhead rotated in the counter direction to the wood feed. The wood was fed manually from right to left, while the cutterhead rotated in an anti-clockwise direction.

The orientation of the wood grain to the cutting direction influences both the surface quality of the wood as well as the degree of wear on the tool faces. Three main orientation directions are possible and these are shown in figure 7.1(a). An intermediate orientation has also been included. Additional intermediate directions are also possible. The directions are defined by the direction of movement of the cutter with respect to the grain orientation as: • A: perpendicular to the grain • B: longitudinal to the grain • C: parallel to the grain

(a) (b) Figure 7.1. (a) Main cutting directions with respect to wood grain orientation. (b) Components of the direct cutting force [2].

In the woodcutting tests the cutter movement was in direction A. In orientation A, the expenditure of forces is greatest since the bonding between the wood grains is strongest in this direction [2].

7 Discussion 119

Hence, the maximum wear is expected on the woodcutting knives. With increasing wear and bluntness of the knives, the usually high strength and stretchable wood fibres, yield under the cutting forces and are prone to rupture more lengthwise than crosswise [2]. This leads not only to bad surface finishes in the wood, but shorter service lifetimes for the tools. The interaction in the cutting edge-wood contact region is thus expected to have an enormous effect on the type and extent of degradation mechanisms which emerge. The degradation mechanisms found on the worn cutting zones are clear evidence of the extent of the damage inflicted on the knives for this type of grain orientation even though only a few meters of wood were cut within very short cutting time intervals during this study. The cause of different types of chuncking and holes (for example see figure 6.11) found along the cutting zone as well as the degree of jaggedness of the wear profile can be attributed to this rupture of the wood fibres.

The type of interacting forces which are expected to be exerted in the wood-tool contact region during the cutting process were reviewed in chapter 4. The main components of the direct cutting force are shown in figure 6.1(b). These forces which are generally very high together with a possible change in orientation of fibre rupture during cutting may be the cause of the smeared regions seen along the rake face of the woodcutting knives when used to cut cured Spruce (see figure 6.9).

7.1.2 Influence of the wood type on the wear process

The woods cut during this investigation represent the three main wood categories, namely, fresh wood, cured wood and manufactured wood products. As was highlighted in chapter 4, each wood type is expected to cause different types of wear mechanisms in the cutting zones of the hardmetal woodcutting knives. This trait was evidently clear when comparing the worn cutting zones of the knives used to cut chipboard and those used to cut the natural woods. Before explaining the different wear mechanisms found it is useful to explain the nature of the contact made between the wood types and the cutting edges, since this directly influences the incurred damage.

When cutting wood, chip formation occurs by rupture of the wood fibres. In the cutting of natural woods a continuous chip is formed which slides over the rake face as cutting proceeds. In the machining of wood products such as chipboard, no continuous fibre is formed as the chipboard machines by fracture. Therefore in the case of the chipboard, there is direct contact between the cutting edge and the wood, whereas in the case of natural wood, direct contact between the wood and the cutting edge is restricted. This is depicted in figure 7.2. These two set-ups influence the relative magnitude of the various forces which emerge in the contact zone during the cutting process. In the cutting of chipboard there is no force exerted on the rake face due to chip motion, whilst in the cutting of natural wood, the direct cutting force exerted on the cutting zone is lower. This implies that more damage can be inflicted on the cutting zone during the cutting of chipboard, while more damage is inflicted on the rake faces of the knives used to cut the natural woods. Differing levels of damage severity on the rake faces was observed during microscopic examination of the tool cutting edges and was found to be in agreement with these expectations.

7 Discussion 120

(a) (b) Figure 7.2. Differences in the amount of contact between cutting edge and wood during the cutting of (a) chipboard and (b) natural wood.

The types of wear mechanisms found on the cutting knives were similar to those reported for tungsten carbide woodcutting tools in literature. It was observed that abrasion is the dominant wear process for both hardmetals under all conditions tested. However, additional information has been found in this investigation, since the hardmetals have finer carbide grain sizes than those reported in literature.

The cutting of chipboard led to a significant degree of fragmentation of the tungsten carbide grains, which appeared to be one of the dominant wear mechanisms. The tungsten carbide grains were fractured into smaller than nominal size grains, encompassing a wide size range, which can be attributed to an interaction between the abrasive nature of the board and the direct cutting forces in the contact zone. Many clusters of carbide grains were seen to be attached to the underlying hardmetal with no cobalt binder between them. This observation suggests a further wear mechanism operating during the machining of chipboard which is the preferential removal of cobalt from between the carbide grains. This removal of the cobalt, causes a decrease in the bonding forces within the matrix and facilitates carbide grain removal. It leaves the carbide grains exposed to further abrasive action which often leads to a rounding off of the sharp grain edges. These exposed carbide grains reach a critical point where they are no longer able to resist the shear forces being exerted on them, resulting in cracking and eventual removal from the matrix.

This mechanism of preferential cobalt removal has been cited in literature for conventional tungsten carbide alloys which have grains sizes larger than 1 micron and relatively large binder mean free paths. This mechanism proposes that the cobalt binder is extruded from between the carbide grains due to oscillation of the carbide grains which is caused by fluctuating forces during the cutting of the chipboard. This type of mechanism is then dependent on the hardmetal binder mean free path and the carbide grain size. It is unlikely to be valid for very small mean free paths as the carbide grains would have very little room to oscillate. Therefore, this mechanism of cobalt removal is not thought valid for the hardmetals investigated in this study. For this study, it is proposed that preferential removal of the binder occurs on a microscale. Small fractured chipboard particles were found in the worn cutting zones. These particles can act as micro-abrasives which act on the carbide and binder phases separately. Due to the lower hardness of the binder phase, it is preferentially removed from the matrix by mechanical action, leaving the surrounding carbides unsupported where they can be easily removed by further machining. This mechanism would explain why more clusters of carbide grains attached only on their undersides, were found in the cutting zones of the UFW3 hardmetal than the UFW10 hardmetal. The UFW3 has less cobalt and a smaller mean free path, hence there is less cobalt to remove from between the carbide grains and thus exposing a greater number of carbides to abrasive action.

Due to the very fine-grained microstructure it is also proposed that microcracking of the matrix occurs due to the abrasive nature of the chipboard contents. This cracking of the matrix facilitates the removal of whole sections of material by further abrasive action. This mechanism of

7 Discussion 121 microcracking will be explained in more detail in section 7.3.1 where the worn surfaces of the hardmetals under three body abrasive wear showed similar wear features.

The type of fragmentation of the carbide grains described for the cutting of chipboard was not found in the worn zones of the tungsten carbide tools during the cutting of natural woods. In these worn zones, the mechanism of carbide grain pull-out was more dominant even though some cracked carbide grains were observed. Carbide grain pull-out, either individually or in clusters of matrix material, is proposed due to the great number of ductile cobalt lips and holes found in the worn zones, especially during the cutting of the Spruce woods. Whether or not preferential removal of the cobalt occurs for the Spruce woods is difficult to ascertain. It is possible that during the cutting of the fresh Spruce, that the wood extractives removed the cobalt via a corrosive mechanism. However, the wear rates of the two wood types are so similar that it is believed that this mechanism is only minor if not absent altogether. In fact, no significant differences could be found between the two types of Spruce woods. This may be attributed to the small lengths of wood cut and the high cutting speeds employed. In literature differences in wear rates and wear mechanisms between cured and fresh wood of the same species is noted after cutting numerous lengths of wood [31].

Possible evidence for the corrosive effect of the wood extractives was seen when cutting the fresh Oak. It was found that the surface of the tungsten carbide worn cutting zones were denuded of cobalt in most areas. The tungsten carbide grains were also found to have retained their original sharp edges and very few cracked grains were found. This indicates that the cobalt was selectively removed from the matrix. The reasons for this could be due to an interaction between corrosive attack and mechanical removal. During the wear process, the acid begins its attack on the binder which leads to a softening of the matrix and a weakening of the bonds within the cobalt binder, as well as between the cobalt and the tungsten carbide. This weakening of the surface facilitates removal of the hardmetal by mechanical wear. The holes seen in the worn surfaces would also allow sub-surface corrosive attack to occur, which leads to a weakening of the underlying matrix and to greater material removal rates. This could explain why the fresh Oak which is more acidic gave rise to higher wear rates than the Spruce woods. The harder Oak would also assist in mechanical removal of the cobalt binder to a greater degree than the softer Spruce woods.

7.1.3 Comparison between the wear response of the two hardmetal woodcutting knives

In order to compare the wear response of the two hardmetals with different binder contents during woodcutting, it is important to consider the general wear progression first. In general, it was found that the sharp cutting edge wears off rapidly after only 1m of wood is cut, after which the knife wears at a progressively slower rate as more wood is cut. After cutting 6m of wood the wear rate had decreased by a factor of 10. This type of behaviour may be linked to applied stress and possibly work-hardening effects. At the beginning of the cutting process, the cutting edge makes asperity contact with the wood surface, which leads to the generation of high localized stresses. The magnitudes of the stresses generated are related to the forces produced in the wear region which are influenced by the high cutting speeds used. This initial contact creates high shear forces between the cutting edge and the wood and due to the relatively small thickness of the original cutting edge, it is rapidly removed. The surface area of the cutting zone on the knife has now increased and thus the forces produced during the cutting process are now distributed over a larger surface area and hence the force per hardmetal particle has been reduced. Consequently, the hardmetal is better able to resist wear and hence the wear rates fell.

From microscopic investigations of the worn surfaces it was seen that abrasion is the dominant wear mechanism for both hardmetals in cutting all the types of wood. The abrasion process is

7 Discussion 122 generally known to cause work-hardening of the abrading surfaces and work-hardening increases during the abrasion process until a critical strain is reached at which point the metal surface yields and fractures. It is thus likely that the worn surfaces work-hardened during the cutting process which would lead to a slowing down of the wear rates. It was however not possible to measure the hardness of the worn zones due to the microscale surface areas of the cutting zones.

During the woodcutting process, the cutterhead in which two woodcutting knives are clamped, rotates at high speeds. This rotational movement of the cutterhead may imply that the knives are continually impacting against the wood each time they enter the wood and cut through it. In addition, this movement exposes the hardmetal to both static and cyclic stresses. Although, the rotational speeds used are high enough to ensure that a smooth surface finish is obtained by means of clean fibre rupture [2], it is possible that during the initial contact of each knife, its impact resistance is of great importance. It was also noted that the profiles of the cutting zones had varying degrees of jaggedness, with the degree of jaggedness increasing with decreasing speed. This is may be a further indication that the knives impact against the wood leading to brittle or high stress fatigue fracture of the knife edges.

It is believed that the impact resistance of the hardmetal plays a role since the 3wt% cobalt hardmetals(UFW3) show greater initial blunting values than the 10wt% cobalt hardmetals(UFW10). -3/2 The UFW10 grade (KIc = 9.87MNm ) with its greater binder content is better able to resist the -3/2 initial impact than the UFW3 grade (KIc = 8.64MNm ), by absorbing greater impact energy in terms of elastic and plastic deformation. The UFW3 grade is harder and therefore more brittle and thus less able to sustain deformation. The only exception to this trend was found in the case of cutting chipboard where the UFW10 had a greater blunted edge than the UFW3 grade. This is most likely linked to an interaction between the impact and abrasion processes during the cutting process.

It has been previously mentioned that the contact area between the knives and the wood is greater in the case of chipboard than between the natural woods and the knives (see figure 7.2). During cutting of chipboard, in addition to impact, there is abrasive contact between the cutting edge and the wood. This abrasive contact is limited during the cutting of natural woods, and thus the role of impact resistance plays a greater role. Thus, it is more likely that due to the larger cobalt content and lower hardness of the UFW10 grade that it would suffer greater abrasive wear than UFW3. In cutting chipboard, the role of impact is less than that of abrasion. This also explains the general wear response of the two knives with respect to the type of wood cut. The UFW10 grades had generally high wear rates when used to cut chipboard and lower wear rates when used to cut the natural woods. The reverse was found to be true for the UFW3 alloys. Additionally, the UFW10 knives had lower wear rates than the UFW3 knives when used to cut the natural woods with the reverse being true in the case of chipboard.

Although, the UFW3 grades have higher blunting values than the UFW10 grades when cutting the natural woods, fewer abrasion scratches were found on the rake faces of the UFW3 grades. This is most likely due to the higher hardness of the UFW3 grade which is able to withstand the abrasive motion of the chips better than the UFW10 grades. However the UFW3 knives showed a greater tendency to chip than the UFW10 knives, and this can be attributed to their more brittle nature, because of their high hardness or lower fracture toughness. In addition, the wear profiles of the UFW10 grade used to cut chipboard showed more uneven wear regions than the UFW3 knives. This may be attributed to the greater amount of cobalt present in the UFW10 knives, which is then more susceptible to the abrasive nature of the chipboard than the 3wt% cobalt UFW3 knives.

7 Discussion 123

7.1.4 Comparison of the initial and the steady state wear stages

Due to the lack of data for the steady state wear stage it is impossible to draw any meaningful comparison between the two wear stages. The one significant comparison possible is that the hardmetal knives with the FeCoNi-binder performed better than the Co-binder knives. The wear rates were 20-50% lower and the worn zones were smoother. The cobalt binder knives showed substantial lateral cracking in the worn zones which was markedly absent from the FeCoNi-cutting zones. The exact cause of these cracks is unclear, but may be the result of a fatigue process due to the continuous exposure to cyclic stresses during the cutting process. In work done by Sailer [138] on the fatigue lives of these hardmetals, he showed that the FeNiCo-binder alloys had longer fatigue lives than the Co-binder knives.

A possible comparison between the initial and steady state wear process is thought a comparison of the wear profiles, which are much smoother in the steady wear stage. This indicates that the initial stages are really an incubation phase for the knives. A secondary comparison can be made for the UFW10 grade used to cut fresh Oak, which represents the incipient stage, and cured Oak which represents the steady stage. A comparison of the cutting edge width values, albeit for different wood types, show a factor of 4 difference between the cutting edge widths after 6m and after 3048m of wood has been cut. This indicates that after the initial rapid wearing off of the sharp edge, the knives then wear at slower rates which remain approximately constant even after hundreds of meters of wood have been cut. This result would however need to be validated by more experimentation.

7.2 Corrosion response of the hardmetals

The corrosion response of the tungsten carbide hardmetals can be described by comparing the hardmetal to a galvanic couple, in which the cobalt assumes the anodic role, therefore providing cathodic protection to the tungsten carbide. This assigning of roles is based on the tungsten carbide having a more noble position in the electrochemical series compared to the cobalt. In addition, from the micrographs shown in chapter 6.3, it was seen that the cobalt is selectively removed from the hardmetal matrix during corrosion. Due to the tannic acid being a weak organic acid, it is possible that the galvanic effect initiates and is limited to the phase boundaries between the cobalt and the tungsten carbide leading to intergranular attack.

Since no corrosion products were found on the surface, nor could they be detected by microscopic analysis, it is difficult to define the nature of the reaction between the cobalt and the tannic acid, which itself is an organic acid with a very complex structure. In simple terms, the anodic reaction can be described by metal dissolution taking place according to M M+ + e-. Hydrogen evolution is the probable cathodic reaction, although it may not be the only one. In addition, due to the exposure to air, the reduction of oxygen may also be taking place. The solutions were not de- aerated during corrosion testing, so that they would be representative of the environment found during the wear tests. The pH range for the tannic acid solutions used was between 2-4. It is possible that at the higher pH values, due to the acid being a very weak organic acid, that oxygen reduction takes place to a greater degree than hydrogen evolution.

It is possible that the cobalt metal forms chelation reactions with the tannic acid. Chelation is defined as the formation of a complex chemical between a metal ion and two or more polar groupings of a singe molecule, by which the metal ion is firmly bound with an organic molecule to

7 Discussion 124 form a ring structure [30]. Most metals form chelation reactions and thus the probability of a metal forming chelates with an organic acid must be high. The chemical properties of iron and cobalt are known to be similar, so it is expected that cobalt will undergo chelation with an organic acid as readily as iron. In the case of the iron chelates, the compounds formed are generally stable and insoluble. Due to the lack of corrosion products from the corrosion testing, it appears that the chelates formed between the cobalt and the tannic acid are soluble.

The influence of the impurities present in the acid-sand solutions were reflected in an increase in the corrosion rates by approximately two decades. The impurities exacerbate corrosive attack on the hardmetal. This was most likely caused by an increase in the conductivity of the tannic acid solution which led to a more aggressive attack on the hardmetal microstructure. The likelihood of the from the impurities forming any significant corrosion compounds is regarded as negligible, since the impurities were detected in amounts of less than 3g/ml.

No true passivation occurs in tannic acid in all the concentrations tested. Even if passivation did occur in these solutions, the passive film would not provide any protection during the wear process, since it would be removed by abrasive action. Therefore, replacing the binder phase with more corrosion resistant binders which offer protection by passive film formation would not be useful in woodcutting applications.

The corrosion rate is influenced by the relative areas of the cathode and the . In general, the relative area of the tungsten carbide is far greater than the cobalt, thus the corrosion resistance of these alloys is generally very good. The corrosion rate is then controlled by the amount of the cobalt binder phase and it’s properties. In this study it was found that higher cobalt contents and larger binder mean free paths led to higher corrosion rates. Since only three alloys were tested, no generalizations can be made for all hardmetals in tannic acid. However, in separate corrosion studies on hardmetals by Human [90] and Ghandahari [98], it was shown that the corrosion rate increased with increasing binder content and mean free path when using electrolytes such as sulphuric and phosphuric acid. However, earlier it was stated that the galvanic effect may be limited to the grain boundaries, hence it is expected that hardmetals with more phase boundaries would have higher corrosion rates. This is inconclusive and would have to be proven with a larger variety of grain size distributions.

Corrosive attack was found to be localized, i.e., areas on the surface were selectively attacked. The exact reasons for this localized attack is unclear. It may be linked to the properties of the binder phase such as dislocation densities and the amount of dissolved tungsten in the binder phase. A high dislocation density would favor corrosion and it is possible that where the binder mean free path is thinnest, that the binder is under greater constraint due to the surrounding carbide grains and hence attack may begin in these areas. Dissolved tungsten in the binder phase would lead to a decrease in the corrosion rate [90]. It is possible that the distribution of the dissolved tungsten may not be uniform across the binder, allowing some regions to be attacked faster. In addition, the possibility of gas bubble formation on the surface during testing may prevent or increase corrosion.

7 Discussion 125

7.3 Evaluation of the wear response during the three body abrasion tests

7.3.1 Evaluation of the general wear process for the investigated system

During the wear test, the abrasive particles make contact with the hardmetal surface as they are pressed onto the hardmetal by interaction of the abrading wheel and the applied normal force. This leads to a wear set-up that can be considered to be a predominantly compressive mode stress application. As the wear scar increases the applied stress decreases as is illustrated in figure 7.3.

5 100N 50N 4

3

-1 2 y = 100x

1

y = 50x-1

0 0 20 40 60 80 100 120 Wear scar area / mm2 Figure 7.3. Variation of applied stress as the wear scar surface area increases.

Abrasion introduces high shear strains into the surface material. The shear strain decreases with depth into the bulk and the depth of deformation should generally be proportional to the depth of indentation of the abrasive particle. The energy of frictional work expended in abrasion is largely accounted for by plastic work done by the deformed surface region. As a result of this strain hardening, the flow stress of the deformed surface material may be two or three times that of the bulk. The magnitude of the flow stresses depend on the response of the hardmetal to the imposed shear strains. For this reason it is generally believed that it is better to correlate wear to the surface hardness than the bulk hardness. Deformation of the surface leads to wear after a critical strain has been reached. At this critical point the work-hardened layer is no longer able to sustain further deformation with the result that surface and subsurface cracking is initiated followed by minute losses of material. The dynamic nature of abrasion is shown in that material is removed at this critical strain whilst material below the surface is yielding. Thus materials with a high work- hardening capacity which are able to attain ultimate hardness while plastically accommodating the imposed stresses and resisting microfracture would have lower wear rates.

The response of the hardmetal to wear depends on the size of the hardmetal phases compared with the scale of deformation caused by individual abrasive particles and to the deformation characteristics of the hardmetal. The size, shape and hardness of the abrasive particles affect the wear. A quartz sand particle is substantially greater than the size of the hard phase and the binder mean free path. The grain size of the quartz sand is approximately 212 microns. Thus it is about

7 Discussion 126

400 times greater than the single tungsten carbide grain. Therefore during wear, one sand particle displaces a great number of carbide crystals and binder islands simultaneously. This causes the material to behave like a homogenous solid which implies that the bulk material properties, i.e. carbide plus binder, will influence the wear rate to a greater extent than the properties of the separate phases.

The quartz sand used was angular and had an irregular shape such that it is expected to deform the surface in a variety of ways. For example, it is possible that sharp edges of the sand particles are able to penetrate between the grains of the tungsten carbide and preferentially remove the cobalt binder. The rate at which the cobalt can be removed is linked to the area of cobalt exposed to attack and the ease with which the abrasive particle edges can penetrate between the carbide grains. It is expected that larger mean free paths would facilitate cobalt removal by this type of abrasive action. For the alloys investigated it is more likely that this mechanism would occur more in alloy FW15 since it had the larger binder mean free path. However, in general the alloys investigated have small man free paths, hence this type of mechanism leading to material loss is expected to be limited.

The extent to which the quartz sand particles can deform the hardmetal microstructure is partly dependent on the ratio of the hardness of the abrasive to the hardness of the hardmetal surface(Ha/Hm). For the investigated hardmetals the ratio is 0.55 for grade UFW10 and 0.69 for grade FW15. A decreasing ratio implies that it is more difficult for the abrasive to penetrate the metal surface. This agrees with the lower wear rates observed for the UFW10 grade. According to Hutchings [108] and ZumGahr [105], the wear rate becomes more sensitive to this ratio when Ha/Hm is less than 1 and, more importantly when this ratio is less than 1.2 the wear process can be classified as ‘low wear’. In this low wear region it is expected that the abrasive particles cannot easily indent the surface, keeping penetration to a minimum, which results in a low surface roughness and low wear rates.

The deformation characteristics of the hardmetal can be assessed by considering the deformation properties of the individual phases. The tungsten carbide phase(WC) is completely constant, stoichiometric and homogeneous [69]. The WC phase is able to accommodate considerable plastic deformation without fracture. One of the main reasons is due to its slip system which is able to maintain continuity under extreme deformation. It has been claimed that of the five independent systems necessary to produce shape changes required, tungsten carbide has four active systems [69].

In most hardmetals, the cobalt binder phase is present in the fcc form. During deformation the fcc changes to a hcp form which is harder. The fcc-hcp transformation occurs with relative ease and is related to the low stacking fault energy of the binder phase which has been measured to be less than 20mJm-2 [69]. In addition, the low stacking fault energies give rise to increased dislocation densities during surface deformation and high work hardening rates. The interaction between the deformation and the dislocations is also expected to cause cracking in the adjoining carbide grains or along them. For the combined hardmetal, at the onset of deformation during wear, the carbide deforms plastically by a glide mechanism [139]. Since tungsten carbides form a skeleton, a small amount of plastic deformation in individual grains can cause large distortions in the bulk skeleton. The binder phase distorts through fcc to hcp formation by the glide of partial dislocations [139]. As deformation proceeds glide becomes difficult and after a certain fraction of fcc to hcp forms, cracks start to form. Simultaneously carbide grains start to crack. As strain increases, the cracking continues. The fcc binder acts to impede crack propagation, but the formed hcp decreases the ability of the binder to deform and also contributes to cracking.

7 Discussion 127

The interaction between these deformation characteristics and the tribosystem parameters control the dominant wear mechanisms in the hardmetal. Wear mechanisms describe the energetic, chemical and physical material interactions between the elements in a tribosystem which lead to structural changes in the contact surfaces. The type of wear mechanisms which occur depend on the properties of the tribosystem. In the tribosystem investigated, abrasion is the dominant damage mechanism. Abrasive wear mechanisms can be divided into four main types, namely microploughing, microcracking, microfatigue and microcutting. Detailed descriptions of these mechanisms are found in chapter 4.5.1.

In the investigated tribosystem, the mechanism of microploughing is discounted as a major mechanism since significant plastic deformation is limited due to the small binder mean free paths and the optimum support provided by the tungsten carbide skeleton. Additional confirmation that plastic deformation is limited is shown by the surface roughness measurement parameter Rz. The intensity of the tribological loading by particle wear is characterized by the depth of the furrows produced in the hardmetal surface. With increasing furrow depth, the contact surface area between abrasive particle and hardmetal surface increases and so does their interaction and the wear rates. The furrow depth is represented by the surface roughness parameter Rz. The measured values for Rz lie between 0.5 to 5 microns. Comparing this value to the grain size of a typical sand particle, 212 micron, shows that exceptionally little penetration of the surface occurs. Hence plastic deformation in the form of microploughing is likely to be minimal.

It is believed that the remaining three wear mechanisms act simultaneously in removing material from the surfaces. Microfatigue is possible due to the continual movement of abrasive across the exposed surface. The microstructures are wear resistant and therefore the contact time between abrasive and microstructure is long, exposing the matrix to repeated impact by the abrasive particles, until the cyclic normal and tangential forces applied to the microstructure by the abrasive particles cause the material shear stress to be exceeded resulting in crack initiation of the matrix. It is most likely that cracking originates in the binder phase since the tungsten carbide grains are known to be able to accommodate a considerable amount of plastic deformation before fracturing. It is usually assumed that the cobalt binder would plastically deform before fracturing. However, due to the small binder mean free paths of the hardmetals tested, plastic flow is restrained and any flow stresses generated by binder deformation lead to crack initiation and propagation. Thus microfatigue leads to microcracking, and surface cracking was found as shown by the micrographs in chapter 6.5.

Microcutting of the surface is possible by fragmented tungsten carbide grains. Microcutting induced by the quartz sand particles is not considered to be likely due to the lower hardness of the abrasive compared to the hardmetal. The hard fragmented carbide grains can act as micro-abrasive particles which may lead to the preferential removal of the cobalt binder as noted on some of the micrographs in chapter 6.5. It is aided by the normal force pushing the hardmetal against the wheel. Preferential removal of the cobalt binder phase leaves the tungsten carbide grains poorly supported and exposed to further abrasive action by the sand particles. Due to repeated abrasive particle action small cracks may form in the carbide grains which eventually, through repeated attack by means of a fatigue mechanism, exceed their shear strength and crack. Consequently they can be easily dislodged from the surface. Finer-grained carbides are more able to resist such shearing forces since there is a greater interlocking between the grains. Hence this mechanism should be more dominant in the FW15 grade. Removal of the cobalt phase may also lead to relaxation of the internal compressive stresses in the carbide grains causing them to crack and become unhinged. Once individual grains are removed, enhanced removal of adjacent grains occurs because they are now less rigidly supported. Pull-out of the carbide grains is then governed by the amount of binder phase present, i.e. the more binder phase the longer it will take to be preferentially removed to the extent where carbide grain pull-out is possible.

7 Discussion 128

Microcracking may also occur at a sub-surface level and extremely fine cracks are also possible due to the material and abrasive properties. Crack propagation may be limited due to cracks becoming arrested possibly at the numerous grain edges and corners. The material removal rates measured all the alloys are low which imply that microcutting and microcracking do not always lead to large amounts of material removal. It is also possible that indentation of the particles leads to a small amount of elastic deformation of the bulk material.

A consequence of partial phase removal may be tribofilm formation which was seen on the worn surfaces in the wet conditions. They may be a product of cyclic plastic deformation or they could be chemical reaction products. Larsen-Basse [112] described the mechanism which leads to a reaction film forming on the wear surfaces during two-body abrasive wear due to binder extrusion arising from the compressive stresses applied to the carbide grains during wear. This extruded binder is than smeared across the surface by further abrasive action. Since smearing of the binder is not expected to occur to such a great extent in the investigated tribosystem, it is unlikely that this is the origin of the tribofilms. Some tribofilms were also transparent, which leans more toward them being chemical reaction products. This could however not be confirmed during this study, thus the exact nature and origin of these films remain a mystery. These tribofilms will lead to a lowering of material loss due to corrosion, since it reduces the exposed tungsten carbide cathodic areas which necessary to sustain an electrochemical reaction.

The wear mechanisms are similar for standard and corrosive wear, although in the case of corrosive wear there is a combined interaction between binder removal via mechanical means and corrosive attack which is discussed in section 7.4.3. This combined interaction leads to a greater degree of carbide grain fall-out and exposure of the carbide grains to abrasive action, thus increasing the material removal rate from the surface and hence the measured wear rate. Due to the type of wear mechanisms which are found and the extent to which they occur, they would lead to greater material removal rates for the FW15 grade. The UFW15 grade has a higher fracture toughness than the UFW10 grade which is linked to the amount of cobalt and the binder mean free path length. For the FW15 grade there is decreasing plastic constraint in the layer of the ductile phase, thus plastic flow becomes easier, local stress concentrations are relieved and both crack initiation and crack propagation are impeded. To account for the decrease in strength of the UFW10 grade it is necessary to consider the observed increase of fracture incidence at the carbide grain boundaries [61]. With smaller carbide grains, as well as less binder phase, there is an increasing area of carbide to carbide contact, i.e. an incomplete surrounding of carbide grains by the cobalt binder. Cracking of these carbide-carbide boundaries are either due to the carbide- carbide grain boundaries being weak or due to stress concentrations caused by the uneven distribution of the binder phase. At any rate the onset of cracking at the boundaries implies that the local strength of the hardmetal has been exceeded. Thus grade UFW10 is expected to have a higher degree of microcracking than grade FW15. The extent of microcracking is also related to the higher hardness of UFW10 which makes it more brittle and thus more prone to cracking.

7.3.2 Influence of fluids on the wear process

A substantial difference between the wear rates under dry conditions and those under wet conditions was measured, such that wear in the wet environment is more benign than wear under dry conditions. Friction forces, co-efficients of friction and surface roughness values were all higher under the dry conditions. The presence of a fluid in the investigated tribosystem modified the mechanics of the contact between the abrasive particle and the hardmetal surface. Fluids flush the wear debris from the system which would otherwise act as micro-abrasive particles and wear the surface on a micro-scale. Fluids also decrease the effectiveness of the abrading action since they

7 Discussion 129 limit particle contact, by reducing the strengths of the junctions formed. This type of lubricating effect of the contact between the particle and the hardmetal surface results in a reduction of the particle tangential force resulting in less tangential deformation below the wear surface and thus less wear. During abrasion, there is a tendency of the abrasive particles to fragment and embed into the metal surface. This was seen on the micrographs in chapter 6.5. A fluid reduces the tendency of the abrasive particles to embed in the surface. By lowering the friction between surface and particle, the fluids lead to less particles cutting into the surface and hence to lower wear rates. Fluids affect the wear rate of a material either as agents of simple lubrication as in the standard wear tests using water or as agents of corrosion as displayed in the corrosive wear tests.

7.3.3 Influence of corrosion on abrasive wear

Wear and corrosion processes consist of a variety of mechanical and chemical mechanisms, which when combined often result in significant increases in material degradation. The question which may arise during corrosive wear testing is then: is the corrosive wear an additive effect or a synergistic one. By this is meant if the volume loss due to the separate processes of corrosion and wear were added up, would they give the volume loss due to the combined process? For the tribosystem investigated, the answer is illustrated in figure 7.4 for hardmetals UFW10 and FW15 where the material loss due to the separate processes of corrosion and standard wear have been calculated. In addition, the volume loss due to corrosive wear is given. By adding the volume losses due to corrosion and due to standard wear, it is found that the volume loss due to corrosive wear is higher than the addition of the separate processes. For the UFW10 grade, the corrosive volume loss is 50% higher and for the FW15 grade, it is 20% higher This shows that the corrosive wear for these hardmetals under these conditions is more than an additive effect, it is a synergistic effect. Thus when the corrosive medium entered this particular wear system, the wear process was accelerated due to the combined effect of corrosion and wear.

100000

7719 9373 10000 2114 1403 1000

100

13,9 9,23 10

1 corrosion wear corrosive corrosion wear corrosive UFW10wear FW15 wear

Figure 7.4. Material loss due to the various tribological conditions for the hardmetals UFW10 and FW15.

The relative magnitude of the wear-corrosion synergism depends on the electrolyte, the alloy and the tribological conditions. The aggressiveness of the electrolyte would affect the corrosion rate significantly. In the system investigated, tannic acid was used. It is an organic acid, which even though the pH range was between 2-4 for the solutions tested, the volume of material loss was

7 Discussion 130 extremely small. This is further confirmed by comparing the volume losses for the materials tested in pure tannic acid solutions and in acid solutions to which the quartz sand had been added. The material loss due to the addition of the sand is approximately 300 times higher. This implies that the impurities in the sand increased the conductivity of the acid solution which led to a more aggressive attack of the microstructure. This implies that for the investigated system, that the role of the corrosion is less significant compared to the abrasive role in material degradation.

In the corrosive-abrasive wear system, two factors which are expected to influence the corrosion rate are surface roughness changes and work hardening of the worn surface. During the wear tests, the surface roughness increased. Hence the contact area between the abrasive particles and the hardmetal surface increased, which subsequently exposed more metal to abrasive action. According to Faraday’s equation the corrosion rate is a function of the surface area. As the surface area increases the corrosion rate increases. Consequently the increase in surface roughness during abrasive wear facilitates the corrosion process which increases material degradation.

The hardmetals tested work-harden during the wear process which was confirmed by measuring the macro-hardness of the worn surfaces after wear testing. An increase in the surface hardness was found for both the hardmetals under all conditions investigated. Work-hardening is associated with an increase in the dislocation density of a metal, which in turn is associated with an increase in the surface free energy. The surface is in a higher state of disorder with a lower average atomic co-ordination. Thus according to the Arrehnius Rate Law, the corrosion rate would increase. The initial energy state of the surface has increased due to an increase in the dislocation density, hence the activation energy barrier to be overcome for corrosion to proceed has decreased, therefore making it easier for the hardmetal to be corroded and reach its final energy state. These two effects were confirmed by doing corrosion tests on unworn specimen and on worn specimen and the results are shown in figure 7.5. The increase in corrosion rates are evident due to the increase in surface roughness which increases the surface area, and due to the increased dislocation density resulting from work-hardening. Wear shifts the corrosion potential into the more active region. Since the measured roughness values are low, it is expected that the influence of work hardening is greater on the corrosion rate.

100000

10000

1000

100

10 UFW10-unworn UFW10-worn FW15-unworn FW15-worn

Figure 7.5. Influence of surface roughness and dislocation density on the corrosion rates of hardmetals UFW10 and FW15.

7 Discussion 131

The corrosive role is further influenced by cracking of the worn surface, which was noted in the microstructure analysis of the worn surfaces. It was seen that fine cracks, pits, depressions and holes occur at random on the worn surface. These features provide access routes into the material for the acid-sand solution, where local increases in anodic activity are created thus permitting corrosive attack of the sub-surface material, whilst abrasion continues on the surface. This kind of sub-surface corrosion would also facilitate material removal due to abrasion, since it effectively dissolves some of the cobalt, thereby causing a reduction in the bonding within the sub-surface matrix, making material removal easier. As the surface layer has a high dislocation density, the acid would react readily with the cobalt and in addition since it is trapped in the hole, the contact time is increased which also facilitates corrosion.

This subsurface attack may be a form of crevice corrosion. In the holes and cracks which run into the surface, the acid-sand solution seeps into the material and initially the solution in the crevice is the same as the bulk solution. The metal dissolves and reduction of an oxidant present in the solution occurs. Initially separation of the cathodic and anodic sites are negligible. Transport of the reduced oxidant may become restrictive resulting in a reduction of the oxidant within the crevice to be faster than the rate at which it can be replenished. Thus the cathodic reaction that may accompany the metal dissolution reaction occurs only on the outer surface rather than inside the crevice where metal dissolution occurs at a high rate. This separation of anodic and cathodic reactions can also produce large internal resistance drops in the corrosion circuits. The crevice region undergoes both solution composition and potential changes. This causes separation of the cathodic and anodic sites leading to localized corrosion.

Evidence for increased material removal rates under corrosive wear due to the presence of holes was shown by comparing the worn surfaces of the hardmetals under the standard wear conditions to that under the corrosive wear conditions. Although the wear mechanisms were found to be similar under both conditions, a closer examination revealed the differences. More holes and depressions were noted in the hardmetals under the corrosive wear condition. This was also reflected in the measured surface roughness which was higher under the corrosive wear conditions. In the corrosive wear system, the cobalt binder is not only removed mechanically by the abrasive particles, but it is also leached out of the matrix by the acid. This combined interaction would lead to a greater degree of carbide grain removal. Thus the material removal rate increases, leading to higher wear rates being recorded.

An increase in the tannic acid concentration led to increased wear rates for both hardmetals. From the corrosion tests, it was seen that the corrosion rates increased with increasing tannic acid concentration, which was also associated with decreasing pH. Thus, it is expected that the attack on the microstructure would increase, which would lead to faster material removal rates. If the tannic acid concentration were continually increased, the influence of the corrosion rate may eventually reach a limiting value and remain constant. The tribological system is expected to reach a point at which increasing the tannic acid solution will not corrode the hardmetal any faster because the rate of metal dissolution reaches a compromise between the corrosion and abrasion processes. It is postulated that the wear rate may eventually decrease with further increases in tannic acid concentration after reaching this compromise. This is linked to an increase in the tannic acid viscosity as the concentration is increased. Subsequently, a test was conducted in which the acid concentration was increased to the point where it reached an almost gel-like state. This solution was then used for a wear test. The wear rate significantly decreased, since the abrasive particles were hindered from making contact with the hardmetal surface. As the viscosity increases the acid begins supporting part of the normal loading by means of the pressures within the acid which arise from the viscous forces within the liquid that are generated by the relative motion between the abrasive particles and the hardmetal surface.

7 Discussion 132

7.3.4 Influence of applied normal force and abrading wheel speed on abrasive wear

It was found that increasing the normal applied load and the abrading wheel speed led to an increase in the standard and corrosive wear rates of both hardmetals. A linear increase was found with load, while an exponential increase was found with speed. The corrosive wear rates were more sensitive to the varied parameters. A variety of interaction factors influence these responses including surface roughness, temperature effects, applied stresses, work hardening, time and contact properties between the abrasive and the hardmetal surface.

The surface roughness increased as the load and speed increased. This implies an increase in the contact area between the abrasive particles and the hardmetal, which leads to increased wear rates. In abrasion the abrasive particle forms an indentation and cuts or displaces material as it moves across the surface. The extent to which penetration occurs depends on factors such as abrasive, hardmetal and tribosystem properties which was discussed in section 7.4.1. The initial contact between the abrasive and the hardmetal at low loads and speeds starts with asperity contact. As the asperities elastically and plastically deform during wear, the contact area increases and the morphology of the abrasive and hardmetal change which leads to increases in the wear rates. At higher loads and speeds, the greater contact area between abrasive and hardmetal is more like a grooving effect leading to higher wear rates, since the rate at which the asperities and morphology changes is greater than at the lower loads, thus causing an increase in wear rate as the load and speed increases. Thus brittle fracture of the hardmetal surface is expected to increase with increasing loads and speeds. The harder UFW10 grade is better suited to resist this situation than the coarser-grained FW15 grade and thus has a better wear resistance under both increasing load and speed.

In addition, as was mentioned in section 7.4.3, an increase in the contact area through a roughness increase, facilitates corrosion, leading to an increased corrosion sensitivity as the load and speed increases. The change in surface roughness was greater under increasing loading than under increasing speed. This is not unusual as the loading remained constant during the speed variation and the extent of material penetration is expected to increase more with increasing applied force than wheel speed. This is also confirmed by the increasing friction force with increasing load, which remained constant with increasing speed.

Temperature often plays a role in wear processes, especially when the speeds employed are high. As the speed varies, temperature changes may occur and the test environment is influenced, such that hardmetal properties and abrasive medium properties are affected. In addition, an increase in temperature favors corrosion kinetics. Attempts were made to measure the temperature during the wear processes, however, due to the system set-up and the abrasion process, it proved to be unsuccessful. However, it is unlikely that temperature had any significant influence on the tribosystem employed since the speeds used were very low. The temperatures were probably higher under the dry testing than the wet testing, but its effect on the wear response is considered to be negligible.

As the load increased, the applied stress increased and hence the thickness of the work-hardened layer is expected to increase. Faster speeds are also known to lead to thicker work hardened layers. Work-hardening, as was mentioned in section 7.4.3, is associated with an increase in the dislocation density of the metal, which facilitates the corrosion rate. Hence with thicker work- hardened layers as the force and speed increase, the corrosion rate is expected to increase. This added to the synergistic effect between corrosion and abrasion increasing as the load and speed increased. It is also possible that the wear rate may decrease as the surface layer continues to

7 Discussion 133 work-harden – it becomes more resistant to the abrasive process. This implies longer contact times between the acid-sand solution and hence the role of corrosion may increase. It is also expected that the wear rate may eventually reach a limit at which point no further increase is measured and the surface will wear at a constant rate.

A critical amount of deformation is required to initiate wear and this process takes time. At this point deformation begins and a wear rate can be determined for the sample. The period until penetration of the first abrasive particles can be termed an initiation wear stage. For low loads and speed, this initiation stage lasts longer than at higher loads and speeds. This initiation stage can be measured online during the wear test. It has been shown for example that for the FW15 grade the time taken before a stationary wear stage is reached is approximately 30 minutes using a load of 20N, while at a load of 100N the time required is between 10-15minutes. The initiation times for the UFW10 are much longer.

At lower loads and speeds the standard and corrosive wear rates are similar. This may be attributed to the amount of surface damage being minimal, hence the factors which facilitate corrosion, such as surface area, work hardening and cracks, are not present in significant amounts. Another factor may be the acid properties, in that tannic acid is very weak, and if for example a stronger acid such as hydrochloric acid were used, the wear rates may have been significantly different even at low loads and speeds. An additional reason for the similar wear rates may be linked to the time period during which the test was run. It is possible that if the test was run for extensively long time periods, the effect of the corrosion may become more significant at lower loads and may even result in an inverse effect to what is shown in the trend of increasing corrosive wear rate with increasing load and speed. By this is meant that the standard and corrosive wear rates would be further apart initially and come together later as the load and speed increase. The reasoning behind this, is that if given enough time to abrade the surface at low loads and speeds, the surface would roughen and work harden and cracking of the hardmetal will also occur. These factors then provide a means for the acid to interact with the material. Since the rate of material removal is small at lower loads and speeds, the acid would have longer contact times with the hardmetal surface and the corrosive role should increase and result in increases of the wear rate.

7.3.5 Comparison between the wear response of the investigated hardmetals

In all the tribosystems tested, the fine-grained FW15 hardmetal displayed higher wear rates than the ultrafine-grained UFW3 and UFW10 hardmetals. The wear resistance of the alloys is primarily provided by the carbide phase, which increases with decreasing cobalt content. Evidence for a higher wear resistance with higher carbide volumes and smaller tungsten carbide grain sizes can be found in the work of Herr [84]. The finely dispersed carbide phase causes an increase in the flow stress and hence the wear resistance is increased. Additionally the cobalt binder is highly constrained between the carbide grains thus increasing its yield strength. With less binder in the hardmetal, the carbide grain boundary strength becomes more important since a higher contiguity implies more interaction between carbide-carbide grains boundaries during wear. The fine-gained hardmetal is tougher than the ultrafine alloys, hence its resistance to fracture and cracking is greater. However, in the abrasive wear set-up, hardness is more important than toughness to resisting wear, as is shown by the lower wear rates for the harder but less tougher hardmetals. These combined properties of phase distribution, hardness and toughness to resisting wear is further elucidated by the type of wear mechanisms which have been found and has been discussed in section 7.4.1. The role of the abrasive has also been highlighted in previous sections. All these interconnecting factors give rise to the higher wear rates of the FW15 grade. The extent

7 Discussion 134 to which they influence the wear rates was further shown by the higher sensitivity of the FW15 grade to increasing normal applied load and increasing abrading wheel speed.

The ultrafine grades showed larger differences between dry wear rates and wet wear rates. This difference can be directly related to the wear mechanisms and to the ratio of the hardness of abrasive to the hardness of the hardmetal. This ratio is lower for the ultrafine-gained alloys due to their higher hardness values. The influence of the wear debris and its subsequent removal by fluids has previously been discussed. It is therefore expected that the removal of the wear debris during wet abrasion leads to a reduction in the wear rates for all the alloys. However, since the hardness ratio is tenuous for the FW15 grade, material degradation is still expected to be high leading to less of a difference between the dry and wet wear rates obtained.

The wear rates measured under corrosive wear conditions were higher than those measured under standard wear conditions for all the alloys. In neutral environments, like water, the corrosion rate of the binder is extremely low and thus the main mechanisms of carbide loss is by mechanical wear and grain pull-out. However decreasing the pH to an acid environment results in enhancement of cobalt removal and subsequent higher wear rates. The difference between the standard and corrosive wear rates is higher for the ultrafine-grained alloys than the fine-grained alloy. The change from water to tannic acid produced an increase in the wear rates of 20, 50 and 80% respectively for alloys FW15, UFW10 and UFW3. This suggests that the UFW3 grade is relatively more sensitive to the corrosive medium than the FW15 grade. The factors which influence this behaviour are once again linked to the properties of the tribosystem. For the FW15 grade the role of the corrosion may be limited due to abrasion playing more of a dominant role due to the higher hardness ratio between abrasive and hardmetal. The material removal rate is fast enough to limit the interaction time between the acid and the metal. For the ultrafine alloys the material removal rates are low, which allows more interaction time between the acid and the hardmetal surface. Since there is less cobalt present and the binder mean free paths are smaller, removal of the cobalt binder by corrosive attack should become easier and lead to greater carbide removal. This does not imply that the ultrafine alloys will have higher overall wear rates, but simply that the change from a neutral to an acid environment would lead to a greater difference in the respective wear rates for finer-grained hardmetals with low cobalt contents. The sensitivity of the ultrafine alloys to the tannic acid was further observed when the tannic acid concentration was increased.

The wear rate of the UFW10 grade was found to be more sensitive to increasing tannic acid concentration than the FW15 grade. More specifically, the rate at which the wear rate increased as the acid concentration increased was higher for alloy UFW10. This was an unexpected result, since it was assumed that as the tannic acid concentration becomes stronger the FW15 grade would be more susceptible to the corrosive-wear synergistic effect. This assumption was based on the higher corrosion rates observed for the FW15 grade as the tannic acid concentration increased. Additionally the higher degree of surface deformation which was reflected in the surface roughness values was expected to facilitate the corrosion rate. The material removal rates for the FW15 are also greater which means that that fresh surfaces are continually being exposed to corrosive attack. However, an additional parameter which influences the corrosion rate is time, i.e. the contact time between the acid-abrasive solution and the specimen surface. The corrosive wear rates of the FW15 grade are approximately 4 times greater than that of the UFW10 grade as the acid concentration increases. This implies that for a given time period, the amount of material removed from the FW15 surface is greater than from the UFW10 surface. This in turn indicates that the contact time between the acid-abrasive solution and the UFW10 surface is longer, hence more time is granted for the corrosive-abrasive fluid to interact with the surface.

7 Discussion 135

In section 7.4.3. various factors were mentioned which facilitate corrosion, such as surface roughness, dislocation density and cracks which act as inlets for the solution. In addition to these factors, there is now the added aspect of time, which according to Faraday’s equation is proportional to the corrosion rate. The longer a surface is exposed to a corrosive substance the more it can interact. It is thus proposed that in addition to the abrasion process occurring and the influence of the corrosion on the surface increasing with increasing tannic acid concentration, the longer surface-to-acid-abrasive solution contact times facilitate higher rates of material removal as the acid concentration increases. As was mentioned in section 7.4.3, the acid attacks the sub- surface deformed layer result by weakening of the matrix bonds, hence with more time this weakening of the sub-surface material should be greater for the UFW10 grade. Hence the relative amount of material removed as the tannic acid concentration increases is higher for UFW10 than FW15 grade.

8 Conclusions 136

8 Conclusions

In this study a systematic investigation of the abrasive wear and corrosive-abrasive wear performance of tungsten carbide hardmetals used in woodcutting has been carried out in which some of the microstructural, environmental and operational parameters which control the overall wear process have been quantified.

This was done by: • investigating the wear response of ultrafine-grained tungsten carbide hardmetal woodcutting knives during commercial woodcutting operations • investigating the wear response of tungsten carbide hardmetals in closely controlled laboratory conditions in which some of the parameters which are believed to control the wear process have been examined more closely

The most important general conclusions from the woodcutting tests are: • The sharp cutting edge wears off rapidly within cutting 1m of wood. Although this effect does not appear to influence the performance of the knives as woodcutting proceeds, the implications of this breakage of the sharp edge can be extended to the production of the knives. Any modifications to the cutting edge or the introduction of new finishing techniques need to consider this effect. If for example, coating of the knives to improve its wear resistance is suggested, (which is currently being considered in the industry), then the effects of this high initial wear on the coating properties need to be investigated. • The wear rates are influenced by the hardmetal and wood properties and the interaction between the two during woodcutting. Factors such as area of contact, test set-up, wood fiber orientation, cobalt content and chip formation dictate the wear response. • A new mechanism of material removal in the machining of chipboard has been proposed which is applicable to ultrafine-grained hardmetals. This mechanism is based on the finer phase distribution and resulting strengthening characteristics.

The most important general conclusions from the corrosion tests are: • the galvanic effect in which the cobalt is preferentially removed may be limited to the phase boundaries between the tungsten carbide and the cobalt due to the weak tannic acid strength. • no true passivation occurs in any of the systems tested resulting in rapid removal of the binder phase. • it is proposed that the cobalt metal forms soluble chelates with the tannic acid hence the lack of corrosion products and passive film formation. • corrosive attack is localized with no clear evidence for the cause.

8 Conclusions 137

The most important general conclusions from the abrasive wear tests are: • the wear rates are governed by the abrasive, hardmetal and tribosystem properties. Any changes in one of these parameters would result in significantly different wear responses. • The low wear rates recorded can in part be attributed to the inability of the quartz sand particles to penetrate the hardmetals. This inability is related to the low relative hardness of the abrasive in comparison to the hardmetal hardness. • the influence of the tannic acid on the wear response is minimal. This can be attributed to the low aggressiveness of the acid strength. As a result abrasion is the dominant damage mechanism for all the systems investigated. • corrosive wear is a synergistic effect, which increases with increasing acid strength, loading and speed conditions. • the wear process facilitates corrosion by increased surface areas, work-hardening effects and the type of wear mechanisms which initially emerge. • the ultrafine-grained hardmetals are more sensitive to corrosive attack under abrasive wear conditions due to the low material removal rates which allow longer contact times between the acid and metal which facilitates corrosion. • wear mechanisms for investigated hardmetals are surface fatigue followed by microcracking. Microcutting by wear debris also present. Material removal characterized by microscale removal of cobalt, crack formation and removal of material clusters.

Altogether this work provides some insights into the wear behaviour of fine-grained tungsten carbide hardmetals during the initial stages of woodcutting. In addition, a quantitative description of the wear response under three body abrasive wear in dry, neutral and a weak corrosive environment has been provided, which is not restricted to the woodcutting environment, but can be expanded to include similar application fields in which hardmetals are subjected to three body abrasive wear. 9 Zusammenfassung 138

9 Zusammenfassung

In dieser Studie wurde eine systematische Untersuchung des Verhaltens von Wolframkarbid Hartmetallen, die bei der Holzzerspanung verwendet werden, unter Abrasivverschleiß und Korrosions-Abrasiv-Verschleiß durchgeführt. Einige der mikrostrukturellen, Umgebungs- und Betriebs-Parameter, welche den gesamten Verschleiß-Prozess steuern, wurden quantitativ bestimmt.

Dieses wurde folgendermaßen getan: • durch die Untersuchung des Verschließverhaltens ultrafeinkörniger Wolframkarbid-Hartmetall Schneidmesser im Rahmen des kommerziellen Einsatzes in der Holzzerspanung. • durch die Untersuchung des Verschließverhaltens ultrafeinkörniger Wolframkarbid- Hartmetallen unter genau kontrollierten Laborbedingungen, bei der einige für den Verschleißprozess potentiell entscheidende Parameter genauer betrachtet wurden

Die wichtigsten allgemeinen Schlußfolgerungen, die aus den Holzzerspanungsversuchen entnommen werden können, sind wie folgt: • die scharfe Schneide verschleißt zunächst schnell, innerhalb des ersten Meters Schnittlänge im Holz. Obgleich dieser Effekt nicht die Leistung der Messer während des weiteren Schneidprozesses beeinflußt, kann dieses Brechen der scharfen Schneidkante bereits bei der Produktion von Messern Beachtung finden. Bei allen Änderungen der Schneidengeometrie oder bei der Einführung neuer Fertigungstechniken zur Endbearbeitung muss dieser Effekt berücksichtigt werden. Wenn zum Beispiel das Beschichten der Messer zur Verbesserung ihrer Verschleißbeständigkeit vorgeschlagen wird (wie es in der Industrie z. Z. geschieht), dann müssen die Effekte dieses hohen Anfangsverschleißes auf die Eigenschaften der Beschichtung weiter untersucht werden. • die Verschleißraten werden durch die Hartmetall- und Holz-Eigenschaften und die Wechselwirkung zwischen beiden während des Schneidprozesses beeinflußt. Faktoren wie die Grösse des Kontaktbereiches, der Versuchsaufbau, die Orientierung der Holzfasern, der Kobaltgehalt des Hartmetalls und die Spanausbildung bestimmen das Verschleißverhalten. • ein neuer Mechanismus des Materialverlusts beim Schneiden von Spanplatten wurde vorgeschlagen, welcher auf ultrafeinkörnige Hartmetalle anwendbar ist. Dieser Mechanismus basiert auf der feineren Phasenverteilung und der daraus resultierenden Verfestigung.

Die wichtigsten allgemeinen Schlußfolgerungen, die aus den Ergebnissen der Korrosionsversuche entnommen werden können, sind wie folgt: • der galvanische Effekt, bei dem das Kobalt vorzugsweise entfernt wird, kann wegen der schwachen Stärke der Tanninsäure auf die Phasengrenzen des Wolframkarbides mit dem Kobalt begrenzt werden. • in keinem der untersuchten Systeme tritt eine wirkliche Passivierung auf. Daraus resultiert ein schneller Abtrag des Kobalt-Binders. 9 Zusammenfassung 139

• als Mechanismus wird vorgeschlagen, daß das Kobalt mit der Tanninsäure lösliche Chelate bildet. Daraus folgt das Fehlen von Korrosionsprodukten und der Ausbildung einer Passivschicht. • der korrosive Angriff ist lokalisiert, die Ursache hierfür ist noch ungeklärt.

Die wichtigsten allgemeinen Schlußfolgerungen, die aus den Ergebnissen der Verschleißversuche entnommen werden können, sind wie folgt: • die Verschleißraten werden durch das Abrasivum, das Hartmetall und die Eigenschaften des Tribosystems bestimmt. Jede mögliche Änderung an einem dieser Parameter würde erhebliche Unterschiede im Verschleißverhalten ergeben. • die niedrigen Verschleißraten werden teilweise der Unfähigkeit der Quarzsandpartikel in das Hartmetall einzudringen, zugeschrieben. Dieses Unvermögen hängt mit der relativ niedrigen Härte des Abrasivums im Vergleich zum Hartmetall zusammen. • der Einfluß der Tanninsäure auf das Verschleißverhalten ist minimal. Dies kann der geringen Aggressivität der Säurestärke zugeschrieben werden. Infolgedessen ist der abrasive Verschleiß der dominierende Schädigungsmechanismus bei allen untersuchten Systemen. • Korrosions-Verschleiß ist ein synergistischer Effekt, der sich bei Zunahme der Säurestärke, der Belastung und der Geschwindigkeiten erhöht. • der Verschleißprozess erleichtert Korrosion durch Vergrößerung der Oberflächen, Verfestigungseffekte und die Art der Verschleißmechanismen, die zuerst auftreten. • die ultrafeinkörnigen Hartmetalle sind für den korrosiven Angriff bei abrasivem Verschleiß empfindlicher wegen der niedrigen Verschleißrate, die längere Kontaktzeiten zwischen der Säure und dem Metall bedingt und somit die Korrosion erleichtert. • Verschleißmechanismen bei den untersuchten Hartmetallen sind die Oberflächenermüdung gefolgt von der Bildung von Mikrorissen. Microspanen durch Verschleißtrümmer ist auch möglich. Der Materialabtrag ist gekennzeichnet durch den mikrostrukturellen Abbau von Kobalt, Spaltbildung und die Entfernung der Bruchstücke.

Zusammengefasst stellt diese Arbeit einige Einblicke in das Verschleißverhalten feinkörniger Hartmetalle während der Anfangsstadien der Holzzerspanung zur Verfügung. Zusätzlich wird eine quantitative Beschreibung des Verhaltens unter drei-Körper-Abrasivverschleiß in trockener, neutraler und schwach korrosiver Umgebung gegeben, die nicht nur auf den Prozess der Holzzerspanung eingeschränkt ist, sondern auch ausgedehnt werden kann, um ähnliche Anwendungsfelder einzuschließen, in denen Hartmetalle einem drei-Körper-Abrasivverschleiß unterworfen werden.

10 References 140

10 References

1. Brookes K.J.A. World directory and handbook of hard metals. 6th Edition, International Carbide Data, London, (1996). 2. Maier G. Holzspannungslehre und werkzeugtechnische Grundlagen. Vogel Buchverlag, Würzburg, (2000). 3. Davis J.R.D. ASM Speciality Handbook: Tool Materials. ASM International, (1995) 433-434. 4. Klamecki B.E. A review of wood cutting tool wear literature. Holz als Roh-Werkstoff, 37 (1979) 265-276. 5. Bayoumi A. E. and Bailey J.A. An analytical and experimental investigation of the wear of cemented carbide cutting tools in the presence of dilute organic acids. Wear 94 (1984) 29-45. 6. Bailey J.A., Bayoumi A. and Stewart J.S. Wear of some cemented tungsten carbide tools in machining oak. Wear 85 (1983) 69-79. 7. Bayoumi A. E. and Bailey J.A. Comparison of the wear resistance of selected steels and cemented carbide cutting tool materials in machining wood. Wear 105 (1985) 131-144. 8. Bayoumi A. Bailey J.A. and Stewart J.S. Comparison of the wear resistance of various grades of cemented carbides that may find application in wood machining. Wear 89 (1983) 185-200. 9. Kirbach E. and Chow S. Chemical wear of tungsten carbide cutting tools by western red Cedar. Forest Products Journal 26(3) (1976) 44-48. 10. http://www.forestworld.com 11. Hillis W.E. Wood extractives and their significance to the Pulp and Paper Industries. Academic Press, New York, (1962). 12. Kivimaa E. Was ist die Abstumpfung der Holzbearbeitungswerkzeuge? Holz als Roh-Werkstoff 10 (1952) 425-428. 13. Smith W.F. Principles of and engineering. 3rd Edition, McGraw-Hill Inc. USA (1996). 14. Nordström J. and Bergström J. Wear testing of saw teeth in timber cutting. Wear 250 (2001) 19-27. 15. http://www.forestprod.org 16. Barz E. and Breier H. Kurzverfahren zur Prüfung der Verschleißwirkung und der Zerspanbarkeit von Holz und Holzwerkstoffe. Holz Roh-Werkstoff, 29 (1971) 142-149. 17. Klamecki B.E. Electrical effects in wood cutting tool wear. Holz Roh-Werkstoff. 36 (1978) 107-110. 18. Prokes S. Einfluß der Ausgangsqualität von Schneiden aus Werkzeugstahl und aus gesintertem Hartmetall auf die Abstumpfung. Holztechnologie 2 (1961) 235-238. 19. Barz E. Standzeitverlängerung bei Holzbearbeitswerkzeugen durch Hartverchromung. Holz Roh- Werkstoff 24 (1966) 593. 20. Barz E. and Breier H. Kurzverfahren zur Prüfung der Verschleißwirkung und der Zerspanbarkeit von Holz und Holzwerkstoffe. Holz Roh-Werkstoff 27 (1969) 148-152. 21. Sheik-Ahmad J.Y. and Bailey J.A. The wear characteristics of some cemented tungsten carbides in machining particleboard. Wear (1999) 225-229. 22. Pahliltzsch G. and Dziobek K. Beobachtungen über das Abstumpfungsverhalten beim Fräsen von Schichtstoff-Verbundplatten. Holz Roh-Werkstoff 23 (1965) 121-125. 23. Mackenzie W.M. and Cowling R.L. A factorial experiment in transverse plane (90/90) cutting of wood. Part 1, Cutting force and edge wear. Wood Science 3 (1971) 204-213. 24. Mackenzie W.M. and Karpovich H. Wear and blunting of the tool corner in cutting a wood-based material. Wood Science and Technology 9 (1975) 59-74. 25. Pahlitzsch G and Schulz K. Schnittkraftmessung und Schneidenabstumpfung beim Hobeln von Holz mit Kreisender Schnittbewegung. Holz Roh-Werkstoff 15 (1957) 159-170. 10 References 141

26. Noguchi M., Fugiwara K and Sugihara H. Use of ultrasonic vibration in turning wood. Wood Science 5 (1973) 211-222. 27. High temperature wear of WC-Co tools in machining particleboard and medium density fibreboard. A critical review. North Carolina State University. 28. Lewandowski, C. Masters thesis. North Carolina State University (1997). 29. Betteridge, W. Cobalt and its Alloys. Halsted Press, New York, (1982). 30. http://www.holistic-health-med.com/Chelation.html 31. Tsai G.S.C and Klamecki B.E. Separation of abrasive and electrochemical mechanisms in woodcutting. Wood Science 12(4) (1980) 236-242. 32. Mackenzie W.M. and Karpovich H. The frictional behaviour of wood. Wood Science and Technology 2 (1968) 138-152. 33. Sugihara H. Wood cutting in S.Kadita(ed) Wood Technology Tokyo, (1961) 329-380. 34. Huber H. Tool wear influences by the contents of particleboard. 72-86. 35. Pahlitzsch G. and Jostmeier H. Beobachtungen über das Abstumpfungsverhalten beim Fräsen von Spanplatten. Holz Roh-Werkstoff 22 (1964)a 139-146. 36. Pahlitzsch G. and Jostmeier H. Weitere Beobachtungen über das Abstumpfungsverhalten und den Einfluß der Schnittgeschwindigkeit beim Fräsen von Spanplatten. Holz Roh-Werkstoff 22 (1964)b 424- 429. 37. Bridges R.R. A quantitative study of some factors affecting the abrasiveness of particleboard. Forest Products Journal 21[11] (1971) 39-41. 38. Rackwitz G. Der Einfluß von Leimstreckmitteln auf die Abstumpfung von Werkzeugschneiden bei der Bearbeitung von Lagenhölzern. Holz Roh-Werkstoff 20 (1962) 398-403. 39. Prakash L.J. Application of fine grained tungsten carbide-based cemented carbides. International Journal of Metals and Hard Materials 13 (1995) 257-264. 40. Larsen-Basse J. et al. (eds) Science of Hard Metals, Adam Higler, (1986) 883-895. 41. Jia K and Fischer T.E. Abrasion resistance of nanostructured and conventional cemented carbides. Wear 200 (1996) 206-214. 42. Galli E. Important considerations when using cemented carbide in woodworking applications. Proceedings of Saw Technology, Berkeley California, October (1991). 43. Pahlitzsch G and Dziobek K. Untersuchungen über das Abstumpfungsverhalten eines Schneidzahnes. Holz Roh-Werkstoff 26 (1968) 162-170. 44. Reynolds D.D. et al. Cutting characteristics and power requirements of chain saws. Forest Products Journal 20(10) (1970) 28-34. 45. Pastor H. Proc. European Conference, European Powder Metallurgy Association, Shrewsbury, 3 (1996). 46. Schroter K. USA Patent 1549615, October 31 (1923). 47. Jia K. and Fischer T.E. Microstructure, hardness and toughness of nanostructured and conventional WC-Co composites. Nanostructured Materials 10[5] (1998) 875-891. 48. Snowball R.F. and Milner D.R. Powder Metallurgy 11 (1968) 23-40. 49. Hinnuber J. et al. An electron-microscopy and x-ray investigation of the milling of tungsten carbide/cobalt mixtures. Powd. Metall. 8 (1961) 1-24. 50. Trent E. M. Powder Metallalurgy 9 (1962) 322. 51. Lewis D. and Lindley J. Journal of the American Society 49 (1966) 49. 52. Carrol D.F. Processing and properties of ultrafine WC/Co hard materials. Proc. 14th Plansee Conf., Eds. G. Knerringer, P. Rödhammer and P. Wilhartitz, Plansee AG, Reutte, Austria, 2 (1997) 168-182. 10 References 142

53. Porat R. et al. Investigations of the sintering mechanisms for cemented carbides based on nanocrystalline powders. Proc. Euro Powd. Metall. Conf., European Powder Metallurgy Association, Shrewsbury (1996) 101-107. 54. Uhrenius B. et al. A study of the Co-W-C system at liquidus temperatures., Scandanavian Journal of Metallurgy 5 (1976) 49-56. 55. Kear B. H. and McCandlish L. E. Chemical Processing and Properties of Nanostructured WC-Co Materials. Nanostructured Mater. 3 (1993)19-30. 56. Exner H.E. Physical and chemical nature of cemented carbides. International Materials Review 24[4] (1979) 149-173. 57. Saller H. A. et al. US Patent (1964) 687 842. 58. French D.N. and Thomas D.A. Hardness anisotropy and slip in WC crystals. Transactions of the AIME 233 (1965) 950-952. 59. Gurland J. The fracture strength of sintered tungsten carbide-cobalt alloys in relation to composition and particle spacing. Transactions of the Metallurgical Society of AIME 227 (1963) 1146-1150. 60. Suzuki H. and Kubota H. The influence of binderphase composition on the properties of WC-Co cemented carbides. Planseeberichte für Pulvermetallurgie 14 (1966) 96-109. 61. Gurland J. and Bardzil P. Relation of strength, composition and grain size of sintered WC-Co alloys. Transactions of the AIME 203 (1955) 311-315. 62. Gurland J. Transactions of the AIME Journal of Metals (1954) 289. 63. Exner H.E. and Gurland J. A review of parameters influencing some mechanical properties of tungsten carbide cobalt alloys. Powder Metallurgy 13 (1970) 13-31. 64. Bartolucci-Luyckx S. Proc. 1st Int. Conf. Sci of Hard Mater., Eds. R. K. Viswanadham, D. F. Rowcliffe and J. Gurland, Plenum Press New York (1983) 629. 65. Arato, P. Sinter-hip treatment of hard metal from nanocrystalline WC/Co powder. Proc. 14th Plansee Conf., Eds. G. Knerringer, P. Rödhammer and P. Wilhartitz, Plansee AG, Reutte, Austria, 2 (1997) 658- 664. 66. H. Suzuki et al. Strength Increase in WC - 10% Co Cemented Carbides Due to Elimination of Large Residual Pores. Planseeber. Pulvermet. 2 (1975)121-130. 67. Fang Z. and Eason J.W. Nondestructive evaluation of WC-Co composites using magnetic properties. International Journal of Powder Metallurgy 29[3] (1993) 259-265. 68. Wirmark G. and Dunlop G.L. Phase transformations in the binder phase of Co-W-C cemented carbides. Proc. 1st International Conference on the Science of Hard Materials (1981) 311-325. 69. Roebuck B. and Almond E.A. Deformation and fracture processes and the physical metallurgy of WC-Co hardmetals. International Materials Review 33[2] (1988) 90-110. 70. Exner H.E. Qualitative and quantitative interpretation of microstructures in cemented carbides. Proc. 1st International Conference on the Science of Hard Materials (1981) 233-251. 71. O’Quigley D.F.G. et al. An empirical ranking of a wide range of WC-Co grades in terms of their abrasion resistance measured by the ASTM Standard B611-85 test. International Journal of and Hard Materials 15 (1997) 73-79. 72. Chermant J.L. and Osterstck F. Fracture toughness and fracture of WC-Co composites. Journal of Material Science 11 (1976) 1939-1951. 73. Namakura M. and Gurland J. Journal of Metallurgical Transactions 11A (1980) 141. 74. Viswanadam R.K. et al. Quantitative fractography of WC-Co by Auger spectroscopy. Journal of Material Science 16 (1981) 1029. 75. Nidikom A. and Davis T.J. Proc. 10th Plansee Conf., Ed. H. M. Ortner, Metallwerk Plansee, Reutte, Austria, (1981) 28. 76. Pickens J.R. and Gurland J. The fracture toughness of WC-Co alloys in SENB specimens precracked by electro-discharge machining. Materials Science and Engineering 33 (1978) 135. 10 References 143

77. Schuhmacher G. and Ostermann G. Cobalt, 4 (1974) 77-92. 78. Rosso M. et al. Cemented carbides: WC grades compared to TiC grades. 13th International Plansee Seminar, Eds. H. Bildstein and R. Eck, Metallwerk Plansee Reutte, 2 (1993) 250-263. 79. Aronsson B. and Aschan L.J. Met. Powd. Rev. 40 (1985) 785. 80. Egami A. et al. Morphology of vanadium carbide in submicron hardmetals. Proc. 13th Plansee Conf., Eds. H. Bildstein and R. Eck, Metallwerk Plansee Reutte, 3 (1993) 639-648. 81. Luyckx S. and Alli Z. Additives in cemented carbides. (private communication). 82. Nordgren A. and Melander A., Powd. Metall. 31 (1988) 189. 83. Fischmeister H. F. and Exner H. E. Gefügeabhängigkeit der Eigenschaften von Wolframkarbid-Kobalt- Hartliegierungen. Archiv für das Eisenhüttenwesen 37 (1966) 499-510. 84. Herr M. Tribologisches Verhalten von ultra-feinkörnigen Hartmetallen mit verschiedenen Binderliegerungen. Ph.D. thesis. University Erlangen-Nürnberg (2002). 85. Lee H.C. and Gurland J. Hardness and deformation of cemented tungsten carbide. Materials Science and Engineering 33 (1978) 125-133. 86. Warren R. and Johannesson B. International Journal of Refractory Metals and Hard Materials 3 (1984) 187. 87. O’Qicgley D.G.F. et al. New results on the relationship between hardness and fracture toughness of WCCo hardmetal. Materials Science and Engineering A 209 (1996) 228-230. 88. Schubert W.D. et al. General aspects and limits of conventional ultrafine WC powder manufacture and hardmetal production. International Journal of Refractory Metals and Hard Materials 13 (1995) 281-296. 89. Jia K. The effect of fine microstructure on the wear and relevant mechanical properties of cemented carbide. PhD thesis. Stevens Institute of Technology, (1996). 90. Human A.M. The corrosion of tungsten carbide-based cemented carbides. Ph.D. thesis. Technische Hochschule Darmstadt (1994). 91. Stern M. Surface area relationships in polarization and corrosion. Corrosion 14 (1958) 329-332. 92. Pourbaix M. Atlas of equilibrium diagrams. Pergammon London 325 (1966). 93. Tomlinson W.J. and Linzell C.R. Anodic polarization and corrosion of cemented carbides with cobalt and nickel binders. Journal of Materials Science 23 (1988) 914-918. 94. Gulbransen E.A. and Andrew K.F. The kinetics of the oxidation of cobalt. Journal of Electrochemical Society 98 (1951) 241-251. 95. Kröncke G. and Masing G. Studien über das Verhalten der Eisenmetalle im Elktrolyten. Werkstoffe und Korrosion 3 (1953) 86-95. 96. Heathcock C.J. and Ball A. Cavitation erosion of cobalt-base alloys, cemented carbides and surface treated low alloy steels. Wear 74 (1981-82) 11-26. 97. Voorhies J.D. Electrochemical and chemical corrosion of tungsten carbide (WC). Journal of Electrochemical Society 119(2) (1972) 219-222. 98. Ghandehari M.H. Anodic behaviour of cemented WC-6% Co alloy in phosphoric acids solutions. Journal of Electrochemical Society. 127(10) (1980) 2144-2147. 99. Hellsing M. High resolution microanalysis of binder phase in as sintered WC-Co cemented carbides. Material Science and Technology 4 (1988) 824-829. 100. Tomlinson W.J. and Ayerst N.J. Anodic polarization and corrosion of WC-Co hardmetals containing small amounts of Cr3C2 and /or VC. Journal of Materials Science 24 (1989) 2348-2354. 101. Fernandes P.J.L. et al. Does the cobalt mean free path affect the corrosion behaviour of WC-Co? Journal of Hard Materials 3 (1992) 185-194. 102. Ringas C. et al. Corrosion behaviour of ion implanted WC-Co and WC-Ni alloys in acid and chloride containing media. Surface Engineering 6[3] (1990) 194-198. 10 References 144

103. Mori G. et al. Influencing the corrosion resistance of cemented carbides by addition of Cr3C2, TiC and TaC. 15th International Plansee Seminar, Reutte, Austria, 2 (2001) 222-236. 104. Enqvist H. Microstructural aspects on wear of cemented carbides. Ph.D. thesis. Uppsala University, Sweden (2000). 105. Zum Gahr K.H. Microstructure and wear of materials. Elsevier Science Publishers, The Netherlands, (1987). 106. Yust C.S. Tribology and wear, International Materials Review 30[3] (1985) 141-154. 107. Peterson M.B. and Winer W.O. Wear Control Handbook. ASME New York, (1980). 108. Hutchings I. Tribology: Friction and wear of engineering materials. Edward Arnold London (1992). 109. Czichos H and Habig K.H. Tribologie Handbuch: Systemanalyse, Prüftechnik, Werkstoffe und Konstruktionselemente. Wiesbaden, Vieweg Verlag (1992). 110. Rabinowicz E. Penetration hardness and toughness indicators of wear resistance. I Mech E conference publications (1987) 197-201. 111. Larsen-Basse J. Effect of composition, microstructure, and service conditions on the wear of cemented carbides. Journal of Metals (1983) 35-42. 112. Larsen-Basse J. Binder extrusion in sliding wear of WC-Co Alloys. Wear 105 (1985) 247-256. 113. Gant A.J. and Gee M.G. Wear of tungsten carbide-cobalt hardmetals and hot isostatically pressed high speed steels under dry abrasive conditions. Wear 251 (2001) 908-915. 114. Feld H. and Walter P. Beitrag zur Kenntnis des Mineral-Hartmetall-Verschleißes. International Journal of Materials Technology 7 (1976) 300-303. 115. Larsen-Basse J. Wear of hard metals in rock drilling: a survey of the literature. Powder Metallurgy 16 (1973) 1-32. 116. Larsen-Basse J. Effects of hardness and local fracture toughness on the abrasive wear of WC-Co alloys. I Mech E conference Publications (1987) 277-282. 117. Larsen-Basse J. Resistance of cemented carbides to sliding abrasion: role of binder metal. eds. Viswanadham R.K., Rowcliffe D.J. and Gurland J., Science of Hard Materials Plenum Press, (1983) 707- 813. 118. Cuddon A. and Allen C. The wear of tungsten carbide-cobalt cemented carbides in a coal ash conditioner. Wear 153 (1992) 375-385. 119. Reshetnyak H. and Kübarsepp J. Structure sensitivity of wear resistance of hardmetals. International Journal of Refractory Metals and Hard Materials 15 (1997) 89-98. 120. Larsen-Basee J. Friction in two-body abrasive wear of a WC-Co composite by SiC. Wear 205 (1997) 231-235. 121. Batchelor A.W. and Stachowiak G.W. Predicting synergism between and abrasive wear. Wear 123 (1988) 281-291. 122. Nöel R.E.J. and Ball A. On the synergistic effects of abrasion and corrosion during wear. Wear 87 (1983) 351-361. 123. Barker K.C. and Ball A. Synergistic abrasive corrosive wear of chromium-containing steels. British Corrosion Journal 24 (1989) 222-228. 124. Kim K.Y. et al. An electrochemical polarization technique for evaluation of wear-corrosion in moving components under stress. in K.C. Ludema(eds), Wear of Materials, ASME New York, (1981) 772-778. 125. Kotlyar D.C. et al. Simultaneous corrosion and abrasion measurements under grinding conditions. Corrosion 44 (1988) 221-228. 126. Madsen B.W. Standard guide for determining amount of synergism between wear and corrosion. ASTM G119-93 (1994). 127. Fink J.B. et al. Electrochemical aspects of WC-Co drill bit wear. Wear 108 (1986) 97-101. 10 References 145

128. Tomlinson W.J. and Molyneux I.D. Corrosion, erosion-corrosion and the flexural strength of WC-Co hardmetals. Journal of Materials Science 26 (1991) 1605-1608. 129. Wentzel E.J. and Allen C. The erosion-corrosion resistance of tungsten carbide hard metals. International Journal of Refractory Metals and Hard Materials 15 (1997) 81-87. 130. Larsen-Basse J. and Liang H. Probable role of abrasion in chemo-mechanical polishing of tungsten. Wear 233-235 (1999) 647-654. 131. Underwood E.E. Quantitative Stereology. Addison-Wesley, Reading MA, (1970). 132. Tillwick D.L. and Joffe I. Magnetic properties of Co-W alloys in relation to sintered WC-Co compacts. Scripta Metallurgica 7 (1973) 479-484. 133. Schedler W. Harmetall für den Praktiker. VDI-Verlag, Düsseldorf, (1988). 134. Exner H.E. The influence of sample preparation on Palmqvist’s method for toughness testing of cemented carbides. Transactions of the Metal Society of AIME 245 (1969) 677-683. 135. Shetty D.K. et al. Indentation fracture of WC-Co cermets. Journal of Materials Science 20 (1985)1873-1882. 136. Palmqvist S. Rissbildungsarbeit bei Vickers-Eindrückeen als Maß für die Zähigkeit von Hartmetallen. Archiv für das Eisenhüttenwesen 33 (1962) 629-634. 137. Collenz A. Wear properties of ultrafine hardmetals with complex Fe-Ni-Co binder. Diplomarbeit. Universität Erlangen-Nürnberg (2001). 138. Sailer T. Ultrafeinkörnige Hartmetalle mit Co-Binder und alternativen Bindersystemen – Korrelation von Mikrostruktur und mechanischen Verhalten undet monoton ansteigender und zyklisch wechselnder Beanspruchung. Ph.D. thesis. University of Erlangen-Nürnberg (2002). 139. Sarin V.K. and Johannesson T. On the deformation of WC-Co cemented carbides. Metal Science 9 (1975) 472-476.

Wenn ein Eisen stumpf wird und an der Schneide ungeschliffen bleibt, muß man mit ganzer Kraft arbeiten. Aber Weisheit bringt Vorteil und Gewinn. Prediger 10:10

Danksagung

An dieser Stelle möchte ich allen danken, die direkt oder indirekt am Enstehen dieser Arbeit beteiligt waren. Mein besonderer Dank gilt dabei: • Herrn Professor Dr. H.-G. Sockel für die Übertragung und Betreuung dieser Arbeit und für viele Lösungen im wissenschaftlichen wie im praktischen Bereich • Frau Professor S. Luyckx, der Universität von Witwatersrand, Johannesburg, die das Projekt mitbetreut hat, für Ihre Unterstützung und die vielen guten Anregungen • Herrn Professor C. Allen, der Universität von Kapstadt, für sein stetes Interesse an meinem Projekt sowie die kritische Durchsicht meiner Dissertation • dem Forschungszentrum Jülich für die finanzielle Unterstützung des Projektes • der Firma Tigra und insbesondere Herrn Dr. R. Schulte und Herrn Dr. W. Feld für die Bereitstellung von Probenmaterial und für die vielen fruchtbaren Diskussionen • Herrn Professor Dr. P. Schmuki, der das Zweitgutachten meiner Arbeit geschrieben hat und der außerdem immer Zeit hatte, für Diskussionen zum Thema Korrosion • Herrn Dr. Michael Herr für sein Hilfe beim Versuchsaufbau • Herrn P. Fink für seine Hilfe bei der Holzzerspannung • der Firma Müller u. Seybold Sägewerk u. Holzhandlung, Forchheim, für die Bereitstellung verschiedener Holzarten • der Firma Osram, Schwabmünchen, für die Bereitstellung Ihre Feld Emission Rasterelektronmikroskop und Ihre freundliche Hilfe • Herrn Udo Schlierf und den anderen Mitarbeitern des Lehrstuhls für Korrosion und Oberflächentechnik, die bei der Durchführung der Korrosionsversuche mitgeholfen haben • Frau Dr. M. Sephton, der Universität von Witwatersrand, Johannesburg, für wichtige Diskussionen über das Thema Korrosion • meinen derzeitigen und ehemaligen Kollegen, darunter Paul Bellendorf, Andrea Collenz, Martina Felice, Günter Freimann, Heinz-Werner Höppel, Gregor Korn, Dr. Serge Kursawe, Werner Langner, Ingrit Lutz, Siphilisiwe Ndlovu, Dr. Phillip Pott, Dr. Vicky Pugsley, Dietmar Puppel, Dr. Dony Sailer, Sanjiv Shrivastava und Massimo Tolazzi für die freundliche Arbeitsatmosphäre • die „Kinderzimmer“ -Bewohner, Björn Backes, Ralf Nützel und Irena Topic, die meine letzen Monate am Lehrstuhl sehr unterhaltsam gemacht haben • all meinen Freunden für die seelische und moralische Unterstützung • weiterhin gilt mein besonderer Dank meinen Eltern und Bruder für ihre tatkräftige Unterstützung, nicht nur für meine Arbeit sondern auch in Sachen „Leben in Deutschland“...

Lebenslauf

Name: Natasha Sacks Geboren am: 16 Februar 1973 Geburtsort: Johannesburg, Südafrika Familienstand: ledig

Schule: 1979-1985 TC Esterhuysen Primary School, Johannesburg, Südafrika 1986-1990 Chris Jan Botha High School, Johannesburg, Südafrika

Studium: 1991-1996 University of Cape Town, Südafrika Abschluss: BSc. Eng (Mat) Materials Engineering 1997-1999 University of Cape Town, Südafrika Abschluss: MSc. (Eng) Materials Engineering

Beruf: seit 1999 Wissenschaftlicher Mitarbeiterin am Lehrstuhl I des Institut für Werkstoffwissenschaften (Prof. Dr. H. Mughrabi) in der Arbeitsgruppe von Prof. Dr. H.-G. Sockel