<%BANNER%>

Lubrication of High Current Density Metallic Sliding Electrical Contacts

Permanent Link: http://ufdc.ufl.edu/UFE0024957/00001

Material Information

Title: Lubrication of High Current Density Metallic Sliding Electrical Contacts
Physical Description: 1 online resource (145 p.)
Language: english
Creator: Bares, Jason
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: brushes, copper, current, fibers, friction, lubrication, profilometry, tribology
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Proper lubrication of sliding electrical contacts is necessary for efficient operation of electric motors and generators. In many cases the electric brush, which is responsible for transferring current between the stationary and moving parts of the circuit, is composed of a lubricant, such as graphite, to aid in reduction of friction and brush wear. For high current applications, the presence of a lubricating film can be detrimental to the electrical conductivity. Decoupling of graphite lubrication from the electric brush was investigated in ambient air environments. Low brush wear rates (on the order of 10^-11 m/m and lower) and low friction coefficients (0.15) were achieved at 40 A/cm^2, but brush wear increased at a current density of 200 A/cm^2. Copper fiber brushes offer improvements in electrical conductivity due to low bulk resistivity and large numbers of independent contact points. However, copper fiber brushes sliding on a copper rotor require humid non-oxidizing operating environments to reduce friction and wear. In the present study, copper fiber brush wear rates in humid carbon dioxide environments were comparable to graphite lubricated monolithic silver brushes, while brush electrical losses per ampere conducted were significantly lower with vapor phase lubricated copper fiber brushes (0.07 W/A) than with graphite lubricated silver brushes (0.7 W/A). From X-ray photoelectron spectroscopy analysis, chemisorbed carbon dioxide on the sliding surfaces reacted to form a carbonate species. Multiple monolayers of water adsorbed on to the sliding surfaces from the environment also aided in lubrication. Proper cooling of the sliding bodies was critical to maintaining the adsorbed water layers on the metal surfaces in humid environments. Brush wear showed a general trend of higher wear rates for positive brushes compared to negative brushes at high sliding velocities (greater than 1 m/s). Worn copper brush fibers examined by electron microscopy showed gross plastic deformation and shearing at fiber surfaces after sliding in humid carbon dioxide. The importance of tribofilm composition in the copper-copper humid carbon dioxide system suggested that other vapor phase additives could potentially be used to form and replenish thin carbon films on the metal sliding surfaces.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jason Bares.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Sawyer, Wallace G.
Local: Co-adviser: Dempere, Luisa A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024957:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024957/00001

Material Information

Title: Lubrication of High Current Density Metallic Sliding Electrical Contacts
Physical Description: 1 online resource (145 p.)
Language: english
Creator: Bares, Jason
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: brushes, copper, current, fibers, friction, lubrication, profilometry, tribology
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Proper lubrication of sliding electrical contacts is necessary for efficient operation of electric motors and generators. In many cases the electric brush, which is responsible for transferring current between the stationary and moving parts of the circuit, is composed of a lubricant, such as graphite, to aid in reduction of friction and brush wear. For high current applications, the presence of a lubricating film can be detrimental to the electrical conductivity. Decoupling of graphite lubrication from the electric brush was investigated in ambient air environments. Low brush wear rates (on the order of 10^-11 m/m and lower) and low friction coefficients (0.15) were achieved at 40 A/cm^2, but brush wear increased at a current density of 200 A/cm^2. Copper fiber brushes offer improvements in electrical conductivity due to low bulk resistivity and large numbers of independent contact points. However, copper fiber brushes sliding on a copper rotor require humid non-oxidizing operating environments to reduce friction and wear. In the present study, copper fiber brush wear rates in humid carbon dioxide environments were comparable to graphite lubricated monolithic silver brushes, while brush electrical losses per ampere conducted were significantly lower with vapor phase lubricated copper fiber brushes (0.07 W/A) than with graphite lubricated silver brushes (0.7 W/A). From X-ray photoelectron spectroscopy analysis, chemisorbed carbon dioxide on the sliding surfaces reacted to form a carbonate species. Multiple monolayers of water adsorbed on to the sliding surfaces from the environment also aided in lubrication. Proper cooling of the sliding bodies was critical to maintaining the adsorbed water layers on the metal surfaces in humid environments. Brush wear showed a general trend of higher wear rates for positive brushes compared to negative brushes at high sliding velocities (greater than 1 m/s). Worn copper brush fibers examined by electron microscopy showed gross plastic deformation and shearing at fiber surfaces after sliding in humid carbon dioxide. The importance of tribofilm composition in the copper-copper humid carbon dioxide system suggested that other vapor phase additives could potentially be used to form and replenish thin carbon films on the metal sliding surfaces.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jason Bares.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Sawyer, Wallace G.
Local: Co-adviser: Dempere, Luisa A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024957:00001


This item has the following downloads:


Full Text

PAGE 1

LUBRICATION OF HIGH CURRENT DENSITY METALLIC SLIDING ELECTRICAL CONTACTS By JASON A. BARES A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2009 Jason A. Bares 2

PAGE 3

To Mom, Dad, and my grandparents 3

PAGE 4

ACKNOWLEDGMENTS I would like to thank all of the members of my supe rvisory committeeDr. Bourne, Dr. Dempere, Dr. Hahn, Dr. Perry, and Dr. Phillpotfor their helpful and productive discussions. I would like to thank my advisor, Dr. Sawyer, for his guidance and instructi on. I had the privilege of working with an outstanding group of graduate students, and I would like to thank all of the current and former members of the University of Florida Tribology Labora tory for sharing their knowledge and making my graduate studies a memorable experience. I would also like to acknowledge the efforts of Nicola s Argibay and Dr. Pamela Dickrell, both of whom have made significant contributions to this study of sliding electrical contact s. I would like to thank Bret Windom, Greg Dudder, and Ben Raterman for th eir collaborations, Neal Sondergaard and Roy Dunnington for lending insight and experience to our discussions, and the Major Analytical and Instrumentation Center and its staff for use of th eir facilities. Finally, I would like to thank my entire family (Mom, Dad, Jennifer, Jordan, Am ber, Jered, and my grandparents) for their constant support. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................11 CHAPTER 1 BACKGROUND AND MOTIVATION................................................................................13 Introduction to Sliding Electri cal Contact Applications.........................................................13 Metal Fiber Brushes for High Current Applications..............................................................18 2 EXPERIMENTAL PROCEDURES.......................................................................................26 Decoupled Solid Lubrication of Sliding Electrical Contacts..................................................26 Brush-on-Rotor Tribometer Design................................................................................26 Solid Lubricant Preparation.............................................................................................28 Alternative Solid Lubricants fo r Sliding Electrical Contacts..........................................29 Environmental Effects on Copper Fiber Brush Sliding Electrical Contacts...........................30 Copper Fiber Brush Characterization..............................................................................30 Brush-on-Rotor High Sliding Velocity Testing..............................................................31 Environmental brush-on-ro tor tribometer design.....................................................32 Rotor subcooling......................................................................................................36 High sliding velocity test conditions........................................................................39 Polarity effects on arcing in copper fiber brush contacts.........................................39 Reciprocating Low Sliding Velocity Testing with Copper Fiber Brushes......................40 Reciprocating tribometer design..............................................................................41 X-ray photoelectron spectroscopy (XPS) characterization of wear tracks...............44 Low sliding velocity subcool study..........................................................................45 Pentanol vapor phase lubrication.............................................................................46 Single copper fibers in humid carbon dioxide.........................................................48 3 EXPERIMENTAL RESULTS...............................................................................................61 Decoupled Graphite Lubrica tion of Sliding Electrical Contacts in Ambient Air...................61 Polytetrafluoroethylene (PTFE) Solid Lubr icants for Sliding Electrical Contacts in Ambient Air.................................................................................................................... ....63 Characterization of Copper Fiber Brushes.............................................................................63 Fiber Brush Packing Frac tion and Fiber Hardness..........................................................63 Metal Fiber Brush Contacts.............................................................................................64 Fiber Brush Stiffness.......................................................................................................65 5

PAGE 6

High Sliding Velocity Studies of Copper Fiber Brush Sliding Electrical Contacts...............66 Copper Fiber Brush Sliding Electrica l Contacts in Humi d Carbon Dioxide Operating Environments..............................................................................................66 Polarity Effects on Arcing in Copper Fiber Brush Contacts...........................................68 Low Sliding Velocity Studies of Copper Fiber Brush Sliding Electrical Contacts................69 Copper Fiber Brush on Copper Fl at in Humid Carbon Dioxide.....................................69 Temperature Study in Humid Carbon Dioxi de and Humid Argon Environments..........73 XPS of Copper Sliding Surf aces in Humid Carbon Dioxide..........................................74 Single Copper Fiber on Copper Disk in Humid Carbon Dioxide...........................................76 Friction Coefficient a nd Contact Resistance...................................................................76 Electron Microscopy of Worn Fiber Surfaces.................................................................76 1-Pentanol as a Vapor Phase Lubricant..................................................................................78 1-Pentanol Vapor Phase Lubricatio n of Quartz-Silicon Sliding Pair..............................78 1-Pentanol Vapor Phase Lubricatio n of a Copper-Copper Sliding Pair..........................80 4 DISCUSSION................................................................................................................... ....103 Decoupled Solid Lubrication of Sliding Electrical Contacts................................................103 Graphite....................................................................................................................... ..103 PTFE and PTFE/In Composites....................................................................................105 Copper Fiber Brush Sliding Electrica l Contacts in Humid Environments...........................106 High Sliding Velocities.................................................................................................106 Low Sliding Velocities..................................................................................................110 Composition of Tribofilms Formed on Copper in Humid Carbon Dioxide..................114 Vapor Phase Lubrication of Metal Sliding Electrical Contacts with 1-Pentanol..........119 Polarity Effects on Friction and Contact Resistance.....................................................120 Summary of Brush Electrical Losses............................................................................122 Rotor Wear....................................................................................................................1 23 Theories on Brush Wear Mechanisms in Metal Fiber Brush Sliding Contacts.............125 Arcing and electromigration...................................................................................126 Oxidation................................................................................................................128 Mechanical wear....................................................................................................130 5 SUMMARY AND CONCLUSIONS...................................................................................135 LIST OF REFERENCES.............................................................................................................138 BIOGRAPHICAL SKETCH.......................................................................................................145 6

PAGE 7

LIST OF TABLES Table page 3-1 Summary of tribological testing with copper fiber brushes and polytetrafluoroethylene (PTFE) lubricants........................................................................86 3-2 Copper counterface wear rates for the negative counterface current condition.................92 3-3 Copper counterface wear rates for th e positive counterface current condition.................92 7

PAGE 8

LIST OF FIGURES Figure page 1-1 Schematic drawing of electric mo tor slip ring with brush contacts...................................23 1-2 Schematic drawing of lamellar graphite structure.............................................................23 1-3 Electric brush materi al capability plot...............................................................................24 1-4 Plots of mechanical, electrical, and tota l losses per ampere for metal fiber brush contacts..............................................................................................................................25 2-1 Line drawing of brush-rotor tribometer for testing of solid lubricants [53]......................50 2-2 Exploded view of brush thruster and load cell assembly line drawing..............................51 2-3 Schematic drawing of the electrical circuit for the brush-rotor tribometer.......................52 2-4 Photograph of as-received copper fiber brush (no flam e-sprayed zinc coating)...............52 2-5 Line drawing of testing apparatus for copper fiber brush stiffness measurements. ..........53 2-6 Line drawing of the modified brush-rotor tribometer for environmental testing. Base and frame have been removed for clarity [54]...................................................................54 2-7 Line drawing of brush holder assembly from brush-rotor environmental tribometer [54].....................................................................................................................................55 2-8 Schematic diagram of electrical circuit for brush-rotor environmental tribometer [54].....................................................................................................................................56 2-9 Plot of isotherms from 0 to 100C for rela tive humidity versus partial pressure of H2O....................................................................................................................................57 2-10 Schematic line drawing of linear recipro cating tribometer with in situ profilometry capabilities................................................................................................................... ......57 2-11 Line drawing of brush holder assembly and stage assembly fo r linear reciprocating tribometer..................................................................................................................... ......58 2-12 Schematic drawing of the wear track created on the counterface by the linear reciprocating tribometer.....................................................................................................59 2-13 Modified setup of linear reciprocating tribometer to accommodate samples for X-ray photoelectron spectroscopy (XPS) analysis.......................................................................60 2-14 Photograph of the testing apparatus for single fiber experiments......................................60 8

PAGE 9

3-1 Friction coefficient and brush wear for graphite lubricated silver brush sliding electrical contacts in ambient air........................................................................................83 3-2 Energy dispersive X-ray spectroscopy (E DS) of worn silver brush surfaces....................84 3-3 Scanning white light interferometry image of graphite transfer film on copper rotor surface................................................................................................................................84 3-4 Micrographs used in calculation of copper fi ber packing fractions...................................85 3-5 Load versus deflection plot for a filamentous copper brush in compression....................85 3-6 Brush wear and voltage drop from tribological testing of co pper fiber brushes conducted on the environmental brus h-rotor tribometer in a humid CO2 environment....87 3-7 Micrographs of copper fibe r brushes after high sliding speed (5 m/s) test on the environmental brush-rotor tribometer in humid CO2.........................................................88 3-8 Surface profiles of worn copper fibers obtained by scanning white light interferometry after high sliding speed testing (5 m/s) in humid CO2...............................88 3-9 Arc lifetimes for positive and negative brush polarities....................................................89 3-10 Atomic emission spectra collected from an arcing copper fiber brush contact while sliding in ambient air......................................................................................................... 90 3-11 Average contact resistance and friction coefficient for low speed (0.01 m/s) copper fiber brush on copper counterface sliding in humid CO2 at current densities of 0 and 180 A/cm2..........................................................................................................................91 3-12 Calculated volume loss for a copper count erface sliding against a copper fiber brush in humid CO2 environment at 0.01 m/s.............................................................................92 3-13 Scanning electron microscopy (SEM) wear track characterizati on after tribological testing of copper fiber brushes at low sliding velocities (0.01 m/s) and 180 A/cm2 in humid CO2.........................................................................................................................93 3-14 Average friction coefficient and contact re sistance for a copper fiber brush in low speed (0.01 m/s) reciprocating sliding against a polis hed copper counterface in humid CO2 and humid argon environments......................................................................93 3-15 XPS analysis of copper counterface wear tracks after low speed (0.01 m/s) sliding in humid CO2 environments...................................................................................................94 3-16 Current-voltage (I-V) sweep of single copper fiber on copper disk contact after sliding in humid CO2 environment....................................................................................95 3-17 SEM micrographs of individual worn copper fibers after sliding in humid CO2. ...........95 9

PAGE 10

3-18 SEM micrographs of worn copper fiber cr oss sections created by focused ion beam milling........................................................................................................................ ........96 3-19 Transmission electron microsocpy (TEM) a nd EDS analysis of lift-outs from worn copper fiber surfaces.......................................................................................................... 97 3-20 Average friction coefficient for each reci procating cycle of quartz-on-silicon sliding in dry argon and 1-pentanol saturated argon environments...............................................98 3-21 Profilometry of wear track formed on s ilicon during reciproca ting sliding against a quartz ball in dry argon and 1pentanol saturated argon....................................................99 3-22 Friction coefficient for copper-on-copper reciprocating sliding (no current) in 1pentanol saturated argon e nvironment over 100 cycles...................................................100 3-23 Profilometry images of polished copper counterface showing evolution of wear track in 1-pentanol saturated argon environment......................................................................100 3-24 Average line profiles for wear track formed on copper counterface in 1-pentanol saturated argon environment............................................................................................101 3-25 Contact resistance and fric tion coefficient for copper fiber brush on copper flat low sliding speed (0.015 m/s) test in humid CO2 (blue) and 1-pentanol saturated argon (black) environments.......................................................................................................102 4-1 SEM image and EDS dot maps of sphe rical and rod-like st ructures found on the surface of a worn copper fiber br ush after testing in humid CO2 environment...............132 4-2 Water adsorption study on Cu2O using ambient pressure XPS (adapted from Salmeron [82])................................................................................................................. 133 4-3 Schematic drawing of surface films pres ent on a copper-copper sliding contact in humid CO2.......................................................................................................................133 4-4 Summary of calculated brush Ohmic loss per ampere conducted for various brush and lubricant pairs............................................................................................................ 134 10

PAGE 11

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LUBRICATION OF HIGH CURRENT DENSITY METALLIC SLIDING ELECTRICAL CONTACTS By Jason A. Bares August 2009 Chair: W.G. Sawyer Major: Materials Science and Engineering Proper lubrication of sliding electrical cont acts is necessary for efficient operation of electric motors and generators. In many cases the electric brush, which is responsible for transferring current between the stationary and m oving parts of the circuit, is composed of a lubricant, such as graphite, to aid in reduction of friction and brush wear. For high current applications, the presence of a lubricating film can be detrimental to the electrical conductivity. Decoupling of graphite lubrica tion from the electric brush was investigated in ambient air environments. Low brush wear rates (on the order of 10^-11 m/m and lower) and low friction coefficients (0.15) were achieved at 40 A/cm^2, but brush wear incr eased at a current density of 200 A/cm^2. Copper fiber brushes offer improvements in electrical conductivity due to low bulk resistivity and large numbers of independent contact points. However, copper fiber brushes sliding on a copper rotor require humid non-oxidi zing operating environments to reduce friction and wear. In the present study, copper fibe r brush wear rates in humid carbon dioxide environments were comparable to graphite lubricated monolithic silver brushes, while brush electrical losses per ampere conducted were si gnificantly lower with vapor phase lubricated copper fiber brushes (0.07 W/A) than with graphi te lubricated silver br ushes (0.7 W/A). From 11

PAGE 12

X-ray photoelectron spectroscopy an alysis, chemisorbed carbon dioxide on the sliding surfaces reacted to form a carbonate species. Multiple monolayers of water adsorbed on to the sliding surfaces from the environment also aided in lubri cation. Proper cooling of the sliding bodies was critical to maintaining the adsorbed water laye rs on the metal surfaces in humid environments. Brush wear showed a general trend of higher wear rates for positive brushes compared to negative brushes at high sliding velocities (greater than 1 m/s). Worn copper brush fibers examined by electron microscopy showed gross plastic deformation and shearing at fiber surfaces after sliding in humid carbon dioxide. The importance of tribofilm composition in the copper-copper humid carbon dioxide system suggest ed that other vapor phase additives could potentially be used to form and replenish th in carbon films on the me tal sliding surfaces. 12

PAGE 13

CHAPTER 1 BACKGROUND AND MOTIVATION Introduction to Sliding Electrical Contact Applications The passage of current across a dynamically changing sliding interface poses a number of material design considerations. Electrical cont acts typically utilize soft, noble metals which create large contact areas and lo w contact resistances. Sliding contacts tend to make use of high hardness materials to minimize contact area as well as low shear strength, electrically insulating lubricating layers to reduce fric tion. In material selection for sliding electrical contacts, the tradeoffs between mechanical and electrical prope rties become apparent. Optimization of sliding electrical contact material se lection is often governed by the requirements of the specific application. Sliding electrical contacts are commonly found in electric motors and generators [1-11]. A stationary brush acts as an electrode, transfer ring current across an interface to (or from) a rotating body (rotor, slip ring, or commutator). Electric motors rely on the in teraction between moving charged particles and magnetic fields to produce mechanical motion. For example, Figure 1-1 shows a schematic drawing of a slip ring from an electric motor. Current enters the slip ring at the positive brush, travels axially along the slip ring, and then leaves the slip ring at the negative brush. The interacti on between the axial current flow through the slip ring and the radial magnetic field creates a to rque which rotates the slip ring. The Lorentz force (F) is given by Equation 1-1, where q is the charge on the particle v is the velocity of the particle, and B is the magnitude of the magnetic field. FqvB (1-1) 13

PAGE 14

In Figure 1-1, current flow directio n is given rather than electron flow direction, so the particle charge is positive; thus, the direction of rotation is clockwise. The role of the brushes is to maintain a completed electrical circuit with the rotating body. Electrical brush design has evolved greatly sin ce their initial use in generators and later electric motors. Kendall, McNab, and Wilkin [6] performed a review of current collection systems. The earliest electrical brushes (19th century) were constructed of metal fibers, generally copper, because of their high conductivity [6, 12]. Although these metallic brushes had low contact resistance, the wear rate s, as well as wear rates of th e opposing surfaces, were extremely high. The poor tribological perf ormance of metal brushes direc tly led to the development of carbon-based electrical brushes in the early 1900s. Carbon brushes are still in use today in low power electric motors. One of the more common forms of the carbon brush is the electrographitic brush. In general, lamellar solids such as graphite, molybdenum disulfide (MoS2), and tungsten disulfide (WS2) are excellent solid lubricants. The lamellar structure of graphite was first discovered by Bragg in 1928 by x-ray diffraction [13]. Th e graphite crystal structure, space group P63/mmc, consists of parallel layers (basal planes) of sp2 bonded carbon atoms in a hexagonal array, as depicted in Figure 1-2. The laye rs are stacked in an ABAB fash ion to yield a hexagonal crystal structure. The sp2 bonding configuration creates a stable local chemical environment for each carbon atom. The carbon-carbon bond length within a single layer is 1.42 [14]. Graphite basal planes are weakly held together by van der Waals forces. The weak interlayer forces allow individual layers to easily slide past each ot her, resulting in low shear strength for the bulk material and generally low friction. 14

PAGE 15

The tribological properties of graphite ar e not intrinsic and are highly dependent on operating environment. This phenomenon was firs t encountered during World War II in electric motors onboard high altitude aircraft [15-18]. Graphite br ushes from electric motors experienced unexpectedly high wear rates at th e higher altitudesthis rapid wear of carbon brushes became known as dusting. It was determined that the absence of oxygen, carbon dioxide, and water vapors at hi gh altitudes led to high friction and poor wear resistance of graphite brushes [18-20]. A t ypical wear rate for aircraft generator dusting is 5 10-8 m/m, while automotive starter brushes under normal us e wear at a rate of approximately 1 10-9 m/m [21]. Other studies have estab lished improved performance of graphite-based brushes in humid carbon dioxide (CO2) environments [22-25]. Even sma ll additions of water vapor to the operating environment can improve the tribologi cal performance of carbon materials [26]. One of the first theories regarding the low friction of graphite was devised by Bragg [13] based on diffraction studies. He proposed that weak interplanar forces allowed for easy shearing of the bulk material and low fric tion behavior. Although this is true in ambient environments, low friction does not persist in all environments, as demonstrated by the dusting of the graphite brushes from high-altitude aircra ft motors. A revision to Bragg s lattice shear theory proposed by Rowe [27] discussed the penetr ation of vapor molecules into the graphite lattice to form an intercalation layer. The presen ce of an intercalation layer w ould reduce the attractive forces between graphite layers, resulti ng in decreased shear strength a nd lower friction. In addition, the intercalation layer would be expected to increas e the spacing between layers. X-ray diffraction studies of graphite exposed to water vapor have not shown the pr edicted lattice expansion due to intercalation layer formation [ 14, 28]. Deacon and Goodman [29] proposed that atoms along the edges of basal planes controlled the frictional properties of gra phite through edge-face and edge15

PAGE 16

edge interactions with neighboring basal planes Edge sites, which have a single dangling bond, can readily react with molecules in the gas phase to create a low energy surface. Zaidi [30] reported surface energies for graphi te face and edge sites of 0.2 J/m2 and 5 J/m2, respectively. If the dangling bonds remain unsaturated, they w ill increase the attractive forces between the basal planes. Therefore, by saturating the da ngling bond sites through ga s phase adsorption, low friction can be achieved accordi ng to Deacon and Goodman [29]. Figure 1-3 is a plot of the eff ective operating range, in terms of sliding velocity and current density, for various electric br ush materials (adapted from Ke ndal, McNab, and Wilkin [6]). Graphite brushes can function effi ciently over a wide range of sliding velocities owing to their excellent tribological prope rties; however, graphite brushes are limited to low current densities. Johnson and Moberly [22] found go od friction and wear performa nce with electrical-grade graphite brushes at the expense of electrical losses. The bulk re sistivity of graphite combined with the resistance of graphite transfer films fo rmed during sliding lead to significant electrical losses for the system. Temperature rises in resp onse to Ohmic heating of the graphite brushes are another obstacle to high current operation. Graphite brush wear increases with increasing temperature because of the loss of adsorbed ga s species which provide graphite with its low friction and low wear behavior. Additionally, the heat produced through frictional and Ohmic heating must be continuously dissipated through some coo ling mechanism, which puts an additional burden on the system. Additions of metal fillers have been used to improve the electrical conductivity of graphite brushes. The graphite component of metal-graphite composit e brushes provides low friction transfer films while the metal component provid es a low resistance pathway for current flow. The improvement in conductivity over carbon br ushes comes at the expense of mechanical 16

PAGE 17

losses. Metal-graphite brushes are typically li mited to lower sliding velocities than carbon brushes (see Figure 1-3). Metal-graphite compos ite brushes have been extensively studied for high current density applications [1, 7, 16, 23-25, 31-41]. McNab [7] provides a thorough review of voltage drop, friction coefficient, a nd wear rate data for various metal-graphite electrical grade brushes under a wi de range of operating conditions. Metal-plated carbon fiber brushes offer furthe r improvements in elec tric brush technology [6, 42]. Fiber brushes provide a large number of independent poi nts of contact (approximately one contact spot per fiber [12]) while monolithic brushes are restricted to a much smaller number of contact spots (on the order of ten contact spots per brush [33]). As a result, fiber brushes can operate under lower applied loads than mono lithic brushes without increases in contact resistance. Fiber brushes are more compliant than monolithic brushes and are better able to track eccentricities in the roto r surface at high sliding velocities. The presence of carbon in the brush material aids in the reduction of friction, but these metal-plated carbon fiber brushes still suffer significant electrical losses at high currents largely due to the bul k resistance of the carbon-based brush and the resistance of carbon transfer films formed during sliding. Many high power motor applications have been limited by the lack of an electric brush design that is sufficient for high sliding speeds an d high currents. Liquid metals, such as sodium potassium and low melting point bism uth-based alloys, have been th e subject of study for current collection systems [3, 6, 43-45]. Wear is not a concern with liquid meta l current collectors. High current densities (up to 3100 A/cm2 [6]) can be managed with liquid metals since the current is distributed over a large uniform area as opposed to being concentrated at a brush contact. Liquid metals have certain drawbacks, namely that they require careful handling and protective inert environments due to their reactivity [7]. 17

PAGE 18

Metal Fiber Brushes for High Current Applications Recently, high current electric brush research ha s reverted back to metal fiber brushes. The large number of independent contact spots allow for lower brus h pressures to be used with metal fiber brushes, which helps to reduce overa ll mechanical losses de spite the high friction relative to graphite containing brus hes. A number of studies have shown that metal fiber brushes can efficiently operate in humid, non-oxidizing envi ronments with wear rates of approximately 10-11 m/m [8, 12, 46-48]. Brush wear is typically determined through length measurements, and brush wear rates are quantified by the ratio of cha nge in length to total sliding distance. Wear rate is an important proper ty in fiber brush design [ 12]. For a wear rate of 10-11 m/m at a sliding velocity of 5 m/s, the total brush wear will be less than 2 mm per year. Brushes must be able to accommodate this wear without negatively impacti ng performance. Because metal fiber brushes do not use solid lubricants, they demonstrate great potential for high pow er current collection systems. Three factors contribute to the resistance of electrical brushe s, as shown in Equation 1-2 [12]. 0 BFRRRR C (1-2) The total brush resistance (RB) is given by the sum of the bulk resistance (R0), the surface film resistance (RF), and the constriction resistance (RC). Constriction resistance refers to the constriction of current flow lines at indivi dual contact spots [49]. All three components contribute to the total resistance for graphite brushes. For metal fi ber brushes, the bulk resistance and constriction resistan ce are very small and can be assumed to be negligible [12]. Therefore, the total resistance for metal fiber brushes is dominated by the surface films. 18

PAGE 19

Although not shown in the electric brush materi al capability plot of Figure 1-3, metal fiber brushes have operated at velocities (up to 40 m/ s [46]) and current densities (up to 1240 A/cm2 [8, 48]) comparable to other elect ric brush materials. The filame ntous nature of the brush gives it distinct advantages over monolithic brushes. Monolithic carbon brushes require higher loads than fiber brushes to maintain continuous contact with the rotor or slip ring surface, and higher loads result in higher wear and higher frictional losses. Carbon brushes al so generate significant amounts of carbon debris that can be deposited throughout the machine. The carbon dust can create alternate pathways for current flow, leading to ina dvertent shorting or grounding of the circuit [2]. Debris from metal fiber brush w ear is much denser than carbon dust and, although capable of creating electrical s horts, metal debris is less easily transported throughout the machine. Binders in carbon brushes can degrad e at high temperatures typical of high current operation, further reducing wear resistance of carbon brushes. Kuhlmann-Wilsdorf [47] analyzed specific brush loss, defined as the heat generated per ampere conducted, to determine the optimal brush pressure and current density for metal fiber brush contacts. Mechanical losses, due to frictio nal heating, and electrical losses, due to Ohmic heating, contribute to the overa ll heat loss of a brush cont act. The mechanical loss (LM) per unit of current is a function of friction coefficient (), sliding velocity ( ), brush force (P), and current (I), and is given by Equation 1-3. MP L I (1-3) The electrical loss (LE) per unit of current is a function of the overall brush resistance and current, and is given by Equation 1-4. EBLR I (1-4) 19

PAGE 20

The total loss for a brush contact is the sum of th e mechanical and electric al terms, as shown in Equation 1-5. TMLLL E (1-5) Kuhlmann-Wilsdorf [47] applied this analysis to gold fiber brushe s operating in humid argon environments. Shown in Figure 1-4 are sche matic plots of the brus h losses per ampere as a function of pressure and current density. Mech anical losses increase with increasing brush pressure while electrical losses decrease with increasing brush pressure. The total brush resistance, RB, is dependent on the real ar ea of contact, which in turn is dependent on the applied load. As the load increases, the real area of cont act increases due to an increase in the size and number of contact spots through deformation of the mating surfaces The increased contact area gives rise to decreased brush re sistance and decreased electrical losses. In terms of current density, electrical losses are di rectly proportional to the current and mechanical losses are inversely proportional to the current. Minimization of total losses occurs when mechanical and electrical losses per ampere ar e approximately equivalent. At high current densities electrical losses dominate total brush losses. Gubser [5] lists typical values for high power (25 MW) electri c motors of approximately 50,000 A and 500 V, with linear sliding velocities of approximately 15 m/s. The superiority of metal fiber brushes over graphite and metal-graphite composite brushes at high current s is apparent when considering Figure 1-4. The low friction coefficients and low m echanical losses for graphitebased brushes are far outweighed by the high electr ical losses. While the total resistance of graphite-based brushes is composed of bulk re sistance, film resistance, and constriction resistance, only film resistance significantly cont ributes to the total resistance of metal fiber brushes. Moreover, the film resi stance of a metal fiber brush is orders of magnitude lower than 20

PAGE 21

the resistance of the transfer f ilm formed under graphite-containi ng brushes [33]. The inherently low resistance of metal fiber brushes compared to other potential brush materials becomes increasingly important at high cu rrent densities. Metal fiber br ushes are ideally suited for high current, low voltage applications because of thei r low bulk resistivity, low film resistivity, and low constriction resistance due to the la rge number of contact spots [47]. Operating environment is a fundamental part of efficient metal fiber brush operation. Oxidation of the metal contact surfaces can be detrimental to high current applications. Nonoxidizing cover gases are typically used to minimi ze oxide film growth. Inert gas environments can also be used to prevent surface reactions which may yield other non-conductive surface films. However, lubricating surface films are desired to prevent adhesion in the metal-on-metal contact and improve brush lifetimes through reduced friction and wear. Operating environments saturated with water (greater than 95% relative humidity) produce thin layers of adsorbed water on the sliding surfaces which act as a boundary lubricant. Metal fiber brushes have also been tested under high currents in ambient air [2, 50] and dry inert envi ronments [51] with reasonable success. Water-saturated CO2 environments, which have been successfully employed with high current metal-graphite brushes [23-25, 33, 37, 39], have also been utilized with metal fiber brushes [8, 48, 52], although the mechanisms by which humid CO2 environments lubricate metal fiber brush sliding electrical contacts are not well understood. The primary goal of this study was to examine lubrication methods for high current density metallic sliding electrical contacts. Solid lubric ants have traditionally been incorporated into electrical brush materials. Decoupling of the solid lubrican t component from the metal brush was investigated for lubrication of high current density sliding contacts in ambient air. Solid lubricant transfer films can eff ectively lubricate metallic sliding electrical contacts without the 21

PAGE 22

aid of a protective environment. Graphite was studied because of its relatively high conductivity compared to most solid lubrican ts. Polytetrafluoroethylene (PTFE), which is a poor electrical conductor, was selected for study be cause of its ability to form ve ry thin lubricating transfer films. Film resistivity is viewed as the limiting f actor for use of solid lubricants in high current applications. Independent brush and lubricant loading were tested as a means of reducing transfer film thickness and re ducing electrical losses. An alternative to solid lubr ication is vapor phase lubri cation of high current metallic sliding electrical contacts. Va por phase lubrication of metal fi ber brush contacts on a metal slip ring present a promising combination for high curr ent operation. Adsorbed gas species are used to reduce friction and wear while minimizing elec trical losses. The adsorbed films formed through vapor phase lubrication are much thinner than the solid lubrican t transfer films, and accordingly, the resistance across the adsorbed films is lower as well. Various surface characterization techniques were used in this study to examine the in teraction between copper fiber brush sliding electri cal contacts and a humid CO2 operating environment and to understand the chemistry of tribofilms formed in the contacts. Rotor wear wa s studied in situ at high current densities (180 A/cm2) in a controlled humid CO2 environment using non-contacting profilometry. Polarity effects on contact resistan ce, friction, and wear we re also investigated over a range of sliding velocities for copper fiber brush contacts. Ultimately, the purpose of the vapor phase lubrication study wa s to relate surface chemistry to the wear mechanisms and electrical properties of copper fiber brush slidin g electrical contacts. 22

PAGE 23

Figure 1-1. Schematic drawing of electric motor slip ring with brush contacts (adapted from Superczynski and Waltman [44]). The radial magnetic field ( B ) interacting with the axial current flow ( I ) through the slip ring creates a torque which drives slip ring rotation in the clockwise direction. Figure 1-2. Schematic drawing of lamellar graphite structure. 23

PAGE 24

Figure 1-3. Electric brush materi al capability plot (adapted from Kendall, McNab, and Wilkin [6]). 24

PAGE 25

A B Figure 1-4. Plots of mechanical (LM), electrical (LE), and total (LT) losses per ampere for metal fiber brush contacts (adapted from Kuhl mann-Wilsdorf [47]). A) Dependence of losses on brush pressure. B) Dependen ce of losses on current density. 25

PAGE 26

CHAPTER 2 EXPERIMENTAL PROCEDURES Decoupled Solid Lubrication of Sliding Electrical Contacts Brush-on-Rotor Tribometer Design A brush-rotor tribometer was designed to test solid lubricants in a unidirectional sliding brush-on-rotor configuration [53]. Figure 2-1 presents an isometric line drawing of key tribometer components (frame components ar e removed for simplicity). A 500 W (0 RPM) motor was used to drive a copper disk dire ctly mounted to a keyed shaft. The disk was made of solid copper (UNS C11000, 99.9% pure) and had a diameter of 152.4 mm and thickness of 25.4 mm. A polyetheretherketon e (PEEK) hub was designed to isolate the motor from current flow through the brush/rotor electrical path. Th e brushes were located on independent tracks on the lateral surface of the disk and were mounted to an electrically controlled pneumatic air cylinder and thruster system for load applica tion. The compressed-air-driven pneumatic system provided brush load and lubricant lo ad control in the range of 0 N. Figure 2-2 provides a detailed view of the thruster assembly, including additional sensors and instrumentation [53]. The brush/thruster assemblies were mounted to the frame through individual six-channel multi-axis load cells. Th e load cells, rated for a maximum load of 200 N and maximum torque of approximately 10 N-m, were obtained from Mechanical Technology, Inc. (Watertown, MA). Each brush/thruster assembly was fitted with a magnetic field sensing linear variable differential transformer (LVDT) to measure linear wear of the brushes. The LVDTs, from RDP Electrosense (Pottstown, PA) had a travel of 10 mm in the linear output signal range and an uncertainty of 5.2 m. All brush wear rates were reported in nondimensional units of linear brush wear per unit of sliding distance (m/m), which is traditional for electrical brush research. 26

PAGE 27

The graphite solid lubricant was applied to the rotor surface using electro-pneumatic thruster assemblies similar to those used for the brushes, with the exception of a load cell. A description of the solid lubrican t preparation can be found in the Solid Lubricant Preparation section. The lubricant applic ation force was approximated using the pneumatic cylinder diameter and measured voltage input to the calibrate d regulator system. A LVDT was mounted to each lubricant applicator to measure lubrican t wear. For long duration tests, graphite pins were replaced as needed. Without stopping the ro tor, the load was removed from the depleted lubricant pins, and the lubrican t applicators were lowered away from the rotor. The used graphite pins were removed and replaced with new pins. The new gra phite pins were then brought into contact with the rotor surface, and load was reapplied. Lubricant pin load was maintained at approximately 5 N. Monolithic silver (99.95% Ag) and silver alloy (92.5% Ag, 7.5% Cu by weight) brushes for conducting current into and out of the copper ro tor were used in this study. Silver brushes were selected to investigate the transfer of ma terial between brush and rotor surfaces. Silver alloy brushes were utilized in additional tes ting because the alloy had a higher hardness than pure silver and detailed study of material transfer was no longer n eeded. Material for the brushes was obtained from Surepure Chemetals (Florham Par k, NJ). Cylindrical silver and silver alloy rods (6 mm diameter) were cut to lengths of approximately 10 mm for each brush. The contacting faces of the metal brushes were machined to a nominal area of 27 mm2. A K-type thermocouple was attached to the side of each brush, and a non-contacting infrared temperature sensor was used to measure the temperature of the copper rotor. The copper rotor surface was lathe-turned and lightly sanded with 240-grit silicon carbide paper. To ensure that the radius of curvature of the brushes matched the curvature of the rotor, the brushe s were lightly loaded 27

PAGE 28

against the rotor, which was wrapped with 240grit silicon carbide paper, and the rotor was slowly turned by hand. The paper was then remove d, and the graphite lubricant pins were loaded against the freely rotating rotor to develop a transfer film. When a uniform graphite transfer film was visible on the rotor surface, the brushes were loaded against the rotor to begin the run-in process. Brush normal load was maintained at 5.5 N (2.0 105 Pa), and the linear sliding speed of the rotor surface was 1.6 m/s. All testing with graphite lubr icants was conducted in ambient air. A 0 A, 0 V DC Instek power supply (o btained from Fotronic Corporation; Melrose, MA) was used to provide the desired curre nt. A diagram of the el ectrical circuit for the brush-rotor system is shown in Figure 2-3 [53]. For this stud y, the brush that had electrons impinging on its sliding surface (Figure 2-3, left) was referred to as the positive brush, and the brush that was emitting electrons from its sliding surface (Figure 2-3, right) was referred to as the negative brush. Current flow direction, wh ich is opposite the direction of electron flow, occurred from the positive brush, through the ro tor, to the negative brush. The brush and lubricant pin holders had insulating spacers to separate the sensors (LVDTs and load cells) from the current path. Voltage drop was measured across the brush-rotor contacts, and current through the brush path was measur ed using an ammeter. Contact resistance, which represented the combined resistance across both brush contacts, was calculated based on the measured voltage drop and measured current. Brush norma l and frictional loads, br ush and lubricant wear, voltage drop, current, brush temperature, and ro tor temperature were recorded using LabView software. Solid Lubricant Preparation Solid lubricant pins were formed from gra phite powder (Dixon Southwestern Graphite, Lakehurst, NJ) with an average particle size of 2 m. A mass of 1000 mg of graphite powder 28

PAGE 29

was used for each pin. The powder was compressed in a cylindrical titanium mold (6.4 mm inner diameter, 25.4 mm outer di ameter, 50.8 mm long) using a cy lindrical drill blank (6.4 mm diameter) and a Carver pneumatic press. The po wder mold was held at a load of 22,200 N for 30 s at room temperature. No additional binder was us ed to hold the graphite particulates together. After removing the graphite pin from the mold, the faces of the pin were lightly sanded to remove the flash and eccentricities other irregular ities left over from the molding process. The final length of the lubricant pi n was approximately 15 to 20 mm. Alternative Solid Lubricants for Sliding Electrical Contacts In addition to graphite, polytetrafluoroethyl ene (PTFE) and PTFE-based solid lubricants were also investigated due to their ability to form thin, lubricious transfer films. Of particular interest was a PTFE-indium (PTF E/In) composite (10 vol% indium). Although PTFE is a very poor conductor, thin transfer films permit meta l-metal contact at th e sliding interface for adequate electrical conductivit y. The addition of indium to PTFE created a conductive phase within the transfer film to further reduce cont act resistance. Because indium has a low shear strength, the frictional properties of the system s hould not be adversely affe cted. The brush-rotor tribometer (Figure 2-1, Figure 2-2, and Figure 2-3) was used to te st the effectiveness of PTFE and PTFE/In as lubricants for sliding electrical contacts. Samples of PTFE and PTFE/In were machined into pins from bulk material to fi nal dimensions of appr oximately 8 mm by 10 mm by 20 mm. The sliding surface of the lubricant pi ns had a surface area of approximately 80 mm2. The base of each pin was machined into a 6 mm diameter cylinder (approximately 5 mm in length) to fit into the lubricant applicator holders. Filamentous copper brushes, supplied by SSI T echnologies, Inc., were used with the PTFE and PTFE/In solid lubricants. A photograph of one of the as-received copper fiber brushes is shown in Figure 2-4. The brush consisted of a bundle of vertically aligned copper fibers 29

PAGE 30

wrapped in a copper wire mesh. Each indi vidual copper fiber had a diameter between approximately 60 and 70 m. The copper fiber bundle was soldered to a solid copper base, which was used as a conductive pathway to deliv er current to the fiber pack. The fiber bundle measured approximately 10 mm in length, 5 mm in width, and 10 mm in height. More details on the properties and characterizat ion of the copper fiber brushe s can be found in the Results (Chapter 3) section titled Char acterization of Copper Fiber Brushe s. The brushes had a slight radius of curvature (~76 mm) mach ined into their surfaces to ma tch the curvature of the copper rotor. Environmental Effects on Copper Fib er Brush Sliding Electrical Contacts Copper Fiber Brush Characterization As-received copper fiber brushes were phot ographed and characterized using scanning electron microscopy (SEM) prior to testing. SEM micrographs were used to calculate the fiber packing fraction of each brush and qualitativ ely evaluate the surf ace morphology of the asreceived fibers. Because brush wear was quantified by length change, brush mass was not recorded prior to testing. Brush wear rate de termination from mass measurements tends to be unreliable because of many factors, including debris accumulation in the void spaces between fibers. Prior to tribological testing, unworn copper fibers were sectioned and removed from a brush for microhardness measurements. The fibers were affixed to a stainless steel sample clip using a cyanoacrylate adhesive. The sample clip and fibers were then mounted in quick-curing epoxy with the fibers vertically aligned. Th e epoxy-mounted samples were prepared through wet-grinding with progressively finer silic on carbide paper from 400-grit through 600-grit to create transverse cross sections of the coppe r fibers. Final polishing was performed using alumina particle slurries (0.9, 0.3, and 0.05 m) on a synthetic velvet cloth. A CSM Micro30

PAGE 31

Combi Tester (CSM Instruments; Needham, MA) was used to evaluate the microhardness of the as-received copper fibers. A Vickers diamond tip i ndenter was used for all measurements. The applied load was linearly ramped from 0 to 98 mN over 30 s, followed by a 10 s hold at 98 mN, and finally a linear decrease of load from 98 to 0 mN over 30 s. Because the fiber diameter was approximately 70 m, only one indent could be ma de per fiber under the given load. An optical microscope (Leica DM LM) was used to image the indents. Dimensions of each indent were measured and used to calculate the hardness. The stiffness of an as-received copper fi ber brush was measured under compressive loading using the testing appara tus shown in Figure 2-5. A copper fiber brush was attached to the underside of the PEEK flexure. The flexure was dead-weight loaded to bring the brush into contact with a flat copper surface. A vertically mounted LVDT was attached to the backside of the flexure to measure displacement. Voltage output from the LVDT was recorded using LabView data acquisition software. Applied load was incrementally increased to 20 N and then incrementally decreased to 0 N to generate lo ading and unloading curves for the copper fiber brush. Brush-on-Rotor High Sliding Velocity Testing In the present study, high sliding velocities were defined as greater than 1 m/s. Accordingly, low sliding velocities were defined as less than 1 m/s. Applications which would potentially make use of copper fiber brush slid ing electrical contacts typically have sliding velocities in the range of 1 to 100 m/s. Tribological testi ng of copper fiber brushes in humid environments was performed at both high and low s liding velocities to address practical issues as well as investigate more fundamental questions. 31

PAGE 32

Environmental brush-on-rotor tribometer design A second iteration of the customized brush-rotor tribometer was designed and constructed for high sliding velocity testing in controlled environments [54]. Of particular interest were nonoxidizing gas environments with high relative humi dity (greater than 90% RH). To maintain a controlled test environm ent, the tribometer was designed to fit inside of a large glove box (Vacuum Atmospheres Company; Hawthorne, CA). The glove box was equipped with automated pressure control system and vacuum feedthroughs which fac ilitated exchange of samples without compromising the test environment. A line drawing of the key tribometer compone nts is shown in Figure 2-6. The base and frame components have been removed from the drawing for clarity. The rotor was machined from a copper disk (UNS C11000) to final dimensions of 152.4 mm diameter and 25.4 mm thickness. The copper rotor was mounted onto a cartridge spindle w ith collet (SKF, model 2750C) and driven by a timing belt-and-pulley syst em using a stepper motor (Parker Hannifin, model ZETA83-135-MO), allowing for surface slidi ng speeds up to 10 m/s. An encoder (BEI, model XH535F) mounted on the spindle shaft was used to couple data with rotor angular position. Filamentous copper brushes, provided by SSI Technologies, were used for all environmental studies. The fiber bundle meas ured approximately 10 mm in length, 5 mm in width, and 10 mm in height. The diameter of individual fibers was between 60 and 70 m, and the fiber packing fraction was approximately 0.5. The copper fiber brushes used in tribological testing were the same as that shown in Figure 2-4 with the addition of a flame-sprayed zinc coating on the exterior of the copper mesh wr apped around the fiber bundle. Further details on the copper fiber brushes can be found in the Results (Chapter 3) section t itled Characterization of Copper Fiber Brushes. 32

PAGE 33

As in the previous version of the brush-ro tor tribometer, the environmental brush-rotor tribometer was a two brush system with both brus hes sliding on the radial surface of the rotor. The brushes were offset so that each brush had an independent track in which to slide. The brushes were mounted to flexible copper foil cantil evers (approximately 0.9 mm thick). Details of the brush holder assembly and components can be seen in the li ne drawing in Figure 2-7. A spring-loaded differential variable reluctance tr ansducer (DVRT; Microstrain, Inc.; Williston, VT) was mounted to a rigid support behind each brush. The spherical-tipped plunger of the DVRT rested against the back of the brush and measured linear brush displacement. Each DVRT had a stroke length of 8 mm and a resolu tion of 2.0 m. A micr ometer-driven linear stage (Parker Hannifin Corporati on) was used to position the br ush holder assembly, bring the brush into contact with the rotor su rface, and apply a normal force. The combination of the copper foil cantil ever deflection and DVRT spring compression provided the normal load for the brush. Calibration pr ior to testing verified that the normal load varied linearly over the range of loads investigat ed. The stiffness of the brush assemblies were 0.759 N/mm for the left brush assembly and 0.707 N/ mm for the right brush assembly. Periodic adjustments to the brush load were made to co rrect for brush wear. No rmal loads and friction loads were measured using single-axis fle xure load cells (Strain Measurement Devices; Wallingford, CT) in the configur ation shown in Figure 2-7. Rigi d PEEK brackets insulated the load cells from the current path. The load cells had a range of 0 to 10 N with a resolution of 4 mN. The exposed surfaces of the load cells were coated with a water-proof silicone sealant to prevent moisture absorpti on during prolonged exposure to humid environments. Ambient temperature and humidity inside the glove box were measured with a thermohygrometer (model RH411; Omega; Stamford, CT). A recirculating chil ler bath (ThermoFlex 33

PAGE 34

1400; Thermo Scientific) with a te mperature stability of .1 C was used to actively cool the copper rotor. Cooling water was supplied to th e rotating copper disk through a rotary union attachment (Duff-Norton; Charlotte, NC) on the face of the copper disk. Water traveled from the recirculating chiller thro ugh the rotary union inlet to a cavity inside of the copper rotor and then back out through the rotary union outlet to the recirculating ch iller. A second recirculating chiller supplied temperature controlled water to a heat exchanger in side the glove box which regulated the ambient temperature. A fan was pos itioned so that it circulated gas over the heat exchanger to maintain consistent temperature throughout the envir onment chamber. An adhesive K-type thermocouple was attached to each copper foil adjacent to the brush base plate to provide a measure of the brush temperature. The glove box was backfilled with CO2 gas, and a slight positive pressure was maintained. An oxygen sens or (DF-310 E; Delta F Corp.; Woburn, MA) was used to monitor oxygen levels inside the glove box. After pur ging the test environment, the oxygen concentration was below 100 ppm for all tests. Beakers of distilled water were placed inside the glove box to saturate the environment with water vapor. A schematic drawing of the electrical circuit is shown in Figure 2-8. Current was provided by a 0 A, 0 V DC power supply. The copper cantilevers were used as conductive pathways to transmit current to the brushes. The voltage drop across each brush contact was measured independently. Voltage tap wires were attached at the base of both brushes using ring terminals and the brush mounting screws. A th ird dead-loaded filamentous copper brush was installed as a voltage tap on the rotor and was positio ned on the back face of the rotor so as not to interfere with the other two brushes. As show n in Figure 2-8, there were a total of three independent voltage measurem ents in the circuit. 34

PAGE 35

A full-scale high current density motor typically has a large number of electric brushes operating in parallel. The tribom eter described here was limited to a two brush system (with a third brush acting as a voltage pickup). This design presented difficulties in terms of maintaining constant current flow. The dynamics of the sy stem can cause the brushes to bounce, thereby breaking the electrical circuit a nd interrupting current flow. To simulate multiple parallel current paths in the two brush system, a resistor wa s constructed from co iled copper tubing and connected in parallel with the brush circuit. The resistance of the parallel resistor was approximately half of the total resistance across the brush-rotor contacts ; this ensured that a majority of the current (approximately two-th irds) from the power supply passed through the parallel resistor while the remaining current (approximately one-third) passed through the brushes and rotor. Two Hall Effect sensors (Amp Loc; Goleta, CA), each rated for 300 A, were used to determine the current split between the pa rallel resistor path and the brush-rotor path. The current flow through the brushes ( iB) was measured with one Ha ll effect sensor, while the total current output from the power supply ( i ) was measured by the second sensor. The current through the parallel path ( iP), which was not a measured value, was calculated by subtracting the brush current from the total current. The voltage drop across each brush-rotor cont act was calculated by taking the difference between the measured brush voltage and ro tor voltage. According to Figure 2-8, V1 is the voltage measured at the positive brush, V2 is the voltage measured at the rotor, and V3 is the voltage measured at the negative brush. The voltage drop across the positive brush interface ( VPB) is simply the difference between V1 and V2, while the voltage drop across the negative brush interface (VNB) is the difference between V2 and V3. 35

PAGE 36

The bulk resistance of the copper rotor and the br ushes is negligible when compared to the contact resistance of the brush-ro tor interface. Using Ohms Law, the contact resistances for the positive brush (RPB) and negative brush ( RNB) were calculated according to Equations 2-1 and 22. 12 PB PB BBVVV R ii (2-1) 2 NB NB BBVV V R ii 3 (2-2) A scanning white light interferometer (SWL I; Ambios Technology, model Xi-100) with a 10X Mirau objective was mounted di rectly above the rotor, as s hown in Figure 2-6. A two-axis linear stage (Parker Hannifin, model 4400) was used to position the interferometer above the two wear tracks. The rotor was stopped at various points throughout testing, and the SWLI was used to measure the topography of the rotor surface. The scan area was 504 m by 504 m. Scan locations were selected inside both of the roto r wear tracks and compared to a location away from the wear tracks (representative of the original rotor surface). To operate the SWLI in a high humidity environment, the body of the inte rferometer was back-filled with dry CO2 to provide a positive pressure and prevent moisture accumulation inside of the instrument. Additionally, an adhesive Kapton heater (Omega, model KH) was wrapped around the objective to heat it to a temperature above the ambient temperature and pr event water from condensing on the lens. The entire glove box rested on pneumatic vibration isol ators. Multiple scans of each wear track were acquired and averaged together. Rotor subcooling Independent ambient and rotor temperature control made it possible to regulate local relative humidity at the s liding contacts. For this study, sust aining a rotor temperature below the 36

PAGE 37

dew point temperature was referred to as roto r subcooling. Because the operating environment was approximately at 100% RH, the ambient temperature was assumed to be the dew point temperature. The amount of subcooling was quan tified as the difference in temperature between the brush/rotor contact and the ambient (glove box) environment temperature. To enhance condensation at the contacting interface during tribological testing, the rotor temperature was maintained approximately 2C below the ambient temperature. The derivation of the relationship between temp erature and the saturated vapor pressure of water was described by Gaskell [55]. The Clausi us-Clapeyron equation, as shown in Equation 23, applies to equilibrium between the vapor phase and a condensed phase. 2lnH dPdT RT (2-3) The ideal gas constant, R is 8.3144 Jmol-1K-1, while P is the vapor pressure, T is the temperature, and H is the change in molar enthalpy. The change in molar heat capacity was calculated for the change of state from liquid to vapor. Values for the molar heat capacities (units of J/K) of water in the vapor ( Cp,H2O(v)) and liquid ( Cp,H2O(l)) states were taken from Gaskell [55]. 2 235 ,() ,()3010.71100.3310 75.44pHOv pHOlcT c 2T2T Thus, the change in molar heat capacity for the transition from liquid to vapor is the difference between Cp,H2O(v) and Cp,H2O(l). 35 ()45.4410.71100.3310plvcT Given that the change in enthalpy for evaporation ( Hevap) is 41,090 J at the boiling temperature of 373 K, the change in enthalpy for evaporati on at any temperature is given by the following expression. 37

PAGE 38

,, 3 7 3( 373 T evapTevap plvHHc)dT2T d TT 35 37341,09045.4410.71100.3310T evapTHT 32 51 ,57,38345.445.35100.3310evapTHTT Substituting values into Equation 2.3 and integrating yields the following expressions for the saturated vapor pressure of wate r as a function of temperature. 2 235 23 35 1 257,38345.445.35100.3310 ln 57,38345.44ln5.35100.1610 lnsatHO satHOP RTRTRRT TT PC RTRRRT The constant of integration, C1, can be evaluated at the boilin g point, where both the temperature (T = 373 K) and saturated vapor pressure (P = 101,325 Pa) are known. The result is a value of 62.14 Pa for the constant, C1. Thus, the final equation relating the saturated vapor pressure of water (in Pa) to temperatur e is given by Equation 2-4. 235 257,38345.44ln5.35100.1610 ln 62.14satHOTT P RTRRRT (2-4) Relative humidity is the ratio of the vapor pres sure of water to the sa turated vapor pressure at a given temperature, expressed as a percenta ge. As can be seen in Figure 2-9, relative humidity has a strong dependence on temperature. Assuming the operating environment is at a temperature of 20C and a constant water part ial pressure of 2,000 Pa, the relative humidity would be approximately 85%. If the temperatur e of the operating environment increased to 30C with the partial pressure of water remaining cons tant, the relative humidity would decrease to a value of 47%. 38

PAGE 39

Relative humidity becomes increasingly important when examining the local environment. As the temperatures of the sliding surfaces incr ease due to frictional he ating and Ohmic heating, the gas temperature near the sliding surface will increase accordingly. Because the partial pressure of water remains constant, the relati ve humidity must decrease in response to the temperature rise. The amount of water adsorbed on to the sliding surfaces is directly dependent on the local relative humidity. Therefore, to maximize water adsorption, the rotor temperature was maintained several degrees below the ambient temperature. Although the brushes were not actively cooled, the rotor cooling system successf ully transferred heat away from the brush contacts to maintain brush temperat ures near the ambient temperature. High sliding velocity test conditions For the brush-rotor tribometer, the brush norma l load was maintained at approximately 0.7 N (1.4 104 Pa). Brush normal load was periodically adjusted to correct for decreases due to brush wear. The sliding velocity was consta nt at 5 m/s, and the current density was approximately 180 A/cm2. Prior to testing, the gl ove box was purged with dry CO2 gas. During testing, the relative humidity was greater than 95% and the oxygen concentration was below 100 ppm. The ambient temperature inside th e glove box was approximately 26C. Rotor temperature was approximately 24C (2C subcool). Brushes were examined prior to and after testing using scanning electron microscopy and scanning white light interferometry. Polarity effects on arcing in copper fiber brush contacts The effects of arc discharges on material tran sfer in the copper fiber brush system were investigated for the positive brush and negative brush polarities. The PEEK flexure shown in Figure 2-5 was used to study copper fiber brushes in a stationary switching contact in ambient air environments. Current was sourced through the contact using a 0 A, 0 V DC Instek power supply (Fotronic Corporat ion; Melrose, MA). The flexure was loaded and unloaded 39

PAGE 40

repeatedly to create a switching contact. Light emitted from arcing events at the contact was collected using a fiber optic positioned approximately 1 cm from the interface. The emitted light was analyzed using a TriVista triple spectrome ter from Princeton Instruments. Voltage signal was recorded from a photomultiplier tube (P MT) tuned to the 521.8 nm Cu(I) emission line. Positive and negative brush polarities were investigated for the same brush contact. The environmental brush-rotor tribometer s hown in Figure 2-6 was used to study arcing under copper fiber brush contacts while sliding in a controlled environment. Testing was performed in both ambient air and humid CO2 environments at a sliding velocity of 2.5 m/s. Current was sourced through the brush-rotor contact using a 0 A, 0 V DC power supply. The electrical circuit for the system remained th e same as that shown in Figure 2-8. The brush load was reduced to nearly zero to intentiona lly induce visible arcing events at the sliding interface. A fiber optic was positioned parallel to the sliding contact to collect light emitted from the arc discharges. The collected light was analyzed using the TriVista triple spectrometer from Princeton Instruments. Current flow direction was reversed several times to examine polarity effects on arcing at the same sliding interface. Reciprocating Low Sliding Velocity Testing with Copper Fiber Brushes Tribological studies of electric brush materials are typically pe rformed at sliding velocities ranging from 1 to 100 m/s [2, 8, 46, 48] because these velocities are commonly encountered in the intended applications. Measurement difficulties arise in high sliding velocity tests from system dynamics. Vibrations can cause the brushes to skip over the rotor surface, which influences measurements of fric tion and contact voltage drop. Moreover, vibrations can induce additional material removal mechanisms. If, wh en the brush breaks contact, the voltage drop across the newly formed gap between the brush a nd rotor is large enough, an arc will form. The 40

PAGE 41

use of low sliding velocities was intended to simplify the study of slidin g electrical contacts by removing unwanted dynamic effects that influence critical measurements. Reciprocating tribometer design The tribological behavior of copper sliding electrical contacts was investigated at low sliding velocities using a linear reciprocating tribometer inside of an acrylic environmental enclosure [56]. A line drawing of the linear reci procating tribometer is shown in Figure 2-10. The tribometer was integrated with a scanni ng white light interferometer (Zygo NewView 5032) for in situ measurements of counterface topography and wear. The interfer ometer had a vertical scan range of up to 5 mm and had a height resolution of 0.1 nm. The copper fiber brush was made by rolling a co pper wire mesh into a cylindrical bundle and soldering the bundle at one end. The diamet er of the brush was approximately 3 mm, and the diameter of an individual fiber was between 60 and 70 m. The brush was held in place by an aluminum cantilever brush holder, as shown in Figure 2-11. The aluminum brush holder was mounted to a PEEK backing plate, which was attach ed to a six-channel multi-axis load cell (JR3, Inc.; 100 N maximum load) for measuring normal and frictional forces. Normal load was applied through elastic deflection of the double leaf spring flexure; load adjustment was controlled manually through a micrometer-driven vertical stage. The copper counterface was prepared using we t-grinding with progres sively finer silicon carbide papers, starting with 600-grit and ending with 1200-grit. Final polishing was performed with alumina particle slurries of 0.9, 0.3, and 0.05 m particle sizes on synthetic velvet cloth. The final average surface roughness of the po lished counterface was less than 50 nm. The polished counterface was mounted to a copper b acking plate which was bolted to the copper cooling stage. A temperature controlled wa ter circulator (ThermoFlex 1400 from Thermo Scientific) was used to supply cooling water to the cooling stage and regulate the sample 41

PAGE 42

temperature. An adhesive K-type thermocouple was affixed to the copper backing plate to measure the sample temperature. The cooling stage was attached to the linear stage, and the entire stage assembly was driven by a steppe r motor with a speed range of 0-140 mm/s and positional accuracy of 1.5 m. A position encoder was used to measure the position of the linear stage. Because of the fine posit ional resolution of the stepper motor, the same location ( 2 m) within the wear track was analyzed repeatedly with the interferometer. A 0 A, 0 V DC Instek power supply (Fotronic Corporation; Melrose, MA) was used to source current through the brush-counterface contact. A current carrying wire from the power supply was soldered to the back of the c opper fiber brush, and a se cond current wire from the power supply was connected to the copper backing plate of the counterface using a ring terminal (as shown in Figure 2-11). The voltage drop across the contact was measured using the data acquisition hardware and recorded in the cust omized software. Wires were soldered to the copper fiber brush and attached to the copper b acking plate of the counterface to make the voltage measurement. Circulated cooling water in the copper cooling stage was used to offset the resistive heating from the current flow throu gh the circuit. The load cell was isolated from the electrical circuit by the PEEK plat e behind the aluminum brush holder. The tribometer was enclosed in an acrylic environment chamber (not shown in Figure 210) to provide control over the operating enviro nment. The chamber had dimensions of 216 mm (width) by 279 mm (length) by 114 mm (height). Testing was primarily conducted in humid carbon dioxide environments. Dry carbon dioxide gas was bubbled through a beaker of distilled water before entering the chamber. A resistive heater was wrapped around the beaker to increase the temperature of the water bath and increase the water content of the gas stream. The temperature of the water beaker was regulated using a K-type thermocouple and an Omega 42

PAGE 43

temperature controller. The chamber was back -filled with humid car bon dioxide gas until the oxygen concentration was below 100 ppm and the relative humidity was above 95%. Relative humidity and ambient temperature inside th e chamber were measured using a thermohygrometer (Omega RH411). Oxygen concentration in the operating environment was measured with an oxygen sensor from De lta F Corporation (DF-310 E). A flexible Kapton heater was wrapped around the body of the load cell, and the load cell was heated to approximately 10C above the ambient temperature to prevent water from condensing inside the load cell. Similarly, a flexible Kapton heater was wrapped around the objective of the interferometer to eliminate conde nsation on the lens. The chamber was outfitted with numerous feedthroughs for the gas supply, cooling water, sensors, and data acquisition. A hole in the top of the chamber allowed the interf erometer objective to be lowered into position to scan the sample surface. To minimize envi ronmental contamination through this opening, a piece of latex was stretched ar ound the objective and attached to the top of the chamber. After loading the brush and counterface samp les on to the tribometer, the stage was positioned in line with the interferometer objectiv e to scan the surface prior to testing. The counterface was then positioned under the brush, and the brush was lowered into position with an applied normal load of 1.0 N (nominal pressure of 1.4 105 Pa). The counterface was subcooled, unless specified otherwise, by approxima tely 2C. Before sliding was initiated, the power supply was turned on and the desired current was sourced through the contact. The track length was 5 mm for all studies. A schematic of the counterface wear track is shown in Figure 212. The area of the wear track analyzed by non-c ontacting profilometry is shaded in Figure 2-12. The velocity profile for a single reciprocating cycle is shown above the wear track. Short periods of acceleration and deceler ation occur at the ends of the wear track while most of the 43

PAGE 44

sliding occurs at a consta nt velocity. For all data recorded fr om this series of tests, only data acquired during periods of constant velocity sliding was analyzed (periods of acceleration were neglected). The LabView data acquisition software recorded normal load, friction load, and contact voltage as a function of position as well as average values per cycle for all three. Sample temperature, ambient temperature, and relative hu midity were also recorded for each cycle. A sliding velocity of 0.01 m/smore than two orde rs of magnitude lower than the previously described high sliding velocity testswas used for these studies. Two independent tests were conducted for a total of 45,000 cycles, with a new polished counterface used for each test. Testing was pe rformed for the positive counterface (electrons leaving brush surface and impinging on counterface surface) and negative c ounterface (electrons leaving counterface surface and impinging on brush su rface) current flow directions. Each test was divided into four segm ents: 15,000 cycles at 0 A/cm2; 10,000 cycles at 180 A/cm2; 10,000 cycles at 0 A/cm2; and 10,000 cycles at 180 A/cm2. Surface profiles of the copper counterface wear tracks were obtained at each change in current using the scanning white light interferometer. Counterface wear was quantified by subtracting th e wear track surface profiles from the surface profile of the orig inal surface obtained at cycle 0. X-ray photoelectron spectroscopy (XPS) characterization of wear tracks X-ray photoelectron spectroscopy (XPS) is a surface sensitive characterization tool used to identify species and their local chemical envi ronment. XPS measures the kinetic energy of photoelectrons emitted from a sample when subjected to a photon source. Because the mean free path of electrons in a solid is particularly small, the sampling volume only extends to a depth of ~10 nm below the surface. The kinetic energy of each electron is charact eristic of the element from which it was emitted. Based on the composition of the sample, certain energies are much more prevalent and appear as peaks in the spectra. Slight shifts in the peak energies of an 44

PAGE 45

element are indicative of the chemical state of the atom. Thus, bonding information in addition to elemental identification can be obtained from XPS. The setup of the linear reci procating tribometer shown in Figure 2-11 was modified to accommodate samples for XPS analysis [56]. Ra ther than using a copper backing plate, the counterface was mounted directly to a custom desi gned stainless steel XPS platen (Figure 2-13). The XPS platen was clamped to the copper co oling block, and the current lead for the counterface was attached to the copper cooling block. When testing was completed, the counterface/platen assembly was removed from the environment chamber, transported under dry nitrogen, and placed directly in the XPS chambe r. This design minimized handling and exposure of the sample to the outside environment dur ing transportation. Materials exposed to the ambient environment have a thin layer of adventitious carbon deposited on the surface, which produces a characteristic peak at 284.6 eV in XPS analysis [57] XPS analysis was used to correlate the CO2 operating environment with tribofilm ch emistry, so minimization of potential sources of carbon contamination was a priority. Low sliding velocity subcool study The linear reciprocating tribometer (Figure 210 and Figure 2-11) was also used to study the effects of subcooling on friction and contac t resistance for a copper brush on copper flat sliding electrical contact [56]. As describe d previously, the subcool is the difference in temperature between the brush/roto r contact and the testing environm ent. The temperature of the contact was controlled using a variable temperat ure refrigerated circulating bath which supplied water to the cooling stage, and the contact temp erature was measured using an adhesive K-type thermocouple attached to the c opper backing plate. The ambien t temperature and humidity were measured using a thermo-hygrometer (Omega model RH411). The ambient temperature (~26C) and relative humidity (~95%) remained c onstant during testing. The five different 45

PAGE 46

subcoolings, quantified as difference between samp le and environment temperature, investigated were: -4, 0, +4, +8, and +12C. Each subc ool was maintained for 2,000 cycles, and friction coefficient and contact resistance we re recorded for each cycle. Testing was performed in humid carbon di oxide and humid argon environments to examine the role of carbon dioxide in the interfacial chemistr y. The copper fiber brush and copper counterface were positioned as shown in Figure 2-11. A current density of 7 A/cm2 was used to minimize heating of the contact and im prove temperature stability. The polarity was brush positivecurrent flowed from the brush to the counterface. The sliding velocity was 0.01 m/s, and the track length was 5 mm. The br ush normal load was set to 1.0 N (1.4 105 Pa nominal pressure); adjustments to the load we re made to account for thermal expansions and contractions. Pentanol vapor phase lubrication Because high current density sliding electri cal contacts require thin lubricating films, potential lubricants were limited to gas phase additives. Other than water saturated environments, 1-pentanol saturated gas environm ents were also investigated. Monolayers of adsorbed 1-pentanol from the gas phase have been shown to effectiv ely lubricate silicon microelectrical mechanical systems (MEMS) [ 58, 59]; however, there were no previous studies in the literature reviewed that utilized 1-pentanol in the lubrication of metallic sliding electrical contacts. The linear reciprocating tribometer shown in Figure 2-10 was used to study the effectiveness of 1-pentanol as a gas phase lubricant for self-m ated copper sliding electrical contacts. Dry argon was bubbled th rough a beaker of 1-pentanol at ambient temperature. The gas delivery system was modifi ed so that the stream of pe ntanol-saturated argon impinged directly on the copper counterface. The flow rate of pentanol-saturated argon gas was 2.5 L/min. The measured relative humidity inside th e chamber was less than 1%, and the oxygen 46

PAGE 47

concentration was below 50 ppm. The electrical circuit remained the same as previously described. When investigating polarity effects, a current of 12.7 A (180 A/cm2) was used. Recirculating chilled water was supplied to the cooling stage to offset the Ohmic heating and maintain a stable temperature. The test setup was verified usi ng a 6.3 mm diameter quartz (SiO2) sphere sliding against a silicon counterface ([100] orienta tion; 619 m thick) in a pentanol-saturated atmosphere for comparison to other studies from literature. Cond itions were matched to those used by Asay et al. for the same sliding pair [59]. The applie d normal load was 0.39 N (contact pressure of 270 MPa). The sliding velocity was 0.015 m/s, and the track length was 10 mm. The chamber was backfilled with pentanol-saturated argon (dry argon bubbled through 1-pentanol) at a constant flow rate of 2.5 L/min. Friction coefficient was recorded for each cycle. Scanning white light interferometry was used to measure the wear tr ack topography intermittently throughout the test. For the copper-on-copper sliding pair in the 1-pentanol envi ronment, a copper pin with a radius of curvature of approxima tely 4 mm was used along with a polished copper counterface. Testing was performed under the same operating conditions (pressure of 270 MPa, sliding speed of 0.015 m/s, no current) as the quartz on silicon e xperiment. Evolution of the wear track on the copper counterface was recorded by sca nning white light interferometry. A 3 mm diameter copper fiber br ush was used in comparing the pentanol-saturated argon environment to the water-saturated carbon dioxi de environment. The current density was 180 A/cm2, and the brush was positively biased (current flow from brush to c ounterface). Friction coefficient and contact resistance were recorded for each cycle. Wear track topography was not measured in this experiment. 47

PAGE 48

Single copper fibers in humid carbon dioxide A shortcoming of tribological testing with multi-filament metal brushes is the ex situ analysis of worn brushes and worn brush fibers The conditions encountered by the entire brush are well documented, but each fiber may experien ce very different conditions from neighboring fibers. There is no way of knowing how long an individual fiber within a brush remained in contact with the rotor or the loading conditions it experienced. Current tends to flow preferentially through fibers located along the peri meter of the brush [60], so the current density through an individual fiber can vary greatly from that of surroundi ng fibers. Assumptions can be made to estimate average values, but little is known about the history of the fibers that are being analyzed. To overcome this problem, a set of experiment s were conducted to look at a single copper filament in unidirectional sliding against a c opper disk while passing current in a controlled environment. This ensured that the fiber bei ng characterized had a known history (load, sliding distance, and current). Figure 2-14 shows a photograph of the te sting apparatus for the single copper fiber experiments. A 70 m diameter copper wire was attached to a glass cantilever. Capacitance probes measured the deflections of the cantilever. Based on the stiffness of the cantilever in the vertical and horizontal directi ons, normal and frictional forces were calculated from the displacements. A current of 40 mA (~1000 A/cm2) was sourced through the cont act during sliding for both the positive brush and negative brush current flow directions. The nominal current of 40 mA was selected based on the estimated current fl ow through a single fiber of a complete brush subjected to 180 A/cm2, assuming only one-third of the fibe rs are in contact and capable of conducting current. A second copper fiber brush, which acted as the pick-up brush (Figure 214), was loaded against the radial surface of the polished copper disk to complete the electrical 48

PAGE 49

circuit. A single fiber was also tested for the same sliding distance under no current. The normal load applied to the fiber was 500 N (nominal pressure of 1.3 105 Pa); the single fiber load was estimated from nominal pressures used with multi-filament brushes assuming one-third of the fibers in contact. The copper disk was polished to a final surface roughness of less than 70 nm, and the sliding velocity of the di sk was constant at 0.16 m/s. The tribometer was enclosed within an environment chamber which was backfilled with water saturated CO2 (98% RH). At several points during test ing, sliding was paused and a current-voltage (I-V) sweep of the contact was performed. The range of curre nt for the I-V sweep was 0 to 40 mA. After completion of wear testing, fibers were removed from the environment chamber and stored in a desiccator. The sliding surf ace of each fiber was carbon coated to prevent oxidation or contamination. Worn fibers were then charac terized using the dual-b eam focused ion beam scanning electron microscope (FIB/SEM). A platinum layer was deposited on top of the carbon layer to protect surface features from ion damage. Milling with the FIB was used to create cross sectional views of the fiber surface parallel to the direction of sliding. Further milling was performed to create electron transparent sa mples suitable for examination by transmission electron microscopy (TEM) and energy di spersive X-ray spectroscopy (EDS). 49

PAGE 50

Figure 2-1. Line drawing of brus h-rotor tribometer for testing of solid lubricants with sliding electrical contacts in ambient air [53]. 50

PAGE 51

Figure 2-2. Exploded view of brush thruster and load cell assembly line drawing with key components labeled [53]. 51

PAGE 52

Figure 2-3. Schematic drawing of the electrical circuit for the brush-rotor tribometer. Current, supplied by the current source, travels from the positive brush through the copper rotor to the negative brush. Conversely, the direction of electron flow is from the negative brush to the positive brush. The br ushes are electrically insulated from the load cells and LVDTs [53]. Figure 2-4. Photograph of as-r eceived copper fiber brush (no flame-sprayed zinc coating). 52

PAGE 53

Figure 2-5. Line drawing of te sting apparatus for copper fiber brush stiffness measurements. The plunger from the LVDT mounts to the backside of the flexure to measure deflections under dead-weight loading. Fle xure was also used to study arcing under a stationary copper fi ber brush contact. 53

PAGE 54

Figure 2-6. Line drawing of the modified brush-rotor tribometer for environmental testing. Base and frame have been removed for clarity [54]. 54

PAGE 55

Figure 2-7. Line drawing of br ush holder assembly from brush-rotor environmental tribometer [54]. 55

PAGE 56

Figure 2-8. Schematic diagram of electrical circ uit for brush-rotor environmental tribometer [54]. 56

PAGE 57

Figure 2-9. Plot of isotherms fr om 0 to 100C for relative humid ity versus partial pressure of H2O. Curves were generated using the Clausius-Clapeyron e quation (Equation 2-3). Figure 2-10. Line drawing of linear reciprocating tribometer with in situ profilometry capabilities [56]. 57

PAGE 58

Figure 2-11. Line drawing of brush holder assembly and stage a ssembly for linear reciprocating tribometer and positioning of scanning white light interferometer (SWLI) objective [56]. 58

PAGE 59

Figure 2-12. Schematic drawing of the wear track created on the counterface by a 3 mm diameter brush with a 5 mm track length. The profilometry scan area, which was approximately in the middle of the wear track, is shown by the shaded box. The velocity profile for a single cycle is displa yed above the wear track. Short periods of acceleration and decelerati on occur at the ends of the wear track [56]. 59

PAGE 60

Figure 2-13. Modified setup of linear reciprocat ing tribometer to accommodate samples for XPS analysis. Figure 2-14. Photograph of the testing apparatus for single fiber experiments. A single copper fiber was loaded against a rotating copper disk. Testing was performed in a humid CO2 environment. A four-wire circuit was us ed to supply current to the contact and measure the voltage drop acr oss the sliding interface. 60

PAGE 61

CHAPTER 3 EXPERIMENTAL RESULTS Decoupled Graphite Lubrication of Slid ing Electrical Contacts in Ambient Air Decoupled graphite lubrication was studied as a means of reducing friction and wear of monolithic silver brush sliding electrical cont acts in ambient air environments. Results are shown in Figure 3-1 for a variable current test at a sliding velo city of 1.6 m/s [53]. Average friction coefficient for the positive brush was 0.16 while the average friction coefficient for the negative brush was 0.21. The observed trend of hi gher friction for the negative brush could not be confirmed. Repeated testing of the graphite lubric ated silver brush sy stem showed friction coefficient to have no dependence on brush pol arity. Large fluctuations in the friction coefficients tended to coincide with replacement of the graphite lubricant pins. No definitive correlation between friction coefficien t and current density was observed. Because of the frequent changes in current de nsity, the brush wear data of Figure 3-1 was temperature compensated for thermal expansions in the system. At a current density below 150 A/cm2, brush wear was so low that wear rates c ould not be accurately quantified based on the resolution of the displacement trans ducers. Sliding distances at 150 A/cm2 were not great enough to reliably calculate a wear rate. At 200 A/cm2, brush wear increased rapidly for both the positive and negative brushes. The steady-st ate positive brush wear rate was 1 10-9 m/m, and the steady-state negative br ush wear rate was 3 10-10 m/m. The wear behavior was shown to be reversible, as decreases in current dens ity caused the wear rate to decrease. Visual inspection of the worn brush and ro tor surfaces after tes ting confirmed that a graphite transfer layer had formed on the brushe s as well as the rotor su rface. Worn silver brushes were characterized by energy dispersive X-ray spectrosc opy (EDS), and spectra from the positive and negative brush surfaces are shown in Figure 3-2 [53]. The composition of the 61

PAGE 62

positive brush surface (Figure 3-2 A) was prim arily silver and carbon with trace amounts of oxygen. The carbon signal originated from the tran sferred graphite layer on the brush surface. Oxidized wear debris likely contributed to the oxygen signal. The negative brush spectrum (Figure 3-2 B) revealed a distin ctly different composition for the negative brush surface. In addition to silver and carbon, significant amounts of copper were detected on the negative brush surface. Trace amounts of oxygen were observed in the negative brush spectrum as well. Wear rate measurements for low current densities were difficult to obtain from the variable current test (Figure 3-1) due to the low sliding distances. An additional test was performed using decoupled graphite lubrication of monolithic silver alloy (92.5 wt% Ag, 7.5 wt% Cu) brushes at 40 A/cm2 [53]. After run-in of the brushes under no current, the steady-state positive and negative brush wear rates (cal culated over approximately 900 km of sliding) were 2 10-11 and 6 10-12 m/m, respectively. Friction coefficients fo r both brushes were initially high during run-in but gradually stabilized at 0.15 for the majority of the test. The combined contact resistance for both brush contacts during sliding ranged from 100 to 150 m at 40 A/cm2. For higher current densities (greater than 150 A/cm2), contact resistances were generally in the range of 30 to 50 m during sliding. At the end of the 40 A/cm2 test, contact resistance measurements were made with a stationary rotor (no sliding), and the co mbined contact resistance for both brushes was 32 m The graphite transfer film thickness on the rotor surface was examin ed by scanning white light interferometry. A surface profile of the edge of the ne gative rotor track (under the positive brush) is shown in Figure 3-3. The edge of the graphite transfer film is oriented horizontally through the middle of the image. Although the transfer film visually appeared to be uniform, the thickness of the transfer film based on profilometry measurements varied from approximately 1 62

PAGE 63

to 5 m across the rotor surface. Graphite film thicknesses on the positive and negative rotor track surfaces were comparable. Polytetrafluoroethylene (PTFE) Solid Lubric ants for Sliding Electrical Contacts in Ambient Air Decoupled lubrication of copper fiber br ush sliding electrical contacts with polytetrafluoroethylene (PTFE) and a PTFE-indium composite (PTF E/In) was studied in ambient air environments using the brush-rotor tribom eter shown in Figure 2-1. Results of the tribological study are summarized in Table 3-1. With PTFE as the lubricant, the positive brush (2 10-8 m/m) had a higher wear rate than the negative brush (1 10-8 m/m). The positive brush lubricant pin also wore at a higher rate than the negative brush lubricant pin. Brush wear rates and lubricant wear rate s were lower with the PTFE/In composite compared to PTFE. With PTFE/In, the positive brush wear rate (7 10-9 m/m) was over an order of magnitude higher than th e negative brush wear rate (6 10-10 m/m). Wear of the PTFE/In lubricant pins did not va ry significantly with brush polar ity. The addition of indium to PTFE did not significantly increase coefficient of friction. Moreover, the addition of indium did not appreciably decrease contact resistance m with indium compared to 16 m without indium. Characterization of Copper Fiber Brushes Fiber Brush Packing Fraction and Fiber Hardness A photograph of an as-received filamentous c opper brush is shown in Figure 2-4. The packing fraction of the brush fibers was calcu lated using scanning electron microscopy (SEM) and image analysis software, as shown in Figure 3-4. Packing fraction was defined as the ratio of total fiber surface area to the total brush surface area. Image analysis software was used to identify the fiber tips based on th e contrast of the SEM microgra phs. As can be seen in the 63

PAGE 64

image on the right of Figure 3-4, the image anal ysis method overestimated the surface area of some fibers while underestimating for others. Th erefore, manual counting of fibers was used to verify the results from the image analysis softwa re. The average fiber diameter was calculated to be approximately 70 m. The packing fraction was estimated by dividing the total fiber area (total number of fibers multiplied by the average area per fiber) by the area of the field of view in the micrograph. Estimates based on manual fibe r counting supported the results obtained by image analysis software. All brushes had a p acking fraction between 0.43 and 0.62; in general, the packing fraction of most brushes was 0.50 0.05. These packing fractions were high compared to the recommended packing fracti ons, generally around 0.2, for high current metal fiber brushes [12]. Hardness measurements were made on a total of 5 as-received copper fibers. Prior to tribological te sting, fibers were transversely sectioned, mounted, and polished. A Vickers diamond tip indenter and an applied lo ad of 98 mN were used. The average fiber hardness was calculated to be 820 80 MPa. Metal Fiber Brush Contacts Kuhlmann-Wilsdorf [12] derived a set of equations to descri be the contact behavior of metal fiber brushes. The transition pressure ( ptrans) between elastic and plastic contact spots for a metal fiber brush is a function of the fiber packing fraction ( f ) and the fiber hardness ( H ) and is given by Equation 3-1. 4310transp f H (3-1) Substituting measured values for packing fraction and hardness of the as-received copper fiber brushes resulted in a transition pressure of 0.12 MPa. Kuhlmann-Wilsdorf [12] also defined a constant, as the ratio of the macr oscopic brush pressure ( pB) to the transition pressure (ptrans) according to Equation 3-2: 64

PAGE 65

B transp p (3-2) For optimal metal fiber brush performance, a value of less than or equal to 0.5 is recommended [12]. For testing with solid lubricants, the value of was approximately 0.8 because high brush pressures were needed to reduce electrical losses and Ohmic heating. For tests involving vapor phase lubricati on, the brush pressures and resulting values were influenced by the limitations of the tribometers. For a copper fiber brush with a packing fracti on of 0.5 and fiber diameter of 70 m under a nominal pressure of 1.4 104 Pa (see high sliding velocity test conditions in humid CO2), the load on an individual fiber was calc ulated to be approximately 0.1 mN If it is assumed that only one-third of all fibers are in contact at any time, the load per fiber becomes 0.3 mN. Assuming a plastic contact, the contact area under an indivi dual fiber was estimated by dividing the fiber normal load by the fiber hardness (820 MPa), th e result of which was a contact area of 0.4 m2. Thus, for a circular contact spot the contact spot radius was es timated to be 350 nm under the given conditions for copper fiber brush sliding electrical contacts. Fiber Brush Stiffness The stiffness of the copper brush fiber bundle was measured using the polyetheretherketone (PEEK) flexure shown in Fi gure 2-5. The brush wa s brought into contact with a copper counterface, and then dead wei ght loads were applied to the flexure while deflections were measured using a linear va riable differential transformer (LVDT). The resulting loading and unloading curves for the co pper brush are shown in Figure 3-5. From the loading curve, the brush stiffn ess was calculated to be 150 N/mm; from the unloading curve, the brush stiffness was calculated to be 190 N/mm For comparison, the stiffness of the PEEK cantilever when the brush was not in contact was 2 N/mm. 65

PAGE 66

High Sliding Velocity Studies of Copper Fiber Brush Sliding Electrical Contacts Copper Fiber Brush Sliding Electrical Contac ts in Humid Carbon Dioxide Operating Environments Shown in Figure 3-6 are plots of brush wear and contact voltage drop versus sliding distance for high sliding speed (5 m/s) testing of copper fiber brushes on the environmental brush-rotor tribometer in a humid CO2 operating environment. Current (180 A/cm2) was introduced to the system after 250 km of sliding. The positive brus h wear rate was initially 2 10-10 m/m after the introduction of current. B ecause brush wear was quantified by measuring linear displacement, brush wear da ta was influenced by thermal expansions. This was evident in the negative brush wear data, partic ularly in the initial 500 km of sliding. Negative deviations in brush wear represented an increase in brush length likely resulting from thermal expansions. Contributions from Ohmic heating of the brush may have accounted for this effect. The brush wear data had a periodic nature due to temperature fluctuations wh ich appeared to be influencing displacements and subsequently brush wear measurements. The periodicity of the fluctuations was approximately 430 km, which, at a sliding velo city of 5 m/s, corresponded to a sliding time of approximately 24 hours. Rotor temperatur e measurements also showed temperature fluctuations which had a period icity of 24 hours. Using brush wear measurements taken at 24 hour intervals, the positive brush wear rate was calculated to be 3 10-11 m/m, and the negative brush wear rate was ca lculated to be 4 10-12 m/m. From Figure 3-6 B, the contact voltage drop fo r the negative brush was consistently higher than the positive brush voltage drop. Contact volta ge was plotted rather than contact resistance because of the magnitude of the voltage drops The softening voltage for copper electrical contacts is 120 mV [61]. Around 1900 km of sliding, the negative brush voltage drop approached the softening volta ge and maintained a value ne ar 120 mV for several hundred 66

PAGE 67

kilometers of sliding. Around 3000 km of slidi ng the negative brush contact voltage decreased while the positive brush contact voltage in creased. Data was not plotted beyond 3300 km because visible arcing was observed at the positive brush contact. The average contact voltages for the positive and negative brushes over the en tire test were 34 mV and 95 mV, respectively. Average contact resistance for the positive brush was 0.4 m and average contact resistance for the negative brush was 1.1 m SEM micrographs of the worn copper fiber brus hes are shown in Figure 3-7. The fiber tips of brushes from both polarities experienced signifi cant plastic deformation, so much so that the brush surface upon visual inspection appeared solid. Many of the fibers were no longer independent. Neighboring fibers were joined togeth er at the surface, and debris filled in the gaps between fibers. For this particular test, the fi ber surfaces of the differe nt polarities had very distinct appearances. Worn fibers from the negative brush had smooth, featureless surfaces, as seen in Figure 3-7 A. Worn fibers from the positive brush had much rougher surfaces (Figure 37 B), and it was difficult to dist inguish individual fibers becaus e the fibers had become welded together at the surface. The difference in su rface roughness of the positive and negative fibers was confirmed by scanning white light interferomet ry. Surface profiles of a negative brush fiber and positive brush fiber are shown in Figure 3-8. Pit-like features were observed on the surface of the positive brush fiber. The surface depressions had a depth of 1 to 2 m. The negative brush fiber surface was free of any pit-like features. The images of Figure 3-7 and Figure 3-8 were obtained after stoppage of the test due to visible arcing at the positive brush interface. The morphology of the positive brush fiber surfaces shown in Figure 3-7 B was not typical of most positive brushes, only those brushes where arcing was visible at the interface. For instances when arcing was not observed, the 67

PAGE 68

positive brush surface appeared similar to the ne gative brush surface depicted in Figure 3-7 A and Figure 3-8 A. In most cases, the amount of debris accumulation around the positive brush fibers was greater than that around negative brush fibers, likely due to the higher wear rates of the positive brushes. Polarity Effects on Arcing in Copper Fiber Brush Contacts Arcing spectra were collected from a stationary flexure (Figure 2-5) outfitted with a copper fiber brush and copper plate. The electrical contact was intermittently broken, creating visible arcing events. All spectra from th e stationary flexure were recorded in ambient air environments for both the positive and negative brush polaritie s. Figure 3-9 shows arc lifetimes for the positive brush and negative brush polarities (in ambient air) which were representative of all arcing events at each polarity. The arc lifetim es for positive and nega tive polarities were not statistically different. The combined averag e lifetime for both polarities was 5.7 s, and the range of all recorded arc lifetimes was approximately 1 to 16 s. The distribution of arc lifetimes showed no dependence on polarity for th e stationary switching c ontact in open air. Using the environmental brush-rotor tribom eter, induced arcing events under sliding copper fiber brush contacts were st udied in ambient air and humid CO2 environments. The atomic emission spectra shown in Figure 3-10 were collected in an ambient air environment. The brushes used in this study had a flame-spra yed zinc coating on the exterior of the fiber bundle. Figure 3-10 shows differences in zinc intensity based on polarity; positive brush spectra showed significantly higher con centrations of zinc than negati ve brush spectra. The current direction was reversed multiple times, and arci ng spectra from the positive brush consistently showed higher zinc intensities. Intensities fo r the copper species were comparable for the positive and negative brushes. The same trend was observed for arcing events while sliding in humid CO2 environmentscopper intensities for posit ive and negative polarity spectra were 68

PAGE 69

similar, while zinc intensities for the positive spectra were significantly higher than the negative spectra. Efforts were made to embed the fibe r optic within the copper fiber bundle to collect light emitted from arcing events; however, the in tensity of light collected by positioning the fiber optic within the fiber bundle was too low to be analyzed. The high packing fraction of the copper fiber brushes likely obstructed the fiber op tic from arcing events at the sliding interface. Low Sliding Velocity Studies of Copper Fiber Brush Sliding Electrical Contacts Copper Fiber Brush on Copper Flat in Humid Carbon Dioxide A series of low sliding velocity test s (0.01 m/s) were conducted in humid CO2 environments. The primary goals of these ex periments were to obtain carefully controlled measurements of friction coefficient, contact resistance, and wear in humid CO2 environments for copper fiber brushes sliding on copper and compare values to those measured at high sliding velocities (~5 m/s). The use of low sliding velocities was designed to minimize system dynamic effects and reduce potential causes for error in friction and resistance measurements. Throughout the literature on s liding electrical contacts, characterization of the mating surface to the brush (in this case, a copper roto r) had been largely neglected. Using the linear reciprocating tribometer shown in Figure 2-10 in conjunction with its in situ profilometry capabilities, wear of the rotor surface was studied in a humid CO2 environment. Data shown in Figure 3-11 [56] corresponds to a low sliding velocity test with a copper fiber brush on a polished copper counterface in humid CO2. The data points represent the average contact resistance and average friction coefficient for each cycle. During th e initial 5,000 cycles of sliding, friction coefficient increased to appr oximately 0.34 before gra dually decreasing. The increase in friction for the positive counterf ace between cycles 5,000 and 15,000 was due to drift in the zero of the load cell. Upon re-zeroing the load cell at cycle 15,000, the measured friction coefficient for ensuing cycles was ~0.24. After 45,000 cycles of sliding in humid CO2, the 69

PAGE 70

friction coefficient for the both the positive and ne gative counterfaces had stab ilized at a value of ~0.23. Contact resistance showed no clear depend ence on the direction of current flow. For the positive and negative counterface, the contact resi stance tended to decrease with an increasing number of cycles. Of particular interest was the increase in friction in the positive counterface test between cycles 20,000 and 25,000, highlighted by the box in the lower left portion of Figure 3-11. The timing of the friction increase corresponded to an unintentional increase in the sample temperature, shown in the plot on the right of Figure 3-11. Up to cycle 20,000, the copper counterface was subcooledthe temperature of the copper counterface sample (28C) was approximately 1C below the ambient temperature inside the environmenta l enclosure (29C). Around cycle 20,000, the recirculating chiller experi enced an unexpected failure in which the water pump ceased operating. Without a steady supply of cooling water, the counterface temperature increased from approximately 28C to 35C due to resistive heating of the sample while conducting 180 A/cm2 through the contact. As the te mperature increased, the friction coefficient for the positive counterface increased from 0.25 to 0.29. The ambient temperature also increased slightly above 30C but remained lower than the sample temperature. Prior to cycle 25,000, a new recirculating water chiller wa s installed, and cooling to the sample was restored. As a result, the sample temperatur e decreased below 28C (s ubcooled once again) and the friction coefficient decreased below 0.25. In situ measurements of the copper counterface wear tr ack topography and volume loss were made using a scanning white light interferomet er. By comparing each wear track profile to the original surface profile (cyc le 0), the material volume loss was calculated. The total volume 70

PAGE 71

loss ( Vol ) was calculated by summing the change in volume ( Vol ) over all pixels in the scan area (I) according to Equation 3-3. I i iVolVol (3-3) The volume loss per pixel was found by determin ing the change in surface height over the number of sliding cycles and multiplying the height change by the pixel area as shown in Equation 3-4, where hi,n is the height of pixel i at cycle n, hi,n+m is the height of pixel i at cycle n+m and x and y are the lateral dimensions of the pixel. ,, iininmVolhhxy (3-4) Volume loss was calculated at the end of each te st segment of constant current sliding and plotted in Figure 3-12 [56]. The data points of Figure 3-12 are representative of the measured volume loss for only the scan area depicted in Figu re 2-12, which is a fract ion of the entire wear track. Although the cumulative volume loss for the negative counterface was higher than the cumulative volume loss for the positive counterface, the difference was not caused by current effects. The largest discrepancy in volume loss for the two samples occurred initially when both were sliding under no current (between cycles 0 and 15,000). Beyond cycle 15,000, the trends in volume loss were similar for the positive and negative counterface. The counterface wear rate for each constant current segment of th e test was calculated using Equation 3-5, where Vol is the volume loss in mm3, F is the applied load in N, d is the total sliding distance in m, and K is the wear rate in mm3/Nm. Vol K Fd (3-5) The applied load was assumed to be constant at 1.0 N. Because the measured volume loss was obtained from a fraction of the track length, the sliding distance used in the wear rate calculation 71

PAGE 72

was scaled appropriately to the width of the scan area. The calculated wear rates are listed in Table 3-2 for the negative counterface and Table 3-3 for the positive counterface. The highest wear rates for both samples occu rred during the initial 15,000 cycles of sliding when current was not being passed through the contact. At a current density of 180 A/cm2, counterface wear rate did not show a dependence on the direction of cu rrent flow. Wear rates for the positive and negative counterfaces at 180 A/cm2 ranged from 2.6 10-7 to 8.8 10-7 mm3/Nm. These values were comparable to wear rates between cycles 25,000 and 35,000 when current was not passing through the contact. The uncertainty in the wear volume, u( Vol ), was calculated acco rding to Equation 3-6. 2 22 2 in inmuVolIxyuhIxyuh,2 (3-6) The uncertainty in the height measurement is not dependent on the cycle number; therefore, u( hi,n) is equivalent to u ( hi,n+m), and Equation 3-6 can be simplif ied to the expression shown in Equation 3-7. 2 22 uVolIxyuh (3-7) For the scanning white light interferometer (Zygo NewView 5032), the uncertainty in the surface height measurement, u( h), was conservatively estimated to be 10 nm. Given that the total number of pixels was 2.15 106 and the area per pixel was 2.03 m2, the uncertainty in volume loss was calculated to be 4.2 10-8 mm3. Referring to Figure 3-12, the uncertainty is approximately three orders of magnitude below the measured volume losses and, thus, too small to be depicted on the plot. After testing, the worn copper counterface samples were examined using SEM. Micrographs of the negative counterface and pos itive counterface samples are shown in Figure 3-13. The micrographs are oriented such that the sliding direction is ver tical. The wear track 72

PAGE 73

regions of the samples exhibited signs of plas tic deformation. Wear track morphology was not significantly different for the positive and negative polarities. For both po larities, material was transferred from the brush to the counterface, ev idenced by the numerous plateaus aligned in the sliding direction. Profilometry confirmed that th e height of many of thes e plateaus was greater than the original surface. Scratches oriented in the sliding direction indicated that abrasive wear or possibly third body wear contributed to material removal. Although not shown in the micrographs of Figure 3-13, plate-li ke wear debris was observed at the end points of each wear track. Temperature Study in Humid Carbon Di oxide and Humid Argon Environments To investigate the role of CO2 in tribofilm chemistry, the copper fiber brush on copper flat sliding geometry shown in Figur e 2-10 was studied in humid CO2 and humid argon (> 95% RH) environments over a range of temp eratures. Ambient temperature (T0) remained relatively constant at 26C. Counterface temperature (TS) was varied from 22C to 38C. Each temperature was maintained for 2,000 cycles of sliding. Friction coefficient and contact resistance are plotted for the different temperat ures in Figure 3-14 [56]. Each data point in Figure 3-14 represents the average steady-state value for a given temperature. Error bars represent the standard deviation in each aver age value. Error bars were omitted when the standard deviation was smaller than the size of the data point. For the average friction coefficients, standard deviations below 0.01 were omitted. In the humid CO2 environment, the lowest recorded fr iction coefficients occurred when the sample temperature was at or below the ambi ent temperature. An increase in sample temperature to 30C (TS T0 = +4C) resulted in an increase in friction coefficient in humid CO2. Friction coefficient continued to increase as sample temperature increased further. In the humid argon environment, the lowest friction coefficient was observed when the sample 73

PAGE 74

temperature was equal to the ambient temperature. When the sample temperature increased above the ambient temperature in humid ar gon, friction coefficient increased. At all temperatures examined, friction coefficient for a copper fiber brush on copper flat sliding contact was lower in humid CO2 environments than humid argon envi ronments. Friction coefficient was more sensitive to temperature changes in the argon environment as indicated in Figure 3-14 by the increasing differential in friction coefficients as temperature increased. In terms of contact resistance, the optimal sample temperature was equal to the ambient temperature for both the CO2 and argon environments. When the sample temperature decreased below the ambient temperature (TS T0 < 0), the contact resistance increased in both environments. When the sample increased above the ambient temperature (TS T0 > 0), the contact resistance increased in bot h environments. The change in contact resistance was more significant for increasing temperature than decreasing temperature, although only one temperature below ambient (22C) was investigate d. Contact resistance was generally lower in humid CO2 environments than humid argon environments for a given temperat ure. At a sample temperature of 26C (TS T0 = 0C), the average contact re sistance in humid argon (0.45 m ) was approximately 1.7 times higher than th e average contact resistance in humid CO2 (0.27 m ). At the highest sample temper ature investigated (38C; TS T0 = +12C), the average contact resistances in CO2 and argon were nearly the same.92 and 1.97 m respectively. X-ray Photoelectron Spectroscopy (XPS) of Copper Sliding Surfaces in Humid Carbon Dioxide The wear track regions of three copper counterface samples tested in humid CO2 were analyzed using X-ray photoelect ron spectroscopy (XPS) [56]. The three samples represented three different current flow conditionsno cu rrent, positive counterface, and negative counterface. Spectra obtained from the three samples are shown in Figure 3-15. Shown in 74

PAGE 75

Figure 3-15 A is the carbon 1 s spectrum for the no current sample [56]. The no current sample spectrum was representative of all three samples, so the positive and negative counterface spectra were omitted for clarity. Several carbonaceous speci es were detected within the wear tracks of all three samples. The highest intensity peak of the carbon 1 s spectra was at 284.6 eV and was attributed to amorphous carbon [57]. Peak in tensities for chemisorbed carbon dioxide (CO2 -) and carbonates (CO3) were observed between 286 a nd 290 eV [62-64]. Methoxy (OCH3) and formate (HCOO ) species, which may have contributed intensity within the energy range examined, could not be spectroscopica lly distinguished [63, 65, 66]. A CF2 species with a peak intensity at approximately 292 eV was dete cted on all samples. It was hypothesized that the CF2 species originated from fluorinated grease used in the drawing of the copper fibers for the brush, and trace amounts of th e grease were transferred from the brush to the counterface during testing. XPS analysis of the coppe r fiber brush showed significantly higher concentrations of the CF2 species on the fibers compared to the counterface. The copper 2 p3/2 spectra for the wear track region of all three samples are shown in Figure 3-15 B [56]. Each spectrum was normalized by the total integrated copper intensity. The close proximity in binding energy of copper and cuprous oxide (Cu2O) prevented quantification of relative amounts of each in the wear tracks [57, 67]. Shown in Figure 3-15 C are the Cu LMM Auger spectra for all three sample s [56]. Transitions characteris tic of metallic copper and Cu2O were identified at kinetic energies of 918.8 eV and 916.5 eV, respectivel y [57, 67]. Although the spectra of Figure 3-15 C appear similar, differentiation yields additional information on the relative amounts of metallic copper and Cu2O. By comparing the peak-to-peak ratios of Cu2O to Cu for the differentiated spectra of all thr ee samples, differences in composition become apparent. The Cu2O to Cu ratio for the no current counterface was 1.1, while the negative 75

PAGE 76

counterface had a ratio of 1.2. The ratio for th e positive counterface was significantly lower at 0.6. This comparison suggest s a relative absence of Cu2O in the positive count erface wear track. Single Copper Fiber on Copper Di sk in Humid Carbon Dioxide Friction Coefficient and Contact Resistance The single fiber sliding contact proved to be ve ry sensitive to run-out and eccentricity of the disk surface. This sensitivity was observed in the normal load signal. For an initial applied load of 500 N, the average load during slidi ng was 450 N with vari ations of 50 N. Average friction coefficients for the positive brus h and negative brush configurations were 0.22 and 0.24, respectively. For comparison, a copper fi ber tested under no current had an average friction coefficient of 0.23. These values of friction coefficient fo r copper-on-copper under unidirectional sliding in humid CO2 were similar to those obtain ed under reciprocating sliding. Test geometry appeared to have little effect on the frictional behavior of the copper-on-copper system in humid CO2. For the positive brush and negative brush sa mples, current-voltage (I-V) sweeps were performed at several points during testing. Data from one such I-V sweep is shown in Figure 316. In all sweeps, the contact displayed Ohmic be havior. The slope of the best-fit line through the current-voltage data yielded a co ntact resistance of approximately 0.8 for both the positive brush and negative brush current directions. For a contact force of 500 N, the contact resistance of 0.8 Ohms for copper-on-copper in humid CO2 is similar to other reported contact resistances for precious me tal contact pairs [68-70]. Electron Microscopy of Worn Fiber Surfaces Worn fiber surfaces examined by SEM showed distinct differences based on current. As seen in Figure 3-17, the surfaces of the positive and negative fibers are more heavily deformed than the no current fiber. Large amounts of w ear particles and deformed material are present 76

PAGE 77

along the perimeter of the fiber. This may indicat e that the current -carrying fibers softened due to resistive heating at such hi gh current densities, resulting in increased plastic deformation. However, the contact voltage remained below the softening voltage for copper. Another possible cause is the loss of water from the sliding interf ace due to resistive heating. With less of a lubricating water layer, shear ing of the fiber surfaces would be expected to increase. Cross sections of the worn fiber surfaces we re created using focused ion beam milling to examine the subsurface microstructure of the fibers after sliding in humid CO2. All cross sections were made parallel to the direction of sliding. Prior to milling, a platinum layer was deposited on the area of intere st of the fiber to protect th e surface from ion damage. Micrographs are shown in Figure 3-18 with the pl atinum layer and copper fiber identified for the no current, negative, and positive fibers. The sl iding interface is located below the platinum layer. The wire drawing process created an el ongated grain structure in the fibers which was most noticeable away from the sliding interf ace. Near the sliding interface, there was a definitive reorientation of the grain structure in th e direction of sliding (horizontal, as viewed in Figure 3-18). Additionally, the grain size near the interface was significantly refined compared to the bulk material. The approximate depth of the reoriented and refined grain structure appeared to be greatest in the positive fiber and least in the nega tive fiber. However, it should be noted that only one location was examined on each fiber surface. To make a more definitive statement about the depth of the deformed materi al layer on the fiber surfaces, multiple locations on each fiber should be sampled. In the positive and negative fibers, resistive heating may have contributed to grain growth. Th is effect was most evident aw ay from the sliding interface because mechanical deformation likely refined the larger grains near the surface. 77

PAGE 78

Samples for transmission electron microscopy (TEM) analysis were prepared from the areas shown in Figure 3-18. During the sample lift -out procedure, the sample from the positive fiber fractured; thus, the positive fiber sample wa s not included in the TEM analysis. Bright field TEM micrographs for the no current and negative fibers are shown in Figure 3-19. A protective carbon layer deposited on the fiber su rfaces after wear testi ng is visible in the micrographs. The copper fiber and deposited carbon layer are labeled in Figure 3-19 A and Figure 3-19 B for clarity. The sliding interface wa s located directly beneath the carbon layer. On the negative fiber, there was a distinctive layer approximately 50 nm in thickness present at what appeared to be the sliding interface. EDS analysis (Figure 3-19 B; right) showed this layer to be carbon-rich. Only trace amounts of oxygen were detected in this layer, and it was also deficient in copper when compared to the bulk of the fiber. Contact resistance data from I-V sweeps (Figure 3-16) of the single fibers suggest ed that any oxide on the sliding interface would be very thin and non-uniform, as the conduction path was dominated by metal-metal contacts. The resistivity of Cu2O films is more than 7 orders of magnitude higher than pure copper [61, 71]. The micrograph of the no-curre nt fiber also showed what appe ars to be an interfacial layer approximately 30 nm thick. The interfacial layer on the no-current fiber was nearly identical in composition to the carbon layer. Both layers exhibited high concentrations of carbon with low concentrations of copper and only trace amounts of oxygen. 1-Pentanol as a Vapor Phase Lubricant 1-Pentanol Vapor Phase Lubrication of Quartz-Silicon Sliding Pair The use of 1-pentanol as a vapor phase addi tive was investigated with a quartz (SiO2) on silicon sliding pair. Test conditions were desi gned to replicate the study performed by Asay, Dugger, Ohlhausen, and Kim [59]. A dry argon envi ronment was used as a comparator. Testing 78

PAGE 79

was performed on the linear reci procating tribometer integrated with a scanning white light interferometer (see Figure 2-10). Average friction coefficients for quartz-on-si licon sliding in dry argon and 1-pentanol saturated argon are plotted in Figure 3-20. In both environments, friction was initially low approximately 0.13. Between cycles 20 and 30, the friction coefficient in the dry argon environment rapidly increased from 0.13 to approxi mately 0.8. For subsequent cycles in the dry argon environment, the friction rema ined at higher values and varied significantly. In the 1pentanol saturated environment, friction coeffi cient remained at a value of approximately 0.13 for 1000 cycles of reciprocating sliding. Changes in the friction coefficient were relate d to evolution of wear track topography and major wear events. Non-contacting profilometry images of the wear track on the silicon wafer were acquired at set intervals throughout the tests: for cycles 1 10, image every cycle; for cycles 11 100, image every 10th cycle; for cycles 101 1000, image every 100th cycle. A sampling of the images is shown in Figure 321 A for the dry argon environment and Figure 3-21 B for the 1-pentanol saturated argon environmen ts. The unworn silicon surface had an average roughness of approximately 1 nm. After 10 cycles of sliding, the wear tr ack was clearly visible in profilometry images from both environments. Some plastic deformation had occurred at cycle 10; the depth of the material removal was several nanometers and there was little or no observable debris. This low wear regime exhi bited low friction as well. In the dry argon environment between cycles 20 and 30, the silic on surface underwent significant changes. The depth of the wear track increas ed from several nanometers to several hundred nanometers. Substantial amounts of debris were generated in and around the wear track. Wear and debris generation likely contributed to the rapid increase in friction observed over th is period of time. 79

PAGE 80

Figure 3-21 A shows an image taken at cycl e 100 in the dry argon environment which was representative of the changes which took place during the transi tion from low friction to high friction. The height scale for the cycle 100 imag e of Figure 3-21 A was changed from 3 nm to 700 nm to capture the severe wear events a nd resulting debris accumulation. Therefore, the color scale in the cycle 100 image is not equivale nt to the color scale in the cycle 0 and cycle 10 images. In dry argon, the observed generation of debris corresponded to high friction in the sliding contact. The rapid transition from low to high friction may be due to breakthrough of the native oxide layer on the silicon. Conversely, the wear track on the silicon surf ace in the 1-pentanol saturated environment showed no signs of severe wear or debris genera tion over 1000 cycles of sliding, indicating the na tive oxide layer remained intact. The images in Figure 3-21 B show only minor plastic deformation in the wear track, and Figure 3-20 shows c onsistently low friction coefficient (~0.14) in the 1-pentanol environment. 1-Pentanol Vapor Phase Lubricatio n of a Copper-Copper Sliding Pair Although studies have shown 1-pentanol to be an effective vapor phase additive for silicon sliding contacts [58, 59], and that behavior was independently verified using the test setup described in the present study, there are no publis hed studies of 1-pentan ol as a vapor phase lubricant for copper-on-copper slid ing electrical contact s. A follow-up to the quartz-on-silicon study in 1-pentanol saturated ar gon was performed using the same setup (Figure 2-10) and test conditions, but substituting a c opper-on-copper sliding pair. Recorded friction coefficient for copper-on-c opper sliding in 1-pentanol saturated argon environment under no current is shown in Figur e 3-22. The initial friction coefficient was approximately 0.24. Similar to quartz-on-silicon in dry argon, the friction coefficient rapidly increased at approximately cycle 30. This friction increase was likely due to br eakthrough of the 80

PAGE 81

native oxide on the copper surface. After break through, friction coefficient was extremely high because of metal on metal contact and high adhesi ve forces. In situ profilometry confirmed a significant wear event o ccurred between cycles 30 and 40. The images in Figure 3-23 show some plastic deformation in the wear track region at cycle 30. By cycle 40 (Figure 3-23; bottom left), the wear track depth ha d increased to 4 m in some locations. The gross plastic deformation and material removal observed at cycle 40 corresponde d to higher friction coefficients. Because the operating environmen t was non-oxidizing, it would be expected that high wear rates would result in metal-metal cont act and high friction. Wear rate of the copper counterface was quantified according to Figure 3-24. The average line prof ile across the wear track was taken at cycle 100 and s ubtracted from the average line profile of the same wear track location at cycle 0. The volume of material removed is indicate d by the shaded region in Figure 3-24. The resulting wear rate for the copper counterface was 8 10-3 mm3/Nm. Nominal pressures for metal fiber brushes are typically well below th e pressure (270 MPa) used in studying vapor phase lubrication with 1pentanol. To compare 1pentanol environments to humid CO2 environments, a test was performed usi ng a copper fiber brush sliding against a polished copper flat at a nomi nal brush pressure of 1.4 105 Pa and a current density of 180 A/cm2. The average friction coefficient and contac t resistance for each cycle are plotted in Figure 3-25. For the test involving 1-pentanol, at cycle 1000 the environment was intentionally changed to dry argon, and then reverted back to 1-pentanol saturated argon at cycle 1500. This was done to investigate the ability of 1-pentanol to aid in recovery of the system after high friction had been established. The 1-pentan ol saturated environm ent showed minimal improvements in friction compared to the dr y argon environment; in dry argon the friction coefficient stabilized at appr oximately 0.65, while in 1-pentanol saturated argon the friction 81

PAGE 82

coefficient stabilized between 0.45 and 0.50. Af ter operating in dry argon for 500 cycles, the friction coefficient decreased to approximately 0.45 upon the re-introduction of 1-pentanol into the environment. Measured friction coefficients for the copper fiber brush in humid CO2 (~ 0.25) were significantly lower than friction coefficients in 1-pentanol saturate d argon for the same conditions. Contact resistances for the humid CO2 and 1-pentanol saturated argon environments were comparable. After 2000 cycles of sliding, c ontact resistance for the copper sliding pair in humid CO2 stabilized around 3.0 m while the same sliding pair in 1-pentanol saturated argon reached a value of approximately 3.1 m Tests run for longer durations (10,000 cycles or more) showed that contact resistance continued to decrease with incr easing sliding distance. 82

PAGE 83

Figure 3-1. Friction coefficient (t op) and brush wear (bottom) for graphite lubricated monolithic silver brush sliding electrical contacts in ambient air. Current density (upper axis) was varied from 0 200 A/cm2. Brush wear significan tly increased at 200 A/cm2. Steady state wear rates are shown for the positive and negative brushes at 200 A/cm2 [53]. 83

PAGE 84

A B Figure 3-2. EDS analysis of worn silver brush surfaces. A) Positive brush surface spectrum primarily showed presence of transferred graphite. B) Negative brush surface spectrum showed significant amounts of copper along with graphite [53]. Figure 3-3. Scanning white light interferometry image of gra phite transfer film on copper rotor surface (positive brush track). Graphite film thickness varied from 1 m to a maximum of approximately 5 m. 84

PAGE 85

A B Figure 3-4. Micrographs used in calculation of copper fiber packi ng fractions for brushes. A) SEM micrograph of as-received copper fiber brush surface. B) SEM micrograph with false coloring of fiber surfaces for calcul ation of packing frac tion. Fiber packing fraction for this particular brush was 0.48. Figure 3-5. Load versus deflecti on plot for a copper fiber brush in compression. Brush stiffness was in the range of 150 to 190 N/mm. 85

PAGE 86

Table 3-1. Summary of tribological testing with copper fi ber brushes and PTFE-based solid lubricants. Lubricant Brush polarity Brush wear rate (m/m) Lubricant wear rate (m/m) Friction coefficient Contact resistance (m ) PTFE Positive 2 10-8 1 10-7 0.21 0.04 PTFE Negative 1 10-8 4 10-8 0.27 0.02 16.2 3.0 PTFE/In Positive 7 10-9 9 10-10 0.25 0.03 PTFE/In Negative 6 10-10 1 10-9 0.22 0.01 14.3 1.5 86

PAGE 87

A B Figure 3-6. Results of tri bological testing of copper fiber brushes conducted on the environmental brush-rotor tribometer in a humid CO2 environment at a sliding velocity of 5 m/s and current density of ~180 A/cm2. A) Brush wear plotted for positive (black) and negative (gray) copper fibe r brushes. B) Voltage drop plotted for positive (black) and negative (gra y) copper fiber brushes. 87

PAGE 88

A B Figure 3-7. SEM micrographs of c opper fiber brushes after high sliding speed (5 m/s) test on the environmental brush-rotor tribometer in humid CO2. A) Secondary electron image of worn negative brush surface. B) Secondary electron image of worn positive brush surface. A B Figure 3-8. Surface profiles of worn copper fibers obtained by scanning white light interferometry after high sliding speed testing (5 m/s) in humid CO2. A) Worn fiber from negative brush. B) Worn fiber from positive brush. 88

PAGE 89

Figure 3-9. Arc lifetimes for positive and negative brush polarities plotted against photomultiplier tube (PMT) voltage signal t uned to the 521.8 nm Cu(I) emission line. Polarity did not have a significant effect on arc lifetimes for the stationary copper fiber brush contact. 89

PAGE 90

Figure 3-10. Atomic emission sp ectra collected from arcing even ts with a zinc-coated copper fiber brush sliding contact in ambient air. The same brush contact was observed under both positive and negative polarities. Zinc was much more prevalent in the positive brush arcing spectrum than the negative brush arcing spectrum. Spectra collected in humid CO2 environments were comparable to those collected in ambient air. 90

PAGE 91

Figure 3-11. Average contact re sistance and friction coefficient for low speed (0.01 m/s) copper fiber brush on copper sliding in humid CO2 at current densities of 0 and 180 A/cm2. Testing was performed for both current fl ow directionspositive counterface (black) and negative counterface (gray). Highlighted on the right is a subset of the recorded temperatures for the positive counterface test. Note the increase in sample temperature following cycle 20,000 and the re sulting increase in friction for the positive data set [56]. 91

PAGE 92

Figure 3-12. Calculated volume loss for a c opper counterface sliding against a copper fiber brush in humid CO2 environment at 0.01 m/s. Volume loss was determined using in situ profilometry measurements of the c ounterface wear track. Lines are drawn to connect data points for clarity [56]. Table 3-2. Copper counterface wear rates for the negative counterface current condition in humid CO2 at low sliding velo city (0.01 m/s) [56]. Cycle number Sliding distance (m) Current density (A/cm2) Wear rate (mm3/Nm) 0 15,000 0 150 0 2.1 10-6 15,001 25,000 150 250 -180 6.9 10-7 25,001 35,000 250 350 0 4.1 10-7 35,001 45,000 350 450 -180 5.7 10-7 Table 3-3. Copper counterface wear rates for th e positive counterface current condition in humid CO2 at low sliding velocity (0.01 m/s) [56]. Cycle number Sliding distance (m) Current density (A/cm2) Wear rate (mm3/Nm) 0 15,000 0 150 0 1.6 10-6 15,001 25,000 150 250 +180 2.6 10-7 25,001 35,000 250 350 0 1.0 10-6 35,001 45,000 350 450 +180 8.8 10-7 92

PAGE 93

A B Figure 3-13. SEM wear track char acterization after tribol ogical testing of c opper fiber brushes at low sliding velocities (0.01 m/s) and 180 A/cm2 in humid CO2. A) SEM micrograph of negative copper counterface (under positive brush) wear track. B) SEM micrograph of positive copper counterface ( under negative brush) wear track. Figure 3-14. Average friction coe fficient (left) and contact resist ance (right) for a copper fiber brush in low speed (0.01 m/s) reciproca ting sliding against a polished copper counterface in humid CO2 and humid argon environments. Sample temperature (TS) was varied while maintaining a constant ambient temperature (T0) within the chamber. Error bars represent the standard deviations in the average values [56]. 93

PAGE 94

A B C Figure 3-15. XPS analysis of copper counterface wear tracks after low speed (0.01 m/s) sliding in humid CO2 environments under the following current conditions: no current, positive counterface (+180 A/cm2), and negative counterface (-180 A/cm2). A) The carbon 1s spectrum for the no current sample; spectrum was repres entative of all samples, so positive and negative counterface spectra were not displayed. B) Copper 2p3/2 spectra for no current (solid black), negative counterface (dash-dot blue), and positive counterface (dash orange) wear tracks. C) Copper L3M45M45 Auger spectra for all three wear tracks. The kinetic energies corresponding to Cu and Cu2O transitions are discernable [56]. 94

PAGE 95

Figure 3-16. Current-voltage (I -V) sweep of single copper fibe r on copper disk contact after sliding in humid CO2 environment. Applied normal load was 500 N. Figure 3-17. SEM micrographs of worn singl e copper fibers after sliding in humid CO2. From left to right: no current, negative, and positive fibers. 95

PAGE 96

Figure 3-18. SEM micrographs of worn single coppe r fiber cross sections created by focused ion beam milling. A platinum layer was deposit ed on each fiber surface prior to milling to protect surface features from ion damage The sliding interface is located below the platinum layer. 96

PAGE 97

A B Figure 3-19. TEM analysis of lift-outs from worn single copper fiber surfaces. A) TEM micrograph of no current fiber surface show ing location of EDS an alysis (black dot) and corresponding EDS spectrum. B) TEM micrograph of negative fiber surface showing location of EDS analysis (blue dot) and corresponding EDS spectrum. 97

PAGE 98

Figure 3-20. Average friction coefficient for each reciprocating cycl e of quartz-on-silicon sliding in dry argon and 1-pentanol saturated argon environments. 98

PAGE 99

A B Figure 3-21. Profilometry of wear track formed on silicon during sliding against a quartz ball. A) Wear track profiles at cycles 0, 10, and 100 in a dry argon environment. For cycle 100 in dry argon, the height scale was incr eased to nm because of extreme changes in surface topography. B) Wear track profiles at cycles 0, 10, and 100 in an argon environment saturated with 1-pe ntanol (height scale nm). 99

PAGE 100

Figure 3-22. Friction coefficien t for copper-on-copper reciprocati ng sliding (no current) in 1pentanol saturated argon environment over 100 cycles. Figure 3-23. Profilometry images of polished copper counterface showing evolution of wear track in 1-pentanol saturate d argon environment: cycle 0 (upper left), cycle 30 (upper right), cycle 40 (lower left), and cycle 100 (lower right). Color scale shows heights ranging from -4 to +4 m. 100

PAGE 101

Figure 3-24. Average line profiles for wear track formed by copper-on-copper sliding in 1pentanol saturated argon envi ronment. Volume loss was calculated by subtracting the profile for cycle 100 from the profile for cy cle 0 and extrapolating over the length of the wear track. 101

PAGE 102

Figure 3-25. Contact resistance ( upper plot) and friction coefficien t (lower plot) for copper fiber brush on copper flat low sliding sp eed (0.015 m/s) test in humid CO2 (blue) and 1pentanol saturated argon (black) environm ents. For the 1-pentanol saturated argon environment, 1-pentanol was intentiona lly removed between cycles 1000 and 1500 to determine if reintroducing 1-pe ntanol to the environment would lead to recoverable friction. 102

PAGE 103

CHAPTER 4 DISCUSSION Decoupled Solid Lubrication of Sliding Electrical Contacts Graphite Testing of monolithic silver brushes with decoupled graphite lubrication in ambient air environments demonstrated the limitations of graphite-based sliding el ectrical contacts for high current applications. Many electrical brushes ut ilize a composite of gr aphite and a conductive metal to achieve a combination of low resistance, low friction, and low wear in a sliding electrical contact application. The present studied examined how decoupling graphite lubrication from a conductive metal brush could be used to imp rove electrical effici ency without increasing mechanical losses. Ideally, through careful se lection of the applied lubricant load, thin, protective transfer film s would be created without increasin g contact resistance, allowing for efficient operation at high currents without the restrictions of a non-oxidizing humid environment. Over the range of currents investigated, fric tion coefficient did not exhibit a dependence on polarity or current magnitude with graphite lubricated silver brushe s. Lancaster and Stanley [72] found friction coefficient of an electrographitic ca rbon and copper sliding pair to be independent of polarity. Other studies have noted polarity effects in graphite-copper slid ing electrical pairs [73, 74]. Electric fields have been shown to cause lubricant deterior ation, which adversely affected frictional properties [ 75, 76]. Graphite transfer films in the present study showed no adverse reactions or decomposition unde r current densities up to 200 A/cm2. Brief excursions of high friction were the result of lubricant pin failures. Subseque nt removal and replacement of the failed lubricant pins reduced the friction co efficient and demonstrated the ability of the system to recover to its stable operating point. 103

PAGE 104

At current densities below 150 A/cm2, graphite effectively lubr icated the silver brush contacts, yielding wear rates that could not be accurately quantified with the given instrumentation. At 200 A/cm2, a critical threshold was cro ssed where brush wear increased notably and a bifurcation of the positive and ne gative brush wear rate s occurred. Higher wear rates for the positive brush compared to the nega tive brush are commonly found in the literature related to metal-graphite el ectrical brushes [6, 23, 24, 34, 42] Energy dispersive X-ray spectroscopy (EDS) analysis of the worn brush su rfaces (Figure 3-2) showed notable differences in composition based on polarity. Copper concentra tions on the negative brush face were higher than that of the positive brush face. Mechanical transfer of copper from the rotor to the brush surfaces should be independent of current flow direction. The inability of the monolithic brush to track imperfections in the rotor surface may cau se small breaks in the electrical circuit at the brush contacts. Dynamic contact resistance meas urements were up to five times higher than stationary contact resistance measurements, suggesting dynamic effects may interrupt the conduction path across the brush-rotor contact. Small gaps created between the brush and rotor may initiate arcing events which further enhance material removal. Under the negative brush, electrons impinge on the copper ro tor. Copper atoms vaporized through arcing events can then be redeposited on the negative brush surface. El ectric field enhanced mi gration of metal atoms (electromigration) has also been cited as a po ssible material transfer mechanism in sliding electrical co ntacts [77, 78]. The primary shortcoming of the decoupled graph ite lubrication system is the inability to achieve sufficiently low electrical losses. The thickness of the graphite transfer films (Figure 33) prevents intimate metal-on-metal contact. Th ick transfer films are preferred for brush wear reduction, but they also limit the conductivity at the interface. The elec trical losses of the 104

PAGE 105

graphite system will be discussed in more detail in a following section titled Summary of Brush Electrical Losses. Decoupled gr aphite lubrication of monolithic metal brushes appears to be suitable for low current applications. One issue th at remains to be addressed is lubricant wear. Without constant replenishment of the graphite transfer film, severe adhesive wear and cold welding between brush and rotor will occur. The graphite lubricant pins need to be periodically inspected and replaced. Such maintenance would be an additiona l obstacle to implementation of decoupled graphite lubricati on in practical applications. Polytetrafluoroethylene (PTFE) and PTFE/In Composites Decoupled graphite lubrication of solid electr ical contacts successfully yielded low wear and low friction; however, the thickness of the transfer films produced unacceptably high contact resistance for high current applications. The use of polytetrafluoroet hylene (PTFE) as an alternative lubricant was investigat ed due to its ability to form thin, lubricious transfer films. The results shown in Table 3-1 indicate that PTFE lubrication produced reasonably low friction coefficients ranging from 0.21 to 0.27. Brush wear rates with PTFE were on the order of 10-8 m/m, which was extremely high considering wear rates on the order of 10-11 m/m and lower have been achieved with metal fiber brushes [8, 12, 46, 48, 50, 54]. The normalized brush pressure ( of ~0.8) was higher that what is typically recommended for me tal fiber brush operation. A decrease in pressure would likely yield lowe r wear rates but increased electrical losses. The addition of indium to PTFE was intended to increase the conductiv ity of the transfer film formed on the rotor surface while having minimal impact on the frictional behavior of the system. Table 3-1 shows that friction coefficien t with the PTFE/In composite was comparable to PTFE. The low shear strength and the low volum e percent (10%) of indi um are likely reasons for the similar frictional behavi or of PTFE and PTFE/In. The electrical conduc tivity of the PTFE/In transfer films was not si gnificantly better than PTFE. The average contact resistance 105

PAGE 106

during sliding was 16 m with PTFE and 14 m with PTFE/In, although th e standard deviations proved the difference to be statistically insignific ant. The low volume percent of indium in the composite may not have been sufficient to create conductive pathways thr ough the transfer film. Lubrication with the PTFE/In composite pr oduced increased brush lifetimes over PTFE lubrication. Compared to PTFE, the PTFE/In co mposite reduced the positive brush wear rate by 65% and the negative brush wear rate by 94%. Th e addition of indium to the PTFE lubricant also considerably decreased the we ar rates of the lubricant pins fo r both brushes. Wear rates for graphite lubricant pins with monolithic silver brushes ranged from 1 10-8 to 3 10-8 m/m. Ex situ observations of worn brushes showed PT FE and PTFE/In infiltration throughout the fiber bundles. Thus, much of the measured lubricant w ear is due to uptake of lubricant in the fiber brushes. Overall, PTFE/In composites yielde d improved wear performance for copper fiber brush sliding electrical contacts but little enhancement of electrical conductivity compared to PTFE lubrication. Copper Fiber Brush Sliding Electrical Contacts in Humid Environments High Sliding Velocities Using the brush-rotor tribometer (Figure 2-6) and a sliding velocity of 5 m/s, brush wear was measured for over 3000 km of sliding in a controlled humid CO2 environment. Displacement transducers attached to the backside of the brush holders were used to make the displacement measurements and quantify wear. Several factors complicated displacement-based brush wear measurements. Metal fiber brushes under constant load tended to splay at the sliding interface. Additionally, high curr ent densities created temperature fluctuations in the brushes and surrounding environment which caused brushes and brush holders to thermally expand and contract. Splaying could not be accounted for in the brush wear data. For long term testing (hundreds of km of sliding), it was assumed that after initial splaying, the brush reached a steady106

PAGE 107

state where splaying was no longer si gnificant relative to brush wear ; therefore, brush wear rates were calculated based on data from the steady-stat e regimes. Thermal fluctuations were evident in the brush wear data of Figure 3-6. Becau se the period of the temperature cycling was approximately 24 hours, brush displacement data poi nts were selected at 24 hour intervals and a linear regression line was fit to the data points to calculate the steadystate wear rate. The positive brush wear rate of 3 10-11 m/m and negative brush wear rate of 4 10-12 m/m were comparable to values from literature for metal fiber brushes in humid non-air environments [8, 12, 46, 48, 50, 54]. During high sliding speed testing, positive brus hes tended to wear at a higher rate than negative brushes. At the end of long tests, such as the one shown in Figure 3-6, brush appearance and performance differed from the as -received state. Deformation of the fiber surfaces and collection of debris in the void spaces between fibers caused the fiber brushes to behave like solid brushes. The scanning electr on microscopy (SEM) images shown in Figure 3-7 provided evidence of the solid nature of the brush surfaces. This may have accounted for the macroscopic arcing events observed at the positive br ush interface near the end of the test. Metal fiber brushes were designed to be compliant. The load was dispersed over a large number of independent contact points, allowing the fiber brus h to follow imperfections in the rotor surface. If the fibers were no longer able to act independently, the brushes would lose their ability to track the rotor surface. Because electrons were impi nging on the positive surface, material removal was intensified at the positive brush fiber tips. For arcing across a small gap, also termed a metal vapor arc, the positive surface expe riences a net material loss rela tive to the negative surface [7981]. The arcing damage observed on the positiv e brush in Figure 3-8 B was not common among 107

PAGE 108

all brushes examined by SEM. The surfaces of most worn positive brushes appeared similar to the surfaces of worn negative brushes. The inherent difficulties in making friction force measurements with copper fiber brushes at high sliding velocities were discussed by Argibay et al. [54]. Brushes were operated in a humid CO2 environment under unidirectional sliding agai nst a copper slip ring. At a velocity of 2.5 m/s, friction coefficients for the negative brush varied greatly between 0.08 and 0.34 while friction coefficients for the positive brush ra nged from 0.17 to 0.42. When the current density was increased from 120 to 180 A/cm2, the positive and negative brush friction coefficients bifurcated, with the positive brush friction coe fficient approximately twice the value of the negative brush friction coefficient. Decreasing the slid ing velocity from 2.5 m/s to 0.5 m/s while maintaining a current density of 180 A/cm2 removed the frictional bifurcation of the system, as friction coefficients for both brushe s varied around 0.20 at this speed. Argibay et al. [54] also utili zed lower velocity (0.5 m/s) sliding direction reversals to remove load cell bias from the frictional load measurements. Average friction coefficients calculated from lower speed motor reversals were approximately 0.20 for both brushes and were independent of current density. Load cell bias likely accounted for the drift in the positive and negative brush friction measurements at 2.5 m/s. Similar trends were observed in the contact resistance data from the same testpositive a nd negative brush resistances bifurcated when current density increased from 120 to 180 A/cm2, and the bifurcation disappeared when sliding velocity decreased from 2.5 to 0.5 m/s while maintaining 180 A/cm2. Although accurate measurements of friction and resistance are difficult to obtain at high slidi ng velocities, potential sources of error can be reduced by decreasing the sliding velocity and reversing the direction of sliding to account for bi ases in the system. 108

PAGE 109

The zinc coating on the exterior mesh wrapping of the copper fiber brushes was used as a means of tracking material transfer between the brush and rotor surfaces. Discoloration of the rotor surface under the positive brush was attributed to transfer of zinc from brush to rotor. Although it was not clear what effect zinc had on brush wear or contac t resistance, it was apparent that zinc was involved in the surf ace chemistry. After removing worn copper fiber brushes from the humid CO2 environment and allowing brushes to dry, fibers were examined by SEM and EDS. A number of rod-shaped and sphe rical features were obse rved on the worn brush fiber surfaces. Shown in Figure 4-1 is an SEM micrograph of the struct ures (upper left) with corresponding EDS dot maps for copper (upper right), oxygen (low er left), and zinc (lower right). The surface growths had higher oxygen contents, lower copper concentrations, and slightly higher zinc concentrations compared to the rest of the copper fiber surface. Though the exact composition could not be determined, the sp ectroscopic analysis suggested the features in Figure 4-1 may be a zinc oxide or zinc carbonate. Calculatio n of the Gibbs free energy of formation of ZnO and Cu2O at room temperature yielded va lues of -363.5 and -198.2 kJ/mol, respectively; therefore, formation of ZnO is favored over formation of Cu2O at room temperature. Deng et al. [63] investigated the reac tion products formed on a copper surface with 0.1 monolayer of zinc in a CO2 + H2O environment. The presence of zinc on the copper surface facilitated the formation of carbonate species and depleted chemisorbed CO2. Deng et al. [63] also noted that adsorbed oxygen r eacted with zinc to form ZnO. The role of zinc in arcing sliding contacts was also investigated on the brush-rotor tribometer. As seen in the atomic emission spect ra of Figure 3-10, higher concentrations of zinc species were observed in arcing events at the pos itive brush interface comp ared to that of the negative brush interface. The presence of zinc in the arcing spectra was not unexpected. As 109

PAGE 110

mentioned previously, plentiful amounts of zinc we re present on the exterior of the fiber bundle. Arcing was visually more prevalent around the peri meter of the brush, which coincides with the highest concentrations of zinc as well as the preferred path for current flow through a fiber brush [60]. It appeared that zinc wa s preferentially ionized under the positive brush polarity. At the positive brush interface, electrons leave the ro tor surface and impinge on the brush surface. Conversely, at the negative brush interface, elect rons leave the brush surface and impinge on the rotor surface. Therefore, at the positive brush interface, material from the brush was ionized more readily than rotor material; thus, higher concentrations of zinc were observed in the positive brush arcing spectrum. For the nega tive brush interface, the ionized material predominantly came from the rotor surface, and the negative brush spectrum appeared zinc deficient relative to the positive brush spectrum. Low Sliding Velocities While high sliding velocity testing is ideal in terms of evaluating elec trical brush materials for practical applications, low sliding velocity testing is better suite d for studying fundamental questions on current direction effects and surf ace chemistry. The low-speed reciprocating test geometry used in this study (Figure 2-10) rem oved many of the complexities of the high-speed brush-on-rotor tribometer (Fig 2-6) which lim ited measurement and analysis options. For example, in the brush-rotor tribometer, the vibr ations induced by operati ng at high revolutions per minute made it inherently difficult to make accurate force and voltage measurements. Performing tests on self-mated copper sliding electr ical contact systems at low sliding velocities generated more reliable measurements in addition to providing an opportunity to evaluate system dynamics effects on data obtaine d at high sliding velocities. Brush wear could not be accurately quantified in the lower sliding velocity testing of copper fiber brushes. Changes in brush mass and brush length were too small to be measured. 110

PAGE 111

Over 45,000 cycles of sliding, the total sliding distance was 450 m. Assuming a relatively low brush wear rate of 10-11 m/m, the total change in brush length would be 4.5 nm. Significant increases to the total sliding distance would be required to obtain meaningful brush wear measurements at the lower sliding velocity. In situ profilometry of a copper counterface s liding against a copper fiber brush in a humid CO2 environment at 180 A/cm2 allowed for interrupted wear m easurements of a simulated rotor surface without disturbing the testing environm ent. Volume loss calculations showed that counterface wear was independent of brush polarit y. In high sliding speed testing, positive brush wear rates were generally higher than negative brush wear rates. This polarity dependence trend was not observed for counterface wear at low sliding velocities. The nominal brush pressure used for the low speed study (1.4 105 Pa) was higher than the brush pressure for the high speed study (1.4 104 Pa). The higher brush pressure may have reduced electrically-induced wear mechanisms and increased mechanical wear. Additionally, the low sliding speeds reduced the probability of the brush fibers breaking contact with the counter face. This should also favor mechanical material removal over other current-induced mechanisms. The micrographs of the copper wear tracks generated in humid CO2 (Figure 3-13) lend some insight as to possible wear mechanisms for copper brush sliding electr ical contacts at low sliding velocities. No features indicative of arcing, such as ablation craters or droplets of resolidified material, were observed on the wear track surfaces. In certain locations, material buildup within the wear track was observed due to material transfer from the brush to counterface. The material transfer was obser ved on both the positive and negative counterfaces and did not have a dependence on the current flow direction. Profilometr y confirmed that the features had a greater height th an the original polished surface. Material was likely transferred 111

PAGE 112

from brush to counterface through an adhesive wear mechanism, which will be discussed in more detail in a following section titled The ories on Brush Wear Mechanisms in Metal Fiber Brush Sliding Contacts. Water adsorption was critical to the effici ent operation of self-mated copper sliding electrical contacts at low and high sliding velocities. Salmeron [82] studied the adsorption behavior of water on clean copper and Cu2O surfaces using a speci ally designed X-ray photoelectron spectrosco py (XPS) system that operated unde r ambient gas pressures. XPS allowed for differentiation between oxygen pres ent in the oxide, oxygen in hydroxyl (OH) groups, and oxygen in adsorbed water. The th ickness of the water layer was determined by measuring the areas under the peaks for each co nstituent. Salmeron [82] concluded that the formation of dangling hydrogen bonds is critical to water adsorption on a surface. For a Cu2O surface, molecular water adsorption was preced ed by passivation of the surface with hydroxyl groups, as shown in Figure 4-2 (adapted from Sa lmeron [82]). Small concentrations of oxygen vacancies at the oxide surface acted as adsorption sites for water. Upon dissociation of the water molecules, the vacant sites became saturated with hydroxyl groups. Molecular water films proceeded to grow due to strong interactions wi th the hydroxyl groups. As shown in Figure 4-2, the water layer thickness on Cu2O increased approximately linearly with increasing relative humidity. At 44% RH, approximately 2.5 monol ayers of water were present on the surface. The operating environment used in testing of copper-on-copper sliding electrical contacts at the University of Florida was pred ominantly a non-oxidizing cover gas (CO2 or argon) saturated with water. The oxygen concentration in the opera ting environment was typically between 50 and 100 ppm, and it was actively monito red during testing. XPS analysis of copper wear tracks confirmed the presence of an oxide (Cu2O) on all samples examined. Thus, the 112

PAGE 113

mechanisms and trends for water adsorption on a Cu2O surface (Figure 4-2) suggested by Salmeron were applicable to copper sliding electrical contacts in humid environments. Because water adsorption on Cu2O is dependent on relative hu midity, changes in operating conditions that affected either temperature or vapor pressure ultimately affected water adsorption on the sliding bodies. Figure 3-11 pr ovides an example of such a situation. As the temperature of the positive counterface increased (see cy cles 20,000 to 25,000), the friction coefficient increased due to a decrease in the water film thickness in the sliding contact. When cooling was restored prior to cycle 25,000, friction coeffici ent decreased as a result of increased water adsorption. Subcooling was demonstrated to be an eff ective method of maintaining adsorbed water films on the copper sliding surfaces A surface is subcooled if its temperature is below the dew point temperature, which is approximately equivalent to the ambient temperature at humidities approaching 100% RH. In Figure 3-14, the lowe st friction coefficients in the humid CO2 environment occurred when the copper c ounterface was subcooled. When counterface temperature increased, friction coefficient correspondingly increase d. The increase in counterface temperature decreased the local relative humidity near the counterface which in turn decreased the adsorbed water f ilm thickness. Similar results were observed in the humid argon environment, although a decrease in friction occu rred as the counterface temperature increased from 22C to 26C. Contact resistance in humid argon was also higher at a counterface temperature of 22C compared to 26C, which may suggest that the system was still being run-in and the native oxides on the brush and counterfa ce were still present in the sliding contact. Subcooling was also shown to be an effec tive means of reducing contact resistance for copper-copper sliding electrical contacts in humid environments. The optimum temperature for 113

PAGE 114

minimizing electrical losses was equivalent to the ambient temperature (26C) in both humid CO2 and humid argon environments. Increases in c ounterface temperature resulted in increased contact resistance for both environments. In itially it was believed that higher sample temperatures and less adsorbed water would incr ease material removal, resulting in more metalmetal points of contact (as well as higher friction) and lower contact resistance. However, the contact resistance data of Figure 3-14 did not agree with this hypothe sis. Clearly, water adsorption played a role in the electrical conduc tivity of the sliding co ntact, as the observed trends were not dependent on the cover gas. Composition of Tribofilms Formed on Copper in Humid Carbon Dioxide One of the goals of the temperature study s hown in Figure 3-14 was to understand the role of CO2 and its effects on the frictiona l and electrical properties of the self-mated copper system. Adsorbed water is an important component of th e system, but the function of the cover gas, specifically CO2, is not well understood. A water-saturat ed argon environment was used as a control to determine if CO2 was actively participating in the formation of surface films. These results definitively showed that CO2 was more than a non-oxidizing cover gas. Significant differences in the sliding behavior of the copper fiber brush on copper flat system were observed when comparing humid CO2 and humid argon environments at low sliding velocities. At all temperatures examined, friction coefficients were substantially lower in humid CO2 than humid argon. Moreover, the contact resistances were generally lower for humid CO2 than humid argon. Surface species other than adsorbed water contributed to the frictio nal and electrical properties of copper-on-copper sliding contacts. In addition to water acting as a boundar y lubricant, reactions between CO2, H2O, and the copper surface formed lubricating surface species. XPS analysis was used to characterize the tribofilms formed unde r copper-copper sliding electrical contacts in humid CO2 environments. XPS is a very surface sensitive analytical tool 114

PAGE 115

and ideal for characterization of thin surface films. The spectrum of Figure 3-15 A shows several carbonaceous species present in the wear tr acks of all samples. The same brush material was used for testing in humid CO2 and humid argon, so the CF2 species would not account for the differences in friction. The peak at 284.6 eV assigned to amorphous carbon may be influenced by adventitious carbon contamination, a result of brief sample exposure to the ambient environment during sample transfer fr om test chamber to XPS chamber. The two constituents from the carbon 1 s spectrum that are reaction products involving gas phase CO2 are the carbonates (CO3) and chemisorbed negatively charged carbon dioxide (CO2 -). Of the two species, the carbonates on the c opper wear track surfaces would seem the more plausible explanation for the lubricating behavior observed in humid CO2. A study by Wu et al. [83] concluded that CO2 was an effective gas phase lubricant for steel sliding c ontacts due to the formation of iron carbonate and/or iron bicarbon ate compounds on the sliding surfaces. Wu et al. [83] reported friction coe fficients of approximately 0.24 for steel-on-steel in CO2 environments. Transmission electron microscopy and energy dispersive X-ray spectroscopy analysis of worn single copper fibers (Figur e 3-19) from tribological testing in humid CO2 environments showed a surface la yer on the copper fibers that wa s primarily composed of carbon with some copper. This agreed well with X PS analysis which found the copper counterface wear track surface composition to be predomin antly carbonaceous species (amorphous carbon, CO2 -, and CO3). Deng et al. [63] extensively studied react ion film formation on polycrystalline copper surfaces at room temperature in the presence of CO2 and H2O using ambient pressure XPS, the results of which are summarized briefly. In 0.1 torr of CO2, a clean copper surface was shown to be active toward CO2. In the carbon 1 s spectrum, three carbonaceous species were identified 115

PAGE 116

(binding energies shown in pa renthesis): carbonate (289.3 eV), chemisorbed negatively charged CO2 (288.4 eV), and hydrocarbon (284.4 eV). Othe r than the hydrocarbon fragments, which were attributed to residual contamination, chemisorbed CO2 was most prevalent on the copper surface and was approximately 4 times as prev alent as the carbonate. From the oxygen 1s spectrum, the following surface species were id entified: carbonate (531.9 eV), chemisorbed negatively charged CO2 (531.4 eV), surface hydroxyl (530.8 eV), and chemisorbed oxygen (529.8 eV). No evidence was found for the formation of Cu2O in 0.1 torr of CO2. The addition of 0.1 torr H2O to 0.1 torr CO2 altered the surface ch emistry slightly. In addition to the species observed in 0.1 torr CO2, three other species were identified: formate (HCOO ), methoxy (O CH3), and adsorbed molecular water. The presence of the carbonate, CO2 -, and hydrocarbon species from Deng et al. [63] correlated well with results from XPS analysis of copper-on-copper slidin g films generated at high current densities in humid CO2 environments. In the present study of sliding electrical contacts, the hydrocarbon peak was assigned to 284.6 eV rather than 284 .4 eV [57]. Of the sliding film components (excluding CF2), the hydrocarbon was most prevalent, followed by CO2 -, and carbonate was least prevalent, similar to the trend noted by Deng et al. [63]. Formate (287.3 eV) and methoxy (285.2 eV) species [63] were not identified in the carbon 1 s spectra from the sliding films due to the inability to di stinguish between the two species; however, it is likely that both formate and methoxy contributed to the intensity, part icularly between 285 and 287.5 eV. Cu2O was present in all wear tracks but not detected on copper exposed to CO2 and water according to Deng et al. [63]. The presence of Cu2O in the wear tracks was attributed to the low concentrations of oxygen (50 100 ppm) in the operating environment reacting with freshly exposed material as w ear debris was generated. 116

PAGE 117

Copperthwaite et al. [62] proposed the reac tion shown in Equation 4-1 (also discussed by Deng et al. [63] and Browne et al. [84]) for the formation of carbonates on copper. CO2 + CO2 CO3(a) + CO(g) (4-1) For this reaction to occur, CO2 gas must first be adsorbed on to the copper surface and converted to CO2 -. The activation of CO2 to CO2 is believed to be the critical step. CO2 is unstable, and it has been shown to act as a pr ecursor for reactions involving CO2 on iron and palladium [62, 85-90]. The CO2 species can be stabilized through solvation with nearby CO2 molecules [62]. The result of this reaction is the formation of a carbonate species on the surface and carbon monoxide (CO) which desorbs from the surface. The carbonate can also be reduced through the reaction shown in Equation 4-2. CO3(a) C(a) + 3 O(a) (4-2) This reaction yields products of adsorbed ca rbon and oxygen. The carbon may contribute to the peak intensity observed at 284.6 eV, which had previously been attributed to hydrocarbon contamination [63]. The degr ee to which carbon produced thr ough Equation 4-2 influenced the intensity at 284.6 eV is not known. According to Deng et al. [63], in reaction films the chemisorbed oxygen remained on the copper surface as chemisorbed oxygen. In sliding films, the chemisorbed oxygen may react to form Cu2O. Deng et al. [63] also studied the surface chemistry of a Cu2O film in the presence of 0.1 torr CO2 at room temperature. They noted that under the given conditions, CO2 did not react on the Cu2O surface. Neither CO2 nor carbonate species were detected on the surface using XPS. In the case of the copper fiber brush sliding c ontacts, reactions shown in Equations 4-1 and 4-2 must have occurred in the wear track where clean copper adsorption sites were being generated. 117

PAGE 118

X-rays impinging on the sample surface and photoelectrons emitted from the surface in XPS can alter the surface chemistr y of the copper samples. Deng et al. [63] noted progressive formation of carbonates on a Cu2O surface after analyzing the sa me location multiple times. Browne et al. [84] also dete cted a chemisorbed species (CO2 -) on copper surfaces formed through X-ray induced chemical transitions. Th e electric fields and high current densities passing through the copper fiber sliding electrical contacts may be able to modify the tribofilm chemistry in a similar manner. Reactions involving gaseous CO2 and adsorbed water must also be considered. Scholer and Euteneuer [91] studied the corrosion of c opper pipes in deionized water, and found the corrosion rate was highly dependent on the amount of dissolved CO2 in the water. A fraction (~1%) of the dissolved CO2 reacted with water to form carbonic acid (H2CO3). Dissociation of carbonic acid created HCO3 ions and hydrogen ions (H+), which broke down oxide on the copper surface by reacting with oxyge n to form water. The pH of condensed water inside the environment chambers of the present study was measured to be in the range of 5 to 6, which is slightly acidic. Contac t resistance in humid CO2 environments was generally lower than the contact resistance in humid argon environments (Figure 3-14), which supported the proposed mechanism of Scholer and Euteneue r [91] for the dissolution of Cu2O in solutions of CO2 and H2O. Through the surface analysis and literature revi ew discussed, a basic model of the chemical nature of copper-on-copper sliding contacts in humid CO2 can be constructed. Figure 4-3 shows a schematic drawing of a microscale c opper-copper contact spot in humid CO2. The drawing is not to scale and is intended onl y to help visualize the composition of the sliding contact. The worn copper surfaces (both positive and ne gative surfaces) showed formation of Cu2O due to 118

PAGE 119

small concentrations of oxygen (~50 100 ppm) in the sliding environmen t. Surface species formed through reactions with the humid CO2 environment included carbonates and amorphous carbon. The formation of these reaction films on copper has been confirmed by Deng et al. [63], although the reactions were found to occur on copper rather than Cu2O surfaces. In high humidity environments, the sliding interface wa s composed of multiple monolayers of water, which contributed to lubrication a nd formation of reaction films. Vapor Phase Lubrication of Metal Sliding Electrical Contacts with 1-Pentanol Adsorption of 1-pentanol produces lubric ating films under sliding conditions which improve the wear resistance of silicon sliding contacts [58, 59]. Asay, Dugger, Ohlhausen, and Kim [59] have shown that the l ubricating film formed in the w ear track was composed of high molecular weight oligomeric products (long-chain hydrocarbons). This reaction film was formed as a direct result of sliding and not solely due to adsorption of alcohol molecules [59]. Figure 321 shows the improvement in wear resistance for a quartz-silicon sliding pa ir due to addition of 1-pentanol to the operating environment. After 1,000 cycles of sliding, the depth of the wear scar in the 1-pentanol environment is only several nanometers, and the average friction coefficient over the entire test is 0.14. The 1-pentanol saturated argon environment was selected for study with sliding electrical contacts because it does not c ontain components that will readily oxidize copper and it can produce thin layers of carbonaceous species. Ideally, this scenario would yield very low contact resistance for high current applications while minimizing friction and wear of the copper surfaces. However, results have indicated that this is not the case. For a copper-on-copper sliding pair at a contact pre ssure of 270 MPa, severe wear and adhesion between the sliding bodies, characteristic of cold we lding of a self-mated metal contact, occurred af ter only ~40 cycles of sliding in the 1-pentanol saturated argon environment. The presence of adsorbed 1119

PAGE 120

pentanol molecules initially aided in the protectio n of the native oxide on the copper. However, after breakthrough of the surface oxide, 1-pentanol adsorption failed to reduce friction and wear when the contact was predominantly copper slidi ng against copper. Using a copper fiber brush and a lower nominal pressure of 1.4 105 Pa, cold welding behavior was not observed in the 1pentanol environment, but the measured fricti on was high relative to measured friction in the humid CO2 environment. The lack of change in contact resistance when the environment was switched from 1-pentanol saturated argon to dry argon (Figure 3-25) suggested only minor amounts of 1-pentanol were adsorbed in the wear track. Longer tests, approximately 10,000 cycles or more, are needed to properly quantif y counterface wear in 1pentanol saturated argon and compare to wear in humid CO2. Polarity Effects on Friction and Contact Resistance The present study of copper fiber brushes in humid CO2 environments at low sliding velocities (0.01 m/s) showed that friction coefficient and contact resistance did not depend on the direction of current flow. Howe ver, at high sliding velocities (5 m/s), the negative brush contact resistance was generally higher than the positive brush contact resistance (Figure 3-6). After significant accumulation of damage to the brus hes and visible arcing at the positive brush interface, the positive brush cont act resistance was higher than th e negative brush resistance for the high sliding velocity test. Other studies of copper fiber brushes noted pol arity effects in measured voltage drops and friction coefficients; however, the results we re inconsistent, making it difficult to discern mechanisms responsible for this behavior. Argiba y et al. [54] observed higher contact resistance for the negative brush upon increa sing current from 120 to 180 A/cm2. In a study of copper fiber brushes in humidified CO2 environments, Reichner [48] found the positive brush had a higher voltage drop than the negative brush, except when both brushes ran in the same track, in which 120

PAGE 121

case the negative brush had the higher voltage dr op. Reichner [48] also examined copper fiber brushes in humidified CO2 sliding against a silver slip ri ng and found the negative brush voltage drop to be higher than the positive brush. In te rms of friction, Reichne r [48] reported for the copper-on-copper system higher friction coeffici ent for the negative brush compared to the positive brush; for the copper-on-silver sliding pa ir, friction coefficient did not show current direction dependence. Boyer, Noel, and Chabrerie [46] observed that positive fiber brush voltage drops were consistently higher than negative fiber brush voltage drops in a humid nitrogen environment. Lee [92] investigated copper/graphite coated co pper fibers as potenti al brush materials for sliding electrical contacts. Lee [92] observed in humid CO2 environments that the negative brush voltage drop was higher than the positive brush and postulated that the difference in voltage drop was due to the electric field aff ecting adsorption of polar molecules (such as H2O) on the slip ring surface. In a study of silver-coated carbon fiber brushes by McNab and Gass [42], positive brush voltage drops were consistently higher than the negative brush voltage drops for the same current in nitrogen, helium, and air environments, but voltage drops for both brushes were approximately the same in CO2 environments. Studies of sliding electrical co ntacts involving monolithic brushes, typically metal-graphite composites, have also reported polarity effects where positive br ush voltage drops were higher than negative brush voltage drops [23, 33, 39]. Ba sed on the results of the present study and the lack of agreement among similar studies from literature, there was no apparent correlation between brush polarity and either friction or contact resistance. System to system variations in test geometry, brush design, and brush holder desi gn may increase the variability of the reported results. 121

PAGE 122

Summary of Brush Electrical Losses The Ohmic losses per brush per ampere c onducted were calculate d for the different brush/lubricant pairs investigated and plotte d in Figure 4-4. Ohmic losses normalized by the current conducted through the br ush contact were calculated a ccording to Equation 1-4. Error bars represent the calculated combined uncerta inty in the Ohmic loss for each system based on the standard deviations in recorded current and resistance measurements. Where applicable, values for positive and negative brushes were averaged. Normalization of brush losses by current conducted allowed different test geometries with different nominal currents to be directly compared. The highest electrical losses occurred with the monolithic silver brushes and decoupled graphite lubrication. The combination of few co ntact spots under the monolithic brush and thick graphite transfer films produced brush electrical losses of appr oximately 0.71 W/A. Electrical losses with the graphite lubricant were notab ly reduced by replacing the monolithic silver brushes with copper fiber brushes under the same loading conditions. Graphite-lubricated copper fiber brushes were only briefly studied at relatively low current densities (20 A/cm2) with minimal success; further testing is needed to de termine if this system could operate at higher current densities. Lubrication of copper fiber brush contacts with PTFE and PTFE/In composites also produced thick transfer films that resulted in high electrical losses. The addition of indium to PTFE did not significantly reduce electrical losses. Vapor phase lubrication of copper fiber brus h sliding electrical contacts delivered the lowest electrical losses. In water-saturated CO2 environments, higher losses were calculated for high sliding speeds (0.07 W/A) co mpared to low sliding speeds (0.03 W/A), partially due to the discrepancy between positive and negative brushes at high sliding speeds. For the data set shown in Figure 3-6, electrical loss for the positive brush at high speeds was 0.04 W/A while 122

PAGE 123

electrical loss for the negative brush at high sp eeds was 0.10 W/A. The electrical loss for the single copper fiber test in humid CO2 (setup shown in Figure 2-14) was calculated to be 0.03 W/A (not displayed in Figure 4-4), which is eq uivalent to the other values obtained for low sliding speeds in humid CO2. Low speed testing of vapor phase lubrication using 1-pentanol also generated low electrical losses for copper-on-copper sliding. Further testing in 1-pentanol saturated argon environments at high slid ing velocities is still needed. Minimization of electrical loss per brush is cr itical in high current motor and generator applications which make use of thousands of brush contacts. The most promising method to achieve this result is through vapor phase lubr ication of a filamentous metal brush sliding electrical contact. As shown in Equation 1-2, the total resist ance for metal fiber brushes is dominated by the film resistance. The lubricat ing films generated from the vapor phase are thin and uniform, and the films can be continually replenished. Investigations of other vapor phase additives could lead to reduced electrical and m echanical losses for high current brush contacts. Rotor Wear The study of brush wear is commonly emphasi zed over rotor wear b ecause worn brushes require replacement long before the rotor. An ex amination of sliding elec trical contact literature revealed a lack of quantitative wear measurements of the opposing surfaces to the brushes. The present study utilized non-contacting scanning white light interferometry measurements to evaluate a copper rotor surface in humid CO2 environments. In the low sliding velocity (0.01 m/s) study of environmental effects on copper fiber brushes, a polished copper counterface was used to simulate the copper rotor surface. Wear rates were initially high er without current than at 180 A/cm2. These higher wear rates were likely caused by slight misalignments of the brush and counterface during the run-in period. Re sults obtained from sliding under 180 A/cm2 in humid CO2 environments showed that rotor wear was not dependent on current flow direction at 123

PAGE 124

low sliding velocities. Changes in the wear track topography, quantified by the average surface roughness, were also inde pendent of the current flow direction. These results differed from the observed wear behavior of the copper rotor surface at higher sliding velocities reported by Argibay et al [54]. After an exte nded test in a humid CO2 environment for 100 hours at 2.5 m/s, the copper rotor was examined using a scanning white light interferometer. The negative rotor surf ace (under the positive brush) and positive rotor surface (under the negative brush) had average roughness values of 1.7 and 4.3 m, respectively. For comparison, an area of the rotor that wa s not affected by brush wear had an average roughness of 1.3 m. Vibrations can cause fibers and even the en tire brush to bounce ove r the rotor surface, breaking the electrical pathway and increasi ng the propensity for ar cing. Arcing has a welldefined material transfer polarity dependence, as outlined by Slade [81]. In metal vapor arcs, which are characterized by small contact gaps electrons impinging on the positive electrode vaporize and ionize metal atoms. The positive me tal ions are accelerated toward the negative electrode where they generally stick to the surf ace. Experimental studies of arcing in opening and closing electrical contacts us ing radioactive tracer elements m easured net material gain for the negative electr ode [79, 80]. The difference in topography between the posi tive and negative rotor surfaces may have been caused by the direction of current flow. At the negative br ush contact, electrons leave the brush and impinge on the positive rotor surface. The stream of electrons ablates atoms from the positive surface, resulting in a net loss of materi al. At the positive brush contact, electrons leave the negative rotor surface and im pinge on the positive brush. T hus, the negative rotor surface 124

PAGE 125

would be subject to less damage from impinging electrons than th e positive rotor surface, and the negative rotor surface would maintain a lower surface roughness. Betz [31] also observed distinct differences in copper rotor surface roughness under copper-graphite composite brushes of different polarities. At current densities below approximately 5 A/cm2, the roughness of the positive and nega tive rotor surfaces were similar. At 10 A/cm2 and above, the positive rotor surface roughness was greater th an the negative surface roughness. Betz [31] proposed that the roughness difference was due to a third body wear mechanism involving oxidized copper debris. Theories on Brush Wear Mechanisms in Metal Fiber Brush Sliding Contacts Several studies of high current density copper fiber brushes reported brush wear anisotropy where the positive brush wore at a higher rate than the negative brush [46, 50, 54]. Many studies of high current density electrical brush materials containing graphite have also reported higher wear rates for positive brushes than negative brushes [6, 23, 24, 34, 42]. A smaller number of studies reported higher wear fo r the negative brush or no dependence of brush wear on polarity [31, 49, 93]. McNab [7], in a review of high power brush research, showed current direction did not have a distinguishable effect on wear of metal-graphite brush materials. The general trend of positive brushes wearing at a higher rate than negative brushes at high sliding velocities was confirmed in the pres ent study. Throughout the literature on electrical brushes, a number of theories have been pres ented regarding brush w ear and the effects of polarity on brush wear [24, 31, 34, 38, 94]. The proposed wear mechanisms which apply to metal fiber brushes were broadly grouped into three categories: arcing and electromigration, oxidation, and mechanical. The support for each of the various theories was briefly discussed in terms of results from literature and results from the previous chapter. Shortcomings of each theory were analyzed as well. 125

PAGE 126

Arcing and electromigration As discussed by Slade [81], the polarity depende nce of arcing-based material removal is largely based on the size of the contact gap and the species ionized within the gap. For metal vapor arcs (small gaps), electr on bombardment vaporizes material from the positive electrode surface. Electrons then ionize metal vapor atoms to sustain the arc. The metal ions travel toward the negative electrode and deposit on the surface, resulting in a net material gain for the negative electrode. Metal vapor ar cs can form in air at standard temp erature and pressure if the contact gap is less than 5 to 10 electron mean free paths (less than 10 m) [95]. For gaseous arcs, characterized by larger contact gaps, ambient ga s atoms in the contact region become ionized. The gas ions are then accelerated toward the negative electrode surface where material is removed rather than deposited, resulting in a net material loss for the negative electrode [81]. Higher wear rates for positive brushes compared to negative brushes have been documented throughout the literatu re for various electric br ush materials and operating environments [6, 23, 24, 34, 42, 46, 50]. Polarity e ffects on brush wear clea rly are not dependent on material pairs or environment. The affinity for high positive brush wear and material transfer from positive to negative suggests that if arcing is responsible, a metal vapor arc (small gap) rather than a gaseous arc (large gap) was likely formed at the brush-rotor interface. An example of intentional material removal through combined arcing and mechanical means is brush electrodischarge mechanical machining, which uses current flow through a rotating metal fiber brush to remove material from a positively biased workpiece [96]. In the present study of sliding el ectrical contact pairs at high sliding velocities, Figure 3-6 shows positive brush wear rate was higher than ne gative brush wear rate for copper fiber brushes at 180 A/cm2 in humid CO2. Visible arcing was observed at the positive brush interface at the 126

PAGE 127

very end of the test. The micrographs of Fi gure 3-7 and profilometry images of Figure 3-8 reveal the damage inflicted upon the positive brush fibers as a result of intense arcing. Material transfer from positive to negative su rfaces was noted with the monolithic silver brushes as well as the zinc-coated copper fiber br ushes, both of which displayed greater wear for the positive brush polarity. EDS spectra from Fi gure 3-2 illustrate the high concentration of copper transferred from the positive rotor surfac e to the negative brush, while only a trace amount of copper transferred from the negative ro tor surface to the posi tive brush. With zinccoated copper fiber brushes, the negative rotor surface (under the positive brush) was discolored and the positive rotor surface (under the negative brush) appeared normal. The discoloration on the negative rotor surface was gray, suggesting that zinc from the brush coating had been transferred to the rotor. Electromigration theory, as reviewed by Ho and Kwok [97], encompasses the effects of electric fields on diffusion in metals. Several studies of brush wear have mentioned electromigration as a possible reason for positive and negative brush wear anisotropy [77, 78]. In a study of silver-graphite electric brushes, Johnson [77] attributed asymmetry in contact resistance and asymmetry in area fractions of silv er on brush faces to electromigration of silver. Visible material transfer from positive to nega tive surfaces was also observed by Boyer, Noel, and Chabrerie [46] with copper fiber brushe s on a gold-plated copper rotor; copper was deposited on the rotor surface beneath positive brushes, and gold was removed from the rotor surface under negative brushes. However, the brush wear mechanism for copper-on-copper sliding proposed by Boyer, Noel, and Chabrerie [46] was related to ox idational wear and not electromigration. Johnson and Taylor [23], in thei r work with silver-graphite composite brushes, visually observed larger covera ge areas of silver on the nega tive brush and negative rotor 127

PAGE 128

surfaces relative to the positive brush and positiv e rotor surfaces. The pr eference for silver transfer to the negatively biased surfaces was attributed to the electric field influencing migration of positive silver ions to the negative surf aces. Similarly, Brown, Kuhlmann-Wilsdorf, and Jesser [50] stated that polarity has a distinct effect on brush wear based on observations of positive brush material transfer to the negative ro tor surface and the fact that positively charged metal ions are attracted to th e negatively charged surfaces. Ex situ analysis of sliding el ectrical contact surf aces has generated little physical evidence to support arcing. Roughening of the negative roto r surface has been obser ved in some studies of sliding electrical contacts [ 31, 54]. With respect to brush wear, Kuhlmann-Wilsdorf [51] reported no difference in metal fi ber brush performance between tests with alternating current and direct current. Worn brush fibers generally appear free of craters or resolidified metal droplets, similar to the negative brush fibers in Figure 3-7 A. The influence of mechanical wear may obscure features characteristic of arcing in switching contacts. Furt her testing at both low and high sliding velocities with dissimilar metals under high current densities is needed to better understand wear mechanisms related to arcing and ion migration. Oxidation XPS analysis of copper surfaces after sliding in humid CO2 showed that oxidized copper (Cu2O) was present in the wear tracks. Oxidation of a metal surface is dependent on the electric field established across the oxide layer. Since oxidation occurs by the diffusion of ions through the oxide scale, the magnitude and polarity of the electric field can influence ion transport. At the positive brush surface, oxidation may be enha nced due to the externally applied field. Diffusion of positively charged metal ions toward the oxide/gas interface is aided by the external field at the positive brush. At the negative brus h surface, oxidation may be inhibited because the electric field opposes the motion of the ions to wards the oxide/gas interface. The same holds 128

PAGE 129

true for the positive and negative rotor surfaces. In testing with metal-plated carbon fiber brushes, Kendall, McNab, and Wilkin [6] correlate d preferential oxidation of the positive brush to higher wear of the positive brush. Testing performed in a reducing environment (4% H2, 96% N2) reduced the difference between positive and nega tive brush wear rates, but the positive brush wear rate was still a factor of four highe r than the negative brush wear rate [6]. In a study of metal fiber brushes in humid e nvironments, Boyer, Noel and Chabrerie [46] consistently observed higher voltage drops for the positive brush af ter reversing current direction multiple times. It was hypothesized that higher voltage drops resulted from enhancement of oxidation at the positive brush fibe r tips due to the electric fiel d, and brush wear rate, which was also higher for the positive brush than negative brush, was directly related to oxide formation rate. Boyer, Noel, and Chabrerie [46] formul ated a modified Cabrer a-Mott [98] model to account for the influence of the external electric field on oxide formati on. Using the model under the assumption that brush wear rate was equi valent to oxide growth rate, they calculated the average oxide thickness at the in terface to be 1 to 2 monolayers [46]. Results from the present study provide little su pport for enhanced oxidation at the positive brush surface. Current-voltage sweeps of a copper fi ber sliding in a humid CO2 environment (Figure 3-16) showed that contact resistance, whic h is directly related to film thickness, was not dependent on polarity. Low sliding velocity st udies, as shown in Figure 3-11, confirmed that contact resistance was not polarity dependent for copper-on-copper. XPS analysis of wear tracks from the low sliding velocity study found the relative amount of Cu2O on the positive wear track to be lower than both the negative wear track and the no current wear track. High sliding velocity studies of copper-on-copper in humid CO2 found higher contact resistance for the negative brush. 129

PAGE 130

As described in the previous section titled Polarity Eff ects on Contact Resistance and Friction, copper fiber brush sliding electrical co ntacts in humid environments typically do not exhibit voltage drops that are dependent on curren t flow direction [8, 48, 54]. Of course, testing environments can vary wildly among a group of independent studies. Different humidities and different cooling systems can greatly affect wate r adsorption at the brush contacts. The present study actively monitored oxygen levels in the testing environments (generally between 50 and 100 ppm of oxygen), while most controlled-environmen t studies of sliding el ectrical contacts did not report this information. Meticulous contro l and monitoring of operating environments is necessary to understand the relationship betw een surface phenomena such as oxidation and overall brush performance. Mechanical wear Assuming operating conditions for all brushe s from the same test are equivalent, mechanical wear alone can not account for the di fferences in wear between positive and negative brushes. However, mechanical wear does feature prominently in the wear behavior of metal fiber brushes for high current applications, as discussed by Brown, Kuhlmann-Wilsdorf, and Jesser [50] as well as other sources [12, 99]. The proposed wear mechanism for metal fiber brushes sliding on metal slip rings involves localized adhesive wear at contact spots through interlocking of microscale roughne ss and shearing of wear partic les. The distinctive smooth appearance of worn fiber surfaces and the collec tion of wear particles along the trailing fiber edges were cited as evidence for the adhesive/shearing mechanism. The worn copper fibers in Figure 3-17 exhi bit characteristics c onsistent with the adhesive/shearing wear mechanis m. All three worn fiber surfaces had a smooth appearance. Wear particles along the trailing e dge of the fibers on the verge of detaching appeared to be severely mechanically deformed from shearing. The micrograph of negative brush fibers from 130

PAGE 131

the high sliding velocity test in Figure 3-7 A shows similar signs of shearing at the fiber surfaces. Micrographs of the copper counter face wear tracks from low slid ing velocity testing in humid CO2 (Figure 3-13) further support the adhesive wear mechanism. Debris particles were sheared from the fiber surfaces, and in some cases the ad hesive forces were strong enough to prevent the debris particles from being pushed to the edges of the wear track. As expected, the morphology of the positive and negative wear track surfaces appeared the same. The combination of a mechanical shearing wear mechanism and an el ectrical-based wear mechanism likely accounted for the disparity between positive and negative brush wear rates. 131

PAGE 132

Figure 4-1. SEM image (upper left) of spherical and rod-like structures found on the surface of a worn copper fiber brush after testing in humid CO2 environment. EDS dot maps show the spatially resolved distribution of copper (upper right), oxygen (bottom left), and zinc (bottom right). The rods and s pheres are primarily composed of oxygen and zinc. 132

PAGE 133

Figure 4-2. Water adsorption study on Cu2O using ambient pressure XPS (adapted from Salmeron [82]). Water layer thickness (blu e; left axis) plotted as a function of relative humidity (RH). At very low relative humidity, the Cu2O surface became saturated with hydroxyl groups. Figure 4-3. Schematic drawing of surface films present on a copper-copper sliding contact in humid CO2 (not drawn to scale). Adsorbed water, as well as amorphous carbon, carbonates, and cuprous oxide, were identifi ed in sliding films formed on positive and negative copper surfaces in humid CO2. 133

PAGE 134

Figure 4-4. Summary of calculate d brush Ohmic losses per ampere conducted for various brush and lubricant pairs. Values are shown fo r both high sliding velo city (high) and low sliding velocity (low) testi ng with copper fiber brushes. Error bars represent the combined uncertainty in the calculated losse s. Uncertainty in resistance and current were estimated from the standard deviation of the measured values. 134

PAGE 135

CHAPTER 5 SUMMARY AND CONCLUSIONS Sliding electrical contacts present a unique design challenge involving tradeoffs between mechanical losses and electrical losses based on ma terial and environmental selection. Electrical losses dominate high current sliding electrical contacts for motor and generator applications. Sacrifices to the lubrication scheme must be ma de to improve the electrical efficiency of the system. Decoupled graphite lubrication of mo nolithic silver brushes proved effective at maintaining low friction (~0.2) a nd low wear (less than 2 10-11 m/m) in ambient air environments at 40 A/cm2. Thick graphite films (~ 1 to 5 m), while detrimental to electrical conductivity, were effective at incr easing brush lifetime. This met hod appeared to be best suited for low current applications. Brush wear incr eased significantly at a current density of 200 A/cm2, and brush electrical losses were significantly higher than electrical losses with metal fiber brushes. Use of polytetrafluoroethylene (PTFE) and PTFE-indium solid lubricants with copper fiber brushes reduced electrical lo sses at the expense of brush w ear. Reductions of brush load could be used to decrease brush wear if increa ses in electrical losses could be tolerated. Copper fiber brushes are compliant, have low bulk resistivity, and have a large number of independent contact points. These properties ma ke copper fiber brushes model current collectors for high current applications. The use of high humidity carbon dioxide operating environments with copper fiber brushes produced the best co mbination of low friction, low wear, and low electrical losses. Copper brush wear rates in humid carbon dioxide at 180 A/cm2 were below 3 10-11 m/m, and electrical losses (0.07 W/A per brush) were lower than electrical losses with solid lubricants at significantly higher brush pressures. The chemistry of the surface films formed on co pper sliding bodies played an important role in the tribological perfor mance of the system. Adsorbed water films acted as a boundary 135

PAGE 136

lubricant and reduced adhesion. Cooling the copper sliding surfaces to temperatures at or below the ambient temperature in a high humidity envi ronment minimized mechanical and electrical losses. X-ray photoelectron spectroscopy identi fied a number of carbonaceous species on the worn surfaces of copper samples after sliding in humid carbon dioxide Reactions involving chemisorbed carbon dioxide created carbonate spec ies on the worn surfaces. Carbon detected on the surface may have resulted from decomposition of the carbonate species but may also have been a contaminant from brief exposure to th e ambient environment. Transmission electron microscopy and energy dispersive X-ray spectros copy analysis of worn copper fiber surfaces confirmed the presence of a carbon-rich surface layer. Cuprous oxide (Cu2O) was found on worn copper surfaces as well. Current-voltage sweeps of copper-copper contacts in humid carbon dioxide suggested that the oxide had little effect on contact resistance. Vapor phase lubrication with 1pentanol saturated argon envir onments was investigated as an alternate means of producing thin, carbonaceou s lubricating films. Although 1-pentanol was shown to effectively lower fricti on and reduce wear in a quartz-sili con sliding contact, it did not offer any improvements over water-saturated carbon dioxide in te rms of copper-copper lubrication. Both frictional loss es and electrical losses were hi gher in 1-pentanol saturated argon compared to humid carbon dioxide for low sliding speed testing of a copper fiber brush contact. Regardless of brush material, lubricant, and operating environment, positive brushes generally wore at a higher rate than negative brushes. Worn brushes and counterfaces were examined using electron microscopy to analyze m echanisms related to brush wear anisotropy. Worn copper fiber surfaces showed signs of plastic deformation through shearing. Cross sections of fiber surfaces created by focused ion beam milling also showed that the deformation extended approximately 1 m below the surface. Physical evidence suggested an adhesive 136

PAGE 137

wear mechanism for copper fiber brush slidi ng electrical contacts in humid carbon dioxide environments. However, mechanical mechanis ms failed to account for the polarity dependence of brush wear. Electrically induced wear, such as micro-scale arcing, may account for the disparity between positive and negative brush wear, but supporting evidence for arcing was lacking. Damage associated with arcing was rare ly observed on worn brush or rotor surfaces, yet visual observations indicated a net transfer of material from positive to negative surfaces. In testing with monolithic silver br ushes and graphite, a significant amount of copper from the rotor was detected on the negative brush surface using spectroscopic techniques. Moreover, zinc from zinc-coated copper fiber brushes was visually observed on the rotor surface under the positive brush while the rotor surface under the negative br ush appeared free of zinc. Micrographs of copper fiber brushes after thousands of kilomete rs of sliding showed substantial amounts of debris compacted between the fibers. Accumulated debris inhibited the fibers from acting independently, decreasing the ability of the brush to follow the rotor and increasing susceptibility of the brush contact to arcing. Future studies of metal fiber brush wear mechanisms should utilize dissimilar material pairs to examine the effects of polarity on material transfer. 137

PAGE 138

LIST OF REFERENCES [1] M. Brennan, Z. Eliezer, W. F. Weldon, H. G. Rylander, H. H. Woodson, The testing of sliding electrical contacts for homopolar generators, IEEE Transactions on Components, Hybrids, and Manufacturing Technology CHMT-2 (1979) 111-115. [2] W. Elger, Development of metal fiber el ectrical brushes for 500 kW SSMG sets, Naval Engineers Journal 117 (2005) 37-44. [3] D. U. Gubser, Superconducting homopolar motor and conductor development, JOM 48 (1996) 30-34. [4] D. U. Gubser, Naval applications of SC magnet systems, Physica C 341-348 (2000) 25252528. [5] D. U. Gubser, Superconducting motors and generators for naval a pplications, Physica C 392-396 (2003) 1192-1195. [6] P. W. Kendall, I. R. McNab, G. A. Wilk in, Recent developments in current collection, Physics in Technology (1975) 117-126. [7] I. McNab, Pulsed high-power brush research, IEEE Transactions on Components Hybrids and Manufacturing Technology 1 (1978) 30-35. [8] P. Reichner, Metallic brushes for extreme high-current applications IEEE Transactions on Components, Hybrids, and Manufact uring Technology 3 (1980) 21-26. [9] R. Soulen, D. Gubser, T. Francavilla, N. Sondergaard, J. Walters, J. Chafe, Recent developments for ship propulsion using high temperature superconductors, Industrial Ceramics 21 (2001) 103-105. [10] R. Thome, W. Creedon, M. Reed, E. Bo wles, K. Schaubel, Homopolar motor technology development, Proceedings of the IEEE Po wer Engineering Society Transmission and Distribution Conference 1 (2002) 260-264. [11] W. E. Yohe, W. A. Nystrom, Illustrative modern brush applications in Electrical Contacts: Principles and Applications, ed. P. G. Slade, Marcel Dekker, New York, 1999, p.873-883. [12] D. Kuhlmann-Wilsdorf, Metal fiber brushes in Electrical Contacts: Principles and Applications, ed. P. G. Slade, Marcel Dekker, New York, 1999, p.943-1017. [13] W. H. Bragg, An Introduction to Crystal Analysis, Bell, London, 1928, p.64. [14] B. K. Yen, B. E. Schwickert, M. F. Toney, Origin of the low friction behavior in graphite investigated by surface X-ra y diffraction, Applied Physics Letters 84 (2004) 4702-4704. [15] R. H. Savage, Graphite lubrication, Journal of Applied Physics 19 (1948) 1-10. 138

PAGE 139

[16] E. I. Shobert, Sliding electrical contacts (graphitic type lubrication) in Electrical Contacts: Principles and Applications, ed. P. G. Slade, Marcel Dekker, New York, 1999, p.839-872. [17] W. Campbell, Lubrication of electrica l contacts, IEEE Transactions on Components Hybrids and Manufacturing Technology 1 (1978) 4-16. [18] D. Ramadanoff, S. W. Glass, High-altitude brush problem, Transactions of the American Institute of Electrical Engineers 63 (1944) 825. [19] R. H. Savage, D. L. Schaefer, Vapor lubri cation of graphite slidi ng contacts, Journal of Applied Physics 27 (1956) 136-138. [20] J. L. Johnson, J. L. McKinney, Electrical-power brushes for dry inert-gas atmospheres, IEEE Transactions on Parts, Materials and Packaging 7 (1971) 62-69. [21] E. I. Shobert, Carbon Brushes, Chemical Publishing Co., New York, 1965, p.127. [22] J. Johnson, L. Moberly, Hi gh Current Brushes .1. Effect of Brush and Ring Materials, IEEE Transactions on Components Hybrid s and Manufacturing Technology 1 (1978) 3640. [23] J. Johnson, O. Taylor, High-Current Brushes .4. Machin e Environment Tests, IEEE Transactions on Components Hybrids a nd Manufacturing Technology 3 (1980) 31-41. [24] P. Lee, J. Johnson, High-Current Brushes .2. Effects of Gases and Hydrocarbon Vapors, IEEE Transactions on Components Hybrid s and Manufacturing Technology 1 (1978) 4045. [25] I. McNab, J. Johnson, High-Current Brus hes .3. Performance Evaluation for Sintered Silver-Graphite Grades, IEEE Transactions on Components Hybrids and Manufacturing Technology 2 (1979) 84-89. [26] B. K. Yen, Influence of water vapor a nd oxygen on the tribology of carbon materials with sp2 valence configurat ion, Wear 192 (1996) 208-215. [27] G. W. Rowe, Frictional behavior of bor on nitride and graphite, Wear 3 (1960) 274-285. [28] R. D. Arnell, D. G. Teer, Lattice parameters of graphite in relati on to friction and wear, Nature 218 (1968) 1155. [29] R. F. Deacon, J. F. Goodman, Lubricati on by lamellar solids, Proceedings of the Royal Society of London, Series A, Mathematical and Physical Sciences 243 (1958) 464-482. [30] H. Zaidi, F. Robert, D. Paulmier, Influence of adsorbed gases on the surface energy of graphite: consequences on the friction behavior, Thin Solid Films 264 (1995) 46-51. [31] D. Betz, Relationship between contact re sistance and wear in sliding contacts, IEEE Transactions on Parts, Hybrids, and Packaging 10 (1974) 32-37. 139

PAGE 140

[32] J. M. Casstevens, H. G. Rylander, Z. E liezer, Friction and wear characteristics of powder metallurgy copper-graphite brushes at hi gh sliding speeds, Wear 49 (1978) 169-178. [33] S. Dillich, D. Kuhlmann-Wilsdorf, Effects of surface films on the performance of silvergraphite (75 wt% Ag, 25 wt% C) electric br ushes, IEEE Transactions on Components, Hybrids, and Manufacturing Technology 3 (1980) 37-41. [34] Y. Feng, M. Zhang, Y. Xu, Effect of the el ectric current on the fric tion and wear properties of the CNT-Ag-G composite s, Carbon 43 (2005) 2685-2692. [35] P. K. Lee, High-current brush material de velopment, Part I: Sintered metal-coated graphite, IEEE Transactions on Components, Hybrids, and Manufacturing Technology CHMT-3 (1980) 4-8. [36] H. Liu, R. Carnes, J. Gully, Measurement a nd prediction of brush interface temperature at sliding electrical contacts, Proceedings of the Thirty-Eighth IEEE Holm Conference on Electrical Contact s (1992) 143-148. [37] E. Rabinowicz, P. Chan, Wear of silver-g raphite brushes against va rious ring materials at high-current densities, IEEE Transactions on Components, Hybrids, and Manufacturing Technology CHMT-3 (1980) 288-291. [38] J. Schreurs, J. Johnson, I. McNab, HighCurrent Brushes .6. Eval uation of Slip Ring Surface-Films, IEEE Transactions on Components Hybrids and Manufacturing Technology 3 (1980) 83-88. [39] R. Slepian, High-Current Brushes .5. Subdivided Monolithic Brushes at Very HighCurrent Levels, IEEE Transactions on Com ponents Hybrids Transactions on Components Hybrids and Manufacturing Technology 3 (1980) 56-62. [40] O. S. Taylor, P. Reichner, Mechanical lo ad aspects of high-curre nt brush system design, IEEE Transactions on Components, Hybrid s, and Manufacturing Technology CHMT-2 (1979) 95-99. [41] Y. Watanabe, Sliding contact characteris tics between self-lubricating composite materials and copper, IEEE Transactions on Component s, Hybrids, and Manufacturing Technology 16 (1993) 442-448. [42] I. McNab, W. Gass, High-current density car bon fiber brush experiments in humidified air and helium, IEEE Transactions on Component s, Hybrids, and Manufacturing Technology 3 (1980) 26-30. [43] D. Maribo, N. A. Sondergaard, Further studies of a low-melting point alloy used in a liquid metal current collector, IEEE Transactions on Components, Hybrid s, and Manufacturing Technology CHMT-10 (1987) 452-455. [44] M. Superczynski, D. Waltman, Homopolar motor with high temperature superconductor field windings, IEEE Transactions on Superconductivity 7 (1997) 513-518. 140

PAGE 141

[45] G. Talmage, J. S. Walker, S. H. Brown, N. A. Sondergaard, H. Branover, S. Sukoriansky, Liquid-metal flows in sliding electrical contact s with arbitrary magnetic-field orientations, Physics of Fluids A 3 (1991) 1665. [46] L. Boyer, S. Noel, J. P. Chabrerie, Electr ochemically activated wear of metal fibre brushes, Wear 116 (1987) 43-57. [47] D. Kuhlmann-Wilsdorf, Gold fibre brushe s their promise for future high-technology applications, Gold Bulletin 16 (1983) 12-20. [48] P. Reichner, High current tests of metal fiber brushes, IEEE Transactions on Components, Hybrids, and Manufacturing Technology 4 (1981) 2-4. [49] R. Holm, Electric Contacts, Theory and Applications, Spring er-Verlag, Berlin, 1976, [50] L. Brown, D. Kuhlmann-Wilsdorf, W. Jesser, Testing and evaluation of metal fiber brush operation on slip rings and commutators IEEE Transactions on Components and Packaging Technologies 31 (2008) 485-494. [51] D. Kuhlmann-Wilsdorf, D. Alley, Co mmutation with metal fiber brushes, IEEE Transactions on Components, Hybrids, a nd Manufacturing Technology 12 (1989) 246-253. [52] D. Kuhlmann-Wilsdorf, What role for contac t spots and dislocations in friction and wear, Wear 200 (1996) 8-29. [53] J. A. Bares, N. Argibay, P. L. Dickrell, G. R. Bourne, D. L. Burris, J. C. Ziegert, W. G. Sawyer, In situ graphite lubrication of metallic sliding electrical contacts (Article in Press), Wear (2009) doi:10.1016/j.wear.2009.1003.1024 [54] N. Argibay, J. A. Bares, W. G. Sawyer, Asymmetric wear behavior of self-mated copper fiber brush and slip ring slid ing electrical contacts in a humid carbon dioxide environment (Article in Review), Wear (2009) [55] D. R. Gaskell, Introduction to the Thermodynamics of Materials: Third Edition, Taylor & Francis, Philadelphia, 1995, p.568. [56] J. A. Bares, N. Argibay, N. Mauntler, G. J. Dudder, S. S. Perry, G. R. Bourne, W. G. Sawyer, High current density copper-on-copper sliding electrical co ntacts at low sliding velocities, Wear 267 (2009) 417-424. [57] J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, Physical Elec tronics USA, Inc., Chanhassen, MN, 1995, p.261. [58] D. B. Asay, M. T. Dugger, S. H. Ki m, In-situ vapor-phase lubrication of MEMS, Tribology Letters 29 (2008) 67-74. 141

PAGE 142

[59] D. B. Asay, M. T. Dugger, J. A. Ohlh ausen, S. H. Kim, Macroto nanoscale wear prevention via molecular adsorp tion, Langmuir 24 (2008) 155-159. [60] L. Boyer, Etude du mode de rpartition du courant dans un contact lectrique glissant multifilamentaire, Revue de Physique Applique, Paris 20 (1985) 183-190. [61] P. G. Slade, Electrical Contacts: Principl es and Applications, Marcel Dekker, New York, 1999, p.1022. [62] R. G. Copperthwaite, P. R. Davies, M. A. Morris, M. W. Roberts, R. A. Ryder, The reactive chemisorption of carbon dioxide at magnesium and copper surfaces at low temperature, Catalysis Letters 1 (1988) 11-20. [63] X. Deng, A. Verdaguer, T. Herranz, C. Weis, H. Bluhm, M. Salmeron, Surface chemistry of Cu in the presence of CO2 and H2O, Langmuir 24 (2008) 9474-9478. [64] G. Moretti, G. Fierro, M. L. Jacono, P. Porta, Characterization of CuO-ZnO catalysts by Xray photoelectron spectroscopy: Precursors, calcined and reduced samples, Surface and Interface Analysis 14 (1989) 325-336. [65] H. Bluhm, M. Havecker, A. Knop-Gericke, E. Kleimenov, R. Schlogl D. Teschner, V. I. Bukhtiyarov, D. F. Ogletree, M. Salmeron, Methanol oxidation on a copper catalyst investigated using in situ x-ray photoelectr on spectroscopy, Journal of Physical Chemistry B 108 (2004) 14340-14347. [66] S. Gnther, L. Zhou, M. Hvecker, A. Knop-Gericke, E. Kleimenov, R. Schlgl, R. Imbihl, Adsorbate coverages and surface reactivity in me thanol oxidation over Cu(110): An in situ photoelectron spectroscopy study, The Jour nal of Chemical Physics 125 (2006) 114709. [67] S. Poulston, P. M. Parlett, P. Stone, M. Bowker, Surface oxidation and reduction of CuO and Cu2O studied using XPS and XAES, Surf ace and Interface Analysis 24 (1996) 811820. [68] H. Hosaka, H. Kuwano, K. Yanagisawa, El ectromagnetic relays: concepts and fundamental characteristics, Sensors a nd Actuators A 40 (1994) 41-47. [69] B. L. Pruitt, W.-T. Park, T. W. Kenny, Measurement system for low force and small displacement contacts, Journal of Microe lectricalmechanical Systems 13 (2004) 220-229. [70] J. Schimkat, Contact measurements providi ng basic design data for microrelay actuators, Sensors and Actuator s A 73 (1999) 138-143. [71] A. Parretta, M. K. Jayaraj, A. D. Nocera, S. Loreti, L. Quercia, A. Agati, Electrical and optical properties of copper oxide films prepared by reactive RF magnetron sputtering, Physica Status Solidi (A) 155 (1996) 399-404. [72] J. K. Lancaster, I. W. Stanley, The e ffect of current on the friction of carbon brush materials, British Journal of Applied Physics 15 (1964) 29-43. 142

PAGE 143

[73] A. Senouci, J. Frene, H. Zaidi, Wear mechanism in graphite-c opper electrical sliding contact, Wear 225-229 (1999) 949-953. [74] D. Paulmier, H. Zaidi, H. Salliot, E. Hounkponou, Effects of electr ical current and its polarity on the surfaces of copper and carbon in contact, Surface Science 251/252 (1991) 769-772. [75] H. Prashad, Investigation of damaged ro lling-element bearings and deterioration of lubricants under the influence of el ectric fields, W ear 176 (1994) 151-161. [76] H. Prashad, Tribology in electrical environments, Lubrication Science 13 (2001) 359-369. [77] J. L. Johnson, Sliding monolithic brush sy stems for large currents, IEEE Transactions on Components, Hybrids, and Manufactur ing Technology CHMT-10 (1987) 3-17. [78] D. Kuhlmann-Wilsdorf, Uses of theory in the design of sliding electrical contacts, Proceedings of the Thirty-Seventh IEEE Ho lm Conference on Elect rical Contacts (1991) 1-24. [79] F. Lewellyn-Jones, Erosion process and assessment of electrical contact materials, Proceeding of the 9th Intern ational Conference on Electrical Contact Phenomena (ICECP) (1978) 249-254. [80] P. G. Slade, An investigation of curr ent interruption in low voltage circuits, IEEE Transactions on Parts, Hybrids, and Packaging PHP-5 (1969) 56-61. [81] P. G. Slade, The consequences of arcing in Electrical Contacts: Principles and Applications, ed. P. G. Slade, Marcel Dekker, New York, 1999, p.487-533. [82] M. Salmeron, Water growth on metals a nd oxides: binding, dissociation and role of hydroxyl groups, Lawrence Berkeley National Laboratory Paper LBNL-1076E (2008) 120. [83] X. Wu, P. Cong, H. Nanao, I. Minami, S. Mori, Tribological behaviors of 52100 steel in carbon dioxide atmosphere, Trib ology Letters 17 (2004) 925-930. [84] V. M. Browne, A. F. Carley, R. G. Coppe rthwaite, P. R. Davies, E. M. Moser, M. W. Roberts, Activation of carbon dioxide at bismuth, gold and copper surfaces, Applied Surface Science 47 (1991) 375-379. [85] H. Behner, W. Spiess, G. Wedler, D. Borgmann, Interaction of carbon dioxide with Fe(110), stepped Fe(100) and Fe(111), Surface Science 175 (1986) 276. [86] H.-J. Freund, H. Behner, B. Bartos, G. Wedler, H. Kuhlenbeck, M. Neumann, CO2 adsorption and reaction on Fe(111): an a ngle resolved photoemission (ARUPS) study, Surface Science 180 (1987) 550-564. 143

PAGE 144

[87] M. Pirner, R. Bauer, D. Borgmann, G. Wedler, CO2 adsorption and reaction on polycrystalline Fe films, Su rface Science 189-190 (1987) 147-160. [88] J. Wambach, G. Odorfer, H.-J. Freund, H. Kuhlenbeck, M. Neumann, Influence of alkali co-adsorption on the adsorption and reacti on of CO2 on Pd(111), Surface Science 209 (1989) 159-172. [89] G. Wedler, W. Kiessling, D. Borgma nn, Influence of coad sorption of oxygen on adsorption of CO2 on Fe(111), Vacuum 41 (1990) 93-95. [90] S. Wohlrab, D. Ehrlich, J. Wambach, H. Kuhlenbeck, H.-J. Freund, Promoter action of alkali in the activation of CO2 on Pd( 111): a HREELS case study, Surface Science 220 (1989) 243-252. [91] H. Schler, H. Euteneuer, Corrosion of copper by deionized cooling water, European Accelerator Conference (EPAC) (1988) 1067-1068. [92] P. K. Lee, High current brush material deve lopment, Part II: Metal-solid lubricant coated wires, IEEE Transactions on Components Hybrids, and Manufacturing Technology CHMT-4 (1981) 114-121. [93] A. Bouchoucha, S. Chekroud, D. Paulmier Influence of the electrical sliding speed on friction and wear processes in an electrical contact copper-stainless steel, Applied Surface Science 223 (2004) 330-342. [94] D. Paulmier, A. Bouchoucha, H. Zaidi, Wear mode of a copper wire during a sliding contact with copper/chrome steel: influence of oxygen, Vacuum 41 (1990) 2230-2232. [95] P. G. Slade, The arc and interruption in Electrical Contacts: Pr inciples and Applications, ed. P. G. Slade, Marcel Dekker, New York, 1999, p.433-486. [96] B. Nowicki, R. Pierzynowski, S. Spaldo, Th e superficial layer of parts machined by brush electro-discharge mechanical machining (B EDMM), Proceedings of the Institution of Mechanical Engineers: Part B, Journal of Engineering Manufacture 218 (2004) 9-15. [97] P. S. Ho, T. Kwok, Electromigration in metals, Reports on Progress in Physics 52 (1989) 301-348. [98] N. Cabrera, N. F. Mott, Theory of the Ox idation of Metals, Reports on Progress in Physics 12 (1949) 163-184. [99] Y. J. Chang, D. Kuhlmann-Wilsdorf, A case of wear particle formation through shearingoff at contact spots interl ocked through microroughness in adhesive wear, Wear 120 (1987) 175-197. 144

PAGE 145

BIOGRAPHICAL SKETCH Jason A. Bares was born in Belgium, Wiscons in, and graduated from the University of Wisconsin-Madison in 2005 with a Bachelor of Science degree in materials science and engineering. He then went on to pursue his Doctor of Philosophy in materials science and engineering at the University of Florida, graduating in 2009. 145