Citation
Determination of metallo-organic and particulate wear metals in lubricating oils associated with hybrid ceramic bearings by inductively coupled plasma mass spectrometry

Material Information

Title:
Determination of metallo-organic and particulate wear metals in lubricating oils associated with hybrid ceramic bearings by inductively coupled plasma mass spectrometry
Creator:
Russell, Robin Ann, 1970-
Publication Date:
Language:
English
Physical Description:
xi, 184 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Digestion ( jstor )
Ions ( jstor )
Lubricating oils ( jstor )
Mass spectra ( jstor )
Mass spectrometers ( jstor )
Microwaves ( jstor )
Nebulizers and vaporizers ( jstor )
Plasmas ( jstor )
Random errors ( jstor )
Signals ( jstor )
Ceramic bearings ( lcsh )
Chemistry thesis, Ph. D ( lcsh )
Dissertations, Academic -- Chemistry -- UF ( lcsh )
Lubricating oils -- Analysis ( lcsh )
Mass spectrometry ( lcsh )
Plasma spectroscopy ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1997.
Bibliography:
Includes bibliographical references (leaves 176-183).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Robin Ann Russell.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
028064501 ( ALEPH )
37819121 ( OCLC )

Downloads

This item has the following downloads:


Full Text










DETERMINATION OF METALLO-ORGANIC AND PARTICULATE WETAR
METALS IN LUBRICATING OILS ASSOCIATED
WITH HYBRID CERAMIC BEARINGS BY
INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY












By

ROBIN ANN RUSSELL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1997
































In loving memory of my father, Bobby James Russell
April 14, 1935 - December 29, 1995














ACKNOWLEDGMENTS


Words can not express my gratitude towards my graduate research advisor, Dr. James D. Winefordner. I consider myself fortunate he allowed me to conduct research under his direction for the past few years. His consideration for my professional and personal well being was always evident. Dr, Benjamin W. Smith also played a large part in the completion of my graduate research. His innovative and thought provoking ideas have made me a creative and independent person inside and outside the laboratory. He has taught me many things, most notably the importance of patience and perseverance. This project brought many frustrating experiences that I would not have survived if not for Scott Baker. His unending patience and knowledge gave me the strength to continue even when I thought all hope was lost. I wish him and Amy all the best. Matthew DellaVecchia and Christopher Hurt were great sources of help in the laboratory as undergraduate students. Matt in particular experienced the struggles with me and always kept my spirits high. There are many others that have helped me through the past few years through constant support and encouragement. A few of these special people are Leslie King, Andrea Croslyn, Cynthia Schilling, Gretchen Potts, Bryan Castle, Arthur Besteman, Igor Gornushkin, Ricardo Aucelio, Kobus Visser, Wei Hang, Xiaomei Yan, Michael Bartberger, Gayanga Weerasekera, Angie Barker, Chibao Le, and Jim Lovelace. Jeanne Karably must receive special recognition as a source of information and strength. She was always able to answer any question I may have had, even those not pertaining to



iii








graduate school. I must also give thanks to my undergraduate academic and research advisor, Dr. Louis G. Daignault. His support for the past eight years has been constant and heartfelt. He has helped me in so many ways throughout my entire higher education experience and for that I am grateful.

I would like to thank Dr. Jim Adair and David Mitchell for supplying me with used lubricating oil samples used on the RCF tester.

The research contained in this dissertation would not have been completed in a timely fashion if not for the support of the machine and electronic shops of the chemistry department. In particular, Joe Shalosky, Gary Harding, Dailey Burch, and Steve Miles were always willing to assist me in trouble shooting the ICP-MS. They gave me knowledge that can not be learned in textbooks.

I would also like to acknowledge the support of my family, especially Terry, Tommy, Jogina, and my mother, Mildred. They understood my physical absence in their lives during my final year of graduate school. Even though I was not around very much, they let me know they believed in me.

It is necessary to also mention the help I received from Tom Musselman from Finnigan MAT. Although many SOLA problems were unexplainable, Tom did his best to help me keep the ICP-MS -running.

Finay, I would like to thank the Air Force Office of Scientific Research and the Texaco Foundation for financial support.


iv









ACRONYMS


AAS AE S AFS ATD, CE DI DIN

ETV F1

FT-ICR-MS


Atomic Absorption Spectrometry Atomic Emission Spectrometry Atomic Fluorescence Spectrometry Arizona Test Dust Capillary Electrophoresis Direct Insertion Direct Insertion Nebulization Electrothermal Vaporization Flow Injection Fourier Transform - Ion Cyclotron Resonance Mass Spectrometry Gas Chromatography Hydride Generation High Performance Liquid Chromatography Inductively Coupled Plasma Inductively Coupled Plasma - Mass Spectrometry Inductively Coupled Plasma - Optical Emission Spectrometry Ion Chromatography Ion Trap - Mass Spectrometry In-Torch Vaporization Laser Ablation


GC HIG

HPLC ICP ICP-MS

ICP-OES


IC

IT-MS ITV LA


V









LC Liquid Chromatography

LOD Lim-it of Detection

NAA Neutron Activation Analysis

PLC Programmable Logic Controller

PTFE Polytetrafluoroethylene

PSIM Particle Size Independent Method

RCF Rolling Contact Fatigue

RDE Rotating Disk Electrode

RSD Relative Standard Deviation

SEM Secondary Electron Multiplier

SIA Sequential Injection Analysis

SOAP Spectrochemical Oil Analysis Program

SRM Standard Reference Material

SFC Supercritical Fluid Chromatography

TOF-MS Time-of-Flight - Mass Spectrometry









TABLE OF CONTENTS


page


ACKNOWLEDGMENTS ..................................................................... iii

ACRONYMS.................................................................................. v

ABSTRACT ..................................................................................X

CHAPTERS

I INTRODUCTION ...................................................................I

2 BACKGROUND ....................................................................... 7

Oil Analysis ......................................................................... 7
Principles of Inductively Coupled Plasma - Mass Spectrometry............... 19
History ......................................................................... 19
Inductively Coupled Plasmas.................................................. 22
Mass Spectrometry ............................................................ 28
Mass analyzer function..................................................... 30
Ion optics ................................................................... 30
Quadrupole mass analyzer ................................................. 32
Detectors in ICP-MS ...................................................... 33
Inductively Coupled Plasma Coupled with Mass Spectrometry............ 36
Background ions ...................................................... 43
Interferences in ICP-M[S................................................... 45
Sample introduction ....................................................... 46

3 MICROWAVE DIGESTION........................................................ 54

Introduction ....................................................................... 54
Theory.............................................................................. 56
Instrumentation .................................................................... 59
Microwave Oven Calibration.................................................. 61
Microwave Acid Digestion Vessel ............................................ 67

4 DETERM[INATION OF WEAR METALS IN AQUEOUS SOLUTIONS ...... 71

Instrumental Set-up ................................................................. 71
Inductively Coupled Plasma...................................................... 71
Quadrupole Mass Spectrometry ................................................. 72
ICP-M[S Interface ................................................................. 72


vii








Sample Introduction ............75'
Experimental Parameters ............................................................. 77
Chemicals ....................................... .......................... .......77
G lassw are . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7 8
Experimental Parameters Optimrization .........................................78
A cid E ffects .. .... .......... .. ...... ..... .... ...... ..... .. 86
Data Analysis.................................................................. . .,86
Error Analysis .................................................................. ..98


5 DETERMINATION OF METALLO-ORGANIC WEAR METALS IN
OIL SAMIPLES .......................................................... ........ ... 106

ICP-MS Instrumental Set-up ................................................106
Microwave Digestion Instrumentation.............................................. 106
Experimental Parameters ............................................................ 107
Chemicals ......................................................................... 107
G lassw are . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..108
Microwave Acid Digestion Procedure.............................................. 108
Data Analysis ....................................................................... 111I
Error Analysis ........................................................................ 117

6 DETERMI[NATION OF PARTICULATE WEAR METALS IN OIL
S A M P L E S ............... ........ ............ ..... ... ... .. . 1 27

Instrum entation ............................... .. ..............127
Experimental Parameters ............................................................ 127
Chemicals ......................................................................... 128
Glassware......................................................................... 128
Preparation of Standards ......................................... 128
HN03 and H202 Digestion of Fe, Mg, Y, Ni, Mo ............................. 128
H2S04 and H202 Digestion of Al ..................................134
HC1 and H202 Digestion of Cr ............................................134
HF and H202 Digestion of Ti .............................................135
HF and H202 Digestion of Cr .............................................137
HIFand H202 Digestion ofAl .................................................... 137
A cid E ffects ............................................ ......143
Error Analysis........................................................................ 147

7 DETERMINATION OF WEAR METALS IN USED LUBRICATING
OILS ................................................................................ 152

Introduction ..................................................................... ....152
Instrumentation....................................................................... 152


viii









Experimental Parameters ............. ...........153)
M aterials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. 153
Collection of Oil Samples ......................................................... 156
M icrowave Acid Digestion .......................... ............ 158
D ata A nalysis ....... ... ........... ....... ..... . ... ......... 158


8 CONCLUSIONS AND FUTURE WORK .......................... 169

Conclusions.................................................. .... ... .. ..... 169
F uture W ork .. .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . .. . . .171

REFERENCES .............................................................................. 176

BIOGRAPHICAL SKETCH ............................................................ 184


ix















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

DETERMIfNATION OF METALLO-ORGANIC AND PARTICULATE WEAR
METALS [N LUBRICATING OILS ASSOCIATED WITH HYBRID CERAMIC BEARINGS BY INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY


By

Robin Ann Russell

August 1997


Chairman: James D. Winefordner
Major Department: Chemistry

It is possible to increase both the performance and operating environment of jet engines by using hybrid ceramic bearings. Our laboratory is concerned with investigating lubricating fluids for wear metals associated with silicon nitride ball bearings and steel raceways. Silicon nitride is characterized by low weight, low thermal expansion, high strength, and corrosion resistance. These attributes result in longer engine lifetimes than when metalic ball bearings are used. Before the routine use of ceramic ball bearings can be realized, the wear mechanisms of the materials should be thoroughly understood. One important variable in determining wear degradation is the concentration of metal present in the lubricating oils used with the bearings. A complete method for analyzing used lubricating oils for wear metal content must accurately determine all metal forms present.


x








Oil samples pose problems for routine analysis due to complex organic matrices Nebulizing these types of samples into an Inductively Coupled Plasma - Mass Spectrometer introduces many problems including clogging of the sample cone with carbon and increasing interferences. In addition, other techniques such as Atomic Absorption Spectrometry and Atomic Emission Spectrometry are particle size dependent. They are unable to analyze particles greater than 10 Lrm in size.

This dissertation describes a method of analyzing lubricating oils for both metalloorganic and particulate species by ICP-MS. Microwave digestion of the oil samples eliminates the need for elaborate sample introduction schemes as well as the use of a modified carrier gas.

Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y have been determined in both aqueous and organic media. Metallo-organic solutions of these metals were successfully digested, nebulized into the ICP, and the singly charged ions measured by mass spectrometry. Metal particulates in oil matrices have also been quantitatively determined by the above method. Linear analytical curves were obtained for these elements from the detection limits (- 1 ppb) to greater than 1 ppm.

Used lubricating oil samples were also analyzed by microwave digestion ICP-MS. Oil samples were collected from a Rolling Contact Fatigue tester. Two bearing systems were evaluated: M50 steel balls on an M50 steel rod, and Si3N4 balls on an M50 steel rod. Improved operating conditions were obtained when the Si3N4 balls were used, which corresponds to longer engine lifetimes.


Xi














CHAPTER 1
INTRODUCTION


Ceramic materials are currently being used for a wide variety of applications including aircraft and aerospace components, machine tools, medical equipment, semiconductor processing equipment, and automotive components. They are replacing components that are currently constructed out of metallic materials due to their many advantageous characteristics, such as low density, the ability to operate in high temperature environments (550 - 800 'C), and corrosion resistance.

The United States Air Force is actively searching for replacement materials for ball bearings in jet engines. In particular, silicon nitride (Si3N4) is finding extended use in bearing systems that employ ceramic rolling elements with steel raceways. It is desirable to use silicon nitride ball bearings over other ceramic materials due to the manner in which they fracture. Like steel bearings, Si3N4 fractures by spalling, or a gradual, pitting-induced fatigue, which has been linked to surface crack formation on the rolling element. Other ceramics fail more catastrophically, by total fracture of the material.

Aeronautic applications traditionally use ball bearings constructed out of steel, such as M-50. Silicon nitride has several characteristics that exceed those of steel including a superior resistance to corrosion, lower mass (40 % less dense), increased hardness, the ability to withstand higher temperatures, and nonmagnetism, which


I





2


eliminates the problems of electric arc damage. In addition, Si3N4 is compatible with a wider range of lubricants than steel ball bearings. The low mass of silicon nitride results in a reduced centrifugal ball load and enables it to be subjected to much higher speeds (1), At temperatures greater the 550 'C, stainless steel begins to lose its hardness and stability, In contrast, Si3N4 maintains its hardness up to temperatures of 1100 'C. This results in a material that can be run at higher speeds, higher service temperatures, and with less lubrication.

Several studies have already been conducted using Si3N4 in hybrid bearings (2,3). It has been determined that the replacement of stainless steel with Si3N4 as the rolling element material in ball bearings results in longer service life due to reduced wear of the ball bearings.

Silicon nitride is fabricated by a variety of methods, the most common being hot isostatic pressing (4). In the first step of manufacturing, silicon nitride powder is Mixed with sintering aides such as yttrium oxide (Y203), magnesium oxide (MgO), and aluminum oxide (A1203). The resultant powder is then mixed with a binder and shaped into spheres by pressing it into a mold. The material is then sintered at 1,700 - 1,800 TC in an inert gas atmosphere, such as nitrogen gas, and is subjected to hot isostatic pressing. An oxynitride liquid is formed upon heating the mixture, dissolving the a-silicon nitride particles. 03silicon nitride precipitates out of the liquid and constitutes the bulk of the Si3N4 densification product. The final step involves finishing the Si3N4 spheres with a diamond grinding wheel.


A








Before metallic components can be replaced with SON ball bearings, the wear mechanisms of this material must be fully understood. One way to do this is to study the lubricating oils associated with these systems. The determination of wear metals in used lubricating oils gives a good indication of the wear of engine parts (5). Often an increase in concentration of a metal is indicative of wear that has occurred before it is visually apparent. As the ball bearings wear over time, elements that went into the construction of the ceramics will appear in the lubricant. Such elements include silicon, nitrogen, and any sintering aids such as yttrium, magnesium, or aluminum. It is also possible to observe in the used lubricating oil any elements that make up the raceway material or other engine components. It is the goal of this research to qualitatively and quantitatively determine wear metals in lubricating oils associated with the wear Of Si3N4 hybrid bearings. These include Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y. Wear metals can take two forms: metalloorganic and particulate. It is desirable for a technique to be able to detect particulate wear metals over a wide range of sizes, Lrm - mmn diameters, since particles in the oil can also further degrade component surfaces. Preventative maintenance programs can be established by determining the cause of failures before detrimental effects are experienced. This constitutes repairing or replacing failing parts as well as simple oil changes. The successful implementation of preventative maintenance programs has the potential to save millions of dollars and possibly lives each year.

Metals in used lubricating oils originate from three sources: wear, contamination, and additives. Wear metals are defined as the products of frictional deterioration or corrosion and can be traced back to the composition of the engine under inspection (6).






4


Dirt and leaks in the system can lead to contamination, while additives are put into lubricating oils to help reduce wear.

As discussed earlier, Al, Mg, and Y are common sintering aides used for the manufacturing of silicon nitride. As shown in Table 1, the presence of iron indicates wear from engine components, chromium indicates ring wear, nickel indicates wear of plating components, and molybdenum and titanium indicate wear of bearing alloys. It is desirable to detect these elements in lubricating oils at a level that precludes complete engine failure.

Current techniques used by the United States Air Force for determining wear metals in used lubricating oils are Atomic Absorption Spectrometry (AAS) and Atomic Emission Spectrometry (AES). Both of these techniques give excellent qualitative analyses for wear metals. They are, however, unreliable when the metals are in the form of particulates. AAS and AES are unable to accurately determine particulate wear metals greater than a few micrometers in diameter. In addition, other atomic analytical techniques such as Inductively Coupled Plasma - Mass Spectrometry (ICP-MS) enable lower metal concentrations to be determined. This results in earlier wear detection and improved maintenance programs.

This dissertation describes a novel method for determining metallo-organic and particulate wear metals in lubricating oils by Inductively Coupled Plasma -Mass Spectrometry preceded by microwave digestion. Inductively Coupled Plasma -Mass Spectrometry is an ideal method for determining trace levels of wear metals. The low detection limits achieved by this technique enable preventive maintenance programs to be established since it can detect metal content at very low levels. In addition, microwave






5


digestion converts all metallic species, into the aqueous phase, thereby eliminating the particle size dependence of the current methods.





6


Table 1. Type of Wear Indicated by Element.


Element Indicated Condition

Al Indicates wear of Si3N4

Mg Indicates wear of Si3N4

Y Indicates wear of Si3N4

Fe Indicates wear from engine components

Cr Indicates ring wear

Ni Indicates wear of plating components

Mo Indicates wear of bearing alloy

Ti Indicates wear of bearing alloy














CHAPTER 2
BACKGROUND


Oil Analysis

Numerous techniques have been employed for the elemental analysis of oil matrices including Neutron Activation Analysis (NAA) (7), Colorimetry (8), Atomic Fluorescence Spectrometry (AiFS) (9,10,11), X-ray Fluorescence Spectrometry (12,13), Inductively Coupled Plasma - Mass Spectrometry (ICP-MS) (14), Atomic Absorption Spectroscopy (AAS) (15,16,17), and Atomic Emission Spectrometry (AES) (18,19,20). The latter two techniques have been used the most in conjunction the United States Air Force Spectrometric Oil Analysis Programs (SOAP). Therefore most of the literature regarding lubricating oil analyses concerns AAS and/or AES. Atomic Absorption Spectrometry has been used with a flame or a furnace for atomization and has the advantages of being rapid with simple sample preparation, wide linear dynamic range, low instrumentation cost, and the ability to analyze liquid organic samples directly. Two of the more popular means of preparing oils for flame AAS are direct aspiration of the sample after dilution with an appropriate organic solvent, and burning the sample and then extracting the residue with inorganic standards. Anwar et al. directly aspirated crude oil samples into a flame AAS after diluting them with a solvent system consisting of butan-2one (for desolvation of the organic matter), water (to accommodate the inorganic metal content), hydrochloric acid (to decompose or dissolve the metal into the aqueous phase),


7





8


metal content), hydrochloric acid (to decompose or dissolve the metal into the aqueous phase), propan-2-ol (to maintain the homogeneity of the sample solutions), and ethanol (to blend the other constituents) (21). For comparative reasons, a dry ashing protocol was also performed on the oil samples. These two procedures gave vastly different results for crude oil samples. Samples that were ashed before introduction into the flame gave lower results than those obtained by the mixed solvent system due to the losses that occurred during the combustion process. Although good results were obtained with the mixed solvent method, it was found to be highly dependent on the solubility of the oil samples, homogeneity of the solution, solvent composition, and the time and temperature of the oil/solvent solution.

Udoh has shown the determination of calcium, magnesium, and zinc in unused lubricating oils by flame atomic absorption spectrometry after ashing the samples (22). Sample preparation consisted of heating the oil samples on an evaporating dish, combining the residue with hydrochloric acid, concentrating the solution on a steam bath, and then diluting the sample once again. Phosphorous is known to interfere with the determination of alkaline earth metals by flame AAS. Therefore, lanthanum and strontium were added to several of the oil concentrates to determine the effects of their concentration on the calcium and magnesium signals. The Ca concentration exhibited a linear correlation with both increasing La and Sr concentrations, while the Mg concentration was not affected by addition of these two elements.

Ashing of oil samples can also be accomplished by a graphite rod or filament atomizer (23,24). These methods have several advantages over flame AAS such as






9


greater sensitivity, better detection limits, smaller sample volumes, and particle size independence, which allowed for total metal content analyses. Reeves et al. determined silver, chromium, copper, iron, nickel, lead, and tin in used jet-engine and reciprocatingengine oils using AAS with a graphite rod atomizer (25,26). Comparison of their results with other laboratories showed good correlation. Dilution of the oil samples with isooctane was required to obtain better sensitivity for the analysis of iron and copper. Bratzel et al. employed a graphite rod as an atomizer for AAS for the determination of lead in petroleum and petroleum products (27). Direct analysis of oil samples as well as the extraction of organic species into an aqueous phase were presented. Direct analyses were performed after diluting petroleum samples with xylene. The reversed extraction procedure consisted of a xylene-aqueous nitric acid system with the addition of solid dithizone to increase the rate of extraction. The carbon rod atomriization method resulted in a detection limit of 2 x 10-12 g for lead. The sensitivity of the technique was affected by the type of lead salt and solvent used for both standard and sample preparation, choice of wavelength, and interferences.

Other means of preparing petroleum samples for AAS analyses have also been used considerably. Decomposition and cold-vapor methods were reported by Knauer et al for the determination of mercury in petroleum and petroleum products (28). The first method required several hours to complete and involved a closed system digestion with nitric acid and sulfuric acid. Another method presented was based on a Wickbold oxyhydrogen combustion set-up where the volatilized mercury was collected in an acidic permanganate solution. This was also lengthy and arduous. The collected solutions from






10


both methods were analyzed by cold-vapor AAS. Although mercury concentrations down to 5 - 10 ng/g could be detected, several problems occurred with these methods. First, sample contamination from glassware and reagents was widespread. In addition, when independent laboratories used these methods, the results exhibited a great deal of variability. The elaborate sample preparation procedures were thus nontrivial.

The above methods for AAS work well for metallo-organic samples. However, not all samples containing particulate matter can be accurately analyzed by AAS. Saba et al have reported that engine component failure has occurred as a result of the inability to detect large wear particles (29). Jantzen et al (30) and Kauffmnan et al. (3 1) reported that direct AAS could only accurately detect iron wear particles smaller than about 1 L.i Jantzen's laboratory compared direct flame AAS, AAS after incineration, and colorimetry for iron content in oil samples. Direct ALAS gave values 50% lower than colorimetry when the oil samples contained iron particles larger than 2 4m.

Saba et al. used AAS with a graphite furnace atomnizer was to analyze iron, copper, aluminum, magnesium, and other important wear metals in aircraft lubricating oils

(32). Standards and samples were diluted with kerosene before insertion into the graphite tube. Oil samples analyzed by this method contained metallo-organic and metal particulates. Metal powder suspensions were also analyzed by a particle size independent method (PSM to determine the actual metal concentration in the samples. The PSIM consisted of digesting the samples in aqua regia, shaking for 30 seconds, and then agitating in an ultrasonic bath at 65 'C for 45 minutes (33). The mixture was then cooled and diluted with methyl isobutyl ketone in isopropyl alcohol. Graphite furnace AAS was





I1I


shown to be superior to conventional methods since it was capable of analyzing wear metal particles. However, the precision of the graphite furnace AAS method decreased with increasing particle size due to the inability to obtain a representative sample in the small amounts (microliter sample size) placed into the furnace.

Atomic Emission Spectrometry has also been widely used for the determ'lination of trace elements in oil matrices. Atomization of the analytes is usually accomplished with a Rotating Disk Electrode (RDE) (34,35) or an Inductively Coupled Plasma (ICP) (36,37). In RDE-AES, a rotating electrode is submerged in an oil sample, while a fixed electrode is positioned above it. Emission data can be obtained once the sample is vaporized with a high voltage arc. Gambrill el al. used RDE-AES for the determination of Pb, Fe, Ba, P, Ca, and Zn in new and used lubricating oils (38). A graphite disk, 12.7 mm in diameter and 3.18 mim thick, rotating at various speeds was used as the lower electrode while a 6. 35 mm graphite rod served as the fixed upper electrode. The best analytical curves were obtained when high rotation speeds of 15 rpm were used.

Compared to ICPs, the Rotating Disk Electrode technique was simpler to operate and more stable. In addition, larger particles could be analyzed by this method and oil samples could be studied without any sample preparation. Plasmas do, however, offer several advantages over flames and RDEs. The higher temperatures and longer residence times encountered by the sample aerosol leads to a greater degree of atomization than that seen in flames, which results in lower detection limits. For these types of analyses, the oil samples require dilution with a suitable organic solvent such as xylene, kerosene, methyl isobutyl ketone, or toluene to reduce the viscosity of the sample before introduction into





12


the plasma. Bangroo et a[ analyzed wear metals and additive elements in lubricating oils by AAS and ICP-AES after subjecting the samples to a rigorous treatment that included ultrasonicating with hydrochloric acid and hydrogen peroxide, diluting with xylene or toluene, agitating for thirty minutes, and then separating the aqueous phase from the organic phase (39). The procedure was then repeated on the resulting oil residue. Although experimental data agreed with certified values, the sample preparation steps were time consuming and subject to several losses. Brown et al. used ICP-AES for the determination of twenty-one elements in new and used lubricating oils (40). The samples were first diluted with xylene before introduction into the plasma, which gave a low bias when compared to other sample preparation methods. Although this method allowed for sub ppm detection limits and multielement analysis, variable wear particle size could result in imprecise results when using this method.

Modification of the sample introduction scheme for AES has also been investigated by many researchers (41). Algeo et al. used a modified Babington nebulizer to introduce oil samples into an ICP-AES (42). This type of sample introduction allowed viscous oils to be nebulized more efficiently and without organic solvent dilution. However, lower results were obtained when compared to ashing-AAS for real world samples due to the particle size dependence of the method.

AES is also an inadequate technique to use for determining total metal content when particles are present in oil samples. In ICP-AES, particles greater than 10 4m in size may pass through plasma without being atomized (39). Kauffman et aL analyzed metal powder suspensions by direct current plasma spectrometry preceded by acid






I13


dissolution (43). A mixture of HF/FH-N03 was used for dissolution of the suspensions. Percent recoveries for Al, Cr, Cu, Fe, Mg, Mo, Ni, Pb, Si, Sn, and Ti were 2 - 48 % without acid dissolution. The use of acid increased the percent recoveries to 89 - 102 %.

The latest method that has been used for the determination of wear metals in oil matrices is Inductively Coupled Plasma - Mass Spectrometry (ICP-MS). It is an attractive technique for this application due to its numerous advantages which include low limits of detection (down to pptr), large linear dynamic range (up to five orders of magnitude), simple spectra, multielement analysis capabilities, and simple sample introduction methods. It is not desirable, however, to introduce organic samples directly into the mass spectrometer because of increased interferences and carbon deposition on the sample cone orifice and ion optics. When an organic sample is analyzed by mass spectrometry, more background peaks are observed than with aqueous samples. Polyatomic ions containing carbon can also increase the level of interferences, as shown in Table 2 (44). It is extremely difficult to determine trace levels of elements such as magnesium, calcium, titanium, chromium, and iron when large levels of polyatomic ions interfering with the elemental masses are present in the plasma.

Although organic solvents are ionized completely in the high temperature plasma, carbon has a tendency to condense on the cooled surface of the sample cone orifice and even on the ion optics. This can be minimized by adding oxygen to the carrier gas. The oxygen atoms combine with carbon atoms in the plasma to form carbon dioxide which is expelled from the system before deposition occurs. The exact amount Of 02 needed to accomplish this is a tedious task. Too much 02 can prematurely degrade the sample cone,





14


Table 2. Interfering Ions Found When Organic Matrices are Introduced into the ICP-MS.
(Adapted from 44)


Species Mass Interfering
____ ___ ____ ___Ion
12C+24 Mg
13 C2+, 12 C21{ 25 Mg 12CNW 26 Mg
12 CO+, N2+ 28 Si
1C0+44 Ca
12COOH, '3CO2+ 45 Sc 13 C021-1i 46 Ti, Ca Arl2c+ 52 Cr
Ar~+53 Cr
,arO+ 56 Fe
ArOWi 57 Fe
4OAr2+ 80 1 Se





15


while too little can result in carbon build-up. The appropriate amount Is dependent on many variables of the system including the physical characteristics of the organic solvent, solvent flow rate, spray chamber temperature, and plasma conditions (45).

In addition, organic solvents have high vapor pressures which can overload the plasma causing it be become unstable. There are several operating parameter

modifications that can be performed to help with plasma stability. One parameter is the liquid flow rate. Reducing the solution uptake from 1.0 mI/mmn to 0.5 mllmin decreases the amount of solvent in the plasma and thus increases its stability (44). Organic samples also require more power to sustain the plasma than aqueous ones. As a result, an increase of 500 W is typically needed.

There are many sample introduction schemes for ICP-MS that enable one to analyze liquid organic samples such as direct aspiration (46,47), electrothermal vaporization (ETV) (48), in-torch vaporization (ITV) (49), high-performance liquid chromatography (HPLC) (50,5 1), gas chromatography (GC) (52,53), liquid chromatography (LC) (54,55), ion chromatography (IC) (56,57), capillary electrophoresis

(CE) (58), supercritical fluid chromatography (SFC) (59), laser ablation (LA) (60), cryogenic desolvation (6 1), membrane desolvation (62,63), direct insertion (DI) (64), flow injection (Fl) (65,66), direct injection nebulization (DIN) (67,68), and sequential injection analysis (SIA) (69). Of these techniques, direct aspiration, ETV, GC, and SIA have been used with ICP-MS for the analysis of petroleum samples. When direct aspiration is used with ICP-MS, the sample is converted into an aerosol and then transported to the plasma for vaporization, atomnization, and ionization. The sample is either combined with an





16


organic solvent or carried through a dissolution or digestion procedure prior to being nebulized. The digestion of oil samples can be tedious and time-consuming, with some digestions requiring hours to complete. Olsen el al. directly aspirated oil samples into the ICP-MS by nebulization (70). An extraction procedure using xylene was carried out on the oil samples prior to sample introduction. The accuracy of the technique was investigated by subjecting NIST Standard Reference Material (SRM) 1634b to the same procedure as the samples. Errors for several elements were rather large when the experimental and certified values were compared. In fact, trace concentrations of chromium in an organic matrix could not be determined by direct ICP-MS due to the large variation in the interfering oAr12 CH on mass 53, the most abundant isotope of Cr.

Al-Swaidan used ICP-MS for the determination of vanadium and nickel in crude oil products by first extractmig the analytes from the organic solution to an aqueous one (14,71). This method employed the addition of xylene or methyl isobutyl ketone to reduce the viscosity of the sample. Extraction of the metals was facilitated with the aid of nitric acid. Recoveries of added V and Ni were good, with percent relative deviations less than 4%. Bettinelli et aL. analyzed fuel oils by ICP-MS preceded by conversion of the oils into aqueous samples by acid mineralization in a microwave oven (72). This group obtained detection limits in the sub ppm region for numerous metals including Al, Cr, Fe, and Ni.

Tandem sample introduction schemes have also been used for the determination of trace metals in oil matrices. Very small sample volumes (WJ) can be used when a graphite furnace is used for vaporization for ICP-MS detection. Other advantages of vaporization with an ETV include better sensitivity;, reduced molecular ion interferences; reduced oxide





17


formation, reduced nonspectroscopic interferences; and the ability to directly analyze samples containing of liquid organic material, strong acids, and high dissolved solids. The sample can be dried, pyrolyzed, and vaporized prior to entering the plasma. Thus, many interferences can be eliminated by step-wise heating of the sample In addition, plasma conditions for atomnization and ionization can be optimized independently of those used for vaporization. As a result, sample introduction parameters can be better controlled. Escobar et al. coupled ETV with ICP-MS to determine Al, Mg, Fe, and Y metallo-organic compounds in lubricating oils (73). The oil samples were diluted with xylene before injection into the graphite furnace, which allowed the metallic species to be separated from the organic matrix. Thus, organic material present in the sample was vented to the atmosphere and not allowed to enter the plasma. Detection limits obtained by this method were in the ppb region. ETV-ICP-MS has also been used for the determination of mercury in petroleum by Osborne (74). Oil samples were mixed with various alkanes and injected into a graphite furnace for vaporization. A detection limit of 3 ppb for Hg in pentane was reported. However, the relative standard deviation associated with this value was 50%.

Electrothermal vaporization has a few undesirable characteristics that make it a difficult method of sample introduction for ICP-MS. Precision can be poor due to the difficulty of pipetting oil solutions into a graphite furnace. In addition, various elements require different experimental conditions for efficient vaporization and transport into the plasma. Much work must be done to ensure that optimum ashing and vaporizing procedures are used. Carriers must also be added to many samples to efficiently transport





18


the sample from the furnace to the plasma. Multielement determinations by ETV-LCP-MS are difficult due to the limited time available for mass scanning and the complexity of finding suitable compromise conditions for multielement mixtures.

Alonso el at used ion chromatography ICP-MS to analyze spent nuclear fuels for fission products and actinides (57). The outlet of the separation column was connected directly to a cross-flow nebulizer. Separation of Cs, Ba, and the actinides was facilitated with nitric acid. Cesium and barium were separated on one column while a different column was used for separation of the actinides. The only way to eliminate isobaric interferences in the determination of the above elements was to chemically separate the interfering nuclides. Three individual separation procedures were required to isolate Cs and Ba, the lanthanides, and the actinides.

Sequential injection analysis has been used with ICP-MS by AI-Swaidan to analyze Saudi Arabian crude oil for lead, ni'ckel, and vanadium (69). SIA differs from flow injection in that a selector valve allows sample and reagent zones to be sequentially injected into a channel. In flow injection, the sample zone is injected into a flowing carrier stream and reagents are combined with it in route to the detector. SIA has several advantages over FL including minimization of reagent use and waste production since the flow is not continuous. In addition, corrosive solvents are handled in a closed system.

Finally, most analyses of lubricating oils quote detection limits in the low ppm or mid ppb region. A technique with lower detection limits would permit earlier detection of wear metals in lubricating oils, and therefore improve the quality of maintenance programs. A method of converting oil solutions into aqueous ones would be





19


advantageous. Aqueous solutions are stable for long amounts of time and can be easily nebulized for a variety of detection schemes.




Principles of Inductively Coupled Plasma - Mass Spectrometry



History

The first Inductively Coupled Plasma (ICP) sustained in atmospheric pressure flowing gases was accomplished by Reed in 1961 (75). He sustained plasmas using various support gases such as argon, and various argon mixtures using oxygen, hydrogen, and helium. Reed introduced the first application of inductively coupled plasmas at atmospheric pressure: independently growing sapphire, zircomia, and niobiumn crystals

(76). Powder was fed into the ICP through the center of the torch. The crystals had a diameter of 5 - 15 mm and a length of 30 - 90 mm. Growth rates were 20 - 50 mm/hr. It was soon realized that the plasma could be used in other applications. In particular, it could function as superior excitation source for atomic emission spectrometry (AES). During the past 30 years, ICP-AES has become one of the most widely used atomic spectrometric methods.

Gray at the University of Surrey was first in illustrating the union of plasmas and mass spectrometers (77). He coupled an atmospheric pressure dc arc plasma to a quadrupole mass spectrometer and obtained ig/ml detection limits in aqueous solutions. Complete vaporization, atomnization, and ionization was not possible with this system due to the core gas temperature of the plasma, which was about 5000 K. This set-up only









worked well for those elements with Ionization energies below 8 eV. In addition, ffirther problems were experienced with inter-element and matrix effects, and sample introduction. With their high temperatures, inductively coupled plasmas were the next logical source of ion production for mass spectrometric detection.

Gray, along with Date from the British Geological Survey, Fassel and Houk from Ames Laboratory at Iowa State University, and Douglas from Sciex became the pioneers in ICP-MS as they worked towards developing a sensitive technique for elemental determination. Houk et al published the first manuscript describing the determination of trace elements with an ICP-MS in 1980 (78). The hardware of the ICP-MS used by this group was similar to that employed in modem instruments. An ultrasonic nebulizer was used to produce an aerosol of the liquid solution. The sample aerosol was then transported to the plasma for vaporization, atomnization, and ionization. The sample cone had an orifice of 50 tm and was water cooled. Mass spectra were presented that illustrated the isotopic capabilities of ICP-MS. Analytical curves were constructed which extended almost four orders of magnitude. In addition, detection limits for Mg, Cr, Mn, Co, Cu, Rb, As, and Y in the range of 0.002 - 0.06 pLg'ml were obtained. One problem observed in the initial experiments was condensation of solid material in or near the sample cone aperture.

The size of the orifice in the sample cone was varied by several researchers to determine its effect on sampling. Larger sampling orifices initially appeared advantageous since they could accommodate larger ion beams and therefore transmit a greater number of analyte ions to the detector. It was determined that at apertures greater than 100 im, a






21

secondary discharge developed in front of the sample cone (78,79). Further ionization by collisions can occur in the secondary discharge and cause high photon levels to reach the detector; high populations of doubly charged ions; production of ions with a higher kinetic energy and energy spread, resulting in peak splitting; and erosion of the sampler orifice. The secondary discharge was the result of arcing between the plasma and the sample cone caused by the large RE voltage swing in the plasma relative to the grounded sample cone. Conventional load coils are grounded at one end while the other end is free to undergo voltage swings at the applied radio frequency. These voltage swings induce a magnetic field in the plasma and couple capacitively into the discharge. When the plasma swings positive, a negative voltage appears in the plasma and a second discharge is generated at the sampling cone. The secondary discharge can be overcome by grounding the load coil at the center (80). The ends of the coil are allowed to go through equal but opposite voltage swings, and cancel each other out. The secondary discharge can also be overcome by using the typical end-grounded load coil and smaller aperture sample cones.

The first commercial instrument of this type was introduced at the 1983 Pittsburgh Conference by Sciex, Inc. There are now several LCP-MS manufacturers, offering varying types of mass analyzers. The combination of inductively coupled plasmas and mass spectrometers provides many advantages in atomic spectroscopy such as multielemental analyses, isotopic capabilities, simple spectra, small sample requirements, low detection limits, and a large linear dynamic range.








Inductively Coupled Plasmas

An ICP is an electrodeless discharge formed in a gas, maintained by energy coupled to it from a RE generator. A quartz torch is placed inside a three turn copper induction, or load, coil that is connected to a high frequency radio frequency (RE) generator. RE generators used for inductively coupled plasmas are operated at a frequency of 27.12 or 40.68 MI-Lz with output levels up to 5 kW. The most commonly used support gas for plasmas is argon, which flows through the center of the torch. By definition, a plasma is an ionized gas. Since Ar is a nonconductor, it is necessary to provide electrons for ionization. A Tesla coil is used to seed the Ar with free electrons. The electrons flow from the Tesla coil to the plasma region by means of the induction coil. Oscillating magnetic fields are generated by the high frequency RE currents flowing in the induction coil. Figure 1 shows the inner axially oriented magnetic forces as well as outer elliptical fields that are formed (81). The induced magnetic fields are time-varying in strength and direction. The axial magnetic fields induce eddy currents which flow in closed annular paths inside the torch. The accelerated electrons and ions encounter resistance to their flow, resulting in Joule or ohmic heating and consequently additional ionization. Since an electron has a mean free path of about I 4m at atmospheric pressure, collisions with argon atoms occur, heat the plasma, and form a bright discharge.

In the center of the plasma, temperatures reach 9,000 - 10,000 K. Another stream of gas, usually Ar, flows through the torch to prevent overheating and subsequent melting of the quartz glassware. It is also used to tune the shape of the plasma for optimum ion beam conditions. This gas flow is known as the intermediate gas.












0o


10


1


H


0


r


1
Ar


Figure 1. Schematic of Magnetic Fields Induced in Inductively Coupled Plasma.
H Denotes Magnetic Field.


23


0





24


Aerosol sample enters the plasma through an injection tube located inside the torch. At high RF frequencies, the plasma takes on an annular shape due to the eddy current flowing more closely to the outer edges of the plasma. This is shown schematically in Figure 2 (8 1). The extent to which the plasma takes on the annular shape is controlled by the frequency of the primary current generator and the carrier gas flow rate. A plasma operating at a RF frequency of 27 MI-Iz requires a carrier gas velocity of

- 1 L/min for optimum sampling.

As illustrated in Figure 2, the sample particles do not pass through the hottest par-t of the plasma. They experience a plasma temperature of 7,000 K, twice that found in the hottest flame. In addition, sample particles generally spend a few milliseconds in the plasma. These two parameters, high temperature and long residence time, results in nearly complete vaporization of analyte and solvent.

Aerosol particles exit the injector tube and enter the plasma where they are first dried by solvent evaporation. The higher plasma temperatures then serve to vaporize, atomize, and ionize the particles. If a 1 ppm solution is aspirated into the plasma through a pneumatic nebulizer at a sample uptake rate of 1 mL/min, the total analyte atom population will be approximately 1 x 10 cm-' (82). The plasma temperatures which the sample particles encounter are high enough to completely (> 99%/) ionize most of the elements in the periodic table. The degree of ionization is a function of many variables including plasma conditions and ionization potentials of the elements. Elements with ionization potentials less than Ar (15.8 eV) are ionized in the argon ICP. As shown in Figure 3, most elements have first ionization potentials in the 5 - 15 eV range.






















0 *0
0 0










Sml



Particles


Figure 2. Sample Aerosol Entering Plasma. (Adapted from 8 1)











24.6
01 F7. 21e
8.301 11.3 1 14.5 1 13.6 I 1I2.


13.6
Li 5.39
Na 5.14


Si


K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
4.34 6.11 6.54 6.82 6.74 6.77 7.44 7.87 7.86 7.64 7.73 9.39 6.0 7.90 9.81 9.75 11.8 14.0
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
4.18 5.70 6.38 6.84 6.88 7.10 7.87 7.37 7.46 8.34 17.58 8.99 5.791 7.33 8.64 19.10 10.5 12.1
Cs Ba La Hf Ta W Re Os Ir Pt Au Hg TI Pb Bi Po At Rn
3.89 15.21 5.58 7.0 17.89 7.98 7.88 18.7 9.10 9.0 19.23 10.441 6.12 17.42 17.23 18.42 10.75


P


S


CI


Ar


Rf I Ha


Sg


Ns


Hs I Mt


Figure 3. Ionization Potential for Selected Elements.


ON


Al 5.99


Be 9.32 Mg 7.65


Fr


Ra 5.28


Ac 6.9


- U - m - m - a - a - a - & - I -


ICe Pr Nd Pm Sm Eu Gd Tb Dy Ho ErITm Ybj Lu1
5.47 5.42 5.49 5.55 5.63 5.671 6.14 5.8Sf 5.93 6.02 6.1016.1816.251 5.43

Th IPa IU INp IPu IAm ICm IBk ICf IEs IFm IMd INo ILr
6.1j 5.91 6.11 M6.22 5.8 6.0 [6.03 6.241 6.31 16.42 16.5116.59 16.661





27


Not all analytes in the sample are extracted and detected by the ICP-MS. Roughly, less than one in a million analyte ions formed in the plasma reach the detector The probability that an analyte is detected is highly dependent on characteristics of the system such as gas temperature, electron number density, electron temperature, and energy transport rates. The processes of vaporization, atomnization, and ionization occur simultaneously in the plasma. Every second more than a million polydisperse aerosol droplets enter the plasma (83). Drops of different sizes undergo the above processes at various positions and time in the plasma. The temperature gradients observed in the plasma result in nonuniform ionization. Simple equations can not be used to describe ionization in the heterogeneous plasma. If the plasma were in local thermodynamic equilibrium, a simple equation could be used to relate the electron number density, temperature, and pressure. In addition, the electron temperature and gas temperature would be identical.

Ionization efficiencies can be calculated from Equation I where M' is the ionized




M+ 2 (2,rmrnkT>)3/2Q~
Q' e h1P)QT Equation 1




element, NV is the neutral element, n, is the electron number density in the plasma (cm-'), m. is the mass of the electron (kg), k is Boltzmann's constant (JfK), Te is the free electron temperature (K), h is Planck's constant (J-s), Q+ and Q0 are the electronic partition ftinctions of the ion and neutral atom (dimensionless), lIP is the ionization potential of the






28


element (eV), and I, is the ionization temperature (K) (84). Equation I assumes the plasma is in local thermodynamic equilibrium when in fact it is not As a result, actual ionization efficiencies may be higher than those predicted based on calculations.

Degrees of ionization, [M] / (Iilv]+[M]) x 100%, are given in Figure 4 for most elements in the Periodic Table, of which most are greater than 90% (85). Values in parentheses are for doubly charges species. The degrees of ionization were calculated using 1 X 101 cnf3 as the value of N~ and 7500 K for both T. and Ti. One can see that even elements like As, Se, and S that are characterized by high ionization potentials can be highly ionized in the ICP. Even though the plasma has a high temperature, ionization conditions are not overly energetic. The temperature and electron number density values do not lead to a large amount of doubly charged ions. For example, Ba has one of the lowest second ionization potentials (10 ev) and only a 9% efficiency for producing doubly charged ions in the ICP.




Mass Spectromety

Numerous types of mass spectrometers have been used with inductively coupled plasmas as the ion source including quadrupole MS (86), high resolution double-focusing magnetic sector MS (87), time-of-flight MS (TOF-MS) (88), ion trap MS (IT-MS) (89), and fourier transform ion cyclotron resonance MS (FT-ICR-MS) (90). Most of the research performed with plasma source mass spectrometry has used quadrupole mass spectrometers. The work presented in this dissertation was carried out on a Finnigan










( M+ --- ) X 100%


He


B I N OjF F Ne
018 0.1 I910 I6x10'
5 - 0-.1


Al
98


Si
85


P
33


S
14


CI
0.9


Ar 0.04


K Ca Sc T1 V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr 100 99(1) 100 99 99 98 95 96 93 91 90 75 98 90 152 33 15 0.6
RbS Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
100 96(4)1 98 99 98 98 196 94 193 93 165 99 96 78 66 129 8.5
Cs Ba La Hf Ta W Re Os I r Pt Au Hg TI Pb Bi Po At Rn 10 91(9) 190(10) 96 95 94 93 78 62 51 ,38 1100 197(01)1 92
a0 a19 a M


OAC


Rf


Ha IS9 INs


Hs


Mt


am Earnam a mama i a a


(M2+ M+ + M )x1%


Figure 4. Ionization Efficiencies for Selected Elements. (Adapted from 85)


H 0.1
Li
100
Na
100


Be
10
75
Mg
98


Fr


Ra


Ce Pr NdIPm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
96(2) 90(10) 99 I97(3) 100 93(7) 99 100 J 99 91(9)1 92(8)

Th Pa IU Np Pu Am Cm Bk Cf Es IFmnMd No Lr
100 100 1H






30


MAT SOLA ICP-MS, which employed the Balzers QMS 511 quadrupole mass analyzer. Therefore, the discussion which follows concentrates on quadrupole mass analyzers



Mass analyzer function

The function of a mass spectrometer is to separate an ion beam into its components according to their mass-to-charge ratios (m/z). The mass-to-charge ratio then can be used to determine the composition of the sample. In atomic mass spectrometry, analyte identification is simplified in that the mass-to-charge ratio directly gives the masses of the species present in the sample since most ions are singly charged. The mass spectra obtained are much simpler than those obtained with molecular mass spectrometry.



Ion optics

A schematic of a quadrupole mass spectrometer used for this research, the Finnigan MAT SOLA ICP-MS, is given in Figure 5. The ion beam from the plasma is directed to the mass analyzer by the ion transfer system, which includes the beam extraction, X deflection plates, Y deflection plates, Y steering plates, and focus. Potentials on the transfer optics help to direct the ion beam into the quadrupole entrance, and deflect unwanted signals from the system. These optics are adjusted to maximize the ion beam which emerges from the mass analyzer.














j -Mass Focus Sample
AnalyzerSteering
PhseMachDeflector Skimmer Cone
DeflctorLensCone

Torch










unting 4+Plasma

~ctor TurboTub
Pump Turbo Tub Rotary


Accelerator


#3


Pu2 Retarding
Lens


U1F Pump #1


Sample Inlet


Figure 5. Schematic Diagram of Finnigan MAT SOLA ICP-MS


Faraday Collector


Ion Co Colle









Quadruvole mass analyzer


Mass separation using a quadrupole mass analyzer is achieved by varying electric fields. The mass analyzer is made up of four long cylinders arranged in a square array, with opposite rods electrically connected. DC and RF voltages are applied to opposite pairs of rods. The DC voltage is positive for one pair and negative for the other pair. The RE voltages applied to the rod are equal in magnitude, but have opposite signs. The mass spectrum is scanned by varying the amplitude of the voltages, while maintaining the ratio of RE to DC constant. The amplitudes of RE and DC voltages are chosen to transmit ions of only one mass-to-charge ratio at a time. All other mass-to-charge ratios have unstable trajectories and are lost from the system. Quadrupole mass spectrometry is therefore not a simultaneous detection method. Fast sequential scanning of many masses can be accomplished by varying the amplitude of the voltages on the quadrupole, while keeping the ratio of RE to DC constant.

The resolution of a mass spectrometer is its ability to distinguish between ions of slightly different mass values. Mass resolution for a quadrupole mass analyzer is set by the ratio of RE to DC voltages. The practical limiting resolution is given in Equation 2.




R = n Equation 2




where R is the resolution, n is the number of RE cycles the ions spend in the RE field, and c is a constant, usually 10 - 25 (91). If ions have too high a kinetic energy when they enter the quadrupole, the number of RE cycles is reduced, which results in degraded








resolution and non-Gaussian peaks that exhibit tailing towards the low mass end Quadrupole mass spectrometers are characterized as having low resolution in that they are only able to separate ions that differ in at least one atomic mass unit.



Detectors in ICP-MS

There are two main types of detectors used in ICP-MS, the Faraday cup and the secondary electron multiplier. These two detectors are shown schematically in Figure 6

(92). The Faraday detector consists of a small metal electrode placed in an open reflecting cup. A 1 count per second (cps) signal on the Faraday corresponds to 5 x 10.17 A, which is equivalent to 6 x 1010 ions per second arriving at the collector. The detecting capability of the Faraday plate is limited by Johnson noise of the measuring resistor and various mechanical contributions. The lower operating limit of the Faraday detector is 10- A.

Unlike multipliers, Faraday cups are robust, have long lifetimes, and can be exposed to the atmosphere without damage. However, they are inherently noisy, with background counts of about 2 x. 10 Another drawback of Faraday detectors is their speed. They are relatively slow when compared to multipliers.

The multiplier is the detector of choice for small signals, those below 10~6 ions/s (10-" A). Secondary electron multipliers (SEM) work on the principle of releasing electrons from surfaces that are bombarded by energetic particles (ions, electrons, neutral atoms, and photons). This phenomenon can best be described with the aid of Figure 7

(93). Ion current arriving at the detector is converted into an electron current on the first electrode, known as the conversion dynode. The electron beam is then amplified by









Faraday Collector
and Amplifier


Beam Deflectors


/


N


Alk


Quadrupole Mass Filter Electron Multiplier


Amplifier and Discriminator
for Ion Counting


Figure 6. Schematic of Faraday and Electron Multiplier Detectors


I


--I


0 A A
1 771

Voltage to Frequency Converter



I












Resolving Slit


- 3000 V


iCO





/~~- //\ Electrons N


Centering


- 5000 V


-1000 V


version nodes


Amplifier and
Recorder


- 600 V


I


- 2000 V


Figure 7. Schematic of Secondary Electron Multiplier.


I r% 0


E~JI I


Bea


m





36


secondary electron emission. One can obtain a gain of up to 108 with a single ion entering the multiplier. The number of charges per ion is directly proportional to the current that is measured and amplified. A single ion gives a pulse of about 108 electrons at the collector.

This pulse is sensed by a fast pre-amplifier. The output is routed to a digital discriminator and counting electronics, which only count pulses above a threshold value. Secondary electron multipliers are the detectors of choice for detecting low concentrations and therefore small signals with quadrupole mass spectrometers. They are fast with a high detection sensitivity, and can detect ion currents as low as 10-19 A. Background signals are usually 50 (cps) or less. However, the response is dependent upon the mass, energy, charge, and chemical nature of the incoming ions. In addition, they have limited lifetimes and can not be exposed to ambient air for long periods of time without damage.





Inductively Coupled Plasma Coupled with Mass Spectrometry


The interface of the Finnigan MAT SOLA ICP-MS used in our laboratory is shown in Figure 5. An ICP-MS undergoes differential pumping to achieve the various vacuum requirements. A high vacuum is needed in a mass spectrometer for two reasons. First, gas molecules in the mass analyzer region can cause scattering of analyte ions. Aside from the reduced ion transmission, this results in peak broadening and reduces the resolving power of the instrument. Background peaks can also appear due to residual gases in the mass spectrometer. This too can be minimized with a good vacuum.





37


The vacuum in an ICP-MS can be reached by various types of pumps including cryogenic, diffusion, mechanical, and turbomolecular pumps, The latter are the most widely used in modem instruments. Pumps work on the principle of molecular densities. A lower molecular density is established in the pumps than in other parts of the instrument. Molecules are then transported from the system into the pump.

The plasma is formed at atmospheric pressure in front of the sample cone. This cone is housed in a water-cooled interface to prevent melting and is located several millimeters from the end of the load coil. The region behind the sample cone is known as the expansion chamber and is pumped down to a vacuum of 2 - 3 mbar. The intermediate chamber, located behind the skimmer cone, achieves a vacuum of about I x 1 0- mbar.

The accelerator cone is found exclusively in the SOLA and is located 10 mm behind the skimmer cone. This third cone in the ICP-MS interface was included to minimize space charge effects (94). It also functions as a differential pumping aperture between the expansion chamber and analyzer chamber, thus minimizing losses due to ion beam scattering. In addition, the accelerator cone focuses ions that pass through it to a fine cross-over line 1 mm behind it, as shown in Figure 8 (94).

The quadrupoles are located behind the accelerator cone. The vacuum in the mass analyzer region is pumped down to about 10-' mbar. It can be seen from Figure 5 that the path the ions take from the plasma to the detector is not a straight one. This is employed to minimize the number of photons that reach the detector. Some instruments also use a photon stop for this purpose.
















Skimmer Cone


Accelerator Cone


Cross Over Point



'-


Figure 8. Ion Bean Trajectories Through the Accelerator Cone.


38





39


The ICP-MS interface is inefficient. Calculations by Hiefije et a. estimate that the sample cone passes only 25% of the gases in the plasma (95). In addition, the skimmer cone passes about 0.73% of the flow through the sample cone. More losses occur as the ion beam is transmitted through the ion optics and mass analyzer Houk et a. reported a single ion is actually detected for about every 106 analyte atoms in the plasma (96). Clearly much work needs to be done on the sampling interface to improve this technique.

The interactions of the plasma with the sample cone are shown in Figure 9 (78). A boundary layer of intermediate temperature is formed between the hot plasma and the cooled sample cone. Between the sampler and skimmer cones, the plasma expands adiabatically and forms a supersonic jet. In the jet, the density and temperature of the gas decrease while the speed of the gas stream increases above the speed of sound. The ions are accelerated to the skimmer cone at supersonic speeds and are thus altered little during their transport to this region. The ions reach the skimmer cone in microseconds after leaving the sample cone. The supersonic jet is shown in detail in Figure 10 (97). The expansion is surrounded concentrically by a barrel shock and perpendicularly by a shock wave called the Mach disk. Shock waves result from collisions between atoms from the jet traveling at fast speeds and the background gas, which functions to reheat the atoms and induce emission. The region inside the shock structure is called the zone of silence. The gas flow in this region is well-characterized. The location of the Mach disk from the sampling orifice can be calculated by Equation 3:





40


Boundary Layer
-1000 K -760 torr


Plasma Stream
-5000 K -760 ton


*0


< 500 K
-3

Supersonic Jet


Figure 9. Interactions of Plasma with ICP-MS Interface. (Adapted from 78)




41


S


Mach Disk




one of Hence


Barrel Shock


Figure 10. Supersonic Jet Expansion. (Adapted from 97)


X


P


D






42



XM= .67D0O PO Equation 3




where X~j is the distance between the Mach disk and the sample cone, Do is the sampling orifice diameter, Po is the ICP pressure, and P1 is the background pressure in the expansion chamber (98). The tip of the skimmer cone is placed inside the Mach disk to reduce the effects of collisions and scattering.

One inherent problem with ICP-MS is a phenomenon known as space charge effects, the result of coulombic repulsions due to a localized volume of positive ions. The plasma is essentially neutral since the ion population is balanced by the electron population. However, the ion optics in the reduced pressure regions are configured for the collection of positive ions. As the ion beam leaves the last cone, the accelerator in the SOLA, the electric field induced by the transfer optics attracts positive ions and repels electrons, which fall out of the ion beam. The ions in the resulting positive beam experience repulsions with each other and produce a defocused ion beam. The lighter ions, with lower kinetic energy, are effected more strongly than the heavy ions and are subsequently lost from the ion beam. Space-charge effects are mass dependent and can have a profound effect on signal intensities when isotope ratios are monitored. Space charge effects can be minimized by decreasing the number of ions in the resulting ion beam. This can be accomplished by detuning the ICP-MS ion optics, operating the ICP so that fewer ions are sampled, or diluting the sample solution. All of the above methods decrease the sensitivity for individual ions.





43


Background ions

The most prominent species found in an Ar ICP are neutral argon atoms at about 1018 cm-3 (99). Ar ions are present at 101 cm-3. Oxygen and hydrogen atoms are the next most abundant species, having a number density of 1015 - 1016 cm-3 Other background ions observed are nitrogen, carbon, and combinations of these with argon and oxygen. Several of the more prevalent polyatomic species are given in Table 3,

Metal oxides are the most commonly formed analyte polyatomic ions observed in the ICP. They can be formed as a result of numerous interactions involving the plasma gases, air entrairnent, solvent, and matrix constituents.

Molecular ions observed in the mass spectrum are thought to be formed during the ion extraction processes. The temperature of the plasma is large enough to dissociate any polyatomics that may form. There are theories as to how molecular ions survive the energetic plasma. It is known that temperatures in the plasma are not uniform, temperature gradients can be up to a few thousand degrees. It is thus believed that some polyatomic ions are formed in the cooled boundary regions of the plasma. Some polyatomic species such as N2' and 02' are most probably formed in the plasma and not dissociated.

Another theory for polyatomic ion production suggests the formation of oxide ions in the cooled expansion chamber, located behind the sample cone. Douglas and French have estimated that about 250 collisions occur between Ar neutral atoms and other species in the expansion chamber (100). Polyatomic ions, such as molecular oxides, have lower





44


Table 3. Prevalent Polyatomic Ions Observed in the ICP-MS.




Ion m/z

0- 16

OH' 17

OH2+ 18

,AX21 20

N' 14

NH' 15

N2' 28

N2HW 29

NO+ 30

NOW 31

02+ 32

Ar+ 40

ArH+ 41

Ar0+ 56

Arc+ 52

ArN+ 54

Ar2+ 80

ArN2+ 68





45


kinetic energies than elemental ions of similar mass. Thus, there is time for polyatomic ions to form in the expansion chamber before entering the mass analyzer.



Interferences in ICP-MS

One of the major characteristics of ICP-MS that limits the determination of certain elements is interferences. Interferences can be classified as spectroscopic or nonspectroscopic (matrix effects). Spectroscopic effects include species such as isobaric overlapping ions and polyatomic ions. An isobaric interference occurs when two or more elements with similar mass isotopes are in a sample matrix. A quadrupole instrument, with only unit resolution, would be unable to distinguish between the two masses. An example is the determination of 4'Ti in a matrix containing 41 Ca. It is not possible, with a quadrupole ICP-MS, to analyze trace amounts of 4'Ti if significant amounts of calcium are present in the sample.

Polyatomic ions are the combination of two or more elements. The classical example is 40M' 60, interfering with the determination of '6 Fe. Many solvents add to the production of polyatomnic ions, as discussed above. Nonspectroscopic interferences are characterized by a large amount of dissolved solids in the sample, or a signal enhancement or suppression. The former can cause short term drift in the signal. Signal enhancements and suppressions are matrix dependent and element specific.

Spectroscopic and nonspectroscopic interferences can be reduced by modifying the sample introduction scheme (e.g. using a dry plasma or desolvating the liquid sample), instrument optimization, mixed gas plasmas or solvents, and instrumental design.





46


Sample introduction

Gas, solid, and liquid samples can be introduced into the plasma for mass spectrometric detection. The outlet of a gas chromatographic column can be coupled with the inlet of the torch to allow the analysis of gaseous samples (101). Other means of introducing samples already in the vapor phase into the plasma include supercritical fluid chromatography (SFC) (102) and hydride generation (HG) (103). Solid samples have been introduced into the ICP for mass spectrometric detection by various sampling methods such as laser ablation (LA) (104), slurry nebulization (105), arc nebulization (106), electrothermal vaporization (ETV) (107), and direct insertion (DI) (108).

Liquid sample introduction was discussed in detail earlier in this chapter. Solutions are normally converted into an aerosol for introduced into the plasma. This is usually accomplished by a ultrasonic (109), pneumatic (110), thermospray (Ill1), or direct injection nebulizer (DIN) (112). Ultrasonic and pneumatic nebulizers are by far the most popular means of introducing liquid samples into the ICP-MS.

Ultrasonic nebulizers have seen widespread use for aerosol production due to their high efficiency. The sample is fed across a piezoelectric crystal that is vibrating in the range of 0. 2 - 10 MIHz. As the sample stream is passed over the crystal, large droplets are broken down into smaller ones. Often the aerosol is then passed through a heating region, which produces dried analyte particles and solvent vapor. A condenser may then be connected to eliminate solvent vapor from the sample aerosol, while the analyte particles are sent to the plasma. Ultrasonic nebulizers produce smaller, more uniform droplets and are less prone to blockage than pneumatic nebulizers. In contrast to pneumatic nebulizers,





47


the high aerosol production efficiency of ultrasonic nebulizers is independent of nebulizer gas flow. This results in better sensitivity and lower detection limits since the analytes have longer residence times in the plasma. The main limitation hindering their use is cost.

The pneumatic nebulizer has been most used with ICP-MS due to its low cost, convenience, and stability. The principle behind these nebulizers is the disruption of a liquid stream by a high velocity gas, resulting in a fine aerosol of sample. Cross-flow nebulizers consists of two capillary tubes located perpendicular to each other. Liquid sample is drawn up one of the tubes and a high-velocity stream of gas is flowed through the other. When the liquid reaches the top of the capillary it is broken into a finely dispersed aerosol. Cross-flow nebulizers are used when the sample solution contains large concentrations of salt, as they are less prone to capillary blockage.

A second type of pneumatic nebulizer that has been used with ICP-MS is the Babington-type nebulizer. The aerosol is produced by flowing a film of liquid sample over the surface of a sphere and flowing gas through an aperture below the film. Because of memory effects, Suddendorf el a[. made modifications to the Babington-type design (113). The new design consisted of constraining the liquid sample in a V-groove and introducing the nebulizer gas from a small hole in the bottom of the groove. This type of nebulizer is resistant to blockage and is thus used when slurry samples need to be analyzed. The most widely employed pneumatic nebulizer for liquid introduction into the plasma is the concentric nebulizer, shown in Figure 11. Liquid sample is introduced perpendicular to the nebulizer gas. The nebulizer gas usually flows through the nebulizer at about 1 L/min. Pneumatic nebulizers are simple to use and are relatively inexpensive. They are, however,






48


characterized by a 1 - 2% efficiency. A peristaltic pump is used to produce a constant liquid flow into the nebulizer. Flow rates lower than 1 mL/mi~n may not allow the aerosol to penetrate the plasma, while higher uptake rates may not provide adequate residence times for the sample particles in the plasma. The resulting aerosol produced contains sample particles with diameters up to 100 4im. Small particles, less than 10 pim, are desired for rapid desolvation, volatilization, and ionization in the ICP. A spray chamber placed between the ICP and nebulier functions to expel large particles from the aerosol. The most popular spray chamber in use is the Scott-type double pass, shown in Figure 12. As the sample aerosol particles enter the spray chamber, the heavier ones fall to the lower walls due to gravitation and are pumped to waste. Smaller particles remain in the argon flow and are carried to the torch. Scott-type spray chambers can be cooled to condense water vapor out of the sample aerosol before it enters the plasma. This results in a decreased solvent loading in the plasma and subsequently decreased background interferences.

The sample uptake rate for pneumatic nebulizers is described by Poiseuille's equation, shown in Equation 4:




7dR4p
Q = -Equation 4
8 i7L



where Q is the liquid flow rate, R is the capillary radius, P is the pressure differential, Yj is the viscosity of the liquid, and L is the capillary length (114). The above equation




49











Aerosol Liquid ** Gas


Figure 11. Schematic of a Concentric Nebulizer.





50


Sample Aerosol In


47


Sample
Aerosol Out 4 Coolant Out 4


-* Coolant In Waste


Figure 12. Scott-type Double Pass Spray Chamber.


7


4





51


assumes the velocity of the liquid at the capillary wall is zero. Actually, the liquid slips over the capillary and therefore a correction must be added to the Poiseuille equation. The R 4 term must be replaced by R4 + 4T1R3/B, where B is the coefficient of sticking friction of the liquid on the capillary wall.

Sample aerosol exits the spray chamber and enters the plasma through the injection tube of a torch. Torches used with ICP-MS are normaly based on the Scott Fassel design (115). Shown in Figure 13, the present design incorporates a 100 mm long quartz torch positioned horizontally. It has an inner diameter of 18 mm and encases two concentric tubes with diameters of 13 mm and 1.5 mm. The coolant gas and intermediate gas are bled in through the side to the torch, whie the nebulizer gas carries the sample aerosol through the center. The torch is usually positioned 10 - 15 mm. from the sample cone.

A iquid sample undergoes several processes as it passes through the nebulizer and spray chamber to the plasma (116, 117). The first process is the generation of a primary aerosol by the nebulizer, which then experiences impaction with the spray chamber to form a secondary aerosol. A tertiary aerosol which is formed by evaporation, impaction, and droplet shattering, coagulation, inertial deposition, gravitational settling and turbulence induced losses, is produced after passage through the spray chamber.

ICP-MS has many advantages including rapid, efficient sample introduction methods, a large degree of ionization of most elements, simple spectra, multielemental. capabilities, low detection limits, and a broad linear dynamic range. As with any analytical technique there are limitations that hinder detection. The ICP-MS is no exception. The interface between the plasma and mass spectrometer is highly inefficient. Only about 1 %





52


100 MM


Sample Aerosol t 18 mm ID
and Nebulizer Gas




Intermediate t Gas Coolant
Gas


Figure 13. Demountable Quartz Torch.


I


I


I






53


of the ions that enter the sample cone exit the skimmer. There are many losses associated with transmitting ions through the ion optics and mass analyzer. The detection efficiency of ICP-MS is extremely low. For every 1 0' ions produced in the plasma, only one reaches the detector. In addition, this technique is expensive to purchase and maintain, it can not accommodate a high degree of total dissolved solids in the sample, and it is subject to numerous interferences.














CHAPTER 3
MICROWAVE DIGESTION



Introduction

Sample preparation procedures are often the most laborious and time consuming steps of an experimental study. This is especially true when the sample is in a nonaqueous matrix. As discussed in Chapter 2, direct analysis of organic samples by ICP-MS is not a trivial matter. The determination of trace levels of elements such as titanium and chromium are impossible by ICP-MS without destruction of the organic matrix.

Metallo-organic compounds can be decomposed to produce aqueous solutions of their respective elements by conventional or microwave heating. Conventional sample dissolution in an open container on a hot plate has been used extensively for the extraction of metallic species (118,119). This type of sample decomposition, however, is prone to numerous losses and contamination. In addition, wet ashing techniques generally require long heating times, with many protocols lasting several hours.

Microwave energy can be used to heat a sample faster and more efficiently than conventional methods such as wet ashing. Abu-Sanira et at were the first to use microwave heating for organic matter decomposition in 1975 (120). Solid samples were placed in open Erlenmeyer flasks or test tubes with hydrogen peroxide and either nitric acid or a nitric acidlperchloric acid mixture. Complete analyte extractions were carried


54





55


out within minutes. This was significantly less than the time required for conventional methods.

The differences in heating times between conventional and microwave digestions can be explained in part due to the heating mechanisms. In conventional systems, a hot plate is used to conductively heat a sample. The sample vessel first absorbs the energy, and then transfers it to the sample. The sample vessel is usually a poor heat conductor, thus resulting in increased heating times. In addition, because vaporization occurs at the liquid surface, convection currents produce a thermal gradient in the sample solution. Thus, only a small fraction of the liquid is at the temperature of the applied heat. As a result, only a minor portion of the sample is above the boiling point of the bulk solution.

Microwave energy, on the other hand, directly heats the entire sample simultaneously without first heating the vessel. The energy travels through a microwave transparent vessel to the sample. The solution reaches its boiling temperature very quickly, which reduces digestion times. Open vessel microwave digestion procedures can be completed in 5 - 15 mmn.

It took nearly ten years after the inception of open vessel microwave digestions for closed vessel microwave heating to be explored as a more effective means of decomposing biological materials (121,122). Smith et al. were the first to use PTFE closed sample vessels for microwave digestions (123). Sulfide samples from mines, mills, and smelters were digested for the determination of copper and nickel. Samples were digested in just 3 min. This was a dramatic improvement over the I - 1.5 hr procedure previously used to





56


dissolve these samples. Although the samples were not completely digested, analyses of the residues showed no trace of copper or nickel.

Sealed vessels offer several advantages over open vessels for Microwave digestion procedures. The energy and rate of a reaction increases with increasing temperature. In closed systems, higher temperatures can be reached which leads to faster reaction times and the decomposition of difficult samples. In addition, the loss of volatile species and the risk of contamination by airborne particles are eliminated by using closed vessels. These systems also give reduced blank levels. This is due in part to the smaller quantities of digestion aides that are used.

Today, closed vessel microwave acid digestion is being used for a wide variety of applications such as biological (124), geological (125), metallurgical (126), environmental (127), and marine (128) samples. An extensive description of microwave sample preparation was given by Kingston and Jassie in the first book on this subject (129). This is still considered the handbook of microwave dissolution.

Microwaves have frequencies in the range 300 - 300,000 MfHz. The Federal Communications Commission established four microwave frequencies that are allowed for industrial, scientific, and medical use. Conventional miicrowave devices used in most homes are operated at 2450 + 13 M4Hz with output energies of 600 - 700 W.



Theory

The dielectric loss theory is used to explain the principles of microwave heating. When microwave energy penetrates a sample, a loss of energy from the electromagnetic





57


field to the sample occurs. The amount of loss depends on the composition of the sample. The rate of this energy loss is dependent upon the sample's dissipation factor, which is the ability of a sample to dissipate microwave energy as heat. The dissipation factor is the ratio of the sample's dielectric loss factor to its dielectric constant, as shown in Equation 5



E"
tan6 5 Equation 5




where tan 8 is the dissipation factor, 6' is the dielectric loss factor, and F' is the dielectric constant (130). The dissipation factor is dependent upon temperature and microwave frequency. A dielectric material is one that obstructs the passage of energy through it due to its composition. The dielectric constant measures the ability of the sample to obstruct energy. The dielectric loss factor measures the ability of a sample to dissipate microwave energy as it penetrates the sample. The use of "loss" here is related to the amount of incoming microwave energy that is lost to the sample by being dissipated as heat.

The greater the dissipation factor, the less microwave penetration at a given temperature and frequency. This is because samples with large dissipation factors rapidly absorb and dissipate microwave energy as it penetrates it. Thus, the greater the dissipation factor, the greater the sample's ability to convert microwave energy into heat. Samples with a significant dipole moment and free rotation are expected to possess large dissipation factors. For example, ice and liquid water have dissipation factors of 0.0009, and 0. 15 79, respectively. The crystal lattice structure of ice restricts the mobility of water and thus makes alignment of the molecules with the microwave field difficult. This






58


explains why liquid water is superior to ice in converting electromagnetic energy into heat. Polytetrafluoroethylene (PTFE), commonly known as Teflon, has a dissipation factor of 0.00015, This is the material of choice for many digestion vessels since it is virtually transparent to microwave radiation.

Microwave energy is lost to a sample in two ways: ionic conduction and dipole rotation. When an electromagnetic field is applied to a sample solution, a conductive migration of ions occurs, which produces a current flow. The solution's resistance to free flow of these migrating ions results in friction, which in turn heats the solution. Although all ions in solution contribute to conduction, the relative concentration of an ion and its mobility in the system governs the magnitude of current carried by it. As a result, size, charge, and conductivity all effect the losses due to ion conductance. As the concentration of an ion increases, current and heat transfer also increase due to greater ion mobility.

The second mechanism in which microwave energy is lost to a sample is dipole rotation. This refers to the alignment of molecules in a sample that have permanent or induced dipole moments. When an applied electric field is increased, the polarized molecules align themselves with the positive poles of the molecules facing the negative poles of the field, and vice versa. Disorder of the molecular arrangement occurs when the microwave energy is removed. At 2450 MlHz, this alignment, followed by disorder, occurs at 4.900 GHz (13 1). The result is very rapid heating due to molecular fiction. Heating due to dipole rotation is dependent upon temperature, viscosity of the sample, and





59


the sample's dielectric relaxation time. The dielectric relaxation time is the time it takes for the molecules to reach 63% of their return to disorder.

While both ionic conductance and dipole rotation influence the efficiency of microwave heating, the temperature determines which process dominates. Initially, dipole rotation controls the dielectric loss to the sample. As the temperature is increased, however, ionic conductance dominates the conversion of microwave energy into heat, The fraction of contribution from each process depends on ionic mobility and concentration. If these two parameters are small, then the dielectric loss is mainly the result of dipole rotation.




Instrumentation

The essential components of a microwave digestion system are the microwave generator (magnetron), waveguide, microwave cavity, mode stirrer, and circulator. A typical microwave instrument is shown in Figure 14. The magnetron produces microwave energy, which then travels through the waveguide to the microwave cavity. The waveguide is constructed out of a reflective material for high transport efficiency. At the exit of the waveguide is the mode stirrer, a fan-shaped blade that distributes the energy in many directions throughout the microwave cavity. Once inside the cavity, the microwaves are reflected back and forth between the walls until all the energy has been absorbed by the sample. Reflected energy can exit the microwave cavity and traverse its path back through the waveguide. A terminal circulator is usually placed at the entrance of the









Dummy Load
I I


Waveg uide


Stii


4

4-


ide ,,er


ample l/essel


I ~.I


(/7)4


Circulator


Magnetron


Figure 14. Schematic of a Microwave Oven


C


A
I I


W






61


waveguide to direct reflected power to a dummy load. This enables the microwave energy to dissipate as heat and prevents damage to the magnetron.



Microwave Oven Calibration

The actual power absorbed by a sample in the microwave oven can be determined by measuring the change in temperature of a given quantity of water. The relationship between power absorbed by the sample and temperature change is shown in Equation 6



P _ CPK KAT m Equation 6




where P is the apparent power absorbed by the sample (W), Cp is the heat capacity of water (cal/g - C), K is a conversion factor from calories to Joules (4. 184 J/cal), AT is Tf Ti ('C), m is the mass of the sample (g), and t is the time (s).

The microwave oven used for sample decomposition was calibrated to determine the approximate power delivered to the oven cavity from the magnetron at each setting on the oven display. Equation 6 can be used for this purpose if it is assumed that the majority of the power delivered to the microwave cavity is absorbed by the sample. The microwave oven was calibrated by measuring the temperature difference of 9 mL of water at each setting on the microwave oven. This sample volume was chosen since it closely simulates an actual sample size. Since heat capacity is dependent upon temperature, it was first necessary to set up a table relating Cp to T. The CRC Handbook contained heat









capacity values at various temperatures, shown in Table 4 (132). However, more exact power calculations could be obtained using smaller increments between points. The values listed in Table 4 were inserted into graphing software to generate a curve and an equation describing it. The generated curve is shown in Figure 15. Equation 7 was used to obtain exact heat capacity values at the various experimental temperatures.



CP= 4.22-O0.00385x +1.74e X2 - 5.49ex3 + 1.40e-7k4 _ Equation 7
2.75e-9x5 + 3.78e-1"X6 - 3.33e'13X7 + 1.67e 5x8
3 .58e'8x9



The CRC Handbook also lists values for the density of water at various temperatures, shown in Table 5. Again, smaller increments would lead to a more exact calculated absorbed power. The values in Table 5 were inserted into a graphing program and a curve was generated. The curve is shown in Figure 16 and the equation describing the curve is given in Equation 8.



p = 1.005 - 0.00212x + 3.58e -X2 - 3.45e-'x3 + 2.02e -X4 _ Equation 8
7.57e-8X5 + 1. 82e-9X6 - 2.72e-1 x7 + 2.26e1 3x8
7. 98e 'x9



The same digestion bomb and PTFE sample vessel used for the metallo-organic digestions was used for the calibration of the microwave oven. Display settings I through 5 were calibrated by measuring the temperature rise in 9 niL of water. Each measurement






63


Table 4. Heat Capacity (J/g * K) at Various Temperatures ('C). (132)


Temperature (0C) Heat Capacity (JIg * K)


4


0

10

20

30

40 50

60 70

80 90 100


4.2 176 4.1921 4. 1818

4.1784 4. 1785

4. 1806 4. 1843 4. 1895

4. 1963 4.2050 4.2 159


I.


Temperature (*C)


Heat Capacity (J/g * K)






64

















4.22



-~4.21~34.20S4.194.18


0 20 40 60 80 100

Temperature (OC)


Figure 15. Heat Capacity (JX/g) versus Temperature ('C).






65


Table 5. Density (glcm') at Various Temperatures ('C). (132)


Temperature (T) I Density (g/CM3)


0.99984 0.99970 0.99821 0. 99565 0.99222 0.98803 0.98320 0.97778 0.97 182 0.9653 5 0.95840


0

10

20 30

40 50 60 70 80 90 100














































a


0O 20 40

Temperature (oc)


Figure 16. Density (g/cm') versus Temperature (0C).


66


1.00



0.99 0.98 027







Og


C.) 0)

U)
C
G)
0


U>


N


a,


0


810 , 100o






67


was taken in triplicate. The results are shown in Figure 17. After each measurement the PTFE sample vessels were cooled with deionized water.



Microwave Acid Digestion Vessel

Microwave acid digestion bombs obtained from Parr Instrument Company, Moline, EL, were used for sample decomposition. The sample cups were constructed out of microwave transparent, thick-walled PTFE. This material is chemically inert, with a melting point of - 306 'C. PTFE also serves as an insulator to confine the heat inside the sample vessel. There is, however, one limitation of using sample vessels constructed of PTFE. This material is known to creep or flow when subjected to high pressure or load. At temperatures less than 150 'C, the effect is negligible, however at higher temperatures it becomes more difficult to maintain tight seals. The result is deformation and a shorter life span of the components.

The outer portion of the bomb assembly that encased the sample vessel was constructed of a microwave transparent, polymer resin. The bomb and liner are shown schematically in Figure 18. The outer mantle contained a compressible pressure relief disc that served as a safety release. The temperature and pressure of the digesting sample should not be allowed to exceed 250 'C or 1200 psi. Values greater than these may result in the sample vessel exploding. If the pressure inside the vessel exceeds 1500 psi, the relief disc compresses the 0-ring seal on the sample vessel and it is blown out. This design prevents total destruction of the bomb. The retaining screw can be used as a crude pressure indicator. The head of the screw is normally flush with the top of the bomb cap.






68


120100

X

__8060




20

-240 20


0 1 2 3 4 5 6

Microwave Setting


Figure 17. Plot of Power Absorbed by the Sample (W) versus Mlicrowave Setting.





69


Pressure Screw


Screw Cap


Outer Casing Teflon 0-ring


elief Disc




Teflon Cover

_____Teflon
Sample Vesse Bottom Plate


Figure 18. Schematic of Microwave Bomb and Liner.






70


The screw extends out of the bomb cap approximately 1/32 in for every 500 psi pressure increase within the sample vessel. The 0-ring has the potential of being blown out after a rise in the screw head of greater than 1/ 16 in.















CHAPTER 4

DETERMINATION OF WEAR METALS IN AQUEOUS SOLUTIONS



Instrumental Set-up

A Finnigan MAT SOLA ICP-MS was used for the elemental analysis of all solutions. A thorough description of the hardware was given in Chapter 2. Specific components are given below.



Inductively Coupled Plasma

The inductively coupled plasma was sustained by a Henry Electronics Radio Frequency Power Generator, Model 2500D (Los Angeles, CA). The RF power was maintained at a frequency of 27.12 MHz, with output powers up to 2.5 kW. The copper induction coil carrying current to the plasma region was obtained from two sources: the chemistry department machine shop and Finnigan MAT.

Chilled water flowed through the copper induction coil at a rate of 3 L/min. The temperature of the water in the system was kept constant at 12 'C by a Neslab (Newington, NH) CFT-75 refrigerated recirculator.


71






72


Quadrupole Mass Spectrometer

The hardware of the ICP-MS including ion optics, mass analyzer, and detectors was described extensively in Chapter 2. The quadrupole mass analyzer used in the SOLA was a Balzers QMS 511. The ICP-MS was equipped with a pulse counting channeltron electron multiplier, Model 4870, from Galileo Electro-optics Corporation (Sturgridge, MA) and a Faraday cup detector.



ICP-MS Interface

The face plate housing the sample cone was kept cool by the same refigerated recirculator used to supply chilled water to the induction coil. The sample cone was fabricated of nickel and copper due to their high thermal conductivity and electrical conductivity, stability, and machinability. The skimmer and accelerator cones were constructed of nickel. The sample cone was located 14 mm from the end of the load coil and had an aperture of 1. 1 mmn. The expansion chamber was pumped down to a vacuum of 2 - 3 mbar by an 18 m3/hr single stage Edwards rotary pump. The skimmer cone was located 8 mm behind the sampler and had an aperture of 0.8 mm. A vacuum of 1 x 10-' mbar was achieved in the intermediate chamber by means of a Baizers TPH 330 turbomolecular pump (Hudson, NJ).

The accelerator cone had an aperture of 1 mmn and was maintained at a voltage of +2 kV. The vacuum in the mass analyzer region was pumped down to 4 x 10-5 mbar by a Baizers TMH 260 turbomolecular pump. A Balzers TPH 062 turbomolecular pump was used to reach a vacuum of < 5 x 10-5 mbar in the detector housing. All three






73

turbomolecular pumps were backed by a 12 m3/hr Edwards mechanical pump. The above pressure values were obtained with Penning ionization gauges, which measured particle densities. Pressures in the corresponding regions were displayed by a Tylan General FC2900V Series mass flow controller (Torrance, CA).

The ICP-MS hardware and computer software were interfaced with a Mitsubishi FX Series Programmable Logic Controller (PLC) (Tokyo) through an RS422 interface port in a Gateway 2000 498DX 66 MFiz personal computer. The PLC controlled interlocks that ensured proper operating conditions to prevent damage to the hardware. The SOLA would shut down if any one of these interlocks were not met. Interlocks found on the SOLA included gas and water flow; coolant and intermediate gas flow; correct positioning of the torch box; adequate pressure in the expansion or intermediate housing; proper operating speed of the turbomolecular pumps in an allocated amount of time, and proper fitting of all instrument panels and lids.

The transfer ion optics were optimized every day with a 1 ppm indium tuning solution. Typical potential values for the optics are given in Table 6.

It was often necessary to put the ICP-MS in standby mode to obtain an acceptable vacuum in the analyzer region within a reasonable amount of time. Normal procedures for lighting the plasma and waiting for the pressure in the analyzer region to fall to 5 x 10 mbar were not always successful. Pumping in standby mode would sometimes enable the ICP-MS to reach an appropriate vacuum in less time (thirty minutes or less). The SOLA was not equipped with a true standby mode in which the vacuum pumps are operated without the plasma being lit. Therefore, a piece of black electrical tape was used to cover






74


Table 6. Typical Transfer Ion Optic Potentials.


Transfer Optic Helipot Setting Voltage

Beam Extraction 2.90 -1500

X Deflection 6.52

X1 -1470

X2 -1530

Y Deflection 4.16

Y+ -1368

Y- -1632

Y Steer 6.24

S+ -1302

S- -1698

Focus 5.60 +89.6

Phase Matching Lens 3.92 -205

Interspace 9.68 +14






75


the sample cone aperture and standby mode was selected from the software, The mass analyzer was able to reach a vacuum of 10-6 mbar in this mode However, since there was no slide valve on the SOLA, the vacuum was broken once the electrical tape was removed, allowing air to be vented into the mass spectrometer. Therefore, the standby mode was only useful if the plasma was lit immediately after removal of the tape.




Sample Introduction

Liquid samples were introduced into the inductively coupled plasma through a recessed tip concentric nebulizer (Precision Glassblowing of Colorado, Englewood, CO). The pneumatic nebulizer accommodated an inlet gas pressure of 30 psi (1 L/min Ar) and a liquid flow rate of 1 mLlmnin. The sample uptake flow was controlled by a Gilson Minipulse 3 peristaltic pump (Middleton, WI). Large solvent droplets were separated from the analyte aerosol by a Scott-type double pass spray chamber (Precision Glassblowing of Colorado, Englewood, CO). The spray chamber contained a I I mixture of antifreeze and water to condense solvent vapor from the sample aerosol. The mixture was kept at a constant temperature of 0 TC by a Neslab (Newington, NI-) RTE- 10 refrigerated bath/circulator.


The sample aerosol entered the plasma through a Fassel-type quartz torch, which contained a demountable injector tube (Precision Glassblowing of Colorado, Englewood, CO). Figure 19 ilustrates the configuration of the nebulizer, spray chamber, and torch. The injector tube of the torch was connected to both the nebulizer and a Nd:.YAG laser ablation sample chamber. The laser ablation chamber was used for separate studies not





















Torch


Spray Chamber


Figure 19. Sample Introduction Set-up Including Nebulizer, Spray Chamber, and Torch


ON


Nebuizer To Laser Ablation NebuizerChamber


Sample Inlet







77

described here. The argon gas flow through the laser was set at 0.00 L/min while the nebulizer gas flow was set to about 1.00 L/min.






Experimental Parameters




Chemicals

High purity water, obtained from a Millipore MilliQ water purification system (1 8M!Q cm specific resistivity) and HPLC grade water (Fisher Scientific, Pittsburgh, PA), were used for preparation of all solutions (Fisher Scientific, Pittsburgh, PA). Optima nitric acid was obtained from Fisher Scientific and used as received. All aqueous standard solutions were made from dilutions of stock solutions. Solutions were made both in the presence and absence of 5% optima nitric acid. A 10 j tg/mL multielement plasma standard (SPEX Industries Inc., Edison, NJ) was used for solutions of Ni and Cr. Aqueous solutions of Al, Fe, Mg, and Ti were prepared by serial dilutions of individual 10 p4g/mL single element standards (High Purity Standards, Charleston, SC). A 1006 .ig/mL Mo standard (SPEX Industries, Inc., Edison, NJ) was used for solutions of this element. Aqueous solutions of indiumn were made by diluting a 1000 Ag/mL plasma standard (SPEX Industries Inc., Edison, NJ). All solutions were made in glass volumetric flasks and immediately transferred to PTFE or plastic bottles.






78


Glassware

Glassware and storage bottles used for all solutions were washed with LIQUINOX (Alconox Inc, New York, NY) and ultrapure water. They were soaked in approximately 20% nitric acid for at least twenty-four hours and then rinsed once again with ultrapure water.



Experimental Parameters Optimization

The peristaltic pump was calibrated to determine the relationship between the setting given on the pump and the actual uptake rate. All tubing connected to the pump was filled with ultrapure water. The coolant, intermediate, and nebulizer argon gas supplies were switched on, the pump was started, and the solution was allowed to flow through the tubing while the water collection was timed. This process was carried out for I min for various pump settings. The results are presented in Table 7. One can see that a solution uptake rate of 1.0 mL/min is accomplished with the peristaltic pump operating at a setting of about 9.90. This setting was used for subsequent experiments.

The efficiency of the concentric nebulizer was also determined by comparing the amount of solution that exits the nebulizer with the amount that is pumped into it. The amount of water used for introduction and collection was determined by weight. A Mettler AE 200 (Mettler Instrument Corporation, Heightstown, NJ) was used for all weighings described here and in later chapters, unless otherwise noted. A dry volumetric flask was weighed before and after the addition of water. The temperature of the water







79


Table 7. Calibration of the Peristaltic Pump.


Peristaltic Pump Solution Uptake Rate

Setting (mLlmin)

4.50 0.545

9.00 1.09

9.91 1.20

10.5 1.27

11 1.33







80

was found to be 24 'C. The weight was converted to volume by calculations involving the density of water.

Collected water was also subjected to the above weighing procedure. The efficiency experiment was repeated for two other solution uptake rates. The results are shown in Table 8. It can be seen that the nebulizer is most efficient at solution uptake rates around 1 mLlmin.

The percent relative standard deviations (%RSD) obtained from mass spectra signal intensities of a 20 ppb multielemnent solution (SPEX Industries, Inc., Edison, NJ) were monitored for several elements to determine the optimum solution uptake rate. The results are given in Table 9. Although the nebulizer is characterized by an increased efficiency at higher flow rates, the sensitivity is consequently reduced at these operating parameters. This can be explained by collisions and coalescence of aerosol droplets (83). Even though the nebulizer is more efficient, the larger particles will have fallen out before they completely traverse the spray chamber, thus reducing the amount of analyte that reaches the plasma.

The nebulizer gas flow was also varied to determine which value gave the maximum signal for a multielement solution. The results are shown in Figure 20. It appears that a wide range of values will give close to maximum signals for several elements. At the time this experiment was performed, the mass flow controller displaying the nebulizer gas flow was somewhat erratic. The values given as optimum flow therefore could not be taken as concrete. The problem of fluctuating mass flow meter was later






81


Table 8 . Determination of Nebulizer Efficiency.


Nebulizer Efficiency


Solution Uptake Rate

(mL/min)


0.43 % 3.5 % 1.6 %


0.6

1.2 2.4






82


Table 9. Effect of Solution Uptake Rate on %RSD for Various Masses.


Solution Uptake Rate % RSD

(mLlmin) Be Mg Ni Co In Ce

0.6 6.1% 2.6% 2.5 % 2.3 % 2.2 % 5.5 %

0.9 3.7% 3.5% 2.2 % 4.1 % 3.7 % 4.4 %

1.2 4.5% 4.8% 8.1 % 3.1 % 4.3 % 3.0 %

2.4 5.5 % 9.8% 5.4 % 3.7 % 4.6% 12.3%





















2.6x106 2.4x 106 2.2xJ06 2.0X106 1.8xl 106 l.6xl 106 1. 4x 16 Q 1.2xl 106



8. Ox 10 6. Ox 103 4. Ox 1 QS

2.0x105-0.88


-UBe
--Mg


-UIn


-4Ce






(so


Figure 20. Effect of Nebulizer Gas Flow Rate on Analyte Signal.


801


0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04

Nebulizer Flow Rate (L/min)






84

circumvented with by-passing the pressure regulator in the gas box. The above experiment was repeated and similar results were obtained.

There was an initial increase in analyte transport rate as the flow rate of the nebulizer gas was increased. This is due to an increase in the number of small diameter droplets in the primary aerosol as well as a decrease in the number of droplets that are lost in the secondary and tertiary production processes. After an optimum nebulizer gas flow rate was achieved, a decrease in the signal was observed with increasing flow rates. This was due to the droplets experiencing smaller residence times in the plasma. The large flow rates pushed the particles though the plasma before complete ionization could occur. In addition, a loss of all droplet sizes occurs in the spray chamber, which reduces the transport efficiency.

Optimization of the incident RF power was attempted, but was not successful. The SOLA had a very small incident power window that could be utilized and still maintain a stable plasma. The ICP-MS used for these studies was not meant for use with "Ccool" plasma conditions (incident powers < 1000 W). As a result, all work was carried out heeding the instrument manufacturer's suggestion of operating the plasma at a forward RF power of 1300 W.

The above results were examined to determine the optimum operating parameters for mass spectra acquisition. The experimental parameters used are shown in Table 10.







85


Table 10. ICP-MS Operating Parameters.


Parameter Value
Plasma
Forward Power 1300 W
Reflected Power <5 W
Coolant Gas 15 L/min
Intermediate Gas 0.9 L/min
Nebulizer Gas I L/min
Solution Uptake 1 mL/min

Data Acquisition
Dwell Time 64 ms
Number of Channels 16
Number of Scans 1







86


Acid Effects

A study to determine the effects, if any, of nitric acid concentration on analyte signal was carried out on aqueous solutions. Nitric acid is the most widely used acid for the preservation of aqueous solutions, and therefore investigated in this study. Several aqueous solutions were made with the analyte concentration held constant at 200 ppb The concentration of the nitric acid was varied from zero to 0.05 M. It can be seen from Figure 21 that in this range, nitric acid has a negligible effect on signal for the elements studied here. The zero (0) on the abscissa in Figure 21 represents no nitric acid added.




Data Analysis

Aqueous standards of Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y were used to construct linear analytical curves from the limit of detection to greater than I .tg/mL. The raw data for the analytical curves were obtained from peak height data from the average of five scans. Both peak height and peak area were investigated. Figure 22-A gives the log-log plot of peak area versus peak height for Mg. The slope of the plot was 1.036. A log-log plot of peak area versus peak height was also constructed for Ni. The corresponding plot is shown in Figure 22-B. The slope of this plot was 1.0715. Both methods of data interpretation gave similar results. For simplicity, peak height data was used for all plots.

As discussed above, aqueous solutions were constructed from serial dilutions from aqueous standards. A mass spectrum of a 1 ppmn solution containing Mg and Al is shown in Figure 23. The sensitivity of the SOLA for Al is slightly higher than for Mg. Figure 24-A and B show mass spectra of a 1 ppm solution containing Cr, Fe, and Nil- and Ti,




























3.Ox1 07 2.5X1 07 2.0x107 '~1.5X107 S.OX1QE -


I


ii-


X


*


i I
-4 -3


-2 -1


0


Log [HNO 3 Concentration (M)]


Figure 21. Effect of HN03 Concentration (M) on 200 ppb Aqueous Solutions.
Arrow Indicates No Acid Added.


87


I


T


+ Cr


I


I


Mg Al


U


A


v y
* Ni
* Ti


A


U
x


Fe Mo


f


X


0.i


- ~~~~~~~1


II


-5























A



.7


38 40 42

Log Peak Height


8.0


7.5


m 7.0a


0


6.0-


44 46


B


E


'3,


64626058-


3.6 4.0 4.5 5.0 5.5 6.0 6-5
Log Peak Height


Figure 22. Log-Log Plots of Peak Area vs. Peak Height. A) Mg B) Ni


00 00


p '7


m d)


(U
4)
a-


36


i





























Al Mg


I * I * I * I


Figure 23. Mass Spectrum of I ppm Mg and Al.


89


2.Oxl 08I


1.5x1O08 k


75

C
P)


1.Oxl08 k


Mg


5OX1 07 k


nn


Mg


23 24 25 26 27 28 29

MlZ


I




Full Text

PAGE 1

DETERMINATION OF METALLO-ORGANIC AND PARTICULATE WEAR METALS IN LUBRICATING OILS ASSOCIATED WITH HYBRID CERAMIC BEARINGS BY INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY By ROBIN ANN RUSSELL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1997

PAGE 2

In loving memory of my father, Bobby James Russell April 14, 1935 December 29, 1995

PAGE 3

ACKNOWLEDGMENTS Words can not express my gratitude towards my graduate research advisor, Dr James D Winefordner. I consider myself fortunate he allowed me to conduct research under his direction for the past few years. His consideration for my professional and personal well being was always evident. Dr Benjamin W. Smith also played a large part in the completion of my graduate research. Elis innovative and thought provoking ideas have made me a creative and independent person inside and outside the laboratory. He has taught me many things, most notably the importance of patience and perseverance. This project brought many fhistrating experiences that I would not have survived if not for Scott Baker. His unending patience and knowledge gave me the strength to continue even when I thought all hope was lost. I wish him and Amy all the best. Matthew DellaVecchia and Christopher Hurt were great sources of help in the laboratory as undergraduate students. Matt in particular experienced the struggles with me and always kept my spirits high. There are many others that have helped me through the past few years through constant support and encouragement. A few of these special people are Leslie King, Andrea Croslyn, Cynthia Schilling, Gretchen Potts, Bryan Castle, Arthur Besteman, Igor Gomushkin, Ricardo Aucelio, Kobus Visser, Wei Hang, Xiaomei Yan, Michael Bartberger, Gayanga Weerasekera, Angie Barker, Chibao Le, and Jim Lovelace Jeanne Karably must receive special recognition as a source of information and strength. She was always able to answer any question I may have had, even those not pertaining to 111

PAGE 4

graduate school. I must also give thanks to my undergraduate academic and research advisor, Dr. Louis G. Daignault. His support for the past eight years has been constant and heartfelt. He has helped me in so many ways throughout my entire higher education experience and for that I am grateful. I would like to thank Dr. Jim Adair and David Mitchell for supplying me with used lubricating oil samples used on the RCF tester. The research contained in this dissertation would not have been completed in a timely fashion if not for the support of the machine and electronic shops of the chemistry department. In particular, Joe Shalosky, Gary Harding, Dailey Burch, and Steve Miles were always willing to assist me in trouble shooting the ICP-MS. They gave me knowledge that can not be learned in textbooks. I would also like to acknowledge the support of my family, especially Terry, Tommy, Jogina, and my mother, Mildred. They understood my physical absence in their lives during my final year of graduate school. Even though I was not around very much, they let me know they believed in me. It is necessary to also mention the help I received from Tom Musselman from Finnigan MAT. Although many SOLA problems were unexplainable, Tom did his best to help me keep the ICP-MS running. Finally, I would like to thank the Air Force Office of Scientific Research and the Texaco Foundation for financial support. IV

PAGE 5

ACRONYMS AAS Atomic Absorption Spectrometry AES Atomic Emission Spectrometry AFS Atomic Fluorescence Spectrometry AID Arizona Test Dust CE Capillary Electrophoresis DI Direct Insertion DIN Direct Insertion Nebulization ETV Electrothermal Vaporization FI Flow Injection FT-ICR-MS Fourier Transform Ion Cyclotron Resonance Mass Spectrometry GC Gas Chromatography HG Hydride Generation HPLC High Performance Liquid Chromatography ICP Inductively Coupled Plasma ICP-MS Inductively Coupled Plasma Mass Spectrometry ICP-OES Inductively Coupled Plasma Optical Emission Spectrometry IC Ion Chromatography IT-MS Ion Trap Mass Spectrometry ITV In-Torch Vaporization LA Laser Ablation V

PAGE 6

LC Liquid Chromatography LOD Limit of Detection NAA Neutron Activation Analysis PLC Programmable Logic Controller PTFE Polytetrafluoroethylene PSIM Particle Size Independent Method RCF Rolling Contact Fatigue RDE Rotating Disk Electrode RSD Relative Standard Deviation SEM Secondary Electron Multiplier SIA Sequential Injection Analysis SOAP Spectrochemical Oil Analysis Program SRM Standard Reference Material SFC Supercritical Fluid Chromatography TOF-MS Time-of-Flight Mass Spectrometry VI

PAGE 7

TABLE OF CONTENTS page ACKNOWLEDGMENTS iii ACRONYMS V ABSTRACT x CHAPTERS 1 INTRODUCTION 1 2 BACKGROUND 7 Oil Analysis 7 Principles of Inductively Coupled Plasma Mass Spectrometry 19 History 19 Inductively Coupled Plasmas 22 Mass Spectrometry 28 Mass analyzer function 30 Ion optics 30 Quadrupole mass analyzer 32 Detectors in ICP-MS 33 Inductively Coupled Plasma Coupled with Mass Spectrometry 36 Background ions 43 Interferences in ICP-MS 45 Sample introduction 46 3 MICROWAVE DIGESTION 54 Introduction 54 Theory 56 Instrumentation 59 Microwave Oven Calibration 61 Microwave Acid Digestion Vessel 67 4 DETERMINATION OF WEAR METALS IN AQUEOUS SOLUTIONS 71 Instrumental Set-up 71 Inductively Coupled Plasma 71 Quadrupole Mass Spectrometry 72 ICP-MS Interface 72 vii

PAGE 8

Sample Introduction 75 Experimental Parameters 77 Chemicals 77 Glassware 78 Experimental Parameters Optimization 78 Acid Effects 86 Data Analysis 86 Error Analysis 98 5 DETERMINATION OF METALLO-ORGANIC WEAR METALS IN OIL SAMPLES 106 ICP-MS Instrumental Set-up 106 Microwave Digestion Instrumentation 106 Experimental Parameters 107 Chemicals 107 Glassware 108 Microwave Acid Digestion Procedure 108 Data Analysis 1 1 1 Error Analysis 117 6 DETERMINATION OF PARTICULATE WEAR METALS IN OIL SAMPLES 127 Instrumentation 127 Experimental Parameters 127 Chemicals 128 Glassware 128 Preparation of Standards 128 HNO3 and H2O2 Digestion of Fe, Mg, Y, Ni, Mo 128 H2S04and H2O2 Digestion of A1 134 HCl and H2O2 Digestion of Cr 134 HF and H2O2 Digestion of Ti 135 HF and H2O2 Digestion of Cr 137 HF and H2O2 Digestion of A1 137 Acid Effects 143 Error Analysis 147 7 DETERMINATION OF WEAR METALS IN USED LUBRICATING OILS 152 Introduction 152 Instrumentation 152 Vlll J

PAGE 9

Experimental Parameters 153 Materials 153 Collection of Oil Samples 156 Microwave Acid Digestion 158 Data Analysis 158 8 CONCLUSIONS AND FUTURE WORK 169 Conclusions 169 Future Work 171 REFERENCES 176 BIOGRAPfflCAL SKETCH 184 IX

PAGE 10

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DETERMINATION OF METALLO-ORGANIC AND PARTICULATE WEAR METALS IN LUBRICATING OILS ASSOCIATED WITH HYBRID CERAMIC BEARINGS BY INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY By Robin Ann Russell August 1997 Chairman: James D. Winefordner Major Department: Chemistry It is possible to increase both the performance and operating environment of Jet engines by using hybrid ceramic bearings. Our laboratory is concerned with investigating lubricating fluids for wear metals associated with silicon nitride ball bearings and steel raceways Silicon nitride is characterized by low weight, low thermal expansion, high strength, and corrosion resistance. These attributes result in longer engine lifetimes than when metallic ball bearings are used. Before the routine use of ceramic ball bearings can be realized, the wear mechanisms of the materials should be thoroughly understood. One important variable in determining wear degradation is the concentration of metal present in the lubricating oils used with the bearings. A complete method for analyzing used lubricating oils for wear metal content must accurately determine all metal forms present. X

PAGE 11

Oil samples pose problems for routine analysis due to complex organic matrices Nebulizing these types of samples into an Inductively Coupled Plasma Mass Spectrometer introduces many problems including clogging of the sample cone with carbon and increasing interferences. In addition, other techniques such as Atomic Absorption Spectrometry and Atomic Emission Spectrometry are particle size dependent They are unable to analyze particles greater than 10 tim in size. This dissertation describes a method of analyzing lubricating oils for both metalloorganic and particulate species by ICP-MS. Microwave digestion of the oil samples eliminates the need for elaborate sample introduction schemes as well as the use of a modified carrier gas. Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y have been determined in both aqueous and organic media. Metallo-organic solutions of these metals were successfully digested, nebulized into the ICP, and the singly charged ions measured by mass spectrometry Metal particulates in oil matrices have also been quantitatively determined by the above method. Linear analytical curves were obtained for these elements from the detection limits (~ 1 ppb) to greater than 1 ppm. Used lubricating oil samples were also analyzed by microwave digestion ICP-MS Oil samples were collected fi-om a Rolling Contact Fatigue tester Two bearing systems were evaluated: M50 steel balls on an M50 steel rod, and Si 3 N 4 balls on an M50 steel rod. Improved operating conditions were obtained when the SisN 4 balls were used, which corresponds to longer engine lifetimes. xi I

PAGE 12

1 / CHAPTER 1 INTRODUCTION Ceramic materials are currently being used for a wide variety of applications including aircraft and aerospace components, machine tools, medical equipment, semiconductor processing equipment, and automotive components. They are replacing components that are currently constructed out of metallic materials due to their many advantageous characteristics, such as low density, the ability to operate in high temperature environments (550 800 °C), and corrosion resistance. The United States Air Force is actively searching for replacement materials for ball bearings in jet engines. In particular, silicon nitride (Si 3 N 4 ) is finding extended use in bearing systems that employ ceramic rolling elements with steel raceways. It is desirable to use silicon nitride ball bearings over other ceramic materials due to the manner in which they fracture. Like steel bearings, Si 3 N 4 fractures by spalling, or a gradual, pitting-induced fatigue, which has been linked to surface crack formation on the rolling element Other ceramics fail more catastrophically, by total fracture of the material. Aeronautic applications traditionally use ball bearings constructed out of steel, such as M-50. Silicon nitride has several characteristics that exceed those of steel including a superior resistance to corrosion, lower mass (40 % less dense), increased hardness, the ability to withstand higher temperatures, and nonmagnetism, which 1

PAGE 13

2 / eliminates the problems of electric arc damage In addition, Si 3 N 4 is compatible with a wider range of lubricants than steel ball bearings The low mass of silicon nitride results in a reduced centrifugal ball load and enables it to be subjected to much higher speeds ( 1 ) At temperatures greater the 550 °C, stainless steel begins to lose its hardness and stability In contrast, Si 3 N 4 maintains its hardness up to temperatures of 1 100 °C This results in a material that can be run at higher speeds, higher service temperatures, and with less lubrication. Several studies have already been conducted using Si 3 N 4 in hybrid bearings (2,3). It has been determined that the replacement of stainless steel with Si 3 N 4 as the rolling element material in ball bearings results in longer service life due to reduced wear of the ball bearings. Silicon nitride is fabricated by a variety of methods, the most common being hot isostatic pressing (4). In the first step of manufacturing, silicon nitride powder is mixed with sintering aides such as yttrium oxide (Y2O3), magnesium oxide (MgO), and aluminum oxide (AI2O3). The resultant powder is then mixed with a binder and shaped into spheres by pressing it into a mold. The material is then sintered at 1,700 1,800 °C in an inert gas atmosphere, such as nitrogen gas, and is subjected to hot isostatic pressing. An oxynitride liquid is formed upon heating the mixture, dissolving the a-silicon nitride particles Psilicon nitride precipitates out of the liquid and constitutes the bulk of the Si 3 N 4 densification product. The final step involves finishing the Si 3 N 4 spheres with a diamond grinding wheel.

PAGE 14

3 Before metallic components can be replaced with Si3N4 ball bearings, the wear mechanisms of this material must be fully understood. One way to do this is to study the lubricating oils associated with these systems. The determination of wear metals in used lubricating oils gives a good indication of the wear of engine parts ( 5 ) Often an increase in concentration of a metal is indicative of wear that has occurred before it is visually apparent. As the ball bearings wear over time, elements that went into the construction of the ceramics will appear in the lubricant. Such elements include silicon, nitrogen, and any sintering aids such as yttrium, magnesium, or aluminum. It is also possible to observe in the used lubricating oil any elements that make up the raceway material or other engine components. It is the goal of this research to qualitatively and quantitatively determine wear metals in lubricating oils associated with the wear of SisN4 hybrid bearings. These include Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y. Wear metals can take two forms: metalloorganic and particulate. It is desirable for a technique to be able to detect particulate wear metals over a wide range of sizes, pm mm diameters, since particles in the oil can also further degrade component surfaces. Preventative maintenance programs can be established by determining the cause of failures before detrimental effects are experienced. This constitutes repairing or replacing failing parts as well as simple oil changes The successful implementation of preventative maintenance programs has the potential to save millions of dollars and possibly lives each year Metals in used lubricating oils originate from three sources: wear, contamination, and additives. Wear metals are defined as the products of fnctional deterioration or corrosion and can be traced back to the composition of the engine under inspection ( 6 ).

PAGE 15

4 Dirt and leaks in the system can lead to contamination, while additives are put into lubricating oils to help reduce wear As discussed earlier, Al, Mg, and Y are common sintering aides used for the manufacturing of silicon nitride. As shown in Table 1, the presence of iron indicates wear from engine components, chromium indicates ring wear, nickel indicates wear of plating components, and molybdenum and titanium indicate wear of bearing alloys. It is desirable to detect these elements in lubricating oils at a level that precludes complete engine failure Current techniques used by the United States Air Force for determining wear metals in used lubricating oils are Atomic Absorption Spectrometry (AAS) and Atomic Emission Spectrometry (AES). Both of these techniques give excellent qualitative analyses for wear metals. They are, however, unreliable when the metals are in the form of particulates. AAS and AES are unable to accurately determine particulate wear metals greater than a few micrometers in diameter. In addition, other atomic analytical techniques such as Inductively Coupled Plasma Mass Spectrometry (ICP-MS) enable lower metal concentrations to be determined. This results in earlier wear detection and improved maintenance programs. This dissertation describes a novel method for determining metallo-organic and particulate wear metals in lubricating oils by Inductively Coupled Plasma Mass Spectrometry preceded by microwave digestion. Inductively Coupled Plasma Mass Spectrometry is an ideal method for determining trace levels of wear metals. The low detection limits achieved by this technique enable preventive maintenance programs to be established since it can detect metal content at very low levels. In addition, microwave

PAGE 16

5 digestion converts all metallic species, into the aqueous phase, thereby eliminating the particle size dependence of the current methods.

PAGE 17

6 Table 1 . Type of Wear Indicated by Element Element Indicated Condition A 1 Indicates wear of Si3N4 Mg Indicates wear of Si3N4 Y Indicates wear of Si3N4 Fe Indicates wear from engine components Cr Indicates ring wear Ni Indicates wear of plating components Mo Indicates wear of bearing alloy Ti Indicates wear of bearing alloy

PAGE 18

CHAPTER 2 BACKGROUND Oil Analysis Numerous techniques have been employed for the elemental analysis of oil matrices including Neutron Activation Analysis (NAA) (7), Colorimetry (8), Atomic Fluorescence Spectrometry (AFS) (9,10,11), X-ray Fluorescence Spectrometry (12,13), Inductively Coupled Plasma Mass Spectrometry (ICP-MS) (14), Atomic Absorption Spectroscopy (AAS) (15,16,17), and Atomic Emission Spectrometry (AES) (18,19,20). The latter two techniques have been used the most in conjunction the United States Air Force Spectrometric Oil Analysis Programs (SOAP). Therefore most of the literature regarding lubricating oil analyses concerns AAS and/or AES. Atomic Absorption Spectrometry has been used with a flame or a furnace for atomization and has the advantages of being rapid with simple sample preparation, wide linear dynamic range, low instrumentation cost, and the ability to analyze liquid organic samples directly. Two of the more popular means of preparing oils for flame AAS are direct aspiration of the sample after dilution with an appropriate organic solvent, and burning the sample and then extracting the residue with inorganic standards. Anwar et al. directly aspirated crude oil samples into a flame AAS after diluting them with a solvent system consisting of butan-2one (for desolvation of the organic matter), water (to accommodate the inorganic metal content), hydrochloric acid (to decompose or dissolve the metal into the aqueous phase). 7

PAGE 19

8 metal content), hydrochloric acid (to decompose or dissolve the metal into the aqueous phase), propan-2-ol (to maintain the homogeneity of the sample solutions), and ethanol (to blend the other constituents) (21). For comparative reasons, a dry ashing protocol was also performed on the oil samples. These two procedures gave vastly different results for crude oil samples. Samples that were ashed before introduction into the flame gave lower results than those obtained by the mixed solvent system due to the losses that occurred during the combustion process. Although good results were obtained with the mixed solvent method, it was found to be highly dependent on the solubility of the oil samples, homogeneity of the solution, solvent composition, and the time and temperature of the oil/solvent solution. Udoh has shown the determination of calcium, magnesium, and zinc in unused lubricating oils by flame atomic absorption spectrometry after ashing the samples (22). Sample preparation consisted of heating the oil samples on an evaporating dish, combining the residue with hydrochloric acid, concentrating the solution on a steam bath, and then diluting the sample once again. Phosphorous is known to interfere with the determination of alkaline earth metals by flame AAS. Therefore, lanthanum and strontium were added to several of the oil concentrates to determine the effects of their concentration on the calcium and magnesium signals. The Ca concentration exhibited a linear correlation with both increasing La and Sr concentrations, while the Mg concentration was not affected by addition of these two elements Ashing of oil samples can also be accomplished by a graphite rod or filament atomizer (23,24). These methods have several advantages over flame AAS such as

PAGE 20

9 greater sensitivity, better detection limits, smaller sample volumes, and particle size independence, which allowed for total metal content analyses Reeves et al. determined silver, chromium, copper, iron, nickel, lead, and tin in used jet-engine and reciprocatingengine oils using AAS with a graphite rod atomizer (25,26) Comparison of their results with other laboratories showed good correlation. Dilution of the oil samples with isooctane was required to obtain better sensitivity for the analysis of iron and copper Bratzel et al. employed a graphite rod as an atomizer for AAS for the determination of lead in petroleum and petroleum products (27). Direct analysis of oil samples as well as the extraction of organic species into an aqueous phase were presented. Direct analyses were performed after diluting petroleum samples with xylene. The reversed extraction procedure consisted of a xylene-aqueous nitric acid system with the addition of solid dithizone to increase the rate of extraction. The carbon rod atomization method resulted in a detection limit of 2 x 10'*^ g for lead. The sensitivity of the technique was affected by the type of lead salt and solvent used for both standard and sample preparation, choice of wavelength, and interferences. Other means of preparing petroleum samples for AAS analyses have also been used considerably. Decomposition and cold-vapor methods were reported by Knauer et al. for the determination of mercury in petroleum and petroleum products (28) The first method required several hours to complete and involved a closed system digestion with nitric acid and sulfuric acid. Another method presented was based on a Wickbold oxyhydrogen combustion set-up where the volatilized mercury was collected in an acidic permanganate solution. This was also lengthy and arduous. The collected solutions from

PAGE 21

10 both methods were analyzed by cold-vapor AAS. Although mercury concentrations down to 5 10 ng/g could be detected, several problems occurred with these methods First, sample contamination from glassware and reagents was widespread. In addition, when independent laboratories used these methods, the results exhibited a great deal of variability. The elaborate sample preparation procedures were thus nontrivial. The above methods for AAS work well for metallo-organic samples However, not all samples containing particulate matter can be accurately analyzed by AAS. Saba et al. have reported that engine component failure has occurred as a result of the inability to detect large wear particles (29). Jantzen et al. (30) and Kauffman et al. (31) reported that direct AAS could only accurately detect iron wear particles smaller than about 1 pm Jantzen’ s laboratory compared direct flame AAS, AAS after incineration, and colorimetry for iron content in oil samples. Direct AAS gave values 50% lower than colorimetry when the oil samples contained iron particles larger than 2 pm. Saba et al. used AAS with a graphite ftimace atomizer was to analyze iron, copper, aluminum, magnesium, and other important wear metals in aircraft lubricating oils (32). Standards and samples were diluted with kerosene before insertion into the graphite tube. Oil samples analyzed by this method contained metallo-organic and metal particulates. Metal powder suspensions were also analyzed by a particle size independent method (PSIM) to determine the actual metal concentration in the samples. The PSIM consisted of digesting the samples in aqua regia, shaking for 30 seconds, and then agitating in an ultrasonic bath at 65 °C for 45 minutes (33), The mixture was then cooled and diluted with methyl isobutyl ketone in isopropyl alcohol. Graphite furnace AAS was

PAGE 22

shown to be superior to conventional methods since it was capable of analyzing wear metal particles. However, the precision of the graphite furnace AAS method decreased with increasing particle size due to the inability to obtain a representative sample in the small amounts (microliter sample size) placed into the furnace Atomic Emission Spectrometry has also been widely used for the determination of trace elements in oil matrices. Atomization of the analytes is usually accomplished with a Rotating Disk Electrode (RDE) (34,35) or an Inductively Coupled Plasma (ICP) (36,37) In RDE-AES, a rotating electrode is submerged in an oil sample, while a fixed electrode is positioned above it. Emission data can be obtained once the sample is vaporized with a high voltage arc. Gambrill et al. used RDE-AES for the determination of Pb, Fe, Ba, P, Ca, and Zn in new and used lubricating oils (38). A graphite disk, 12.7 mm in diameter and 3.18 mm thick, rotating at various speeds was used as the lower electrode while a 6.35 mm graphite rod served as the fixed upper electrode. The best analytical curves were obtained when high rotation speeds of 1 5 rpm were used. Compared to ICPs, the Rotating Disk Electrode technique was simpler to operate and more stable. In addition, larger particles could be analyzed by this method and oil samples could be studied without any sample preparation. Plasmas do, however, offer several advantages over flames and RDEs. The higher temperatures and longer residence times encountered by the sample aerosol leads to a greater degree of atomization than that seen in flames, which results in lower detection limits. For these types of analyses, the oil samples require dilution with a suitable organic solvent such as xylene, kerosene, methyl isobutyl ketone, or toluene to reduce the viscosity of the sample before introduction into

PAGE 23

12 the plasma. Bangroo et al. analyzed wear metals and additive elements in lubricating oils by AAS and ICP-AES after subjecting the samples to a rigorous treatment that included ultrasonicating with hydrochloric acid and hydrogen peroxide, diluting with xylene or toluene, agitating for thirty minutes, and then separating the aqueous phase from the organic phase (39). The procedure was then repeated on the resulting oil residue Although experimental data agreed with certified values, the sample preparation steps were time consuming and subject to several losses. Brown et al. used ICP-AES for the determination of twenty-one elements in new and used lubricating oils (40), The samples were first diluted with xylene before introduction into the plasma, which gave a low bias when compared to other sample preparation methods. Although this method allowed for sub ppm detection limits and multielement analysis, variable wear particle size could result in imprecise results when using this method. Modification of the sample introduction scheme for AES has also been investigated by many researchers (41). Algeo et al. used a modified Babington nebulizer to introduce oil samples into an ICP-AES (42). This type of sample introduction allowed viscous oils to be nebulized more efficiently and without organic solvent dilution. However, lower results were obtained when compared to ashing-AAS for real world samples due to the particle size dependence of the method. AES is also an inadequate technique to use for determining total metal content when particles are present in oil samples. In ICP-AES, particles greater than 10 pm in size may pass through plasma without being atomized (39). Kauffinan et al. analyzed metal powder suspensions by direct current plasma spectrometry preceded by acid

PAGE 24

13 dissolution (43). A mixture of HF/HCI/HNO3 was used for dissolution of the suspensions Percent recoveries for Al, Cr, Cu, Fe, Mg, Mo, Ni, Pb, Si, Sn, and Ti were 2-48 % without acid dissolution. The use of acid increased the percent recoveries to 89 102 % The latest method that has been used for the determination of wear metals in oil matrices is Inductively Coupled Plasma Mass Spectrometry (ICP-MS) It is an attractive technique for this application due to its numerous advantages which include low limits of detection (down to pptr), large linear dynamic range (up to five orders of magnitude), simple spectra, multielement analysis capabilities, and simple sample introduction methods. It is not desirable, however, to introduce organic samples directly into the mass spectrometer because of increased interferences and carbon deposition on the sample cone orifice and ion optics. When an organic sample is analyzed by mass spectrometry, more background peaks are observed than with aqueous samples. Polyatomic ions containing carbon can also increase the level of interferences, as shown in Table 2 (44). It is extremely difficult to determine trace levels of elements such as magnesium, calcium, titanium, chromium, and iron when large levels of polyatomic ions interfering with the elemental masses are present in the plasma. Although organic solvents are ionized completely in the high temperature plasma, carbon has a tendency to condense on the cooled surface of the sample cone orifice and even on the ion optics. This can be minimized by adding oxygen to the carrier gas The oxygen atoms combine with carbon atoms in the plasma to form carbon dioxide which is expelled from the system before deposition occurs. The exact amount of O 2 needed to accomplish this is a tedious task. Too much O 2 can prematurely degrade the sample cone.

PAGE 25

14 Table 2. Interfering Ions Found When Organic Matrices are Introduced into the ICP-MS (Adapted from 44) Species Mass Interfering Ion 24 Mg ’^cisr 25 Mg 26 Mg N2^ 28 Si 44 Ca '^C02ir, ‘^C02^ 45 Sc '^C02H" 46 Ti, Ca 52 Cr 53 Cr ArO^ 56 Fe ArOfT 57 Fe 40At2^ 80 Se

PAGE 26

15 while too little can result in carbon build-up. The appropriate amount is dependent on many variables of the system including the physical characteristics of the organic solvent, solvent flow rate, spray chamber temperature, and plasma conditions (45). In addition, organic solvents have high vapor pressures which can overload the plasma causing it be become unstable. There are several operating parameter modifications that can be performed to help with plasma stability One parameter is the liquid flow rate. Reducing the solution uptake from 1.0 ml/min to 0.5 ml/min decreases the amount of solvent in the plasma and thus increases its stability (44). Organic samples also require more power to sustain the plasma than aqueous ones. As a result, an increase of 500 W is typically needed. There are many sample introduction schemes for ICP-MS that enable one to analyze liquid organic samples such as direct aspiration (46,47), electrothermal vaporization (ETV) (48), in-torch vaporization (ITV) (49), high-performance liquid chromatography (HPLC) (50,51), gas chromatography (GC) (52,53), liquid chromatography (LC) (54,55), ion chromatography (IC) (56,57), capillary electrophoresis (CE) (58), supercritical fluid chromatography (SFC) (59), laser ablation (LA) (60), cryogenic desolvation (61), membrane desolvation (62,63), direct insertion (DI) (64), flow injection (FI) (65,66), direct injection nebulization (DIN) (67,68), and sequential injection analysis (SLA) (69). Of these techniques, direct aspiration, ETV, GC, and SIA have been used with ICP-MS for the analysis of petroleum samples. When direct aspiration is used with ICP-MS, the sample is converted into an aerosol and then transported to the plasma for vaporization, atomization, and ionization. The sample is either combined with an

PAGE 27

16 organic solvent or carried through a dissolution or digestion procedure prior to being nebulized. The digestion of oil samples can be tedious and time-consuming, with some digestions requiring hours to complete. Olsen et al. directly aspirated oil samples into the ICP-MS by nebulization (70). An extraction procedure using xylene was carried out on the oil samples prior to sample introduction. The accuracy of the technique was investigated by subjecting NIST Standard Reference Material (SRM) 1634b to the same procedure as the samples. Errors for several elements were rather large when the experimental and certified values were compared. In fact, trace concentrations of chromium in an organic matrix could not be determined by direct ICP-MS due to the large variation in the interfering ‘‘^Ar'^CH on mass 53, the most abundant isotope of Cr. Al-Swaidan used ICP-MS for the determination of vanadium and nickel in crude oil products by first extracting the analytes fi'om the organic solution to an aqueous one (14,71). This method employed the addition of xylene or methyl isobutyl ketone to reduce the viscosity of the sample. Extraction of the metals was facilitated with the aid of nitric acid. Recoveries of added V and Ni were good, with percent relative deviations less than 4%. Bettinelli et al. analyzed fuel oils by ICP-MS preceded by conversion of the oils into aqueous samples by acid mineralization in a microwave oven (72). This group obtained detection limits in the sub ppm region for numerous metals including Al, Cr, Fe, and Ni. Tandem sample introduction schemes have also been used for the determination of trace metals in oil matrices. Very small sample volumes (|jJ) can be used when a graphite furnace is used for vaporization for ICP-MS detection. Other advantages of vaporization with an ETV include better sensitivity; reduced molecular ion interferences; reduced oxide

PAGE 28

17 formation, reduced nonspectroscopic interferences, and the ability to directly analyze samples containing of liquid organic material, strong acids, and high dissolved solids The sample can be dried, pyrolyzed, and vaporized prior to entering the plasma Thus, many interferences can be eliminated by step-wise heating of the sample In addition, plasma conditions for atomization and ionization can be optimized independently of those used for vaporization. As a result, sample introduction parameters can be better controlled Escobar et al. coupled ETV with ICP-MS to determine Al, Mg, Fe, and Y metallo-organic compounds in lubricating oils (73). The oil samples were diluted with xylene before injection into the graphite furnace, which allowed the metallic species to be separated from the organic matrix. Thus, organic material present in the sample was vented to the atmosphere and not allowed to enter the plasma. Detection limits obtained by this method were in the ppb region. ETV-ICP-MS has also been used for the determination of mercury in petroleum by Osborne (74). Oil samples were mixed with various alkanes and injected into a graphite furnace for vaporization. A detection limit of 3 ppb for Hg in pentane was reported. However, the relative standard deviation associated with this value was 50%. Electrothermal vaporization has a few undesirable characteristics that make it a difficult method of sample introduction for ICP-MS. Precision can be poor due to the difficulty of pipetting oil solutions into a graphite furnace. In addition, various elements require different experimental conditions for efficient vaporization and transport into the plasma. Much work must be done to ensure that optimum ashing and vaporizing procedures are used. Carriers must also be added to many samples to efficiently transport

PAGE 29

18 the sample from the furnace to the plasma. Multielement determinations by ETV-ICP-MS are difficult due to the limited time available for mass scanning and the complexity of finding suitable compromise conditions for multielement mixtures Alonso et al. used ion chromatography ICP-MS to analyze spent nuclear fuels for fission products and actinides (57). The outlet of the separation column was connected directly to a cross-flow nebulizer. Separation of Cs, Ba, and the actinides was facilitated with nitric acid. Cesium and barium were separated on one column while a different column was used for separation of the actinides. The only way to eliminate isobaric interferences in the determination of the above elements was to chemically separate the interfering nuclides. Three individual separation procedures were required to isolate Cs and Ba, the lanthanides, and the actinides. Sequential injection analysis has been used v^dth ICP-MS by Al-Swaidan to analyze Saudi Arabian crude oil for lead, nickel, and vanadium (69). SIA differs from flow injection in that a selector valve allows sample and reagent zones to be sequentially injected into a channel. In flow injection, the sample zone is injected into a flowing carrier stream and reagents are combined with it in route to the detector SIA has several advantages over FI including minimization of reagent use and waste production since the flow is not continuous. In addition, corrosive solvents are handled in a closed system Finally, most analyses of lubricating oils quote detection limits in the low ppm or mid ppb region. A technique with lower detection limits would permit earlier detection of wear metals in lubricating oils, and therefore improve the quality of maintenance programs. A method of converting oil solutions into aqueous ones would be

PAGE 30

19 advantageous Aqueous solutions are stable for long amounts of time and can be easily nebulized for a variety of detection schemes. Principles of Inductively Coupled Plasma Mass Spectrometry History The first Inductively Coupled Plasma (ICP) sustained in atmospheric pressure flowing gases was accomplished by Reed in 1961 (75). He sustained plasmas using various support gases such as argon, and various argon mixtures using oxygen, hydrogen, and helium. Reed introduced the first application of inductively coupled plasmas at atmospheric pressure: independently growing sapphire, zirconia, and niobium crystals (76). Powder was fed into the ICP through the center of the torch. The crystals had a diameter of 5 15 mm and a length of 30 90 mm. Growth rates were 20 50 mm/hr. It was soon realized that the plasma could be used in other applications. In particular, it could function as superior excitation source for atomic emission spectrometry (AES). During the past 30 years, ICP-AES has become one of the most widely used atomic spectrometric methods. Gray at the University of Surrey was first in illustrating the union of plasmas and mass spectrometers (77). He coupled an atmospheric pressure dc arc plasma to a quadrupole mass spectrometer and obtained |ag/ml detection limits in aqueous solutions. Complete vaporization, atomization, and ionization was not possible with this system due to the core gas temperature of the plasma, which was about 5000 K. This set-up only

PAGE 31

20 worked well for those elements with ionization energies below 8 eV In addition, further problems were experienced with inter-element and matrix effects, and sample introduction With their high temperatures, inductively coupled plasmas were the next logical source of ion production for mass spectrometric detection. Gray, along with Date from the British Geological Survey, Fassel and Houk from Ames Laboratory at Iowa State University, and Douglas from Sciex became the pioneers in ICP-MS as they worked towards developing a sensitive technique for elemental determination. Houk et al. published the first manuscript describing the determination of trace elements with an ICP-MS in 1980 (78). The hardware of the ICP-MS used by this group was similar to that employed in modem instmments. An ultrasonic nebulizer was used to produce an aerosol of the liquid solution. The sample aerosol was then transported to the plasma for vaporization, atomization, and ionization. The sample cone had an orifice of 50 |im and was water cooled. Mass spectra were presented that illustrated the isotopic capabilities of ICP-MS. Analytical curves were constructed which extended almost four orders of magnitude. In addition, detection limits for Mg, Cr, Mn, Co, Cu, Rb, As, and Y in the range of 0.002 0.06 |ag/ml were obtained. One problem observed in the initial experiments was condensation of solid material in or near the sample cone aperture. The size of the orifice in the sample cone was varied by several researchers to determine its effect on sampling. Larger sampling orifices initially appeared advantageous since they could accommodate larger ion beams and therefore transmit a greater number of analyte ions to the detector. It was determined that at apertures greater than 100 tim, a

PAGE 32

21 secondary discharge developed in front of the sample cone (78,79). Further ionization by collisions can occur in the secondary discharge and cause high photon levels to reach the detector, high populations of doubly charged ions, production of ions with a higher kinetic energy and energy spread, resulting in peak splitting; and erosion of the sampler orifice The secondary discharge was the result of arcing between the plasma and the sample cone caused by the large RF voltage swing in the plasma relative to the grounded sample cone Conventional load coils are grounded at one end while the other end is free to undergo voltage swings at the applied radio frequency. These voltage swings induce a magnetic field in the plasma and couple capacitively into the discharge. When the plasma swings positive, a negative voltage appears in the plasma and a second discharge is generated at the sampling cone. The secondary discharge can be overcome by grounding the load coil at the center (80). The ends of the coil are allowed to go through equal but opposite voltage swings, and cancel each other out. The secondary discharge can also be overcome by using the typical end-grounded load coil and smaller aperture sample cones. The first commercial instrument of this type was introduced at the 1983 Pittsburgh Conference by Sciex, Inc There are now several ICP-MS manufacturers, offering varying types of mass analyzers. The combination of inductively coupled plasmas and mass spectrometers provides many advantages in atomic spectroscopy such as multielemental analyses, isotopic capabilities, simple spectra, small sample requirements, low detection limits, and a large linear dynamic range.

PAGE 33

22 Inductively Coupled Plasmas An ICP is an electrodeless discharge formed in a gas, maintained by energy coupled to it from a RF generator. A quartz torch is placed inside a three turn copper induction, or load, coil that is connected to a high frequency radio frequency (RF) generator. RF generators used for inductively coupled plasmas are operated at a frequency of 27.12 or 40.68 MFIz with output levels up to 5 kW The most commonly used support gas for plasmas is argon, which flows through the center of the torch. By definition, a plasma is an ionized gas. Since Ar is a nonconductor, it is necessary to provide electrons for ionization. A Tesla coil is used to seed the Ar with free electrons The electrons flow from the Tesla coil to the plasma region by means of the induction coil. Oscillating magnetic fields are generated by the high frequency RF currents flowing in the induction coil. Figure 1 shows the inner axially oriented magnetic forces as well as outer elliptical fields that are formed (81). The induced magnetic fields are time-varying in strength and direction. The axial magnetic fields induce eddy currents which flow in closed annular paths inside the torch. The accelerated electrons and ions encounter resistance to their flow, resulting in Joule or ohmic heating and consequently additional ionization. Since an electron has a mean free path of about 1 pm at atmospheric pressure, collisions with argon atoms occur, heat the plasma, and form a bright discharge. In the center of the plasma, temperatures reach 9,000 10,000 K. Another stream of gas, usually Ar, flows through the torch to prevent overheating and subsequent melting of the quartz glassware. It is also used to tune the shape of the plasma for optimum ion beam conditions. This gas flow is known as the intermediate gas.

PAGE 34

23 I Ar Figure 1. Schematic of Magnetic Fields Induced in Inductively Coupled Plasma H Denotes Magnetic Field.

PAGE 35

24 Aerosol sample enters the plasma through an injection tube located inside the torch. At high RF frequencies, the plasma takes on an annular shape due to the eddy current flowing more closely to the outer edges of the plasma This is shown schematically in Figure 2 (81). The extent to which the plasma takes on the annular shape is controlled by the frequency of the primary current generator and the carrier gas flow rate A plasma operating at a RF' frequency of 27 MHz requires a carrier gas velocity of ~ 1 L/min for optimum sampling. As illustrated in Figure 2, the sample particles do not pass through the hottest part of the plasma. They experience a plasma temperature of 7,000 K, twice that found in the hottest flame. In addition, sample particles generally spend a few milliseconds in the plasma. These two parameters, high temperature and long residence time, results in nearly complete vaporization of analyte and solvent. Aerosol particles exit the injector tube and enter the plasma where they are first dried by solvent evaporation. The higher plasma temperatures then serve to vaporize, atomize, and ionize the particles. If a 1 ppm solution is aspirated into the plasma through a pneumatic nebulizer at a sample uptake rate of 1 mL/min, the total analyte atom population will be approximately 1 x lOÂ’ cm'^ (82). The plasma temperatures which the sample particles encounter are high enough to completely (> 99%) ionize most of the elements in the periodic table. The degree of ionization is a function of many variables including plasma conditions and ionization potentials of the elements. Elements with ionization potentials less than Ar (15.8 eV) are ionized in the argon ICP. As shown in Figure 3, most elements have first ionization potentials in the 5 15 eV range.

PAGE 36

25 o o Sample Particles Figure 2. Sample Aerosol Entering Plasma. (Adapted from 81)

PAGE 37

26 He 24.6 Ne 21.6 Ar 15.8 Kr 14.0 Xe 12.1 Rn 10.75 F 17.4 Cl 13.0 Br 11.8 1 10.5 At o 13.6 s 10.4 Se 9.75 Te 9.10 Po 8.42 N 14.5 P 10.5 As 9.81 Sb 8.64 Bi 7.23 c 11.3 Si 8.15 Ge 7.90 Sn 7.33 Pb 7.42 o CD oo 66S IV Ga 6.0 In 5.79 TI 6.12 Zn 9.39 66 9 PO Hg 10.44 Cu 7.73 Ag 7.58 Au 9.23 Ni 7.64 Pd 8.34 Pt 9.0 Co 7.86 Rh 7.46 Ir 9.10 Mt Fe 7.87 Ru 7.37 Os 8.7 Hs Mn 7.44 Tc 7.87 Re 7.88 Ns Cr 6.77 Mo 7.10 w 7.98 0) CO V 6.74 Nb 6.88 Ta 7.89 Ha Ti 6.82 Zr 6.84 Hf 7.0 Rf Sc 6.54 Y 6.38 La 5.58 Ac 6.9 Be 9.32 Mg 7.65 Ca 6.11 Sr 5.70 Ba 5.21 Ra 5.28 H 13.6 Li 5.39 Na 5.14 K 4.34 Rb 4.18 Cs 3.89 LL Lu 5.43 L. _I Yb 6.25 99 9 ON Tm 6.18 Md 6.59 Er 6.10 Fm 6.51 Ho 6.02 Es 6.42 Dy 5.93 Cf 6.31 Tb 5.85 Bk 6.24 frL9 PO Cm 6.03 Eu 5.67 Am 6.0 Sm 5.63 Pu 5.8 Pm 5.55 Np 6.22 Nd 5.49 U 6.11 Pr 5.42 Pa 5.91 Ce 5.47 Th 6.11 Figure 3 Ionization Potential for Selected Elements.

PAGE 38

27 Not all analytes in the sample are extracted and detected by the ICP-MS Roughly, less than one in a million analyte ions formed in the plasma reach the detector The probability that an analyte is detected is highly dependent on characteristics of the system such as gas temperature, electron number density, electron temperature, and energy transport rates. The processes of vaporization, atomization, and ionization occur simultaneously in the plasma. Every second more than a million polydisperse aerosol droplets enter the plasma (83). Drops of different sizes undergo the above processes at various positions and time in the plasma. The temperature gradients observed in the plasma result in nonuniform ionization. Simple equations can not be used to describe ionization in the heterogeneous plasma. If the plasma were in local thermodynamic equilibrium, a simple equation could be used to relate the electron number density, temperature, and pressure. In addition, the electron temperature and gas temperature would be identical. Ionization efficiencies can be calculated from Equation 1 where is the ionized Ml M“ 2 K m,kT 3/2 V -IP/kT, Equation 1 element, M° is the neutral element, n, is the electron number density in the plasma (cm‘^), m« is the mass of the electron (kg), k is Boltzmann’s constant (J/K), Te is the free electron temperature (K), h is Planck’s constant (J*s), Q* and Q° are the electronic partition functions of the ion and neutral atom (dimensionless), IP is the ionization potential of the

PAGE 39

28 element (eV), and T; is the ionization temperature (K) (84) Equation 1 assumes the plasma is in local thermodynamic equilibrium when in fact it is not As a result, actual ionization efficiencies may be higher than those predicted based on calculations. Degrees of ionization, [NT] / ([M"]+[M]) x 100%, are given in Figure 4 for most elements in the Periodic Table, of which most are greater than 90% (85) Values in parentheses are for doubly charges species. The degrees of ionization were calculated using 1 X 10*’ cm'^ as the value of n« and 7500 K for both Te and Tj. One can see that even elements like As, Se, and S that are characterized by high ionization potentials can be highly ionized in the ICP. Even though the plasma has a high temperature, ionization conditions are not overly energetic. The temperature and electron number density values do not lead to a large amount of doubly charged ions. For example, Ba has one of the lowest second ionization potentials (10 ev) and only a 9% efficiency for producing doubly charged ions in the ICP. Mass Spectrometry Numerous types of mass spectrometers have been used with inductively coupled plasmas as the ion source including quadrupole MS (86), high resolution double-focusing magnetic sector MS (87), time-of-flight MS (TOF-MS) (88), ion trap MS (IT-MS) (89), and fourier transform ion cyclotron resonance MS (FT-ICR-MS) (90). Most of the research performed with plasma source mass spectrometry has used quadrupole mass spectrometers. The work presented in this dissertation was carried out on a Finnigan

PAGE 40

X 100% 29 Figure 4 Ionization Efficiencies for Selected Elements (Adapted from 85)

PAGE 41

30 MAT SOLA ICP-MS, which employed the Balzers QMS 5 1 1 quadrupole mass analyzer Therefore, the discussion which follows concentrates on quadrupole mass analyzers Mass analyzer function The function of a mass spectrometer is to separate an ion beam into its components according to their mass-to-charge ratios (m/z). The mass-to-charge ratio then can be used to determine the composition of the sample. In atomic mass spectrometry, analyte identification is simplified in that the mass-to-charge ratio directly gives the masses of the species present in the sample since most ions are singly charged. The mass spectra obtained are much simpler than those obtained with molecular mass spectrometry. Ion optics A schematic of a quadrupole mass spectrometer used for this research, the Finnigan MAT SOLA ICP-MS, is given in Figure 5. The ion beam from the plasma is directed to the mass analyzer by the ion transfer system, which includes the beam extraction, X deflection plates, Y deflection plates, Y steering plates, and focus. Potentials on the transfer optics help to direct the ion beam into the quadrupole entrance, and deflect unwanted signals from the system. These optics are adjusted to maximize the ion beam which emerges from the mass analyzer.

PAGE 42

31 Figure 5 Schematic Diagram of Finnigan MAT SOLA ICP-MS

PAGE 43

32 OuadaiDole mass analyzer Mass separation using a quadrupole mass analyzer is achieved by varying electric fields The mass analyzer is made up of four long cylinders arranged in a square array, with opposite rods electrically connected DC and RF voltages are applied to opposite pairs of rods. The DC voltage is positive for one pair and negative for the other pair. The RP voltages applied to the rod are equal in magnitude, but have opposite signs The mass spectrum is scanned by varying the amplitude of the voltages, while maintaining the ratio of RF to DC constant. The amplitudes of RF and DC voltages are chosen to transmit ions of only one mass-to-charge ratio at a time. All other mass-to-charge ratios have unstable trajectories and are lost from the system. Quadrupole mass spectrometry is therefore not a simultaneous detection method. Fast sequential scanning of many masses can be accomplished by varying the amplitude of the voltages on the quadrupole, while keeping the ratio of RF to DC constant. The resolution of a mass spectrometer is its ability to distinguish between ions of slightly different mass values. Mass resolution for a quadrupole mass analyzer is set by the ratio of RF to DC voltages. The practical limiting resolution is given in Equation 2. R = c Equation 2 where R is the resolution, n is the number of RF cycles the ions spend in the RF field, and c is a constant, usually 10 25 (91). If ions have too high a kinetic energy when they enter the quadrupole, the number of RF cycles is reduced, which results in degraded

PAGE 44

33 resolution and non-Gaussian peaks that exhibit tailing towards the low mass end Quadrupole mass spectrometers are characterized as having low resolution in that they are only able to separate ions that differ in at least one atomic mass unit. Detectors in ICP-MS There are two main types of detectors used in ICP-MS, the Faraday cup and the secondary electron multiplier. These two detectors are shown schematically in Figure 6 (92) . The Faraday detector consists of a small metal electrode placed in an open reflecting cup. A 1 count per second (cps) signal on the Faraday corresponds to 5 x 10‘‘’ A, which is equivalent to 6 x 10‘® ions per second arriving at the collector The detecting capability of the Faraday plate is limited by Johnson noise of the measuring resistor and various mechanical contributions. The lower operating limit of the Faraday detector is 10‘‘* A. Unlike multipliers, Faraday cups are robust, have long lifetimes, and can be exposed to the atmosphere without damage. However, they are inherently noisy, with background counts of about 2 x lO’. Another drawback of Faraday detectors is their speed. They are relatively slow when compared to multipliers. The multiplier is the detector of choice for small signals, those below 10® ions/s (10'*^ A). Secondary electron multipliers (SEM) work on the principle of releasing electrons from surfaces that are bombarded by energetic particles (ions, electrons, neutral atoms, and photons). This phenomenon can best be described with the aid of Figure 7 (93) . Ion current arriving at the detector is converted into an electron current on the first electrode, known as the conversion dynode. The electron beam is then amplified by

PAGE 45

Beam Deflectors 34 (0 c u. •— ^ c S2 § Q O c c (0 jO jfc 4; Q. E < Figure 6 Schematic of Faraday and Electron Multiplier Detectors

PAGE 46

35 Figure 7. Schematic of Secondary Electron Multiplier

PAGE 47

36 secondary electron emission. One can obtain a gain of up to 10* with a single ion entering the multiplier The number of charges per ion is directly proportional to the current that is measured and amplified A single ion gives a pulse of about 10* electrons at the collector. This pulse is sensed by a fast pre-amplifier. The output is routed to a digital discriminator and counting electronics, which only count pulses above a threshold value Secondary electron multipliers are the detectors of choice for detecting low concentrations and therefore small signals with quadrupole mass spectrometers. They are fast with a high detection sensitivity, and can detect ion currents as low as 10‘‘® A. Background signals are usually 50 (cps) or less. However, the response is dependent upon the mass, energy, charge, and chemical nature of the incoming ions. In addition, they have limited lifetimes and can not be exposed to ambient air for long periods of time without damage. Inductively Coupled Plasma Coupled with Mass Spectrometry The interface of the Finnigan MAT SOLA ICP-MS used in our laboratory is shown in Figure 5. An ICP-MS undergoes differential pumping to achieye the various vacuum requirements. A high vacuum is needed in a mass spectrometer for two reasons. First, gas molecules in the mass analyzer region can cause scattering of analyte ions. Aside fi-om the reduced ion transmission, this results in peak broadening and reduces the resolving power of the instrument. Background peaks can also appear due to residual gases in the mass spectrometer. This too can be minimized with a good vacuum

PAGE 48

37 The vacuum in an ICP-MS can be reached by various types of pumps including cryogenic, diffusion, mechanical, and turbomolecular pumps The latter are the most widely used in modem instruments. Pumps work on the principle of molecular densities. A lower molecular density is established in the pumps than in other parts of the instrument. Molecules are then transported from the system into the pump The plasma is formed at atmospheric pressure in front of the sample cone This cone is housed in a water-cooled interface to prevent melting and is located several millimeters from the end of the load coil. The region behind the sample cone is known as the expansion chamber and is pumped down to a vacuum of 2 3 mbar. The intermediate chamber, located behind the skimmer cone, achieves a vacuum of about 1x10Â’^ mbar. The accelerator cone is found exclusively in the SOLA and is located 10 mm behind the skimmer cone. This third cone in the ICP-MS interface was included to minimize space charge effects (94). It also functions as a differential pumping aperture between the expansion chamber and analyzer chamber, thus minimizing losses due to ion beam scattering. In addition, the accelerator cone focuses ions that pass through it to a fine cross-over line 1 mm behind it, as shown in Figure 8 (94). The quadrupoles are located behind the accelerator cone. The vacuum in the mass analyzer region is pumped dov^ to about 10"Â’ mbar. It can be seen from Figure 5 that the path the ions take from the plasma to the detector is not a straight one. This is employed to minimize the number of photons that reach the detector. Some instruments also use a photon stop for this purpose.

PAGE 49

38 Skimmer Cone Accelerator Cone ////y ^ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ V \ \ Cross Over Point Figure 8. Ion Bean Trajectories Through the Accelerator Cone.

PAGE 50

39 The ICP-MS interface is inefficient. Calculations by Hieftje et al. estimate that the sample cone passes only 25% of the gases in the plasma (95). In addition, the skimmer cone passes about 0.73% of the flow through the sample cone More losses occur as the ion beam is transmitted through the ion optics and mass analyzer Houk et al. reported a single ion is actually detected for about every 10* analyte atoms in the plasma (96) Clearly much work needs to be done on the sampling interface to improve this technique. The interactions of the plasma with the sample cone are shown in Figure 9 (78). A boundary layer of intermediate temperature is formed between the hot plasma and the cooled sample cone. Between the sampler and skimmer cones, the plasma expands adiabatically and forms a supersonic jet. In the jet, the density and temperature of the gas decrease while the speed of the gas stream increases above the speed of sound. The ions are accelerated to the skimmer cone at supersonic speeds and are thus altered little during their transport to this region. The ions reach the skimmer cone in microseconds after leaving the sample cone. The supersonic jet is shown in detail in Figure 10 (97). The expansion is surrounded concentrically by a barrel shock and perpendicularly by a shock wave called the Mach disk. Shock waves result fi-om collisions between atoms from the jet traveling at fast speeds and the background gas, which functions to reheat the atoms and induce emission. The region inside the shock structure is called the zone of silence. The gas flow in this region is well-characterized. The location of the Mach disk from the sampling orifice can be calculated by Equation 3:

PAGE 51

40 <500K -3 <1x10 torr Supersonic Jet Figure 9. Interactions of Plasma with ICP-MS Interface. (Adapted from 78)

PAGE 52

41 Mach Disk Zone of Silence Barrel Shock Figure 10, Supersonic Jet Expansion. (Adapted from 97)

PAGE 53

1/2 X M = 0.67Do VP.; Equation 3 where Xm is the distance between the Mach disk and the sample cone. Do is the sampling orifice diameter, Po is the ICP pressure, and Pi is the background pressure in the expansion chamber (98). The tip of the skimmer cone is placed inside the Mach disk to reduce the effects of collisions and scattering. One inherent problem with ICP-MS is a phenomenon known as space charge effects, the result of coulombic repulsions due to a localized volume of positive ions The plasma is essentially neutral since the ion population is balanced by the electron population. However, the ion optics in the reduced pressure regions are configured for the collection of positive ions. As the ion beam leaves the last cone, the accelerator in the SOLA, the electric field induced by the transfer optics attracts positive ions and repels electrons, which fall out of the ion beam. The ions in the resulting positive beam experience repulsions with each other and produce a defocused ion beam. The lighter ions, with lower kinetic energy, are effected more strongly than the heavy ions and are subsequently lost from the ion beam. Space-charge effects are mass dependent and can have a profound effect on signal intensities when isotope ratios are monitored Space charge effects can be minimized by decreasing the number of ions in the resulting ion beam This can be accomplished by detuning the ICP-MS ion optics, operating the ICP so that fewer ions are sampled, or diluting the sample solution. All of the above methods decrease the sensitivity for individual ions.

PAGE 54

43 Background ions The most prominent species found in an Ar ICP are neutral argon atoms at about 10‘* cm'^ (99). Ar ions are present at lO'’ cm‘^. Oxygen and hydrogen atoms are the next most abundant species, having a number density of 10*’ 10*^ cm ’ Other background ions observed are nitrogen, carbon, and combinations of these with argon and oxygen Several of the more prevalent polyatomic species are given in Table 3. Metal oxides are the most commonly formed analyte polyatomic ions observed in the ICP. They can be formed as a result of numerous interactions involving the plasma gases, air entrainment, solvent, and matrix constituents. Molecular ions observed in the mass spectrum are thought to be formed during the ion extraction processes. The temperature of the plasma is large enough to dissociate any polyatomics that may form. There are theories as to how molecular ions survive the energetic plasma. It is known that temperatures in the plasma are not uniform, temperature gradients can be up to a few thousand degrees. It is thus believed that some polyatomic ions are formed in the cooled boundary regions of the plasma. Some polyatomic species such as and 02 ^ are most probably formed in the plasma and not dissociated. Another theory for polyatomic ion production suggests the formation of oxide ions in the cooled expansion chamber, located behind the sample cone. Douglas and French have estimated that about 250 collisions occur between Ar neutral atoms and other species in the expansion chamber (100). Polyatomic ions, such as molecular oxides, have lower

PAGE 55

Table 3. Prevalent Polyatomic Ions Observed in the ICP-MS Ion m/z 0" 16 oir 17 0 H 2 " 18 20 >r 14 NIT 15 N2^ 28 N2ir 29 NO" 30 Noir 31 02 * 32 At" 40 AtFT 41 ArO" 56 Arc* 52 ArN* 54 AT2* 80 AtN2* 68

PAGE 56

45 kinetic energies than elemental ions of similar mass. Thus, there is time for polyatomic ions to form in the expansion chamber before entering the mass analyzer Interferences in ICP-MS One of the major characteristics of ICP-MS that limits the determination of certain elements is interferences. Interferences can be classified as spectroscopic or nonspectroscopic (matrix effects). Spectroscopic effects include species such as isobaric overlapping ions and polyatomic ions. An isobaric interference occurs when two or more elements with similar mass isotopes are in a sample matrix. A quadrupole instrument, with only unit resolution, would be unable to distinguish between the two masses. An example is the determination of '‘*Ti in a matrix containing '‘*Ca. It is not possible, with a quadrupole ICP-MS, to analyze trace amounts of ‘‘*Ti if significant amounts of calcium are present in the sample. Polyatomic ions are the combination of two or more elements. The classical example is ‘*®Ar‘*0"^ interfering with the determination of ’*Fe. Many solvents add to the production of polyatomic ions, as discussed above. Nonspectroscopic interferences are characterized by a large amount of dissolved solids in the sample, or a signal enhancement or suppression. The former can cause short term drift in the signal. Signal enhancements and suppressions are matrix dependent and element specific. Spectroscopic and nonspectroscopic interferences can be reduced by modifying the sample introduction scheme (e g. using a dry plasma or desolvating the liquid sample), instrument optimization, mixed gas plasmas or solvents, and instrumental design.

PAGE 57

46 Sample introduction Gas, solid, and liquid samples can be introduced into the plasma for mass spectrometric detection. The outlet of a gas chromatographic column can be coupled with the inlet of the torch to allow the analysis of gaseous samples (101) Other means of introducing samples already in the vapor phase into the plasma include supercritical fluid chromatography (SFC) (102) and hydride generation (HG) (103). Solid samples have been introduced into the ICP for mass spectrometric detection by various sampling methods such as laser ablation (LA) (104), slurry nebulization (105), arc nebulization (106), electrothermal vaporization (ETV) (107), and direct insertion (DI) (108). Liquid sample introduction was discussed in detail earlier in this chapter Solutions are normally converted into an aerosol for introduced into the plasma. This is usually accomplished by a ultrasonic (109), pneumatic (110), thermospray (1 1 1), or direct injection nebulizer (DIN) (112). Ultrasonic and pneumatic nebulizers are by far the most popular means of introducing liquid samples into the ICP-MS. Ultrasonic nebulizers have seen widespread use for aerosol production due to their high efficiency. The sample is fed across a piezoelectric crystal that is vibrating in the range of 0.2 10 MHz. As the sample stream is passed over the crystal, large droplets are broken down into smaller ones. Often the aerosol is then passed through a heating region, which produces dried analyte particles and solvent vapor A condenser may then be connected to eliminate solvent vapor fi-om the sample aerosol, while the analyte particles are sent to the plasma. Ultrasonic nebulizers produce smaller, more uniform droplets and are less prone to blockage than pneumatic nebulizers. In contrast to pneumatic nebulizers.

PAGE 58

47 the high aerosol production efficiency of ultrasonic nebulizers is independent of nebulizer gas flow. This results in better sensitivity and lower detection limits since the analytes have longer residence times in the plasma. The main limitation hindering their use is cost. The pneumatic nebulizer has been most used with ICP-MS due to its low cost, convenience, and stability. The principle behind these nebulizers is the disruption of a liquid stream by a high velocity gas, resulting in a fine aerosol of sample Cross-flow nebulizers consists of two capillary tubes located perpendicular to each other. Liquid sample is drawn up one of the tubes and a high-velocity stream of gas is flowed through the other. When the liquid reaches the top of the capillary it is broken into a finely dispersed aerosol. Cross-flow nebulizers are used when the sample solution contains large concentrations of salt, as they are less prone to capillary blockage. A second type of pneumatic nebulizer that has been used with ICP-MS is the Babington-type nebulizer. The aerosol is produced by flowing a film of liquid sample over the surface of a sphere and flowing gas through an aperture below the film. Because of memory effects, Suddendorf et al. made modifications to the Babington-type design (113). The new design consisted of constraining the liquid sample in a V-groove and introducing the nebulizer gas fi-om a small hole in the bottom of the groove. This type of nebulizer is resistant to blockage and is thus used when slurry samples need to be analyzed The most widely employed pneumatic nebulizer for liquid introduction into the plasma is the concentric nebulizer, shown in Figure 11, Liquid sample is introduced perpendicular to the nebulizer gas. The nebulizer gas usually flows through the nebulizer at about 1 L/min. Pneumatic nebulizers are simple to use and are relatively inexpensive They are, however.

PAGE 59

48 characterized by a 1 2% efficiency. A peristaltic pump is used to produce a constant liquid flow into the nebulizer Flow rates lower than 1 mL/min may not allow the aerosol to penetrate the plasma, while higher uptake rates may not provide adequate residence times for the sample particles in the plasma. The resulting aerosol produced contains sample particles with diameters up to 100 pm. Small particles, less than 10 pm, are desired for rapid desolvation, volatilization, and ionization in the ICP. A spray chamber placed between the ICP and nebulizer functions to expel large particles from the aerosol The most popular spray chamber in use is the Scott-type double pass, shown in Figure 12 As the sample aerosol particles enter the spray chamber, the heavier ones fall to the lower walls due to gravitation and are pumped to waste. Smaller particles remain in the argon flow and are carried to the torch. Scott-type spray chambers can be cooled to condense water vapor out of the sample aerosol before it enters the plasma. This results in a decreased solvent loading in the plasma and subsequently decreased background interferences. The sample uptake rate for pneumatic nebulizers is described by PoiseuilleÂ’s equation, shown in Equation 4: 0 = nR*P StjL Equation 4 where Q is the liquid flow rate, R is the capillary radius, P is the pressure differential, ^ is the viscosity of the liquid, and L is the capillary length (114). The above equation

PAGE 60

49 Aerosol Gas Figure 1 1 Schematic of a Concentric Nebulizer

PAGE 61

50 Sample Aerosol Out Coolant Out <4 Sample Aerosol In Figure 12. Scott-type Double Pass Spray Chamber

PAGE 62

51 assumes the velocity of the liquid at the capillary wall is zero Actually, the liquid slips over the capillary and therefore a correction must be added to the Poiseuille equation The R‘‘ term must be replaced by R'* + 4r|R^/B, where B is the coefficient of sticking friction of the liquid on the capillary wall. Sample aerosol exits the spray chamber and enters the plasma through the injection tube of a torch. Torches used with ICP-MS are normally based on the Scott Fassel design (115). Shown in Figure 13, the present design incorporates a 100 mm long quartz torch positioned horizontally. It has an inner diameter of 1 8 mm and encases two concentric tubes with diameters of 13 mm and 1.5 mm. The coolant gas and intermediate gas are bled in through the side to the torch, while the nebulizer gas carries the sample aerosol through the center. The torch is usually positioned 10-15 mm from the sample cone. A liquid sample undergoes several processes as it passes through the nebulizer and spray chamber to the plasma (116, 117). The first process is the generation of a primary aerosol by the nebulizer, which then experiences impaction with the spray chamber to form a secondary aerosol. A tertiary aerosol which is formed by evaporation, impaction, and droplet shattering, coagulation, inertial deposition, gravitational settling and turbulence induced losses, is produced after passage through the spray chamber ICP-MS has many advantages including rapid, efficient sample introduction methods, a large degree of ionization of most elements, simple spectra, multielemental capabilities, low detection limits, and a broad linear dynamic range. As with any analytical technique there are limitations that hinder detection. The ICP-MS is no exception. The interface between the plasma and mass spectrometer is highly inefficient. Only about 1 %

PAGE 63

52 100 mm Sample Aerosol and Nebulizer Gas Gas I 18 mm ID Figure 13. Demountable Quartz Torch

PAGE 64

53 of the ions that enter the sample cone exit the skimmer There are many losses associated with transmitting ions through the ion optics and mass analyzer The detection efficiency of ICP-MS is extremely low. For every 10^ ions produced in the plasma, only one reaches the detector. In addition, this technique is expensive to purchase and maintain, it can not accommodate a high degree of total dissolved solids in the sample, and it is subject to numerous interferences.

PAGE 65

CHAPTER 3 MICROWAVE DIGESTION Introduction Sample preparation procedures are often the most laborious and time consuming steps of an experimental study. This is especially true when the sample is in a nonaqueous matrix. As discussed in Chapter 2, direct analysis of organic samples by ICP-MS is not a trivial matter. The determination of trace levels of elements such as titanium and chromium are impossible by ICP-MS without destruction of the organic matrix. Metallo-organic compounds can be decomposed to produce aqueous solutions of their respective elements by conventional or microwave heating. Conventional sample dissolution in an open container on a hot plate has been used extensively for the extraction of metallic species (118,119). This type of sample decomposition, however, is prone to numerous losses and contamination. In addition, wet ashing techniques generally require long heating times, with many protocols lasting several hours. Microwave energy can be used to heat a sample faster and more efficiently than conventional methods such as wet ashing. Abu-Samra et al. were the first to use microwave heating for organic matter decomposition in 1975 (120). Solid samples were placed in open Erlenmeyer flasks or test tubes with hydrogen peroxide and either nitric acid or a nitric acid/perchloric acid mixture. Complete analyte extractions were carried 54

PAGE 66

55 out within minutes. This was significantly less than the time required for conventional methods The differences in heating times between conventional and microwave digestions can be explained in part due to the heating mechanisms. In conventional systems, a hot plate is used to conductively heat a sample. The sample vessel first absorbs the energy, and then transfers it to the sample. The sample vessel is usually a poor heat conductor, thus resulting in increased heating times. In addition, because vaporization occurs at the liquid surface, convection currents produce a thermal gradient in the sample solution. Thus, only a small fraction of the liquid is at the temperature of the applied heat. As a result, only a minor portion of the sample is above the boiling point of the bulk solution. Microwave energy, on the other hand, directly heats the entire sample simultaneously without first heating the vessel. The energy travels through a microwave transparent vessel to the sample The solution reaches its boiling temperature very quickly, which reduces digestion times. Open vessel microwave digestion procedures can be completed in 5 15 min. It took nearly ten years after the inception of open vessel microwave digestions for closed vessel microwave heating to be explored as a more effective means of decomposing biological materials (121,122). Smith et al. were the first to use PTFE closed sample vessels for microwave digestions (123). Sulfide samples from mines, mills, and smelters were digested for the determination of copper and nickel. Samples were digested in just 3 min. This was a dramatic improvement over the 1 1.5 hr procedure previously used to

PAGE 67

56 dissolve these samples Although the samples were not completely digested, analyses of the residues showed no trace of copper or nickel. Sealed vessels offer several advantages over open vessels for microwave digestion procedures. The energy and rate of a reaction increases with increasing temperature. In closed systems, higher temperatures can be reached which leads to faster reaction times and the decomposition of difficult samples. In addition, the loss of volatile species and the risk of contamination by airborne particles are eliminated by using closed vessels. These systems also give reduced blank levels. This is due in part to the smaller quantities of digestion aides that are used. Today, closed vessel microwave acid digestion is being used for a wide variety of applications such as biological (124), geological (125), metallurgical (126), environmental (127), and marine (128) samples. An extensive description of microwave sample preparation was given by Kingston and lassie in the first book on this subject (129). This is still considered the handbook of microwave dissolution. Microwaves have fi-equencies in the range 300 300,000 MHz. The Federal Communications Commission established four microwave frequencies that are allowed for industrial, scientific, and medical use. Conventional microwave devices used in most homes are operated at 2450 + 13 MHz with output energies of 600 700 W. Theory The dielectric loss theory is used to explain the principles of microwave heating. When microwave energy penetrates a sample, a loss of energy from the electromagnetic

PAGE 68

57 field to the sample occurs. The amount of loss depends on the composition of the sample The rate of this energy loss is dependent upon the sample’s dissipation factor, which is the ability of a sample to dissipate microwave energy as heat. The dissipation factor is the ratio of the sample’s dielectric loss factor to its dielectric constant, as shown in Equation 5 tan5 = — Equation 5 where tan 6 is the dissipation factor, e' is the dielectric loss factor, and e is the dielectric constant (130). The dissipation factor is dependent upon temperature and microwave frequency. A dielectric material is one that obstructs the passage of energy through it due to its composition. The dielectric constant measures the ability of the sample to obstruct energy. The dielectric loss factor measures the ability of a sample to dissipate microwave energy as it penetrates the sample. The use of “loss” here is related to the amount of incoming microwave energy that is lost to the sample by being dissipated as heat. The greater the dissipation factor, the less microwave penetration at a given temperature and frequency. This is because samples with large dissipation factors rapidly absorb and dissipate microwave energy as it penetrates it. Thus, the greater the dissipation factor, the greater the sample’s ability to convert microwave energy into heat. Samples vvdth a significant dipole moment and free rotation are expected to possess large dissipation factors. For example, ice and liquid water have dissipation factors of 0.0009, and 0,1579, respectively. The crystal lattice structure of ice restricts the mobility of water and thus makes alignment of the molecules with the microwave field difficult. This

PAGE 69

58 explains why liquid water is superior to ice in converting electromagnetic energy into heat Polytetrafluoroethylene (PTFE), commonly known as Teflon, has a dissipation factor of 0 00015. This is the material of choice for many digestion vessels since it is virtually transparent to microwave radiation Microwave energy is lost to a sample in two ways: ionic conduction and dipole rotation. When an electromagnetic field is applied to a sample solution, a conductive migration of ions occurs, which produces a current flow. The solutionÂ’s resistance to free flow of these migrating ions results in fnction, which in turn heats the solution. Although all ions in solution contribute to conduction, the relative concentration of an ion and its mobility in the system governs the magnitude of current carried by it. As a result, size, charge, and conductivity all effect the losses due to ion conductance. As the concentration of an ion increases, current and heat transfer also increase due to greater ion mobility. The second mechanism in which microwave energy is lost to a sample is dipole rotation. This refers to the alignment of molecules in a sample that have permanent or induced dipole moments. When an applied electric field is increased, the polarized molecules align themselves with the positive poles of the molecules facing the negative poles of the field, and vice versa. Disorder of the molecular arrangement occurs when the microwave energy is removed. At 2450 MHz, this alignment, followed by disorder, occurs at 4.900 GHz (131). The result is very rapid heating due to molecular fnction. Heating due to dipole rotation is dependent upon temperature, viscosity of the sample, and

PAGE 70

59 the sampleÂ’s dielectric relaxation time. The dielectric relaxation time is the time it takes for the molecules to reach 63% of their return to disorder While both ionic conductance and dipole rotation influence the efficiency of microwave heating, the temperature determines which process dominates. Initially, dipole rotation controls the dielectric loss to the sample. As the temperature is increased, however, ionic conductance dominates the conversion of microwave energy into heat The fraction of contribution from each process depends on ionic mobility and concentration. If these two parameters are small, then the dielectric loss is mainly the result of dipole rotation. Instrumentation The essential components of a microwave digestion system are the microwave generator (magnetron), waveguide, microwave cavity, mode stirrer, and circulator A typical microwave instrument is shown in Figure 14. The magnetron produces microwave energy, which then travels through the waveguide to the microwave cavity The waveguide is constructed out of a reflective material for high transport efficiency At the exit of the waveguide is the mode stirrer, a fan-shaped blade that distributes the energy in many directions throughout the microwave cavity. Once inside the cavity, the microwaves are reflected back and forth between the walls until all the energy has been absorbed by the sample. Reflected energy can exit the microwave cavity and traverse its path back through the waveguide. A terminal circulator is usually placed at the entrance of the

PAGE 71

Dummy Load 60 Figure 14 Schematic of a Microwave Oven

PAGE 72

61 waveguide to direct reflected power to a dummy load. This enables the microwave energy to dissipate as heat and prevents damage to the magnetron Microwave Oven Calibration The actual power absorbed by a sample in the microwave oven can be determined by measuring the change in temperature of a given quantity of water. The relationship between power absorbed by the sample and temperature change is shown in Equation 6 CpKATm P ^ Equation 6 where P is the apparent power absorbed by the sample (W), Cp is the heat capacity of water (cal/g • °C), K is a conversion factor from calories to Joules (4. 184 J/cal), AT is Tf Ti (°C), m is the mass of the sample (g), and t is the time (s). The microwave oven used for sample decomposition was calibrated to determine the approximate power delivered to the oven cavity from the magnetron at each setting on the oven display. Equation 6 can be used for this purpose if it is assumed that the majority of the power delivered to the microwave cavity is absorbed by the sample. The microwave oven was calibrated by measuring the temperature difference of 9 mL of water at each setting on the microwave oven. This sample volume was chosen since it closely simulates an actual sample size. Since heat capacity is dependent upon temperature, it was first necessary to set up a table relating Cp to T. The CRC Handbook contained heat

PAGE 73

62 capacity values at various temperatures, shown in Table 4 (132) However, more exact power calculations could be obtained using smaller increments between points The values listed in Table 4 were inserted into graphing software to generate a curve and an equation describing it. The generated curve is shown in Figure 15 Equation 7 was used to obtain exact heat capacity values at the various experimental temperatures. Cp = 4.22 0.00385X + 1 .74e-^ 5.49eV + 1 40e'Y Equation 7 2.75eY + 3.78e‘V 3.33e‘Y + 1.67e'V* 3.58e‘V The CRC Handbook also lists values for the density of water at various temperatures, shown in Table 5. Again, smaller increments would lead to a more exact calculated absorbed power. The values in Table 5 were inserted into a graphing program and a curve was generated. The curve is shown in Figure 16 and the equation describing the curve is given in Equation 8. p= 1.005 -0.00212x + 3.58eV-3.45e‘V + 2.02eVEquation 8 7.57eV + 1.82e'V 2.72e‘ V + 2.26e‘Y 7.98e'*V The same digestion bomb and PTFE sample vessel used for the metallo-organic digestions was used for the calibration of the microwave oven. Display settings 1 through 5 were calibrated by measuring the temperature rise in 9 mL of water. Each measurement

PAGE 74

Table 4. Heat Capacity (J/g • K) at Various Temperatures (°C). (132) Temperature (°C) Heat Capacity (J/g • K) 0 4.2176 10 4.1921 20 4.1818 30 4.1784 40 4.1785 50 4.1806 60 4.1843 70 4.1895 80 4.1963 90 4.2050 100 4.2159

PAGE 75

Heat Capacity (JK/g) 64 Temperature (°C) Figure 15. Heat Capacity (JK/g) versus Temperature (°C).

PAGE 76

Table 5. Density (g/cm^) at Various Temperatures (°C). (132) Temperature (°C) Density (g/cm^) 0 099984 10 0.99970 20 0.99821 30 0.99565 40 0.99222 50 0.98803 60 0.98320 70 0.97778 80 0.97182 90 0.96535 100 0.95840

PAGE 77

Density (g/cm^) 66 Figure 16 . Density (g/cm^) versus Temperature (°C).

PAGE 78

67 was taken in triplicate. The results are shown in Figure 17 After each measurement the PTFE sample vessels were cooled with deionized water. Microwave Acid Digestion Vessel Microwave acid digestion bombs obtained from Parr Instrument Company, Moline, IL, were used for sample decomposition. The sample cups were constructed out of microwave transparent, thickwalled PTFE. This material is chemically inert, with a melting point of 306 °C. PTFE also serves as an insulator to confine the heat inside the sample vessel. There is, however, one limitation of using sample vessels constructed of PTFE. This material is known to creep or flow when subjected to high pressure or load. At temperatures less than 150 °C, the effect is negligible, however at higher temperatures it becomes more difficult to maintain tight seals. The result is deformation and a shorter life span of the components. The outer portion of the bomb assembly that encased the sample vessel was constructed of a microwave transparent, polymer resin. The bomb and liner are shown schematically in Figure 18. The outer mantle contained a compressible pressure relief disc that served as a safety release. The temperature and pressure of the digesting sample should not be allowed to exceed 250 °C or 1200 psi. Values greater than these may result in the sample vessel exploding. If the pressure inside the vessel exceeds 1500 psi, the relief disc compresses the 0-ring seal on the sample vessel and it is blown out. This design prevents total destruction of the bomb. The retaining screw can be used as a crude pressure indicator. The head of the screw is normally flush with the top of the bomb cap.

PAGE 79

68 Figure 17. Plot of Power Absorbed by the Sample (W) versus Microwave Setting.

PAGE 80

69 Pressure Screw Screw Cap Outer Casing li sp;; VAV .V«SV«S%WfVl.SS\ ^eli eiief Disc Pressure Plate Teflon Cover _ Teflon Sample Vesse Teflon 0-ring Bottom Plate Figure 18. Schematic of Microwave Bomb and Liner

PAGE 81

70 The screw extends out of the bomb cap approximately 1/32 in for every 500 psi pressure increase within the sample vessel The 0-ring has the potential of being blown out after a rise in the screw head of greater than 1/16 in.

PAGE 82

CHAPTER 4 DETERMINATION OF WEAR METALS IN AQUEOUS SOLUTIONS Instrumental Set-up A Finnigan MAT SOLA ICP-MS was used for the elemental analysis of all solutions. A thorough description of the hardware was given in Chapter 2 Specific components are given below. Inductively Coupled Plasma The inductively coupled plasma was sustained by a Henry Electronics Radio Frequency Power Generator, Model 2500D (Los Angeles, CA) The RF power was maintained at a fi-equency of 27.12 MHz, with output powers up to 2.5 kW. The copper induction coil carrying current to the plasma region was obtained from two sources: the chemistry department machine shop and Finnigan MAT. Chilled water flowed through the copper induction coil at a rate of 3 L/min The temperature of the water in the system was kept constant at 12 °C by a Neslab (Newington, NH) CFT-75 refiigerated recirculator. 71

PAGE 83

72 Ouadrupole Mass Spectrometer The hardware of the ICP-MS including ion optics, mass analyzer, and detectors was described extensively in Chapter 2. The quadrupole mass analyzer used in the SOLA was a Balzers QMS 511. The ICP-MS was equipped with a pulse counting channeltron electron multiplier, Model 4870, from Galileo Electro-optics Corporation (Sturgridge, MA) and a Faraday cup detector, ICP-MS Interface The face plate housing the sample cone was kept cool by the same refrigerated recirculator used to supply chilled water to the induction coil. The sample cone was fabricated of nickel and copper due to their high thermal conductivity and electrical conductivity, stability, and machinability. The skimmer and accelerator cones were constructed of nickel. The sample cone was located 14 mm from the end of the load coil and had an aperture of 1.1 mm. The expansion chamber was pumped down to a vacuum of 2 3 mbar by an 18 m^/hr single stage Edwards rotary pump. The skimmer cone was located 8 mm behind the sampler and had an aperture of 0.8 mm. A vacuum of 1 X 10'^ mbar was achieved in the intermediate chamber by means of a Balzers TPH 330 turbomolecular pump (Hudson, NJ), The accelerator cone had an aperture of 1 mm and was maintained at a voltage of +2 kV. The vacuum in the mass analyzer region was pumped down to 4 x 10"Â’ mbar by a Balzers TMH 260 turbomolecular pump. A Balzers TPH 062 turbomolecular pump was used to reach a vacuum of < 5 x lO"Â’ mbar in the detector housing. All three

PAGE 84

73 turbomolecular pumps were backed by a 12 m^/hr Edwards mechanical pump The above pressure values were obtained with Penning ionization gauges, which measured particle densities. Pressures in the corresponding regions were displayed by a Tylan General FC2900V Series mass flow controller (Torrance, CA). The ICP-MS hardware and computer software were interfaced with a Mitsubishi FX Series Programmable Logic Controller (PLC) (Tokyo) through an RS422 interface port in a Gateway 2000 498DX 66 MHz personal computer. The PLC controlled interlocks that ensured proper operating conditions to prevent damage to the hardware The SOLA would shut down if any one of these interlocks were not met. Interlocks found on the SOLA included gas and water flow; coolant and intermediate gas flow, correct positioning of the torch box; adequate pressure in the expansion or intermediate housing; proper operating speed of the turbomolecular pumps in an allocated amount of time, and proper fitting of all instrument panels and lids. The transfer ion optics were optimized every day with a 1 ppm indium tuning solution. Typical potential values for the optics are given in Table 6, It was often necessary to put the ICP-MS in standby mode to obtain an acceptable vacuum in the analyzer region within a reasonable amount of time. Normal procedures for lighting the plasma and waiting for the pressure in the analyzer region to fall to 5 x 10Â’Â’ mbar were not always successful. Pumping in standby mode would sometimes enable the ICP-MS to reach an appropriate vacuum in less time (thirty minutes or less). The SOLA was not equipped with a true standby mode in which the vacuum pumps are operated without the plasma being lit. Therefore, a piece of black electrical tape was used to cover

PAGE 85

74 Table 6. Typical Transfer Ion Optic Potentials Transfer Optic Helipot Setting Voltage Beam Extraction 2.90 -1500 X Deflection 6.52 XI -1470 X2 -1530 Y Deflection 4.16 Y+ -1368 Y-1632 Y Steer 6.24 S+ -1302 s-1698 Focus 5.60 +89.6 Phase Matching Lens 3.92 -205 Interspace 9.68 +14

PAGE 86

75 the sample cone aperture and standby mode was selected from the software The mass analyzer was able to reach a vacuum of mbar in this mode However, since there was no slide valve on the SOLA, the vacuum was broken once the electrical tape was removed, allowing air to be vented into the mass spectrometer Therefore, the standby mode was only useful if the plasma was lit immediately after removal of the tape Sample Introduction Liquid samples were introduced into the inductively coupled plasma through a recessed tip concentric nebulizer (Precision Glassblowing of Colorado, Englewood, CO). The pneumatic nebulizer accommodated an inlet gas pressure of 30 psi (1 L/min Ar) and a liquid flow rate of 1 mL/min. The sample uptake flow was controlled by a Gilson Minipulse 3 peristaltic pump (Middleton, WI). Large solvent droplets were separated from the analyte aerosol by a Scott-type double pass spray chamber (Precision Glassblowing of Colorado, Englewood, CO). The spray chamber contained a 1:1 mixture of antifreeze and water to condense solvent vapor from the sample aerosol. The mixture was kept at a constant temperature of 0 °C by a Neslab (Newington, N?f) RTE-110 refngerated bath/circulator. The sample aerosol entered the plasma through a Fassel-type quartz torch, which contained a demountable injector tube (Precision Glassblowing of Colorado, Englewood, CO). Figure 19 iUustrates the configuration of the nebulizer, spray chamber, and torch. The injector tube of the torch was connected to both the nebulizer and a Nd;YAG laser ablation sample chamber. The laser ablation chamber was used for separate studies not

PAGE 87

76 Figure 19. Sample Introduction Set-up Including Nebulizer, Spray Chamber, and Torch

PAGE 88

77 described here. The argon gas flow through the laser was set at 0 00 L/min while the nebulizer gas flow was set to about 1 .00 L/min. Experimental Parameters Chemicals High purity water, obtained from a Millipore MilliQ water purification system (18MH cm specific resistivity) and HPLC grade water (Fisher Scientific, Pittsburgh, PA), were used for preparation of all solutions (Fisher Scientific, Pittsburgh, PA). Optima nitric acid was obtained from Fisher Scientific and used as received. All aqueous standard solutions were made from dilutions of stock solutions. Solutions were made both in the presence and absence of 5% optima nitric acid. A 10 pg/mL multielement plasma standard (SPEX Industries Inc., Edison, NJ) was used for solutions of Ni and Cr Aqueous solutions of Al, Fe, Mg, and Ti were prepared by serial dilutions of individual 10 pg/mL single element standards (High Purity Standards, Charleston, SC). A 1006 pg/mL Mo standard (SPEX Industries, Inc., Edison, NJ) was used for solutions of this element Aqueous solutions of indium were made by diluting a 1000 pg/mL plasma standard (SPEX Industries Inc., Edison, NJ). All solutions were made in glass volumetric flasks and immediately transferred to PTFE or plastic bottles.

PAGE 89

78 Glassware Glassware and storage bottles used for all solutions were washed with LIQUINOX (Alconox Inc, New York, NY) and ultrapure water They were soaked in approximately 20% nitric acid for at least twenty-four hours and then rinsed once again with ultrapure water. Experimental Parameters Optimization The peristaltic pump was calibrated to determine the relationship between the setting given on the pump and the actual uptake rate. All tubing connected to the pump was filled with ultrapure water. The coolant, intermediate, and nebulizer argon gas supplies were switched on, the pump was started, and the solution was allowed to flow through the tubing while the water collection was timed. This process was carried out for 1 min for various pump settings. The results are presented in Table 7. One can see that a solution uptake rate of 1.0 mL/min is accomplished with the peristaltic pump operating at a setting of about 9.90. This setting was used for subsequent experiments. The efficiency of the concentric nebulizer was also determined by comparing the amount of solution that exits the nebulizer with the amount that is pumped into it. The amount of water used for introduction and collection was determined by weight. A Mettler AE 200 (Mettler Instrument Corporation, Heightstown, NJ) was used for all weighings described here and in later chapters, unless otherwise noted. A dry volumetric flask was weighed before and after the addition of water. The temperature of the water

PAGE 90

79 Table 7, Calibration of the Peristaltic Pump. Peristaltic Pump Setting Solution Uptake Rate (mL/min) 4,50 0.545 9.00 1.09 9.91 1.20 10.5 1.27 11 1.33

PAGE 91

80 was found to be 24 °C. The weight was converted to volume by calculations involving the density of water. Collected water was also subjected to the above weighing procedure The efficiency experiment was repeated for two other solution uptake rates. The results are shown in Table 8. It can be seen that the nebulizer is most efficient at solution uptake rates around 1 mL/min. The percent relative standard deviations (%RSD) obtained from mass spectra signal intensities of a 20 ppb multielement solution (SPEX Industries, Inc., Edison, NJ) were monitored for several elements to determine the optimum solution uptake rate. The results are given in Table 9. Although the nebulizer is characterized by an increased efficiency at higher flow rates, the sensitivity is consequently reduced at these operating parameters. This can be explained by collisions and coalescence of aerosol droplets (83). Even though the nebulizer is more efficient, the larger particles will have fallen out before they completely traverse the spray chamber, thus reducing the amount of analyte that reaches the plasma. The nebulizer gas flow was also varied to determine which value gave the maximum signal for a multielement solution. The results are shown in Figure 20. It appears that a wide range of values will give close to maximum signals for several elements. At the time this experiment was performed, the mass flow controller displaying the nebulizer gas flow was somewhat erratic. The values given as optimum flow therefore could not be taken as concrete. The problem of fluctuating mass flow meter was later

PAGE 92

81 Table 8 . Determination of Nebulizer Efficiency. Solution Uptake Rate Nebulizer Efficiency (mL/min) 0.6 0.43 % 1.2 3.5 % 2.4 1.6 %

PAGE 93

Table 9. Effect of Solution Uptake Rate on %RSD for Various Masses Solution Uptake Rate (mL/min) Be Mg % RSD Ni Co In Ce 0.6 6.1 % 2.6 % 2.5 % 2.3 % 2.2 % 5.5% 0.9 3.7% 3,5% 2.2 % 4.1 % 3.7% 4.4 % 1.2 4,5 % 4.8 % 8.1 % 3.1 % 4.3 % 3.0% 2.4 5,5 % 9.8 % 5,4 % 3,7% 4.6% 12.3%

PAGE 94

83 2.6x106 2.4x106 2.2x106 2.0x106 1.8x106 1.6x10615 1 4x106 c .SP 1.2x106 1.0x106 8.0x103 6.0x103 4.0x103 2.0x103 I — Be I— Mg — Co — In ,— Ni — Ce 0.88 0.90 0.92 0.94 0.96 0.98 1.00 Nebulizer Flow Rate (L/min) 1.02 1.04 Figure 20. Effect of Nebulizer Gas Flow Rate on Analyte Signal.

PAGE 95

84 circumvented with by-passing the pressure regulator in the gas box The above experiment was repeated and similar results were obtained There was an initial increase in analyte transport rate as the flow rate of the nebulizer gas was increased. This is due to an increase in the number of small diameter droplets in the primary aerosol as well as a decrease in the number of droplets that are lost in the secondary and tertiary production processes. After an optimum nebulizer gas flow rate was achieved, a decrease in the signal was observed with increasing flow rates. This was due to the droplets experiencing smaller residence times in the plasma. The large flow rates pushed the particles though the plasma before complete ionization could occur. In addition, a loss of all droplet sizes occurs in the spray chamber, which reduces the transport efficiency. Optimization of the incident RP power was attempted, but was not successful The SOLA had a very small incident power window that could be utilized and still maintain a stable plasma. The ICP-MS used for these studies was not meant for use with “cool” plasma conditions (incident powers < 1000 W). As a result, all work was carried out heeding the instrument manufacturer’s suggestion of operating the plasma at a forward RF power of 1300 W. The above results were examined to determine the optimum operating parameters for mass spectra acquisition. The experimental parameters used are shown in Table 10.

PAGE 96

Table 10, ICP-MS Operating Parameters Parameter Value Plasma Forward Power 1300 W Reflected Power <5 W Coolant Gas 1 5 L/min Intermediate Gas 0,9 L/min Nebulizer Gas 1 L/min Solution Uptake 1 mL/min Data Acquisition Dwell Time 64 ms Number of Channels 16 Number of Scans 1

PAGE 97

86 Acid Effects A study to determine the effects, if any, of nitric acid concentration on analyte signal was carried out on aqueous solutions. Nitric acid is the most widely used acid for the preservation of aqueous solutions, and therefore investigated in this study Several aqueous solutions were made with the analyte concentration held constant at 200 ppb The concentration of the nitric acid was varied from zero to 0,05 M. It can be seen from Figure 21 that in this range, nitric acid has a negligible effect on signal for the elements studied here. The zero (0) on the abscissa in Figure 21 represents no nitric acid added. Data Analysis Aqueous standards of Al, Cr, Fe, Mg, Mo, Ni, Ti, and Y were used to construct linear analytical curves from the limit of detection to greater than 1 jig/mL. The raw data for the analytical curves were obtained from peak height data from the average of five scans. Both peak height and peak area were investigated. Figure 22-A gives the log-log plot of peak area versus peak height for Mg. The slope of the plot was 1.036. A log-log plot of peak area versus peak height was also constructed for Ni. The corresponding plot is shown in Figure 22-B. The slope of this plot was 1.0715. Both methods of data interpretation gave similar results. For simplicity, peak height data was used for all plots. As discussed above, aqueous solutions were constructed from serial dilutions from aqueous standards. A mass spectrum of a 1 ppm solution containing Mg and Al is shown in Figure 23. The sensitivity of the SOLA for Al is slightly higher than for Mg. Figure 24-A and B show mass spectra of a 1 ppm solution containing Cr, Fe, and Ni, and Ti,

PAGE 98

Signal (cps) 87 3.0x1072.5x1072.0x1071.5x1071.0x1075.0x1060 . 0 -Figure 21. I I 3! I * * + Cr • Mg Al Y Ni A Ti Fe X Mo y/i 1 1 1 1 1 1 1 1 r -5 -4 -3 -2 -1 Log [HNO Concentration (M)] Effect of HNO 3 Concentration (M) on 200 ppb Aqueous Solutions. Arrow Indicates No Acid Added.

PAGE 99

88 esjv >ie3cj 6o~] g> 0) X os 0) O) o O) 0) X ra V Q. O) o z QQ ao s QO «3 a u Cu > CO V CO 0> a. Cm O c/i z cu 00 0 1 00 o (S 3 00

PAGE 100

Signal (cps) 89 m/z Figure 23. Mass Spectrum of 1 ppm Mg and Al.

PAGE 101

90 5 (sdo) leuGjs O (sdo) |bu6!S Figure 24. Mass Spectrum of a 1 ppm Aqueous Solution. A) Cr, Fe, and Ni B) Ti

PAGE 102

91 respectively. These mass spectra illustrate one of the greatest advantages of ICP-MS over techniques such as ICP-AES Isotopic information about the sample is readily found from the mass spectrum. In addition, peaks at mass 63 and 64 in Figure 24-A can easily be identified as copper and zinc. It can be seen that ICP-MS spectra are easier to interpret than emission or absorption spectra. Figure 25-A and 25-B show mass spectra of 1 ppm aqueous solutions of Y and Mo. Analytical curves for aqueous solutions of Al, Mg, Fe, and Y are given in Figure 26, and Figure 27 shows the corresponding curves for Cr, Ni, Ti and Mo. Limits of detection were calculated for these elements based on the definition shown in Equation 9 LOD = (3 X ab)/m Equation 9 where Ob is the standard deviation of the blank signal and m is the slope of the analytical curve (133). As shown in Table 11, detection limits in the low ppb region were obtained for Al, Mg, Fe, Cr, and Ni. The %RSD for these measurements range from 2% 6 % over the entire calibration range. Aqueous solutions of Y, Mo, and Ti gave detection limits in the pptr region with about 5% RSD. The measurements of the elements that gave ppb region detection limits were all blank limited. The mass spectrum of 5% HNO 3 (the blank used for the analyses) of Mg, Al, and Fe, shown in Figure 28, and Ti and Cr, shown in Figure 29, show distinct peaks for these elements. The HNO 3 used for the solution preparation contained these elements in detectable levels. The peak at mass 44 is most likely not due to Ti, but which is known to be present in the mass spectrum of 5 % HNO 3 (134). The Ti peak at mass 48 is probably due to contamination on the mass

PAGE 103

80I.X2 I. 92 L J I I 1 I I L S r*» p«> r*» o o o S 5 S S ^ o6 (b ^ (sdo) |bu6!S j_ r**. O s (vi o d s

PAGE 104

I.2xl9>i 7x19 93 Figure 26. Analytical Curves for Aqueous Solutions. A) A1 B) Mg C) Fe D) Y

PAGE 105

30x1(7 94 r§ r§ r (sdo) |eu6|S (sdo) |eu6js o E a. c o c 8 c o (sdo) |eu6|s (sdo) (BuSis Figure 27. Analytical Curves for Aqueous Solutions. A) Cr B) Ni C) Ti

PAGE 106

95 Table 11. Detection Limits for Aqueous Solutions. Element Detection Limit Aqueous (ng/mL) %RSD* A1 0.2 6% Mg 3 2% Fe 2 5% Cr 1 4% Ni 1 2% Y 0.02 3% Mo 0.01 5% Ti 0.03 4% Average of 5 runs at lOOx LOD

PAGE 107

1.0x105 p 2.5x106 96 s S -|S S 0 0 0 s i s cd
PAGE 108

97 Figure 29. Mass Spectrum of Ti and Cr Peaks in 5% HNO3.

PAGE 109

98 spectrometer ion lens and quadrupoles that indicate a need for cleaning The mass spectrum of 5% HNO 3 , shown in Figure 30, also exhibited Ni peaks The detection limit of nickel was also limited by an unavoidable experimental parameter In addition to the unavoidable experimental parameter. In addition to the nickel content of the HNO 3 , the sample cone used in the SOLA, supplied by Finnigan, and contained nickel. The most obvious reason Y and Mo gave better results than the other elements was their relatively low blank levels. As shown in Figure 31, a solution of 5% HNO 3 gave no discernible peaks for Y and Mo. In addition, the quadrupoles were not subjected to these elements as frequently as the other metals in this study. Thus, the wash out times for Y and Mo were substantially less than for the other elements. It should be noted, however, that to regularly achieve pptr range detection limits, the quadrupoles and lenses must be cleaned thoroughly and frequently. The above pptr region detection limits were obtained after cleaning the mass analyzer with a paste of alumina and water. After several months of analyzing the solutions mentioned here as well as others, detection limits for Y and Mo were in the sub ppb region. Error Analysis There are errors associated with every measurement Experimental errors are grouped into systematic and random errors. Systematic errors are usually the result of operator error and can be isolated and reduced. Examples of systematic errors include not properly taring a balance and using a buret that has not been calibrated correctly. It is assumed that systematic errors have been eliminated in this dissertation. Random errors.

PAGE 110

99 Figure 30. Mass Spectrum of Ni Peaks in 5% HNO3.

PAGE 111

5000 p 4500 4000 3500 3000 2500 2000 1500 1000 500 0 100 75 80 85 90 95 100 m/z Figure 3 1 , Mass Spectrum of 5% HNO 3 .

PAGE 112

101 on the other hand, are easier to quantitate. These errors are the ultimate limitation on a particular measurement There were several steps in the preparation of aqueous standards that affected the random errors of the measurement. The flow chart in Figure 32 illustrates how these steps effect the final measurement. A 0.5 ppm solution was prepared from a stock aqueous solution. The random error in this step occurs with pipetting the first aliquot. Aliquots were then dispensed with a transfer pipet or Eppendorf pipet. The tolerance of the transfer pipet depends on the calibrated volume. The stock aqueous solution was purchased from a commercial manufacturer and had a concentration of 10 + 0.05 ^g/mL. For illustration purposes, a 5 mL volumetric pipet with an error of + 0.01 mL will be considered. This volume was dispensed into a volumetric flask and diluted to calibration mark. Several flask sizes were used for the standards. Again for the sake of illustration, a 100 mL flask with a tolerance of + 0.08 mL will be considered. The above dilution is shown in Equation 10. Dilution errors add quadradically and can introduce a significant error. (10 + 0.05 |ig/mL) X (5 + 0.01 mL) = (X) x (100 + 0.08 mL) Equation 10 From Equation 10, a concentration of 0.5 + 0.003 |ig/mL is obtained. The error was calculated by following the example in Equation 11, where Sr is the random error (standard deviation) in the result, R is the result, Sa, Sb, Sc are the random errors in the

PAGE 113

102 Figure 32. Flow Chart Illustrating Random Errors Associated with Preparing Aqueous Standards

PAGE 114

103 individual values and A,B,C are the values. Equation 12 gives the actual numbers used for the calculation. V A/ B) + \CJ kV =f^V +f^V IrJ V 10 J V 5 y uooJ Equation 1 1 Equation 12 The 0.5 + 0.003 |ig/mL solution was further diluted to prepare other aqueous used to construct the analytical curve. The preparation of a 0.1 pg/mL (100 ng/mL) aqueous solution will be used as an illustration. An example of calculating the absolute random uncertainty associated with preparing a 25-fold dilution of a 0. 1 pg/mL aqueous solution is shown in Equation 13 (see Figure 32), " 0.003y I 0.5 y 0.03y 25 > + 0.08 A 2 V 100 ; Equation 13 Solving Equation 13 gives a final concentration of 0. 1 + 0.0006 pg/mL (100 + 0.6 ng/mL). Therefore, the serial dilutions for preparing aqueous standards results in 0 6% theoretical random error Random error calculations were also carried out on the raw data for a 200 ng/mL aqueous solution. Table 12 summarizes the theoretical and experimental random errors. The theoretical random error is that associated with sample preparation only.

PAGE 115

104 Measurement error is not included because it varies experimentally with concentration and element The large difference between experimental and theoretical illustrates a significant portion of the random error is introduced in the measurement stage In all cases, the experimental random error is larger than the theoretical random error

PAGE 116

105 Table 12. Summary of Theoretical and Experimental Random Errors. Metal Experimental Concentration (ng/mL) Experimental Random Error (ng/mL) Theoretical Concentration (ng/mL) Theoretical Random Error* (ng/mL) A1 106 5 100 0.6 Cr 105 2 100 0.6 Fe 119 4 100 0.6 Mg 104 17 100 0.6 Mo 90 9 100 0.6 Ni 105 2 100 0.6 Ti 100 4 100 0.6 Y 100 3 100 0.6 Based on sample preparation only

PAGE 117

CHAPTER 5 DETERMINATION OF METALLO-ORGANIC WEAR METALS IN OIL SAMPLES ICP-MS Instrumental Set-up The instrumental parameters for the determination of metallo-organic wear metals in lubricating oils are identical to those used for the analysis of aqueous solutions. The sample introduction scheme, plasma operating conditions, mass spectrometer configuration, and data acquisition are given in detail in Chapter 4. Microwave Digestion Instrumentation A General Electric Model # J ES65T 001 microwave oven was used for all digestions. Microwave acid digestion bomb Model #4781 (Parr Instrument Company, Moline, IL), with a capacity of 23 mL, and Model #4782, with a capacity of 45 mL, was used for sample decompositions. These digestion cups have a maximum dry organic sample load of 0.1 g and 0.2 g, respectively. In addition, Parr recommends a maximum 3 mL and 6 mL of concentrated (70%) nitric acid for the digestion of organic materials. Values greater than these may result in unstable conditions and should be avoided 106

PAGE 118

107 Experimental Parameters Chemicals Water from the same Millipore MilliQ water purification system used to obtain ultrapure water for aqueous solutions was also used in the preparation of digested metallo-organic samples. Optima nitric acid, 30% certified A.C.S, hydrogen peroxide, and HPLC grade water (Fisher Scientific, Pittsburgh, PA) were used as received, Metallo-organic oil standards were digested and used for the construction of analytical curves. Individual metallo-organic oil standards of 1000 pg/g Cr, Mo, Ni, and Ti (High Purity Standards, Charleston, SC), and 500 pg/g Al, Mg, Fe, and Y (Johnson Matthey, Ward Hill, MA) were used to prepare oil samples for these elements. It is imperative that these samples be stored in the dark. Several metallo-organic standards were left out in the laboratory. Several months later the samples gave low results for the particular metal present in the sample. The Cr sample had developed a gray solid at the bottom of the sample bottle. Blank matrix oil was obtained from High Purity Standards and Johnson Matthey. Standard Reference Material 1084a and 1085a were obtained from NIST (Gaithersburg, MD).

PAGE 119

108 Glassware Glassware and storage bottles were cleaned in an identical manner to that described in Chapter 4 for the preparation of aqueous solutions. Microwave Acid Digestion Procedure Bettinellli et al. described a microwave acid digestion procedure for the extraction of several elements from fuel oil (72). A PTFE sample vessel was used for the decomposition of fuel oil. Nitric acid and hydrogen peroxide were used as the digestants. Six samples were digested simultaneously in different vessels. This procedure was attempted in our laboratory without success. Initially, 5 mL of nitric acid and 3 mL of 30% hydrogen peroxide were placed in the digestion vessel with 252 mg of metalloorganic aluminum. Approximately one minute into the heating cycle given in the above manuscript, the bomb exploded. The plastic bomb fractured in half and the rotating glass plate was shattered. The door of the microwave oven was blown open, which ceased the microwave energy from continuing. Damage to the microwave oven was limited to a blown fuse. The combination of several variables may have caused the mishap. First, the sample size used exceeded the maximum organic sample load for the particular vessels used. The 23 mL bomb liners were initially used and have a maximum digestant (e.g. HNO3 and H2O2) capacity of 5 mL, which was also exceeded. In addition, the bombs had been stored in the laboratory for an undetermined amount of time. The complete history

PAGE 120

109 of the samples used with them was unknown The pressure release screw did not activate, which would have prevented fracturing of the bomb. The faulty screw may have been damaged due to the continued use of HCl many years ago Hydrochloric acid has been known to cause the release screw to stick and therefore not function properly Modifications to the microwave digestion procedure given above were investigated. Several parameters such as sample size, amount of digestants, and heating times were attempted until an adequate procedure was established. The final heating schedule is presented in Table 13 below. Digested metallo-organic oil samples were prepared by weighing 25 60 mg of oil directly into the PTFE microwave digestion bomb liner. Sample loads in excess of this resulted in the release of the pressure disc of the bomb due to a dangerously high pressure build-up. Nitric acid (5 mL) and 30% hydrogen peroxide (4 mL) were then added. It was found that volumes less than these sometimes resulted in incomplete digestion. The bomb was sealed and subjected to the microwave procedure given in Table 13. After the last heating step, the bomb was water cooled for 1 hour Solutions were diluted with water and immediately quantitatively transferred to PTFE or plastic bottles for storage. Blank oil digestions were also performed by digesting matrix oil with the same amounts of nitric acid and hydrogen peroxide give above The resulting aqueous solutions were diluted in a similar fashion as the metallo-organic samples.

PAGE 121

Table 13. Microwave Acid Digestion Procedure. Step Approximate Power (W) Time (s) 1 20 80 2 40 30 3 0 40 4 40 30 5 0 40 6 40 30 7 0 40 8 40 30 9 20 20

PAGE 122

Ill Data Analysis After the metallo-organic oil samples were microwave digested and converted into aqueous solutions, they were then measured by Inductively Coupled Plasma Mass Spectrometry, Mass spectra of a digested multielement metallo-organic sample containing Mg and Al, and Ti, Cr, Fe, and Ni are shown in Figure 33-A and B, respectively. Figure 34 shows mass spectra of digested metallo-organic Y and Mo Linear analytical curves were constructed for Al, Mg, Fe, and Y, shown in Figure 35, and Cr, Ni, Ti, and Mo, shown in Figure 36, The analytical curves for digested metallo-organic samples shown in Figures 35 and 36 are plotted on the same graph as the aqueous solutions. It is clear that the digestions were complete due to overlapping graphs with similar slopes. Table 14 shows the results of the metallo-organic study. The digested metallo-organic samples gave similar results as the aqueous solutions with regards to detection limits and %RSD. Detection limits for Al, Mg, Fe, and Cr are in the low ppb region, while those of Ni, Y, Mo, and Ti are in the pptr region, with the %RSD of these measurement ranging from 3% to 10%. NIST Standard Reference Material 1084a and 1085a were used to evaluate the accuracy of the microwave acid digestion method. The reference materials were subjected to the microwave digestion procedure described above. For each element at least three NIST samples were independently digested and compared to the signals of the digested metallo-organic samples. Figure 37 shows mass spectra of NIST SRM 1085a The results

PAGE 123

112 Figure 33. Mass Spectra of Digested Metallo-Organic Oil. A) 650 ng/mL Mg and 326 ng/mL A1 B) 654 ng/mL Ti, 650 ng/mL Cr, 650 ng/mL Fe, and 654 ng/mL Ni

PAGE 124

113 (sdo) |bu6|S Figure 34. Mass Spectra of Digested Metallo-Organic Oil. A) 8.3 pg/mL Y B) 654 ng/mL Mo

PAGE 125

114 Figure 35. Analytical Curves for Digested Metallo-Organic Oils. A) A1 B) Mg C) Fe D) Y

PAGE 126

115 8 I f § I M 8 (scb) |«u6«s Figure 36. Analytical Curves for Digested Metallo-Organic Oil. A) Cr B) Ni C) Ti D) Mo

PAGE 127

116 Table 14. Figures of Merit for Metallo-Organic Samples. Element Detection Limit^ Digested Oil (ng/mL) %RSD* Detection Limit* (ng/mL) %RSD* A1 1 6% 2 <20% Mg 2 4% 2 3-27% Fe 3 10% 1.3 <10-25% Cr 4 4% N/A N/A Ni 0.2 2% N/A N/A Y 0.01 3% 0.7 15% Mo 0.07 6% N/A N/A Ti 0.02 3% N/A N/A Average of 5 runs at lOOx LOD ^Present Work ^TV-ICP-MS Data from (73)

PAGE 128

117 of this study are shown in Table 15 For the most part, experimental concentrations show good agreement with certified values. A comparison of microwave acid digestion ICP-MS with ETV-ICP-MS work carried out previously in this laboratory is shown in Table 14 (73). This clearly shows increased precision of the present method over electrothermal vaporization Problems associated with tedious pipetting oil samples into the graphite furnace have been eliminated. Determination of Cr, Ni, Mo, and Ti were not carried out by Escobar et al. and therefore not available (N/A) for comparison. A comparison of the present data with that of other researches for the analysis of trace metals in lubricating oils is given in Table 16. The microwave acid mineralization procedure used by Bettinellli et al for the determination of trace metals in fuel oil is described above (72). Anderau et al. analyzed lubricating oils by ICP-OES. The samples were diluted with kerosene and then directly nebulized into the ICP (37). Fischer et al. used a Babington V-groove nebulizer in conjunction with a heated spray chamber for the determination of metals in edible and lubricating oils (41). Graphite furnace AAS was carried out by Lukas (35). It can be seen from Table 16 that for the trace multielemental analysis of lubricating oils, microwave acid digestion followed by ICP-MS gives the lowest detection limits. Error Analysis The preparation and microwave acid digestion of metallo-organic samples is more complex than the preparation of aqueous standards. Therefore, more random errors are

PAGE 129

118 Table 15. NIST SRM 1084a and 1085a Results. Element SRM NIST 1084a (Itg/g) Exp. Cone, (ng/g) % Error SRM NIST 1085a (lAg/g) Exp. Cone, (lig/g) % Error Mg 99.5 95.5 4.0% 296.0 292.8 1.1% Fe 98.9 108.3 9.5% 296.8 300.3 1.2% Ni 99.7 102.3 2.6% 302.9 302.8 0.03% Cr 98.3 98.9 0.6% 296.3 292.1 1.4% Ti 100.4 85.3 15% 305.1 297.1 0.67% Mo 100.3 98 2 2.0% 302.9 299.2 2.6% Al^ 104 105 0.81% 289 254 12% ^A1 concentration not certified, just given for reference

PAGE 130

Table 16. Comparison of Microwave Digestion ICP-MS with Other Techniques for the Determination of Trace Metals in Lubricating Fluids. 119 1— 1 a ^ 0.04 0.0002 0.01 0.004 0.05 N/A 0.03 0.3 ICP-OES^ High Temp. Nebulization (Pg/g) 0.328 0.003 0.032 0.051 0.131 N/A N/A 0.045 ICP-OES* Direct Nebulization (mg/L) 0.021 0.05 0.018 0.003 0.07 N/A 0.043 N/A c g to ^ 0.87 N/A 1.28 0.10 0.12 N/A 0.03 N/A o o • c s >1’^ s r i s && — cNmO^qoo 0^000 c o E V Crf) ^ Um r ^ k. O u
PAGE 131

120 introduced Figure 38 gives a flow chart illustrating the random errors associated with preparing microwave acid digested oil samples. An example calculation of the random errors in the final solution is given below. A sample of approximately 30 + 0.014 mg was taken from the original 1000 + 5 pg/g metallo-organic commercial standard. This sample was then digested with HNO3 and H 2 O 2 . There was negligible random error associated with the introduction of the digestants. HNO3 and H2O2 are not the limiting reagents in the decomposition of the organic sample. Therefore, a fraction of the digestants will remain after the decomposition. Although drastic inconsistencies with the amount of digestants added would affect the completeness of the digestion, the final dilution was the most important with regards to accurately calculating concentrations. After the microwave digestion procedure was complete, the sample was quantitatively transferred to a volumetric flask. The sample vessel was rinsed several times with fi-esh water and these washings added to the same volumetric flask. Therefore, it could be assumed that the errors associated with transferring the sample from the vessel to the volumetric flask was negligible The digested metallo-organic samples were usually diluted to 25 + 0 03 mL Calculations show that the original 1000 + 5 pg/g metallo-organic standard was converted to a 1.200 + 0.007 pg/mL aqueous solution. The random errors associated with the sample preparation steps add quadradically and are shown in Equation 14.

PAGE 132

121 Stock Solution 1 000 + 5 )ig/mL r Weighing Stock Solution 30.00 + 0.014 mg 1 r Addition of HNO and H 0 3 I 2 negligible r Sample Transfer negligible 1 First Dilution 25 ± 0.03 mL First Dilution 1 .200 + 0.007 ug/mL r Pipetting Standard Solution 10 + 0.02 mL r Second Dilution 50 + 0.05 mL r Random Error in Final Solution 240 + 1 ng/mL Random Error in First Dilution 1 .200 + 0.007 ng/mL Figure 38. Flow Chart Illustrating the Random Errors Associated with Digesting Oils Containing Metallo-Organic Wear Metals

PAGE 133

122 f + 0.014 V 30 + "o.o3y V 25 J Equation 14 The 1.200 + 0 007 iig/mL solution was further diluted to prepare other solutions used to construct an analytical curve The preparation of a 0.240 pg/mL (240 ng/mL) aqueous solution is used as an illustration. A 10 mL aliquot was taken from the concentrated solution and pipetted into a 25 + 0.03 mL volumetric flask The final concentration of the digested sample was 240 + 1 ng/mL. The random error associated with the dilution of the 1 .200 + 0.007 pg/mL solution is given in Equation 15 rsR_y _r o.oo7y ^ro.o2y ^ro.o5y . rJ ” I 1.2 J 10 ) Equation 15 Therefore, the preparation of microwave acid digested metallo-organic samples resulted in 0.4% theoretical random error (sr/R x 100) in the final solution. The experimental random errors were also calculated for each element under investigated. Table 17 summarizes the theoretical and experimental random errors. The theoretical random error excludes the measurement error, which changes with element and concentration. It is the random error based on sample preparation only. The experimental random error was larger than the theoretical random error for each element studied. Random errors associated the preparation of microwave acid digested NIST SRM samples were also calculated. In this case, only the experimental random errors were

PAGE 134

123 Table 17. Summary of Theoretical and Experimental Random Errors, Metal Experimental Concentration (ng/mL) Experimental Random Error (ng/mL) Theoretical Concentration (ng/mL) Theoretical Random Error* (ng/mL) A1 237 10 245 1 Cr 305 44 274 3 Fe 295 30 263 1 Mg 380 35 320 2 Mo 239 7 238 1 Ni 254 3 229 1 Ti 288 6 299 2 Y 118 5 121 0.7 *Based on sample preparation only

PAGE 135

124 calculated A certificate of analysis was shipped with the NIST SRM oil samples Therefore, the experimental results were back calculated to obtain the concentration of wear metal in the original sample. Tables 18 and 19 summarize the random errors associated with the preparation of NIST SRM 1084a and 1085a, respectively. The confidence intervals for the experimental concentrations were calculated from Equation 16. p = X ± Equation 16 The value of t was taken as 4.303, the value for a 95% confidence level of the average of 3 data points. It can be seen that the experimental concentrations fall within the confidence limits.

PAGE 136

125 Table 18. Summary of Random Errors and Confidence Intervals Associated with the Preparation of Microwave Acid Digested NIST SRM 1084a. Metal Theoretical Concentration and Random Error* (tig/g) Experimental Concentration and Random Error* (tig/g) Experimental Concentration and Confidence Interval (tig/g) % Error^ Cr 98.3 + 0.8 98 9 + 9.8 98.9 ±38 0.61% Fe 98.9+1.4 108.3 ±11 108.3 ±8 9.5% Mg 99.5 ± 1.7 95.5 + 15 95.5 ±33 4.0% Mo 100.3 ± 1.4 98.2+15 98.2 ±37 2.1% Ni 99.7+1.6 102.3 + 8.4 102.3 ±45 2.6% Ti 100.4+1.3 85.3 ±5.3 85.3 ±52 15% Standarc deviation of concentration ^ % Error |Experimental Theoretical] X 100 Theoretical

PAGE 137

126 Table 19. Summary of Random Errors and Confidence Intervals Associated with the Preparation of Microwave Acid Digested NIST SRM 1085a Metal Theoretical Concentration and Random Error* (lig/g) Experimental Concentration and Random Error* (lig/g) Experimental Concentration and Confidence Interval (Pg/g) % Error^ Cr 296.3 + 3.3 292.1 ±22 292. 1 ± 56 1.4% Fe 296.8 ±2.7 300.3 ± 18 300.3 ±28 1.2% Mg 296.0 ±3.1 292.8 ±26 292.8 ±52 1.1% Mo 302.9 ±4.1 299.2 ±43 299.2 ± 16 1.2% Ni 302.9 ±6.8 302.8 ±65 302.8 ± 142 0.03% Ti 305.1 ± 10.0 297.1 ±42 297.1 + 74 2.6% * Standard deviation of concentration ^ % Error = |Experimental Theoretical| X 100 Theoretical

PAGE 138

CHAPTER 6 DETERMINATION OF PARTICULATE WEAR METALS IN OIL MATRICES Instrumentation The instrumental parameters for the determination of particulate wear metals in lubricating oils are identical to those used for the analysis of metallo-organic samples. This includes all ICP-MS variables as well as the microwave digestion hardware. These parameters were described extensively in Chapters 4 and 5. In addition, a Perkin Elmer AD2B Microbalance and a Mettler AE 200 balance were used to weigh out the metal particles. Experimental Parameters Chemicals The Millipore MilliQ water purification system described previously was used to obtain ultrapure water. Optima nitric acid, optima hydrochloric acid, optima hydrofluoric acid, optima sulfuric acid, 30% certified A.C.S. hydrogen peroxide, and HPLC grade water (Fisher Scientific, Pittsburgh, PA) were used as received. 127

PAGE 139

128 Blank oil (High Purity Standards, Charleston, SC) was used to prepare the standard oil suspensions. Metal particles were obtained from Goodfellow Cambridge Limited (London). The diameters of the particles are given in Table 20. The size of the particles ranged from 2 pm (Ni and Mo) to 500 pm (Y). Glassware Glassware and storage bottles were cleaned in an identical manner to that described in Chapter 4 for the preparation of aqueous solutions. All digested solutions were made in glass volumetric flasks and immediately transferred to PTFE or plastic bottles. Preparation of Standards HNOj and H 7 O 7 Digestion of Fe. Mg. Y, Ni, Mo It was not possible to purchase standards consisting of metal particulates suspended in an oil medium. Therefore standards were constructed by doping blank oil with metal particles. The first attempt at preparing standards was not successful. Initially 40 pg of Fe was weighed into 46 mg of blank oil. The PTFE microwave acid digestion sample vessel was too large to place directly onto the microbalance. Very small pieces of PTFE were cut and placed on the microbalance. The metal particles were weighed directly onto the PTFE slabs, and then transferred to the sample vessel. The blank oil was

PAGE 140

129 Table 20. Diameters of the Particles Used for Preparing Standards. Metal Particle Diameter A1 max: 5 fim Mg max: 50 pm Fe mean: 8 pm Y max: 500 pm Cr max: 5 pm Ni mean: 2 pm Ti max: 150 pm Mo mean: 2 pm

PAGE 141

130 then added to the sample vessel, on top of the metal particles. The digestants, 2 mL of nitric acid and 1 mL of hydrogen peroxide, were added to the sample bomb liner as well. The sample was microwave digested and cooled for one hour in a water bath. Once cooled, the sample was diluted with water. Serial dilutions of this sample were made and an analytical curve was constructed. Two separate NIST Standard Reference Oil samples were subjected to the digestion procedure outlined in Chapter 5 to test the accuracy of the microwave acid digestion procedure. These samples were compared to the analytical curve constructed from the particle doped oil samples. Calculations showed that NIST SRM 1084a and 1085a gave concentrations that were 46% and 54% low, respectively. It was concluded that this was due to the many errors associated with weighing and transferring the particles from their original sample bottle to the PTFE weighing slab and finally to the bomb. Another error encountered in weighing was the limit of the microbalance. The balanced was only able to be read to the hundredths place for the milligram scale. Expense of these types of balances prevented purchasing an instrument that would allow for more significant figures to be read. The quantity of metal particles was increased even further to give 2 significant figures, instead of 1 and the microwave digestion procedure was repeated. A quantity of 0. 12 mg of Fe particles and 53 mg of blank oil were digested. The analysis of NIST SRM 1084a and 1085a resulted in calculated concentrations that were 19% and 35% low, respectively. Weighing even larger and larger quantities of metal particles did not increase the accuracy of the procedure. This technique required a steady hand. A significant

PAGE 142

131 portion of the errors resulted in the transference of the particles to the PTFE slab and then to the sample vessel. The particles used in these experiments developed a charge and would be lost from the microspatula very easily. In addition, the microbalance was not housed in the same laboratory as the ICP-MS. Added steps in the sample preparation procedure, therefore, increased the risk for contamination. Another metal, Al, was subjected to the microwave acid digestion procedure to determine if the above errors were due poor experimental techniques or just element specific. A quantity of 0. 16 mg of Al and 61 mg of blank oil were digested with nitric acid and hydrogen peroxide. Calculated concentrations for these samples gave extremely low results (> 70% loss of analyte), indicating either an incomplete digestion or loss of the sample before the digestion procedure. Another Al particle digestion was carried. The amount of nitric acid was increased from 2 mL to 3 mL, and the amount of hydrogen peroxide was increased from to 1 mL to 2 mL. After digesting and cooling, the sample was diluted and serial dilutions were performed. Most of the earlier digested samples were clear yellow. However, this solution, with increased amounts of HNO3 and H 2 O 2 , was clear with no yellow color. Calculations of NIST SRM gave concentrations that were approximately 2 7% low. Microwave acid digestions were repeated with Al particles several times. The results were not reproducible. Most digestions resulted in low results. Although increased amounts of digestants sometimes improved the digestion procedure, another factor was effecting the results. The low results could not be explained by incomplete digestion alone since increasing the amount of digestants did not reproducibility increase

PAGE 143

132 the recovery. The loss of sample during the weighing procedure was still critical. In addition, the low results were not element specific. The amount of digestants seemed to be a good place to start modifying the procedure. The amount of nitric acid and hydrogen peroxide were altered in varying combinations to develop an adequate procedure. The size of the sample vessel limited the total amount of digestants to 5 mL. Therefore, the amounts of HNO3 and H2O2 could not be increased any further. Instead two individual digestions were performed on the same sample. Aluminum particles and blank oil were added to the sample vessel, and then 3 mL HNO3 and 2 mL H2O2 were added. The sample was microwave digested and allowed to cool. The sample was then decanted into a volumetric flask. The residue was digested once again with the same amount of digestants. Once the sample had cooled, it was decanted into the same volumetric flask. The sample was diluted and treated in a similar manner as the previous digested samples. This procedure resulted in more complete digestions. The double digestion procedure was also performed on Mg particles. The result was a more complete digestion, albeit a time-consuming one. Larger microwave acid digestion bombs and sample vessels were used to enable greater amounts of digestants to be used in a single digestion. The new microwave digestion procedure used 5 mL HNO3 and 4 mL H2O2. Particles were still being weighed in the same tedious manner of weighing onto a piece of PTFE on the microbalance and then transferring the PTFE slab to the sample vessel. The new method involving an increase in the amount of HNO3 and H2O2 was attempted for several particle/oil suspensions such as Fe, Al, and Mg. Erratic and non-reproducible results were obtained.

PAGE 144

133 Therefore, the low recoveries were due to weighing errors as well as incomplete digestions. Once again it was thought that increasing the sample size even further in conjunction with increased quantities of digestants might help the metal recoveries. A large batch of Fe particles suspended in oil was prepared. The concentration of the suspension was 3400 ppm Fe. Before an aliquot was taken, the sample was placed in a mechanical mixer (Spex Industries, Princeton, NJ) to distribute the particles throughout the oil. However, the particles settled out of the oil very quickly. Even though sampling was performed immediately after the suspension was removed from the mixer, this appeared not very accurate. A less concentrated batch of Fe particles suspended in oil was also prepared. Microwave acid digestion of that sample also gave low recoveries. The search began for a more accurate method of preparing and sampling the oil suspension. It was decided to digest a larger sample size (10-20 mg). Iron particles suspended in blank oil were digested with HNO 3 (5 mL) and H 2 O 2 (4 mL). After the microwave acid digestion procedure was carried out, the sample was cooled for 1 hour. The Fe particles and blank oil were completely digested. It was discovered, however, that the procedure was not always reproducible, sometimes the particles were not completely digested. This was due to a bad 0-ring on the PTFE sample vessel, which did not allow for a good tight seal. Once the 0-ring was replaced, the digestions worked very well. Ten separate microwave digestions were carried out. The average recovery for Fe was 99% with a relative standard deviation of 14%.

PAGE 145

134 Mg, Y, Ni, and Mo particles in oil were microwave acid digested in a similar manner with encouraging results. The recoveries for Mg, Y, Ni, and Mo were 98% with 24% RSD, 117% with 20% RSD, 118% with 1 1% RSD, and 120% with 1 1% RSD. H 2 SO 4 and H 7 O? Digestion of A1 The microwave acid digestion procedure described above was then applied to A1 particles suspended in blank oil. The sample, however, was not completely digested with this procedure. Several parameters were varied to determine the optimum digestion procedure for Al. The sample size was reduced and the amount of HNO3 and H2O2 increased. These modifications did not fully digest the sample. It was determined that AI2O3 build-up on the surface of the particles hindered the digestions. Another acid, H2SO4, was then investigated for the digestion. The Al particles dissolved immediately upon addition of the sulfuric acid. The reaction, however, was quite exothermic. The outside wall of the PTFE sample vessel was hot to the touch. The vessel expanded so greatly that it would not fit into the digestion bomb. Reduction of the amount of H2SO4 used resulted in completely digesting the sample when subjected to the microwave heating program. Recovery for Al was 89% with a 17% RSD. HCl and Digestion of Cr Suspensions of Cr particles in blank oil were also incompletely digested using HNO3 and H2O2 as the digestants. Sulfuric acid was attempted in conjunction with hydrogen peroxide, also with incomplete digestion. However, Cr particles dissolved

PAGE 146

135 almost immediately with HCl and H 2 O 2 used as digestants. The sample was microwave digested and then cooled in a water bath, diluted and then analyzed by ICP-MS. Ten independent digestions were carried out. Again, the signals obtained from the digested samples were compared against an aqueous analytical curve. Calculations revealed a 108% Cr recovery with a 21% RSD. HF and H7O7 Digestion of Ti Finally, an attempt to microwave digest Ti particles in blank oil with HNO 3 , HCl, H2SO4, and aqua regia in conjunction with H2O2 was made. All of these mixtures gave incomplete digestions. The last resort was to use HF to digest a Ti suspension. Hydrofluoric acid was avoided initially due to its ability to etch glass. It has been known to readily etch glass nebulizers and quartz torches. It appeared, however, that no other acid would digest the Ti particles. In fact, the literature contains many articles that consistently use HF for the dissolution of Ti (43, 134). When HF and H 2 O 2 were added to a Ti particle/oil suspension and microwave acid digested, the particles were completely digested and the resulting solution was clear white. Average percent recoveries for 10 Ti digestions was 109% with an 16% RSD. At this point there were 4 different microwave acid digestion procedures used for the determination of wear metals in lubricating oils. They are summarized in Table 21. It was the goal of this research to develop a simple procedure for the multielemental determination of wear metals in lubricating fluids. The use of 4 digestion procedures per sample to analyze 8 metals is time-consuming and laborious. It would be advantageous to

PAGE 147

136 Table 21. Summary of the Acids Used for the Microwave Digestion of Oil Suspensions. Metal Acid Used for Digestion Mg HNO3 Fe HNO3 Y HNO3 Ni HNO3 Mo HNO3 A1 H2SO4 Cr HCl Ti HF

PAGE 148

137 have just 1 method for the determination of all metals. It was evident that more than 1 procedure would have to be used. HF and H 7 O 7 Digestion of Cr Microwave digestion of a chromium suspension was attempted with HF and H 2 O 2 . After the cooling period, the bomb was opened and the sample was found to be incompletely digested. As discussed above, Cr particles would not digest in either HNO3 or H2SO4. Therefore the digestion of Cr particles in oil had to be carried out with HCl and H 2 O 2 . HF and H 2 O 2 Digestion of A1 The digestion of A1 suspensions with HF and H 2 O 2 was also investigated. Aluminum suspensions were found to be completely digested in these two digestants. The average of 10 digestions gave a percent recovery of 99% with a 30% RSD. Further analyses of A1 in lubricating oils were carried out by digesting the sample with HF and H2O2. Table 22 summarizes the results of the microwave acid digestion procedures for the determination of Fe, Mg, Y, Ni, Mo, Cr, Al, and Ti. All data is the average of 10 independent digestions. Mass spectra for digested oil suspensions of the Mg and Fe, and Ni and Y are shown in Figure 39 and Figure 40. Figures 41 and Figure 42 show the mass spectra of Mo and Cr, and Al and Ti digested oil suspensions.

PAGE 149

138 Table 22. Recoveries of Digested Wear Metals in Oil Matrices. Analyte Ave* % Recovery Relative Uncertainty Fe 99% 14% Mg 98% 24% Y 117% 20% Ni 118% 11% Mo 120% 11% Cr 107% 21% A1 99% 29% Ti 109% 16% Â’Average of 10 digestions Fe, Mg, Y, Ni, and Mo digestions were carried out with HNO3 and H 2 O 2 Cr digestions were carried out with HCl and H 2 O 2 A1 and Ti digestions were carried out with HF and H 2 O 2

PAGE 150

1x10? 139

PAGE 151

1.4x10? 140 in s o 9i in N u5 ^ o n! in » -I 1 I L & & & ^ 8 6 T^ eo (sdo) |eu6js T“ 8 (6 'Bb c VO QQ 'eb c 00 t/i c ,2 ‘S c 4> o. CO 3 C/2 73 i T3 V 4-* CO V 00 5 •3 9i % o <41 o B 2 o. c/2 CO s O u ^3) £

PAGE 152

141 O (sdo) leuSfs Figure 41. Mass Spectrum of Microwave Acid Digested Particle Suspensions. A) 120 ng/mL Mo B) 1 12 ng/mL Cr

PAGE 153

iOlXOC 142 Figure 42. Mass Spectrum of Microwave Acid Digested Particle Suspensions. A) 136 ng/mL A1 B) 1 15 ng/mL Ti

PAGE 154

143 It should be noted that weighing the metal particles was not a simple process. The metal particles would many times develop a charge and adhere to the PTFE vessel. This would not cause a problem if they were inside the vessel. However, many times they would coat the outside of the sample vessel. It was impossible to remove them without acid treatment. In addition, the particles would also adhere to the balance pan. It was discovered that moving the balance away from direct air currents greatly reduced the errors associated with weighing. Acid Effects A study was carried out to determine if acid remaining in the sample after the microwave digestion procedure was completed influenced metal recoveries. This study was similar to the HNO 3 study described in Chapter 4. As shown in Figure 21, the addition of HNO 3 did not significantly effect the signal from individual 200 ppb aqueous solutions of Fe, Mg, Ni, Y, and Mo. The effect of HF concentration on A1 and Ti, and HCl concentration on Cr was determined by keeping the concentration of analyte constant and varying the acid concentration. The results are shown in Figure 43 and Figure 44 for HF and HCl, respectively. The zero (0) on the abscissa in Figures 43 and 44 represents no acid added. The range of acid concentration was not arbitrarily chosen. Digested samples were titrated with NaOH to determine the amount of acid remaining after the microwave digestion procedure was completed. Table 23 shows the average amount of acid in the digested particle suspensions was <2.1 x 10 M. The concentrations given in Table 23 are the average of 3 independent microwave acid digestions and titrations. In addition.

PAGE 155

Signal (cps) 144 4.0x107 3,5x107 3.0x107 2.5x107 2.0x107 1.5x107 1.0x107 5.0x106 0.0 I 0 1 1 1 1 ' 1 ' r -4 -3 -2 LOG [HF Concentration (M)] Al • Ti Figure 43. Effect of HF Concentration on Analyte Signal. Arrow Indicates No Acid Added.

PAGE 156

Signal (cps) 145 1.2x1071.0x1078.0x1066.0x1064.0x1062.0x1060 . 0 I I Cr I 1 1 1 1 r -5 -4 -3 -2 -1 LOG [ HF Concentration (M)] Figure 44 . Effect of HCl Concentration on Cr Signal. Arrow Indicates No Acid Added.

PAGE 157

146 Table 23. Summary of Titrations of Microwave Acid Digested Samples with NaOH. Analyte Acid Added Acid Concentration After Digestion and Dilution (M) Mo HNO3 1.0 X 10*^ Ni HNO3 2.1 X 10'^ Y HNO3 1.0 X lO'* Fe HNO3 undetectable Mg HNO3 undetectable Cr HCl 1.4 X 10'^ A1 HF 2.0 X 10'^ Ti HF 3.2 X 10Â’^

PAGE 158

147 the values represent the diluted samples, those that were directly aspirated into the ICPMS. The effect of HNO3, HF, and HCl on the respective analyte signal is negligible. Therefore, the digested oil suspensions were compared to the appropriate aqueous signals with no attempt to matrix match the aqueous standards. Error Analysis As discussed in Chapters 4 and 5, all measurements contained random error. The microwave acid digestion of metal particles suspended in lubricating oils introduced random errors not present in the digestion of commercial standards. The oil suspension standards were prepared by adding very small diameter particles to blank oil. The flow chart in Figure 45 illustrates the errors associated with preparing and digesting these types of samples. An example calculation of the corresponding random error is shown in Equation 17. ro.oi4Y ro.osy rJ I 20.00J V 25 J Equation 17 The first dilution gave a concentration of 800 + 1 pg/mL. This solution was further diluted prior to ICP-MS analysis. The random error associated with the final dilution is shown in Equation 18. The concentration of the final solution was 320 + 4 ng/mL, or 1.3% error, which is larger than the random error calculated for the preparation of aqueous and metallo-organic samples.

PAGE 159

148 i Sample Transfer Negligible Error i Random Error i First Dilution to 25 + 0.03 mL 800 + 1 |ig/mL i Pipetting Standard Solution 20 ± 0.25 pL i Random Error in Final Dilution to 50 + 0.05 mL 320 + 4 ng/mL Figure 45. Flow Chart Illustrating Errors Associated with Sample Preparation and Digestion.

PAGE 160

149 Equation 18 Table 24 summarizes the metal particle recovery results along with the random error calculations based from the standard deviations. The theoretical random error is based on calculations associated with sample preparation only. As discussed in Chapter 4, the measurement error is not included. Table 25 presents the metal particle recoveries along with the confidence interval. Calculations were carried out according to Equation 16, with a confidence limit of 95% for 10 observations. Once again, the theoretical calculations do not include measurement errors. Most of the experimental recoveries fall within the theoretical recovery, 100%. The Q-test was performed on the raw data to determine if any data points could be considered as outliers and not be included in the calculations. The Finnigan MAT SOLA sometimes gave erratic signals. The Q-test, at the 90% confidence limit, was applied to the data. It was determined that several signals could be discarded, however, the final calculations were not significantly affected. Review of Table 22 shows 5 of the 8 metals studied gave recoveries that were within 10% of the true values. Three metals, Ni, Mo, and Y gave high recoveries. Yttrium gave an average recovery that fell within the confidence limit. All three ICP-MS interface cones are constructed of nickel. It is not uncommon for erosion of the cones to produce high mass spectrum signals for nickel, which is a true systematic error.

PAGE 161

150 Table 24. Summary of Metal Recoveries with the Theoretical and Experimental Random Errors. Metal Experimental Recovery* Experimental Random Error^ Theoretical Random Error^ A1 99% 29 1.3 Cr 107 % 21 0.6 Fe 99% 14 0.6 Mg 98% 24 0.8 Mo 120% 11 0.4 Ni 118% 11 0.5 Ti 109 % 16 0.9 Y 117% 20 1.1 * Average of 10 independent microwave acid digestions ^ Based on standard deviation ^ Based on sample preparation only

PAGE 162

151 Table 25. Summary of Experimental and Theoretical Metal Recoveries and Confidence Intervals. Metal Experimental Recovery and Confidence Interval*^ Theoretical Recovery and Confidence Interval^ A1 99 ±29 100 ± 0.96 Cr 107 ±21 100 ±0.57 Fe 99± 11 100 ±0.43 Mg 98± 17 100 ±0.77 Mo 120 ± 8 100 ± 0.40 Ni 118± 11 100 ± 0.46 Ti 110± 16 100 ± 0.70 Y 117±20 100± 1.1 * Average of 10 independent microwave acid digestions ^ Calculated at 95% confidence limit * Based on sample preparation only

PAGE 163

CHAPTER 7 DETERMINATION OF WEAR METALS IN USED LUBRICATING OILS Introduction A microwave digestion inductively coupled plasma mass spectrometric method for the determination of metallo-organic and particulate wear metals in lubricating oils was described in Chapters 5 and 6. Commercial standards and samples prepared in the laboratory were used to evaluate the validity of the method. The goal of this research was to develop a technique for the determination of all forms of wear metals at low levels in used lubricating fluids. The following describes the analysis of used lubricating oils obtained from David Mitchell in the Material Science Department at the University of Florida, Gainesville, FL. The Rolling Contact Fatigue experiments were performed by him. Instrumentation The ICP-MS and microwave acid digestion instrumentation for the determination of wear metals in used lubricating oils are identical to those used for the analysis of metallo-organic and particulate oil samples. These parameters were discussed fiilly in Chapters 5 and 6. 152

PAGE 164

153 A NTN-Bower Rolling Contact Fatigue (RCF) tester (Ann Arbor, MI) with a three-ball-on-rod configuration was used for simulated wear of the lubricating oils. This type of instrument is widely accepted as a cost and time effective means of studying the wear of bearings (136). A schematic of the RCF tester is shown in Figure 46. A cylindrical test specimen is rotated around three ball specimens. The balls are separated by a retainer and are loaded against the rod specimen. Oil can be flowed through the RCF tester, between the balls and rod, while a load is applied to the balls. The RCF tester had found much use as a means of investigating the wear patterns of balls and rods at various applied loads. Some types of RCF tester allow temperature to be varied as well as pressure, thereby simulating the wear in numerous types of environments. Experimental Parameters Materials All materials used for the microwave acid digestion of used lubricating oils were described in Chapters 4 and 5. Balls and rods constructed of M50 VIM VAR steel were obtained fi'om MRC Specialty Balls, Winsted, CT. Table 26 shows the composition of M50 VIM VAR. The rods had a diameter of 9.5 mm (3/8 in) and the balls had a diameter of 12.7 mm (1/2 in). Toshiba TSN43 silicon nitride balls were used for the experiments. They were supplied to MRC Specialty Balls, and were purchased fiÂ’om them. The Si 3 N 4 balls and rods were hot isostatically pressed with Y2O3 and AI2O3 added as sintering aides.

PAGE 165

154 Rod Specimen Bearing ^ Cups Lubricant Load \ t Load Balls Figure 46. Three-Ball-on-Rod RCF Tester

PAGE 166

155 Table 26. Composition of M50 VIM VAR steel. Element Composition Fe Majority C 0.85 % Mn 0.35 % Si 0.40 % Cr 4.25 % V 1.00% Mo 4.25 %

PAGE 167

156 Exxon synthetic oil ETO 2380 (MIL-L-23699) jet engine oil (Exxon Chemicals, Houston, TX) was purchased from the Gainesville Regional Airport, Gainesville, FL. The ASTM standard for lubrication contamination for ball bearings is Arizona Test Dust (ATD), which was supplied by Powder Technology Inc., Burnsville, MN. The composition of ATD is given Table 27. The oil used for the RCF tester experiment was prepared by adding 1 g of ATD to 4 L oil. Aluminum oxide was fabricated by Dr. Jim AdairÂ’s research group in the Material Science Department at the University of Florida. Collection of Oil Samples Three independent experiments were carried out using the RCF tester. The first experiment used an M50 steel rod and M50 steel balls as the test specimens and the other two experiments used M50 rod and SisN 4 balls. Arizona Test Dust was added to unused jet engine lubricating oil for the systems of M50-M50 and M50-Si3N4. Aluminum oxide was also added to unused lubricating oil for the M50-Si3N4 system. The purpose of adding ATD and AI 2 O 3 was to facilitate wear in a timely manner. Stresses of 5.19 GPa and 5.97 GPa were applied to the M50-M50 systems and M50-Si3N4 systems, respectively. The rods in both cases were rotated at 3600 rpm. In the case of M50 steel balls rotating along an M50 steel rod, the lubricating oil doped with ATD flowed through the RCF tester and was collected after 440 hr had elapsed. A fresh, ATD doped, unused oil sample then flowed through a system of Si 3 N 4 balls and an M50 rod. Oil was sampled after 835 hr. An unused oil sample containing ATD was also obtained.

PAGE 168

Table 27. Composition of Arizona Test Dust. Chemical Weight % SiOj 65-76 AI2O3 11 17 Fe 203 2.5 -5.0 Na20 2-4 CaO 3-6 MgO p 1 Ti02 0 1 0 V2O3 0.10 ZrO 0.10 BaO 0.10

PAGE 169

158 Microwave Acid Digestion All oil samples were placed in a mechanical mixer prior to microwave acid digestion. As described in Chapter 6, three microwave acid digestions were carried out to determine all elements under investigation: HNO3/H2O2 for Mg, Fe, Mo, Ni, and Y, HCI/H2O2 for A1 and Ti; and HF/H2O2 for Cr. Two blank oil samples and all used oils were subjected in triplicate to the microwave acid digestion procedure described in Chapters 5 and 6. A total of 63 microwave acid digestions were carried out and the resulting aqueous solutions analyzed by ICP-MS. Data Analysis Analytical curves were constructed from aqueous standards for those elements showing a change in signal between the sample and blank mass spectra. A change in signal indicated wear. The ATD doped lubricating oil samples for both M50-M50 and M50-Si3N4 systems exhibited only two metals that gave a change in concentration detectable by ICP-MS: Fe and Ti. The concentrations of Fe and Ti in the original used ATD oil samples are given in Table 28. The M50-M50 bearings exhibited a greater amount of iron wear than the M50Si 3 N 4 bearings, even after running half the time. This was expected since the M50 is constructed primarily of iron. The advantage of using Si 3 N 4 balls on M50 raceways is clearly evident. The existence of wear metal particles in the lubricating oil can lead to further wear and fatigue of the engine components. Thus, the use of Si 3 N 4 ball bearings lead to improved engine conditions due to less wear.

PAGE 170

159 Table 28. Comparison of M50-M50 and M50-S13N4 Systems with ATD on RCF Tester. Fe Ti Mg Ni Y Mo A1 Cr M50-M50 440 hr 4 ± 2 pg/g 4 + 2 pg/g NDC NDC

PAGE 171

160 Titanium exhibited twice the wear in twice the time when the M50-Si3N4 bearings were compared to the M50-M50 bearings. This leads to the conclusion that replacing M50 balls with Si 3 N 4 does not lead to a significant change in operating conditions for Ti. The two systems produced the same amount of Ti wear. The other metals studied, Al, Cr, Mg, Mo, Ni, and Y, showed no change in signal when the used oil was compared to the fresh oil. Mass spectra of Al, Cr, Mg, and Mo; and Ni, Y, Fe, and Ti for the M50-M50 ATD system are shown in Figures 47 and 48, respectively. Figures 49 and 50 show the corresponding mass spectra for the M50-Si3N4 ATD system. With the exception of Mo and Y, distinct peaks were seen in the mass spectra of the oils, however no concentration change was observed between the blank and sample oils. The third system studied consisted of an M50 rod and Si 3 N 4 balls with AI 2 O 3 added to the lubricating oil. The results are shown in Table 29. A change in mass spectrum signal was observed for Fe, Mg, Ni, Y, and Mo. As expected, the level of these metals in the used lubricating oil increased with time spent on the RCF tester, which signified increased wear. The used lubricating oil sample did not exhibit Cr or Al mass spectra signals that were significantly different from the unused, AI2O3 doped oil. Mass spectra of Fe, Mg, Ni, and Y; and Mo, Cr, Al, and Ti are shown in Figure 51 and Figure 52, respectively. A fourth study of M50-Si3N4 bearings with AI2O3 doped lubricating oil was unavailable for analysis. The effects of ATD and AI 2 O 3 can be compared based on the wear they induced on the M50-Si3N4 bearings systems. The ATD system was operated for over a total of 835 hr and the AI 2 O 3 system was operated for over 154 total hours. Therefore, doping

PAGE 172

7x1 0» 161 (sd3) |eu6is Figure 47. Mass spectra of Used Lubricating Oil from M50-M50 ATD System. A) A1 B) Cr C) Mg D) Mo

PAGE 173

7x10* 162 I — i — S 2 S i A A 3 (sdo) iBufts Figure 48. Mass Spectra of Used Lubricating Oil from M50-M50 ATD System. A) Ni B) Y C) Fe D) Ti

PAGE 174

7x10* 163 Figure 49. Mass Spectra of Used Lubricating Oil from M50-Si3N4 ATD System. A) A1 B) Cr C) Mg D) Mo

PAGE 175

1 ^ 10 * 164 % & & & & 3 i 4 i i 4 ^ m m ^ (sd9) leuSis Figure 50. Mass Spectra of Used Lubricating Oil from M50-Si3N4ATD System. A)Ni B) Y C) Fe D)Ti

PAGE 176

165 0) H Uh C o J= Z o \r\ s C^x O I' E E 3 C/3 0\ 00 00 00 'Si) 'Si) 'ob 3. 3. 3. — ts 00 o o o o o o +1 +1 +1 O CO 'T o o o o o o' Z 00 00 'ob 'ob 00 =S=s.'ob ^ 00 3. o d — +1 +1 +1 0\ >T) \0 o d C30 s S Q e 1 00 H § § § 4) b 3. 3. 3. VO CO VO CS — ' +1 +1 +1 VO £ Tf uo After 40 hr After 74 hr After 307 hr

PAGE 177

3Qk10' 166 Figure 51. Mass Spectra of Used Lubricating Oil from M50-Si3N4Al2O3 System. A) Fe B)MgC)Ni D)Y

PAGE 178

1 0x104 167 Figure 52. Mass Spectra of Used Lubricating Oil from M50-Si3N4Al2O3 System. A) Mo B) Cr C) A1 D)Ti

PAGE 179

168 lubricating oil with aluminum oxide resulted in increased wear in less time when compared to Arizona Test Dust. This was true for Fe, Mg, Ni, Y, and Mo. Titanium exhibited the same amount of wear with AI 2 O 3 and ATD. Chromium and aluminum exhibited no wear with both dopants over the time specified. An increase in aluminum concentration might have occurred; however, no change in mass spectrum signal was observed. The addition of a large concentration of A1 hindered the determination of small concentration changes. ATD is primarily silicon. Therefore, A1 seems to result in more severe degradation in a shorter amount of time than Si.

PAGE 180

CHAPTERS CONCLUSIONS AND FUTURE WORK Conclusions Inductively Coupled Plasma Mass Spectrometry (ICP-MS) is quickly becoming the technique of choice when multielemental determinations are required. ICP-MS is able to detect low levels of many elements very quickly. The analysis of aqueous solutions with a low total dissolved solid content has become almost trivial. Much attention is now focused on the analysis of more complex matrices such as organic solutions and solids. As discussed in Chapter 2, organic samples require modifications either to the sample preparation procedures or the instrumental set-up before analyses can be carried out. Decomposition of an organic matrix can be used to convert metallo-organic species into water soluble salts. Traditionally, these types of acid digestions were carried out in an open vessel that required hours or days to complete. The advent of closed vessel digestion instrumentation has decreased reaction times to a few minutes. Coupling microwave acid digestion to ICP-MS provides an accurate and precise method of determining trace metals in organic solvents. The microwave acid digestion step enables particulate wear metals to be accurately determined. Otherwise, large particles would either clog the concentric nebulizer or fall out of the ICP before being vaporized, atomized and ionized. The use of ICP-MS as a detection method allows metals 169

PAGE 181

170 at low concentrations to be detected. The is advantageous for the implementation of preventative maintenance programs by the United States Air Force. Current methods of wear metal detection. Atomic Absorption Spectrometry and Atomic Emission Spectrometry, are characterized by detection limits in the ppb region. A technique that can detect wear metals at lower levels would allow for earlier wear metal detection, and thus indicate fatigue before failure could progress any further. Depending on the nature of the wear, this could be a life saving technique. Methods have been developed to determine metallo-organic and particulate wear metals in lubricating oils. The accuracy of the technique for metallo-organic samples was verified by analysis of NIST Standard Reference Material 1084a and 1085a. Oil suspension standards are not commercially available. Therefore, standards were prepared in the laboratory. Chapter 6 discussed the many problems encountered while preparing these types of standards. Three microwave acid digestion procedures were developed for the determination of metallo-organic and particulate wear metals in lubricating oils. Nitric acid was used for the digestion of Mg, Fe, Ni, Mo, and Y, hydrofluoric acid was used for the digestion of A1 and Ti, and hydrochloric acid as used for the digestion of Cr. All three procedures used hydrogen peroxide to aid in the oxidation of the organic matter. The accuracy of the digested metal particle suspensions was determined by calculating recoveries for the specific metals. Recoveries fell within the limits set by the 95% confidence intervals.

PAGE 182

171 Aqueous and metallo-organic analytical curves were plotted on the same graph as the metal particle data, shown in Figures 53 and 54. It can be seen that slopes of these three graphs are similar to one another, another indication of complete digestion of the oil samples. Used lubricating oil samples were also analyzed my Inductively Coupled Plasma Mass Spectrometry preceded by microwave acid digestion. Two bearing systems were evaluated: M50 steel balls on an M50 steel rod, and Si 3 N 4 balls on an M50 steel rod. Both systems were studied with lubricating oils doped with Arizona Test Dust, while the M50-Si3N4 system was also studied with AI 2 O 3 doped lubricating oil. The replacement of M50 steel balls with one constructed out of Si 3 N 4 led to a decrease in Fe concentration. The other metals studied, Al, Cr, Mg, Mo, Ni, Ti, and Y, showed no change in concentration between the two bearing systems. Thus, improved operating conditions can be reached with Si 3 N 4 ball bearings. This corresponds to longer engine lifetimes. The lubricating oil degradation effects of ATD and AI2O3 were also compared. Aluminum oxide produced a larger amount of wear in less time than the ATD, which was mostly comprised of silicon. Therefore, Al results in more wear than Si. Future Work More reproducible microwave heating could be applied to oil samples by replacing the rotating glass plate that was destroyed in earlier experiments. The microwave field inside the sample cavity is not homogenous. The microwave acid digestions described in

PAGE 183

172 (•da) ituB® (•da) lauBis Figure 53. Aqueous, Digested Metallo-Organic, and Digested Particulate Suspension Analytical Curves. A) A1 B) Mg C) Fe D)Y

PAGE 184

173 o (sds) |eu6is (9do) leuOis c S Figure 54. Aqueous, Digested Metallo-Organic, and Digested Particulate Suspension Analytical Curves. A) Cr B) Ni C) Ti D) Mo

PAGE 185

174 this dissertation were carried out by placing the microwave bomb in the same comer of the cavity each time. However, this does not allow each sample to experience even heating throughout the microwave field. Placing the bomb on a rotating place inside the cavity would allow for more reproducible digestions due to each sample being exposed to similar microwave fields. It would also be advantageous to reduce the three microwave acid digestions procedures to one procedure. The multielemental capability of ICP-MS detection was not fully exploited due to the three time-consuming digestions that were performed on each sample. Microwave digestion with an acid mixture should be investigated. A mixture of HF will be needed for the digestion of Ti. The erratic nature of the SOLA could be circumvented by using an internal standard. As mentioned in Chapter 6, the SOLA would sometimes give signals that were not representative of the concentration of the sample. Five measurement were averaged per sample, and a Q-test was applied to the suspect signals. However, internal standardization has been shown to give more precise data when using ICP-MS as a detection method. Better detection limits can be obtained by desolvating the sample before it is introduced into the plasma. The removal of water should be attempted by repetitive heating and cryogenic cooling of the sample aerosol as described by Alves et al. (61). Removal of water would help to lower the detection capability for Fe. Iron is interfered by ArO* at mass 56. Reduction of water will greatly decrease the abundancy of ArO* in the plasma, and thus improve the detection limit of Â’^e.

PAGE 186

175 The analysis of used lubricating oils should be investigated further. Verification of the results could be accomplished by Atomic Emission Spectrometry. Additional used lubricating oil samples from the M 50 -M 50 and M 50 -Si 3 N 4 bearing systems with longer RCF run times should be obtained and analyzed by microwave digestion ICP-MS. It will be interesting to note how many hours elapse before wear metals other than iron are evident. Used lubricating oil samples fi-om a Si3N4-Si3N4 bearing system should also be collected and analyzed by the above method. It is expected that less wear metals will be detected with Si3N4-Si3N4 bearing systems than with the two previous bearing systems. It would be interesting to apply microwave acid digestion followed by ICP-MS to the analysis of used lubricating oils from jet engines. This would illustrate the applicability of the technique. The direct impact of detecting metallo-organic and particulate wear metals in lubricating fluids at concentrations lower than current methods (AES or AAS) would be then realized.

PAGE 187

REFERENCES (1) Mori, T.; Nishida, K.; Inoue, K.; Yasui, H.; Takebayashi, H.; Tanimoto, K, Proceedings of the Third International Symposium Cermaic Materials and Components for Engines, Las Vegas, Nevada, 1988, p. 1459. (2) Lucek, J W. Proceedings of the Gas Turbine and Aeroengine Congress and Exposition, Brussels, Belgium, 1990, p. 1. (3) Galbato, A.T.; Cundill, R.T.; Harris, T.A. Proceedings of STLE, Philadelphia, Pennsylvania, 1992, p. 1. (4) Takebayashi, H.; Johns, T.M.; Rokkaku, K.; Tanimoto, K. SAE Technical Paper Series 1990, 901629. (5) De la Guardia, M. At. Spectrosc. 1984, 5(4), 150. (6) Brown, R.J. Spectrochim. Acta 1983, 38, 283. (7) Hogtahl, O.T.; Melsom, S. Anal. Chem. 1966, 38, 1414. (8) Burrows, J.A.; Heerdt, J.C.; Willis, J.B. Anal. Chem. 1965, 37, 579. (9) Miller, R.L.; Fraser, L.M.; Winefordner, J.D. Appl. Spectrosc. 1971, 25(4), 477. (10) Johnson, D.J.; Plankey, F.W.; Winefordner, J.D. Can. J. Spectrosc. 1974, 19, 151. (11) Wynn, T.F.; Clardy, P.; Vaughn, L., Bradshaw, J.D.; Bower, J.N.; Epstein, M S.; Winefomder, J.D. Anal. Chim. Acta 1981, 124, 155. (12) Maddox, W.E.; Kelliher, W.C. Adv. X-ray Anal. 1986, 29, 497. (13) Zararsiz, A.; Kirmaz, R., Arikan, P. Nucl. Instr. andMeth. in Phys. Res B 1996 108, 385. (14) Al-Gadi, A. A.; Al-Swaidan, H.M. Anal. Lett. 1990, 23, 1757. (15) King, A.D.; Hilligoss, D.R.; Wallace, G.F. At. Spectrosc. 1984, 5(4), 189. 176

PAGE 188

177 (16) Brodie, K.G.; Matousek, J.P. Ami Chem. 1971, 43, 1557. (17) Gonzalez, M.C., Rodriguez, A.R.; Gonzalez, V. Microchem. J. 1982, 27, 188. (18) Merryfield, R.N.; Loyd, R.C. Anal. Chem. 1979, 51, 1965. (19) Granchi, M.P., BiggerstafF, J.A.; Hilliard, L.J.; Grey, P. Spectrochim. Acta 1987, 42B, 169. (20) Fassel, V.A.; Peterson, C.A.; Abercrombie, F.N.; Kniseley, R.N. Anal. Chem. 1976, 48, 516. (21) Anwar, J.; Nahid, F.; Kirmani, Z.U. Jour. Chem. Soc. ofPak. 1990, 12, 249. (22) Udoh, A.P. Talanta 1995, 42, 1827. (23) Runnels, J.H.; Merryfield, R., Fisher, H.B. Anal. Chem. 1975, 47, 1258. (24) Chuang, F.S.; Winefordner, J.D. Appl. Spectros. 1974, 28,215. (25) Reeves, R.D., Molnar, C.J.; Glenn, M.T.; Ahlstrom, J.R.; Winefordner, J.D. Anal. Chem. 1972, 44, 2205. (26) Chuang, F.S.; Patel, B.M.; Reeves, R.D.; Glenn, M.T.; Winefordner, J.D. Can. J. Spectrosc. 1973, 18, 6. (27) Bratzel, Jr., M B.; Chakrabarti, C.L. Anal. Chim. Acta 1972, 61, 25. (28) Knauer, H.E.; Milliman, G.E. Anal. Chem. 1975, 47, 1263. (29) Saba, C.S.; Centers, P.W. Lubr. Eng. 1995, 51(9), 111. (30) Jantzen, E.; Buck, V.; Kaegler, S.H. Wear 1983, 87, 331. (31) Kauffinan, R.E.; Saba, C.S.; Rhine, W.E.; Eisentraut, K.J. Anal. Chem. 1985, 28, 400. (32) Saba, C.S.; Rhine, W.E.; Eisentraut, K.J. Appl. Spectrosc. 1985, 39(4), 689. (33) Brown, J.R.; Saba, C.S., Rhine, W.E.; Eisentraut, K.J. Anal. Chem. 1980, 52, 2365. (34) Fry, D.L. Appl. Spectrosc. 1956, 10, 65.

PAGE 189

178 (35) Lukas, M. Trans ASME 1993, 115, 620. (36) Botto, R.I. J. Anal. At. Spectrom. 1993, 5, 51. (37) Anderau, C.; Fredeen, K.J.; Thomsen, M.; Yates, D.A. At. Spectrosc. 1995, 16, 79. (38) Gambrill, C.M.; Gassmann, A.G., OÂ’Neill, W.R. Anal. Chem. 1951, 23, 1365. (39) Bangroo, P.N.; Jagga, C.R.; Arora, H.C.; Rao, G.N. At. Spectrosc. 1995, 16(3), 118 . (40) Brown, R.J. Spectrochim. Acta 1983, 38B, 283. (41) Fischer, J.L.; Rademeyer, C.J. J. Anal. At. Spectrom. 1994, 9, 623. (42) Algeo, J.D.; Heine, D.R.; Phillips, H.A.; Hoek, F.B.G.; Schneider, M R.; Freelin, J.M., Denton, M B. Spectrochim. Acta 1985, 40B, 1447. (43) Kauffman, R.E.; Saba, C.S.; Rhine, W.E.; Eisentraut, K.J. Anal. Chem. 1982, 54, 975. (44) Hutton, R.C. J. Artal. At. Spectrom. 1986, 1, 259. (45) Lowe, D.S.; Stahl, R.G. Anal. Proc. 1992, 29, 111. (46) Hausler, F. Spectrochim. Acta 1987, 42B, 63. (47) Paschal, D.C.; Caldwell, K.L.; Ting, B.G. J. Anal. At. Spectrom. 1995, 10, 367. (48) Richner, P., Wunderli, S. J. Anal. At. Spectrom. 1993, 8, 45. (49) Karanassios, V.; Drouin, D.; Reynolds, G.G. Spectrochim. Acta 1995, 50B, 415. (50) Cairns, W.R.L.; Ebdon, L.; Hill, S.J. Fresenius' J. Anal. Chem. 1996, 255, 202. (51) Larsen, E.H.; Stiirup, S. J. Anal. At. Spectrom. 1994, 9, 1099. (52) Hintelmann, H.; Evans, R.D.; Villeneuve, J.Y. J. Anal. At. Spectrom. 1995, 10, 619. (53) Hill, S.J. Anal. Proc. 1992, 29, 399. (54) Inoue, Y., Kawabata, K.; Suzuki, Y. J. Anal. At. Spectrom. 1995, 10,363.

PAGE 190

179 (55) Shum, S.C.K.; Neddersen, R.; Houk, R.S. Analyst 1992, 117, 577, (56) Magnuson, M.L.; Creed, J.T.; Brockhoff, C.A. J. Anal. At. Spectrom. 1996, 11, 893. (57) Alonso, Sena, F.; Arbore, P.; Betti, M.; Koch, L. J. Anal. At. Spectrom. 1995, 10, 381. (58) Kinzer, J.A. Olesik, J.W., Olesik, S.V. Anal. Chem. 1996, 68, 3250. (59) Byrdy, F.A.; Caruso, J.A. Environ. Sci. Technol. 1994, 28, 528A. (60) Ghazi, A.M.; McCandless, T.E.; Vanko, D.A.; Ruiz, J. J. Anal. At. Spectrom. 1996, 11, 667. (61) Alves, L.C.; Minnich, M.G.; Wiederin, D.R.; Houk, R.S. J. Anal. At. Spectrom. 1994, 9, 399 (62) Backstrom, K.; Gustavsson, A. Spectrochim. Acta 1989, 44B, 1041. (63) Hitchen, P.; Hutton, R.; Tye, C. J. Autom. Chem. 1992, 14, 17. (64) Karanassios, V., Horlick, G. Spectrochim. Acta 1989, 44B, 1345. (65) Cairns, W.R.L.; McLeod, C.W., Hancock, B. Spectroscopy 1997, /2(4), 16. (66) Woller, A.; Garraud, H.; Martin, F.; Donard, O.F.X.; Fodor, P. J. Anal. At. Spectrom. 1997, 12, 53. (67) Shum, S.C.K.; Pang, H.; Houk, R.S. Anal. Chem. 1992, 64, 2444. (68) Wiederin, D.R.; Smith, F.G.; Houk, R.S. Anal. Chem. 1991, 63, 219. (69) Al-Swaidan, H.M. Talanta 1996, 43, 1313. (70) Olsen, S.D.; FUby, R.H.; Brekke, T.; Isaksen, G.H. Analyst 1995, 120, 1379. (71) Al-Swaidan, H.M. Anal. Lett. 1993, 26, 141. (72) Bettinelli, M., Spezia, S.; Baroni, U.; Bizzarri, G. J. Anal. At. Spectrom. 1995, 10, 555. (73) Escobar, M.P., Smith, B.W.; Winefordner, J.D. Anal. Chim. Acta 1996, 320, 11. (74) Osborne, S.P. Appl. Spectrosc. 1990, 44, 1044.

PAGE 191

180 (75) Reed, T.B. J. Appl. Phys. 1961, 32, 821. (76) Reed, T.B. J. Appl. Phys. 1961, 32, 2534. (77) Gray, A.L. Analyst 1975, 100, 289. (78) Houk, R.S.; Fassel, V.A.; Flesch, G.D.; Svec, H.J.; Gray, A.L.; Talyor, C.E. Anal. Chem. 1980, 52, 2283. (79) Gray, A.L.; Date, A.R. Dyn. Mass Spectrom. 1981, 6, 252. (80) Douglas, D.J.; French, J.B. Spectrochim. Acta 1986, 41, 197. (81) Fassel, V.A.; Kniseley, R.N. Anal. Chem. 1974, 46, 1155A. (82) Hieftje, G.M. J. Anal. At. Spectrom. 1992, 7, 783. (83) Olesik, J.W. Anal. Chem. 1996, 68, 469A. (84) Douglas, D.J. In Inductively Coupled Plasmas in Analytical Atomic Spectrometry Montaser, A.; Golightly, D.W. (Eds.), New York: VCH Publishers, Inc., 1992, 2nd ed., 614. (85) Houk, R.S. Anal. Chem. 1986, 58, 97 A. (86) Selby, M.; Hieftje, G.M. Am. Lab. 1987, 19, 16. (87) Vanhaecke, F.; Moens, L.; Dams, R.; Papadakis, I.; Taylor, P. Aruil. Chem. 1997, <59, 268. (88) Mahoney, P.P.; Ray, S.J.; Hieftje, G.M. Appl. Spectrosc. 1997, 51, 16A. (89) Koppenaal, D.W.; Barinaga, C.J.; Smith, M.R. J. Anal. At. Spectrom. 1994, 9, 1053. (90) Milgram, E.; Watson, C.; Eyler, J. Procedings of the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, Georgia, 1995, p. 1088. (91) Douglas, D.J.; Houk, R.S. Proc. Analyt. Atom. Spectrosc. 1985, 8, 1. (92) SOLA I. C.P./G.D. Mass Spectrometer Operators Manual, Finnigna MAT Ltd: San Jose, 1993.

PAGE 192

181 (93) Roboz, J. Introduction to Mass Spectrometry New York: Interscience Publishers, 1968 , p.l60. (94) Turner, P.J. In A pplications of Plasma Source Mass Spectrometry Holland, G.; Eaton, AN. (Eds), Cambridge; The Royal Society of Chemistry, 1991, p.71. (95) Hieftje, G.M.; Vickers, G.H. Anal. Chim. Acta 1989 , 216, 1. (96) Houk, R.S.; Shum, S.C.K.; Wiederin, D.R. Anal Chim. Acta 1991 , 250, 61. (97) Douglas, D.J. In Inductively Coupled Plasmas in Analytical Atomic Spectrometry Montaser, A.; Golightly, D.W. (Eds ), New York; VCH Publishers, Inc., 1992 , 2nd. ed. p. 629. (98) Jarvis, K.E.; Gray, A.L.; Houk, R.S. Handbook of Inductively Coupled Plasma Ma.s.s Spectrometry New York: Chapman and Hall, 1992, p. 26. (99) Gray, A.L. In A pplications of Inductively Coupled Plasma Mass Spectrometry Date, A.R.; Gray, A.L. (Eds.), New York: Chapman and Hall, 1989 , p. 17. (100) Douglas, J.B.; French, J.B. J. Anal At. Spectrosc. 1988 , 3, 743. (101) Moens, L.; De Smaele, T.; Dams, R.; Van Den Broeck, P.; Sandra, P. Anal Chem. 1997 , 69, 1604. (102) Vela, N.P.; Olson, L.K.; Caruso, J.A. Anal Chem. 1993 , 65, 585A. (103) Janghorbani, M.; Ting, B.T.G. Anal Chem. 1989 , 61, 701. (104) Watling, R.J.; Lynch, B.F.; Herring, D. J. Anal At Spectrosc. 1997 , 12, 195. (105) Ebdon, L.; Foulkes, M.; Sutton, K. J. Anal At Spectrosc. 1997, /2, 213. (106) Jiang, S.-J.; Houk, R.S. Spectrochim. Acta 1987 , 42, 93. (107) Gregoire, D C. Can. J. Anal Set Spectrosc. 1997 , 42, 1. (108) Moens, L.; Verrept, P., Boonen, S.; VanHaecke, F.; Dams, R. Spectrochim. Acta 1995 , 50, 463 . (109) Koropchak, J.A. Spectroscopy 1993 ,
PAGE 193

182 (111) Thomas, C.; Jakubowski, N.; Stuwer, D.; Broekaert, J.A.C. J. Anal. At. Spectrosc. 1995, 10, 583. (1 12) Christodoulou, J.; Kashani, M.; Keohane, B.M.; Sadler, P.J. J. Anal. At. Spectrosc. 1996, 11, 1031. (1 13) Suddendorf, R.F.; Boyer, K.W. Anal. Chem. 1978, 50, 1769. (114) Greenfield, S.; Montaser, A. In Inductively Coupled Plasmas in Analytical Atomic Spectrometry Montaser, A.; Greenfield, S. (Eds ), New York: VCH Publishers, Inc., 1992, 2nd ed., p. 224. (115) Scott, R.H.; Fassel, V.A.; Kniseley, R.N.; Nixon, D.E. Arml. Chem. 1974, 46, 75. (116) Browner, R.F.; Boom, A.W. Anal. Chem. 1984, 56, 875A. (1 17) Olesik, J.W.; Bates, L.C. Spectrochim. Acta 1995, 50, 285. (118) Sulcek, A. ; Povondra, P. Methods of Decomposition in Inorganic Analysis Boca Raton, FL: CRC Press, Inc., 1989. (1 19) Matusiewicz, H.; Sturgeon, R E. Prog. Analyt. Spectrosc. 1989, 12, 21. (120) Abu-Samra, A.; Morris, J.S.; Koirtyohann, S.R. Anal. Chem. 1975, 47, 1475. (121) Fernando, L.A.; Heavner, W.D.; Gabrielli, C.C. Anal. Chem. 1986, 58, 511. (122) Fischer, L.B. Anal. Chem. 1986, 58, 261. (123) Smith, F.; Cousins, B.; Bozic, J.; Flora, W. Anal. Chim. Acta 1985, 177, 243. (124) Lam, J.W.H.; McLaren, J.W.; Methven, B.A.J. J. Anal. At. Spectrosc. 1995, 10, 551. (125) Chakraborty, R.; Das, A.K.; Cervera, M.L.; De la Guardia, M. Anal. Lett. 1997, 20, 283. (126) Coedo, A.G.; Dorado, T.; Fernandez, B.J.; Alguacil, F.J. Anal. Chem. 1996, 68, 991. (127) Yoshinaga, J.; Morita, M. J. Anal. At. Spectrosc. 1997, 12, 417. (128) Nakashima, S.; Sturgeon, R.E.; Willie, S.N.; Berman, S.S. Analyst 1988, 112, 159.

PAGE 194

(129) Kingston, H.M.; Jassie, L.B. Introduction to Microwave Sample Preparation American Chemical Society: Washington, DC, 1988, Chapter 2. 183 (130) Neas, E.D., Collins, M.J. In Introduction to Microwave Sample Preparation Kingston, H.M.; Jassie, L.B. (Eds ), American Chemical Society: Washington, DC, 1988, Chapter 2. (131) Kingston, H.M.; Water, P.J. Spectroscopy 1992, 7(9), 20. (132) CRC Handbook of Chemistry and Physics Lide, D.R. (Ed ), 71st ed., CRC Press: Boca Raton, FL, 1990. (133) Ingle, Jr., J.D.; Crouch, S.R. Spectrochemical Analysis Prentice-Hall, Inc.; New Jersey, 1988, p. 10. (134) Tan, S.H.; Horlick, G. Appl. Specctrosc. 1986, 40, 446. (135) Saba, C.S.; Eisentraut, K.J. Anal. Chem. 1977, 49, 454. (136) Glover, D. Rolling Contact Fatigue Testing of Bearing Steels. ASTM STP 771 J.J.C. Hoo (Ed ), American Society for Testing and Materials, 1982.

PAGE 195

BIOGRAPHICAL SKETCH Robin Ann Russell was bom in Charlotte, North Carolina, on May 31, 1970. A few months later her family moved to Harrisburg, NC, where Robin resided for the next twenty-three years. Robin graduated from Northwest Cabarrus High School in Concord, NC, in 1988 and began her college experience at the University of North Carolina at Charlotte in early 1989. She continued her education at the University of Florida under the direction of Dr. James D. Winefordner in 1993. In August of 1997, Robin obtained her doctorate of philosophy in chemistry with a minor in environmental engineering sciences. She began her professional career at Exxon Chemical Company in Baytown, Texas as a chemist. 184

PAGE 196

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fiilly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. James D. Winefordne/ Chairman /^Graduate Research Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. David H. Powell Associate Scientist of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Robert T. Kennedy O Associate Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is firlly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. J. Eric Enholm Associate Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. J^ph J. Delffno Professor of Environmental Engineering Sciences

PAGE 197

This dissertation was submitted to the Graduate Faculty of the Department of Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1997 Dean, Graduate School