analysis of trace metals in lubricating oil using - Encore

DETERMINATION OF TRACE METALS IN LUBRICATING OILS BY
ICP-OES
by
HAILE ARAYA TEKIE
Submitted in partial fulfilment of the requirements for the degree
MAGISTER TECHNOLOGIAE: CHEMISTRY
in the
Department of Chemistry
FACULTY OF SCIENCE
TSHWANE UNIVERSITY OF TECHNOLOGY
Supervisor: Prof RI McCrindle
Co-supervisor: Prof PJJG Marais
Co-supervisor: Dr AA Ambushe
February 2013
ii
DECLARATION BY CANDIDATE
“I hereby declare that the dissertation submitted for the degree Magister
Technologiae: Analytical Chemistry, at Tshwane University of Technology, is my own
original work and has not previously been submitted to any other institution of higher
education. I further declare that all sources cited or quoted are indicated and
acknowledged by means of a comprehensive list of references”.
Haile Araya Tekie
Copyright © Tshwane University of Technology (2013)
iii
DEDICATION
This work is dedicated to my elder brother, Goitom, for his unrestricted support and
encouragement during the entire study.
iv
ACKNOWLEDGEMENTS
First of all I would like to express my thanks and glory to Almighty God, who provided
me with strength and inspiration throughout my life.
My profound gratitude and appreciation goes to: my supervisor, Prof RI McCrindle, for
making this study possible through his wealthy academic experience. He was always
there and patient with me any time throughout my study period, whenever I had
problems; my co-supervisor, Prof PJJG Marais, for his positive attitude and guidance;
my co-supervisor, Dr AA Ambushe, for his unrelenting assistance in organizing the
study and critical review of the dissertation; Tshwane University of Technology and
the National Research Foundation, for funding the project; The Department of
Chemistry for providing a rich atmosphere to work on my studies and the opportunity
of teaching assistantship.
My sincere thanks are also expressed to: my wife, Meaza Habtegiorghis, my son
Henok and daughter Delina, for their understanding, encouragement and patience
during my study; my parents, sisters and brothers, for their support and
encouragement. Among them are: Ato A Tesfagiorghis and family, Dr T Araya and
family, Dr A Ghebremeskel and family.
I would also like to express my sincere thanks to: Stanley Moyo, for teaching me the
ICP-OES, his positive attitude and team spirit; Mr P Kobue, statistical analyst, for his
kind assistance with data analysis using ANOVA; All the authors referenced
throughout my research work; The unique and diverse Chemistry postgraduate team,
for their friendly cooperation and open discussions in all aspects of life.
Last but not least, my thanks also go to administrative staff members of Tshwane
University of Technology for their kind support throughout my research work.
v
ABSTRACT
Quantification of trace elements in used lubricating oil forms a vital part in monitoring
engine conditions and impact on the environment. In this study, inductively coupled
plasma-optical emission spectrometry (ICP-OES) was used for the determination of
trace elements (Ag, Ba, Cu, Mn and Ni) in lubricating oils. The methodology was
developed intending to minimise the oil’s carbonaceous matter and viscosity.
Accordingly, six oil preparation techniques (xylene dilution, detergent emulsion,
microwave digestion, dry-ashing, wet-ashing and ultrasonic extraction) were
investigated. Optimisation of the factors influencing an ultrasonic extraction and ICPOES operating parameters enabled quantification of the trace metals in oils.
Limits of detection (3Sb/m), in the ng g-1 range, were obtained for each element of
interest using each method investigated. The validity of the methodologies studied
was checked through the analysis of a certified reference material in used oil
(EnviroMAT, HU-1) and Conostan S-21 standards. Good analyte recoveries, ranged
48.3 to 106%, were obtained. Evaluation of the analytical methods studied with
regard to accuracy, precision, LOD, linearity, applicability for routine analysis,
preparation time and cost was made. Based on these evaluations, ultrasonic
extraction has a clear advantage in terms of accuracy, applicability for routine
analysis, time and cost of sample preparation.
vi
CONTENTS
PAGE
DECLARATION -------------------------------------------------------------------------------------- ii
DEDICATION ------------------------------------------------------------------------------------------iii
AKNOWLEDGEMENTS ------------------------------------------------------------------------- iv
ABSTRACT--------------------------------------------------------------------------------------------- v
CONTENTS-------------------------------------------------------------------------------------------- vi
LIST OF FIGURES -------------------------------------------------------------------------------- xi
LIST OF TABLES --------------------------------------------------------------------------------- xiii
LIST OF ABREVIATIONS USED -----------------------------------------------------------xv
GLOSSARY ----------------------------------------------------------------------------------------- xvii
CHAPTER 1: INTRODUCTION --------------------------------------------------------------- 1
1.1
BACKGROUND AND MOTIVATION ----------------------------------------- 1
1.2
PROBLEM STATEMENT -------------------------------------------------------- 5
1.3
HYPOTHESIS ----------------------------------------------------------------------- 6
1.4
OBJECTIVES ----------------------------------------------------------------------- 6
1.4.1
General objective ----------------------------------------------------------------- 6
1.4.2
Specific objectives --------------------------------------------------------------- 7
CHAPTER 2: LITERATURE SURVEY----------------------------------------------------- 8
2.1
INTRODUCTION ------------------------------------------------------------------- 8
2.2
LUBRICATING OIL CHEMISTRY --------------------------------------------- 8
2.2.1
Background ------------------------------------------------------------------------- 8
2.2.2
Refining lubricating oil from crude petroleum-------------------------- 8
2.3
TYPES OF LUBRICATING OILS ---------------------------------------------- 9
vii
2.3.1
Mineral oils -------------------------------------------------------------------------- 9
2.3.1.1
Paraffinic oils----------------------------------------------------------------------- 10
2.3.1.2
Naphthenic oils -------------------------------------------------------------------- 10
2.3.1.3
Aromatic oils ----------------------------------------------------------------------- 10
2.3.2
Synthetic oils --------------------------------------------------------------------- 11
2.4
PROPERTIES OF LUBRICATING OIL ------------------------------------ 11
2.4.1
Viscosity --------------------------------------------------------------------------- 11
2.4.2
Viscosity index ------------------------------------------------------------------ 12
2.4.3
Total acid and base numbers ----------------------------------------------- 12
2.5
SOURCES OF TRACE ELEMENTS IN LUBRICATING OILS ------ 13
2.5.1
Additive elements--------------------------------------------------------------- 13
2.5.2
Wear metals ----------------------------------------------------------------------- 14
2.5.3
Lubricating oil contaminants ----------------------------------------------- 15
2.6
SEQUENTIAL SIMPLEX OPTIMISATION -------------------------------- 16
2.6.1
Introduction ----------------------------------------------------------------------- 16
2.6.2
The simplex calculations ----------------------------------------------------- 17
2.6.3
Rules of the simplex algorithm -------------------------------------------- 17
2.6.4
Boundary violation ------------------------------------------------------------- 18
2.6.5
The convergence criteria ----------------------------------------------------- 18
2.7
INDUCTIVELY COUPLED PLASMA-OPTICAL EMISSION
SPECTROMETRY --------------------------------------------------------------- 18
2.7.1
Introduction ----------------------------------------------------------------------- 18
2.7.2
Instrumentation for ICP-OES ----------------------------------------------- 19
2.7.3
Sample preparation for ICP-OES analysis ----------------------------- 20
2.7.3.1
Dilution of lubricating oil with xylene ---------------------------------------- 20
2.7.3.2
Emulsification ---------------------------------------------------------------------- 21
2.7.3.3
Dry-ashing -------------------------------------------------------------------------- 22
2.7.3.4
Wet-ashing ------------------------------------------------------------------------- 22
2.7.3.5
Microwave-assisted acid digestion------------------------------------------- 23
viii
2.7.3.6
Ultrasound-assisted extraction ------------------------------------------------ 24
2.7.4
Optimisation of ICP-OES operating parameters --------------------- 24
2.7.4.1
Introduction ------------------------------------------------------------------------- 24
2.7.4.2
The optimisation scheme ------------------------------------------------------- 25
2.8
ANALYTICAL LIMITS OF DETECTION ----------------------------------- 25
2.8.1
Limit of detection --------------------------------------------------------------- 26
2.8.2
Limit of quantitation ----------------------------------------------------------- 26
CHAPTER 3: EXPERIMENTAL ------------------------------------------------------------- 27
3.1
INTRODUCTION ----------------------------------------------------------------- 27
3.2
REAGENTS AND STANDARDS -------------------------------------------- 27
3.3
APPARATUS AND INSTRUMENTATION -------------------------------- 28
3.3.1
Apparatus for microwave digestion -------------------------------------- 28
3.3.2
Apparatus for ultrasonic-assisted extraction ------------------------- 28
3.3.3
Setup for ICP-OES instrumentation -------------------------------------- 28
3.4
SAMPLE PREPARATION FOR ICP-OES ANALYSIS ---------------- 29
3.4.1
Xylene dilution ------------------------------------------------------------------- 29
3.4.2
Oil emulsification --------------------------------------------------------------- 30
3.4.3
Dry-ashing ------------------------------------------------------------------------- 30
3.4.4
Wet-ashing ------------------------------------------------------------------------ 31
3.4.5
Microwave-assisted acid digestion --------------------------------------- 31
3.4.6
Ultrasonic-assisted extraction --------------------------------------------- 32
3.5
SELECTION OF ANALYTICAL WAVELENGTHS---------------------- 33
3.6
OPTIMISATION ------------------------------------------------------------------- 33
3.6.1
Simplex optimisation of ICP-OES operating parameters --------- 33
3.6.2
Factorial optimisation of an ultrasonic extraction ------------------ 34
CHAPTER 4: RESULTS AND DISCUSION ------------------------------------------- 36
4.1
INTRODUCTION ----------------------------------------------------------------- 36
ix
4.2
ELEMENTAL WAVELENGTH SELECTION ----------------------------- 36
4.3
OPTIMISATION ------------------------------------------------------------------- 38
4.3.1
Simplex optimisation of ICP-OES operating parameters --------- 38
4.3.2
Factorial optimisation of the ultrasound-assisted extraction for
trace metals in lubricating oils --------------------------------------------- 43
4.3.2.1
Influence of HNO3--------------------------------------------------------------------------------------------------43
4.3.2.2
Influence of aqua regia ---------------------------------------------------------- 45
4.3.2.3
Influence of HNO3:H2O2 (2:1) ------------------------------------------------- 48
4.3.2.4
Influence of HNO3:HCl (1:1) --------------------------------------------------- 50
4.3.2.5
Influence of sonication time ---------------------------------------------------- 51
4.4
ANALYSIS OF UED LUBRICATING OILS BY ICP-OES ------------- 52
4.4.1
Introduction ----------------------------------------------------------------------- 52
4.4.2
Linearity of calibration curves --------------------------------------------- 52
4.4.3
Determination of limits of detection and quantitation ------------- 53
4.4.4
Determination of trace elements in used oil samples -------------- 55
4.5
ANALYSIS OF QUALITY CONTROL SAMPLES BY ICP-OES ---- 60
4.5.1
Introduction ----------------------------------------------------------------------- 60
4.5.2
Accuracy and precision ------------------------------------------------------ 61
4.5.3
Validation criteria applied ---------------------------------------------------- 66
CHAPTER 5: CONCLUSION AND RECOMMENDATIONS ---------------- 73
5.1
INTRODUCTION ----------------------------------------------------------------- 72
5.2
ACHIEVEMENT OF THE OBJECTIVES----------------------------------- 72
5.3
CONTRIBUTION OF THE STUDY ------------------------------------------ 77
5.4
SHORTCOMING OF THE STUDY ------------------------------------------ 77
5.5
RECOMMENDATIONS FOR FURTHER RESEARCH ---------------- 77
REFERENCES -------------------------------------------------------------------------------------- 79
x
ANNEXURE A: DETAILS OF THE SIMPLEX ALGORITHM -------------- 86
A.1
DEFINITION OF THE SIMPLEX ALGORITHM -------------------------- 86
A.2
TYPES OF THE SIMPLEX ALGORITHM --------------------------------- 86
A.2.1
Fixed step simplex algorithm ----------------------------------------------- 86
A.2.2
Variable step simplex algorithm ------------------------------------------- 86
A.3
RULES OF THE SIMPLEX ALGORITHM --------------------------------- 87
A.3.1
Rules of the fixed step simplex algorithm ----------------------------- 87
A.3.2
Rules of the variable step simplex algorithm ------------------------- 88
ANNEXURE B: ADDITIONAL TABLES ---------------------------------------- 89
ANNEXURE C: ADDITIONAL FIGURES --------------------------------------- 95
xi
LIST OF FIGURES
Figure 1.1
Figure 2.1
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8
Figure 4.9
Bathtub curve showing the wear pattern of an internal combustion
engine --------------------------------------------------------------------------------- 2
Schematic representation of a typical ICP-OES instrument----------- 19
Determination of the most sensitive analytical line for
(a) Ag 328.068, (b) Ba 455.404, (c) Cu 324.754, (d) Mn 257.611,
and (e) Ni 341.476 --------------------------------------------------------------- 37
Progress of the simplex optimisation when applied to
(a) forward power, (b) plasma, (c) auxiliary and (d) nebuliser
flow, for Cu in xylene ------------------------------------------------------------ 41
Simplex optimisation progress showing trial number versus
SBRs for Mn in aqueous solution -------------------------------------------- 42
Influence of HNO3 on the extraction of (a) Ag, (b) Ba, (c) Cu,
(d) Mn, and (e) Ni, in used oils ------------------------------------------------ 44
Influence of aqua regia on the extraction of (a) Ag, (b) Ba,
(c) Cu, (d) Mn, and (e) Ni, in used oils -------------------------------------- 46
Calibration curves constructed using 0-50 µg g-1 Ag in 25%
aqua regia (m/v) ------------------------------------------------------------------ 47
Influence of HNO3:H2O2/2:1 on the extraction of (a) Ag, (b) Ba,
(c) Cu, (d) Mn, and (e) Ni, in used oils -------------------------------------- 49
Influence of HNO3:HCl/1:1 on the extraction of (a) Ag, (b) Ba,
(c) Cu, (d) Mn, and (e) Ni, in used oils ------------------------------------- 50
Calibration curve for the determination of Cu (324.754) in xylene -- 52
Figure 4.10
Mean concentrations of trace elements, in sample A of used
lubricating oil, analysed by ICP-OES ---------------------------------------- 55
Figure 4.11
Mean concentrations of trace elements, in sample B of used
lubricating oil, analysed by ICP-OES ---------------------------------------- 55
Figure 4.12
Mean concentrations of trace elements, in used engine oil
(15W-40) analysed by ICP-OES---------------------------------------------- 56
Figure 4.13
Mean concentrations of trace elements, in used engine oil
(20W-50) analysed by ICP-OES---------------------------------------------- 57
Figure 4.14
Mean concentrations of trace elements in used gear oil (EP 90),
working for about 12,000 km in a TOYOTA COROLA motor car ---- 57
Figure 4.15
Mean concentrations of trace elements in used gear oil (EP 90),
working for more than one year in a OPEL CORSA motor car ------- 58
Figure 4.16
Ultrasonic extraction of Ba under constant conditions of 10 mL
xii
aqua regia and 120 min sonication time ------------------------------------ 63
Figure A.1
Possible moves in the variable size simplex algorithm ----------------- 87
Figure C.1
Progress of the simplex optimisation of the ICP-OES when
applied to (a) plasma flow, (b) forward power, (c) auxiliary flow
and (d) nebuliser flow for Mn in aqueous matrix ------------------------- 95
Calibration curve for the determination of Ag in xylene diluted oils - 96
Figure C.2
Figure C.3
Figure C.4
Calibration curve for the determination of Ba in microwave
digested oils ------------------------------------------------------------------------ 96
Calibration curve for the determination of Cu in emulsified oils ------ 97
Figure C.5
Calibration curve for the determination of Mn in dry-ashed oils ------ 97
Figure C.6
Calibration curve for the determination of Ni in wet-ashed oils ------ 98
Figure C.7
Calibration curve for the determination of Cu in ultrasonically
extracted oils ----------------------------------------------------------------------- 98
xiii
LIST OF TABLES
Table 2.1
Boiling ranges of crude oil fractions ------------------------------------------- 9
Table 2.2
Common lubricating oil additives and their functions ------------------- 13
Table 2.3
Common wear metals and their origins ------------------------------------- 14
Table 2.4
Common lubricating oil contaminants and their origins ---------------- 15
Table 3.1
Ramp to temperature heating program of the MARS 5
microwave -------------------------------------------------------------------------- 32
Table 3.2
Instrumental limits of control variables for the simplex scheme ------ 34
Table 3.3
Setting of control variables for the modified simplex in
aqueous matrix -------------------------------------------------------------------- 34
Table 3.4
Setting of control variables for the modified simplex in
xylene matrix ----------------------------------------------------------------------- 34
Table 3.5
Setting of the modified simplex algorithm ---------------------------------- 34
Table 3.6
Factors and levels used in the full factorial experimental design
(43) of the ultrasonic extraction scheme ------------------------------------ 35
Table 4.1
Analytical lines established for the determination of trace metals --- 38
Table 4.2
Simplex optimised plasma conditions in aqueous matrix -------------- 39
Table 4.3
Simplex optimised plasma conditions in xylene matrix ----------------- 39
Table 4.4
Optimum conditions of an ultrasonic-assisted extraction -------------- 43
Table 4.5
Linear regression equations obtained by all methods studied -------- 53
Table 4.6
Radial ICP-OES limits of detection and quantitation of the
trace elements investigated ---------------------------------------------------- 54
Statistical comparisons of the analytical methods studied
using ANOVA ---------------------------------------------------------------------- 59
Table 4.7
Table 4.8
Analysis of variance for Cu results ------------------------------------------- 60
Table 4.9
Comparison of Cu by treatments (methods) ------------------------------ 60
Table 4.10
Mean recoveries (±SD) obtained from the analysis of
Conostan S-21 standard by ICP-OES -------------------------------------- 61
Table 4.11
Mean recoveries (±SD) obtained from the analysis of
EnviroMAT (certified reference material) by ICP-OES------------------ 63
xiv
Table 4.12
Analytical criteria applied for method validation -------------------------- 67
Table B.1
Simplex optimisation progress data of ICP-OES operating
parameters, for Ag in xylene matrix ------------------------------------------ 89
Simplex optimisation progress data of ICP-OES operating
parameters, for Mn in aqueous matrix -------------------------------------- 91
Table B.2
Table B.3
Sensitivity study of ICP-OES lines for the elements investigated---- 92
Table B.4
The full factorial experimental design (43) applied to the
optimisation of the ultrasound-assisted extraction of metals in
lubricating oils --------------------------------------------------------------------- 93
xv
LIST OF ABREVIATIONS USED
AAS:
atomic absorption spectrometry
AGMA:
American gear manufacturers association
ANOVA:
analysis of variance
ASTM:
American society for testing materials
CCD:
charge coupled device
CID:
charge injection device
DCP:
direct current plasma
GF-AAS
graphite furnace- atomic absorption spectrometry
GNP:
gross national product
ICCD:
intensified charge coupled device
ICP-MS:
inductively coupled plasma-mass spectrometry
ICP-OES/AES: inductively coupled plasma-optical/atomic emission spectrometry
IDL:
instrument detection limit
ISO:
international organisation for standardisation
JOAP:
joint oil analysis program
LOD:
limit of detection
LOQ:
limit of quantitation
MDL:
method detection limit
MHz:
mega hertz
MIP:
microwave induced plasma
OECD:
organisation for economic co-operation and development
PAG:
polyalkylene glycols
PAH:
polycyclic aromatic hydrocarbon
PAO:
polyalphaolefins
PCB:
polychlorinated biphenyl
PCT:
polychlorinated terphenyl
QC:
quality control
RDE-OES:
rotating-disc electrode sparks optical emission spectrometry
xvi
RF:
radio frequency
SBR:
signal to background ratio
SUS:
saybolt universal seconds
TAN:
total acid number
TBN:
total base number
ZDDP:
zinc dialkyl dithiophosphate
xvii
GLOSSARY
Glossary of terms as applied to this study:
Algorithm:
A prescribed set of rules or procedures for the solution of a
problem in discrete steps.
Atomization:
Process of dissociating vaporised sample molecules into free
atoms.
Blank solution:
A solution that doesn’t contain a detectable amount of the analyte
of interest. Usually used for calibration purposes.
Calibration curve:
A plot or equation that describes the relationship between the
variable that is measured to indicate the presence of an analyte
and the concentration of the analyte.
Decay:
Any process in which an excited atom or ion changes to a lower
energy state.
Desolvation:
Process of removing the solvent molecules from a sample
droplet, resulting in a dried sample particle.
Detector:
A photosensitive device that absorbs photons and converts them
into electrons with certain efficiency. These photoelectrons can
be measured to indicate the intensity of the incoming light.
Excitation:
A process, in which an electron is promoted to a higher energy
level, resulting in an atom or ion said to be in an excited state.
Flash point:
The lowest temperature at which the lubricant emits enough
vapour to ignite in the presence of flame but not continue to burn.
Injector:
The centremost tube of an ICP torch through which the sample
aerosol is introduced to the plasma.
Interference:
Any thing that causes the signal from an analyte in a sample to
be different from the analyte signal for the same concentration of
that analyte in a single element calibration solution.
Matrix matching:
An approach to interference correction in which the major
chemical compositions of the standards, blanks and samples are
xviii
made identical aiming to cancel out the effect of sample matrix on
the analytical results.
Matrix:
The components of the sample other than the elements of
interest.
Nebulisation:
The process of creating aerosols from liquid samples using
mechanical or other forces.
Optimisation:
The process by which the response of a system is improved to its
best possible response.
Pour point:
The lowest temperature at which the paraffinic wax content of the
oil becomes crystallizes and loses its flow characteristic.
Real sample:
Refers to used oil samples collected from local garages for
investigation.
Sonochemistry:
The branch of chemistry which concerns with chemical changes
caused by or involving sound, particularly ultrasound energy.
Tesla coil:
An electrical device used to create a high-voltage spark; often
used for igniting an ICP discharge.
Wavelength:
The distance between two adjacent peaks of a monoenergetic
electromagnetic wave.
Working range:
The range of concentrations over which reliable analyte
determinations can be made.
1
CHAPTER 1
INTRODUCTION
1.1
BACKGROUND AND MOTIVATION
Lubricating oils, such as engine, gear, hydraulic and turbine oils, are primarily used to
reduce friction and reducing heat generation between contacting surfaces (Jha,
2005). They are generally formulated from mineral base oils, obtained from one of the
fractions of crude oil distillation. These petroleum based lubricants are composed
primarily of paraffinic, naphthenic and aromatic hydrocarbons, distilled between 300
and 400 oC. A variety of oil additives, introduced to enhance the oil’s performance,
also contribute to the final composition of the lubricating oil (Aucelio et al., 2007b;
Barbeira & Ferreira, 2009). Lubricants that have been used in machine components
for a period of time have their composition altered because of the breakdown of
additives, contamination with the products of combustion and the addition of metals
from component wear and tear. Although it is difficult to generalise about the exact
composition of used engine oil, the major components include aliphatic and aromatic
hydrocarbons including phenol, naphthalene, benzanthracene, benzopyrene and
fluoranthene (Basham et al., 1997).
The main sources of metals in lubricating oil include additives, wear metals and
contaminants (Bombelka & Nham, 1991). Additive metals are deliberately blended
into the final product, in order to enhance the oil’s performance (Aucelio et al.,
2007b). Some of the most common oil additives include: anti-wear, anti-rust, oxidation
inhibitors, viscosity index improvers, dispersants, anti-foaming agents, extreme
pressure additives and corrosion suppressors (Thermo Elemental, 2002; Thomson,
2007). Contaminants including metals, fuel, glycols, water and debris can be
introduced into the lubricating oil, most often through dust, leaks or residual metal
pieces (Jha, 2005). Wear metal residues occur when moving metal surfaces come
2
into contact with each other. The sources of wear could result from either physical or
chemical mechanisms (Aucelio et al., 2007b; Kuokkanen et al., 2008). When the
lubricated metal surfaces rotate, the wear metal concentrations slowly increase due to
normal wear processes. However, if the concentration of one or more metals in the oil
suddenly increases, it may be an indication of excessive wear of that component
(Body et al., 2005). The onset of wearing is often represented with a bathtub curve as
shown in Figure 1.1, in which the machine initially experiences running-in-wear,
followed by normal wear and finally fast abnormal wear. This eventually leads to
extreme component damage. Evaluations have been carried out in the United States
of America, England and Germany to estimate the costs due to friction and wear.
These were found to be between 2 to 7% of the gross national product (GNP) of the
country in which the oil was studied (Kuokkanen et al., 2008).
Figure 1.1: Bathtub curve showing the wear pattern of an internal combustion
engine (Kuokkanen et al., 2008)
Lubricating oils, particularly those used in an internal combusting engine, perform
their function under the adverse conditions of high temperature and pressure.
Fortunately, lubricating oils can be designed or modified according to the amount and
3
nature of the metals present in the machine component (Chausseau, Lebouil-Arlettaz
& Thomas, 2010). In general, the most important functions of lubricating oils, as
mentioned in the literature by Gedeon (1999) and Thermo Elemental (2002), include:

Reduction of the overall system friction that subsequently reduces heat
generation and formation of wear particles.

Removal of considerable amounts of heat from a system.

Protection of metal parts from corrosion and rust.

Removal of internally generated debris and contaminants from the system.

Transmission of mechanical power as in hydraulic systems.

Sealing of machine components against dust, dirt and water intrusion.

Provision of insulation in transformer applications.
New engine oil contains lighter hydrocarbons that may cause a short-term toxicity to
aquatic species, whereas used engine oil contains more metals and heavy polycyclic
aromatic hydrocarbons (PAHs). Aromatic hydrocarbons are acutely hazardous
products that can cause chronic toxicity and are often carcinogenic (Basham et al.,
1997). Used oils are extremely hazardous to the environment due to their persistent
nature and their tendency to become disseminated in soil and water. A significant
amount of used oil enters the environment due to reservoir leaks and sudden spills or
illegal disposal. Bartz (cited by Vahaoja, 2006) found in 1998, that worldwide, more
than 12 million tons of used oil is released into the environment annually. Once used
oil has been spilled or enters sewers, it seeps to underground water and causes
severe pollution. For example, one litre of used oil has the capacity to pollute one
million litres of water, rendering it undrinkable (Jha, 2005).
The analysis of used lubricating oil dates back to the early 1940s. It was done by
railway companies, in an effort to detect wear in heavy diesel locomotive engines.
Lubricant inspection and testing has been used for many years to help diagnose the
internal condition of oil‐wetted components. Due to the expansion of basic research
and technology, lubricating oil analysis has evolved and is becoming the most widely
4
implemented form of proactive maintenance technology (Analysts, 2008; Milner,
2009). Regular oil testing is now an integral part of the maintenance plan for shipping,
aviation, the military, oil refineries, mining operations, processing and chemical plants
(Body et al., 2005; Milner, 2009). Lubricating oil that has serviced moving parts of
machine components for a period of time reflects the exact condition of the assembly
of the machine. This is because wear particles, detached from various types of metal
parts, as well as combustion products, are suspended in the lubricating oil and
circulate with it. In this way, the oil composition becomes the working history of the
machine. This forms the basis of oil analysis used to determine wear status through
the level of contamination (Vahaoja, 2006; Kuokkanen et al., 2008).
Oil analysis provides a means of predicting impending failure without dismantling the
machine. Oil analysis can also detect fuel dilution, oil contamination, antifreeze leaks,
excessive bearing wear and incorrect application of lubricants (Analysts, 2008). Early
detection of wear status and level of contaminants through oil analysis reduces repair
bills, increases machinery life, reduces unscheduled downtime, extends lubricant life
and improves safety and environmental awareness (Mobil, 2009).
Elemental analysis of lubricating oil is often difficult, due to the complexity of the oil
matrix, its high organic load and viscosity. Hence, a very sensitive analytical
technique is required to be able to monitor small changes in concentration of the key
elements (Aucelio et al., 2007b). This technique should be sufficiently sensitive,
accurate, reliable, easily automated and computer controlled. The most common
analytical techniques employed for used oil analysis include rotating-disk electrode
spark-optical
emission
spectrometry
(RDE-OES),
laser-induced
breakdown
spectroscopy (LIBS), graphite furnace-atomic absorption spectrometry (GF-AAS),
inductively coupled plasma - optical emission spectrometry (ICP-OES) and inductively
coupled plasma-mass spectrometry (ICP-MS) (Body et al., 2005). These techniques
eliminate complex dissolution procedures and can be used for multi-element analysis.
In addition, LIBS can be used for analysis of samples of any type (solid, liquid and
gas) with minimal or no sample preparation (Body et al., 2005; Agnelli et al., 2009).
5
Most spectrometric techniques require some kind of sample pre-treatment prior to
elemental analysis, in order to make the sample compatible with the sample
introduction system. The choice of such preparation procedure is critical for the
success of the analytical methodology (Aucelio, Silveira & Souza, 2004). Several oil
pre-treatment approaches have been proposed by Angnes, Munoz and Oliveira
(2006), Ana et al. (2007) and Aucelio et al. (2007b) and are listed below:
1.2

Sample dilution with appropriate organic solvents.

Conversion of the sample into oil-in-water emulsion.

Dry and wet ashing.

Microwave assisted acid digestion.

Ultrasonic-assisted extraction.
PROBLEM STATEMENT
Methods developed for the analysis of used oil can play a vital role in monitoring the
condition of engines and may help to improve overall equipment effectiveness.
International organisations like the American Society for Testing Materials (ASTM),
the International Organisation for Standardisation (ISO) and the Organisation for
Economic Co-operation and Development (OECD), have developed standards for oil
analysis (Vahaoja, 2006). Despite this, there still are major analytical challenges
associated with oil analyses. The complexity of the oil matrix and its high organic load
and viscosity, complicate the accurate determination of trace elements. Moreover,
various analytical techniques for lubricating oil analysis give rise to different results
(Aucelio, Silveira & Souza, 2004; Filho & Netto, 2009; Goncalves, Gonzalez & Murillo,
1998). This should be investigated to determine the reasons for the differences.
Traditionally, organo-metallic compounds have been employed as standards, aiming
to compensate for matrix effects and allow the use of simple and direct external
calibration. However, organic standards are expensive and may lead to inaccurate
results due to concentration changes, arising from the evaporation and deposition of
these species on the walls of the storage container (Aucelio et al., 2007b).
6
To address the above analytical challenges, the following research questions need be
investigated through experimental studies:

Which sample pre-treatment technique minimises the organic load of the
oil?

What are the optimal ICP-OES operating parameters that give the most
sensitivity?

Is it possible to extract inorganic analytes from used lubricating oil samples
using ultrasonic energy generated from an ultrasonic bath? If so, what
extraction conditions could be applied?

Different sample preparation methods often yield different results for the
same sample type. What could be the possible reason for such differences?
1.3
HYPOTHESIS
Trace elements in lubricating oil can be determined accurately by spectrometric
techniques if an appropriate sample pre-treatment has been carried out.
1.4
OBJECTIVES
1.4.1
General objective
The main objective of this work is to investigate methods used for the accurate
determination of trace elements in lubricating oils.
In this study, determination of trace metals (Ag, Ba, Cu, Mn and Ni) in six different
used lubricating oil samples was performed by ICP-OES. The reason for choosing
these elements was because of:

Organo-metallic compounds of Ag, Ba and Ni are commonly used as oil
additives (Aucelio et al., 2007b).
7

Ag and Ba are among the elements regulated as hazardous under Resource
Conservation and Recovery Act (RCRC) (Eslling et al., 1995).

Cu and Mn are wear metals and at the same time they are also oil
contaminants (AGAT laboratories, 2004).
1.4.2
Specific objectives
The specific objectives are to:

compare and evaluate the efficiencies of six sample pre-treatment methods
(xylene dilution, emulsification, microwave assisted digestion, dry and wet
ashing and ultrasonic extraction) in terms of accuracy, precision and
analytical drawbacks using ANOVA;

optimise ICP-OES operating parameters both for aqueous and organic
matrices, using a modified simplex algorithm;

optimise factors affecting the ultrasonic extraction of analytes in used
lubricating oils, employing full factorial design;

establish interference free analytical lines for the determination of selected
trace metals in used lubricating oil;

determine limits of detection and quantitation for the selected trace metals in
used lubricating oil; and

perform method validation employing certified reference materials.
8
CHAPTER 2
LITERATURE SURVEY
2.1
INTRODUCTION
This literature review provides details of the fundamental concepts of lubricating oil.
The concepts include: lubricating oil chemistry, types and functions of lubricating oil,
properties of lubricating oil, and sources of elements in lubricating oil. In addition,
details of optimisation schemes and method for oil analysis are included.
2.2
LUBRICATING OIL CHEMISTRY
2.2.1
Background
Lubricating oil, sometimes referred to as “lube oil”, is usually a liquid substance
introduced between two moving surfaces, primarily to reduce friction and wear
(Castro & Priego-Capote, 2004). It is one of the fractions of crude oil distillation,
containing 20 to 70 carbon atoms per molecule, often in an extremely complex
arrangement of straight chains and side chains of five and six membered ring
structures (New Zeeland Institute of Chemistry, 2002).
2.2.2
Refining lubricating oil from crude petroleum
Crude oil is a naturally occurring mixture, consisting largely of hydrocarbons formed
by bacterial decomposition of animal and plant remains, deposited about a hundred
million years ago. Though the composition of crude oil is highly dependent on the
source, it usually contains 2% lubricating oil (New Zeeland Institute of Chemistry,
2002). Crude oil is refined by a process of fractional distillation, in two stages, to drive
off the various fractions as vapour. The first-stage, known as crude distillation,
9
reduces the number of compounds within each fraction and provides different
qualities of fuels. The final products derived from this distillation stage are raw
gasoline, kerosene, and diesel fuel. The product that does not distil off in the first
distillation stage, that is, the residuum, distils in the second stage. Correct distillation
of the residuum yields oils with different boiling points, including lubricating oil. Since
the lubricating oil fraction is very stable and difficult to volatilize, heating the oil to the
temperature required for distillation will result in decomposition (Gedeon, 1999; New
Zeeland Institute of Chemistry, 2002). Hence, the lubricating oil fraction is best carried
out under vacuum distillation, which reduces the boiling point from 500 oC to about
300 oC. In Table 2.1 the boiling ranges of crude oil (New Zeeland Institute of
Chemistry, 2002) are listed.
Table 2.1:
Boiling ranges of crude oil fractions (New Zeeland Institute of Chemistry,
2002)
Crude component
Gasoline
Kerosene
Diesel
Lubricating oil
Boiling Range (oC)
40 – 190
190 - 260
260 - 330
330 – 400 (under vacuum)
2.3
TYPES OF LUBRICATING OILS
2.3.1
Mineral oils
The base stock of mineral or petroleum lubricating oils is derived primarily from crude
oil refining (Castro & Priego-Capote, 2004). These mineral base stocks are complex
mixtures, which cannot typically be defined by detailed compositional information.
They are studied by their process history, physical properties, and product use
specifications (Petroleum HPV Testing Group, 2003). Depending on the prevailing
composition, mineral oils are further classified into three categories: namely paraffinic,
naphthenic and aromatic oils.
10
2.3.1.1
Paraffinic oils
Paraffinic oils are predominantly characterised by long straight chain hydrocarbons
derived from crude petroleum. The hydrogen and carbon atoms in paraffinic oils are
linked in a long linear pattern, similar to a chain. Paraffinic oils contain paraffinic wax
and are the most widely used base stock for lubricating oils. When compared with
naphthenic oils, paraffinic oils have characteristics of excellent stability (higher
resistance to oxidation), higher pour point and viscosity index, low volatility, high flash
point, low specific gravity and tend to be waxy (Gedeon, 1999; New Zeeland Institute
of Chemistry, 2002).
2.3.1.2
Naphthenic oils
Naphthenic oils are predominantly characterised by a molecular structure composed
of rings of hydrocarbons derived from crude petroleum. The hydrogen and carbon
atoms in naphthenic oils are linked in a circular pattern to form their characteristic
saturated ring structure (Gedeon, 1999). Unlike paraffinic oils, naphthenic oils have
low pour point and don’t contain wax. Naphthenic oils are generally applied for narrow
temperature ranges, where a low pour point is required, such as refrigeration oil.
When compared to paraffinic oils, naphthenic oils have good stability, lower pour
point and viscosity index, higher volatility, lower flash point, higher specific gravities
and are highly carcinogenic (Gedeon, 1999; New Zeeland Institute of Chemistry,
2002).
2.3.1.3
Aromatic oils
Aromatic oils include polycyclic aromatic compounds (PACs), some of which are
heterocyclic and incorporate S, N, and O. These oils are characterised by their high
volatility and low oxidative stability. In general, aromatics are the unwanted
components of the mineral base stock, which lower the performance of the lubricant.
11
Usually, they are removed from the base stock by a process of hydro-cracking
(Petroleum HPV Testing Group, 2003).
2.3.2
Synthetic oils
Synthetic oils are produced by chemical synthesis rather than from the refinement of
petroleum based oils. When compared with mineral oils, synthetics are generally
superior with respect to excellent oxidation stability and viscosity index, much lower
pour point (as low as -46 oC) and lower coefficient of friction. Synthetic lubricants can
be used both in extremely low and high temperature applications. The main
disadvantage of synthetic lubricants is their high initial cost, which is about three
times that of petroleum based lubricants. However, the high cost of synthetic oils is
compensated for by their prolonged serviceability, which is about three times longer
than mineral oils (Gedeon, 1999).
2.4
PROPERTIES OF LUBRICATING OIL
Details of the most important properties of lubricating oil are described in the literature
by Gedeon (1999) and Analysts (2008). These include the viscosity, viscosity index,
total acid number and total base number of the oil.
2.4.1
Viscosity
The most important physical property of a lubricant is its viscosity. It is the fluid’s
resistance to flow. Viscosity and resistance to flow are directly proportional, that is, a
high viscosity gives rise to a high resistance to flow and vice versa. Viscosity and
pressure have also a direct relationship, whereas viscosity and temperature are
inversely proportional. The viscosity measurement most commonly carried out is that
of kinematic viscosity, which is the ratio of the fluid’s absolute viscosity to its density
and is expressed in centistokes (cSt) at conventionally accepted temperatures of 40
12
and 100 oC. It can also be expressed in centipoises (cP) and Saybolt Universal
Seconds (SUS), depending on the test method used. Lubricants are distinguished
from each other by their viscosity ranges, known as oil grades. The most familiar oil
grading systems include the Society of Automotive Engineers (SAE), the American
Gear Manufacturers Association (AGMA), the International Organisation for
Standards (ISO) and the American Society for Testing and Materials (ASTM).
2.4.2
Viscosity index
The viscosity index (VI) is a number that indicates how the oil viscosity changes with
temperature. Since the introduction of viscosity index in 1929, a reference paraffinic
base stock has been assigned a VI of 100 and a naphthenic base stock a VI of zero.
A high viscosity index implies small oil viscosity changes with temperature, while a
low viscosity index implies high viscosity changes. The knowledge of oil’s viscosity
index is a crucial factor when selecting a lubricant for an application at extremely hot
or cold temperatures.
2.4.3
Total acid and base numbers
The total acid number (TAN) or neutralisation number, measures the total amount of
acidic material in lubricating oils. An increase in the TAN of a working lubricant, with
respect to a new lubricant, is an indication of the occurrence of oil oxidation or
contamination by acidic products. Lubricants applied at extreme temperature usually
become oxidised and subsequently produce organic acids. The TAN is thus a good
diagnostic tool for indicating when an oil change is required. In contrast, the total base
number (TBN) measures the total alkaline content in the lubricating oil. Lubricants
applied in internal combustion engines are often designed with additives offering
higher alkaline content in order to neutralise the acid produced from oil oxidation. A
significant decrease in the lubricant’s alkaline content is a sign of low acid-neutralising
ability or depletion of the additives that enhance the oil’s alkalinity.
13
2.5
SOURCES OF TRACE ELEMENTS IN LUBRICATING OIL
This section provides a general orientation on oil additives, wear metals and
contaminants, which are considered as the most important sources of trace elements
in oils or oil tests.
2.5.1
Additive elements
Organo-metallic compounds containing Ag, B, Ba, Bi, Ca, Cd, Co, Cr, Fe, Mg, Mo, Ni,
P, Sb, and Zn are deliberately added to lubricating oil to increase performance.
Almost all lubricating oils are blended with additives in controlled amounts (Aucelio et
al., 2007b). However, additives alone cannot establish oil quality with respect to
oxidation resistance, emulsification, pour point and viscosity index. Therefore, oil
additives need to be blended with premium quality base stock, in order to yield quality
lubricants (Gedeon, 1999). The most common lubricant additives and their functions
are illustrated in Table 2.2.
Table 2.2:
Common lubricating oil additives and their functions (AGAT laboratories,
2004; Analysts, 2008; BITOG, 2010; Chausseau, Lebouil-Arlettaz &
Thomas, 2010; Evans, 2010)
Additive elements
Ag
B
Ba
Ca
Fe
K
Mg
Mn
Mo
Na
Ni
P
Sb
Zn
Functions
Antifriction and coatings.
Corrosion, oxidation, wear and water inhibitor.
Detergent, corrosion and rust inhibitor.
Detergent, dispersant and alkalinity increaser.
Antifriction in rolling elements or bearings.
Corrosion inhibitor.
Detergent, dispersant and alkalinity increaser.
Combustion improver and smoke suppresser in residual and
distillate oil.
Anti-wear and EP additive (in special oils).
Corrosion inhibitor.
Antifriction alloy and coatings
Anti-wear, anti-oxidant, corrosion inhibitor and EP additive.
Anti-wear and anti-oxidant additive.
Anti-wear, anti-oxidant and corrosion inhibitor additive.
14
According to BITOG (2010) the three functions of oil additives are to:

protect lubricated metal surfaces from rust and corrosion;

extend the range of lubricant applicability; and

extend lubricant life.
2.5.2
Wear metals
Wear is the progressive damage resulting in material loss due to the relative contact
between adjacent working parts (Gedeon, 1999). Motion types that cause wearing are
sliding, rolling, impact loading, vibration and liquid flow with or without solid particles
(Kuokkanen et al., 2008). During normal operations, wearing is considered to be
inevitable. However, excessive friction causes premature wear, which results in
significant economic loss due to equipment failure, cost of replacement parts and
downtime (Gedeon, 1999). The wear products are composed of the same material as
the metal surfaces from which they originated (Gedeon, 1999; Kuokkanen et al.,
2008). The most common wear metals and their sources are listed in Table 2.3.
Table 2.3:
Common wear metals and their origins (AGAT laboratories, 2004;
Analysts, 2008; BITOG, 2010; Chausseau, Lebouil-Arlettaz & Thomas,
2010)
Wear metals
Al
Cr
Cu
Fe
Mg
Mn
Mo
Pb
Si
Sn
Zn
Origins
Engine block, oil coolers, pistons, push rods, blowers, oil pump bushing,
bearing cage and gears.
Cylinders, piston rings, compression rings, gears, crank shaft, seals,
piston rings and plating materials.
Bearing, oil coolers, connecting rods, piston pins, gears, clutch plate,
valves, radiators, bushes and injector shields.
Cylinders, piston rings, engine block, valve train, oil pump bearings,
spring gears, rolling elements, shaft and rust.
Component housing, wear of cylinder liner and gears.
Wear of cylinder liner, blower, exhaust and intake systems, shafts and
valves.
Wear of bearing alloys, surface coating in some piston rings.
Bearing, seals, clutch and solder.
Engine block and gasket sealants.
Bearings, wrist and piston pins, solder, piston overlay seals and bushes.
Wear of galvanised piping and brass alloy.
15
2.5.3
Lubricating oil contaminants
Contamination is usually associated with the introduction of foreign substances into
the oil system. Particulate matter, water, fuel dilution and anti-freeze are some of the
most common lubricant contaminants (New Zeeland Institute of Chemistry, 2002;
Analysts, 2008). Contaminants act as catalysts to facilitate the process of oil
degradation and system component deterioration. Severe oil contamination
deteriorates the system’s operation, life and reliability. The Oklahoma State University
reported (Jeremic & Macuzic, 2004) that if a hydraulic fluid is kept ten times cleaner, a
fifty times extension in hydraulic pump life would be possible. Some of the most
common lubricant contaminants and their sources are listed in Table 2.4
Table 2.4:
Common lubricating oil contaminants and their origins (AGAT
laboratories, 2004; Analysts, 2008; BITOG, 2010; Chausseau, LebouilArlettaz & Thomas, 2010)
Wear metals
Origins
Al
B
Ba
Ca
Cr
Cu
K
Mg
Mn
Mo
Na
As aluminium silicate from ingested sand or dust.
Coolant and grease leaks.
From diesel fuel (smoke depressant) and grease contamination.
From hard water, road salts and grease contamination.
From chromate coolant treatment and dirty intake.
Coolant leak.
Coolant leak.
From hard water and road salts.
From detergent additive in unleaded gasoline.
From grease and some coolants (anti-corrosion).
Grease and coolant additive, road salt and ingested dirt.
P
Coolant leak.
Pb
Regular (leaded) gasoline leak.
Si
From grease, sand or dust breathers.
Sn
From oil coolers.
16
2.6
SEQUENTIAL SIMPLEX OPTIMISATION
2.6.1
Introduction
A good analytical method acts on parameters that influence the response of the
instrument at settings that give maximum readings. The traditional univariate method
that optimises one parameter at a time, while holding the others constant, often fails
to find the optimum conditions for interdependent parameters (Allman, 1995).
Multivariate approaches, such as factorial design, can be applied for such
interdependent parameters, but are more complicated, tedious and time-consuming.
The simplex optimisation technique is the most popular approach for true optimisation
of interrelated parameters with ease and speed (Cave, Ebdon & Mowthorpe, 1980;
Carpenter & Ebdon, 1986). The two most important aspects of the simplex approach
are the small number of experiments required for the initial simplex (k+1) and the
efficiency of finding the optimum (Deming et al., 1999).
Galley, Hieftje and Horner (1995) used both net signal and signal-to-noise ratio (SNR)
as response functions to optimise four operating parameters of an ICP-OES (rf
power, intermediate argon flow, nebuliser flow and sample uptake rates) using the
CaII and MgI lines and automated simplex optimisation. Cave, Ebdon and Mowthorpe
(1980) successfully optimised five operating parameters of an ICP-OES (rf power,
coolant gas flow, injector gas flow, plasma gas flow and observation height) using
variable size simplex, employed both signal-to- background ratio (SBR given in Eq.
2.1), which is a measure of the range of total signal intensity and background relative
to the background values, and SNR, as criterion of merits. Carpenter and Ebdon
(1986) also optimised three ICP-OES parameters (rf power, injector gas flow and
observation height) for two sample introduction–torch configurations using variable
size simplex and SBR as a response function. They obtained small but significant
limits of detection for the six elements studied. Salin and Sartoros (1997) also used
SBR as a figure of merit to optimise ICP operating parameters (rf power and
observation height) for seven elements using a simplex algorithm.
17
SBR =
2.6.2
(Eq. 2.1)
The simplex calculations
The various projections of the simplex algorithm moving away from the rejected trial
(W) are calculated using the following equations (Deming et al., 1999):
P=
(Eq. 2.2)
R = P + (P-W)
(Eq. 2.3)
E = P + (P-W) = R + (P-W)
(Eq. 2.4)
C+ = P + +(P-W)
(Eq. 2.5)
C- = P - -(P-W)
(Eq. 2.6)
where, P is the centroid; R the reflection; E the expansion; W the worst trial; C+ the
positive contraction; C- the negative contraction;  the reflection coefficient (usually 1);
 the expansion coefficient (usually 2); + the positive contraction coefficient (usually
0.5); and - the negative contraction coefficient (usually 0.5)..
2.6.3
Rules of the simplex algorithm
The simplex procedure is designed to force the simplex experiments away from
regions of poor response to towards the region of optimum response, guided by a
well defined series of rules and moves, given in Annexure A. The possible moves of
the modified simplex algorithm are depicted in Figure A.1 of Annexure A (Deming et
al., 1999).
18
2.6.4
Boundary violation
The simplex algorithm may require to carry out an experiment outside the boundary
variables, which is termed a boundary violation. If the boundary constraints are set for
convenience and are not absolute, removing the boundary constraints can be
considered. However, if the boundaries have been set because of safety
considerations, the boundaries should not be removed. Alternatively, the investigator
can simply assign the out-of-bounds vertex with an infinitely bad response. This
action automatically forces the simplex to undergo a negative contraction (C-) that
brings the simplex back within bounds (Deming et al., 1999).
2.6.5
The convergence criteria
The simplex experimentation must be halted at some point of convergence. The k+1
rule, proposes that if a vertex were in a simplex k+1 times and was not about to be
rejected due to the possibility of recording an incorrect good response, then the
progress of the simplex should be halted and the vertex re-evaluated. The threshold
criteria also suggest that the experiment should stop when the threshold is reached.
Further experimentation is unjustified. This rule is widely used in most optimisation
projects (Deming et al., 1999).
2.7
INDUCTIVELY COUPLED PLASMA-OPTICAL EMISSION
SPECTROMETRY
2.7.1
Introduction
Inductively coupled plasma-optical emission spectrometry (ICP-OES) is a technique,
based on electrons of an excited atom emitting energy, at a given wavelength, as
they return to their ground state. The energy emitted by the excited atoms or ions is a
characteristic of the atom. Wavelength separation takes place in the spectrometer.
19
The intensity of the emission radiation, which is directly proportional to the
concentration of these species, is measured by the aid of advanced detectors (Boss
& Fredeen, 2004).
2.7.2
Instrumentation for ICP-OES
The main components of a typical ICP-OES instrument setup are as depicted in
Figure 2.1. These include: the ICP-discharge, a highly ionised gas where excitation
and ionisation takes place; ICP-torch, a quartz tube used to support and introduce the
sample into an ICP-discharge; rf generator, a device providing the power for the
generation and sustainment of the ICP-discharge; nebuliser, a device used to create
aerosols from liquid samples; spray chamber, a device that removes large droplets
from the aerosol; and computer system, where instrument control and data
processing takes place.
Figure 2.1: Schematic representation of a typical ICP-OES instrument (Boss &
Fredeen, 1997)
20
2.7.3
Sample preparation for ICP-OES analysis
Few analytical techniques allow direct sample analysis with minimal or no sample
pre-treatment. Most analytical techniques require the sample used for analysis to be
in solution form. This sample solution should not contain a high level of insoluble
particles and its viscosity should be reasonable (close to water). Unfortunately, most
of the pre-treatment approaches to samples are relatively labour intensive and timeconsuming, in particular for large samples (Kuokkanen et al., 2008). Consequently,
60% of the total time required for complete analysis is spent in sample preparation.
This also accounts for 30% of the total analysis error (Oliveira, 2003; Costa et al.,
2005).
The purpose of sample pre-treatment for ICP-OES analysis is primarily to minimise
the organic load and viscosity of oils. Samples with high organic load affect the
stability of the ICP, increase background emission and reduce the level of energy
suitable for ionisation and excitation. These factors are closely related to the
sensitivity of the ICP-OES and to the correlation between sample and standard
signals that form the calibration curve (Aucelio, Silveira & Souza, 2004).
2.7.3.1
Dilution of lubricating oil with xylene
The traditional dilution method for oil samples, using organic solvents, is simple and
rapid. It is convenient for rapid estimation of indicator metals. Organic solvents such
as xylene and kerosene are typical solvents used with ICP-OES. Instrument
calibration for the simple dilution method, has to be carried out using organo-metallic
standards. The obvious drawback of this method is the high cost of these standards
and the inability of the organic solvents to dissolve particles larger than 10 μm. This
may cause loss of sensitivity for the emission spectrometers, during the nebulisation
process (Kuokkanen et al., 2008). Aucelio, Silveira and Souza (2004) diluted oil
samples using kerosene to determine refractory elements by ICP-OES. Chausseau,
Lebouil-Arlettaz and Thomas (2010) also diluted oil samples using kerosene, to
21
determine wear metals and additives using ICP-OES. They obtained results with a
deviation of less than 5% from the expected concentration. Goncalves, Gonzalez and
Murillo (1998) applied xylene to dilute oil samples for the determination of metals by
AAS.
2.7.3.2
Emulsification
A detergent emulsion of oil-in-water is formed by adding water and surfactant to oil
samples, in order to convert them into homogenously dispersed micro-droplets in the
aqueous phase. This approach does not require destruction of the organic matter or
use of large amounts of organic solvents; however, it reduces the organic content of
samples. When the oil sample is homogenously dispersed in the water phase, the
mixture behaves similar to an aqueous solution (Aucelio et al., 2007b). Acidification of
the emulsion converts insoluble particles, organo-metallic species and oxides into
dissolved inorganic species (Aucelio, Silveira & Souza, 2004). Goncalves, Gonzalez
and Murillo (1998) tested different emulsification methods, with and without acid
treatment, and emphasised the importance of carrying out acid treatment in order to
dissolve solid metal particles. Aucelio, Silveira and Souza (2004) also compared the
use of detergent emulsions with microwave-assisted acid digestion and kerosene
dilution, for the analysis of refractory elements in lubricating oils. They observed
certain advantages when using emulsification in comparison with the traditional pretreatment methods. Leocadio, Silveira and Souza (2008) proposed detergent
emulsion formation with biodiesel, for the determination of Ca, Cu, Fe, Mg, Mn, Na,
and P, using radial and axial ICP-OES. They used a mixture that had 1.0 g biodiesel,
0.2 mL conc. HNO3, 6% Triton x-100 and water. They were able to achieve at least 12
h of detergent stability. Goncalves, Gonzalez and Murillo (1998) treated oil samples
with an acid mixture, water and nonylphenol, as surfactant, to determine metals by
AAS. They found good agreement between calibration curves of aqueous and
emulsified standard solutions.
22
2.7.3.3
Dry-ashing
The dry-ashing technique is based on the organic material being ignited in air or
oxygen, in order to determine the inorganic elements present in the organic
substance. The organic substance reacts to form gaseous carbon dioxide and water
vapour, leaving the inorganic materials behind as solid oxides. Usually, ashing is
performed using crucibles or an evaporating dish of platinum or fused silica, in a
muffle furnace. Dry-ashing cannot be applied to determine volatile elements such as
Hg, As or Cd, as they can be lost even at relatively low temperatures (Frame, Frame
& Robinson, 2005). The dry-ashing procedure is extremely time-consuming; however,
it ensures that the organic matrix is mineralised and the total metal content (coming
from both physical and chemical wear) is converted to simple water-soluble species.
These water-soluble species allow the use of inorganic standards for calibration
(Aucelio et al., 2007b). Goncalves, Gonzalez and Murillo (1998) used a method in
which the treatment of 5 g of oil sample took over 5 h. The sample was first heated
until dryness, when no more smoke appeared; two digestion temperatures of 250 oC
for 1 h and 600 oC for 4 h were applied. The final residue was dissolved with
concentrated HCl and diluted with water. Similarly, Agarwal, Sharma and Singh
(2006) took over 6 h to treat 5 g of oil sample. The oil was first heated on a hot plate
until completely dry, after which two digestion temperatures of 450 oC for 4 h and
650 oC for 2 h were applied to samples in a muffle furnace. The final residue was then
dissolved with concentrated HCl and diluted with water.
2.7.3.4
Wet-ashing
The wet-ashing technique is based on the organic material being degraded, using
high ignition temperatures, strong acids and atmospheric pressure, in a muffle
furnace. The acids used in the digestion process are evaporated and the resulting
residue is dissolved with a mild acid. Hydrogen peroxide may be added at the end of
the ashing procedure to improve digestion efficiency. Concentrated H2SO4, HCl and
HNO3, alone or as mixtures, are widely used for the mineralisation of oil samples.
23
When compared to dry-ashing, wet-ashing is relatively quicker, usually taking from 1–
3 h (Kuokkanen et al., 2008). Goncalves, Gonzalez and Murillo (1998), heated 5 g of
an oil sample in concentrated H2SO4 to dryness, after which two digestion
temperatures of 250 oC for 1 h and 600 oC for 4 h were applied. The final residue was
then dissolved with concentrated HCl. Further, they proposed a modified wetdigestion method, in which silica gel and calcium chloride were used as auxiliary
reagents.
2.7.3.5
Microwave-assisted acid digestion
Microwave-assisted acid digestion is carried out in closed and pressurised vessels
employing various acids or acid mixtures. It is an efficient digestion technique for a
variety of industrial and biological samples (Jankowski et al., 2001). When compared
to classical digestion methods, microwave digestion offers better recoveries of volatile
species, minimises the organic load and risk of sample contamination, gives faster
sample decomposition, better reproducibility (Kuokkanen et al., 2008), minimal
volume of reagents, better working environment and a very high digestion
temperature for shorter digestion times at relatively low pressures (Agazzi & Pirola,
2000). Using closed microwave vessels, HNO3 digestion at elevated temperature and
pressure, can rapidly decompose the oil matrix. However, it produces a significant
amount of pressure in the vessels due to the formation of digestion by-products, CO2,
NO, NO2 and H2O. This limits the sample size that can be digested (Essling et al.,
1995). Aucelio, Silveira and Souza (2004) used microwave digestion, employing
HNO3 and H2O2, to determine refractory metals in oils. A two-stage microwave
programme was suggested to digest 0.2 g of oil, in which pressure was released
between digestion stages. The digestion process lasted 55 min, excluding the 30 min
cooling time. Essling et al. (1995) proposed microwave acid digestion involving HNO 3
to digest 0.5 g of waste oil in two digestion stages. The digestion time was rather
long, 120 min plus additional cooling time.
24
2.7.3.6
Ultrasound-assisted extraction
Ultrasound energy is an alternative means of analyte extraction from biological and
environmental samples. Probe and ultrasonic bath are the most common ultrasonic
devices for sonochemistry. Variables of this technique that affect cavitation bubbles
and hence the analyte extractions include: the frequency and intensity of ultrasound,
type of solvent, bubbled gas and external temperature and pressure. The ultrasonic
probe appears to be the choice of ultrasonic device as it offers the best extraction
efficiencies, in less time and with less acid. An ultrasonic bath is the most widely
available and cheapest source of ultrasonic irradiation. The volume of water bath,
sample position, percentage of detergent in the bath, and the bath temperature, are
some of the most important variables. The major limitations of an ultrasonic bath is
that ultrasound waves have to cross the sample vessel, which results in less
cavitation efficiency and hence less analyte extraction. The ultrasonic electromagnetic
transducer, which is often situated in the centre of the bath, also causes variation of
cavitational efficiency within locations of the bath (Capelo, Maduro & Vilhena, 2005).
2.7.4
Optimisation of ICP-OES operating parameters
2.7.4.1
Introduction
A good analytical method has a high level of sensitivity, low detection limit and high
accuracy and precision. However, the procedure of finding the optimum parameters
becomes complicated, particularly when more than two parameters need to be
optimised or if they are interdependent (Galley, Hieftje & Horner, 1995).
Optimisation of the ICP-OES is more crucial, due to the availability of different types
of torches and sample introduction devices (Galley, Hieftje & Horner, 1995). The ICP
operating parameters affecting the performance of plasma conditions that may need
optimisation include rf power, gas flow rates (coolant, plasma and injector),
observation height and solution pump rate or sample flow rate (Cave, Ebdon &
25
Mowthorpe, 1980; Salin & Sartoros, 1997). The ICP is a complex emission source
and the operating parameters are clearly inter-related. For example: observation
height and injector gas flow rate, as well as power and plasma or coolant gas flow
rates, are inter-related. This inter-relationship among the parameters makes the use
of univariate optimisation impossible. Instead, multivariate approaches, such as
factorial design, can be applied but are more complicated, tedious and timeconsuming (Cave, Ebdon & Mowthorpe, 1980; Carpenter & Ebdon, 1986).
Optimisation techniques that have been used with ICPs are the simplex and DavidonFletcher-Powell algorithms (Salin & Sartoros, 1997). Simplex optimisation is the most
popular approach for true optimisation of both argon and nitrogen cooled ICPs, as it
permits a number of inter-related variable parameters to be optimised with relative
ease and speed (Carpenter & Ebdon, 1986).
2.7.4.2
The optimisation scheme
The response functions used with the simplex approach to determine the optimum
instrument operating conditions include: SBR, SNR, precision and accuracy. The
SBR is easily obtained, requires a small number of replicate measurements and is a
good figure of merit that is correlated to a better detection limit. It can be calculated
using equation 2.1 (Salin & Sartoros, 1997).
2.8
ANALYTICAL LIMITS OF DETECTION
For trace elements and environmental analysis, proper judgment, validation and
selection of a procedure, or instrument, is an important issue to obtain information
about the lower limits at which analytes can be detected with confidence. For this
reason, a number of terms and concepts are available, including the limit of detection
and limit of quantitation (FAO Corporate Document Repository, 1998).
26
2.8.1
Limit of detection
The limit of detection (LOD) is the minimum concentration of an analyte that can be
distinguished, with 99% confidence, to be significantly different from the blank. It is
impacted by the analytical method, sample matrix and type of analyte (Ripp, 1996).
One of the most common approaches to establish the LOD is to measure a suitable
number of replicates of the blank solution and apply equation 2.7 (FAO Corporate
document repository, 1998; Aucelio, Silveira & Souza, 2004).
LOD = 3
(Eq. 2.7)
where, Sb is the standard deviation of 10 or 20 replicate measurements of the blank
solution and m is the sensitivity of the calibration curve.
2.8.2
Limit of quantitation
Limit of quantitation (LOQ) is the level above which quantitative responses are
possible with a specified degree of confidence. Like the LOD, LOQ is also affected by
the analytical method, sample matrix and the type of analyte (Ripp, 1996). It is
computed using equation 2.8 from the data obtained using the blank analysis.
LOQ = 10
(Eq. 2.8)
where, Sb is the standard deviation of 10 or 20 replicate measurements of the blank
solution and m is the sensitivity of the calibration curve.
27
CHAPTER 3
EXPERIMENTAL
3.1
INTRODUCTION
In this chapter detailed description of the instruments, reagents and standards used
throughout the study are given. In addition, sample preparation procedures and
optimisation schemes are also briefly described.
3.2
REAGENTS AND STANDARDS
High purity doubly-deionised water, obtained using a Millipore Rios 5® reverse
osmosis and a Millipore Milli-Q Academic® deioniser system (resistivity 18.2 MΩ cm,
Millipore, Bedford, MA) was used throughout this project. High purity, 65% HNO3,
30% H2O2 (m/m), 32% HCl, 98.08% H2SO4 and xylene (all AR grade, from SMM
Instruments Pty. Ltd., Johannesburg) were used to prepare all solutions. A non-ionic
surfactant (Triton X-100), obtained from Merck (Darmstadt, Germany), was used for
detergent formation. Aqueous standard solutions were prepared from 1000 µg g-1
stock solution, containing Ag, Ba, Cu, Mn and Ni in 5% HNO3 (Teknolab AB, Kolbotn,
Norway). A organo-metallic standard (Conostan S-21, Conoco Specialty Products
Inc., Ponca City, OK, USA), with analyte concentration of 500 µg g-1 in oil, was used
for the xylene dilution method. EnviroMAT used oil certified reference material, HU-1,
(SCP SCIENCE, Quebec, Canada), was used to validate the analytical methods
applied to quantify trace metals in used lubricating oils. All glassware was washed
with detergent and water. After being rinsed with deionised water three times, they
were soaked in 10% HNO3 (v/v) solution for 24 h. The glassware was then rinsed
three times with doubly-deionised water and oven-dried before use. Six used
lubricating oil samples (sample-A, sample-B, 15W, 20W, EPO and EPT) were
collected from a local garage. The first two are unknown grade used oils, the next two
28
(15W40 and 20W50) multipurpose engine oils and the latter two (EP 90) are multipurpose gear oils.
3.3
APPARATUS AND INSTRUMENTATION
3.3.1
Apparatus for microwave digestion
A MARS 5 microwave digestion system (CEM Corporation, USA) was employed for
the mineralisation of lubricating oils. At full power, the MARS 5 delivers approximately
1200 W of microwave energy at a magnetron frequency of 2450 MHz. The MARS 5
was equipped with an ESP-1500 plus pressure control and RTP-300 plus
temperature control system. Teflon EasyPreps vessels, allowing a maximum
decomposition pressure of 800 psi at 240 oC were used to digest oil samples. These
vessels do not employ vent nuts or membranes, as in Xp-1500 plus vessels, but
require typical ramping times of 25 min to perform effectively.
3.3.2
Apparatus for ultrasonic-assisted extraction
An ultrasonic bath (Ultrasons, J.P. Selecta, Barcelona, Spain) was used for the
ultrasonic-assisted extraction and agitation of oil samples. Ultrasonic bath
parameters, like the intensity of the ultrasonic energy and temperature, which were
supposed to play a great role in sample sonication, were monitored automatically by
company default. Volumetric flasks of 100 mL were used as reaction vessels for the
extraction procedure. Burette supports were used to place the reaction vessels in a
selected location in the bath.
3.3.3
Setup for ICP-OES instrumentation
Analysis of used lubricating oils for trace metals was carried out using a simultaneous
SPECTRO ARCOS ICP-OES with radial plasma viewing (Spectro Instruments, Kleve,
Germany). The ICP-OES utilises 32 linear CCD detectors in a Paschen-Runge mount
29
for the simultaneous recording of wavelengths between 130 and 770 nm. The sample
introduction system was a Scott type double-pass spray chamber and a cross flow
nebuliser. A peristaltic pump with Solvent-flex1 PVC tubing was used to feed the
nebulisation system with sample solution. A fixed one-piece ICP torch with 1.8 mm
internal diameter injector tube was applied. The argon plasma was sustained by an
air-cooled “free-running” rf generator, at 27.12 MHz with a power output of 0.7 to 1.7
kW. The ICP operating parameters, pumping speed, integration time, rf power, torch
position and gas flows (coolant, auxiliary and plasma flow) were monitored by Spectro
Smart Analyser Vision Software (Spectro Instruments, Kleve, Germany).
3.4
SAMPLE PREPARATION FOR ICP-OES ANALYSIS
Sample introduction for ICP-OES analysis requires the sample to be a homogenous
liquid that can be aspirated easily using a nebuliser. The primary purpose of the
preparation was to minimise the oil’s organic load, viscosity and suspended solid
particles, which affect the plasma stability, reduce ionisation and excitation energy,
increase background and possibly block the nebuliser (Aucelio, Silveira & Souza,
2004).
3.4.1
Xylene dilution
An aliquot of used oil sample (2 g) was accurately weighed and diluted to final mass
of 40 g with xylene. A blank solution was also prepared in the same way, but using
base oil 75. Masses of 0.08, 0.4, 0.8, 1.6 and 2.0 g of organo-metallic stock solution
with analyte concentration of 500 µg g-1 were accurately weighed and made up to 2 g
with base oil 75, and diluted to 40 g with xylene. These solutions formed a series of
calibration standards, containing 1, 5, 10, 20 and 25 µg g-1 of analyte. Check
standard, having analyte concentration of 10 µg g-1, was prepared in a similar manner
as the calibration standards. Spiked organo-metallic standard and a certified
reference material (CRM) were also prepared in a similar manner as the real oil
samples. The oil content was kept at 5% in all types of solutions (blanks, calibration
30
standards, organo-metallic standards, CRM and real oil samples), in order to match
the viscosities.
3.4.2
Oil emulsification
Oil-in-water detergent emulsions were prepared using a specific sequence to
guarantee homogeneity and stability. An aliquot of 0.5 g of used oil was placed in a
25.0 mL volumetric flask. A volume of 0.5 mL concentrated HNO3 was added to
dissolve some metals and decomposes organo-metallic species. The mixture was
then agitated in an ultrasonic bath for 5 min for homogenisation. A mass of 1.0 g
Triton X-100 was added to stabilise the oil-in-water emulsion. The resulting mixture
was then shaken gently, after which Milli-Q water was added under continuous
agitation until a final mass of 10.0 g was obtained. The resulting mixture was finally
shaken vigorously and agitated in an ultrasonic bath for 5 min to homogenise the
contents.
Calibration standards were prepared in a similar manner as the real samples, but
using the required quantity of aqueous stock standard solution. The blank solution
was also prepared in a similar manner as the real samples, but using the reagents
only.
3.4.3
Dry-ashing
An aliquot of used lubricating oil (2 g) was accurately weighed in a porcelain crucible
and heated on a hot plate until completely dry or no more fumes appear. The dried
sample was then exposed to a two stage muffle furnace heating program until
completely ashed: 450 oC for 11 h (overnight) and 550 oC for 1.5 h. The crucibles
were partially opened during heating on a hot plate and completely closed during
muffling, in order to minimise sample contamination. The resulting ash was dissolved
in 2 mL concentrated HNO3 and finally diluted to 40 g with doubly deionised water.
Corresponding blank solutions were prepared in the same manner as real samples,
31
but using blank oil. Calibration standards were prepared from aqueous stock solution
and matrix matched contains the acids used to dissolve the ash.
3.4.4
Wet-ashing
A mass of 2 g used lubricating oil was accurately weighed in a porcelain crucible.
After addition of 2 mL H2SO4, the sample was subjected to heating on a hot plate until
completely dried or no more fumes appeared. The dried sample was then subjected
to a two-stage muffle furnace heating program until completely ashed: 450 oC for 2 h
and 550 oC for another 2 h. The crucibles were partially opened during heating on a
hot plate and completely closed during muffling, aiming to minimise sample
contamination. The resulting ash was dissolved in 1.5 mL of concentrated HNO3, and
diluted to 40 g with doubly deionised water.
For Ba analysis, 2 g of oil sample was accurately weighed into a porcelain crucible.
After addition of 0.5 mL HNO3, the crucible was heated on a hot plate. While the
sample was heating, an additional 1 mL of HNO3 was added, drop-wise, to assist with
complete oxidation of the organic matrix. The oxidant, HNO 3, was introduced dropwise to minimise sample splashing during heating. The dried sample was then
subjected to a two-stage muffle furnace heating program until completely ashed: 450
o
C for 2 h and 550 oC for another 2 h. The resulting ash was dissolved in 1.5 mL of
concentrated HNO3 and finally diluted to 40 g with doubly deionised water.
Corresponding blank solutions were prepared together with the real samples, using
the blank oil and reagents. Calibration standards were prepared from aqueous stock
solution and matrix matched to contain the acids used during the entire wet-ashing
procedure.
3.4.5
Microwave-assisted acid digestion
An aliquot of used oil (0.2 g) was accurately weighed into an EasyPreps vessel liner.
After an addition of 4 mL HNO3, the vessels were assembled on the turntable and
32
subjected to digestion based on the MARS 5 heating program given in Table 3.1. In
each batch, six vessels were used for simultaneous digestion. When the digestion
program was complete, the vessels were allowed to cool and vented correctly. After
venting, the digestate was transferred to a pre-cleaned high-density polypropylene
(HDPP) vessel. It was then diluted to 20 g with doubly deionised water.
Corresponding method blanks were digested in a similar manner as the real samples
using only HNO3. A series of working standards were prepared from a 1000 µg g-1
inorganic stock solution and matrix matched for the oxidant used in real oil samples.
Table 3.1:
Stage
1
3.4.6
Ramp to temperature heating programme of the MARS 5 microwave
Max. Power
Power
Ramp time
Pressure
Temperature
o
Hold time
(W)
(%)
(min)
(psi)
( C)
(min)
1600
100
30
800
200
10
Ultrasonic-assisted extraction
A mass of 2 g used lubricating oil was accurately weighed and placed in a precleaned volumetric flask. An optimised amount of extractant solution was then added
and the resulting mixture irradiated at the optimum sonication time in an ultrasonic
bath. The volumetric flasks, containing the sample, were kept stationary within the
bath at a position of maximum irradiation. To guarantee complete sample irradiation,
only four samples were sonicated simultaneously. After sonication, the supernatant
liquid was separated from the solid phase using a centrifuge at 2500 rpm for 15 min.
The resulting solution was finally diluted to 40 g with doubly deionised water for
analysis by ICP-OES. Corresponding method blanks were also prepared in the same
manner but using blank base oil as sample. With each series of extraction, quality
control (QC) standards were also prepared by adding a known concentration of
organo-metallic standard or CRM to the blank oil and ensuring the same procedure
as the real samples. Working standards ranging from 1-50 µg g-1 were prepared from
33
a 1000 µg g-1 inorganic stock solution and matrix matched for the extractant solutions
used in real oil samples.
3.5
SELECTION OF ANALYTICAL WAVELENGTHS
A known concentration of sample solution was scanned using each line from the ICPOES menu, aiming to establish a reliable analytical line for each element studied. The
results of this analysis were used to calculate the SBR of the elements at each
wavelength evaluated. Finally, the line with minimum potential spectral interferences
and maximum analytical performance was selected for each element. These served
during the entire ICP-OES sample analysis.
3.6
OPTIMISATION
3.6.1
Simplex optimisation of ICP-OES operating parameters
The forward power, coolant flow, auxiliary flow and nebuliser flow rates were used as
control variables with SBR as response function for the modified simplex optimisation
scheme. Although the torch height is another important ICP control variable to be
considered for optimisation, it was kept constant at 10 mm below the load coil for
aqueous and 0 mm for organic matrices because of limitations with the instrumental
Set up. A signal integration time of 30 s and a pumping rate of 30 rpm were used
throughout the optimisation procedure. All the remaining ICP operating parameters
were held constant. Boundary limits were set for each control variable (Table 3.2)
based on the instrument’s capability. Coordinate of the starting vertex and step-size
for each control variable and sample matrix was set as shown in Tables 3.3 and 3.4.
These served as reference point for the generation of the other vertexes. Table 3.5
lists the setting of the modified simplex algorithm used throughout optimisation. A
known concentration of standard solution was prepared for each sample matrix and
analysed at the simplex generated experimental conditions by ICP-OES. The results
obtained from this analysis were used to calculate the SBR of elements at each
34
vertex used. The simplex was terminated when there was little difference in SBR
values between consecutive experimental trials. That is, when the graph of SBR
values versus trial number converged to some point in the graph.
Table 3.2:
Instrumental limits of control variables for the simplex scheme
Boarder limits for the analysis in:
Control variables
Aqueous matrix
Xylene matrix
Forward rf power (W)
950-1600
1300-1650
Coolant Ar flow (L min-1)
11-20
12-20
Auxiliary Ar flow (L min-1)
0.6-3.0
1.2-3.0
Nebuliser flow (L min-1)
0.5-2.5
0.5-2.5
Table 3.3:
Setting of control variables for the modified simplex in aqueous matrix
Control variables rf power Coolant flow Auxiliary flow Nebuliser flow
(W)
(L min-1)
(L min-1)
(L min-1)
Step size
150
1
0.5
0.4
Reference value
1400
13
1
1
0
0
1
1
Decimals
Table 3.4:
Setting of control variables for the modified simplex in xylene matrix
Control variables RF power Coolant flow Auxiliary flow Nebuliser flow
(W)
(L min-1)
(L min-1)
(L min-1)
Step size
100
2
0.5
0.1
Reference value
1600
16
2.2
0.6
0
0
1
2
Decimals
Table 3.5:
Setting of the modified simplex algorithm
Re-evaluation
α
β-
β+
ϒ
No. of Variables + 3
1
0.5
0.5
2
35
3.6.2
Factorial optimisation of an ultrasonic extraction
Ultrasonic energy, generated from an ultrasonic bath, was thoroughly studied for its
capability to extract analytes in used lubricating oil samples. A full factorial
experimental design of 43 was proposed to study the effects of sonication time,
extractant solution ratio and volume. This is illustrated in Table B.4 of Annexure B.
The extracting solutions selected for this study were 65% HNO3, 30% H2O2 and 32%
HCl in different proportions. The maximum analyte recovery was used as a response
criterion and the three factors, together with the four levels, were used as control
variables for the optimisation scheme. Table 3.6 lists the factors and levels used in
the factorial optimisation design.
Table 3.6:
Factors and levels used in the full factorial experimental design (4 3) of
the ultrasonic extraction scheme
Factors
Sonication time (min)
Acids mixture ratio (HNO3:H2O2:HCl, v/v)
Volume of acid mixture (mL)
Levels
30
60
90
120
2:1:0
1:0:0
1:0:1
1:0:3
3
6
8
10
A known concentration of organo-metallic standard was sonicated in each set of
experimental condition and the resulting supernatant solution analysed by ICP-OES.
The results obtained from this analysis were used to calculate the recovery efficiency
of each experimental condition applied. The mass of sample (2 g), volume of water
bath (2000 mL) and volume of detergent (2%) were kept constant throughout the
optimisation scheme.
36
CHAPTER 4
RESULTS AND DISCUSSION
4.1
INTRODUCTION
In this chapter all the experimental results of the optimisation schemes are reported
as well as the determination of trace elements in used oil and corresponding figures
of merit, i.e., optimum ICP operating parameters, optimum ultrasonic extraction
conditions, level of trace elements in used lubricating oils, quality control (QC) results,
limits of detection and limits of quantitation are briefly reported.
4.2
ELEMENTAL WAVELENGTH SELECTION
Lubricating oil samples were scanned at various analytical lines from the instrument
(ICP-OES) menu in order to test their sensitivity with regard to the oil’s matrix. Based
on the measured analyte’s peak signal and background emissions, the SBR was
calculated for each element at each analytical wavelength, as shown in Table B.3 of
Annexure B. The analytical lines listed in Table 4.1 were finally established based on
the minimum potential spectral interferences (Figures 4.1(a-e)) and maximum SBR
values of the elements investigated. For Ni, seven different wavelengths were
investigated for sensitivity; however, it was not possible to find a line entirely free of
interfering species. Figure 4.1 (e) indicates that there was an interfering species
around the Ni line, at wavelengths between 341.35 and 341.4 nm. In addition, the
background of the Ni line was enhanced (background shift), which subsequently
could reduce the SBR values. Background correction was applied to minimise the
background shift.
37
(a)
(b)
(d)
Figure 4.1:
(c)
(e)
Determination of the most sensitive analytical line for (a) Ag 328.068, (b)
Ba 455.404, (c) Cu 324.754, (d) Mn 257.611, and (e) Ni 341.476
38
Table 4.1:
Analytical lines established for the determination of trace metals
Element
Wavelength (nm)
Ag
328.068
Ba
455.404
Cu
324.754
Mn
257.611
Ni
341.476
4.3
OPTIMISATION
4.3.1
Simplex optimisation of ICP-OES operating parameters
Data was obtained for both sample matrices within the ICP boundary limits, discussed
in Section 3.6.1. The optimum instrumental operating conditions for each element, in
both aqueous and organic matrices, were determined and are presented in Tables
4.2 and 4.3, respectively. According to regression analysis of the results, an increase
of the coolant gas flow rate was generally shown to decrease the forward power,
which consequently increased the SBRs of the elements. Alternatively, increasing the
coolant gas flow rate was observed to cause contraction and elongation of the
plasma, which agreed with the finding of Carpenter and Ebdon (1986). The SBR of
the elements appeared to decrease as the forwarded power increased; i.e.,
increasing the power resulted to an increase in the background greater than that of
the signal, a finding which is supported by Salin and Sartoros (1997). Interactions
between the plasma power and nebuliser flow rate showed a negative correlation,
where as the auxiliary and nebuliser gas flow rates a positive correlation, for most of
the elements investigated. Increasing the auxiliary gas flow rate caused positive
influence to the SBR of the elements studied; i.e., for every point increase in the
auxiliary gas flow rate, the SBR of elements increased by 2145 and 947 points for
organic and aqueous solutions, respectively. Likewise, a simultaneous increase in the
plasma power and nebuliser flow rate resulted in an enhanced SBR values of the
elements studied, i.e., for every point of a simultaneous increase of the plasma power
39
and the nebuliser gas flow rate, the SBR of elements increased by 3.06 and 0.536
points for organic and aqueous solutions, respectively.
Table 4.2:
Simplex optimised plasma conditions in aqueous matrix
Element
Wavelength
(nm)
rf power
(W)
Plasma flow
(L min-1)
Auxiliary flow
(L min-1)
Nebuliser flow
(L min-1)
Ag
328.068
1141
18
1.2
0.7
Ba
455.404
1238
12
0.9
0.9
Cu
324.754
1113
11.7
0.94
1.27
Mn
257.611
1359
13
0.9
0.9
Ni
341.476
1289
15
1.1
0.9
1228
13.9
1.0
0.93
Compromise optimum
Table 4.3:
Element
Simplex optimised plasma conditions in xylene matrix
Wavelength
rf power
Plasma flow
Auxiliary flow
-1
Nebuliser flow
(nm)
(W)
(L min )
(L min )
(L min-1)
Ag
328.068
1473
15
2.4
0.8
Ba
455.404
1532
17
2.07
0.8
Cu
324.754
1586
16
2.28
0.9
Mn
257.611
1549
17
2.42
0.9
Ni
341.476
1560
20
1.89
0.7
1540
17
2.21
0.82
Compromise optimum
-1
For every point of simultaneous increase in coolant and nebuliser flow rates, the
SBRs of elements were shown to increase by 30.5 points. In contrast, a simultaneous
increase in the auxiliary and nebuliser flow rates resulted in decreased SBR values
for the elements. Moreover, increasing the auxiliary gas flow rate was observed to
have a significant effect in stabilising plasma conditions during an analysis of organic
samples by preventing carbon depositions at the injector tube, which agrees with the
finding of Marais (1987). During the optimisation process, an auxiliary gas flow rate
lower than 1.4 L min-1 appeared to cause a significant amount of carbon deposit,
40
which totally blocked the injector tube within minutes. All these findings lead to a
conclusion that the auxiliary gas flow rate was more important for the enhancement of
the SBRs of elements, which is more pronounced for organic analysis. During
aqueous analysis, sample nebulisation at low plasma power (in most cases less than
1300 W) and relatively high nebuliser flow rate (> 0.9 L min -1) appeared to cause
instrumental drift, which also increased the background more than the signal. When
the instrument drifted, the plasma became darker.
Higher values for the forward power, plasma and auxiliary gas flow rates, were
recorded for organic samples (Table 4.3), confirming the finding of Marais (1987).
This is primarily due to the high organic vapour loadings exerted by the organic
matrix, which can absorb rf power of the plasma and thus reducing the excitation
temperature and analytical signal (Chirinos, Fernandez & Franquiz, 1998). Further
observation of the optimised ICP conditions indicated good agreement between the
optimum conditions established for Mn 257.611 line and the compromise conditions.
This phenomenon was previously reported by Marais (1987). Used lubricating oil,
blank and quality control samples were analysed using the optimised ICP operating
conditions and produced LODs in the ng g-1 level for all elements studied.
Progress of the simplex optimisation and the variable sizes of the steps are illustrated
in Figure 4.2 for the determination of Cu in xylene and in Figures 4.3 for Mn in water.
After some experimental trials, the variable size simplex was observed to converge
around the sub-optimal region and spin around this best vertex. The forward power,
plasma, auxiliary and nebuliser flow rates appeared to rotate around the sub-optimal
regions of 1570-1630 W, 15-17 L min-1, 2.0-2.35 L min-1 and 0.8-1.0 L min-1,
respectively (Figure 4.2 (a-d)). These sub-optimal regions clearly indicate the location
of the optimum, which was marked by higher SBR values and denser data points.
Additional simplex optimisation progress of Mn, for aqueous solution, can also found
in Figure C.1 (Annexure C).
41
Cu 324.754
Cu 324.754
SBR
60
60
40
40
SBR
SBR
80
SBR
80
20
20
0
0
14
1520 1540 1560 1580 1600 1620 1640 1660 1680 1700
15
18
Plasma flow rate (L min )
(a)
(b)
Cu 324.754
Cu 324.754
SBR
80
SBR
80
60
60
40
SBR
SBR
17
-1
Forward RF power (W)
40
20
20
0
0
1.4
16
1.6
1.8
2.0
2.2
2.4
-1
Auxiliary gas flow rate (L min )
(c)
2.6
0.5
0.6
0.7
0.8
0.9
1.0
1.1
-1
Nebulizer flow rate (L min )
(d)
Figure 4.2: Progress of the simplex optimisation when applied to (a) forward power,
(b) plasma, (c) auxiliary and (d) nebuliser flow rates, for Cu in xylene
The simplex software provided plots of control variables and response functions
against trial numbers, as illustrated in Figure 4.3. Progress of the optimisation
process could thus be monitored. The modified simplex underwent a number of
expansion and contraction moves within 16 successive experimental trials to reach
the sub-optimum region (Figure 4.3). Starting from trial number 17, the SBR values
Figure 4.3: Simplex optimisation progress showing trial numbers versus SBRs for Mn in aqueous solution
42
43
were shown to converge around the sub-optimal region and continued to almost
rotate around the same range, for the next 13 trials.
4.3.2
Factorial optimisation of the ultrasonic-assisted extraction for
trace metals in lubricating oils
Optimisation of the factors influencing an ultrasonic extraction of analytes in used oils
was conducted applying the full factorial (43) experimental design, shown in Table B.4
(Annexure B). Details of the factors and levels used in the optimisation scheme are
provided in Section 3.6.2. Sonication of oil samples using the full factorial design
produced supernatant solutions, which could easily be aspirated into the plasma.
Analysis of this supernatant solution by ICP-OES and employing aqueous calibration
standards, gave quantifiable data. These data were used to compute the recovery
efficiency of each experimental condition, as depicted in Figures 4.4 to 4.8. The
maximum analyte recovery was finally used as criteria to establish optimum extraction
conditions, as listed in Table 4.4. These optimum conditions are therefore
recommended for the extraction of analytes in used oils using an ultrasonic bath.
Table 4.4:
Optimum conditions of an ultrasound-assisted extraction
Element
Sonication time
(min)
Acid mixture
(HNO3:H2O2:HCl)
Volume of acid
(mL)
Ag
120
1:0:3
10
Ba
60/120
1:0:0/1:0:3
10
Cu
120
1:0:3
10
Mn
120
1:0:3
10
Ni
120
1:0:3
10
4.3.2.1
Influence of HNO3
Extracting reagents with volumes of 3, 6, 8 and 10 mL, which made 7.5, 15, 20 and
25% (v/m) of the final solution, respectively, were studied for their analyte extracting
efficiency. Used oil samples sonicated with HNO3 alone appeared yellow in colour
44
and gave poor recoveries for Ag (1.7-4.5%) and Ba (15.9-79.9%), as shown in
Figures 4.4 (a-b). However, HNO3 was shown to have enhanced performance for the
extraction of Cu, Mn and Ni, with recoveries ranging from 52.3 to 106.6%, as
illustrated in Figures 4.4 (c-e).
(a)
(b)
(c)
(d)
(e)
Figure 4.4: Influence of HNO3 on the extraction of (a) Ag, (b) Ba, (c) Cu, (d) Mn, and
(e) Ni, in used oils
45
Pearson correlation function was applied for these recoveries to study the effect of
HNO3. Based on the calculated Pearson correlation coefficient (r) values, increasing
the volume of HNO3 from 3 to 10 mL, while the sonication time was kept constant at
30 min, produced a relatively small positive effect on the extraction of all the elements
investigated, with the exception of Ni. A simultaneous increase in the volume of HNO3
(3-10 mL) and sonication time (30-90 min) promoted a strong positive interaction for
all the elements studied. However, increasing the sonication time from 90 to 120 min
together with an increase in the volume of HNO3 from 3 to 10 mL caused strong (r = 0.81) and moderate (r = -0.67) negative interactions for the extraction of Cu and Mn,
respectively.
4.3.2.2
Influence of aqua regia
Used oil samples extracted using aqua regia (HNO3:HCl/1:3) resulted in clear
solutions. As a strong oxidising agent, aqua regia gave good analyte recoveries of all
the elements studied, except Ba. These are depicted in Figures 4.5 (a-e). An increase
in volume of aqua regia, while the sonication time was kept constant at 30 min,
strongly promoted the recoveries of Ag and Ba. Simultaneously, it strongly
suppressed the recoveries of Cu and Mn. When the sonication time was increased to
60 min, the efficiency for extraction of Ag (53.9-106%), Ba (20.5-31.4%) and Mn
(95.9-103%) was substantially increased with an increase in aqua regia volume from
3 to 10 mL. In general, a simultaneous increase in the volume of aqua regia and
sonication time resulted in improved extraction efficiency in all the elements studied,
except for Ni.
The recovery of Ag was found to increase approximately by 50% with an increase in
the volume of aqua regia from 3 to 10 mL. The increase in the volume of aqua regia
possibly introduced more chloride ions into the solution, which might contribute to the
formation of silver anionic complex (AgClx1-x), in solution, rather than the precipitation
of AgCl. For this context, concentrations of Ag ranging from 0 to 10 µg g-1 in 25%
46
aqua regia (v/m) formed a clear solution and hence resulted in a good linearity (r2 =
0.99993) calibration curve, as depicted in Figure 4.6. This implied that
(a)
(c)
(b)
(d)
(e)
Figure 4.5: Influence of aqua regia on the extraction of (a) Ag, (b) Ba, (c) Cu, (d)
Mn, and (e) Ni, in used oils
47
Ag 328.068
Ag 328.068
120000
50000
40000
80000
y = 2316.8x + 13245.6
2
r = 0.93116
60000
Signal (cps)
Signal (cps)
100000
40000
30000
20000
10000
0
0
0
10
20
30
40
y = 4966.15x-298.01
2
r = 0.99993
20000
50
0
2
-1
100000
100000
80000
80000
2
y = -647.6+5480.8x-53.4x -0.22x
2
r = 0.99976
Signal (cps)
Signal (cps)
120000
3
60000
20000
0
0
20
30
40
2
0
50
3
y =2.13+4392.4x+128.4x -8.2x +0.095x
2
r = 1.0000
40000
20000
10
10
Ag 328.068
Ag 328.068
0
8
Concentration (µg g )
120000
40000
6
-1
Concentration (µg g )
60000
4
10
20
30
40
4
50
-1
-1
Concentration (µg g )
Concentration (µg g )
Ag 328.068
120000
Signal (cps)
100000
80000
60000
2
3
y =2.13+4392.4x+128.4x -8.2x +0.095x
2
r = 1.0000
40000
4
20000
0
0
10
20
30
40
50
-1
Concentration (µg g )
Figure 4.6: Calibration curves constructed using 0-50 µg g-1 of Ag in 25% aqua regia
(m/v)
48
a reasonably low concentration of Ag is unlikely to form AgCl residue if kept in excess
aqua regia (chloride), a finding which is supported by Gaines (2004). In contrast,
calibration standards containing 25 and 50 µg g-1 of Ag in 25% aqua regia (v/m) were
observed to form more precipitate of AgCl, and so significantly deviated from the
certified values (r2 = 0.93116). The linear calibration curve, illustrated in Figure 4.6,
clearly indicated the degree of deviation from a straight line or loss of linearity when
high concentrations of Ag (25 and 50 µg g-1 of Ag in 25% aqua regia (v/m)) were
used.
4.3.2.3
Influence of HNO3:H2O2 (2:1)
Used oil samples extracted by HNO3:H2O2 (2:1, v/v) appeared pale yellow in colour.
The difficulties encountered with this reagent was that more pressure in the reaction
vessels was formed, which caused an explosion if the pressure in the flask was not
frequently released, particularly during the first 30 min of sonication. The use of
HNO3:H2O2 as extracting solution had similar effect on the recoveries of the elements
studied, when compared with HNO3 alone. This extracting reagent yielded very poor
recovery values for Ag and Ba, ranging from 0.7 to 73.2%, as illustrated in Figures 4.7
(a-b). However, HNO3:H2O2 (2:1) gave good recoveries of Cu, Mn and Ni, ranging
from 74.6 to 105%, as depicted in Figures 4.7(c-e). Increasing the volume of
HNO3:H2O2/2:1 from 3 to10 mL while the sonication time was kept constant at 30 min,
produced a small positive effect (r = 0.02 to 0.33) for the extraction efficiency for all
the elements studied, except for Ag, which was negatively affected (r = -0.22).
Increasing the sonication time to 60 min, together with an increase in the volume of
reagent, moderately enhanced (r = 0.49 to 0.74) the extraction efficiency for Ba, Cu
and Mn and caused a low negative influence (r = -0.015 to -0.12) for Ag and Ni.
However, at constant sonication time of 90 min, increasing the volume of reagent
from 3 to 10 mL moderately enhanced (r = 0.36 to 0.77) the extraction efficiency of all
the elements investigated.
49
(a)
(c)
(b)
(d)
(e)
Figure 4.7: Influence of HNO3:H2O2 (2:1) on the extraction of (a) Ag, (b) Ba (c) Cu,
(d) Mn and (e) Ni, in used oils
50
4.3.2.4
Influence of HNO3:HCl (1:1)
Oil samples extracted using HNO3:HCl (1:1) were pale yellow in colour. This
extracting solution gave good recovery for Cu, Mn, and Ni, as depicted in Figure 4.8.
(a)
(c)
(b)
(d)
(e)
Figure 4.8: Influence of HNO3:HCl (1:1) on the extraction of (a) Ag, (b) Ba (c) Cu, (d)
Mn and (e) Ni, in used oils
51
The recoveries of Ag and Ba (Figures 4.8(a-b)) were not satisfactory, but had
improved when compared to the recoveries obtained using HNO3 alone or
HNO3:H2O2 (2:1). At constant sonication time (30 min) but increasing the volume of
the reagent (3 to10 mL), caused a strong negative effect for the extraction efficency
for Cu, Mn and Ni (r = -0.74), and a moderate negative effect for Ag and Ba (r = 0.50). However, increasing the sonication time to 60 min, together with an increase in
the volume of HNO3:HCl (1:1), produced a strong positive association for Ag, Ba and
Mn (r = 0.73-0.95) and moderate for Cu and Ni (r = 0.67). A strong positive interaction
(r = 0.82-0.88) was also observed for all the elements studied, except Ni (r = -0.48),
when the volume of this reagent was increased, while the sonication time was kept
constant at 90 min. The recoveries of Ag, Cu, Mn and Ni generally increased with an
increase in the volume of HNO3:HCl (1:1) from 3 to 10 mL.
4.3.2.5
Influence of sonication time
The effect of sonication time on the extraction of analytes in used lubricating oils was
studied at 30, 60, 90 and 120 min intervals. Increasing the sonication time was shown
to increase the temperature of the bath, and hence that of the sample solution. The
ultrasonic extraction efficiency of the elements studied, except Ni, was enhanced with
an increase in sonication time together with an increase in aqua regia volume. When
using HNO3 as extractant, increasing the sonication time was generally observed to
depress the recoveries of Ba and Ni, but enhanced the recovery of Ag. Increasing the
sonication time, for analyte extraction with HNO3:H2O2 (2:1), generally increased the
recoveries of Ag, Cu and Mn. When using low volumes of the extracting reagents, 3
to 6 mL, increasing the sonication time was found to weakly influence the extraction
efficiency of the elements investigated. Similarly, increasing the sonication time,
employing low volumes (3 to 8 mL) of aqua regia and HNO3:HCl (1:1) as extracting
reagents, depressed the recovery of Ag. A simultaneous increase in sonication time
and volume of extracting reagent was generally found to enhance the extracting
efficiency of the elements studied. This was critical when aqua regia and HNO3:HCl
(1:1) were used as extracting reagents.
52
4.4
ANALYSIS OF USED LUBRICATING OILS BY ICP-OES
4.4.1
Introduction
The methodology for the analysis of lubricating oils by ICP-OES involved simultaneous
preparation of samples and calibration standards, choosing suitable sample
introduction assemblies, establishing analytical wavelengths, selecting or optimising
instrument operating parameters, constructing calibration curves, correcting for
background emission and finally the analysis of real samples.
4.4.2
Linearity of calibration curves
Organo-metallic working standards were used to construct calibration curves for the
oil dilution, whereas aqueous standards were used for the remaining five methods.
Using both organo-metallic and aqueous standards, it was possible to establish a
linear calibration curves (r2 = 0.9993-1.0000) for all elements studied. A typical
calibration curve of Cu, in xylene matrix, is depicted in Figure 4.9, which is
representative of all the other elemental calibration curves for each method.
Signal intensity (cps)
Cu 324.754
2000000
1500000
y = 74483.7x - 7134.4
2
r = 0.99996
1000000
500000
0
0
5
10
15
20
25
-1
Concentration (µg g )
Figure 4.9: Calibration curve for the determination of Cu (324.754) in xylene
53
Additional calibration curves can be found in Annexure C (Figures C.2-C.7). Linear
regression equations obtained for the elements of interest in each method studied are
illustrated in Table 4.5.
Table 4.5:
Linear regression equations obtained by all methods studied
Element
λ (nm)
Ag
328.068
Ba
455.404
Cu
324.754
Mn
257.611
Ni
341.476
4.4.3
Method
Dilution
Microwave
Emulsion
Dry-ashing
Wet-ashing
US extraction
Dilution
Microwave
Emulsion
Dry-ashing
Wet-ashing
US extraction
Dilution
Microwave
Emulsion
Dry-ashing
Wet-ashing
US extraction
Dilution
Microwave
Emulsion
Dry-ashing
Wet-ashing
US extraction
Dilution
Microwave
Emulsion
Dry-ashing
Wet-ashing
US extraction
Equation
y = 43597.7x + 4151.92
y = 6663.4x - 1272.2
y = 15043.75x + 807.64
y = 14286.5x - 791.92
y = 7331.8x + 832.7
y = 20388.99x + 3178.58
y = 94142.2x - 30597.5
y = 25430.6x - 6774.9
y = 38213.75x + 78.4
y = 37315.3x - 4082.3
y = 61705.06x + 46291.7
y = 61705.1x + 46291.7
y = 74483.7x - 7134.5
y = 18012.98x + 387.38
y = 29900.5x + 2331.8
y = 23618.9x + 2430.07
y = 22198.96x + 4965.4
y = 26004.63x + 6368.87
y = 92190.4x - 31947.8
y = 12523.96x - 1265.18
y = 43500.58x + 9470.22
y = 25826x - 2157.8
y = 7548.85x + 1097.4
y = 32632.9x + 133.96
y = 14107.6x + 1696.1
y = 2202.7x - 662.51
y = 2184.63x + 111.43
y = 2495.67x - 532.93
y = 3087.66x + 156.63
y = 3087.66x + 156.63
r2
0.99903
0.99995
0.99996
0.99998
0.99994
0.99999
0.99992
0.99987
0.9999
0.99997
0.99973
0.99973
0.99996
0.99999
0.99998
1.00000
0.99999
0.99999
0.99962
0.99996
0.99995
0.99998
0.99997
0.99990
0.99955
0.99984
0.99996
0.99996
0.99999
0.99999
Determination of limits of detection and quantitation
The data obtained from 10 consecutive nebulisations of the blank solution were used
to calculate the limits of detection and quantitation applying equations 2.7 and 2.8,
respectively. The limits of quantitation obtained using the six methods (Table 4.6)
54
were in the range of 3.01 to 117 ng g-1, which permits trace levels of wear metals to be
quantified and a trend to be observed, even at low levels.
Table 4.6:
Radial ICP-OES limits of detection and quantitation of the trace
elements investigated (n=10)
Parameter
(ng g-1)
Ag
(328.068)
Ba
(455.404)
Cu
(324.754)
Mn
(257.611)
Ni
(341.476)
Xylene dilution
LOD
LOQ
5.38
17.9
8.24
27.5
2.97
9.91
2.20
7.32
26.0
86.8
Microwave
digestion
LOD
LOQ
1.92
6.38
1.07
3.58
6.92
23.1
1.16
3.85
18.0
60.2
Emulsion
LOD
LOQ
4.79
16.0
4.06
13.5
4.36
14.5
1.18
3.92
20.6
68.7
Dry ashing
LOD
LOQ
1.79
5.96
0.904
3.01
12.7
42.3
1.36
4.55
5.60
18.7
Wet ashing
LOD
LOQ
4.50
15.0
2.87
9.56
8.14
27.1
3.32
11.1
15.4
51.4
Ultrasonic
extraction
LOD
LOQ
4.13
13.8
2.36
7.87
35.1
117
2.64
8.79
22.9
76.2
*Literature
LOD
7.00
2.00
60.0
1.4
20.0
Method
*Estimated LOD for radial ICP-OES analysis (Inorganic Ventures, 2012)
The LODs obtained in this study were compared with the estimates of detection limits
proposed by Inorganic Ventures (2012). Based on these comparisons, all the six
methods investigated gave better LODs for Ag and Cu. The microwave and dryashing methods offered much better LODs (0.904-18.0 ng g-1) for all the elements
studied. The wet-ashing method for determining Ba and Mn; and the ultrasonic
extraction for Mn and Ni, gave slightly higher LODs. However, they gave better LODs
for the other three elements. The emulsification method for Ba and xylene dilution for
Ba, Mn and Ni, yielded slightly higher LODs.
55
4.4.4
Determination of trace elements in used oil samples
Analysis of used oil samples prepared using each of the six methods studied (xylene
dilution, microwave digestion, oil emulsification, ultrasonic extraction, dry and wetashing) yielded the results depicted in Figures 4.10 to 4 15. Based on these results,
both unknown grade used oil samples (samples A and B), exhibited high
concentrations of Cu and Mn (Figures 4.10 and 4.11).
Figure 4.10: Mean concentrations of trace elements, in sample A of used lubricating
oil, analysed by ICP-OES
Figure 4.11: Mean concentrations of trace elements, in sample B of used lubricating
oil, analysed by ICP-OES
56
The concentration of Cu, ranged from 3.3 to 6.1 µg g-1, found in the 15W-40 engine
oil prepared by microwave digestion, wet-ashing and ultrasonic extraction methods
(Figure 4.12). Analysis of 20W-50 used engine oil, prepared using the six methods
studied, had higher concentration of Mn, ranging from 96.5 to 203 µg g-1 (Figure
4.13). Analysis of used gear oil (EP 90), obtained from a TOYOTA COROLA motor
car, had concentrations of Cu ranging from 7.66 to 20.3 µg g-1 (Figure 4.14). Similarly,
analysis of used gear oil (EP 90), obtained from a OPEL CORSA motor car, had high
concentration of Ba (8.05-53.6 µg g-1) and Cu (20.5-67.2 µg g-1) (Figure 4.15).
Figure 4.12: Mean concentrations of trace elements, in used engine oil (15W-40),
analysed by ICP-OES
In all analysed samples, high concentration of Cu and Mn were found. According to
AGAT laboratories (2004), a high level of Cu could result from wearing of bearings,
bushings, oil coolers, radiators, camshaft thrust washers, gears, valves, clutch plate
and sealing compounds. The significant level of Mn might have also been due to wear
of cylinder liner, blower, exhaust and intake systems, shafts and valves. “Leak” of
unleaded gasoline could also increase Mn level in oils (AGAT laboratories, 2004).
Since Mn is introduced as a combustion improver and smoke suppressor in residual
and distillate oils, a “leak” of this component into the oil system could increase Mn
concentration (Chausseau, Lebouil-Arlettaz & Thomas, 2010).
57
Figure 4.13: Mean concentrations of trace elements, in used engine oil (20W-50),
analysed by ICP-OES
Figure 4.14: Mean concentrations of trace elements in used gear oil (EP 90) working
for about 12,000 km in a TOYOTA COROLA motor car
58
Figure 4.15: Mean concentrations of trace elements in used gear oil (EP 90), working
for more than one year in a OPEL CORSA motor car
The results obtained using the six methods (Figures 4.10 to 4.15) were statistically
compared with each other using the analysis of variance (ANOVA), in order to
determine any significant differences between the means. The overall ANOVA result,
obtained from comparison of treatments (methods) against variables (elements), is
illustrated in Table 4.7. From the ANOVA table of Cu (Table 4.8), it can be observed
that the sum of squares between groups (530.67) is very close to the total sum of
squares (542.33), which indicate mean variation between the methods used for Cu
determination. Likewise, the high F-value (109.25) or low p-value (0.000) also
indicates statistically significant differences in mean between the methods studied
when compared at the 95% confidence level. Table 4.9 illustrates comparison of Cu
results, by treatments. Based on this table, comparison between microwave digestion
and dry-ashing as well as xylene dilution and oil emulsification, for Cu results, gave pvalues of 0.083 and 0.011, respectively. In general, more than 65% compared
combinations of any two methods gave statistically similar concentrations for Ag, Ba
and Mn at the 95% confidence level. Comparisons between microwave digestion
versus dry ashing, microwave digestion versus ultrasonic extraction, microwave
digestion versus emulsion and dry-ashing versus ultrasonic extraction gave
59
statistically similar concentrations of Cu at the 95% confidence level. Likewise, xylene
dilution versus oil emulsion and wet-ashing versus ultrasonic extraction yielded
statistically similar concentrations of Ni at the 95% confidence level. These ANOVA
results for Cu also can represent the other four elements studied. Collectively, the
concentrations of trace elements obtained using the six methods showed significant
differences at the 95% confidence level, as illustrated in Table 4.7.
Table 4.7:
Statistical comparisons of the analytical methods studied using ANOVA
(α = 0.05)
P values
No. Methods compared
Ag
Ba
Cu
Mn
Ni
1
Microwave vs dilution
0.000
0.000
0.001
0.000
0.000
2
Microwave vs emulsion
0.000
0.000
1.000
0.000
0.000
3
Microwave vs dry-ashing
0.000
0.000
0.083
0.000
0.000
4
Microwave vs wet ashing
0.000
0.000
0.000
0.000
0.000
5
Microwave vs us extraction
0.000
0.000
0.472
0.000
0.000
6
Dilution vs emulsion
1.000
1.000
0.011
1.000
1.000
7
Dilution vs dry-ashing
0.148
0.158
0.000
1.000
0.000
8
Dilution vs wet-ashing
1.000
0.702
0.000
1.000
0.000
9
Dilution vs us extraction
1.000
0.480
0.000
1.000
0.000
10
Emulsion vs dry-ashing
0.375
1.000
0.005
1.000
0.000
11
Emulsion vs wet-ashing
1.000
1.000
0.000
1.000
0.000
12
Emulsion vs us extraction
1.000
1.000
0.023
1.000
0.000
13
Dry-ashing vs wet-ashing
0.589
1.000
0.000
1.000
0.000
14
Dry-ashing vs us extraction
1.000
1.000
1.000
1.000
0.000
15
Wet-ashing vs us extraction
1.000
1.000
0.000
1.000
1.000
16
All
0.000
0.000
0.000
0.000
0.000
*p values > 0.05 (in bold captions) indicate statistical similarity, whereas P values < 0.05
indicate significant differences at the 95% confidence level.
60
Table 4.8:
Analysis of variance for Cu results
Source
SS
df
Between groups 530.673084 5
MS
F
Prob > F
106.134617
109.25
0.0000
Within groups
11.6579972 12 0.971499767
Total
542.331081 17 31.9018283
Table 4.9:
Row meanCol mean
Emulsion
Comparison of Cu by treatments (methods)
Dilution
Emulsion
Microwave
Dry-ashing
Wet-ashing
3.61573
0.011
Microwave
4.93138
1.31564
0.001
Dry-ashing
7.64891
1.000
4.03318
0.000
Wet-ashing
17.5839
0.005
13.9682
0.000
Ultrasonic
extraction
6.89097
0.000
3.27524
0.000
0.023
2.71753
0.083
12.6526
0.000
1.95959
0.472
9.93504
0.000
-0.757939
1.000
-10.693
0.000
4.5
ANALYSIS OF QUALITY CONTROL SAMPLES BY ICP-OES
4.5.1
Introduction
Analytical methods developed for sample analysis should be validated to ensure
reproducibility of the data. In this study, six analytical methods were investigated for
their ability to determine trace elements in used lubricating oils. The accuracy of these
developed methods was checked through the analysis of EnviroMAT (HU-1), used oil
certified reference material (CRM). Organo-metallic standard, added to the base oil,
was also analysed and served as a comparative approach to the CRM.
61
4.5.2
Accuracy and precision
Appropriate spikes of both organo-metallic standard (Conostan S-21) and certified
reference material (CRM) of used oil, which employed external calibrations, were
analysed by ICP-OES. The quantitative data obtained was used to evaluate the
accuracy and precision of the methods studied. The recoveries were computed,
based on the measured and consensus values of the standards. Statistical analysis of
these data using the Student's t-test indicated good agreement between the certified
and measured values at the 95% confidence level, for most of the elements studied.
All the methods investigated gave comparable results for most of the elements
studied, as illustrated in Tables 4.10 to 4.11.
Table 4.10: Mean recoveries (±SD), obtained from the analysis of Conostan S-21
standard, by ICP-OES (n=3)
Ag
(µg g-1)
Ba
(µg g-1)
Cu
(µg g-1)
Dilution
Spiked
Obtained
%Recovered
500
497±11
99.3±2.2
500
486±17
97.2±3.5
500
494±7.1
98.8±1.4
Emulsion
Spiked
Obtained
%Recovered
500
432±5.8
86.4±1.3
500
500
456±11
462±2.3
91.1±2.4 92.3±0.62
Method
Description
Mn
(µg g-1)
Ni
(µg g-1)
500
500
489±4.2
498±2.9
97.8±0.84 99.7±0.59
500
497±11
99.3±2.3
500
471±2.9
94.0±0.75
Spiked
500
500
Obtained
27.1±1.3 447±8.7
%Recovered 5.41±0.30 89.4±1.7
Spiked
500
500
Obtained
245±23
457±22
%Recovered 48.9±5.0 91.4±5.2
500
464±8.9
92.8±1.8
500
422±23
84.4±5.5
500
521±8.2
104±1.6
500
411±21
82.2±5.2
500
460±7.6
92.0±1.5
500
185±27
36.9±5.8
Wet-ashing
Spiked
Obtained
%Recovered
500
382±21
76.3±4.3
500
476±30
95.1±6.1
500
433±18
86.5±3.6
500
499±1.4
99.7±0.27
500
100±6.2
20.0±1.2
Ultrasonic
extraction
Spiked
Obtained
%Recovered
500
506±11
101±2.3
500
241±45
48.1±9.1
500
471±7.1
94.2±1.4
500
501±7.1
100±1.3
500
465±9.3
93.1±1.8
Microwave
Dry-ashing
62
All the methods studied, except microwave digestion and dry-ashing, generated good
recoveries of Ag for spiked Conostan standards. On the other hand, all methods
produced poor recoveries of Ag from the CRM. In all cases, the precision of Ag was
generally good, with %RSDs ranging from 1.53 to 5.62%, except for dry-ashing
method (16.2%). The variation in accuracy could be because of the low level of Ag in
the CRM, and to some extent, the possible risk of chloride contamination, which
probably had reacted with the limited spike level to form insoluble AgCl. All studied
methods, except oil emulsification (48.3% accuracy), were found to be convenient for
the determination of Ba in CRM, with good accuracy. Even so, oil emulsification,
employing spikes of Conostan standards, produced good recovery for Ba (91.1%).
This variation in accuracy primarily occurred due to the low level of Ba in the CRM
samples. In addition, the relatively high viscosity of the CRM, which caused emulsion
to be stable for short periods, probably prevented the release of Ba from the organic
matrix. Viscosity matching that employed less viscous base oils, was necessary.
However, this was not attempted due to the limited sample mass required to form a
stable emulsion, and the low level of Ag and Ba analytes in the CRM.
Unlike ultrasonic extraction of CRMs, ultrasonic extraction of Conostan standards
yielded a low recovery for Ba (48.1%). To study this variation, five recovery
experiments, containing 5.95, 5.16, 0.463, 0.453 and 0.233 µg g-1 Ba analytes were
performed. The first two additions were from the Conostan S-21 standard and the
latter three from the CRM. Under constant extraction conditions of 10 mL aqua regia
and 120 min sonication time, these additions gave Ba recoveries of 42.9, 48.1, 87.4,
93.2 and 96.3%, respectively. These results are illustrated in Figure 4.16. The above
observations led to a conclusion that under the same extraction conditions, the
recovery of Ba decreased as the spike level increased. This could be due to the
limitation of the extracting agent (aqua regia) to isolate Ba analytes from the organic
matrix as the spiked level exceeded some limit, in this case about 0.45 µg g-1.
63
Table 4.11: Mean recoveries (±SD), obtained from the analysis of EnviroMAT
(certified reference material in used oil), by ICP-OES (n=3)
Method
Description
Certified
Obtained
%Recovered
Certified
Emulsion
Obtained
%Recovered
Certified
Microwave
Obtained
digestion
%Recovered
Certified
Dry-ashing Obtained
%Recovered
Certified
WetObtained
ashing
%Recovered
Certified
Ultrasonic
Obtained
extraction
%Recovered
Xylene
dilution
Ag
(µg g-1)
Ba
(µg g-1)
13.0
2.21±0.13
17.0±1.2
13.0
3.93±0.36
30.2 ± 2.5
13.0
6.55±0.08
50.4 ± 1.4
13.0
2.34±0.15
18.0±2.9
13.0
1.42±0.15
10.9±1.2
13.0
6.29±0.17
48.4±0.7
9.00
9.22±0.85
102±7.2
9.00
4.35±0.25
48.3±2.3
9.00
8.62±0.27
96.2±0.64
9.00
9.20±0.32
102±6.4
9.00
8.56±0.03
95.1±1.0
9.00
8.67±0.47
96.3±2.4
Cu
(µg g-1)
Mn
(µg g-1)
3132
18.00
3312±26
14.5±0.31
80.3±3.2
106±3.0
3132
18.00
3290±191 13.25±0.79
73.6±3.5
105±5.2
3132
18.0
3248±5.2 18.5±0.68
104±3.8
103±0.48
3132
18.0
3173±117
14.8±1.2
81.9±1.7
101±5.9
3132
18.0
3314±67
18.9±0.84
106±1.2
105±4.8
3132
18.0
3064±25
14.3±0.84
79.5±2.2
97.8±2.9
Ni
(µg g-1)
45.0
37.5±0.70
83.3 ±3.3
45.0
43.2±1.6
96.0±3.2
45.0
43.0±1.7
95.6±2.6
45.0
21.8±2.5
48.3±3.8
45.0
42.9±4.2
95.2±9.0
45.0
46.0±4.6
102±13
*a. Recovery values in normal print fall within the confidence limit (CL) adopted (95%).
*b. Recovery values in bold fall within the tolerance interval (TI) adopted (95%).
*c. Recovery values in italics fall outside the TI adopted (95%).
% Recovered
100
96.3%
93.2%
87.4%
80
60
48.1%
42.9%
40
20
0
0.233
0.453
0.463
5.16
5.95
Concentration (µg g-1)
Figure 4.16: Ultrasonic extraction of Ba under constant conditions of 10 mL aqua
regia and 120 min sonication time
64
All the studied methods were found to be useful for the determination of Cu in both
sample spikes (CRM and Conostan S-21). Microwave digestion and wet-ashing
procedures allowed Mn recoveries from the CRM that fell within the confidence
interval adopted (95%). The other four methods yielded Mn recoveries out of the
confidence interval, but within the tolerance interval adopted at 95%. However, all the
studied methods gave enhanced recoveries of Mn, ranging from 82.2 to 104%, when
spikes (additions) of Conostan standard, containing 4.98 to 15.0 µg g-1 of Mn, were
used. The ultrasonic extraction method gave good recovery of Mn (100%) for
additions of Conostan standard, containing 5.18 µg g-1 Mn, but generated
unsatisfactory Mn recovery (79.5%) in the CRM, containing 0.466 µg g-1 of the
analyte. The studied methods, except dry-ashing, generated quantitative recoveries
of Ni for both spiking standards. The dry-ashing approach caused a significant loss of
Ni using both type of additions, which became worse (49.8 to 7.67% accuracy) when
the spike level increased from 1.18 to 10 µg g-1. This loss of Ni was probably caused
by the volatilisation that might have occurred during the intense heating and muffling
processes.
Like dry-ashing, wet-ashing method also produced poor recovery for Ni (38.7%) when
only HNO3 was used as oxidising agent. However, wet-ashing, involving concentrated
H2SO4 as oxidising agent, offered better recovery of Ni that fell within the confidence
limit adopted (95%). Taking the above into consideration, the loss in Ni was primarily
caused by the formation of volatile Ni compounds, which could be retained by
sulphated ashing. This phenomenon is in agreement with the finding of Aucelio et al.
(2007a). Another observation is that the analysis of QC samples, containing around 1
µg g-1 of Ni, ashed using 2 mL of H2SO4, yielded better reproducibility of Ni (95.2%).
However, increasing the concentration of Ni to 5.79 µg g-1, for a given volume of
H2SO4 (2 mL) resulted in an abrupt loss of Ni (a loss of 79%). Though it needs further
investigation, this phenomenon indicates the need for monitoring the quantity of
H2SO4 that can efficiently retain a given concentration of Ni. Sample adsorption onto
the vessel walls of the porcelain, which was unavoidable, could also have played a
65
part in the loss of Ni. All in all, it can be concluded that the dry-ashing approach is
inefficient for the determination of Ni.
When the recovery results obtained using both QC samples were evaluated, all the
methods studied, except dry-ashing, were found to be suitable for the determination
of the elements investigated with a reasonable degree of accuracy. Ultrasonic
extraction and xylene dilution for the determination of Ag, with recovery of 99.3 to
101%, wet-ashing, xylene dilution, microwave digestion and dry-ashing methods for
Ba (92.8-99.6%), xylene dilution, oil emulsification, microwave digestion and
ultrasonic extraction for Cu (96-102%), microwave digestion, wet-ashing, ultrasonic
extraction and xylene dilution for Mn (89.1-104%) and oil emulsification, microwave
digestion, ultrasonic extraction and xylene dilution for Ni (91.5-97.6%), were found to
be the most suitable choices.
Analysis of QC samples in each method was performed in three replicates. These
replicated results were used to calculate the %RSD, which was used to evaluate the
precision of each method investigated. The dilution method for Conostan standards
gave good precision, with %RSDs below 3.6. However, it produced low precise
recoveries for Ag and Ba (7.36 and 7.03 %RSD, respectively) in CRM samples. This
low precision for Ag and Ba occurred most probably due to their low level in the CRM.
The oil emulsification approach generated good precision of results (< 5.00 %RSDs)
for both QC samples, with the exception for Ag (8.40 %RSD) in CRM samples. The
precision of results for both microwave digested QC samples was very good, always
lower than 5.00 %RSDs. In contrast, the dry-ashing method exhibited lower precision
of results (always >5.00 %RSDs) for both QC samples. Reasonably good precision of
results was obtained for both wet-ashed QC samples, except for Ag and Ni (>5.00
%RSDs). Precision of results obtained from ultrasound-assisted extraction method
were generally good (<5.00 %RSDs), with the exception for Ni (12.6 %RSD) in CRM
and Ba (18.9 %RSD) in Conostan standard samples. In general, the determined
precision of results for all methods studied was good (<5.00 %RSDs), except for the
66
dry-ashing technique. The possible reason for the variation in accuracy and precision
of each method is described in Section 4.5.3 below.
4.5.3
Validation criteria applied
The six analytical methods investigated were validated in terms of accuracy,
precision, linearity, LOD and LOQ. Details of the five validation criteria applied is
illustrated in Table 4.12. Calibration curves constructed for the ultrasonic extracted
oils, which involved a series of aqueous standards, gave good linearity with r 2 ranging
from 0.9997-1.000. The calculated LOQs for the ultrasonic extraction method were
between 7.87 and 117 ng g-1, which can allow the determination of trace levels of
wear metals in lubricating oils. Analysis of both QC samples, prepared using the
ultrasound-assisted extraction, generally gave good analyte recoveries of all the
elements studied. The precision of these recoveries was also fairly good, with
%RSDs ranging from 2.26-2.76%, except for Ni (12.6%). The slight variations in
accuracy and precision of these results might have been due to the position of the
sample within the ultrasonic bath; that is, the closer the sample was to the ultrasonic
transducer, the more it was reproducibly extracted. This phenomenon was reported in
the works of Angnes, Munoz & Oliveira (2006). The probability of sample loss and
contamination during filtration and transferring the digestate to and from centrifuge
tubes could also have contributed to variations in accuracy and precision.
The oil dilution method, which employed a series of organo-metallic working
standards for the analysis of oil samples, gave good linearity calibration curves with r 2
ranging from 0.9997 to 0.9998. The LOQs determined for this method ranged from
7.32 to 86.8 ng g-1. These LOQs can allow the determination of trace levels of wear
metals in lubricating oils. Analysis of xylene diluted QC samples yielded satisfactory
recoveries (98.1-102%) for all the elements studied. The precision, as given by
%RSD, of these results was generally good, ranging from 0.591 to 2.85%, except for
Ba, which was 7.03% (Table 4.12).
67
Table 4.12: Analytical criteria applied for method validation
Method
Xylene
dilution
Emulsion
Microwave
digestion
Dry ashing
Wet
ashing
Ultrasonic
extraction
Validation
criteria
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
% Recovery
% RSD
Linearity (r2)
LOD (ng g-1)
LOQ (ng g-1)
Trace elements in lubricating oils
Ag
99.3±2.1
2.16
0.9997
5.38
17.9
86.4±1.3
1.53
0.9999
4.79
16.0
50.4±1.4
2.72
0.9999
1.92
6.39
48.9±5.0
16.2
0.9998
1.79
5.96
76.3±4.3
5.62
0.9999
4.50
15.0
101±2.3
2. 26
0.9997
4.13
13.8
Ba
102±7.2
7.03
0.9998
8.24
27.5
91.1±2.4
2.66
0.9998
4.06
13.5
96.2±0.64
0.669
0.9997
1.07
3.58
102±6.4
6.33
0.9998
0.904
3.01
95.1±1.1
1.10
0.9999
2.87
9.56
96.3±2.4
2.49
0.9997
2.36
7.87
Cu
106±3.0
2.85
0.9998
2.97
9.91
105±5.2
4.99
1.0000
4.36
14.5
104±3.8
3.63
1.0000
6.92
23.1
101 ± 5.9
5.80
1.0000
12.7
42.3
106 ± 1.2
1.15
1.0000
8.14
27.1
97.8 ± 2.9
2.97
1.0000
35.1
117
Mn
97.8±0.84
0.862
0.9996
2.2
7.32
99.3±2.3
2.35
0.9999
1.18
3.92
103±0.48
0.467
0.9999
1.16
3.85
81.9±1.7
2.06
0.9999
1.36
4.55
105±4.8
4.57
0.9997
3.32
11.1
100±1.3
2.76
1.0000
2.64
8.79
Ni
99.7±0.59
0.591
0.9998
26.0
86.8
96.0±3.2
3.30
0.9999
20.6
68.7
95.6±2.6
2.71
0.9997
18.0
60.2
48.3±3.8
7.89
1.0000
5.60
18.7
95.2±9.0
9.47
0.9998
15.4
51.4
102±13
12.6
0.9999
22.9
76.2
The wet-ashing method generated good analyte recoveries for Ba, Cu, Mn and Ni,
ranging from 95.1 to 106%. Even so, the recovery of Ag was found to be
unsatisfactory (76.3%), most probably due to chloride contamination. The precision of
results obtained by the wet-ashing method were generally good (1.10-4.57 %RSD),
with the exception of Ag (5.62%) and Ni (9.47%). The possible reasons for these
slight variations in precision and accuracy may include:
68

The observed inability of the ash to dissolve completely in commonly used
acids (HNO3, HCl, H2SO4, aqua regia and HF). Unlike used oil samples,
EnviroMAT and Conostan standards were found to form insoluble brown
residues, even after dilution. The formation of residue was more pronounced
for additions (spikes) of Conostan standards than CRMs. The residue was
further tested if complete dissolution could be possible using hot or cold
HNO3, HCl, aqua regia and HF, but neither was successful. HF etched the
porcelain crucible without affecting the residue.

Sample adsorption onto the reaction vessel walls was unavoidable and
might have resulted in sample loss.

Sample contamination possibly occurred during open heating procedure.

The formation of precipitates of BaSO4 and AgCl, which occurred when
using H2SO4 (as oxidant) and HCl (as solvent) during the wet-ashing
procedure, might have caused difficulties for sample aspiration.
A series of aqueous working standards for the analysis of wet-ashed oils gave good
linearity calibration curves with r2 between 0.9997 and 1.000. The LOQs obtained by
this method ranged from 9.56 to 51.4 ng g-1, which indicates that the method can be
used for the determination of trace levels of wear metals in lubricating oils.
The dry-ashing method for oil preparation yielded good analyte recoveries for Ba and
Cu that fell within the confidence level adopted at 95%. However, this method yielded
unsatisfactory recoveries of Ni, Ag and Mn (48.3, 48.9 and 81.9%, respectively). The
recovery of Ni, in both QC samples, was totally out of the tolerance interval adopted
of 95% confidence level. This loss of Ni was probably caused by the volatilisation that
might have occurred during the intense heating and muffling processes. Generally,
this method was found to be inefficient for the determination of Ni in used oils. The
precision of these results was poor, with %RSDs ranging from 5.80 to 16.2%, except
for Mn (2.06%). The variation of accuracy and precision in the results could be due to:

The inability of the ash to completely dissolve in commonly used acids, such
as HNO3, HCl, H2SO4, aqua regia and HF.

Sample loss due to sample adsorption onto the crucible walls.
69

The possibility of sample contamination caused by the open air heating
procedure.
A series of aqueous working standards for the analysis of dry-ashed oil samples gave
good linear calibration curves with r2 between 0.9998 and 1.000. The determined
LOQs for this method were between 3.01 and 42.3 ng g-1, which indicates that the
method can be used for the determination of wear metals in lubricating oils.
Oil-in-water or oil emulsification method yielded good analyte recoveries (86.4-99.3%)
for all the elements studied when additions of Conostan standards were used.
However, this method yielded unsatisfactory recoveries for Ag and Ba with the CRM.
This variation in accuracy might have occurred due to the low levels of these
elements in the CRM and the problem of viscosity matching experienced during
experimentation. The limitation with viscosity matching occurred due to the low level
of Ag and Ba analytes in the CRM, as well as the limited sample size that can
efficiently form a homogenous and stable emulsion. For this reason, viscous oils
experienced short term emulsion stability, whereas thin oils formed emulsions that
were stable for more than 2.5 h. The precision of results determined for this method
were fairly good, with %RSDs ranging from 1.53 to 4.99%. Working standards for the
emulsification method were prepared from aqueous standards and gave good linear
calibration curves with r2 ranging from 0.9998 to 1.000. The calculated LOQs for this
method were in the range of 3.92 and 68.7 ng g-1. These LOQs will imply that the
method could be used for the determination of trace levels of wear metals in
lubricating oils.
Microwave-assisted acid digestion using pressurised closed vessels yielded good
recoveries (95.6-104%) of all the elements studied, except Ag (50.4%). Silver showed
the worst recovery and this may have been due to its low concentration of that metal
in the CRM and the high dilution factor (about 100x), which led, in the final solution, to
levels close to the LOQ of the method. The precision, as determined by %RSD
(0.467-3.63%) was very good when compared to the other methods. The determined
70
LOQs for the microwave digestion method, were in the low ng g-1 level (3.58-60.2 ng
g-1), which would indicate that the method could be used for the determination of trace
levels of wear metals in lubricating oils. Aqueous working standards prepared for the
analysis of microwave digested samples yielded good linearity calibration curves with
r2 ranging from 0.9997 to 1.000. This achievement of comparatively good results by
the microwave digestion method can be explained in terms of the complete digestion
of the oil and low sample contamination incurred during microwave digestion.
Based on the five validation criteria given in Table 4.12, the method of choice for the
determination of Mn was found to be microwave digestion, for Cu, xylene dilution, for
Ba, wet-ashing, for Ni, oil emulsification and for Ag, ultrasonic extraction. In addition
to the validation criteria given in Table 4.12, the analytical methods investigated were
also evaluated based on their suitability for routine analysis, cost and ease of sample
preparation, as detailed below. The cost of sample preparation, which was based on
approximate costs of reagents and instrumental running costs, was approximated to
be microwave digestion > xylene dilution > wet ashing > dry ashing > ultrasonic
extraction > detergent emulsion.
Xylene dilution method required the use of organo-metallic standard for calibration,
which was more expensive than aqueous inorganic standards. The choice of special
sample tubes that can withstand organic solvents also contributed to the additional
cost of the xylene dilution method. The inability of xylene to dissolve some suspended
large wear particles in used oils and the lack of the method to minimise the organic
content of the oil were the main concerns with the applicability of the xylene dilution
method. These limitations are some of the main factors that could restrict the dilution
method for routine analysis. In contrast, this method offered minimal sample
contamination, was fast and simple.
The pressurised microwave digestion system, which employed inorganic acids as
oxidants, resulted in complete digestion of the oil samples. These digestates were
clear and easily aspirated without affecting plasma stability. Moreover, the closed
71
digestion system also offered minimal sample contamination, which is a major
concern in trace element analysis. In contrast, the initial and running-costs of the
microwave apparatus were found to be high, which could limit its use for routine
analysis. The limited organic sample size recommended for digestion was also a
major drawback for the microwave digestion method. For this reason, it was not
possible to digest the same sample size or use the same dilution factor, as was used
in the other methods. In addition, the closed microwave digestion technique is
moderately unsafe due to the possible occurrence of explosion, promoted by faults
such as vessel leaks, sample size, sample matrix and operator’s error.
The dry-ashing method completely mineralised the oil matrix and resulted in a clear
solution, which could be easily aspirated into the plasma. It was simple to use and
easy to control the procedure. Sample preparation costs were also moderate.
However, sample preparation time was rather long (about 17 h), which could limit its
use for routine analysis. In addition, sample contamination could be a major concern
for this method because of the open heating procedure. Furthermore, this technique
caused loss of volatile species and hence was found to be inefficient for the
determination of Ni.
The wet-ashing method completely destroyed the oil matrix and resulted in a clear
solution. Unlike dry-ashing, the wet-ashing method offered a relatively short sample
preparation time, taking 6 h (2 h heating to dryness and 4 h muffling). Sample heating
and muffling employed in both dry and wet-ashing procedures was relatively
inexpensive when compared to microwave digestion. It gave lower LODs (2.87-15.4
ng g-1), good accuracy (95.1-106%), except for Ag (76.3%) and good precision (1.105.62% RSD), except for Ni (9.47%). Hence, the wet-ashing approach can be used, as
an alternative method to microwave digestion, for routine analysis. In contrast,
sample contamination, precipitation of AgCl and BaSO4 when HCl and H2SO4 were
used, loss of Ni in the absence of H2SO4, sample loss due to foaming and sample
adsorption were some of the major limitations for the wet-ashing technique.
72
The ultrasonic extraction method, using extractant reagents, was found to be effective
for extraction of analytes from used lubricating oil. Initial and running costs for an
ultrasonic bath are inexpensive. The reaction vessels (volumetric flasks) are also
cheap and available in any laboratory. However, finding the location of the ultrasonic
transducer, which generates the ultrasonic energy, was a major limitation for this
method. Consequently, only four samples were used in each batch to guarantee
efficient extraction. When used oil samples were exposed to ultrasonic energy,
formation of intense bubbles inside the reaction vessels could be observed. These
effects suggest the occurrence of an oxidative process in the oil, which was
necessary to liberate the metals from the organic matrix. Samples situated far from
the location of the ultrasonic transducer remained unchanged. However, samples
within the location of maximum irradiation were observed to create pressure that
stressed the reaction vessels and finally formed an organic layer with agglomerated
solid material. The pressure was more pronounced during the first 30 min of
sonication. Therefore, an ultrasonic extraction of oils needs continual follow-up in
order to avoid the possibility of the vessels exploding. For its low cost of sample
preparation, reasonably short preparation time (2 h) and ease of procedures, the
ultrasonic extraction approach can be applied for routine analysis.
73
CHAPTER 5
CONCLUSION AND RECOMMENDATIONS
5.1
INTRODUCTION
This chapter provides summaries of the study outcome or conclusions, achievement
of the objectives, contribution of the study and recommendations for further study.
5.2
ACHIEVEMENT OF THE OBJECTIVES
The main objective of this study was to evaluate methods used for the determination
of trace elements in lubricating oils. The specific objectives of the study were set, as
given in Section 1.4.2, and experimentally tested according to their sequential priority.
The most sensitive analytical lines that can be used for the determination of trace
elements by ICP-OES were investigated prior any kind of sample analysis. The
measured peak signal and background intensities determined by ICP-OES served to
calculate the SBRs for each element. The maximum SBR was then applied as criteria
for line selection. Analytical lines with higher SBR values, together with no interfering
species in the vicinity of the lines, were established. These established analytical
lines served for the entire determination of trace elements in both aqueous and
organic sample matrices.
The modified simplex optimisation approach has been shown to be a rapid and
effective technique for the optimisation of ICP-OES operating conditions. The criteria
of maximum SBR values produced efficient ICP optimum conditions both for aqueous
and organic sample matrices. These optimum conditions guaranteed low background
emission and continuous aspiration of oils without plasma extinction and the build up
of carbon deposits on the injector tip. Evaluation of the ICP optimum conditions
74
indicated that the organic matrix required greater values of forward power, coolant
and auxiliary gas flow rates, as compared to the aqueous matrices. On the other
hand, increasing the auxiliary gas flow rate or a simultaneous increase both in the
coolant and nebuliser flow rates was shown to strongly increase the SBRs of the
elements studied. The performed regression analysis clearly illustrated that the
auxiliary gas flow rate was the most significant variable in generating enhanced SBR
values for the elements of interest, particularly during the analysis of organic samples.
Three factors that were considered to have the greatest impact on ultrasonic-assisted
extraction procedure were studied. A four-level full factorial design (43), with 64 runs,
was developed in order to determine the influence of the factors and their interactions
on extraction efficiency. The criteria of maximum analyte recovery generated efficient
extracting conditions for all the elements studied. Metal recoveries obtained from the
ultrasonic extraction method were quantitative, presenting values better than 93%.
Such values denote that the ultrasonic extraction procedure, proposed in this work, is
efficient in destroying the organic matrix and allowing the release of metals. The
application of the Pearson correlation function to the recovery data simplified the
evaluation of factors that influenced the performance of ultrasound-assisted
extraction. It was demonstrated that the most important interaction between variables
occurred when a simultaneous increase in sonication time (30-120 min) and volume
of the extracting reagent (3-10 mL) was used. This was critical, particularly when
HNO3:HCl (1:1, v/v) and HNO3:HCl (1:3, v/v), were used as sonicating reagents. Of
the four sonicating reagents studied, aqua regia was found to be the most powerful
and universal extracting agent for all the elements studied.
Nebulisations of 10 replicates of blank solutions, prepared together with the real
samples, were used for the determination of the LOD and LOQ for each element. The
obtained LODs and LOQs were in the low ng g-1 level, which indicated that the
methods could be easily be used for the determination of trace metals in lubricating
oils and compares very well with those proposed by Inorganic Ventures (2012).
75
Evaluation of the six analytical methods investigated (xylene dilution, detergent
emulsion, microwave digestion, dry-ashing, wet-ashing and ultrasonic extraction) for
determining trace elements in used lubricating oils showed significant differences
when statistically compared using the analysis of variance (ANOVA) at the 95%
confidence level. Nonetheless, more than 65% compared combinations of any two
methods gave statistically similar concentrations of Ag, Ba and Mn at the confidence
level adopted (95%). Additions of both organo-metallic standard (Conostan S-21) and
CRMs, analysed by ICP-OES, permitted the evaluation of each method studied based
on its accuracy and precision. All the methods studied, except dry-ashing, gave
quantitative recoveries of Ag. In all cases the precision (as measured by the %RSD)
for the determination of Ag in the CRM samples, was found to be poor (7.36-16.2%),
with the exception of results obtained by microwave digestion and ultrasonic
extraction (1.45-2.72%). The low level of Ag in the CRM and sample contamination,
particularly chloride contamination, complicated the accuracy and precision of Ag.
Precipitates of AgCl, Ag2SO4 and BaSO4 were formed during sulphated wet-ashing,
when followed by concentrated HCl dissolution. The method was thus ineffective for
the determination of Ag and Ba. The dry-ashing approach gave unsatisfactory Ag
recoveries. This was probably due to loss caused by sample adsorption onto the
vessel walls, which was unavoidable, and sample contamination caused by prolonged
open heating on a hot plate.
All the studied methods were found to be effective for the determination of Ba in used
oil samples. Evaluation of the recovery results for Ba, obtained by ultrasonic
extraction, indicated a clear dependency on the concentration of the additions. This
concentration dependency could have been due to some limitations of the aqua regia
(extracting agent) to release Ba from the organic matrix when the addition level
exceeded some concentration, in this case about 0.45 µg g-1.
All the studied methods permitted the determination of Cu and Mn in lubricating oils
with good accuracy. Quantitative recovery of Ni was also possible using all the
76
methods, except for the dry-ashing technique. Only the dry-ashing method provided
Ni recoveries below 50% in the two QC samples analysed by ICP-OES. This result
indicated that the metal is possibly volatilised and lost during intense heating
promoted by the dry-ashing programme. This phenomenon for Ni has been reported
by Aucelio et al. (2007a). The loss of Ni in the dry-ashing procedure was observed to
increase with increased levels of Ni; however, the technique of wet-ashing, which
employed oxidation of the organic material using concentrated H2SO4, followed by
HNO3 dissolution, solved the problem of Ni losses. Wet-ashing, which employed
HNO3 as oxidising agent, was found to be ineffective for the determination of Ni, but
effective for Ba. Similarly, sulphated wet-ashing was ineffective for the determination
of Ba due to the formation of insoluble BaSO 4. Based on the above, ashing
procedures for the determination of Ni need to be performed using H2SO4 in order to
reduce the loss of Ni as non-volatile sulphur compounds, most probably NiS, which is
readily soluble in HNO3.
Validation of the studied analytical methods was performed through the analysis of
additions of Conostan standards and EnviroMAT (CRM). The results obtained using
both sample additions were used to determine the accuracy, precision, linearity, LOD
and LOQ of each method. Based on these evaluations, the choice of method for the
determination of Mn in used oil samples was microwave digestion, for Ba, wet-ashing,
for Cu, xylene dilution, for Ni, detergent emulsion and for Ag, ultrasonic extraction.
Further evaluation of these methods in terms of applicability for routine analysis,
sample preparation time and cost (cost of reagents and instrumental running costs)
was performed. Based on these comparisons, ultrasonic extraction has a clear
advantage in terms of accuracy, cost of sample preparation, simplicity of use, safety
and suitability for routine analysis.
77
5.3
CONTRIBUTION OF THE STUDY
Used lubricating oil has always been a difficult sample matrix to dissolve so as to be
efficiently aspirated into the plasma. Its dissolution mainly involves dilution or
oxidation of the organic matter and this is usually achieved by employing several
inorganic acids or organic solvents. In the current study, dissolution of used
lubricating oil samples were carried out using six different analytical procedures that
involved the use of mineral acids, when necessary; xylene dilution, oil emulsification,
microwave digestion, ultrasonic extraction, dry and wet ashing. Despite its low cost,
literature research has shown that ultrasonic energy, generated from an ultrasonic
bath, has not yet fully been exploited for analytical applications. As a result, literature
related to ultrasound-assisted extraction approach, which employ an ultrasonic bath is
very limited. In this study, though the precision of results for Ni was poor (12.6%)
relative to other techniques, lubricating oils extracted using an ultrasonic bath
produced good recoveries for all the other elements studied. This indicates the
strength of ultrasonic energy, generated by the ultrasonic bath, to isolate inorganic
species from complex organic matrices. Hence, the finding of this study is expected to
create some motivation for future research, based on the use of ultrasonic energy.
5.4
SHORTCOMING OF THE STUDY
Working with used lubricating oil sample was a difficult task due to its matrix
complexity and viscosity, which compromises plasma stability. The problems
associated with the viscosity of used oil samples might have probably been resolved
using a V-groove nebuliser, which was not available in our laboratory.
5.5
RECOMMENDATIONS FOR FURTHER RESEARCH
For future research work in this field, further optimisation of the main factors that
influence the ultrasonic extraction of analytes in used lubricating oil samples need to
78
be done. This should include the distribution of ultrasonic energy and sample size;
optimisation of the factors affecting detergent emulsion and optimisation of ICP-OES
operating parameters, including observation height and pumping rate. Further,
optimisation of oxidising agents used for the wet-ashing procedure, particularly
targeted for the determination of Ag and Ba, which were the most troublesome
elements in the current study, would be interesting. It is also imperative that the full
spectrum of just over 20 elements associated with lubricating oils be investigated.
79
REFERENCES
AGARWAL, A. K., SHARMA, M. & SINGH, S. K. 2006. Experimental investigations
of heavy metal addition in lubricating oil and soot deposition in an EGR operated
engine. Applied Thermal Engineering, 26:259-266.
AGAT LABORATORIES. 2004. Oil analysis user guide tribology [Online]. Calgary
Alberta: agatlabs.com. Available from: www.agatlabs.com [Accessed: 24/6/2010].
AGAZZI, A. & PIROLA, C. 2000. Fundamentals, methods and future trends of
environmental microwave sample preparation. Microchemical Journal, 67:337-341.
AGNELLI, J. A. M., GODOI, Q., KRUG, F. J., LEME, F. O., NUNES, L. C., RUFINI, I.
A., SANTOS, D. & TREVIZAN, L. C. 2009. Preliminary studies of laser-induced
breakdown spectrometry for the determination of Ba, Cd, Cr and Pb in toys.
Spectrochimica Acta Part B, 64:573-581.
ALLMAN, M. C. 1995. Sequential simplex optimisation. Rockwell SOC team, 1-14.
ANA, F. W. S., CASSELLA, A. R., CASSELLA, R. J. & SANTELLI, R. E. 2007.
Optimisation of an open-focused microwave oven digestion procedure for
determination of metals in diesel oil by inductively coupled plasma-optical emission
spectrometry. Journal of Hazardous Materials, 149:67-74.
ANALYSTS. 2008. Basics of oil analysis [Online]. Torrance: Analysts Corporation,
Inc. Available from: www.AnalystsInc.com [Accessed 21/10/2010].
ANGNES, L., MUNOZ, R. A. A. & OLIVEIRA, P. V. 2006. Combination of ultrasonic
extraction and stripping analysis: an effective and reliable way for the determination of
Cu and Pb in lubricating oils. Talanta, 68:850-856.
80
AUCÉLIO, R. Q., SILVEIRA, C. L. P. D. & SOUZA, R. M. 2004. Determination of
refractory elements in used lubricating oil by ICP-OES employing emulsified sample
introduction and calibration with inorganic standards. The Japan Society for
Analytical Chemistry, 20:351-355.
AUCÉLIO, R. Q., BRANDÃO, G.P., CAMPOS, R.C., DUYCK, C., GRINBERG, P.,
MIEKELEY, N, & SILVEIRA, C.L.P. 2007a. The determination of trace elements in
crude oil and its heavy fractions by atomic spectrometry. Spectrochimica Acta Part B,
62: 939–951.
AUCÉLIO, R. Q., CAMPOS, R. C. D., MIEKELEY, N., SILVEIRA, C. L. P. & SOUZA,
R. M. D. 2007b. The determination of trace metals in lubricating oils by atomic
spectrometry. Spectrochimica Acta Part B, 62:952-961.
BARBEIRA, P. J. S. & FERREIRA, H. B. P. 2009. Voltametric determination of zinc
in lubricating oils. Energy and Fuels, 23:3048-3053.
BASHAM, W., IRWIN, R. J., MOUWERIK, M. V., SEESE, M. D. & STEVENS, L.
1997. Environmental Contaminants Encyclopedia. Oil: used motor oil entry. Fort
Collins, Clourado: National Park Service, Water Resources Division.
BITOG. 2010. Oil forums [Online]. S.I.: Bobistheoilguy.com. Available from:
http://www.bobistheoilguy.com/cms/index.php?option=com_content&view=article&id=
50&Itemid=56 [Accessed: 16/06/2010].
BODY, D., CHADWICK, B. L., MORRISON, R. J. S. & YAROSHCHYK, P. 2005.
Quantitative determination of wear metals in engine oils using laser-induced
breakdown spectroscopy: a comparison between liquid jets and static liquids.
Spectrochimica Acta Part B, 60:986-992.
81
BOMBELKA, R. M. & NHAM, T. T. 1991. Determination of metals in lubricating oil by
ICP- AES. ICP-AES instruments at work. Varian.
BOSS, C. B. & FREDEEN., K. J. 2004. Concepts, Instrumentation and Techniques
in Inductively Coupled Plasma Optical Emission Spectrometry. 3rd ed. USA: Perkin
Elmer.
CAPELO, J. L., MADURO, C. & VILHENA, C. 2005. Discussion of parameters
associated with the ultrasonic solid-liquid extraction for elemental analysis (total
content) by electrothermal atomic absorption spectrometry: an overview. Journal of
Ultrasonics Sonochemistry, 12:225–232.
CARPENTER, R. & EBDON, L. 1986. A comparison of inductively coupled plasma
torch-sample introduction configurations using simplex optimisation. Journal of
Analytical Atomic Spectrometry, 1:265-268.
CASTRO, M. D. L. D. & PRIEGO-CAPOTE, F. 2004. Analytical uses of ultrasound:
sample preparation. Trends in Analytical Chemistry, 23:644-653.
CAVE, M. R., EBDON, L. & MOWTHORPE, D. J. 1980. Simplex optimisation of
inductively coupled plasmas. Analytica Chimica Acta, 115:179-187.
CHAUSSEAU, M., LEBOUIL-ARLETTAZ, S. & THOMAS, G. 2010. Wear metals and
additive elements analysis in used lubricating oils using ICP-OES [Electronic
Version]. Analytical Instrumentation: 6-7. Retrieved 11/12/ 2010 from
www.horiba.com/scientific/.
CHIRINOS, J., FERNANDEZ, A. & FRANQUIZ, J. 1998. Multielement optimisation
of the operating parameters for inductively coupled plasma-atomic emission
spectrometry with a charge injection device detector for the analysis of samples
82
dissolved in organic solvents. Journal of Analytical Atomic Spectrometry, 13:9951000.
COSTA, L. M., FERREIRA, S. L. C., NOBREGA, J. A. & NOGUEIRA, A. R. A. 2005.
Use of factorial design for optimisation of microwave-assisted digestion of lubricating
oil. Journal of Brazilian Chemical Society, 16(6A):1269-1274.
DEMING, S. N., MORGAN, S. L., PARKER, L. R. & WALTERS, F. H. 1999.
Sequential simplex optimisation. Karlskrona: MultiSimplex AB.
ESSLING, A. M., HUFF, D. R., HUFF, E. A., FOX, I. M. & GRACZYK, D. G. 1995.
Preparation of waste oil for analysis to determine hazardous metals. USA:
Department of Commerce, National Technical Information Service.
EVANS, J. S. 2010. Where does all that metal come from? Technical bulletin (47
ed., pp. 1-6). Wear check Africa: a member of the Set Point Group.
FAO CORPORATE DOCUMENT REPOSITORY. 1998. Guidelines for quality
management in soil and plant laboratories [Online]. S.I.: Fao.org. Available from:
http://www.fao.org/docrep/w7295e/w7295e09.htm#TopOfPage [Accessed:
02/11/2010].
FILHO, V. R. A. & NETO, J. A. G. 2009. Evaluation of lubricating oil preparation
procedures for the determination of Al, Ba, Mo, Si and V by high-resolution continuum
source FAAS. The Japan Society for Analytical Chemistry, 25:95-100.
FRAME, S. E. M., FRAME, G. M. & ROBINSON, J. W. 2005. Undergraduate
instrumental analysis. USA: Marcel Dekker.
GAINES, P. 2004. ICP Operations: common Problems with Ag, As, S, Ba, Pb, and
Cr [Online]. USA: Inorganic Ventures. Available from:
http://www.esslab.com/iv/tech/iv-icp-operations/part13.htmL [Accessed 19/01/2012].
83
GALLEY, P. J., HIEFTJE, G. M. & HORNER, J. A. 1995. Automated simplex
optimisation for monochromatic imaging inductively coupled plasma-atomic emission
spectroscopy. Spectrochimica Acta Part B, 50:87-107.
GEDEON, G. P. E. 1999. Lubricants and hydraulic fluids. Pomona: CED
engineering. (Study guide T05-002).
GONCALVES, I. M., GONZALEZ, A. M. & MURILLO, M. 1998. Determination of
metals in used lubricating oils by AAS using emulsified samples. Talanta, 47:10331042.
INORGANIC VENTURES. 2012. Interactive Periodic Table [Online]. Inorganic
Ventures. Available: http://www.inorganicventures.com/tech/periodic-Table/
[Accessed 30/04/2012].
JANKOWSKI, K., JERZAK, A., SERNICKA-POLUCHOWICZ, A. & SYNORADZKI, L.
2001. Multielement determination of major elements in polymer additives by
microwave induced plasma-atomic emission spectrometry after microwave digestion.
Analytica Chimica Acta, 440:215–221.
JEREMIĆ, B. & MAČUŽIĆ, I. 2004. Proactive approach to oil maintenance strategy.
Tribology in Industry, 26(1, 2):27-31.
JHA, M. K. 2005. Re-refining of used lube oils: an Intelligent and eco-friendly option.
Journal of Indian Chemical Engineering Section B, 47(3):209-211.
KUOKKANEN, T., LAHDELMA, S., ROPPOLA, K., VAHAOJA, P. & VALIMAKI, I.
2008. Wear metal analysis of oils. Critical Reviews in Analytical Chemistry, 38(2):6783.
84
LEOCADIO, L. G., SILVEIRA, C. L. P. D. & SOUZA, R. M. D. 2008. ICP-OES
simultaneous determination of Ca, Cu, Fe, Mg, Mn, Na, and P in biodiesel by axial
and radial inductively coupled plasma-optical emission spectrometry. Journal of
Analytical Letters, 41(9):1615-1622.
MARAIS, P. J. J. G. 1987. Application of an argon-cooled Inductively Coupled
Plasma to the analysis of metals in lubricating oils. MSc dissertation, Cape Town,
University of Cape Town.
MILNER, T. E. (ed). 2009. AMETEK materials analysis division. SPECTRO Live.
Kleve: SPECTRO analytical instruments GmbH.
MOBIL. 2009. Oil analysis: the basics [Online]. S.I.: Exxon Mobil Corporation.
Available from: www.mobilindustrial.com [Accessed: 18/05/2010].
NEW ZEELAND INSTITUTE OF CHEMISTRY. 2002. The re-refining of used
lubricating oils. VII-Energy-B-Re-refining used oils [Online]. Available from:
http://nzic.org.nz/ChemProcesses/energy/78.pdf [Accessed: 27/7/2010].
OLIVEIRA, E. D. 2003. Sample preparation for atomic spectroscopy: evolution and
future trends. Journal of Brazilian Chemical Society, 14(2):174-182.
PETROLEUM HPV TESTING GROUP. 2003. Lubricating oil base stocks category
[Online]. S.I.: petroleumhpv.org. Available from: www.petroleumhpv.org [Accessed:
28/03/2011].
RIPP, J. 1996. Analytical detection limit guidance and laboratory guide for
determining method detection limits. Wisconsin Department of Natural Resources
Laboratory Certification Program.
85
SALIN, E. D. & SARTOROS, C. 1997. Comparison of two objective functions for
optimisation of simultaneous multielement determinations in inductively coupled
plasma spectrometry. Journal of Analytical Atomic Spectrometry, 12:13-19.
THERMO ELEMENTAL. 2002. The analysis of new and used lubricating oils by
atomic absorption spectrometry. Franklin, USA: ThermoElectron.
THOMSON, J. 2007. Waste vacuum pump oil sample preparation and analysis
[Online]. S.n.: npl.co.uk. Available from: www.npl.co.uk/.../waste-vacuum-pump-oilsample-preparation-and-analysis [Accessed 13/4/2010].
VAHAOJA, P. 2006. Oil analysis in machine diagnostics. PhD, thesis, University of
Oulu, Finland.
86
ANNEXURE A
DETAILS OF THE SIMPLEX ALGORITHM
A.1
DEFINITION OF THE SIMPLEX ALGORITHM
A simplex is a geometric figure defined by a number of points or vertices equal to one
more than the number of dimensions of the factor space. If k is the number of
dimensions in the factor space, then a simplex is defined by k+1 points in that factor
space. Each vertex, or corner, of a simplex optimisation experimental design
corresponds to a set of experimental conditions (Deming et al., 1999).
A.2
TYPES OF THE SIMPLEX ALGORITHM
A.2.1
Fixed step simplex algorithm
Fixed step simplex is the original or basic simplex algorithm, which was introduced as
an alternative evolutionary operation method for increasing the productivity of existing
industrial processes (Deming et al., 1999). The progression along the response
surface is in fixed distance units, which is used when there is a need for maximum
control of the experiments (Allman, 1995).
A.2.2
Variable step simplex algorithm
The variable step simplex is also referred to as the modified simplex because several
new rules were added to the basic simplex, to allow the simplex to change its size.
These new rules permit the modified simplex to expand in a direction of more
favourable condition and contract in a direction of unfavourable condition (Deming et
al., 1999). The variable step simplex is not stranded by ridged systems and can
therefore speed up the optimisation progress, prevent the attainment of a false
87
optimum and permit a closer definition of the optimum conditions (Cave, Ebdon &
Mowthorpe, 1980).
A.3
Rules of the simplex algorithm
The possible moves in the modified simplex algorithm are as depicted in Figure A.1.
Each move is governed by a series of rules. The rules that govern both the fixed and
the variable size simplex algorithm which are described in the literature by Allman
(1995) and Deming et al. (1999) are as follows:
E
N
R
C
+
P
CW
Figure A.1:
B
Possible moves in the variable size simplex algorithm. Where P is the
centroid of the hyperface; B the best vertex; N the second worst vertex;
W the worst vertex; R the reflection vertex; E the expansion vertex; C+
the contraction vertex on the R side and C- the contraction vertex on the
W side (Deming et al., 1999)
A.3.1.
Rules of the fixed step simplex algorithm
Rule-1:
Rank the vertex of the first simplex on a worksheet in decreasing order
of response from best to worst and put the worst vertex into the row
labelled W.
Rule-2:
Calculate and evaluate R.
88
Rule-3:
Never transfer the current row labelled W to the next worksheet. Always
transfer the current row labelled N to the row labelled W on the next
worksheet. Rank the remaining retained vertices in order of decreasing
responses on a new work sheet and go to rule 2.
A.3.2.
Rules of the variable step simplex algorithm
The sign “≥” represent “better than or equal to” and “≤” worse than or equal to.
Rule-1:
Rank the vertex of the first simplex on a worksheet in decreasing order
of response from best to worst and put the worst vertex into the row
labelled W.
Rule-2:
Calculate and evaluate R:
A. If N ≤ R ≤ B, use simplex B..NR.
B. If R > B, calculate and evaluate E:
i. If E ≥ B, use simplex B..NE.
ii. If E < B, use simplex B..NR.
C. If R< N:
i. If R ≥ W, calculate and evaluate C+, use simplex R..NC+.
ii. If R < W, calculate and evaluate C-, use simplex R..NC-.
Rule-3:
Never transfer the current row labelled W to the next worksheet. Always
transfer the current row labelled N to W on the next worksheet. Rank the
remaining retained vertexes in order of decreasing responses.
89
ANNEXURE B
ADDITIONAL TABLES
Table B.1:
Trial
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
4RE1
18
19
20
21
22
23
9RE1
24
25
26
27
28
29
30
Simplex optimisation progress data of ICP-OES operating parameters
for Ag (328.068) in xylene matrix
RF
power
Coolant
flow
Aux.
flow
Neb.
Flow
(W)
(L min )
(L min )
(L min )
17
17
17
15
15
18
17
16
16
16
14
17
15
14
14
13
16
15
15
14
13
15
15
13
16
15
14
15
16
16
15
15
1.8
2.3
1.8
2.3
1.8
2.3
1.8
1.5
2.1
2.3
1.6
2.1
2.2
2.4
2
1.9
1.6
2.3
2.1
2.7
3.3
2.4
2.2
2.5
2.1
2.2
2.4
2.2
2.7
3
2.7
2.3
0.55
0.55
0.65
0.65
0.55
0.65
0.7
0.77
0.81
0.94
0.79
0.69
0.81
0.89
0.83
0.9
0.96
0.65
0.73
0.82
0.85
0.91
0.77
0.8
0.81
0.81
0.8
0.81
0.78
0.75
0.78
0.8
1550
1650
1650
1550
1650
1550
1500
1425
1538
1531
1532
1545
1379
1244
1401
1329
1322
1550
1493
1481
1508
1407
1471
1328
1538
1486
1541
1419
1528
1591
1464
1480
-1
-1
-1
Signal
(cps)
919424
968661
1333387
1317785
814311
954903
859457
748565
746312
295580
363387
1304887
899357
BV
900000
BV
BV
1178583
1337671
1181737
BV
555935
1290540
BV
1376599
1259793
1360132
1097991
1384519
1433576
1135965
1120795
Back
ground
(cps)
57783.6
80031.8
55124.6
47136
82862.2
42540.2
28161.3
18025.8
16123.2
22417
22499.8
39772.4
18182.5
-17948.7
--48393.6
29545.2
22032.3
-21564.8
23847.2
-28702.2
24231.4
28646.6
20615.2
25169.4
38877.8
22973.8
22929.9
SBR
T
Ret
Type
14.9
11.1
23.2
27.0
8.8
21.4
29.5
40.5
45.3
12.2
15.2
31.8
48.5
-49.1
--23.4
44.3
52.6
-24.8
53.1
-47
51
46.5
52.3
54
35.9
48.4
47.9
3
2
5
7
1
3
0
6
7
0
0
2
6
0
6
0
0
0
2
7
0
0
7
0
0
3
0
3
7
0
0
2
F
F
F
F
F
R
R
E
R
E
R
CR
E
R
E
R
F
CR
E
R
CR
R
CR
CR
E
R
C-
90
31
32
33
19RE1
34
35
22RE1
36
37
38
39
40
41
27RE1
42
36RE1
43
44
45
36RE2
46
22RE2
47
48
49
50
51
52
53
42RE1
54
36RE3
55
52RE1
56
57
58
36RE1
51RE1
53RE1
1561
1455
1488
1481
1482
1526
1471
1473
1496
1485
1497
1500
1501
1528
1487
1473
1467
1424
1458
1473
1468
1471
1527
1450
1474
1472
1503
1529
1509
1487
1526
1473
1532
1529
1494
1500
1477
1473
1503
1509
15
15
15
14
15
15
15
15
17
15
16
15
16
16
16
15
15
15
16
15
15
15
16
15
16
15
16
16
16
16
17
15
16
16
15
15
15
15
16
16
2.8
2.3
2.7
2.7
2.4
2.7
2.2
2.4
2.1
2.6
2.6
2.2
2.1
2.7
2.5
2.4
3.2
2
2.5
2.4
2.3
2.2
2.8
2.2
2.5
2.4
2.7
2.9
2.7
2.5
2.9
2.4
3
2.9
2.6
2.1
2.1
2.4
2.7
2.7
0.77
0.8
0.78
0.82
0.8
0.78
0.77
0.8
0.76
0.8
0.78
0.65
0.7
0.78
0.87
0.8
1.01
0.84
0.89
0.8
0.8
0.77
0.79
0.83
0.85
0.84
0.82
0.82
0.81
0.87
0.85
0.8
0.77
0.82
0.77
0.6
0.69
0.8
0.82
0.81
1225043
1079768
1171119
1037904
1167012
1206417
1173315
1112030
1235909
1145426
1207034
1072935
1170195
1292941
921434
1552811
BV
1074865
812462
1411804
1386284
1518611
1545182
1167711
1126988
1162315
1355239
1411632
1404011
1068673
1258436
1384091
1578056
1407955
1525195
1103767
1413215
1371318
1379512
1419076
30165.4
21530.7
24207
21810.3
23891.2
27021
22504
17727.6
26488.1
23844
27347
38346.9
34451
28200.5
15167.8
24685
-18249.2
20770
24826.2
25840.7
28908.9
29337.4
20512.8
20853.7
22538.8
21272.6
19638.6
23339.9
20999.8
22929.2
24488.4
32043
27284.7
30571.2
47863.2
32252.2
22477
23954.4
27110.2
39.6
49.2
47.4
46.6
47.8
43.6
51.1
61.7
45.7
47
43.1
27
33
44.8
59.7
61.9
-57.9
38.1
55.9
52.6
51.5
51.7
55.9
53
50.6
62.7
70.9
59.2
49.9
53.9
55.5
48.2
50.6
48.9
22.1
42.8
60.0
56.6
51.3
0
2
0
2
3
0
7
7
0
2
0
2
2
3
7
7
0
2
0
5
2
0
0
2
0
2
0
6
5
2
2
4
0
2
0
2
1
2
2
1
*T Ret stands for “total retained”, REn “n-times re-evaluation” and F “first vertex”.
R
CR
R
CR
CCR
CR
CR
R
R
R
E
R
R
CCCR
CR
C+
R
E
R
R
R
R
R
CR
CR
E
CC-
91
Table B.2:
Trial
No.
Simplex optimisation progress data of ICP-OES operating parameters
for Mn (257.611) in aqueous matrix
RF
power
(W)
1
1400
2
1400
3
1200
4
1200
5
1400
6
1200
7
1400
8
1250
9
1425
10
1538
11
1463
12
1409
13
1317
14
1379
15
1293
16
1235
17
1549
2RE1
1400
18
1461
19
1379
20
1277
21
1308
9RE1
1425
22
1361
23
1299
24
1359
25
1442
26
1303
15RE1 1293
27
1241
28
1307
29
1382
30
1359
31
1191
32
1379
33
1284
Coolant
flow
(L min-1)
12
13
12
13
13
12
13
12
12
11
13
12
13
12
13
13
13
13
13
12
12
13
12
12
13
12
13
13
13
14
14
14
13
14
13
15
Aux.
flow
(L min-1)
1.3
0.8
0.8
1.3
1.3
0.8
1.3
0.9
0.6
0.3
0.8
0.9
0.3
1
0.7
0.6
0.7
0.8
0.8
0.7
0.7
0.8
0.6
0.7
0.4
0.8
0.7
0.9
0.7
1
0.9
1
0.9
1.1
0.8
1
Neb. flow
(L min-1)
0.8
0.8
1.3
0.8
1.3
0.6
0.2
1
0.8
0.8
0.5
0.6
0.6
0.8
0.9
1.1
1
0.8
0.9
0.9
0.8
0.7
0.8
0.9
0.9
0.8
0.9
0.9
0.9
1
0.9
0.9
0.9
0.9
0.9
1
Signal
Back
ground
SBR
T
Ret
Type
(cps)
1591185
1466942
3090.9
247107
19559
840496
BV
BV
1631405
BV
900826
1404959
BV
1616529
661433
26148.8
1264187
1560331
1466942
1146281
1036364
1257025
1705785
1057576
BV
1483471
1380716
805785
680716
173058
795592
599725
1042424
322865
1097521
234050
(cps)
44077
37190
225.9
15427
1652.9
22314
--39669
-35813
45455
-39669
13499
771.3
32231
39669
37190
28926
24242
32231
39669
24242
-33058
30854
17355
15427
3471
17631
13499
21212
7713.5
26446
5289.3
35.1
38.4
12.7
15.0
10.8
36.7
--40.1
-24.2
29.9
-39.8
48.0
32.9
38.2
38.3
38.4
38.6
41.8
38.0
42.0
42.6
-43.9
43.7
45.4
43.1
48.9
44.1
43.4
48.1
40.9
40.5
43.2
5
7
2
3
1
6
0
2
7
0
0
2
0
6
7
0
0
1
2
2
3
0
3
4
0
6
4
0
2
7
7
0
7
0
2
2
F
F
F
F
F
R
R
CR
E
R
C+
R
CR
E
R
F
C+
R
R
R
R
CR
CR
R
R
E
R
R
C+
R
CR
92
34
35
36
37
27RE1
38
28RE1
39
40
1217
1278
1280
1377
1241
1456
1362
1307
1462
Table B.3:
15
13
14
13
14
11
13
14
13
1.1
0.9
1
0.8
1
0.6
0.8
0.9
0.7
1
0.9
0.9
0.8
1
0.8
0.9
0.9
0.8
114270
686501
653719
1503030
163636
1939394
1033333
762534
1818182
2534.4
15427
15427
35262
3966.9
49587
24242
17631
44077
44.1
43.5
41.4
41.6
40.3
38.1
41.6
42.2
40.3
2
0
2
2
2
0
1
1
0
Sensitivity study of ICP-OES lines for the elements investigated
Wavelength Signal
Element
(nm)
(cs-1)
Ag I
328.068
2574617
Ag I
338.289
160000
Ag II
224.641
4300
Ag II
243.779
5508
Ba ll
455.404
6301994
Ba ll
233.527
1670472
Ba ll
230.424
1632506
Cu I
324.754
4145299
Cu I
327.396
478908.2
Cu II
224.700
383126.6
Cu II
219.226
240000
Cu I
219.958
159603
Mn II
257.611
5641026
Mn II
259.373
505785.1
Mn II
260.569
392286.5
Mn II
294.921
349311.3
Mn I
403.076
257135.0
Ni II
231.604
616625.3
Ni II
221.648
556823.8
Ni I
232.003
219454.1
Ni II
227.021
14069.5
Ni II
174.828
19454.1
Ni I
300.249
130000
Ni I
341.476
516239.3
-1
* cs “counts per second”
Back
ground
(cs-1)
Net Signal
(cs-1)
SBR
7407.4
4367
176
268
39886
25310.2
25310.2
14245
29776.7
11910.7
10322.6
10322.6
17094
11570.2
8815.4
8815.4
15757.6
17632.8
15399.5
12009.9
1749.4
2183.6
27369.7
13675.1
2567210
155633
4124
5240
6262108
1645161
1607196
4131054
449131.5
371215.9
229677.4
149280.4
5623931.6
494214.9
383471.1
340495.9
241377.4
598992.5
541424.3
207444.2
12320.1
17270.5
102630.3
502564.2
346.6
35.6
23.4
19.6
157.0
65.0
63.5
290.0
15.1
31.2
22.2
14.5
329.0
42.7
43.5
38.6
15.3
34.0
35.2
17.3
7.0
7.9
3.7
36.8
R
R
C+
R
E
E
R
R
R
93
Table B.4:
Trial
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
The full factorial experimental design (43) applied to the optimisation of
the ultrasound-assisted extraction of metals in lubricating oils
Sonication time
(min)
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
60
60
60
60
60
60
60
60
60
60
60
60
60
60
60
60
90
90
90
90
90
90
90
90
Ratio of extractant
(HNO3:H2O2:HCl)
2:1:0
2:1:0
2:1:0
2:1:0
1:0:0
1:0:0
1:0:0
1:0:0
1:0:1
1:0:1
1:0:1
1:0:1
1:0:3
1:0:3
1:0:3
1:0:3
2:1:0
2:1:0
2:1:0
2:1:0
1:0:0
1:0:0
1:0:0
1:0:0
1:0:1
1:0:1
1:0:1
1:0:1
1:0:3
1:0:3
1:0:3
1:0:3
2:1:0
2:1:0
2:1:0
2:1:0
1:0:0
1:0:0
1:0:0
1:0:0
Volume of extractant
(mL)
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
94
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
90
90
90
90
90
90
90
90
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
1:0:1
1:0:1
1:0:1
1:0:1
1:0:3
1:0:3
1:0:3
1:0:3
2:1:0
2:1:0
2:1:0
2:1:0
1:0:0
1:0:0
1:0:0
1:0:0
1:0:1
1:0:1
1:0:1
1:0:1
1:0:3
1:0:3
1:0:3
1:0:3
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
3
6
8
10
95
ANNEXURE C
ADDITIONAL FIGURES
Mn 257.611
Mn 257.611
50
SBR
45
45
40
40
35
35
30
30
SBR
SBR
50
25
25
20
20
15
15
10
10
11
12
13
14
15
SBR
1150 1200 1250 1300 1350 1400 1450 1500 1550 1600
-1
Forward RF power (W)
Plasma flow rate (L min )
(a)
(b)
Mn 257.611
Mn 257.611
SBR
45
45
40
40
35
35
30
30
SBR
SBR
50
50
25
25
20
20
15
15
10
10
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
-1
Auxiliary gas flow rate (L min )
(c)
SBR
0.4
0.6
0.8
1.0
1.2
1.4
-1
Nebulizer flow rate (L min )
(d)
Figure C.1: Progress of the simplex optimisation of the ICP-OES when applied to
(a) plasma flow, (b) forward power, (c) auxiliary flow and (d) nebuliser
flow rate for Mn in aqueous matrix
96
Ag 328.068
Signal intensity (cps)
1200000
1000000
800000
y = 43597.7x + 4151.9
2
r = 0.99903
600000
400000
200000
0
0
5
10
15
20
25
-1
Concentration (µg g )
Figure C.2: Calibration curve for the determination of Ag in xylene diluted oils
Ba 455.404
1400000
Signal intensity (cps)
1200000
1000000
800000
y = 25430.6x - 6774.9
2
r = 0.99987
600000
400000
200000
0
-200000
0
10
20
30
40
50
-1
Concentration (µg g )
Figure C.3: Calibration curve for the determination of Ba in microwave digested oils
97
Cu 324.754
1600000
Signal intensity (cps)
1400000
1200000
1000000
y = 29900.5x + 2331.8
2
r = 0.99998
800000
600000
400000
200000
0
-200000
0
10
20
30
40
50
-1
Concentration (µg g )
Figure C.4: Calibration curve for the determination of Cu in emulsified oils
1400000
Mn 257.611
Signal intensity (cps)
1200000
1000000
y = 25826x - 2157.8
2
r = 0.99998
800000
600000
400000
200000
0
-200000
0
10
20
30
40
50
-1
Concentration (µg g )
Figure C.5: Calibration curve for the determination of Mn in dry-ashed oils
98
Ni 341.476
350000
Sgnal intensity (cps)
300000
250000
200000
y = 3087.66x + 156.63
2
r = 0.99999
150000
100000
50000
0
0
20
40
60
80
100
-1
Concentration (µg g )
Figure C.6: Calibration curve for the determination of Ni in wet-ashed oils
Cu 324.754
3000000
Signal intensity (cps)
2500000
2000000
y = 26004.63x + 6368.87
2
r = 0.99999
1500000
1000000
500000
0
0
20
40
60
80
100
-1
Concentration (µg g )
Figure C.7: Calibration curve for the determination of Cu in ultrasonically extracted
oils