plasma optical emission spectrometry by charles …
TRANSCRIPT
FUNDAMENTAL STUDIES AND APPLICATIONS IN MICROWAVE-INDUCED
PLASMA OPTICAL EMISSION SPECTROMETRY
BY
CHARLES BRYSON WILLIAMS, III
A Dissertation Submitted to the Graduate Faculty of
WAKE FOREST UNIVERSITY GRADUATE SCHOOL OF ARTS AND SCIENCES
in Partial Fulfillment of the Requirements
for the Degree of
DOCTOR OF PHILOSOPHY
Chemistry
May 2019
Winston-Salem, North Carolina
Approved By:
Bradley T. Jones, PhD, Advisor
Michael D. Gross, PhD, Chair
Christa L. Colyer, PhD
George L. Donati, PhD
Scott M. Geyer, PhD
ii
This work is dedicated to my grandfathers:
LEWIS FRANKLYN SUTTON, PhD,
who taught me the value of knowledge;
CHARLES BRYSON WILLIAMS, Sr,
who taught me the value of hard work.
iii
ACKNOWLEDGMENTS
I wish to thank and acknowledge all those who have made this possible. First, my parents,
Laura and Bryson Williams, who raised me and supported me, my sister Meg, who put up
with me, and my aunt Nancy Sutton who has taken care of me in Winston-Salem. I also
could never have done this without my late grandfather, Lewis Sutton, who made much of
my education possible and who inspired my love of learning, as well as my living
grandfather Charles Williams, Sr, who inspired me by working his way out of poverty into
the American Dream.
I also wish to thank my advisors for all their support: Brad Jones for being willing to take
a student after a hiatus, and for securing a research assistantship for two years, and George
Donati for working tirelessly and putting up with me through melted torches, spilled
solutions, and sloppy calibrations, as I found my footing in the lab. I also greatly appreciate
my lab mates, Jake Carter and John Sloop, for their support and advice on projects and on
keeping the lab running smoothly, as well as for their friendship. And I thank Cliff
Calloway for his advice and perspective during the summer months. I also thank Tom
Whitmann for his hard work in our lab, as well as all the other undergraduate members.
I also thank those collaborators who have supported me through my years here, especially
Tina and Larry McSweeney, and everyone at Agilent Technologies, Paul Elliott and
everyone at CEM, Holly Peterson at Guilford College, and Daniel Goncalves and Renata
Amais from Brazil. I also thank Shiba Adhikari, Hui Li, Chang Lu, Jennifer Buchanan, and
iv
all of our collaborators within the Chemistry Department at Wake, as well as our
collaborators in the Physics Department.
I wish to thank my committee members, Christa Colyer, Michael Gross, and Scott Geyer
for their advice and feedback along the way. I also thank Al Rives and Jon Booze, the
members of the Teaching and Learning Collaborative, as well as my students, for their
support and guidance as I have developed my teaching skills. I also wish to acknowledge
Steve Creager and the Chemistry Department at Clemson University for setting me up with
a solid background in chemistry and teaching me that graduate school was possible.
I wish to thank David Pegg, the Rev. Lawrence Womack, the Rev. Ginny Wilder, and
everyone at St. Anne’s Episcopal Church for providing me a creative and spiritual outlet
in serving as their organist during my time in Winston-Salem.
Finally, and most importantly, I thank God, the Universal Mind and Source of all
knowledge and wisdom, without whose grace I would have been utterly unable to complete
this work.
v
TABLE OF CONTENTS
PAGE
LIST OF TABLES AND FIGURES vi
LIST OF ABBREVIATIONS xi
ABSTRACT xiii
CHAPTER I INTRODUCTION 1
CHAPTER II DETERMINATION OF CALCIUM, POTASSIUM AND SODIUM IN
SOFT DRINKS USING THE 4200 MP-AES 31
CHAPTER III DRY ASHING AND MICROWAVE-INDUCED PLASMA OPTICAL
EMISSION SPECTROMETRY AS A FAST AND COST-EFFECTIVE STRATEGY
FOR TRACE ELEMENT ANALYSIS 41
CHAPTER IV NATURALLY OCCURRING MOLECULAR SPECIES USED FOR
PLASMA DIAGNOSTICS AND SIGNAL CORRECTION IN MICROWAVE-
INDUCED PLASMA OPTICAL EMISSION SPECTROMETRY 66
CHAPTER VI CONCLUSIONS 124
APPENDIX A SUPPLEMENTARY INFORMATION FOR CHAPTER III 126
SCHOLASTICA VITA 136
vi
LIST OF TABLES AND FIGURES
TABLES
Table I. Recently described non-commercial MIP OES systems. ................................... 12
Table II. Recent sample introduction strategies and sample preparation procedures used in
MIP OES. .......................................................................................................................... 17
Table III. Element-Specific Operating Parameters.......................................................... 34
Table IV. Instrument Parameters ..................................................................................... 35
Table V. Soft drink analysis with the 4200 MP-AES. ..................................................... 38
Table VI. Ca, K and Na concentrations in soft drinks determined by the 4200 MP-AES.
........................................................................................................................................... 39
Table VII: Operating conditions used in MIP OES determinations. ............................... 49
Table VIII. Ashing parameters providing the best recoveries for determinations using the
HRA prototype and MIP OES. ......................................................................................... 53
Table IX. Evaluating the accuracy of the HRA-MIP OES procedure by analyzing a
standard reference material of Tomato Leaves (NIST SRM 1573a). ............................... 56
Table X. Comparison between ashing and acid extraction (HRA) with the traditional
microwave-assisted digestion (MAD). ............................................................................. 58
Table XI. HRA results presented as their percent portion of concentrations determined
using the MAD procedure. ................................................................................................ 60
Table XII. Instrumental operating conditions used in MIP OES. .................................... 74
Table XIII. Analyte percent recoveries for 2.0 mg L-1 solutions of Al, Ba, Mn, Sr and Zn
prepared in different matrices. .......................................................................................... 89
Table XIV. The two best signal correction strategies for each sample matrix. ............... 92
vii
Table XV. Comparison of MFC with EC using the same individual Q values while
determining Cu in River Sediment A.............................................................................. 107
Table XVI. Accuracy comparison between MFC and EC. ............................................ 111
Table XVII. Analyte percent recoveries (%) from spiked concentrations in water and food
samples analyzed by MIP OES using MFC or EC. ........................................................ 114
viii
FIGURES
Figure 1. Microwave cavity structure of the Hammer arrangement. ................................. 6
Figure 2. Representation of the Hammer cavity and resonant iris used in the commercial
MIP OES instrument by Agilent Technologies (used in models 4100, 4200, and 4210 MP-
AES). ................................................................................................................................... 7
Figure 3. Schematic representation of the detection system in a monochromator-based MIP
OES instrument (Agilent 4200 MP-AES)........................................................................... 9
Figure 4. Schematic representation of the pre-optics in an Agilent 4200 MP-AES
instrument. .......................................................................................................................... 9
Figure 5. Helios Rapid Ashing unit, courtesy of CEM Corporation................................ 47
Figure 6. Sample basket used with the HRA prototype. .................................................. 47
Figure 7. Sample “sandwich” setup using an aluminum foil at the bottom and a quartz fiber
pad on top loaded in the sample basket. ........................................................................... 52
Figure 8. Effects of nebulization gas flow rate on Mg II / Mg I (a), and N2+ / OH (b) in
MIP OES. .......................................................................................................................... 79
Figure 9. Effects of sodium concentration on the Mg II / Mg I and N2+ / OH signal ratios
at nebulization gas flow rates of 0.6 L min-1 (a) and (b), and 1.0 L min-1 (c) and (d). ..... 80
Figure 10. Effects of nebulization gas flow rate on (a) Mg II (280.271 nm) and Mg I
(285.213 nm), and (b) N2+ (391.439 nm) and OH (308.970 nm). .................................... 80
Figure 11. MIP OES spectra for the OH molecular species (band peak at 308.970 nm).
Each spectrum corresponds to a different plasma / sample introduction condition ......... 82
Figure 12. Correlation between the Mg II / Mg I and N2+ / OH signal ratios in the presence
of sodium at different nebulization gas flow rates (0.6 - 1.2 L min-1) .............................. 84
ix
Figure 13. Relationship between average analyte percent recovery and N2+ / OH signal
ratio in different matrices .................................................................................................. 87
Figure 14. Multi-flow calibration plots for determining (A) Cu in River Sediment A, and
(B) Mn in Tomato Leaves. .............................................................................................. 109
Figure 15. Long-term stability of MFC ......................................................................... 118
Figure S 1. Molecular emission spectra for CN recorded with the plasma off, or as a 1 %
v/v HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7 L
min-1. ............................................................................................................................... 128
Figure S 2. Molecular emission spectra for N2 recorded with the plasma off, or as a 1 %
v/v HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7 L
min-1. ............................................................................................................................... 128
Figure S 3. Molecular emission spectra for N2+ recorded with the plasma off, or as a 1 %
v/v HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7 L
min-1. .............................................................................................................................. 129
Figure S 4. Molecular emission spectra for OH recorded with the plasma off, or as a 1 %
v/v HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7 L
min-1. ............................................................................................................................... 129
Figure S 5. Molecular emission spectra for CN recorded as 1 % v/v HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP. ................................................................................................. 130
Figure S 6. Molecular emission spectra for N2 recorded as 1 % v/v HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP. ................................................................................................. 130
x
Figure S 7. Molecular emission spectra for N2+ recorded as no aqueous solution (no spray
chamber), distilled-deionized water, or 1 % v/v HNO3 (NGFR = 0.7 L min-1) was
introduced into the MIP. ................................................................................................. 131
Figure S 8. Molecular emission spectra for OH recorded as 1 % v/v HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP. ................................................................................................. 131
Figure S 9. Absolute emission signal percent change as Na concentrations in the 0 - 1000
mg L-1 range were introduced into the MIP at a NGFR of 0.6 L min-1. ......................... 133
Figure S 10. Absolute emission signal percent change as Na concentrations in the 0 - 1000
mg L-1 range were introduced into the MIP at a NGFR of 1.0 L min-1. ......................... 133
Figure S 11. Absolute signal ratio percent change as Na concentrations in the 0 - 1000 mg
L-1 range were introduced into the MIP at a NGFR of 0.6 L min-1. ............................... 134
Figure S 12. Absolute signal ratio percent change as Na concentrations in the 0 - 1000 mg
L-1 range were introduced into the MIP at a NGFR of 1.0 L min-1. ............................... 134
xi
LIST OF ABBREVIATIONS
Abbreviation Definition
ACS American Chemical Society
AES Atomic Emission spectrometry (synonymous with OES)
APER Average Percent of Recovery
CCD Charge-Coupled Device
CMP Capacitively-Coupled Microwave Plasma
CRM Certified Reference Material
EC External Standard Calibration
EDTA Ethylene Diamine Tetra-Acetate
EGCM External Gas Control Module
EIE Easily-Ionizable Element
FAAS Flame Atomic Absorption Spectrometry
FAES Flame Atomic Emission Spectrometry
FBN Flow-Blurring Nebulizer
FDA U.S. Food and Drug Administration
HPLC High-Pressure Liquid Chromatography
HRA Helios Rapid Ashing
HR-CS High-Resolution, Continuum Source
ICP Inductively-Coupled Plasma
IS Internal Standard
IUPAC International Union of Pure and Applied Chemistry
LED Light-Emitting Diode
LIBS Laser-Induced Breakdown Spectroscopy
LOD Limit of Detection
MAD Microwave-Assisted Digestion
MEC Multi-Energy Calibration
MFC Multi-Flow Calibration
MICal Multi-Isotope Calibration
xii
MICAP Microwave-Sustained, Inductively-Coupled, Atmospheric-Pressure
Plasma
MINDAP Microwave-Induced Nitrogen Discharge at Atmospheric Pressure
MIP Microwave-Induced Plasma
MP Microwave Plasma
MS Mass Spectrometry
MSC Multispecies Calibration
NGFR Nebulization Gas Flow Rate
NIST National Institute of Standards and Technology
OES Optical Emission Spectrometry
PID Proportional-Integral-Derivative
PN Pneumatic Nebulizer
Q Nebulization Gas Flow Rate
RF Radio Frequency
RSD Relative Standard Deviation
SA Standard Additions
SDA Standard Dilution Analyis
SRM Standard Reference Material
TIA Torche a Injection Axiale (Axial-Injection Torch)
TIAGO Torche a Injection Axiale sur Guide d’Ondes (Axial-Injection
Waveguide Torch)
USN Ultrasonic Nebulizer
VBPN V-Groove Babington-type Pneumatic Nebulizer
xiii
ABSTRACT
Microwave-induced plasma optical emission spectrometry (MIP OES) is a
technique within Atomic Spectrometry which is rapidly growing, following the relatively
recent introduction of a complete commercial instrument based on the Hammer-cavity
MIP. It offers advantages of low cost and simple operation, as well as the ability to run on
air, enabling it to be used in remote areas or places with underdeveloped infrastructure.
Due to the relative novelty of the technique, intensive method development for specific
sample types is necessary to expand the utility of the technique and compensate for some
of its limitations, such as its relatively low robustness and sequential detection. In addition,
developments in plasma diagnostics and novel calibration strategies are also important to
help increase the prominence and utility of the technique. The present research covers
several such applications and efforts to improve instrumentation performance, as well as
more fundamental studies of the properties of the plasma.
In the first project, a method was developed to determine concentrations of Ca, K,
and Na in various soft drinks. Addition and recovery experiments were used to evaluate
the accuracy of the method. This study shows the ability of the technique to withstand
relatively complex matrices, with no sample preparation other than simple dilution, and
produce accurate results with traditional calibration methods.
The second project involved development of a rapid dry-ashing technique to
prepare samples for analysis using MIP. Calcium, Fe, K, Mg, Na and Zn were determined
in complex-matrix samples such as tomato leaves, cheese, butter, peanut butter, infant
formula and biodiesel samples. General agreement was also found between MIP OES
xiv
results from samples decomposed either by the dry-ashing method or by a conventional
microwave-assisted digestion procedure, and accurate results were found when applying
the new, simpler method in certified reference material analyses.
In the third project, several molecular species, naturally occurring in the plasma,
were evaluated for their use as a plasma diagnostic tool. The N2+ / OH emission intensity
ratio was evaluated for identifying the best instrumental operating conditions in MIP OES.
Aluminum, Ba, Mn, Sr and Zn (analytes), and high concentrations of C, Na, Ca, HNO3 and
HCl (sample matrices) were used as models to investigate the effects of complex matrices
on analyte recoveries. The N2+ / OH signal ratio was more sensitive to changes in plasma
conditions than the traditionally-used Mg II / Mg I ratio. Some other advantages include
real-time monitoring capabilities, and the possibility of independently tracking variations
in both plasma and sample introduction. Significant improvements in accuracy were
achieved by employing the analyte-to-molecular species signal ratio, or their product, for
calibration.
In the final project, a novel calibration method, multi-flow calibration (MFC), was
proposed. This strategy involves the use of multiple nebulization gas flow rates in the
analysis to mitigate error caused by employing suboptimal sample introduction conditions,
and to eliminate the need for optimization of conditions. The new calibration method was
applied to the determination of Cr, Cu, Fe, and Mn in water and food samples. Addition-
recovery experiments and certified reference materials were used to validate the method.
Multi-flow calibration presents comparable or superior accuracy to the traditional external
standard calibration (EC), and offers simpler sample preparation than EC, requiring only a
single standard and no modification of the sample introduction equipment.
1
CHAPTER I
INTRODUCTION
REVIEW OF MICROWAVE-INDUCED PLASMA OPTICAL EMISSION
SPECTROMETRY
Charles B. Williams and George L. Donati
This chapter is based on unpublished material originally submitted as part of a
review article published in the Journal of Analytical Atomic Spectrometry, 2017, 32,
1283-1296. This material was excised because it was deemed off-topic from the main
subject of ICP OES. Stylistic variations are due to the requirements of the journal. First
authorship is shared between Charles B Williams and George Donati.
2
BACKGROUND: ATOMIC SPECTROMETRY
Atomic spectrometry involves the measurement of elemental concentrations in
various samples. Common methods used in this field include flame atomic absorption
spectrometry (commonly called AA), flame atomic emission spectrometry (FAES),
graphite furnace atomic absorption spectrometry (GFAAS), inductively-coupled plasma
optical emission spectrometry (ICP OES, also called ICP AES, for atomic emission
spectrometry), and inductively-coupled plasma mass spectrometry (ICP-MS).1 An
increasingly significant method in the field is microwave-induced plasma optical emission
spectrometry (MIP OES).2
The differences between these approaches generally break down to two factors: the
method of atomization/excitation, and the method of detection. Atomization/excitation
sources include flames, graphite furnaces, and plasmas, and detection can be accomplished
by either absorption, emission, or mass spectrometry. Generally, flame and graphite
furnace sources are associated with atomic absorption (using either a continuum-source or
line-source light), whereas plasma sources are associated with either atomic emission or
mass spectrometric detection. Flame and graphite furnace sources, as well as atomic
absorption and mass spectrometry will not be discussed in further detail in this text.
In a typical optical emission spectrometry (OES) system, atoms are excited in a
high-temperature environment such as a flame or plasma. As the electrons relax to a lower
energy state, a photon is emitted. Emitted photons from the analyte are measured using an
electronic detector and the intensity of the emission is directly proportional to the
concentration of the analyte in the sample.1 Each element has a characteristic set of
3
emission wavelengths, so with sufficient resolution, multielement samples can be analyzed
both qualitatively and quantitatively.
The most common atomization/excitation source in atomic spectrometry is the
inductively-coupled plasma (ICP). An ICP is generated from argon gas by a radio-
frequency electromagnetic field at either 27 or 40 MHz.3 This produces a highly stable
plasma with a high electron density and high temperature (7,500-10,000 K). This provides
a good source for atomization, ionization, and excitation of analytes. Because of the high
plasma temperature, most elements are ionized in an ICP.4 Some of its main limitations are
its relatively poor tolerance for organic solvents, and its expense, both in up-front cost and
running cost.
Microwave-induced plasma (MIP) can be generated from a variety of gases, but N2
is used most commonly. Microwave energy is generated from a magnetron, and then
transmitted and amplified by a waveguide cavity.2 Microwave-induced plasma is generally
lower in temperature than ICP, resulting in a predominance of atomic species, rather than
ionic ones. However, MIP is generally less expensive than ICP both in up-front and running
costs. In addition, it is more tolerant of organic solvents and other difficult matrices,
manifesting as matrix effects rather than extinguishing the plasma.
Matrix effects present one of the most significant challenges in atomic
spectrometry. They are caused by concomitant species within the sample which influence
the signal of the analyte, resulting in inaccurate analysis.5–7 Species which commonly cause
matrix effects include easily-ionizable elements (EIEs) such as Na and Ca, inorganic acids
at high concentrations, including HCl and HNO3, and high concentrations of carbon
species.
4
INSTRUMENTATION DEVELOPMENTS IN MIP OES
Microwave-induced plasmas have historically evolved as atomization and
excitation / ionization sources that could potentially be used as alternatives to traditional
ICPs in atomic spectrometry. There are two main approaches to generating microwave
plasmas. When microwave radiation is applied to an electrode, an electric field and a
perpendicular magnetic field are produced. In contact with a neutral gas and seeded
electrons, these fields will generate and sustain a so-called capacitively-coupled microwave
plasma (CMP) at the tip of the electrode. Alternatively, one can generate a microwave-
induced plasma by applying microwave radiation into a resonant structure filled with a
neutral gas. The energy associated with the standing wave inside the resonant cavity is
transferred to the gas, and the electric and magnetic fields generated produce and sustain
the plasma.8 The CMP was described in 1951 by Cobine and Wilbur,9 and the first use of
a 2.45 GHz microwave discharge as excitation source for emission spectrometry was
reported in 1965.10 The Beenakker cavity was introduced ten years later and used in MIP
OES for elemental analysis.11–13 For a historical perspective on the evolution of the
different instrumental arrangements used for microwave plasma generation and its
application in atomic spectrometry, the reader is referred to a thorough review recently
published by Jankowski and Reszke.8 Additional information can also be found in a book
by the same authors.2
The early instruments presented low sensitivity and had difficulty handling liquid
samples, which stemmed from the low microwave applied power (typically 200 - 300 W)
and the plasma non-toroidal shape. Later designs minimized some of these problems, but
low plasma temperatures and relatively poor interaction between plasma and liquid sample
5
often caused significant matrix effects.14–19 As discussed by Jankowski and Reszke,2 five
microwave plasma instruments have been commercially available in the past, but none of
them have generated enough interest to be widely adopted. A more successful system was
released in 2012 by Agilent Technologies (4100 MP-AES). The new MIP OES instrument
is based on a Hammer cavity with a resonant iris and runs on N2 gas.20,21 It has been
successfully used in several applications, and will be discussed in more detail in the next
paragraphs.
Microwave radiation is usually transmitted through a hollow rectangular structure
known as a waveguide. The electric field component of a typical microwave is aligned with
the short axis of the rectangular structure (i.e. height), while the magnetic field component
aligns with its longer axis (i.e. width). In both cases, the field is supported by induced
currents flowing on the walls of the metallic waveguide. The difficulties associated with
controlling the plasma central channel, as well as its non-toroidal shape, are directly related
to the fact that most MIPs are sustained by the microwave’s electric field component. In
the Hammer arrangement,22 this issue is resolved by positioning the magnetic field
component parallel to the torch. It then induces an electric field, which accelerates electrons
and ions causing collisions and generating and sustaining a doughnut-shaped MIP. To
improve sensitivity and minimize matrix effects, two additional modifications are adopted:
(i) N2 replaces monoatomic gases such as Ar and He as the plasma gas, and (ii) the torch
is placed in a resonant iris inside the waveguide. Plasmas composed of diatomic molecules
such as N2 have lower electron densities, which results in a larger skin depth (i.e., more
interaction between the plasma and the sample aerosol). The resonant iris further improves
coupling between the plasma outer layer and the cooler central channel by changing the
6
circular MIP into an elliptical shape. The iris structure also increases the intensity of the
magnetic field, which contributes to higher plasma temperatures. Figure 1 shows a picture
of the Hammer arrangement with the resonant iris, and Figure 2 shows a schematic
representation of the commercial MIP OES system based on this instrument design
(Agilent MP-AES). In the resonant iris, the axial magnetic field (small blue arrows in
Figure 2) and the transverse electric field (large red arrows in Figure 2) are combined to
render an elliptical form to the plasma.
Figure 1. Microwave cavity structure of the Hammer arrangement. Reprinted from Ref. 16. Copyright© 2017 Elsevier B.V. or its licensors or contributors. Reprinted with permission.
7
Figure 2. Representation of the Hammer cavity and resonant iris used in the commercial MIP OES instrument by Agilent Technologies (used in models 4100, 4200, and 4210 MP-AES). Small blue arrows and large red arrows represent the magnetic and electric fields, respectively. Copyright© 2017 Agilent Technologies, Inc. Courtesy of Agilent Technologies, Inc.. Reprinted with permission.
8
The commercial MIP OES instrument is based on a Czerny-Turner monochromator
(600 mm focal length), and a back thinned Peltier-cooled charge-coupled device (CCD)
detector. Figure 3 shows a schematic representation of the detection system. The
monochromator is composed of a fixed 2.5 mm high x 19 µm wide entrance slit, a spherical
parabolic collimating mirror (75 mm diameter), a 90 x 90 mm holographic diffraction
grating (2400 lines / mm and blazing angle optimized at 250 nm), and a spherical parabolic
focusing mirror (87 mm diameter). Considering, for example, an incident angle of 0o, this
system has a spectral bandpass of 0.011 nm and a resolving power of approximately 12,000
at 250 nm. The CCD detector is composed of a 532 wide x 128 high pixel array (each pixel
is 24 x 24 µm). It covers a spectral range between 176 and 1100 nm, and is Peltier-cooled
to 0 oC to minimize dark current noise. Figure 4 shows additional details of the pre-optics.
It is composed of a quartz entrance window (EW), two toric mirrors (M1 and M2), a
stepper-motor-positioned filter wheel assembly (FW), a stepper-motor-positioned flat
mirror (M3), and an exit slit (ES). The entrance window keeps the pre-optics separated
from the plasma and the exhaust system to minimize potential contamination. It also allows
for complete purging of the pre-optics system with N2 to prevent O2 absorption at the UV
region. Bandpass filters in the filter wheel assembly work in concert with the
monochromator to improve resolution. Wavelengths in the 160 - 320, 320 - 530, or 530 -
940 nm range are covered by moving the FW to positions zero (no filter), 1 (UV filter), or
2 (orange filter), respectively. A fourth position blocks the incoming light from the plasma,
and it is used in initialization routines to optimize the system’s dark current. The flat mirror
(M3) enables imaging of different regions of the plasma.
9
Figure 3. Schematic representation of the detection system in a monochromator-based MIP OES instrument (Agilent 4200 MP-AES). Copyright© 2017 Agilent Technologies, Inc. Courtesy of Agilent Technologies, Inc. Reprinted with permission.
Figure 4. Schematic representation of the pre-optics in an Agilent 4200 MP-AES instrument. Copyright© 2017 Agilent Technologies, Inc. Courtesy of Agilent Technologies, Inc. Reprinted with permission. M1, M2, M3 are mirrors, FW is the filter wheel, EW is the entrance window, and ES is the entrance slit to the optics.
10
Different from an echelle-CCD spectrograph, this monochromator-based system
detects analytical signals sequentially. As a consequence, the more analytes monitored, the
longer the analysis. On the other hand, while a compromise condition is used for all
elements in ICP OES, determinations using this sequential MIP OES instrument take place
at optimal nebulization gas flow rate and plasma viewing position for each individual
analyte. Plasma imaging is carried out axially, and is based on stepper motor positioning
of M3 (usually varying between -120 and 120 steps). Most determinations, however, are
carried out at position zero, which approximately corresponds to the center of the plasma.20
One of the main limitations of this MIP OES instrument is that the microwave applied
power is fixed at 1000 W and cannot be changed by the user. Depending on the position of
M3 and the nebulization gas flow rate, plasma temperature and electron number density
(ne) values vary between 4220 - 5360 K, and 0.47 - 3.72 x 1013 cm-3, respectively. Because
of the lower temperatures and ne values when compared with a typical Ar ICP, most
analytes are determined using atomic lines. As expected, the plasma is also less robust than
an ICP (Mg II / Mg I = 0.26 - 2.01), which requires matrix-matching calibration methods
(e.g. standard additions) to ensure adequate accuracy when analyzing some complex matrix
samples.20
Other non-commercial MIP instrumental arrangements have recently been
described in the literature.23–29 A list including a miniaturized, a low-pressure-operated,
and a portable system is presented in Table I. One of the most interesting new
arrangements is the microwave-sustained, inductively-coupled, atmospheric-pressure
plasma (MICAP).29 In this system, a typical ICP quartz torch is positioned concentrically
in a high-purity, high-density alumina (Al2O3) resonator ring. The microwave field
11
generated by a magnetron operating at 1000 W is directed toward the resonator by an
aluminum waveguide. It then reaches an inductive-coupling iris oriented perpendicularly
to the torch. Polarization currents are induced within the resonator as it couples with the
iris, and a corresponding oscillating magnetic field, oriented parallel to the torch, is
generated. The magnetic field induces an oscillating electric field that, similarly to a RF-
powered ICP, accelerates electrons and ions to generate and sustain a toroidal plasma. The
MICAP system runs on N2 or air and is tolerant to solvent loading. It can accept volatile
organic solvents and dissolved solids up to 3% w w-1. For a N2 plasma and determinations
in radial-view using an ultrasound nebulizer and a membrane desolvator, limits of detection
for Al, Ca, Cd, Co, Cr, Fe, Mg, Mn, and Pb were calculated in the 0.03 - 70 µg L-1 range,
with relative standard deviation (RSD) values ranging from 0.7 to 2.0 %. These analytical
characteristics are generally similar to a radial-viewing ICP OES. In fact, different from
the commercial MIP OES and closer to an ICP behavior, the most intense analytical signals
observed for the MICAP arrangement were recorded for ionic lines.
12
Tab
le I.
Rec
ently
des
crib
ed n
on-c
omm
erci
al M
IP O
ES sy
stem
s.
Mic
row
ave
syst
em
Plas
ma
gas
flow
ra
te
(L
min
-1)
Mic
row
ave
appl
ied
pow
er (W
)
Ana
lyte
s Sa
mpl
es
Com
men
ts
Ref
eren
ce
Mic
rost
rip M
IP
He,
0.2
5 40
Br
, C, C
l V
olat
ile h
alog
enat
ed
orga
nic
com
poun
ds
Min
iatu
re M
IP u
sed
as a
de
tect
or fo
r gas
ch
rom
atog
raph
y.
23
MIP
torc
h in
a T
E
rect
angu
lar c
avity
Ar,
0.2
- 1.0
10
0 - 1
80
Ba, C
a,
Cd,
Cu,
Fe, M
g,
Mn,
Ni,
Sr, Z
n
SRM
164
8 (U
rban
Pa
rticu
late
Mat
ter)
, IA
EA
336
(Lic
hen)
, SR
M 2
710
(Mon
tand
Soi
l), IN
CT-
SBF-
4 (S
oy B
ean
Flou
r)
Air-
cool
ed M
IP o
pera
ted
at
low
pre
ssur
e (0
.8 b
ar).
24
MIP
to
rch
in
a
cylin
dric
al c
avity
Ar,
0.4
- 0.5
15
0 A
s, C
d,
Hg,
Mn,
Ni,
P, P
b,
and
Sr
Aqu
eous
solu
tions
in 2
% v
v-1
HN
O3
OES
/ ca
vity
ring
dow
n sp
ectro
scop
y du
al m
ode.
25
MIP
tor
ch i
n an
Oka
mot
o ca
vity
N2,
14.0
; O2,
0 -
1.5
800
- 100
0 C
r A
queo
us so
lutio
ns
prep
ared
in d
ilute
d H
Cl
Ar a
t 0.5
L m
in-1
use
d as
ca
rrier
gas
. Stu
dy o
f pla
sma
char
acte
ristic
s and
com
paris
on
with
ICP.
27
13
MIP
tor
ch i
n an
Oka
mot
o ca
vity
N2,
14.0
; O2,
0 -
1.8
800
- 900
M
n A
queo
us so
lutio
ns
prep
ared
in d
ilute
d H
Cl
Ar a
t 0.5
L m
in-1
use
d as
ca
rrier
gas
. Stu
dy o
f pla
sma
char
acte
ristic
s and
com
paris
on
with
ICP.
26
MIP
tor
ch i
n an
Oka
mot
o ca
vity
N2,
14.0
80
0 - 1
500
Fe
Aqu
eous
solu
tions
pr
epar
ed in
dilu
ted
HC
l N
2 at 0
.5 L
min
-1 u
sed
as
carri
er g
as. I
nves
tigat
ion
of F
e ex
cita
tion
/ ion
izat
ion
mec
hani
sms i
n M
IP a
nd IC
P us
ing
the
Boltz
man
n pl
ot
met
hod.
28
Mic
row
ave-
sust
aine
d IC
P in
a
diel
ectri
c
reso
nato
r rin
g
N2 o
r air,
15.
0 -
18.0
1000
A
l, C
a,
Cd,
Co,
Cr,
Fe,
Mg,
Mn,
Pb, S
r
Aqu
eous
solu
tion
in 0
.1 M
H
NO
3; m
iner
al o
il,
kero
sene
, o-x
ylen
e,
tolu
ene,
ace
toni
trile
, he
xane
, met
hano
l
A ty
pica
l IC
P qu
artz
torc
h is
po
sitio
ned
conc
entri
cally
in a
hi
gh-p
urity
, hig
h-de
nsity
al
umin
a (A
l 2O3)
reso
nato
r rin
g.
The
mic
row
ave
field
indu
ces
pola
rizat
ion
curre
nts i
n th
e re
sona
tor a
nd g
ener
ates
a
mag
netic
fiel
d pa
ralle
l to
the
torc
h. T
he m
agne
tic fi
eld
indu
ces a
n el
ectri
c fie
ld th
at
sust
ains
the
plas
ma.
29
14
MIP OES APPLICATIONS FOCUSED ON SAMPLE INTRODUCTION AND
SAMPLE PREPARATION
Similar to ICP OES, many recent works on MIP OES describe new strategies for
sample introduction. Among these, most involve hydride generation and multi-channel
spray chambers, also known as multi-mode sample introduction systems.24,30–37 Typically,
in these systems, a multi-channel peristaltic pump introduces independent solutions into
the spray chamber, where efficient mixing and chemical reactions take place. Chemical
vapor generation is particularly useful in MIP OES because of the plasma’s lower
temperatures (ca. 5000 K)2,20 when compared with an ICP (6000 - 1000 K).3,4 Introducing
the analytes as gaseous species has three main beneficial effects on sensitivity and
accuracy: (i) sample introduction is significantly improved because it does not depend on
nebulization efficiency, (ii) with no solvent to vaporize, more plasma energy is available
for atomization and excitation, and (iii) negligible matrix effects.38,39 The simplest multi-
mode arrangements use one capillary channel to introduce the sample and a second one for
a solution containing the chemical-vapor-generating species.30–33 Due to its reducing
potential, NaBH4 (stabilized in solution with NaOH) is the most common reagent in HG
applications.38 LODs for this arrangement were calculated in the 1 - 10 µg L-1 range for
As, Ag, Bi, Cd, Cu, Ge, Hg, Mn, Ni, Pb, Rh, Se, Sn and Zn, with RSDs between 8 and 12
%. To improve sensitivity, triple-channel spray chambers typically introduce an acid
solution (e.g. HCl) in addition to the sample and the reducing agent (NaBH4 or
SnCl2).33–35 LODs and RSDs in this case were in the 0.3 - 9 µg L-1, and 5 - 9 % ranges,
respectively. In the quadruple-capillary sample introduction arrangement, acid, reducing
agent, sample, and internal standard solutions can be introduced separately into the spray
15
chamber to improve precision.36 By using internal standardization, RSDs were calculated
in the 2 - 3 % range, with LODs between 1 and 5 µg L-1 for As, Bi, Sb, Se and Sn. Most of
the recently described multi-mode systems use an USN, rather than the traditional
pneumatic nebulization (PN). Although more expensive and prone to memory effects, USN
produces smaller-diameter droplets and a more homogeneous tertiary aerosol, which
results in greater nebulization efficiency when compared with PN (ca. 20 % for USN cf.
ca. 5 % for PN).3 As observed in ICP OES applications, the increased nebulization,
combined with a finer aerosol, allows for more efficient contact and improved reaction
rates between analytes and reagents within the spray chamber. It also enables the analysis
of smaller sample volumes. Matusiewicz and Ślachciński, for example, used all four
channels of a quadruple-capillary system to introduce the sample solution. With more
sample effectively reaching the plasma, low-volume aliquots of liver, sediment, soil and
water samples were readily analyzed.37
To improve sensitivity and sample throughput, some procedures digest the sample
in-flow, or introduce it as a slurry or a powder.40–44 Jankowski et al., for example,
preconcentrated fluoride using a zirconia-based sorbent material, then dried the solid
mixture and introduce it into a lab-made continuous powder introduction chamber at the
base of a MIP torch.40 In a similar procedure, inorganic selenium was extracted from
aqueous samples and preconcentrated using bacteria immobilized on silica gel.41
Ślachciński determined several elements in liver, sediment, soil and coal ash by applying
ultrasound radiation to homogenize the sample slurry before introducing it into a lab-made
MIP OES using two different nebulizers: a V-groove Babington-type nebulizer (VBPN)
and a flow-focusing pneumatic nebulizer.42 To minimize matrix effects, Matusiewicz and
16
Ślachciński digested sample slurries of dogfish liver, milk powder, lichen, barley and
cinnamon immediately prior to analysis by using an in-line flow solubilization system
based on electromagnetic induction heating and a commercial ultrasonic nebulizer.43 In this
system, the sample slurry and nitric acid were pumped through Teflon tubing into a coil of
acid-resistant steel pipe, which was wrapped around a ferrite core. The mixture was then
heated and solubilized in-flow. A similar method is described by the same authors in
another paper, in which a UV micro-reactor is used to digest biological samples in-flow
prior to MIP OES analysis.44 As discussed by the authors, the main advantages of using
these strategies are the lower cost when compared with a conventional microwave-assisted
digestion, and potentially less sample contamination and analyte loss due to less sample
manipulation.
Even though MIP OES is not as sensitive as ICP-MS, it has also been recently used
in chemical speciation.45–47 Matusiewicz and Ślachciński used a microchip-based capillary
electrophoresis system to separate and determine As3+ and As5+, and Cu2+ and
Cu(EDTA)22- in water using a non-commercial MIP OES system.45,46 Barrientos et al.
determined Se and selenomethionine in yeast by coupling a commercial MIP OES system
with HPLC.47
Other sample preparation approaches such as dry-ashing, subcritical microwave-assisted
extraction with water, and ultrasound-assisted extraction were also used to improve
sensitivity and accuracy in MIP OES determinations.48–50 A summary of recent
developments in MIP OES involving sample introduction and sample preparation is
presented in Table II.
17
Tab
le II
. Rec
ent s
ampl
e in
trodu
ctio
n st
rate
gies
and
sam
ple
prep
arat
ion
proc
edur
es u
sed
in M
IP O
ES.
Sam
ple
intr
oduc
tion/
prep
arat
ion
Ana
lyte
s Sa
mpl
es
Com
men
ts
Ref
.
Dua
l cap
illar
y; u
ltras
onic
neb
uliz
er
/ vap
or g
ener
atio
n
Ag,
As,
Au,
Bi,
Cd,
Cu,
Ge,
Hg,
Mn,
Ni,
Pb, P
d,
Rh,
Sb,
Se,
Sn,
Zn
Biol
ogic
al ti
ssue
, hai
r,
soil,
food
, wat
er, w
ine
Seco
nd
capi
llary
in
trodu
ces
redu
cing
ag
ent
(NaB
H4
or
SnC
l 2) fo
r vap
or g
ener
atio
n.
30–3
3
Trip
le
capi
llary
; ul
traso
nic
nebu
lizer
/ va
por g
ener
atio
n
Ag,
As,
Au,
Bi,
Ge,
Hg,
Pd, P
t, R
h, S
b, S
e, S
n
Biol
ogic
al ti
ssue
, soi
l,
wat
er, w
ine,
ferti
lizer
Third
cha
nnel
int
rodu
ces
HC
l
to im
prov
e se
nsiti
vity
.
34,3
5,51
Qua
drup
le
capi
llary
; ul
traso
nic
nebu
lizer
/ va
por g
ener
atio
n
As,
Ba, B
i, C
a,
Cd,
Cu,
Fe,
Ge,
M
g, M
n, P
b, P
b,
Sb, S
e, S
n, S
r, Te
, Tl
, Zn
Wat
er,
Biol
ogic
al
tissu
e,
soil,
or
gani
c
solv
ents
Add
ition
al c
hann
els
used
for
vapo
r ge
nera
tion
and
inte
rnal
stan
dard
izat
ion.
36,3
7
Con
tinuo
us p
owde
r int
rodu
ctio
n F
Wat
er
Ana
lyte
pr
econ
cent
ratio
n on
zirc
onyl
nitr
ate.
40
Bact
eria
pr
econ
cent
ratio
n,
then
cont
inuo
us p
owde
r int
rodu
ctio
n
Se
Wat
er, b
eer
Bact
eria
im
mob
ilize
d on
sili
ca
gel.
41
18
Slur
ry m
icro
-sam
plin
g Ba
, Ca,
Cd,
Cu,
Fe,
Mg,
Mn,
Pb,
Sr,
Zn
Biol
ogic
al
tissu
e,
wat
er, s
oil
Cal
ibra
ted
by
stan
dard
addi
tions
, sl
urrie
s so
nica
ted
prio
r to
anal
ysis
.
42
Pres
suriz
ed
flow
so
lubi
lizat
ion
usin
g el
ectro
mag
netic
in
duct
ion
heat
ing
Ba, C
a, C
d, C
u, F
e, M
g,
Mn,
Na,
Pb,
Sr,
Zn
Biol
ogic
al ti
ssue
U
ses i
n-lin
e di
gest
ion.
43
Mic
roflu
idic
; m
icro
chip
-bas
ed
phot
o-m
icro
-reac
tor,
and
ultra
soni
c
nebu
lizat
ion
Ba, C
, Ca,
Cd,
Cu,
Fe,
Li, M
g, M
n, P
b, S
r, Zn
Urin
e,
wat
er,
Biol
ogic
al fl
uids
Sam
ple
flow
rate
of 9
µL
min
-1.
45
Mic
roch
ip c
apill
ary
elec
troph
ores
is
As3+
and
As5+
, and
Cu2+
and
Cu(
EDTA
) 22-
Wat
er
Buff
er:
Boric
aci
d +
CTA
B;
Sam
ple
Flow
Rat
e 0.
5 µL
min
-
1 .
45,4
6
Ion-
pair
reve
rsed
pha
se H
PLC
with
hydr
ide
gene
ratio
n
Se, S
eMet
Y
east
N
aBH
4 in
trodu
ced
in-li
ne.
Yea
st
sam
ples
re
cove
red
by
47
19
cent
rifug
atio
n, th
en r
eflu
xed
in
met
hane
sulfo
nic
acid
for 1
6 h.
Rap
id d
ry a
shin
g C
a, F
e, K
, Mg,
Na,
Zn
Food
, bio
dies
el
Sam
ples
ash
ed i
n A
l fo
il an
d
quar
tz fi
ber p
ads a
t 500
°C fo
r 3
min
, the
n ex
tract
ed in
10
% H
Cl
befo
re a
naly
sis.
48
Soni
catio
n /
Hea
ting
for
met
al
extra
ctio
n
Cd,
Cr,
Cu,
Mn,
Pb,
Zn
Inor
gani
c fe
rtiliz
er
Soni
cate
d in
50
% v
v-1
HC
l for
10 m
in a
nd h
eate
d at
85
°C
befo
re a
naly
sis.
49
Mic
row
ave-
assi
sted
su
bcrit
ical
wat
er e
xtra
ctio
n
Ba, C
a, C
u, F
e, M
g, M
n,
Na,
Pb,
Sr,
Zn
Plan
t tis
sue
Wat
er a
cidi
fied
with
100
µL
of
HN
O3.
Ana
lyte
ex
tract
ion
at
280
o C
and
90
bar.
Hig
h
pres
sure
ke
eps
wat
er
in
the
liqui
d ph
ase
(sub
criti
cal).
50
20
RECENT APPLICATIONS USING THE HAMMER-CAVITY MIP OES
Although more prone to matrix effects, the N2 MIP is also more tolerant to organic
solvent load than an Ar ICP. Amais et al. have used the commercial Hammer-cavity MIP
OES instrument to determine Cr, Ni, Pb and V in gasoline and ethanol fuel, and Si in diesel
and biodiesel employing external standard calibration.52,53 No sample preparation other
than simple dilution in 1 % v v-1 HNO3 (ethanol samples) or ethanol (diesel and biodiesel
samples) was required for a sample introduction system composed of an inert FBN and a
cyclonic spray chamber. For gasoline, sample and reference standard micro-emulsions
were prepared in 1-propanol to minimize matrix effects. Air was introduced into the plasma
to prevent carbon deposition on the torch and the optics. Accuracy was checked by addition
and recovery experiments, with values in the 91 - 108 % range in most cases. Some
recoveries were outside the 10 % error range for Cr (86 %), Ni (123 %), and Pb (86 %) in
gasoline, and Ni (84 %) in ethanol fuel. The LODs were calculated as 20 µg L-1 for Si, and
between 0.3 and 60 µg L-1 for Cr, Ni, Pb and V. Nelson et al. used the same MIP OES
instrument to analyze several samples of crude oil, and compared the results with values
obtained with ICP OES and ICP-MS.54,55 Sample solutions were prepared by dilution with
o-xylene and homogenization in a mechanic shaker for 30 min. Matrix-matching and
internal standardization were adopted to minimize matrix effects. The standard reference
solutions were prepared with organosoluble standards diluted in o-xylene, mineral oil to
match the sample matrix, a dispersant (oronite) to ensure homogeneous and stable
solutions, and Sc as IS. A low flow of air was added to the N2 MIP to prevent carbon build-
up on the torch and ensure plasma stability. The LODs for Ca, Fe, K, Na, Ni and V were
all < 0.07 mg kg-1, which are comparable with values calculated for ICP OES (i.e. < 0.02
21
mg kg-1). Accuracies were also comparable with ICP OES and ICP-MS, with spiked sample
recoveries between 93 and 107 %, and between 102 and 110 % for a certified reference
material of fuel oil.
MIP OES has also been successfully used to analyze soil, geological samples,
vinegar, animal feed and fertilizers.56–59 Niedzielski et al. described a high throughput
procedure for assessing soil fertility based on Mehlich-3 extraction and Ca, K, Mg and P
determination by MIP OES.56 Up to 100 samples h-1 were analyzed, with LODs between
0.06 and 0.9 mg L-1, RSDs in the 1.0 - 4.6 % range, and accuracies between 92 and 107 %.
In another work, 23 elements in geological samples were determined by MIP OES after
extraction with HF (to assess total concentration), aqua regia (quasi-total concentrations),
or HCl (acid leachable fraction). The LODs were calculated between 0.001 and
0.1 mg L-1, with RSDs and accuracies in the 0.20 - 1.37 %, and 85 - 115 %, respectively.57
Ozbek et al. determined 10 elements in 35 vinegar samples, with LODs between 0.4 and
30 µg L-1, and accuracies in the 93 - 104 % range.58 Similar to behavior observed for other
complex samples, the MIP was stable and well tolerant to the direct introduction of diluted
vinegar samples. The commercial MIP OES system was also used to determine Cu, Fe, Mn
and Zn in animal feed and fertilizers.59 Samples were submitted to either microwave-
assisted digestion with HNO3 (animal feed), or HCl extraction on a hot plate (fertilizers).
A 40-fold dilution was required to minimize matrix effects and provide results comparable
with values obtained with ICP OES and FAAS. The LODs were comparable with ICP OES,
but significantly superior to FAAS: 2 - 4, 2 - 5, and 10 - 40 µg L-1 for MIP OES, ICP OES
and FAAS, respectively.
22
As discussed earlier, MIPs operate at relatively lower temperatures compared with
a conventional Ar ICP,3,20 and often require matrix-matching calibration to ensure accurate
results when analyzing some complex-matrix samples. A simple strategy to minimize
matrix effects in determinations using external standard calibration with aqueous solutions
was recently proposed by Lowery et al.60 In this work, emission signals from naturally-
occurring molecular species in the N2 MIP were used as internal standards to correct for
analytical signal variations due to the sample matrix. Emission signals from N2+ (0–0, B
2Σ+u → X 2Σ+g) and OH (0–0, A 2Σ+ → X 2πi), with emission band heads at 391.4 and 309
nm, were used as molecular internal standards to minimize matrix effects and determine
Ca, K, Mg and Na in biodiesel. Significant improvements in accuracy were observed by
using the Mg / OH and Na / OH signal ratios, and the Ca x N2+, and K x OH signal
relationships as the dependent variable (y-axis) in the calibration curve plot. Biodiesel
samples were simply diluted in 1-propanol, and recoveries for K and Mg employing
aqueous standards and external standard calibration went from 130 and 60 % without signal
correction to 82 and 92 %, respectively, using the molecular probe strategy. Solvent-
matching (i.e., standard reference solutions prepared in 1-propanol) was required for
determining Ca and Na, but recoveries were also improved from 74 and 84 % to 104 and
92 %, respectively, when molecular standardization was employed.
Another approach to minimizing matrix effects in MIP OES determinations is the
previously described SDA calibration method. Goncalves et al. determined 7 elements in
several complex-matrix beverage samples using a procedure based on simple sample
dilution in 1 % v v-1 HNO3 and determination by SDA and MIP OES.61 LODs for Al, Co,
Cr, Cu, Mn, Ni and Zn were in the 10 (Cu) - 500 (Zn) µg L-1 range, and average recoveries
23
for all analytes and samples evaluated were between 90 and 99 %. As a comparison,
average recoveries for the same samples and analytes using the standard additions method
provided values in the 101 - 122 % range.
CONCLUSIONS AND PERSPECTIVES
While MIP OES offers advantages of low cost and simple operation, at present it
cannot match the sensitivity or robustness of ICP OES, preventing MIP from expanding its
foothold in the field of atomic spectrometry. Further development of the technique is
required to improve accuracy and expand the types of samples that can be analyzed.
MIP OES has recently become a cost-effective option for trace multi-element
analysis. Its ability to run on air, combined with its relatively low cost of purchase and
operation show great potential for increasing access to atomic spectrometry to scientists
with limited budgets and limited infrastructure. Therefore, studies developing low-cost,
simple sample preparation strategies are critical to the development of MIP OES and
furthering its adoption. Two such studies are presented here, one focused on a simple,
dilute-and-shoot method for analysis of soft drink samples (Chapter II) and another focused
on the use of a rapid dry-ashing procedure to decompose samples prior to analysis by MIP
OES (Chapter III).
Studies associated with plasma diagnostics and analytical signal variation due to
sample matrix or type of solvent would significantly contribute to expanding the
capabilities of MIP OES. These studies would increase the understanding of the
characteristics of the MIP and would enhance the ability to use these plasma properties for
24
signal correction and for method development purposes, thereby increasing accuracy. A
study in plasma robustness based on the ratio of molecular lines N2+/OH is presented in
Chapter IV as a novel diagnostic tool developed specifically for MIP OES, as opposed to
being transferred from ICP. It contributes to increasing the understanding of the plasma
properties, as well as a method for optimization of operating conditions, and a preliminary
study of the use of molecular species for signal correction.
The development of novel calibration techniques is another strategy to improve the
ability of MIP OES to accurately analyze a broader variety of sample types. Calibration
methods tailored to the specific properties of MIP OES can address its shortcomings better
than more general techniques transferred from other instruments and techniques. In
Chapter V, multi-flow calibration (MFC) is presented as a novel strategy in MIP OES
which builds on the study of robustness presented in Chapter IV. By utilizing multiple
nebulization gas flow rates to perform analysis, the sample is exposed to a variety of
conditions, obviating the need to select a single set of conditions and simplifying method
development.
These studies together present a general advancement of microwave induced
plasma optical emission spectrometry as a technique. It can be coupled with simple, low-
cost sample preparation strategies to expand its analytical utility. In addition, molecular
species can be used to study and diagnose the plasma as well as for signal correction, to
improve accuracy. Finally, new calibration approaches such as multi-flow calibration can
be used to compensate for issues with robustness by exposing samples to a variety of
conditions, improving the quality of the analysis. Thus, the combination of sample
preparation, instrument optimization and calibration methods contributes to a more
25
efficient application of MIP OES and may bring it closer to the performance of traditional
methods such as ICP OES.
26
REFERENCES
1 J. D. Ingle and S. R. Crouch, Spectrochemical analysis, Prentice Hall, Englewood
Cliffs, N.J, 1988.
2 K. J. Jankowski and E. Reszke, Microwave induced plasma analytical
spectrometry, Royal Society of Chemistry, Cambridge, 2010.
3 A. Montaser and D. W. Golightly, Eds., Inductively coupled plasmas in analytical
atomic spectrometry, VCH Publishers, New York, N.Y, 2nd ed., 1992.
4 X. Hou and B. T. Jones, in Encyclopedia of Analytical Chemistry, ed. R. A. Meyers,
John Wiley & Sons Ltd, Chichester, 2000, p. 19.
5 J. W. Olesik and E. J. Williamsen, Appl. Spectrosc., 1989, 43, 1223–1232.
6 J. L. Todoli, L. Gras, V. Hernandis and J. Mora, J. Anal. At. Spectrom., 2002, 17,
142–169.
7 Z. Zhang and K. Wagatsuma, Spectrochim. Acta Part B At. Spectrosc., 2002, 57,
1247–1257.
8 K. Jankowski and E. Reszke, J. Anal. At. Spectrom., 2013, 28, 1196–1212.
9 J. D. Cobine and D. A. Wilbur, J. Appl. Phys., 1951, 22, 835–841.
10 A. J. McCormack, S. C. Tong and W. D. Cooke, Anal. Chem., 1965, 37, 1470–
1476.
11 C. Beenakker, Spectrochim. Acta Part B-At. Spectrosc., 1976, 31, 483–486.
27
12 E. Bulska, P. Tschöpel, J. A. C. Broekaert and G. Tölg, Anal. Chim. Acta, 1993,
271, 171–181.
13 B. D. Quimby and J. J. Sullivan, Anal. Chem., 1990, 62, 1027–1034.
14 R. D. Deutsch and G. M. Hieftje, Appl. Spectrosc., 1985, 39, 19–24.
15 R. D. Deutsch, J. P. Keilsohn and G. M. Hieftje, Appl. Spectrosc., 1985, 39, 531–
534.
16 M. Selby and G. M. Hieftje, Spectrochim. Acta Part B At. Spectrosc., 1987, 42,
285–298.
17 M. Selby, R. Rezaaiyaan and G. M. Hieftje, Appl. Spectrosc., 1987, 41, 749–761.
18 Y. Okamoto, Anal. Sci., 1991, 7, 283–288.
19 Z. Zhang and K. Wagatsuma, J. Anal. At. Spectrom., 2002, 17, 699–703.
20 D. A. Goncalves, T. McSweeney and G. L. Donati, J Anal Spectrom, 2016, 31,
1097–1104.
21 S. Zaidi, M. Denning and R. Urdahl, in 52nd Aerospace Sciences Meeting,
American Institute of Aeronautics and Astronautics, 2014.
22 M. R. Hammer, Spectrochim. Acta Part B At. Spectrosc., 2008, 63, 456–464.
23 A. Kadenkin and J. A. C. Broekaert, J. Anal. At. Spectrom., 2011, 26, 1481–1487.
24 H. Matusiewicz and M. Ślachciński, Spectrosc. Lett., 2011, 44, 128–137.
25 P. Sahay, S. T. Scherrer and C. Wang, Rev. Sci. Instrum., 2012, 83, 095109.
28
26 Y. Arai, S. Sato and K. Wagatsuma, Microchem. J., 2014, 116, 135–141.
27 Y. Arai, S. Sato and K. Wagatsuma, ISIJ Int., 2013, 53, 1993–1999.
28 K. Satoh and K. Wagatsuma, Spectrochim. Acta Part B At. Spectrosc., 2015, 108,
21–27.
29 A. J. Schwartz, Y. Cheung, J. Jevtic, V. Pikelja, A. Menon, S. J. Ray and G. M.
Hieftje, J Anal Spectrom, 2016, 31, 440–449.
30 H. Matusiewicz and M. Ślachciński, Microchem. J., 2010, 95, 213–221.
31 H. Matusiewicz and M. Ślachciński, Spectrosc. Lett., 2010, 43, 172–182.
32 C. K. Tanabe, H. Hopfer, G. Gilleland, A. Liba, S. E. Ebeler and J. Nelson, J. Anal.
At. Spectrom., 2016, 31, 1223–1227.
33 H. Matusiewicz and M. Ślachciński, Spectrosc. Lett., 2013, 46, 315–326.
34 H. Matusiewicz and M. Ślachciński, Spectrosc. Lett., 2014, 47, 415–426.
35 A. F. Lima, M. C. Da Costa, D. C. Ferreira, E. M. Richter and R. A. A. Munoz,
Microchem. J., 2015, 118, 40–44.
36 H. Matusiewicz and M. Ślachciński, Microchem. J., 2017, 131, 70–78.
37 H. Matusiewicz and M. Ślachciński, Microchem. J., 2017, 130, 345–352.
38 P. Pohl and P. Jamroz, J. Anal. At. Spectrom., 2011, 26, 1317.
39 J. Dedina and D. L. Tsalev, Hydride generation atomic absorption spectrometry,
Wiley, Chichester [England], 1995.
29
40 K. Jankowski, A. Jackowska and A. Tyburska, Spectrosc. Lett., 2010, 43, 91–100.
41 A. Tyburska and K. Jankowski, Anal. Methods, 2011, 3, 659–663.
42 M. Ślachciński, Talanta, 2016, 161, 812–818.
43 H. Matusiewicz and M. Ślachciński, J. Braz. Chem. Soc., 2016, 27, 584–590.
44 H. Matusiewicz and M. Ślachciński, Microchem. J., 2015, 119, 133–139.
45 H. Matusiewicz and M. Ślachciński, Microchem. J., 2012, 102, 61–67.
46 H. Matusiewicz and M. Ślachciński, Cent. Eur. J. Chem., 2011, 9, 896–903.
47 E. Y. Barrientos, K. Wrobel, J. C. T. Guzman, A. R. C. Escobosa and K. Wrobel,
J. Anal. At. Spectrom., 2015, 31, 203–211.
48 C. B. Williams, T. G. Wittmann, T. McSweeney, P. Elliott, B. T. Jones and G. L.
Donati, Microchem. J., 2017, 132, 15–19.
49 A. F. Lima, F. F. Lima, E. M. Richter and R. A. A. Munoz, Appl. Acoust., 2016,
103, 124–128.
50 H. Matusiewicz and M. Ślachciński, Microchem. J., 2014, 115, 6–10.
51 H. Matusiewicz and M. Ślachciński, J. Anal. At. Spectrom., 2010, 25, 1324.
52 R. S. Amais, G. L. Donati, D. Schiavo and J. A. Nóbrega, Microchem. J., 2013,
106, 318–322.
53 G. L. Donati, R. S. Amais, D. Schiavo and J. A. Nóbrega, J. Anal. At. Spectrom.,
2013, 28, 755–759.
30
54 J. Nelson, G. Gilleland, L. Poirier, D. Leong, P. Hajdu and F. Lopez-Linares,
Energy Fuels, 2015, 29, 5587–5594.
55 L. Poirier, J. Nelson, G. Gilleland, S. Wall, L. Berhane and F. Lopez-Linares,
Energy Fuels, 2017, 31, 7809–7815.
56 P. Niedzielski, L. Kozak, K. Jakubowski, W. Wachowiak and J. Wybieralska, Plant
Soil Environ., 2016, 62 (2016), 215–221.
57 P. Niedzielski, L. Kozak, M. Wachelka, K. Jakubowski and J. Wybieralska,
Talanta, 2015, 132, 591–599.
58 N. Ozbek, M. Koca and S. Akman, Food Anal. Methods, 2016, 9, 2246–2250.
59 W. Li, P. Simmons, D. Shrader, T. J. Herrman and S. Y. Dai, Talanta, 2013, 112,
43–48.
60 K. L. Lowery, T. McSweeney, S. P. Adhikari, A. Lachgar and G. L. Donati,
Microchem. J., 2016, 129, 58–62.
61 D. A. Goncalves, T. McSweeney, M. C. Santos, B. T. Jones and G. L. Donati, Anal.
Chim. Acta, 2016, 909, 24–29.
31
CHAPTER II
DETERMINATION OF CALCIUM, POTASSIUM AND SODIUM IN SOFT
DRINKS USING THE 4200 MP-AES
Charles B. Williams, Tina McSweeney, Bradley T. Jones, and George L. Donati
This work was published as an Application Note by Agilent Technologies on
www.agilent.com on September 23, 2018 and is reprinted with permission (Appendix A).
Stylistic variations are due to the requirements of the publisher. In particular, the term
MP-AES is used instead of MIP OES to be consistent with the trade name of the
instrument. All of the work was performed by Charles B. Williams. The manuscript was
prepared by Charles B. Williams and edited by George L. Donati, Tina McSweeney, and
Neli Drvodelic.
32
INTRODUCTION
Soft drinks are among the most common beverages consumed in the United States.
Fifty percent of young adults report drinking one or more soft drinks per day [1].
Regulations by the United States Food and Drug Administration (FDA) require reporting
of Ca and Na content of packaged foods and beverages [2], while K content reporting is
optional. FDA regulations for package labeling dictate that K and Na concentrations in the
5-140 mg/serving range should be rounded to the nearest 5 mg. For values over 140
mg/serving, the label can be rounded to the nearest 10 mg. Calcium is reported to the
nearest 10 % of the “recommended daily value” of 1000 mg [2]. These labelling
requirements, as well as the typical quality control analyses carried out by manufacturers,
demonstrate the need for simple, cost-effective, sensitive, accurate, precise and high
sample throughput methods for soft drink analyses.
Current FDA guidelines suggest using microwave-assisted digestion (MAD) and
inductively coupled plasma optical emission spectrometry (ICP OES) for soft drink
elemental analysis [3]. MAD is recommended due to the complexity of this type of sample.
The combination of carbonation, dye additives, artificial flavoring, and high sugar content
contribute to significant matrix effects, which are difficult to overcome for most analytical
techniques [4]. Although effective, MAD is a labor-intensive, expensive, and time-
consuming process, especially due to sample handling and the time required for sample
cool down after digestion.
This application note describes a “dilute and shoot” method for soft drink elemental
analysis based on microwave-induced plasma optical emission spectrometry using
Agilent’s 4200 MP-AES. Sample preparation is not required, eliminating the need for an
33
expensive digestion apparatus as well as some time-consuming procedures associated with
them. The 4200 MP-AES is less expensive to acquire and operate than an ICP OES since
it runs on N2 rather than Ar. It can even run on air if compressed air is supplied to the
optional nitrogen generator. Because the N2 microwave-induced plasma is cooler than an
Ar ICP [5,6], background signals in the visible region of the spectrum are lower, which
allows for limits of detection (LODs) that are comparable to the ones obtained with ICP
OES. The 4200 MP-AES records emission signals sequentially. Therefore analytical
conditions can be optimized for each specific element within the same experiment, which
maximizes efficiency and may minimize potential interferences.
34
EXPERIMENTAL
Instrumentation
All determinations were carried out using the Agilent 4200 MP-AES. A liquid N2
Dewar was used to provide N2 gas to run the microwave-induced plasma. No sample
preparation other than simple dilution with 1 % v/v HNO3 was required. The sample
introduction system is composed of an SPS 4 automatic sampler, solvent-resistant tubing,
a double-pass cyclonic spray chamber, and an inert Flow Blurring nebulizer (OneNeb).
Nebulizer flow rate and plasma viewing position were adjusted for each individual analyte
to optimize recovery. Operating conditions were optimized by selecting one of the spiked
samples, then running the “optimize nebulizer flow” followed by “optimize viewing
position” features in MP Expert. One of the samples was chosen for optimizing the
instrumental operating conditions in order to match matrix conditions and improve
accuracy. The viewing position used in this case was 0 for all elements. The nebulizer flow
rates were 0.90, 1.00 and 1.00 L/min for Ca , K and Na, respectively. Table III and Table
IV list the instrument operating parameters.
Table III. Element-Specific Operating Parameters
Element Wavelength (nm) Nebulizer Flow Rate (L/min)
Ca 393.366 0.9
K 766.491 1
Na 588.995 1
35
Table IV. Instrument Parameters
Number of Replicates 3
Peristaltic Pump Speed 15 rpm
Uptake Time 45 s
Rinse Time 30 s
Stabilization Time 15 s
Background Correction None
No. of Pixels 3
Read Time (all elements) 3 s
Spray Chamber Double pass glass cyclonic
Nebulizer OneNeb Flow-Blurring
Sample Pump tubing Orange/Green
Waste Pump Tubing Blue/Blue
Autosampler SPS-4
Samples and sample preparation
All solutions were prepared using distilled-deionized water (18 M Ω cm, Milli-Q®,
Millipore, Bedford, MA, USA) and trace metal grade nitric acid (Fisher, Pittsburgh, PA,
USA). Single-element stock solutions containing 1000 mg/L of Ca, K or Na (SPEX
CertPrep, Metuchen, NJ, USA) were used to prepare standard reference solutions for
calibration and to carry out spike experiments.
36
Seven different soft drink samples were analyzed. Various popular beverages were
chosen to get a sample of different flavors and types (e.g. diet vs. regular, cola, orange,
ginger ale, etc.). Approximately 2.0-mL aliquots of each of these samples were weighed in
15-mL graduated polypropylene centrifuge tubes using an analytical balance (Mettler AE
100, Hightstown, NJ, USA). For K determination in Diet Dr. Pepper, and Na determination
in Diet Dr. Pepper, Mountain Dew and Schweppes, the same procedure was employed
using 0.2 mL sample aliquots. Sample mass was adopted rather than volume to minimize
any potential bias introduced by residual gas bubbles in the soft drink or inaccuracies due
to viscosity caused by high sugar content. The samples were then diluted to 10.0 mL with
1 % v/v HNO3. Five standard reference solutions (0.2-10 mg/L) and a blank, all prepared
in 1 % v/v HNO3, were used to build the calibration curves. The traditional external
standard calibration method was used in all determinations in order to simplify analysis.
Spike experiments were carried out to evaluate the procedure’s accuracy. Samples
used in this study were prepared in the same way as the unspiked samples. Adequate
volumes of stock solution were added to the samples such that the final concentrations were
1.00 mg/L for Ca, and 2.00 mg/L for K and Na. An intermediate stock solution of 10 mg/L
Ca and 20 mg/L K and Na was prepared from the same standards used to create the
calibration curve. 1 mL was added to the weighed 2.0 mL soft drink aliquots, which was
then completed to 10.0 mL using 1% HNO3.
37
RESULTS AND DISCUSSION
Limits of detection and accuracy
The limits of detection for determinations using Agilent’s 4200 MP-AES were
calculated according to IUPAC’s recommendations as 3 times the standard deviation of the
blank signal (SB) divided by the calibration curve slope (m): LOD = 3SB / m. Twelve
consecutive blank solution (1 % v/v HNO3) measurements were used to calculate SB for
each instrumental condition. The detection limits for Ca, K and Na were 30, 3 and 20 µg/L,
respectively.
The procedure’s accuracy was evaluated by spike experiments. Recoveries were
calculated by comparing expected (concentrations added) and measured (spiked -
unspiked) values. The results for each analyte and sample are presented in Table V.
Recoveries were within 91-110 % using the optimized conditions. The traditional external
standard calibration was used in all cases, which is much simpler and less labor-intensive
than standard additions or internal standardization. The spike recoveries were excellent
with simple external calibration, which are good indicators of the method accuracy. The
method also takes advantage of the MP’s tolerance to high carbon-content matrices, as the
regular (non-diet) drinks included a substantial concentration of sugar. The external gas
control module (EGCM) air injection was not used in this study, but significant carbon
buildup was not found to be a problem on the torch.
38
Table V. Soft drink analysis with the 4200 MP-AES. Values are the recoveries (%) for
spike experiments with 1.00 mg/L Ca, and 2.00 mg/L K and Na.
Sample Ca K Na
Cheerwine 95 91 105
Diet Dr. Pepper 100 106 102
Fanta 110 107 106
Mountain Dew 99 96 100
Pepsi 96 95 100
Schweppes 91 104 103
Sprite 91 106 108
Soft drink concentrations
The same procedure used in the addition and recovery experiment was used to
determine the concentrations of Ca, K and Na in the original (non-spiked) samples. The
results are shown in Table VI. Calcium concentrations were more varied across the
samples. Schweppes ginger ale presented a significantly higher Ca value, which may be
related to the high concentrations of Ca found in the Ginger root (Zingiber officinale) [6].
Sodium and potassium also had wider variations, which may be explained by some
products using a Na or K salt as a preservative, as disclosed on the individual labels.
39
Table VI. Ca, K and Na concentrations in soft drinks determined by the 4200 MP-AES.
Results are the mean ± 1 standard deviation (mg/L, n = 3).
Sample Ca K Na
Cheerwine 6.09 ± 0.10 6.07 ± 0.04 18.15 ± 0.14
Diet Dr. Pepper 3.06 ± 0.02 31.27 ± 0.41 58.58 ± 0.31
Fanta 6.17 ± 0.02 54.63 ± 0.46 3.28 ± 0.18
Mountain Dew 6.28 ± 0.06 21.20 ± 0.23 84.72 ± 0.52
Pepsi 3.30 ± 0.04 14.69 ± 0.20 20.08 ± 0.11
Schweppes
Ginger Ale
23.49 ± 0.07 3.16 ± 0.02 62.47 ± 0.58
Sprite 7.22 ± 0.12 1.03 ± 0.01 79.73 ± 0.51
CONCLUSIONS
The procedure described in this application note is an interesting alternative to the
MAD and ICP OES method recommended by the FDA. It is a fast, cost-effective and
efficient strategy that can be applied by manufacturing laboratories in routine quality
control analyses and determinations associated with package-labeling regulations It also
demonstrates the ability of MP AES to handle complex sample matrices without time-
consuming, expensive sample preparation steps.
40
ACKNOWLEDGMENTS
The authors would like to thank Agilent Technologies and the Department of
Chemistry at Wake Forest University for their support to this work.
REFERENCES
[1] E. Han and L. M. Powell, Consumption patterns of sugar-sweetened beverages in the
United States, J. Acad. Nutr. Diet 113(1) (2013) 43-53.
[2] United States Food and Drug Administration, Code of Federal Regulations: Title 21,
Chapter I, Subchapter B, Part 101: Food Labeling; Vol. Title 21: Food and Drugs § 101.9:
Nutrition Labeling of Food
[3] W. R. Mindak and S. P. Dolan, Inductively Coupled Plasma-Atomic Emission
Spectrometric Determination of Elements in Food Using Microwave Assisted Digestion,
United States Food and Drug Administration Elemental Analysis Manual for Food and
Related Products, 2010.
[4] R. E. S. Froes, W. B. Neto, R. L. P. Naveira, N. C. Silva, C. C. Nascentes and J. B. B.
Silva, Exploratory analysis and inductively coupled plasma optical emission spectrometry
(ICP OES) applied in the determination of metals in soft drinks, Microchem. J. 92(1)
(2009) 68-72.
[5] A. Montaser and D. W. Golightly (Eds.), Inductively Coupled Plasmas in Analytical
Atomic Spectrometry, 2nd ed., Wiley-VCH, New York, 1992.
[6] S. Adel P. R. and J. Prakash, Chemical composition and antioxidant properties of ginger
root (Zingiber officinale), J. Med. Plants Res. 4(24) (2010) 2674-2675
41
CHAPTER III
DRY ASHING AND MICROWAVE-INDUCED PLASMA OPTICAL EMISSION
SPECTROMETRY AS A FAST AND COST-EFFECTIVE STRATEGY FOR
TRACE ELEMENT ANALYSIS
Charles B. Williams , Thomas G. Wittmann, Tina McSweeney, Paul Elliott,
Bradley T. Jones and George L. Donati
The following manuscript was published in Microchemical Journal, 2017, 132, 15-
19, and is reprinted with permission. Stylistic variations are due to the requirements of the
journal. The presented research was conducted by Charles B Williams with assistance from
Thomas G. Wittmann. The manuscript was prepared by Charles B. Williams and edited by
George L. Donati.
42
ABSTRACT
Microwave-induced plasma optical emission spectrometry (MIP OES) is combined with a
simple dry ashing apparatus as a cost-effective alternative for trace element analysis. The
Helios Rapid Ashing (HRA) prototype can reach up to 750 °C using a ceramic radiative
heating element. It allows for sample decomposition in less than 5 min, and requires
inexpensive sample holder materials such as aluminum foil and quartz fiber pads. Samples
were decomposed at 500 oC and the analytes were extracted into a 10 % v/v HCl solution
before analysis by MIP OES. Limits of detection for Ca, Fe, K, Mg, Na and Zn were
calculated as 2, 20, 30, 0.6, 2 and 5 µg/L, respectively. These analytes were determined in
a certified reference material of Tomato Leaves (NIST SRM 1573a) and in challenging
samples such as cheese, butter, peanut butter, infant formula and biodiesel. No statistically
significant differences were observed between certified values and concentrations
determined by the HRA-MIP OES procedure (t-test at a 95 % confidence level). General
agreement was also found between MIP OES results from samples decomposed either by
the HRA or by a conventional microwave-assisted digestion procedure. MIP OES is an
efficient alternative to FAAS, with comparable linear dynamic ranges and significantly
improved sensitivity. It has short start/warmup times (ca. 20 min), runs on inexpensive N2,
and may be a perfect match to the HRA system. The HRA-MIP OES procedure can be a
simple, fast and accurate strategy for inexpensive and effective sample decomposition and
trace element analysis.
43
1. INTRODUCTION
Current United States Food and Drug Administration (US FDA) guidelines require
reporting of Ca and Na content of packaged foods [1]. Reporting of other elements such as
Fe, K, Mg and Zn is optional, but these are important nutrients in human diet, and such
information has become increasingly relevant for health-conscious consumers, particularly
parents selecting food products for their infants and children. Therefore, factors such as
marketing, consumer awareness, and commercial competitiveness has led food
manufacturers to seek reliable and cost-effective methods of determining trace elemental
nutrients in their products.
Most modern atomic spectrometry methods such as flame atomic absorption or
emission spectrometry (FAAS and FAES), inductively coupled plasma optical emission
spectrometry (ICP OES), microwave-induced plasma optical emission spectroscopy (MIP
OES), and inductively coupled plasma mass spectrometry (ICP-MS) require samples to be
introduced as liquid solutions, primarily aqueous solutions. For samples which already are
aqueous, this is trivial and usually involves simple dilution before analysis. On the other
hand, for the majority of samples, an additional time-consuming digestion step is
necessary. One of the most effective approaches to solubilizing complex samples is based
on microwave-assisted digestion (MAD) using acids [2]. Among the many advantages of
MAD are the typically low blanks, efficient digestion of a variety of sample matrices, and
relatively low consumption of reagents [3]. It involves using microwave radiation to
accelerate the decomposition of sample matrices by a strong acid at high pressures and
temperatures. In most MAD procedures, a mineral acid (usually HNO3) is added to a small
mass of sample (i.e. 0.1 - 0.5 g), and the mixture is heated up in a closed Teflon® vessel.
44
Diluted acid solutions may replace the concentrated reagent, and H2O2 may be added to the
digestion mixture to enhance the oxidative power of the acid [4, 5]. For trace element
analysis, MAD is used in association with ICP OES or ICP-MS, which results in highly
accurate and precise determinations.
Despite their many advantages, MAD and ICP-based systems are relatively
expensive to acquire and run. In addition, for a few high-temperature digestions, sample
throughput becomes an issue. Sample cool down after digestion and vessel
decontamination may represent one of the main time-limiting steps in a typical trace
element procedure. A less expensive alternative to MAD is based on dry and/or wet ashing
of samples [6]. With these methods, samples are heated in a crucible in a muffle furnace or
other conventional heating apparatus and reduced to a fine powder containing the residual
minerals from the original matrix. The main limitation of this approach is related to poor
sample throughput. Ashing may take up to several hours, with additional time for sample
cooling, and final solubilization of the residual powder in an aqueous medium before
analysis [7]. For certain applications, additional sample pretreatment is required, which can
lengthen an already long analysis by 12 hours or more [8]. Complete matrix decomposition
is easily achieved with sample ashing, however potential analyte contamination and low
sample throughput have prevented its use in routine trace element analysis.
Recently, a simple and efficient MIP OES system has become commercially
available [9]. This N2-plasma instrument is less expensive to acquire and maintain than
ICP systems. If a nitrogen generator and a conventional air compressor are used, it can
even run on air [10], which significantly reduces running costs and allows operation in
remote locations. For most elements, MIP OES has sensitivities which range between those
45
of FAAS and ICP OES, and it has been successfully used in many applications [10-15].
MIP OES is also advantageous when compared with low-cost flame-based methods (i.e.
FAAS and FAES) because of its improved safety (no flammable gases are required as with
FAAS) and multi-element capabilities.
In the present work, we describe the combination of MIP OES with a simple dry
ashing apparatus (Helios Rapid Ashing prototype, CEM Corporation) as a cost-effective
alternative for trace element analysis. The ashing prototype unit can heat samples up to 750
°C using a ceramic radiative heating element. It allows for sample decomposition in less
than 5 min, and utilizes very inexpensive materials such as aluminum foil and quartz fiber
pads. The efficiency of the dry ashing/MIP OES procedure is evaluated by analyzing
complex sample matrices of cheese, butter, peanut butter, infant formula and biodiesel.
These were chosen as test samples due to their high fat and high sodium contents, which
makes matrix decomposition challenging, and often results in severe matrix effects that can
compromise precision and accuracy [16, 17]. The procedure’s accuracy was evaluated by
determining Ca, Fe, K, Mg, Na and Zn in a certified reference material of tomato leaves
(NIST SRM 1573a). In addition, MIP OES results from samples decomposed either by the
ashing procedure or by a conventional microwave-assisted acid digestion were also
compared.
46
2. EXPERIMENTAL
2.1. Instrumentation
A prototype dry ashing apparatus (Helios Rapid Ashing, CEM Corporation,
Mathews, NC, USA) was used for sample preparation. A picture of the instrument showing
its touchscreen interface and user-programmable heating cycle is depicted in Figure 5.
This unit consists of a tubular furnace section oriented vertically into which a metallic
basket is introduced for sample ashing. The prototype is 25.4 cm high and 21.6 cm in
diameter, and its furnace is capable of sustaining temperatures up to 750 °C. The sample
basket consists of a steel grid at the bottom, onto which the sample holder (e.g. aluminum
foil) is placed, and an aperture-adjustable lid at the top, which controls air flow into the
furnace. A rubber-grip handle completes the sample basket setup (Figure 6). The Helios
Rapid Ashing (HRA) prototype is designed for ease of use as a manual operation. An LED
touchscreen panel on the unit has a timer and allows for temperature setting and control.
The interface relies on the PID control algorithm to quickly reach and hold the set
temperature. The timer is independent and is not synchronized nor does it control the
furnace temperature.
47
Figure 5. Helios Rapid Ashing unit, courtesy of CEM Corporation.
Figure 6. Sample basket used with the HRA prototype.
48
Approximately 7 cm2 sections of either aluminum foil (Reynolds Wrap Heavy
Duty, Reynolds Consumer Products, Lake Forest, IL, USA), or quartz fiber ashing pads
(CEM) were used as sample holders.
For comparison and accuracy evaluation, all samples were also submitted to
microwave-assisted acid digestion using an instrument with quartz digestion vessels
(Discover, CEM).
A MIP OES system (4200 MP-AES, Agilent Technologies, Santa Clara, CA, USA)
equipped with an SPS-4 automatic sampler, an inert Flow Blurring nebulizer (OneNeb),
and a double-pass cyclonic spray chamber was used in all determinations. Nitrogen gas
used by the instrument was provided by a liquid N2 Dewar. Default operating conditions
(Table VII) were used to analyze samples submitted to both dry ashing and MAD.
49
Table VII: Operating conditions used in MIP OES determinations.
Instrumental parameter Operating condition
Microwave frequency
(MHz)
2450
Applied power (kW) 1.0
Peristaltic pump speed (rpm) 15
Integration time (s) 3
Number of replicates 3
Plasma viewing positiona 0
Analyte (wavelength, nm) Nebulization gas flow rate
(L/min)
Ca (393.366) 0.60
Fe (371.993) 0.65
K (766.491) 0.75
Mg (285.213) 0.90
Na (588.995) 0.95
Zn (213.857) 0.45
a The plasma viewing position has no specific unit. It is based on stepper motor
positioning of a mirror. Position 0 approximately corresponds to the center of the plasma
[9].
50
2.2. Reagents, standard reference solutions and samples
Distilled-deionized water (18 MΩ cm, Milli-Q ®, Millipore, Bedford, MA, USA)
was used to prepare all solutions. Trace metal grade nitric acid (Fischer, Pittsburgh, PA,
USA) was used to prepare calibration standards, and to digest samples by MAD. ACS+
grade hydrochloric acid (Fischer) was used to extract the analytes from ashing residues
following sample decomposition by the HRA prototype. The standard reference solutions
used for calibration were prepared in either 1% v/v HNO3 (commercial samples) or 10 %
v/v HCl (certified reference material) by adequate dilution of single-element stock
solutions of Ca, Fe, K, Mg, Na and Zn (1000 mg/L, SPEX CertPrep, Metuchen, NJ, USA).
Various commercial samples were used to evaluate the efficiency of the dry
ashing/MIP OES procedure: cheese (Great Value low moisture part-skim mozzarella, Wal-
Mart Corp., Little Rock, AR, USA), butter (Land O Lakes, St. Paul, MN, USA), peanut
butter (Jif reduced fat creamy, J. M. Smucker, Orrville, OH, USA), infant formula (Parent’s
Choice, gentle, Perrigo Nutritionals, Charlottesville, VA, USA), and biodiesel (prepared
in-lab using soybean oil, methanol, and KOH) [18]. A Standard Reference Material from
the National Institutes of Standards and Technology (Tomato Leaves, NIST SRM 1573a)
was used to evaluate the procedure’s accuracy.
2.3. Sample preparation
Microwave-assisted digestion. Sample aliquots of 0.2 - 0.3 g were accurately weighed
using an analytical balance (Mettler, Toledo, OH, USA), and digested in quartz vessels
with 5 mL of HNO3 50% v/v. The MAD heating cycle was composed of a 4-min
51
temperature ramp to 200 °C, and a 3-min hold at 200 °C using 300 W of power. Samples
were allowed to cool, then quantitatively transferred into 50 mL polypropylene graduated
centrifuge tubes. The digested mixture was then diluted to 25 mL with distilled-deionized
water. Analytical blanks were prepared using the same procedure, and all samples were
digested in triplicate.
Dry ashing and analyte extraction
The HRA prototype was pre-heated to 500 °C to improve repeatability. Sample aliquots of
0.2 - 0.3 g were accurately weighed directly on square sections of either aluminum foil or
quartz fiber pads (sample holders) using an analytical balance (Mettler). Another ca. 7 cm2
section of aluminum foil or quartz fiber pad was placed on top of the sample, and the entire
“sandwich” set was transferred to the steel grid in the sample basket (Figure 7). With the
adjustable lid kept closed at all times, the basket was then lowered into the furnace, and the
timer was set for either 2 or 3 minutes depending on the sample. After the ashing was
complete, a set of forceps was used to remove the sample holder and place it into a 250 mL
snap-fit polypropylene flask (Corning, Corning, NY, USA). The analytes were extracted
using 25 or 30 mL aliquots of a 10 % v/v HCl solution. For quartz fiber pad sample holders,
the HCl solution was simply poured on top of the pad and the extraction vial was shaken
for a few seconds. Aluminum foil sample holders were held above the extraction vial with
forceps and rinsed with the HCl solution using a polypropylene syringe (Becton-Dickinson
& Co, Fraklin Lakes, NJ, USA). After the vial was shaken, the extraction mixture (i.e.
ashing residue, quartz fiber pads and HCl solution) was allowed to sit for approximately 1
hour. Finally, the mixture was gravity-filtered using a small disposable funnel and a 0.45-
µm-pore filter paper (CEM, Matthews, NC, USA). Samples were initially analyzed
52
undiluted. For elements with concentrations higher than 5 mg/L, a 10-fold dilution with
distilled-deionized water was carried out before MIP OES determination. Analytical blanks
were prepared using the same procedure and all samples were processed in triplicate. Table
VIII presents the specific ashing parameters which provided the best recoveries for each
sample (optimization not shown).
Figure 7. Sample “sandwich” setup using an aluminum foil at the bottom and a quartz fiber pad on top loaded in the sample basket.
53
Table VIII. Ashing parameters providing the best recoveries for determinations using the
HRA prototype and MIP OES.
Sample Ashing time (min) Sample holder Extracting
solution
volume (mL)a
Cheese 2 Two quartz fiber pads 30.0
Biodiesel 3 Two sections of aluminum foil 25.0
Butter 3 Two sections of aluminum foil 25.0
Peanut butter 3 Aluminum foil at bottom,
quartz fiber pad on top
25.0
Infant formula 3 Aluminum foil at bottom,
quartz fiber pad on top
25.0
Tomato leaves
(NIST SRM 1573a)
3 Aluminum foil at bottom,
quartz fiber pad on top
25.0
a Extracting solution: HCl 10 % v/v.
54
3. Results and discussion
3.1. Limits of detection and accuracies
Limits of detection (LOD) were calculated according to IUPAC’s recommendation
as three times the standard deviation of the blank solution (SB, n = 15) divided by the
calibration curve slope (m), i.e., LOD = 3SB / m. Similarly, the limits of quantification
(LOQ) were calculated as LOQ = 10SB / m. Considering the HRA extracting agent, the
blank used in these calculations was a 10 % v/v HCl solution. Using the operating
conditions listed in Table VII, LODs for Ca, Fe, K, Mg, Na and Zn were calculated as 2,
20, 30, 0.6, 2 and 5 µg/L, respectively. These concentrations, as well as the LOQs for the
same elements (i.e. 6, 60, 100, 2, 6 and 20 µg/L, respectively), are well below the reporting
ranges required by the US FDA [1]. In addition to the adequate sensitivities of MIP OES,
relatively low LODs are achievable due to the minimal sample dilution associated with
HRA. These values are comparable to a traditional MAD-ICP OES procedure for these
types of sample matrices [19], and significantly superior to FAAS [20].
The accuracy of the HRA-MIP OES procedure was checked by analyzing a
certified reference material of Tomato Leaves (NIST SRM 1573a). Nitric acid and HCl
solutions at 1, 10 and 20 % v/v were evaluated as extracting agent. In this study, HCl
provided better recoveries than HNO3. For the majority of analytes, a 1 % v/v acid
concentration resulted in low recoveries, while no significant differences were observed
between samples extracted with 10 or 20 % v/v HCl (results not shown). To minimize
potential matrix effects, and considering the best recoveries for all analytes evaluated, a 10
% v/v HCl solution was chosen as extracting agent in all subsequent HRA extractions. All
determinations were carried out using the external standard calibration method and
55
standard solutions prepared in 10 % v/v HCl. The results are shown in Table IX. As it can
be seen, no statistically significant difference was found between the reference values and
concentrations determined by HRA-MIP OES (t-test at a 95 % confidence level). Although
efficient and accurate, the procedure’s precision is relatively low. High relative standard
deviations (RSDs) are common in ashing procedures due to more sample manipulation and
less control of the digestion conditions [2]. In the specific case of this study, RSDs were
higher for the reference material than the commercial samples (see Table IX and Table X).
This fact may be related to the different consistencies of each sample and their behavior
during digestion. It is possible that the dry and powdery reference material was more
susceptible to spilling and analyte loss during the HRA heating cycle, which resulted in
higher RSDs and relatively lower concentrations for most analytes.
56
Table IX. Evaluating the accuracy of the HRA-MIP OES procedure by analyzing a
standard reference material of Tomato Leaves (NIST SRM 1573a). Reported values are the
mean ± 1 standard deviation concentrations in the original solid sample (mg/g, n = 3).
Analyte Certified Found
Ca 50.5 44.6 ± 8.1
Fe 0.368 0.337 ± 0.143
K 27.0 26.6 ± 3.5
Mg 12.0 9.7 ± 1.9
Na 0.136 0.122 ± 0.020
3.2. Application to commercial samples and comparison with MAD
Commercial samples of cheese, butter, peanut butter, infant formula, and a lab-
made biodiesel sample were digested by HRA or MAD and the concentrations of Ca, K,
Mg and Na were determined by MIP OES. The results were used to further assess the
efficiency and accuracy of the HRA-MIP OES procedure using challenging sample
matrices. Analyte concentrations determined with both digestion procedures were also
compared to product label information. In this study, calibration curve solutions were
prepared in HNO3 1 % v/v to evaluate the performance of the HRA-MIP OES procedure
in non-matrix-matching conditions. As can be seen in Table X, there is a general agreement
57
between concentrations determined in samples processed by the ashing and acid extraction
(HRA) procedure, and by the traditional microwave-assisted acid digestion (MAD). The
results also broadly agree with label values. The few discrepancies may probably be
explained either by uncertainty due to US FDA rounding requirements [21], or because
label values are based on a different sample lot.
As discussed previously, precision is generally similar for HRA and MAD. Based
on the results presented in Table X, HRA can be considered as efficient as MAD for
applications involving the matrices and analytes evaluated in this work. Table XI shows
the relationship between HRA and MAD results. Values in this table were calculated as
(HRA result / MAD result) x 100. For the majority of samples and analytes, HRA values
were within 90 - 110 % of the MAD results, and all values were in the 84 - 122 % range.
Considering the importance of micronutrients in infant formula, Fe and Zn were
also determined in this sample. For HRA, 0.104 ± 0.009 and 0.063 ± 0.003 mg/g were
found for Fe and Zn, respectively. These results are comparable to values obtained from
samples submitted to MAD, i.e. 0.102 ± 0.002 and 0.067 ± 0.001 mg/g, and slightly higher
than concentrations reported on the product’s label (0.08 and 0.045 mg/g for Fe and Zn,
respectively).
58
Tab
le X
. Com
paris
on b
etw
een
ashi
ng an
d ac
id ex
tract
ion
(HR
A) w
ith th
e tra
ditio
nal m
icro
wav
e-as
sist
ed d
iges
tion
(MA
D).
The a
naly
te
conc
entra
tions
wer
e det
erm
ined
by
MIP
OES
usi
ng th
e ext
erna
l sta
ndar
d ca
libra
tion
met
hod.
Val
ues a
re th
e mea
n ±
1 st
anda
rd d
evia
tion
(mg/
g, n
= 3
).
Sam
ple
Sam
ple
prep
arat
ion
met
hoda
Ca
K
Mg
Na
Che
ese
HR
A
4.70
± 0
.32
0.49
± 0
.03
0.21
4 ±
0.01
0 6.
57 ±
0.1
6
M
AD
4.
84 ±
0.1
3 0.
58 ±
0.0
1 0.
225
± 0.
010
7.09
± 0
.26
La
bel
7 ±
3 0.
89 ±
0.4
5 N
Ab
7.10
± 0
.40
Butte
r H
RA
0.
29 ±
0.0
2 0.
21 ±
0.0
1 0.
019
± 0.
003
6.32
± 0
.09
M
AD
0.
27 ±
0.0
2 0.
21 ±
0.0
2 0.
021
± 0.
001
5.24
± 0
.61
La
bel
NA
b N
Ab
NA
b 6.
40 ±
0.7
0
Pean
ut B
utte
r H
RA
0.
35 ±
0.0
3 3.
47 ±
0.1
9 1.
749
± 0.
232
4.12
± 0
.26
M
AD
0.
36 ±
0.0
2 3.
38 ±
0.0
8 1.
818
± 0.
104
4.00
± 0
.12
La
bel
0.56
± 0
.28
NA
b 1.
60 ±
0.6
0 5.
60 ±
0.3
0
59
Infa
nt F
orm
ula
HR
A
3.81
± 0
.17
5.32
± 0
.17
0.69
1 ±
0.00
8 1.
92 ±
0.0
6
M
AD
3.
82 ±
0.0
6 5.
69 ±
0.0
6 0.
683
± 0.
017
2.02
± 0
.01
La
bel
3.5
4.9
0.36
1.
2
Biod
iese
l H
RA
<
LOD
<
LOD
<
LOD
71
.72
± 3.
46
M
AD
<
LOD
<
LOD
<
LOD
84
.17
± 1.
52
a Lab
el v
alue
s ar
e ba
sed
on th
e pr
oduc
t pac
kagi
ng in
form
atio
n an
d th
e FD
A p
erce
nt re
com
men
ded
daily
val
ues
[22]
. Sta
ndar
d
devi
atio
ns fo
r ‘La
bel’
are
base
d on
the
FDA
roun
ding
rule
s for
pac
kagi
ng, r
athe
r tha
n st
atis
tical
ana
lysi
s [21
].
b Not
ava
ilabl
e.
60
Table XI. HRA results presented as their percent portion of concentrations determined
using the MAD procedure.
Sample Ca K Mg Na
Cheese 97 84 91 93
Biodiesel < LOD < LOD < LOD 85
Butter 107 100 90 122
Peanut Butter 94 103 94 102
Infant Formula 97 93 100 95
61
4. Conclusions
The HRA system is a simple and cost-effective alternative to MAD. It requires no
high-pressure-withstanding flasks, nor any other relatively expensive apparatus and
consumables. It also is a green approach to sample preparation since it uses few reagents,
generates no acid vapors, and produces fewer residues. When compared to traditional
methods of dry- and wet-ashing, the HRA procedure is faster, greener and more efficient.
Ten samples can be digested in less than two hours with no cooling step required. Dry-
ashing in a muffle furnace, for example, requires heating and subsequent cooling overnight
[8].
Similar to other open-system ashing procedures, HRA is more prone to
contamination and analyte loss than MAD. Although no severe contamination was
observed for the analytes evaluated in this study, memory effects may be particularly
critical, as the sample holder grid is not decontaminated between samples. In its current
state, the HRA prototype can only accommodate one sample at a time, and has no
automation capability, which may be expected of a prototype.
MIP OES is an efficient alternative to FAAS. It has short start/warmup times (ca.
20 min), and runs on inexpensive N2. MIP OES has also adequate sensitivities for
applications associated with US FDA requirements for food packaging, and may be a
perfect match to the HRA system. The HRA-MIP OES procedure can be a simple, fast and
accurate strategy for inexpensive and effective sample decomposition and trace element
analysis.
62
Acknowledgements
The authors would like to thank the Department of Chemistry at Wake Forest
University, as well as CEM Corporation and Agilent Technologies for their support to this
work.
References
[1] Code of Federal Regulations: Title 21, Chapter I, Subchapter B, Part 101: Food
Labeling. http://www.ecfr.gov/cgi-
bin/retrieveECFR?gp=1&SID=4bf49f997b04dcacdfbd637db9aa5839&ty=HTML&h=L&
mc=true&n=pt21.2.101&r=PART#se21.2.101_19 (accessed May 25, 2016).
[2] H. M. (Skip) Kingston and S. J. Haswell (Eds.), Microwave-Enhanced Chemistry:
Fundamentals, Sample Preparation, and Applications, 1st ed., American Chemical Society,
Washington, D.C., 1997, 800p.
[3] J. A. Nobrega and G. L. Donati, Microwave-Assisted Sample Preparation for
Spectrochemistry, In R. A. Meyers (ed.) and N. H. Bings (Assoc. ed.), Encyclopedia of
Analytical Chemistry, Wiley, Chichester, 2011, 23p.
[4] G. C. L. Araujo, M. H. Gonzales, A. G. Ferreira, A. R. A. Nogueira and J. A. Nobrega,
Effect of acid concentration on closed-vessel microwave-assisted digestion of plant
materials, Spectrochim. Acta Part B 57 (2002) 2121-2132.
[5] E.I. Muller, J.P. Souza, C.C. Muller, A.L.H. Muller, P.A. Mello, C.A. Bizzi,
Microwave-assisted wet digestion with H2O2 at high temperature and pressure using single
63
reaction chamber for elemental determination in milk powder by ICP-OES and ICP-MS,
Talanta 156-157 (2016) 232-238.
[6] J. Tang, Y. Ying, X.-D. Pan, W. Jiang, P.-G. Wu, Elements analysis of infant milk
formula by ICP-OES: a comparison of pretreatment methods, Accredit. Qual. Assur. 19
(2014) 99-103.
[7] C.J. Amarasiriwardena, I. Jayawardene, N. Lupoli, R.M. Barnes, M. Hernandez-Avila,
H. Hu, A.S. Ettinger, Comparison of digestion procedures and methods for quantification
of trace lead in breast milk by isotope dilution inductively coupled plasma mass
spectrometry, Anal. Methods 5 (2013) 1676-1681.
[8] E. Poitevin, Official methods for the determination of minerals and trace elements in
infant formula and milk products: a review, J. AOAC Int. 99 (2016) 42-52.
[9] D. A. Goncalves, T. McSweeney and G L. Donati, Characteristics of a resonant iris
microwave-induced nitrogen plasma, J. Anal. At. Spectrom. 31 (2016) 1097-1104.
[10] G. L. Donati, R. S. Amais, D. Schiavo and J. A. Nobrega, Determination of Cr, Ni, Pb
and V in gasoline and ethanol fuel by microwave plasma optical emission spectrometry, J.
Anal. At. Spectrom. 28 (2013) 755-759.
[11] R. S. Amais, G. L. Donati, D. Schiavo and J. A. Nobrega, A simple dilute-and-shoot
procedure for Si determination in diesel and biodiesel by microwave-induced plasma
optical emission spectrometry, Microchem. J. 106 (2013) 318-322.
64
[12] A. F. Lima, F. F. Lima, E. M. Richter and R. A. A. Munoz, Combination of sonication
and heating for metal extraction from inorganic fertilizers prior to microwave-induced
plasma spectrometry determinations, Appl. Acoust. 103 (2016) 124-128.
[13] N. Ozbek and S. Akman, Microwave plasma atomic emission spectrometric
determination of Ca, K and Mg in various cheese varieties, Food Chem. 192 (2016) 295-
298.
[14] P. Niedzielski, L. Kozak, M. Wachelka, K. Jakubowski and J. Wybieralska, The
microwave induced plasma with optical emission spectrometry (MIP-OES) in 23 elements
determination in geological samples, Talanta 132 (2015) 591-599.
[15] D. A. Goncalves, T. McSweeney, M. C. Santos, B. T. Jones and G. L. Donati, Standard
dilution analysis of beverages by microwave-induced plasma optical emission
spectrometry, Anal. Chim. Acta 909 (2016) 24-29.
[16] Z. Mester and R. Sturgeon, Sample Preparation for Trace Element Analysis, In D.
Barcelo (Ed.), Comprehensive Analytical Chemistry, Vol. XLI, Amsterdam, 2003, 1338p.
[17] K. O’Hanlon, L. Ebdon and M. Foulkes, Effect of easily ionizable elements on
solutions and slurries in an axially viewed inductively coupled plasma, J. Anal. At.
Spectrom. 11 (1996) 427-436.
[18] F. Motasemi, F. N. Ani, A review on microwave-assisted production of biodiesel,
Renew. Sust. Energ. Rev. 16 (2012) 4719-4733.
65
[19] T. Narukawa, E. Matsumoto, T. Nishimura, A. Hioki, Determination of sixteen
elements and arsenic species in brown, polished and milled rice, Anal. Sci. 30 (2014) 245-
250.
[20] C. Sola-Larrañaga, I. Navarro-Blasco, Optimization of a slurry dispersion method for
minerals and trace elements analysis in infant formulae by ICP OES and FAAS, Food
Chem. 115 (2009) 1048-1055.
[21] C. for F. S. and A. Nutrition, Labeling & Nutrition - Guidance for Industry: A Food
Labeling Guide (16. Appendix H: Rounding the Values According to FDA Rounding
Rules).
http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformati
on/LabelingNutrition/ucm064932.htm (accessed September 29, 2016).
[22] C. for F.S. and A. Nutrition, Labeling & Nutrition - Guidance for Industry: A Food
Labeling Guide (14. Appendix F: Calculate the Percent Daily Value for the Appropriate
Nutrients).
http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformati
on/LabelingNutrition/ucm064928.htm (accessed September 29, 2016).
66
CHAPTER IV
NATURALLY OCCURRING MOLECULAR SPECIES USED FOR PLASMA
DIAGNOSTICS AND SIGNAL CORRECTION IN MICROWAVE-INDUCED
PLASMA OPTICAL EMISSION SPECTROMETRY
Charles B. Williams, Bradley T. Jones and George L. Donati
The following manuscript was published in the Journal of Analytical Atomic
Spectrometry, 2018, 33, 1224-1232, and is reprinted by permission of the Royal Society of
Chemistry. Stylistic variations are due to the requirements of the journal. All of the
presented research was conducted by Charles B. Williams. The manuscript was prepared
by Charles B. Williams and George L. Donati. Supplementary material published online
for this chapter is presented in Appendix A.
67
ABSTRACT
In the present study, we evaluate the N2+ / OH emission intensity ratio as a
diagnostic tool for identifying the best instrumental operating conditions in microwave-
induced plasma optical emission spectrometry (MIP OES). This molecular species signal
ratio is compared with the traditional Mg II / Mg I ratio. Aluminum, Ba, Mn, Sr and Zn
(analytes), and high concentrations of C, Na, Ca, HNO3 and HCl (sample matrices) are
used as models to investigate the effects of complex matrices on analyte recoveries. The
N2+ / OH signal ratio is more sensitive to changes in plasma conditions than the Mg II / Mg
I ratio. Some other advantages include real-time monitoring capabilities, and the possibility
of independently tracking variations in both plasma and sample introduction. For less
complex matrices, the N2+ / OH signal ratio may be used for instrument optimization, which
ensures plasma conditions are as similar as possible when analyzing standard solutions and
samples. For analyses involving severe matrix effects, molecular species such as CN, N2,
N2+ and OH are used for signal correction. Significant improvements in accuracy are
achieved by employing the analyte-to-molecular species signal ratio, or their product, for
calibration. Both the use of the N2+ / OH signal ratio as a diagnostic tool, and of molecular
species for signal correction to minimize matrix effects are simple strategies that may
significantly contribute to expanding the analytical capabilities of MIP OES and facilitating
its application in routine analysis.
68
INTRODUCTION
Microwave-induced plasma optical emission spectrometry (MIP OES) is an
increasingly popular technique in atomic spectrometry, which has been thoroughly
reviewed in a monograph by Jankowski and Reszke.1 Such popularity is associated with
the relatively recent commercial introduction of a complete MIP OES instrument, which
provides adequate sensitivity for most elements as well as low costs of operation. The
commercial instrument features a resonant-iris, hammer-cavity MIP, which runs on
nitrogen, and a sequential monochromator and charge-coupled device (CCD) detector.2,3
Fundamental studies associated with plasma properties of this system are somewhat
limited. The literature has largely been application-based, with only two recent studies
focusing on topics associated with plasma fundamental properties. Goncalves et al., for
example, determined plasma robustness and temperature profiles for varying experimental
conditions.4 In another work, a comprehensive study by Chalyavi et al. characterized the
MIP using Thomson scattering and other techniques to determine electron number density
and temperature, among other properties.5
MIP OES has general features which are similar to the better-known inductively
coupled plasma (ICP) OES. However, the method is not as mature as ICP OES and it has
some marked differences. The primary distinction of relevance to matrix effects is that the
ICP is coupled to a radio-frequency-generated field, whereas the MIP is induced by a
standing wave generated by a magnetron. In an ICP, when the plasma impedance increases
during sample introduction, additional power is drawn to compensate such change. As a
consequence, the energy within the plasma remains relatively constant.6 In the
commercially available MIP OES, when the plasma impedance increases, the power put
69
out by the magnetron does not change, which results in a less energetic plasma.7 Because
of the fixed-power setup, MIP OES is significantly more prone to matrix effects than ICP
OES. Small variations in solution composition can lead to significant fluctuations in the
energy available for sample vaporization and analyte atomization and excitation, which, in
turn, will result in significant changes in emission intensity. Another major difference
between the two plasmas is that MIP runs on N2 rather than Ar. As noted by Chalyavi et
al., because N2 is a molecule, it consumes energy not only in translational degrees of
freedom, but also rotational and vibrational ones, while Ar, as an atom, has only
translational degrees of freedom.5 As a consequence, above ca. 5000 K, increased applied
power does not significantly result in higher temperatures for a N2 plasma, whereas
increased applied power continues to raise temperature in an ICP up to about 10,000 K.
Thus, plasma temperatures are necessarily lower in a N2 MIP than in an Ar ICP.
It is important to note that other MIP devices have been developed for OES
analyses, which can be based on a Beenakker cavity,8 the microwave-induced nitrogen
discharge at atmospheric pressure (MINDAP),9 an Okamoto cavity,10 a surfatron,11 the
torche a injection axiale (TIA) or torche a injection axiale sur guide d'ondes (TIAGO)
designs,12,13 and more recently, the microwave-sustained, inductively coupled,
atmospheric-pressure plasma (MICAP).14 However, given its increasing popularity and
cost-effective applications in different fields, we focus the present study on a commercially
available, Hammer-cavity-based MIP OES instrument.2 There have generally been no
plasma diagnostic tools developed specifically for commercially available MIP OES
systems, with most strategies transferred from ICP. For example, Goncalves et al. used the
concept of robustness as originally proposed by Mermet to describe the stability of the
70
plasma.4,15 According to Mermet’s definition, a robust plasma is one which is highly
tolerant to matrix effects, and which has a Mg ion-to-atom intensity ratio (Mg II / Mg I)
greater than 10. However, such Mg II / Mg I value is impossible to achieve with the
commercially available MIP, as the energy of the plasma is considerably lower. Goncalves
et al. found the Mg II / Mg I ratio to rarely exceed a value of 2, and to commonly be less
than 1 in MIP OES, pointing to a plasma in which most emitting species are neutral atoms
and molecules rather than ions.4 This notion may be confirmed by the analytical
wavelengths recommended by the instrument control software, which are mainly
composed of atomic lines rather than ionic ones. Considering the specific characteristics
of a N2 MIP, particularly the fact that most elements are determined as atoms rather than
ions, maximizing robustness as a strategy to optimize analytical accuracy may not be
effective. Therefore, a more nuanced approach than simply robustness may be appropriate
in MIP OES, possibly including monitoring variations in the plasma on a per-sample basis.
Another aspect associated with robustness is that it involves the addition of Mg as
a test element. It would be advantageous to be able to estimate robustness (or monitor
changes in the plasma) for any solution without having to add a test element. Thus,
naturally occurring molecular species, which are part of the MIP background, may
represent a promising source of information about plasma conditions. In the present study,
we evaluate some of these species as plasma diagnostic tools. In this context, a few studies
describing the use of molecular species (e.g. N2+ and OH) to estimate plasma temperature
can be found in the literature.16-18 However, they are rarely applied to MIP, nor are they
used in the context of atomic spectrometry. The molecular emission lines present in a MIP
were detailed by Jankowski and Reszke, and include peaks for N2, N2+, CH, CN, OH and
71
C2.1 Chalyavi et al. recently described the spectral background for the new commercial
MIP OES system, and note that it is dominated by several molecular species such as NO,
NH, OH, N2, and N2+.5 These species are usually treated as a nuisance, as they can interfere
with atomic emission measurements. Frentiu et al., for example, added methane gas to the
Ar supply of a capacitively-coupled plasma in order to quench interference from OH and
N2 emission bands.19 Chalyavi et al. note that a background modelling algorithm is
included in the instrumental control software of the commercial MIP OES in order to
remove spectral background due to molecular emission bands.5 On the other hand, Lowery
et al. have used molecular species to improve MIP OES’ performance in biodiesel
analysis.20 In this study, emission band peaks from OH and N2+, naturally occurring in the
MIP’s spectral background, were used as “molecular probes” to correct for changes in
emission intensity due to matrix effects. Significant improvements in accuracy were
observed when a ratio or multiplication between the emission signals from the analytes and
the molecular species were used for calibration rather than the analytical signals alone.
As discussed before, matrix effects can be severe in MIP OES determinations due
to typically lower plasma temperatures. Zhang and Wagatsuma investigated the effects of
Na, Ca and HNO3 on emission intensities in an Okamoto-cavity MIP OES.21 They found
that easily ionizable elements (EIEs) such as Na and Ca cause signal enhancement in most
atomic lines presenting lower excitation energies, and signal suppression in ionic lines and
atomic lines of higher excitation energies. The authors suggested that these effects were a
result of a shift in excitation equilibrium, which was caused by an increase in the plasma
electron number density (ne). They also found that HNO3 has a suppressive effect on most
analytical signals. In Zhang and Wagatsuma’s study, the sample was naturally aspirated
72
(i.e. no peristaltic pump used), which may explain part of the effects observed. It is also
important to note that signal intensity suppression due to HNO3 has been shown to be less
severe than the effects caused by the EIEs in that work.
In the present study, we investigate the use of the intensity ratio between the N2+
emission band peak at 391.439 nm and the OH band peak at 308.970 nm as a plasma
diagnostic tool for identifying the best plasma conditions to minimize matrix effects in MIP
OES. We evaluate a commercial MIP OES instrument (Agilent 4200 MP-AES), for which
results may be applicable to other MIP systems. The N2+ signal may be used as a proxy for
the energy available within the plasma, while the OH signal could be a proxy for the
efficiency of the sample introduction system. We examine the effects of high
concentrations of Na, Ca, HNO3, HCl and C on analyte recoveries for atoms and ions
presenting a wide range of excitation energies, i.e. Al, Ba, Mn, Sr and Zn. The N2+ / OH
signal ratio is compared to the traditional Mg II / Mg I ratio, and other plasma naturally
occurring molecular species (including CN, N2, N2+ and OH) are evaluated as signal
correction species to improve accuracy in complex matrix analysis by MIP OES.
73
MATERIALS AND METHODS
Instrumentation
All experiments were carried out on a commercial resonant-iris, Hammer-cavity
MIP OES instrument (4200 MP-AES, Agilent Technologies, Santa Clara, CA, USA).
Nitrogen gas was supplied from a liquid N2 Dewar (N2 99.998 % pure, Air Products and
Chemicals, Allentown, PA, USA). The sample introduction system comprised an SPS 4
autosampler, a double-pass cyclonic spray chamber and an inert OneNeb nebulizer. Each
solution was analyzed in three replicates of a 3-s integration time each. Emission band
peaks for CN (387.147 nm), N2 (337.097 nm), N2+ (391.439 nm) and OH (308.970 nm)
were monitored as potential candidates for plasma diagnostics and signal correction.
Emission spectra for these molecular species are presented in Appendix A (Figs. S1 - S8).
Because these specific molecules are not available as options in the instrument control
software, nearby lines for Fe I, Ti I, Nb I, and Tb II (where I and II represent atomic and
ionic species, respectively) were chosen, and the molecular emission peaks were adjusted
as described in a previously published work.20 Analytical emission lines at 396.152,
455.403, 403.076, 407.771 and 213.857 nm were used for Al I, Ba II, Mn I, Sr II and Zn I,
respectively.
Each solution was analyzed in a range of nebulization gas flow rates between 0.3
and 1.0 L min-1, at intervals of 0.1 L min-1. All emission signals were collected at the center
of the plasma (plasma viewing position zero, as identified by the instrument control
software).4 Atomic emission intensities at 280.271 nm (Mg II) and 285.213 nm (Mg I) were
used to calculate the Mg II / Mg I ratio. In this case, a detector correction factor of 1.1 was
employed according to the method outlined by Dennaud et al., and as previously
74
determined by Goncalves et al..4,22 The N2+ / OH and Mg II / Mg I signal ratios were
compared in various plasma conditions. General instrumental operating conditions are
listed in Table XII. Instrumental operating conditions used in MIP OES..
Table XII. Instrumental operating conditions used in MIP OES.
Instrumental parameter Operating
condition
Microwave frequency (MHz) 2450
Applied power (kW) 1.0
Peristaltic pump speed (rpm) 15
Integration time (s) 3
Number of replicates 3
Nebulization gas flow rate (L min-1) 0.3 - 1.0
Reagents and standard reference solutions
Concentrated acids were obtained from Fischer Scientific: HNO3 (Trace Metals Grade, ca.
16 M), and HCl (ACS+ Grade, ca. 12 M). All solutions were prepared in 18.2 MΩ cm-1
distilled-deionized H2O (Milli-Q, Millipore, Bedford, MA, USA). Single-element stock
solutions of Al, Ba, Fe, Mg, Mn, Sr and Zn (SPEX CertPrep, Metuchen, NJ, USA) were
used to prepare the working solutions. Solid urea (Fisher) was used to investigate the
75
effects of high carbon content in the plasma. In addition, 10,000 mg L-1 stock solutions of
Na and Ca (SPEX) were used to investigate the effects of EIEs on all analytes evaluated.
A series of solutions containing 2 mg L-1 of Al, Ba, Mn, Sr and Zn, and 1 % v v-1
HNO3 was prepared in various sample matrices: 1000 mg L-1 Na, 1000 mg L-1 Ca or 1000
mg L-1 C. Solutions containing 2 mg L-1 of these same analytes were also prepared in 20
% v v-1 HCl or 20 % v v-1 HNO3. Blank and calibration curve standard solutions containing
these analytes were prepared in 1 % v v-1 HNO3.
RESULTS AND DISCUSSION
Determination of the detector response correction factor for calculating the N2+ /
OH signal ratio
We have evaluated four of MIP’s most intense background signal sources as part
of a strategy to monitor plasma conditions and improve accuracy in MIP OES analysis.
Individual signal intensities and different combinations of CN, N2, N2+ and OH signals
were examined, and as discussed in later sections, the best results were achieved when
employing the N2+ / OH signal ratio. To prevent any signal bias due to differences in
detector response at each wavelength, an experiment was carried out to determine the
extent of such a bias and correct it while calculating the N2+ / OH signal ratio. Goncalves
et al. used low-intensity Tb lines near 285.213 and 280.271 nm to minimize any
interference from residual Mg in solution when determining the detector response
correction factor used in Mg II / Mg I experiments.4 Because the molecular species used in
the current study are present in high concentrations as part of the plasma or the sample
76
solvent, no similar method could be used. Alternatively, a 150 W, 15 V light bulb (Osram
HLX 64633) was employed as a continuum source to determine detector sensitivity at
different wavelengths. With the plasma off and the torch removed, the radiation intensity
produced by a light bulb operated at 10.0 A was recorded at 308.958 ± 0.500 nm and
391.470 ± 0.500 nm (n = 3) using the “quick read” function of the instrument. Differences
in radiation intensity due to the blackbody emission from the tungsten filament were taken
into account using Plank’s law (Eq. 3), where 𝐵𝐵𝜆𝜆𝑏𝑏, λ and T represent spectral radiance in
W/sr·cm2·nm, wavelength in nm, and temperature in K, respectively.23 The tungsten
filament temperature used in blackbody radiation calculations was estimated using Eq. 4,
where T, ddp, L/A and i represent temperature (K), potential across the filament (V),
filament’s length-to-area ratio (m-1), and applied current (A).24 In this case, 12.81 V, 1.13
x 106 m-1 and 10.0 A were used to estimate a temperature of 3460 K, which is in agreement
with the bulb’s reported color temperature of 3450 K.
𝐵𝐵𝜆𝜆𝑏𝑏 = 1.190𝑥𝑥1016∙ 𝜆𝜆−5
𝑒𝑒1.438𝑥𝑥107/𝑇𝑇𝑇𝑇 −1 (3)
𝑇𝑇 = 𝑑𝑑𝑑𝑑𝑑𝑑𝐿𝐿𝐴𝐴 ∙ 𝑖𝑖
0.80548
∙ 2.1287 ∙ 108 (4)
The radiation source emits more intensely at 391 than at 309 nm (Eq. 3), so a blackbody
correction factor of 5.21 was used to correct radiation intensities recorded at 391 nm (i.e.
the signal intensity at 391 nm was divided by 5.21 to prevent bias). It was determined that
77
the detector is more sensitive at 391.439 nm than at 308.970 nm. Therefore, a correction
factor of 0.758 must be used when measuring the N2+ / OH signal ratio in our instrument.
This experimentally determined correction factor is in agreement with quantum
efficiencies reported at the respective wavelengths for the back-thinned CCD detector used
in the MIP OES instrument evaluated in this study. The detector response correction factor
calculated using values estimated from the detector’s quantum efficiency graph (available
for the commercial MIP OES) was estimated as approximately 0.8.
78
Effects of a changing plasma on the N2+ / OH and Mg II / Mg I ratios
As discussed earlier, we have evaluated individual signals and different
combinations of the emission intensities for CN, N2, N2+ and OH as plasma diagnostic
tools. Figs. S9 - S12 (Appendix A) show that the best results are achieved with the N2+ /
OH ratio. In addition to providing a consistently greater sensitivity to plasma changes
across all conditions evaluated (see Figs. S11 and S12), using a signal ratio (N2+ / OH in
this case) rather than individual signals allows for a broader application of the method.
Variations in the system will cause different effects in each individual signal. However,
such effects will be proportional in a signal ratio, which makes it more consistent over time
and across instruments, as well as more efficiently transferable between laboratories than
individual signals alone.
As noted in previously published works, the traditional definition of plasma
“robustness” may not fit well with a MIP.5,25,26 However, signal ratios such as the N2+ / OH
may still be used to monitor changes in the plasma due to instrumental fluctuations, or as
different matrices are introduced. Such strategy may then be employed to correct for signal
variation and improve accuracy when analyzing complex-matrix samples. In the present
study, we have compared the performances of the N2+ / OH signal intensity ratio and the
traditional Mg II / Mg I ratio to identify changes in the MIP. Figure 8 shows the effects of
nebulization gas flow rate (NGFR) on Mg II / Mg I and N2+ / OH. As it can be observed,
both signal ratios drop with an increase in NGFR, which is related to more solution
reaching the plasma, changes in electron number density (ne) and plasma cooling.4,5,21
However, the N2+ / OH ratio is more sensitive to these changes than the Mg II / Mg I ratio.
While the latter presents a quadratic relationship with NGFR (Figure 8a), the former has a
79
more intense power correlation with NGFR (Figure 8b). A similar trend is observed when
introducing increasing concentrations of sodium into the plasma at different NGFRs
(Figure 9). In this case, both signal ratios present a quadratic correlation with the
concentration of Na in solution. However, the N2+ / OH ratio is still more sensitive to
changes than Mg II / Mg I, as observed by comparing the curves’ gradients in Figure 9a
and b, and Figure 9c and d.
Figure 8. Effects of nebulization gas flow rate on Mg II / Mg I (a), and N2+ / OH (b) in
MIP OES.
80
Figure 9. Effects of sodium concentration on the Mg II / Mg I and N2+ / OH signal ratios
at nebulization gas flow rates of 0.6 L min-1 (a) and (b), and 1.0 L min-1 (c) and (d).
Figure 10. Effects of nebulization gas flow rate on (a) Mg II (280.271 nm) and Mg I
(285.213 nm), and (b) N2+ (391.439 nm) and OH (308.970 nm).
81
In addition to a potentially more sensitive diagnostic tool, the N2+ / OH ratio may
provide a more independent measure of change in both the N2 MIP conditions and the
sample introduction process than the traditional Mg II / Mg I ratio. The excitation
mechanism to produce N2+* involves species closely associated with the plasma, with the
promotion of N2+ from the ground state (X 2Σg, v) to the excited state (B 2Σu, v) by direct
electron impact and energy exchange reactions from collisions with vibrationally excited
N2 molecules.19,27 On the other hand, the OH radical is mainly produced from water, and
therefore, it is closely related to the nebulization process. In Figure 10b, for example, the
correlation coefficient between the NGFR and the OH signal intensity is R2 = 0.971.
Chalyavi et al. previously demonstrated that the OH emission bands are absent in a dry
MIP, but are easily detected upon introduction of 1% v v-1 HNO3.5 According to Frentiu et
al., OH radical species form following a reaction between dissociated O2 and water
molecules: O + H2O 2 OH.19 Additional evidence to this reaction is presented in Figure
11. When air is introduced into the plasma using the external gas control module (EGCM)
available in the commercial instrument, but water is not present (peristaltic pump speed at
0 rpm), the emission signal for OH at 308.970 nm is not distinguishable from the
background noise (Figure 11a). On the other hand, when no air is added, but distilled-
deionized water is introduced into the MIP at a peristaltic pump speed of 15 rpm and NGFR
of 0.7 L min-1, the emission band peak for OH can be easily identified (Figure 11b). In this
case, enough oxygen may be provided by the atmosphere, which is dragged into the plasma
by the N2 gas flow generating the MIP. As more oxygen gas is made available by
introducing air through the EGCM, with water uptake rate kept constant (peristaltic pump
speed of 15 rpm, and NGFR at 0.7 L min-1), the OH emission signal proportionally
82
increases (Figs. 4c-4e). These results are in agreement with the reaction mechanism
proposed by Frentiu et al..
Figure 11. MIP OES spectra for the OH molecular species (band peak at 308.970 nm).
Each spectrum corresponds to a different plasma / sample introduction condition: (a) no
water introduced (peristaltic pump speed at 0 rpm) and air added to the plasma at a medium
flow rate; (b) water introduced, with no air added to the plasma; (c) water introduced, with
a low air flow rate; (d) water introduced with a medium air flow rate; and (e) water
introduced with a high air flow rate. Conditions for water introduction: peristaltic pump
speed and NGFR of 15 rpm and 0.7 L min-1, respectively. No air flow rate specification
(other than low, medium and high) is available from the instrument’s EGCM.
83
Based on the mechanisms discussed for N2+ and OH formation, it is reasonable to assume
that combining the emission signals from these species into a ratio may be useful for
monitoring plasma and sample introduction variations in a more independent fashion than
with traditional plasma diagnostic tools such as the Mg II / Mg I ratio. As observed in
Figure 10a, especially at lower NGFRs, signals from both Mg lines generally increase with
the NGFR, as they are directly dependent on the amount of Mg going through the sample
introduction system. On the other hand, signals from N2+ and OH go in opposite directions
with increasing NGFR (Figure 10b), which may indicate more independence between their
original sources.
Using the N2+ / OH ratio to optimize instrumental operating conditions
The N2+ / OH and Mg II / Mg I signal ratios compare well, as observed in Figure
12. In this case, solutions containing 5 mg L-1 of Mg with increasing Na concentrations
(50, 100, 200, 500 and 1000 mg L-1) are introduced into the MIP under different
nebulization gas flow rates (0.6, 0.8, 1.0 and 1.2 L min-1). As discussed before, the N2+ /
OH ratio may present distinct advantages when compared with the traditional Mg II / Mg
I method. In addition to a potentially higher sensitivity to changes and the possibility of
more independently monitoring the sample introduction system and the plasma, it is also
convenient for routine applications. The N2+ / OH ratio method may be applied to any
sample, at any time, with no need for an additional sample preparation step (i.e. Mg does
not need to be added to the sample solutions). Thus, plasma conditions may be routinely
monitored (almost simultaneously since this is a fast sequential system) as different
samples are introduced into the MIP.
84
Figure 12. Correlation between the Mg II / Mg I and N2+ / OH signal ratios in the presence
of sodium at different nebulization gas flow rates (0.6 - 1.2 L min-1). Within each flow rate
group, Na concentrations in solution vary in the 50 - 1000 mg L-1 range (from right to left
on the graph).
85
The N2+ / OH signal ratio could also be used to determine the most favorable
conditions to minimize matrix effects. For example, one could monitor this signal ratio in
a range of NGFRs while running a standard solution, and then repeat the procedure for a
sample. The most favorable condition would be the one for which the N2+ / OH ratio
determined for the standard solution most closely matches the one from the sample. At that
specific NGFR, plasma conditions during sample introduction would be as close as
possible to those observed during the introduction of the calibration standard solutions,
which could improve accuracy in MIP OES measurements. Various sample matrices and
five elements with a wide range of excitation energies (i.e. Al, Ba, Mn, Sr and Zn) were
evaluated to investigate this hypothesis. To simulate a routine application, analyte
concentrations were determined using the external standard method, with no matrix-
matching, and calibration solutions prepared in 1 % v v-1 HNO3. Carbon (C as urea), Na
and Ca (all at 1000 mg L-1 each), and HNO3 and HCl (20 % v v-1 each) were separately
evaluated as matrices. Figure 13 shows the effects of each matrix, at various NGFRs, on
accuracy. For the analytes, the y-axis in Figure 13 represents the average percent error of
recovery, which is an average of absolute percent differences from 100 % recovery of all
analytes. For the N2+ / OH signal ratio, the y-axis represents the percent difference between
an average of N2+ / OH calculated for the standard solutions prepared in 1 % v v-1 HNO3
(calibration solutions), and values calculated for individual matrices. Note that, in general,
the lowest average percent errors of recovery coincide with the lowest percent difference
between N2+ / OH ratios calculated for standards and samples. Thus, without having
previous knowledge of the matrix, one could assume that the best overall accuracies may
be achieved at a NGFR of 0.9, 0.3 and 0.6 L min-1 for Al, Ba, Mn, Sr and Zn determination
86
in a matrix containing high concentrations of C, Na and HNO3, respectively (Figs. 6a, 6b
and 6d). On the other hand, the results show that this is not valid for a matrix with1000 mg
L-1 Ca or 20 % v v-1 HCl, in which case the best results would be found at the second lowest
N2+ / OH signal ratio difference (Figs. 6c and 6e). In Figure 13a, b, 6c, 6d and 6e, these
values are: 10.70 and 6.67 (1000 mg L-1 Na, NGFR = 0.3 L min-1); 0.664 and 0.446 (1000
mg L-1 Ca, NGFR = 1.0 L min-1); 1.25 and 1.20 (20 % v v-1 HNO3, NGFR = 0.6 L min-1);
and 0.664 and 0.655 (20 % v v-1 HCl, NGFR = 1.0 L min-1).
87
Figure 13. Relationship between average analyte percent recovery and N2+ / OH signal
ratio in different matrices: (a) 1000 mg L-1 C (as urea), (b) 1000 mg L-1 Na, (c) 1000 mg
L-1 Ca, (d) 20 % v v-1 HNO3, and (e) 20 % v v-1 HCl. Analytes evaluated: Al, Ba, Mn, Sr
and Zn at 2.0 mg L-1.
88
Signal correction using molecular species
Using the N2+ / OH signal ratio to identify the most favorable plasma conditions
and improve accuracy may be efficient for less severe matrix effects. For a matrix with
1000 mg L-1 C, for example, the average percent error of recovery went from 21.0 to 2.0
% by changing the NGFR from 0.4 to 0.9 L min-1 (Figure 13a). The recoveries for Al, Ba,
Mn, Sr and Zn went from 116, 112, 133, 107 and 126 % to 101, 98, 102, 97 and 102 %,
while the percent difference between standard solution N2+ / OH and sample N2
+ / OH went
from 18.0 to 0.4 %, respectively. On the other hand, some matrix effects may be too severe
to be corrected by just changing the NGFR. For Na at 1000 mg L-1, for example, the average
percent error of recovery at the most favorable plasma condition (i.e. NGFR at 0.3 L min-
1) was calculated as 77.0 % (Figure 13b), with Al, Ba, Mn, Sr and Zn recoveries of 112,
199, 130, 229 and 146 %, respectively. Table XIII. Analyte percent recoveries for 2.0 mg
L-1 solutions of Al, Ba, Mn, Sr and Zn prepared in different matrices. shows recoveries for
individual analytes at the best NGFRs, according to Figure 13. As previously observed by
Zhang and Wagatsuma, matrix effects due to EIEs are more pronounced than those due to
inorganic acids.21
89
Table XIII. Analyte percent recoveries for 2.0 mg L-1 solutions of Al, Ba, Mn, Sr and Zn
prepared in different matrices.
Sample matrix NGFR (L min-1)a Analyte Recovery (%)
1000 mg L-1 C 0.9 Al 101
Ba 98
Mn 102
Sr 97
Zn 102
APERb 2.0
1000 mg L-1 Na 0.3 Al 112
Ba 199
Mn 130
Sr 229
Zn 146
APERb 77.0
1000 mg L-1 Ca 1.0 Al 172
Ba 131
Mn 144
Sr 143
Zn 16
APERb 58.0
90
20 % v v-1 HNO3 0.6 Al 86
Ba 81
Mn 84
Sr 80
Zn 107
APERb 16.0
20 % v v-1 HCl 1.0 Al 93
Ba 90
Mn 94
Sr 87
Zn 96
APERb 9.0
a Nebulization gas flow rate (NGFR) providing the lowest average percent error of
recovery.
b Average percent error of recovery: absolute percent difference from 100 %
recovery of all analytes in a 2.0 mg L-1 solution.
91
A simple strategy to improve accuracy when severe matrix effects are present is the
use of MIP naturally occurring molecular species to correct for analytical signal bias.20 In
the present study, emission band peaks for CN (387.147 nm), N2 (337.097 nm), N2+
(391.439 nm) and OH (308.970 nm) were recorded in the same run as the analytical signals,
and then evaluated as signal correction species for improving accuracy while analyzing
different matrices. This strategy was investigated for NGFRs ranging between 0.3 and 1.0
L min-1. The analytical signals (A) were either divided or multiplied by the emission
intensity of the signal correction species (X), i.e. A / X or A · X. The calibration plot was
then built with A / X or A · X on the y-axis, and analyte concentration on the x-axis. The
results for the two best correction strategies for each sample matrix are presented in Table
XIV. The two best signal correction strategies for each sample matrix. The analyte percent
recoveries are associated with a 2.0 mg L-1 solution of each analyte..
For matrices causing less severe effects, a slight improvement in analyte recovery
was observed when using the molecular species correction strategy. The average percent
error of recovery (APER) for determinations in 1000 mg L-1 C and 20 % v v-1 HCl went
from 2.0 and 9.0 % (Table XIII) to 0.4 and 2.0 % (Table XIV), respectively. For the 1000
mg L-1 C matrix, recoveries for all analytes were in the 96 - 105 % when using A / X or
A · X, with X = N2, N2+ or OH. As expected, CN is not suitable for applications with high
C content matrices. Recoveries were in the 23 - 50 %, and 271 - 359 % for A / CN and A·
CN, respectively. For the 20 % v v-1 HCl matrix, analyte recovery was in the 92 - 105 %
range for all analytes by employing any combination of signal correction (i.e. A / X or A ·
X) and using either CN, N2, N2+ or OH.
92
Table XIV. The two best signal correction strategies for each sample matrix. The analyte
percent recoveries are associated with a 2.0 mg L-1 solution of each analyte.
Sample
matrix
Signal correction strategya Analyte recovery (%) APER
(%)b
Al Ba Mn Sr Zn
1000 mg
L-1 C
A / OH 99 100 100 100 100 0.4
NGFR (L min-1) 1.0 1.0 0.6 1.0 0.6
A · N2+ 99 99 100 99 100 1.0
NGFR (L min-1) 0.9 0.6 0.9 0.4 0.9
1000 mg
L-1 Na
A · CN 100 107 92 96 88 7.0
NGFR (L min-1) 0.4 0.4 0.5 0.5 0.3
A · N2 91 108 94 109 110 9.0
NGFR (L min-1) 0.5 0.6 0.6 0.6 0.3
1000 mg
L-1 Ca
A · N2+ 102 97 101 100 88 6.0
NGFR (L min-1) 0.6 1.0 0.9 0.8 0.3
A · CN 123 112 99 119 94 15.0
NGFR (L min-1) 0.6 0.9 0.7 0.8 0.3
20 % v v-1
HNO3
A / CN 101 99 102 98 106 3.0
NGFR (L min-1) 0.3 0.3 0.3 0.3 0.6
A / N2+ 100 100 100 99 109 4.0
NGFR (L min-1) 1.0 0.3 0.7 0.3 0.6
20 % v v-1
HCl
A / N2+ 100 98 98 98 99 2.0
NGFR (L min-1) 0.7 0.3 0.4 0.3 1.0
A · OH 105 99 103 99 99 3.0
NGFR (L min-1) 0.8 0.3 0.4 0.3 0.8 a A represents the analytical signal, and NGFR is the nebulization gas flow rate.
b Average percent error of recovery: absolute percent difference from 100 %
recovery of all analytes in a 2.0 mg L-1 solution.
93
For the other matrices, the molecular species signal correction approach allowed for significant
improvements in accuracy, as observed in Table XIII and Table XIV. In all cases, one of the two
best signal correction strategies, or a combination of them, can provide recoveries in the 90 - 110
% range for all the analytes and matrices evaluated in this study.
Conclusions
The use of the N2+ / OH signal ratio compares favorably with the traditional Mg II / Mg I
ratio as a plasma diagnostic tool. Among its advantages are the higher sensitivity to changes in
instrument conditions, and the quasi-simultaneous, sample-by-sample monitoring capabilities. No
additional solution preparation step is required since N2+ and OH are both plasma naturally-
occurring species. Considering the specific source of each of these molecular species, they may
also be used to optimize plasma / sample introduction conditions and improve accuracy when
analyzing less complex matrices. As the most useful analytical lines in MIP OES are primarily
atomic, maximizing robustness is not an effective method of optimizing accuracy. Instead, matrix
effects can be ameliorated by altering the instrumental operating conditions so that the plasma is
as similar as possible when introducing standard solutions and samples.
On the other hand, simply adjusting the instrumental operating conditions is not sufficient
to improve accuracy in analyses involving severe matrix effects. A simple and efficient strategy to
minimize matrix effects is the use of CN, N2, N2+ or OH for signal correction. The analyst may
just need to evaluate these molecular species in a range of NGFRs and identify the optimal
conditions and most effective signal correction operation as part of their method development.
94
Both the use of the N2+ / OH signal ratio as a diagnostic tool, and of molecular species for
signal correction to minimize matrix effects are simple strategies that may significantly contribute
to expanding the analytical capabilities of MIP OES and facilitating its application.
Conflicts of Interest
There are no conflicts of interest to declare.
Acknowledgements
The authors would like to thank the Department of Chemistry and the Graduate School of
Arts and Sciences at Wake Forest University for their support to this work.
References
1. K. J. Jankowski and E. Reszke, Microwave induced plasma analytical spectrometry, Royal
Society of Chemistry, Cambridge, 2010.
2. M. R. Hammer, Spectrochim. Acta Part B, 2008, 63, 456-464.
3. T. D. Hettipathirana, J. Anal. At. Spectrom., 2013, 28, 1242 - 1246.
4. D. A. Goncalves, T. McSweeney and G. L. Donati, J. Anal. At. Spectrom., 2016, 31, 1097-1104.
5. N. Chalyavi, P. S. Doidge, R. J. S. Morrison and G. B. Partridge, J. Anal. At. Spectrom., 2017,
32, 1988-2002.
6. S. Nowak, J. A. M. Van Der Mullen and D. C. Schram, Spectrochim. Acta Part B, 1988, 43,
1235-1245.
95
7. H. Wiltsche, M. Winkler and P. Tirk, J. Anal. At. Spectrom., 2015, 30, 2223-2234.
8. C. I. M. Beenakker, Spectrochim. Acta Part B, 1976, 31, 483-486.
9. R. D. Deutsch, J. P. Keilsohn and G. M. Hieftje, Appl. Spectrosc., 1985, 39, 531-534.
10. Y. Okamoto, Anal. Sci., 1991, 7, 283-288.
11. S. Luge and J. A. C. Broekaert, Mikrochim. Acta, 1994, 113, 277-286.
12. M. Moisan, G. Sauve, Z. Zakrzewski and J. Hubert, Plasma Sour. Sci. Technol., 1994, 3, 584-
592.
13. M. Moisan, Z. Zakrzewski and J. C. Rostaing, Plasma Sour. Sci. Technol., 2001, 10, 387-394.
14. A. J. Schwartz, Y. Cheung, J. Jevtic, V. Pikelja, A. Menon, S. J. Ray and G. M. Hieftje, J.
Anal. At. Spectrom., 2016, 31, 440-449.
15. J.-M. Mermet, Anal. Chim. Acta, 1991, 250, 85-94.
16. S. Pellerin, J. M. Cormier, F. Richard, K. Musiol and J. Chapelle, J. Phys. Appl. Phys., 1996,
29, 726-739.
17. H. Nassar, S. Pellerin, K. Musiol, O. Martinie, N. Pellerin and J.-M. Cormier, J. Phys. Appl.
Phys., 2004, 37, 1904-1916.
18. M. Tsumaki and T. Ito, AIP Adv., 2017, 7, 125211 (1-7).
19. T. Frentiu, M. Ponta, A. I. Mihaltan, E. Darvasi, M. Frentiu and E. Cordos, Spectrochim. Acta
Part B, 2010, 65, 565-570.
20. K. L. Lowery, T. McSweeney, S. P. Adhikari, A. Lachgar and G. L. Donati, Microchem. J.,
2016, 129, 58-62.
96
21. Z. Zhang and K. Wagatsuma, Spectrochim. Acta Part B, 2002, 57, 1247-1257.
22. J. Dennaud, A. Howes, E. Poussel and J.-M. Mermet, Spectrochim. Acta Part B, 2001, 56, 101-
112.
23. D. Ingle and S. R. Crouch, Spectrochemical analysis, Prentice Hall, Englewood Cliffs, 1988.
24. A. Virgilio, C. K. Healy, J. A. Nóbrega, B. T. Jones and G. L. Donati, Microchem. J., 2013,
110, 758-763.
25. J.-M. Mermet and E. Poussel, Appl. Spectrosc., 1995, 49, 12A-18A.
26. E. Poussel, J.-M. Mermet and O. Samuel, Spectrochim. Acta Part B, 1993, 48, 743-755.
27. J. Henriques, E. Tatarova, F. M. Dias and C. M. Ferreira, J. Appl. Phys., 2011, 109, 023302
(1-8).
97
Multi-flow calibration applied to microwave-induced plasma optical
emission spectrometry
Charles B. Williams, Bradley T. Jones and George L. Donati
The following manuscript was accepted by the Journal of Analytical Atomic
Spectrometry on April 5, 2019, DOI: 10.1039/C9JA00091G. Stylistic variations are due to the
requirements of the journal. All of the presented research was conducted by Charles B Williams.
The manuscript was prepared by Charles B. Williams and edited by George Donati.
98
ABSTRACT
Multi-flow calibration (MFC) is based on a single calibration standard and multiple
nebulization gas flow rates (Q). Analytical signals are recorded at different Q conditions, and
intensities from calibration standard and sample are plot on the x and y axes, respectively. The
analyte concentration in the sample is calculated by multiplying the standard concentration by the
calibration plot slope. In the present work, MFC is used to determine Cr, Cu, Fe and Mn in water
and food samples by microwave-induced plasma optical emission spectrometry. Analyte percent
recoveries for certified reference materials and addition/recovery experiments were in the 91-
112% and 84-134% ranges for MFC and external standard calibration (EC), respectively. The
limits of detection (LODs) for Cr, Cu, Fe and Mn were 20, 5, 7 and 2 µg L-1 using MFC (0.6, 8,
20 and 1 µg L-1 for EC). Precisions were in the 0.9-12.2% and 1.9-23.9% ranges for MFC and EC,
respectively. MFC may minimize matrix effects as it exposes all solutions to a variety of plasma
conditions. This normalizing effect may be capable of improving accuracies compared with EC
for simple to moderately complex matrix samples. One of MFC’s main limitations is the potential
for systematic errors associated with solution preparation (a single calibration standard is used).
Variation in Q may also result in higher LODs and lower sample throughputs compared with EC.
On the other hand, neither negative effects due to a poor choice of Q nor additional experiments
to optimize it are required with MFC.
99
INTRODUCTION
Traditional calibration methods used in atomic spectrometry, such as the external standard
calibration (EC), typically involve preparing standard solutions at a range of known
concentrations, recording the respective analytical signal intensities for those solutions to
determine sensitivity, and then using the sensitivity value to calculate the analyte concentration in
a given sample.1-3 A series of standard solutions, rather than a single one, is generally employed
to minimize uncertainty in the mathematical model used for calibration.2,4 The method of standard
additions (SA) is based on a similar principle, with the difference of adding a series of standard
solutions directly to the sample to control for matrix effects.5,6
Recently, there has been an increased interest in alternative calibration methods to improve
accuracy and sample throughput in atomic spectrometry applications. Some of these methods have
been discussed in a recent review by Carter et al., which is primarily focused on inductively
coupled plasma mass spectrometry (ICP-MS), but also covers some applications involving ICP
and microwave-induced plasma (MIP) as atomization/excitation sources for optical emission
spectrometry (ICP OES and MIP OES, respectively).3 While traditional calibration methods use
analyte concentration as the independent variable and analytical signal intensity as the dependent
variable when building the calibration plot, some of the new strategies take advantage of multiple
variables such as wavelength (or transition energy),7-9 isotopes,10 or polyatomic species11 to
determine the analyte concentration in a sample. Multi-energy calibration (MEC), for example,
makes use of multiple transition energies (corresponding to different emission or absorption
wavelengths), combined with a matrix-matching approach, to improve accuracy in ICP OES, MIP
OES, high-resolution continuum source atomic absorption spectrometry (HR-CS AAS), and laser-
induced breakdown spectroscopy (LIBS).7-9 In ICP-MS, multi-isotope and multispecies calibration
100
(MICal and MSC) employ a similar strategy as MEC in which multiple analyte isotopes or multiple
polyatomic species containing the analyte ion are used to improve the accuracy of the calibration
procedure.10,11
MEC, MICal and MSC involve matrix-matching to compensate for matrix effects.
Therefore, although more efficient than the traditional SA method, the newly developed calibration
strategies still are more involved than EC. In each case, the sample is measured twice, i.e. once by
itself and once as a mixture containing added standard solution. Thus, similar to SA, these recently
described calibration strategies are more efficient at minimizing matrix effects, but present lower
sample throughputs when compared with EC. In the present work, we describe a method that
combines the simplicity of EC with the enhanced precision of employing multiple signal collection
conditions to carry out calibration. In multi-flow calibration (MFC), a single reference solution is
used, with no standard added to the samples. The analyte concentration is varied online by
modulating the amount of sample introduced into the plasma using the instrument’s nebulization
gas, i.e. the analyte concentration in the plasma varies as the nebulization flow rate (Q) changes.
Therefore, rather than preparing several calibration standards (as in EC), the same effect of
generating multiple signal levels to estimate analyte sensitivity and then determine the analyte
concentration in the sample is achieved with a single standard solution in MFC. The mathematical
basis of MFC is generally simple. Consider, for example, the parameters involved in signal
intensity (here represented as the measured output voltage, Eout) for an arbitrary emission line (eqn
(1)):12
𝐸𝐸𝑜𝑜𝑜𝑜𝑜𝑜 = 𝐶𝐶𝐶𝐶𝜖𝜖𝑎𝑎𝑄𝑄𝑒𝑒𝑓𝑓
𝑔𝑔𝑗𝑗𝑔𝑔0𝑒𝑒−𝐸𝐸𝑗𝑗0/𝑘𝑘𝐵𝐵𝑇𝑇 × 𝑉𝑉𝐸𝐸𝑗𝑗0𝐴𝐴𝑗𝑗0 × 𝑌𝑌𝑚𝑚 × 𝑇𝑇𝑜𝑜𝑑𝑑𝑅𝑅(𝜆𝜆)𝐺𝐺 (1)
101
where C, F, 𝜖𝜖𝑎𝑎, Q, ef, gj, g0, Ej0, kB, T, V, Aj0, Ym, Top, R(λ) and G represent analyte concentration,
solution flow rate, atomization efficiency, nebulization gas flow rate, gas expansion factor,
statistical weights of the excited state and the ground state, transition energy, Boltzmann constant,
plasma temperature, volume observed by the monochromator, rate of spontaneous emission,
monochromator collection efficiency, transmittance of the optics, detector responsivity, and gain
of the electronics. For a given instrument in a fixed condition and a given analyte, eqn (1) has
traditionally been simplified as:
𝐸𝐸𝑜𝑜𝑜𝑜𝑜𝑜 = 𝐾𝐾𝐾𝐾 (2)
with K representing a proportionality constant, which incorporates all parameters on the right hand
side of eqn (1) except for C. In the simplest calibration strategy, a single-point calibration,13 one
runs a standard solution for which Eout,std = KCstd (or Istd = KCstd to represent signal intensity rather
than voltage output). Then, the sample is run and Isam = KCsam. Because K is the same in both runs
(considering no matrix effects), the analyte concentration in the sample (Csam) may be determined
by:
𝐾𝐾𝑠𝑠𝑎𝑎𝑚𝑚 = 𝐶𝐶𝑠𝑠𝑠𝑠𝑠𝑠 𝐼𝐼𝑠𝑠𝑎𝑎𝑠𝑠𝐼𝐼𝑠𝑠𝑠𝑠𝑠𝑠
(3)
The primary issue with this simple strategy is the error associated with using a single
measurement for calibration. The analyte concentration in the sample is determined by
interpolation, and the greater the number of calibration points involved the lower the standard
deviation associated with the estimated concentration.4 Thus, working within the traditional
paradigm, several calibration standard solutions are prepared and run to minimize the error in
determining the unknown analyte concentration in the sample. This is done by re-arranging eqn
(3) into eqn (4) and plotting the analyte intensity collected from different standard solutions (which
102
obviously depend on the individual standard concentrations, Istd(Cstd)) versus the analyte
concentration in each standard (Cstd). The slope of that plot will equal Isam / Csam, and the unknown
analyte concentration in the sample can be found by recording Isam from the sample solution and
dividing that signal by the slope (eqn (5)).
𝐼𝐼𝑠𝑠𝑜𝑜𝑑𝑑 = 𝐾𝐾𝑠𝑠𝑜𝑜𝑑𝑑 𝐼𝐼𝑠𝑠𝑎𝑎𝑠𝑠𝐶𝐶𝑠𝑠𝑎𝑎𝑠𝑠
(4)
𝐾𝐾𝑠𝑠𝑎𝑎𝑚𝑚 = 𝐼𝐼𝑠𝑠𝑎𝑎𝑠𝑠𝑠𝑠𝑠𝑠𝑜𝑜𝑑𝑑𝑒𝑒
(5)
Alternatively, if one chooses to keep Cstd constant, re-arrange eqn (3) into eqn (6), and plot
Isam versus Istd as they change with Q, i.e. Isam(Q) vs. Istd(Q), the slope of that plot will be Csam / Cstd.
Therefore, Csam can be easily found by multiplying the concentration of analyte in the standard by
the slope, as represented in eqn (7).
𝐼𝐼𝑠𝑠𝑎𝑎𝑚𝑚 = 𝐼𝐼𝑠𝑠𝑜𝑜𝑑𝑑 𝐶𝐶𝑠𝑠𝑎𝑎𝑠𝑠𝐶𝐶𝑠𝑠𝑠𝑠𝑠𝑠
(6)
𝐾𝐾𝑠𝑠𝑎𝑎𝑚𝑚 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑒𝑒 ∙ 𝐾𝐾𝑠𝑠𝑜𝑜𝑑𝑑 (7)
In the present work, we evaluate the applicability of the MFC method in atomic
spectrometry by determining Cr, Cu, Fe and Mn in water and food samples by MIP OES. Certified
reference materials (CRMs) and addition and recovery experiments, as well as a comparison with
the traditional EC method, are used to validate the MFC strategy.
103
EXPERIMENTAL
Instrumentation
A MIP OES instrument (4200 MP-AES, Agilent Technologies, Santa Clara, CA, USA)
was used in all determinations. It was outfitted with a glass concentric pneumatic nebulizer
(Meinhard, Golden, CO, USA) and a glass double-pass cyclonic spray chamber (Agilent). A liquid
N2 Dewar (99.99% purity, Air Products, Allentown, PA, USA) was used to supply the MIP with
both plasma gas and nebulization gas.
The microwave applied power in this instrument is fixed at 1 kW. For all analytes, the
peristaltic pump speed was set to 15 rpm, the integration time was 3 s, and the plasma observation
position was set to 0 (which corresponds to the center of the plasma).14 Samples were analyzed to
determine Cr (at 425.433 nm), Cu (at 324.754 nm), Fe (at 371.993 nm) and Mn (at 403.076 nm).
In the present proof-of-concept study, the analytes were chosen based on their certified values in
the CRMs available and on the general MIP OES sensitivity. For MFC, each element was
measured at the following Q values: 0.4, 0.5, 0.6, 0.7, and 0.8 L min-1 N2. For EC comparisons,
default Q conditions recommended by the manufacturer were employed, i.e. 0.9 L min-1 for Cr and
Mn, and 0.65 L min-1 for Cu and Fe.
An Ethos UP microwave-assisted digestion system (Milestone, Sorisole, Italy) was used to
decompose the solid samples before analysis.
104
Reference materials, samples, and sample preparation
Distilled-deionized water (18 MΩ·cm, Milli-Q, Millipore, Bedford, MA, USA) and trace-
metal-grade nitric acid (Fisher, Pittsburgh, PA, USA) were used to prepare all working solutions.
All samples and standard solutions were prepared in 1% v/v HNO3. Trace-analysis-grade H2O2
(30% v/v, Sigma Aldrich, Atlanta, GA, USA) was used for sample digestion. Single-element stock
solutions containing 1000 mg L-1 Cr, Cu, Fe or Mn (High Purity Standards - HPS, Charleston, SC,
USA) were used to prepare the standard reference solutions and to carry out addition and recovery
experiments.
Three CRMs were used for method validation: Secondary Drinking Water Metals, River
Sediment A (HPS), and Tomato Leaves (National Institute of Standards and Technology - NIST,
Gaithersburg, MD, USA). Addition and recovery experiments were also carried out using Cheerios
(General Mills, Minneapolis, MN, USA), Oatmeal (The Quaker Oats Co., Chicago, IL, USA), and
two water samples from the Sea of Galilee and the Jordan River. The river water samples were
collected into new, clean 50-mL polypropylene centrifuge tubes and stored in 1% v/v HNO3 until
analysis. Secondary Drinking Water Metals was diluted 10-fold in1% v/v HNO3 before analysis,
and River Sediment A was diluted 5-fold or 100-fold, depending on the analyte, to adequate the
analyte concentrations to the calibration curve range. All the other samples were digested in
triplicate using 1 mL of concentrated HNO3 and 2 mL of H2O2. The heating program used for
microwave-assisted digestion included a 10-min step to reach 180 oC, a 15-min hold at 180 oC and
a 15-min cool down step.
105
RESULTS AND DISCUSSION
MFC strategy
One of the key challenges in MIP OES and other spectrochemical analysis methods is
associated with the optimization of instrument operating conditions to improve accuracy. Because
applied power, plasma gas flow rate, and other conditions are fixed in the commercially available
MIP OES instrument, the chief operating condition which affects the state of the plasma and can
be changed on a per-wavelength basis is the nebulization gas flow rate.15-17 In this context, different
studies have evaluated the effects of operating conditions on accuracy, and sought to identify their
optimal setup to improve the performance of MIP OES.14,18-20 Plasma robustness, according to
several definitions, may be generally improved by reducing Q and adjusting the plasma viewing
position. However, such generally recommended conditions not always result in accurate results
for certain MIP OES applications.14,18 Extremely low Q values may lead to insufficient or
suboptimal amounts of analyte reaching the plasma and to a potential overpopulation of interfering
species. Thus, rather than seeking to select a single optimal condition for each sample matrix, the
MFC method simultaneously employs multiple Q values, which, on average, minimize the
negative effects of a poorly chosen condition on the overall accuracy of the analysis. In principle,
MFC can also be performed by modifying the rotation speed of the peristaltic pump (i.e. by varying
F rather than Q in eqn (1)). However, the commercially available MIP OES control software allows
for no modification of pump speed within a single run. Thus, a method based on such strategy
would require the use of multiple worksheets and large volumes of sample, resulting in
significantly longer analysis times and the generation of large volumes of waste.
Table XV demonstrates the multi-flow advantage when compared with EC. In this case, MFC and
EC carried out at several different Q conditions were used to determine Cu in a sample of River
106
Sediment A. As it can be seen, the MFC result is generally more accurate than any of the individual
values obtained by EC, and superior to the average 91.7% recovery calculated from all Q
conditions used with the traditional method. Because it is more accurate than EC at both 0.60 and
0.70 L min-1, it may be expected that MFC would also provide a more accurate concentration of
Cu than that obtained with EC at the default 0.65 L min-1 condition.
107
Table XV. Comparison of MFC with EC using the same individual Q values while determining Cu in River Sediment A. Analyte percent recoveries (%) refer to the 1 mg L-1 certified concentration in the CRM. Results are shown as mean ± 1 standard deviation (n = 3). MFC calibration standard: 5.00 mg L-1.
Calibration method, Q (L min-1) Recovery R2
MFC, 0.4 - 0.8 96 ± 2 0.99780
EC, 0.4 95 ± 4 0.99954
EC, 0.5 92 ± 1 0.99993
EC, 0.6 88 ± 1 0.99991
EC, 0.7 89 ± 1 0.99972
EC, 0.8 92 ± 1 0.99989
EC, 0.9 94 ± 2 0.99987
108
Typical MFC plots are shown in Figure 14. Copper in River Sediment A (Figure 14A)
and Mn in Tomato Leaves (Figure 14B) were determined using 5.00 mg L-1 standards. From eqn
(7), the concentration of Cu calculated for this particular sample replicate was 0.98 mg L-1, which
corresponds to a recovery of 98% from the certified value of 1 mg L-1. For Mn, considering an
initial sample mass of 0.2024 g and a final solution volume of 20.0 mL, the analyte concentration
in this sample replicate is calculated as 249 mg kg-1, which corresponds to a 101% recovery from
the certified value of 246 ± 8 mg kg-1.
When standard solution and sample present different matrices, the effects of each matrix
on the plasma mechanisms responsible for analytical signal generation may significantly
compromise accuracy in EC applications. The purpose of a matrix-matching calibration method,
such as SA and MEC, is to minimize such effects by preparing standard and sample solutions in
the same physical-chemical environment, i.e. the same matrix.5,7 An alternative approach to
minimizing matrix effects involves the normalization of the plasma rather than the matrix.
Although neither as specific nor as effective as matrix-matching, one may minimize the effects of
different matrices on the plasma by deliberately exposing standard and sample solutions to a
variety of plasma conditions, which is achieved in MFC by employing multiple nebulization gas
flow rates. Such strategy may be compared to the calibration method known as extrapolation to
infinite dilution, with the advantage of requiring no complicated sample preparation procedures
and no additional sample introduction apparatus.21 In addition to its plasma normalizing effects,
MFC may also contribute to mitigating matrix interferences on the nebulization process itself, as
it employs a range of conditions rather than restricting the analysis to a single Q value that may be
more prone to interference.22
109
Figure 14. Multi-flow calibration plots for determining (A) Cu in River Sediment A, and (B) Mn in Tomato Leaves.A 5.00 mg L-1 calibration standard was used in both determinations. The calibration plots refer to a single sample replicate.
110
It is important to emphasize that MFC can never be as efficient as matrix-matching
methods, and may be considered an intermediary strategy between EC and SA, for example. It is
even more straightforward than EC, as it only requires one calibration standard. In addition, it
should perform as well as, and mostly better than, EC for simple and moderately complex matrix
samples such as the ones presented in Table XVI. However, it probably needs to be replaced with
a matrix-matching alternative to ensure accuracy when analyzing complex-matrix samples capable
of producing severe matrix effects. Similar to MEC, MICal and MSC (which use a multivariate
calibration strategy not directly associated to multiple standard concentrations), MFC is more
prone to systematic errors related to solution preparation. Because a single standard calibration
solution is used, any inaccuracy with that solution produces biased results. On the other hand, an
advantage of MFC over MEC, MICal and MSC is that a certified stock solution can be used
directly (i.e. without dilution) if available, whereas the other methods require the addition of a
standard solution to an aliquot of each sample.7-11
T
able
XV
I. A
ccur
acy
com
paris
on b
etw
een
MFC
and
EC
. Ana
lyte
con
cent
ratio
ns a
re re
porte
d as
mea
n ±
1 st
anda
rd d
evia
tion
(n =
3).
Ana
lyte
per
cent
reco
verie
s (%
) fro
m th
e cer
tifie
d va
lues
are s
how
n in
par
enth
esis
. MFC
calib
ratio
n st
anda
rd: 5
.00
mg
L-1 fo
r all
anal
ytes
ex
cept
for C
u in
Tom
ato
Leav
es w
hich
was
0.0
500
mg
L-1.
a Sec
onda
ry d
rinki
ng w
ater
met
als
(HPS
). C
ertif
ied
valu
es:
50,
100
and
50 m
g L-1
for
Cu,
Fe
and
Mn,
res
pect
ivel
y.
b Ri
ver S
edim
ent A
(HPS
). C
ertif
ied
valu
es: 3
00, 1
, 120
0 an
d 8
mg
L-1 fo
r Cr,
Cu,
Fe
and
Mn,
resp
ectiv
ely.
The
sta
ndar
d de
viat
ions
calc
ulat
ed fo
r thi
s sam
ple
are
base
d on
inst
rum
enta
l rep
licat
es.
c Tom
ato
Leav
es (
NIS
T). C
ertif
ied
valu
es (m
g kg
-1):
1.99
± 0
.06,
4.7
0 ±
0.14
, 368
± 7
and
246
± 8
for
Cr,
Cu,
Fe
and
Mn,
resp
ectiv
ely.
d NA
= N
ot a
vaila
ble.
Sam
ple
Cr
Cu
Fe
Mn
MFC
E
C
MFC
E
C
MFC
E
C
MFC
E
C
SDW
M a
NA
d N
A d
50 ±
1
(100
%)
52 ±
1
(104
%)
91 ±
2
(91%
) 92
± 5
(9
2%)
48 ±
1
(96%
) 49
± 1
(9
8%)
CR
M-R
S-A
b 29
1 ±
8 (9
7%)
309
± 11
(1
03%
) 1.
01 ±
0.0
5 (1
01%
) 1.
11 ±
0.1
0 (1
11%
) 12
12 ±
15
(101
%)
1368
± 1
31
(114
%)
8.4
± 0.
2 (1
05%
) 10
.7 ±
2.4
(1
34%
)
NIS
T 15
73a
c <L
OD
1.
67 ±
0.0
6 (8
4%)
5.02
± 0
.21
(107
%)
5.75
± 0
.31
(122
%)
400
± 9
(109
%)
339
± 7
(92%
) 26
2 ±
12
(106
%)
277
± 8
(113
%)
111
112
Limits of detection
The limits of detection (LODs) were based on eqn (7) and calculated as 3 times the standard
error of the MFC calibration curve slope (Sslope), times the concentration of the standard, i.e. LOD
= 3·Sslope·Cstd. A solution of 1% v/v HNO3 was treated as sample in LOD calculations, with 60
data points used for determining Sslope (12 samples measured at 5 different Q conditions). A 5 mg
L-1 standard solution was used in this experiment. The LODs calculated for Cr, Cu, Fe and Mn
were 20, 5, 7 and 2 µg L-1, respectively. Except for Cr, these values are comparable with those
obtained with EC (calculated as three times the standard deviation of a blank solution, n = 12,
divided by the calibration curve slope): 0.6, 8, 20 and 1 µg L-1, respectively.
To investigate the discrepancy in Cr LODs, a 1% v/v HNO3 solution was analyzed at Q
values of 0.4, 0.5, 0.6, 0.7 or 0.8 L min-1, and the respective BG spectrum at the 425.433, 324.754,
371.993 and 403.076 nm regions (i.e. Cr, Cu, Fe and Mn) were evaluated. Percent relative standard
deviations (RSDs) associated with BG signals collected at the different Q conditions were then
calculated for each wavelength. In this case, BG signals recorded at each Q condition were used
to calculate mean, standard deviation and RSD for each of the analytes’ wavelengths. The results
indicate that higher MFC LODs (when compared with EC) are related to higher BG signal RSDs.
The highest values were found for Cr and Mn, with 52 and 25%, while Cu and Fe presented 15
and 18%, respectively. For Cr, for example, there was a 68% decrease in BG signal intensity when
Q was changed from 0.4 to 0.8 L min-1. For Cu, the signal reduction was 30% for the same change
in Q. These results suggest that the BG signal at 425.433 nm is more sensitive to changes in Q
when compared to the other wavelength regions evaluated, which may have resulted in a relatively
high LOD for Cr when applying MFC.
113
The upper limit of the calibration curve, for which analyte concentrations can be accurately
determined using a single standard of 5 mg L-1, was not experimentally determined. However,
preliminary results suggest MFC may be successfully applied within an order of magnitude from
the standard concentration used. In addition, the introduction of samples with concentrations
higher than 100 mg L-1 resulted in the saturation of the instrument’s detector.
Accuracy
The method’s accuracy was evaluated by analyzing CRMs and by addition and recovery
experiments. The results, along with a comparison with values obtained with EC, are presented in
Table XVI and Table XVII. It is important to note that the different recoveries for Cu in River
Sediment A between Table 1 and Table XVI are due to different analysis days and the use of a
single-element (Table 1) or a multi-element (Table XVI) standard for calibration. In addition, the
results shown for River Sediment A (Table XVI), Sea of Galilee and Jordan River (Table XVII)
are based on instrumental replicates, rather than true replicates, due to the limited amount of these
samples available.
114
Tab
le X
VII
. Ana
lyte
per
cent
reco
verie
s (%
) fro
m sp
iked
con
cent
ratio
ns in
wat
er a
nd fo
od sa
mpl
es a
naly
zed
by M
IP O
ES u
sing
MFC
or
EC
. The
resu
lts ar
e pre
sent
ed as
mea
n ±
1 st
anda
rd d
evia
tion
(n =
3).
Ana
lyte
conc
entra
tion
adde
d to
the s
ampl
es =
5.0
0 m
g L-1
. MFC
ca
libra
tion
stan
dard
= 5
.00
mg
L-1.
a The
stan
dard
dev
iatio
ns c
alcu
late
d fo
r the
se sa
mpl
es a
re b
ased
on
inst
rum
enta
l rep
licat
es.
Sam
ple
Cr
Cu
Fe
Mn
MFC
E
C
MFC
E
C
MFC
E
C
MFC
E
C
Sea
of G
alile
e a
107
± 1
118
± 4
109
± 1
113
± 12
10
7 ±
2 10
9 ±
26
112
± 5
103
± 4
Jord
an R
iver
a 10
9 ±
2 12
0 ±
6 10
2 ±
2 11
4 ±
6 95
± 1
10
7 ±
12
103
± 2
106
± 8
Che
erio
s 99
± 3
11
0 ±
8 92
± 7
92
± 5
98
± 1
2 94
± 1
2 97
± 4
10
5 ±
6
Oat
mea
l 97
± 7
10
9 ±
6 93
± 8
91
± 6
10
1 ±
9 10
5 ±
6 94
± 9
10
0 ±
7
115
As expected considering the early discussion on plasma normalization, MFC results are
similar, and often more accurate than those from EC. MFC’s superior performance becomes
evident as a more complex-matrix sample such as Tomato Leaves is analyzed. From Table XVI,
no statistically significant difference between certified and determined values were found for most
analytes (Student’s t-test at the 95% confidence level), except for Fe in both Secondary Drinking
Water Metals and Tomato Leaves for MFC, and all analytes in Tomato Leaves for EC. Analyte
percent recoveries were in the 91-109% and 84-134% ranges for MFC and EC, respectively. In
addition and recovery experiments (Table XVII), analyte percent recoveries were in the 92-112%
and 91-120% ranges for MFC and EC, respectively. No statistically significant difference was
found between MFC and EC results (two-mean Student’s t-test at the 95% confidence level),
except for Cu in Tomato Leaves and Jordan River, and Cr in Sea of Galilee and Jordan River. It is
interesting to note that in all these four cases, EC’s analyte recoveries were higher than 110%,
which may be additional evidence of MFC’s relatively higher accuracy.
To further evaluate the effect of exposing sample and calibration standards to different
plasma conditions on accuracy, the MFC results presented in Table XVI may be compared with
values obtained from a single-point calibration13 using a 5.00 mg L-1 standard and determinations
at the default Q conditions. For Secondary Drinking Water Metals (HPS), recoveries for Cu, Fe
and Mn were 120, 121 and 104%, respectively (100, 91 and 96% for the same elements using
MFC, Table XVI). Similar results were found for River Sediment A (HPS) and Tomato Leaves
(NIST), with 100 and 101% recovery for Cr, 113 and 117% for Cu, 115 and 95% for Fe, and 128
and 108% for Mn, respectively. From Table XVI, the MFC results for these same CRMs were
97% and < LOD for Cr, 101 and 107% for Cu, 101 and 109% for Fe, and 105 and 106% for Mn,
respectively. In comparison with MFC, a single-point calibration using the same reference
116
standard solution generally produces overestimated values, which reinforces the hypothesis of the
plasma normalizing effect associated with the new calibration strategy.
As noted earlier, MFC performs better when analyte concentrations in the sample and the
calibration standard are within one order of magnitude of each other. Therefore, when applying
MFC in a routine analysis, if the analytical signals recorded for the sample are too low or too high
in comparison with those from the calibration standard (i.e. more than one order of magnitude
difference), sample dilution or a higher-concentration standard must be adopted. A similar
procedure is expected in EC applications, i.e. an estimate of analyte levels in the sample is required
before deciding on the range of concentrations covered by the calibration curve. Alternatively, if
the analytical signal is too high, sample dilution is required to fit the calibration curve’s
concentration range. In this context, a 0.0500 mg L-1 standard solution was used with MFC to
determine Cu in Tomato Leaves, given the relatively lower levels expected in this case (Table
XVI). To highlight the effects of standard concentration on MFC accuracy, the same Tomato
Leaves sample replicates used to produce the values shown in Table XVI were analyzed using a
1.00 mg L-1 Cu standard (ca. 20-fold higher than the expected Cu concentration in the sample).
The Cu value found in this case was 5.80 ± 0.30 mg kg-1, which corresponds to a 123% recovery
and is obviously less accurate than the 107% value obtained with a 0.0500 mg L-1 calibration
standard.
117
Precision
Precision for both MFC and EC may be assessed by calculating RSDs from the results in
Table XVI and Table XVII. RSD values are generally lower for MFC when compared with EC,
which may be associated to fewer sources of error when running a single calibration solution in
MFC. In Table XVI, RSDs are in the 1.2-5.0% and 1.9-9.6% ranges for MFC and EC, respectively.
In Table XVII, RSD values are calculated between 0.9 and 12.2% for MFC, and between 3.4 and
23.9% for EC.
Long-term stability
In EC applications, it is generally necessary to re-run the calibration curve after a certain
period of time to minimize the effects of signal drift on accuracy. Thus, a long-term stability
experiment was carried out to evaluate the applicability of the MFC method over time. A solution
containing 5 mg L-1 of each Cr, Cu, Fe and Mn was introduced into the plasma and measured every
10 min for a period of 90 min using MFC. The first measurement (minute 0) was used as the
calibration standard and the subsequent measurements were considered samples. As observed in
Figure 15, accurate results are expected for periods of up to 90 min without the need for
recalibration, except for Cr, which requires recalibration approximately every 30 min. These
results suggest MFC’s recommended procedure for minimizing signal drift is not significantly
different from that routinely adopted for EC. The signal behavior observed for Cr suggests higher
instability. As discussed earlier, Cr results may be related to changes in BG signal over time and a
relatively higher sensitivity of the BG signal at the 425.433 nm region to plasma changes.
118
Figure 15. Long-term stability of MFC. Analyte percent recoveries (%, n = 3) are based on a 5.00 mg L-1 solution measured every 10 min over a period of 90 min. Average values are the mean percent recoveries of all four analytes at each time point.
119
Sample throughput
Considering an analysis involving twenty samples, four analytes, five calibration points for
EC, and the analytical parameters used in the present work, the sample throughputs calculated for
EC and MFC are 27 and 13 samples h-1, respectively. Therefore, EC is significantly faster than
MFC, which is especially due to the different Q conditions required for the new method. However,
these estimates are based on time spent during instrumental measurements, with no consideration
of solution preparation. Thus, for an analysis involving a small number of samples, in which
preparing the calibration standard solutions has a more significant effect on the final sample
throughput, MFC’s speed may be comparable to EC. This is especially valid considering that the
detection step is fully automated and most of the analyst’s time is spent on solution preparation.
Analyte concentrations in the samples
For most samples evaluated, the original analyte concentrations (non-spiked) were lower
than the respective LODs. Detectable values were found only for Fe in Cheerios, with 255 ± 20
and 279 ± 24 mg kg-1 (n = 3) determined by MFC and EC, respectively. In this case, no statistically
significant difference was observed between the different calibration methods (two-mean
Student’s t-test at a 95% confidence level). Both values are also in agreement with the product
labelling, which indicates an approximate Fe mass equivalent to 45% of the recommended daily
value in a 28 g serving size. According to the product’s label and information from the U.S. Food
and Drug Administration,23 the approximate Fe concentration in this sample should be 280 ± 30
mg kg-1.23
120
CONCLUSIONS
Multi-flow calibration is a novel strategy for use in MIP OES and other spectrochemical
analysis methods that allow for variation of nebulization gas flow rates during the analysis. It is an
efficient method, as it employs a single calibration standard and provides accuracies and precisions
comparable to, and often better than, the traditional EC method. Although not as effective as SA
and other matrix-matching strategies, MFC may minimize matrix effects. It exposes samples and
standards to a variety of plasma conditions, which may have a normalizing effect capable of
improving accuracies, in comparison with EC, for analysis involving samples with simple to
moderately complex matrices.
As demonstrated in this proof-of-concept work, the MFC method is an effective alternative
to EC for applications involving Cr, Cu, Fe and Mn, but additional studies are required to evaluate
other analytes and sample matrices. One of its main limitations is the potential for systematic errors
associated with solution preparation, especially considering a single calibration standard is used to
determine analyte concentrations in the samples. Although less time is spent with solution
preparation, MFC presents a lower sample throughput than EC when analyzing a large number of
samples. In addition, signal fluctuations due to the required changes in Q may result in higher
LODs when compared with EC. On the other hand, no additional experiments are required to
optimize Q conditions, and no negative effects on accuracy due to a poor choice of Q are expected
with MFC.
121
CONFLICTS OF INTEREST
There are no conflicts of interest to declare.
ACKNOWLEDGEMENTS
The authors would like to thank High Purity Standards, and the Department of Chemistry
and the Graduate School of Arts and Sciences at Wake Forest University for their support to this
work.
122
REFERENCES
1. L. Cuadros-Rodrıguez, L. Gámiz-Gracia, E. Almansa-López and J. Laso-Sánchez, TrAC Trends
Anal. Chem., 2001, 20, 195-206.
2. L. Cuadros-Rodrıguez, L. Gámiz-Gracia, E. M. Almansa-López and J. M. Bosque-Sendra, TrAC
Trends Anal. Chem., 2001, 20, 620-636.
3. J. A. Carter, A. I. Barros, J. A. Nóbrega and G. L. Donati, Front. Chem., 2018, 6, Art. 504.
DOI:10.3389/fchem.2018.00504.
4. J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry, Ellis Horwood PTR Prentice
Hall, 3rd ed., New York, 1993.
5. W. R. Kelly, K. W. Pratt, W. F. Guthrie and K. R. Martin, Anal. Bioanal. Chem., 2011, 400,
1805-1812.
6. D. A. Goncalves, B. T. Jones and G. L. Donati, Microchem. J., 2016, 124, 155-158.
7. A. Virgilio, D. A. Gonçalves, T. McSweeney, J. A. Gomes Neto, J. A. Nóbrega and G. L. Donati,
Anal. Chim. Acta, 2017, 982, 31-36.
8. R. C. Machado, A. B. S. Silva, G. L. Donati and A. R. A. Nogueira, J. Anal. At. Spectrom.,
2018, 33, 1168-1172.
9. D. V. Babos, A. Virgilio, V. C. Costa, G. L. Donati and E. R. Pereira-Filho, J. Anal. At.
Spectrom., 2018, 33, 1753-1762.
10. A. Virgilio, J. A. Nóbrega and G. L. Donati, Anal. Bioanal. Chem., 2018, 410, 1157-1162.
11. C. B. Williams and G. L. Donati, J. Anal. At. Spectrom., 2018, 33, 762-767.
123
12. J. D. Ingle and S. R. Crouch, Spectrochemical Analysis, Prentice Hall, Englewood Cliffs, 1988.
13. G. J. Kemp, Clin. Chem., 1984, 30, 1163-1167.
14. D. A. Goncalves, T. McSweeney and G. L. Donati, J. Anal. At. Spectrom., 2016, 31, 1097-
1104.
15. N. Ozbek, M. Koca and S. Akman, Food Anal. Methods, 2016, 9, 2246-2250.
16. N. Ozbek, H. Tinas and A. E. Atespare, Microchem. J., 2019, 144, 474-478.
17. S. M. Azcarate, L. P. Langhoff, J. M. Camiña and M. Savio, Talanta, 2019, 195, 573-579.
18. C. B. Williams, B. T. Jones and G. L. Donati, J. Anal. At. Spectrom., 2018, 33, 1224-1232.
19. N. Chalyavi, P. S. Doidge, R. J. S. Morrison and G. B. Partridge, J. Anal. At. Spectrom., 2017,
32, 1988-2002.
20. K. L. Lowery, T. McSweeney, S. P. Adhikari, A. Lachgar and G. L. Donati, Microchem. J.,
2016, 129, 58-62.
21. M. Thompson and M. Ramsey, J. Anal. At. Spectrom., 1990, 5, 701-704.
22. J. L. Todoli, L. Gras, V. Hernandis and J. Mora, J. Anal. At. Spectrom., 2002, 17, 142-169.
23. Center for Food Safety and Applied Nutrition, A Food Labeling Guide: Guidance for Industry,
Food and Drug Administration, College Park, p. 127.
https://www.fda.gov/downloads/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInfor
mation/UCM265446.pdf, (accessed March 12, 2019).
124
CHAPTER VI
CONCLUSIONS
Microwave-induced plasma optical emission spectrometry (MIP OES) is a rapidly-
growing technique in the field of atomic spectrometry. Its chief advantages are its simple,
low-cost operation for routine trace element analysis, and its ability to be used in remote
locations by virtue of its ability to run on air. With intensive method development and
further understanding of the nature of the plasma, the method can be applied to increasingly
complex sample types and matrices, expand access to elemental analysis in remote areas,
and better compete with established techniques such as ICP OES.
When combined with other simple, low-cost sample preparation techniques, the
utility of MIP OES for routine analysis of complex samples can be greatly expanded. MIP
OES coupled with a simple, dilute-and-shoot procedure with soft drink samples shows the
ability of the technique to perform routine elemental analyses in aqueous samples with high
levels of organic concomitants, further enabling the field to expand and potentially
increasing access to safety and quality control testing of food and other consumer products.
Similarly, MIP OES was combined with a prototype rapid dry-ashing unit to facilitate
sample preparation. This expands the utility of the technique to cover even more complex
sample types and further improves access to low-cost elemental analyses.
The development of plasma diagnostic tools is an important factor in both
increasing the understanding of the fundamental properties of the MIP and in expansion of
the analytical utility of the plasma. The development of the N2+/OH ratio as a plasma
diagnostic tool is one of the first strategies developed specifically for MIP OES, as opposed
to being transferred from another technique. This development not only opens up a path
125
for better understanding the MIP, but also enables real-time analysis and signal correction.
It offers a short-cut to extensive trial-and-error method development by characterizing the
change in plasma conditions caused by a specific sample and allowing the analyst to
modulate operating conditions or sample preparation steps accordingly. While further
development of their use as signal-correction species is still required, these molecular
species present potential for use in a successful approach for further improving accuracy
in MIP OES determinations.
Novel calibration strategies are another potentially efficient means of improving
the accuracy of MIP OES. In this context, multi-flow calibration is well-adapted to the
limitations of MIP OES and may contribute to expanding its analytical utility. Although
not as effective as standard additions and other matrix-matching methods, MFC has the
ability to compensate for matrix effects, which improves on the shortcomings of combining
MIP OES with traditional external standard calibration. Because the properties of the
plasma change significantly with the flow rate of the solution entering the MIP,
nebulization gas flow rate can be used in the analyst’s favor. In particular, the performance
of the combined MFC-MIP OES strategy at low concentrations (i.e. ppb-level) is an
important avenue for further development.
The developments described in this dissertation represent a significant step forward
in the expansion of MIP OES as a mainstream technique in atomic spectrometry. The future
of MIP OES as an important tool for routine elemental analysis is promising. While it still
requires improvements relative to other, more established techniques, the gap is closing
and the niche for MIP OES in the atomic spectrometry market is becoming wider and more
secure.
126
APPENDIX A
SUPPLEMENTARY INFORMATION FOR CHAPTER III
NATURALLY OCCURRING MOLECULAR SPECIES USED FOR PLASMA
DIAGNOSTICS AND SIGNAL CORRECTION IN MICROWAVE-INDUCED
PLASMA OPTICAL EMISSION SPECTROMETRY
Charles B. Williams, Bradley T. Jones and George L. Donati
This appendix presents material published online as Electronic Supplementary
Information to accompany Chapter III, by the Journal of Analytical Atomic
Spectrometry, 2018, 33, 1224-1232. All of the presented research was conduced by
Charles B. Williams. The supplementary information section was prepared by Charles B.
Williams and George L. Donati.
127
EMISSION SPECTRA FOR CN, N2, N2+ AND OH
Figs. S1 - S4 show emission spectra recorded in wavelength regions
corresponding to CN, N2, N2+ and OH, respectively. The spectra were collected with the
microwave-induced plasma (MIP) either on or off to demonstrate the origin of these
molecular species in the N2 plasma. The CN emission band corresponds to the B(2∑+) -
X(2∑+) electronic transition, with a band peak at 387.147 nm. The N2, N2+, and OH
emission bands correspond to the C(3∏u) - B(3∏g), B(2∑u+) - X(2∑g
+), and A(2∑+) - X(2∏i
+) electronic transitions, with band peaks at 337.097, 391.439, and 308.970 nm,
respectively.1
Figs. S5 - S8 show the spectra collected for these same molecular species when the
instrument’s spray chamber was removed (i.e., no aqueous solution was being introduced
into the plasma), as well as when distilled-deionized water or 1 % v v-1 HNO3 were
introduced into the MIP. Note that N2+ is originally part of the MIP (Fig. S7), while OH is
mainly produced as an aqueous solution is introduced into the plasma (Fig. S8).
128
Figure S 1. Molecular emission spectra for CN recorded with the plasma off, or as a 1 % v
v-1 HNO3 solution was introduced into the MIP at a nebulization gas flow rate of
0.7 L min-1.
Figure S 2. Molecular emission spectra for N2 recorded with the plasma off, or as a 1 % v
v-1 HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7
L min-1.
129
Figure S 3. Molecular emission spectra for N2+ recorded with the plasma off, or as a 1 %
v v-1 HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7 L
min-1.
Figure S 4. Molecular emission spectra for OH recorded with the plasma off, or as a 1 % v
v-1 HNO3 solution was introduced into the MIP at a nebulization gas flow rate of 0.7
L min-1.
130
Figure S 5. Molecular emission spectra for CN recorded as 1 % v v-1 HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP.
Figure S 6. Molecular emission spectra for N2 recorded as 1 % v v-1 HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP.
131
Figure S 7. Molecular emission spectra for N2+ recorded as no aqueous solution (no spray
chamber), distilled-deionized water, or 1 % v v-1 HNO3 (NGFR = 0.7 L min-1) was
introduced into the MIP.
Figure S 8. Molecular emission spectra for OH recorded as 1 % v v-1 HNO3, distilled-
deionized water (NGFR = 0.7 L min-1), or no aqueous solution (no spray chamber) was
introduced into the MIP.
132
EVALUATION OF INDIVIDUAL EMISSION SIGNALS AND COMBINATIONS
OF CN, N2, N2+ AND OH INTENSITIES AS PLASMA DIAGNOSTIC TOOLS
To evaluate the efficiency of some of the MIP’s naturally occurring species as
diagnostic tools, individual emission signals for CN, N2, N2+ and OH, as well as their
combination in the form of signal ratios (i.e. N2+/OH, N2
+/N2, N2+/CN, N2/OH, N2/CN
and CN/OH) were studied. The performance criterion was the sensitivity to plasma
changes caused by the introduction of high concentrations of Na into the MIP at different
nebulization gas flow rates (NGFR). The molecular species were also compared with Mg
II and Mg I lines, and with the Mg II/Mg I signal ratio. As shown in Figs. S9 - S12, the
N2+/OH signal ratio was the most sensitive to plasma change across all conditions
evaluated. For example, when the Na concentration increased from 200 to 500 mg L-1, at
a NGFR of 0.6 L min-1, the N2+/OH signal ratio dropped more than 43 %, from 1.173 to
0.666 (absolute percent change of 43.2 %). For comparison at the same conditions, the
Mg II/Mg I, N2+/N2, N2
+/CN, N2/OH, N2/CN and CN/OH ratios changed 22.3, 34.1, 21.9,
13.9, 18.6 and 27.4 %, respectively. If a more drastic change takes place, as for example
with the Na concentration increasing from 0 to 500 mg L-1 (not shown in Figs. S9 - S12),
the N2+/OH signal ratio will change 65.2 %, as opposed to 33.2, 51.6, 32.1, 28.1, 40.2 and
48.7 % for Mg II/Mg I, N2+/N2, N2
+/CN, N2/OH, N2/CN and CN/OH, respectively.
Based on these results, and considering Na as a model for EIEs, which are
responsible for some of the most severe matrix effects observed in MIP OES,2 the
N2+/OH signal ratio was further studied as a diagnostic tool in the present work.
133
Figure S 9. Absolute emission signal percent change as Na concentrations in the 0 - 1000
mg L-1 range were introduced into the MIP at a NGFR of 0.6 L min-1. Species 1 - 6
correspond to Mg II (280.271 nm), Mg I (285.213 nm), CN (387.147 nm), N2 (337.097
nm), N2+ (391.439 nm) and OH (308.970 nm), respectively.
Figure S 10. Absolute emission signal percent change as Na concentrations in the 0 - 1000
mg L-1 range were introduced into the MIP at a NGFR of 1.0 L min-1. Species 1 - 6
correspond to Mg II (280.271 nm), Mg I (285.213 nm), CN (387.147 nm), N2 (337.097
nm), N2+ (391.439 nm) and OH (308.970 nm), respectively.
134
Figure S 11. Absolute signal ratio percent change as Na concentrations in the 0 - 1000 mg
L-1 range were introduced into the MIP at a NGFR of 0.6 L min-1. Here, 1 - 7 correspond
to Mg II/Mg I, N2+/OH, N2
+/N2, N2+/CN, N2/OH, N2/CN, and CN/OH, respectively.
Figure S 12. Absolute signal ratio percent change as Na concentrations in the 0 - 1000 mg
L-1 range were introduced into the MIP at a NGFR of 1.0 L min-1. Here, 1 - 7 correspond
to Mg II/Mg I, N2+/OH, N2
+/N2, N2+/CN, N2/OH, N2/CN, and CN/OH, respectively.
135
REFERENCES
1. Z. Zhang and K. Wagatsuma, Spectrochim. Acta Part B, 2002, 57, 1247-
1257.
2. K.P. Huber, G. Herzberg, Constants of Diatomic Molecules, in: P.J.
Linstrom, W.G. Mallard (Eds.), NIST Chemistry WebBook, NIST Standard Reference
Database, Num. 69, National Institute of Standards and Technology, Gaithersburg, MD,
USA, 2016 (Available at http://webbook.nist.gov (retrieved May 4, 2018)).
136
SCHOLASTICA VITA
CHARLES BRYSON WILLIAMS, III
BORN: December 17, 1992, Winston-Salem, NC
UNDERGRADUATE STUDY: Clemson University
Clemson, South Carolina
B.S. Chemistry, 2015
GRADUATE STUDY: Wake Forest University
Winston-Salem, North Carolina
PhD. 2019
SCHOLASTIC AND PROFESSIONAL EXPERIENCE:
Graduate Teaching Assistant
Wake Forest University, 2015-2017
Graduate Research Assistant
Wake Forest University, 2017-2019
Organist-Choir Director
Ascension Episcopal Church, 2013-2015
Organist
St. Anne’s Episcopal Church, 2016-2019
PROFESSIONAL SOCIETIES:
Society for Applied Spectroscopy, 2018-Present
137
HONORS AND AWARDS:
Merck Index Award, 2015
Glaxo-Smith-Kline Assistantship, 2015-2016
Alumni Student Travel Award, 2017, 2018
American Institute of Chemists Graduate Student Award, 2019
PUBLICATIONS:
G.L. Donati, R.S. Amais, C.B. Williams, Recent advances in inductively coupled plasma optical emission spectrometry, Journal of Analytical Atomic Spectrometry. 32 (2017) 1283–1296. doi:10.1039/C7JA00103G. C.B. Williams, T.G. Wittmann, T. McSweeney, P. Elliott, B.T. Jones, G.L. Donati, Dry ashing and microwave-induced plasma optical emission spectrometry as a fast and cost-effective strategy for trace element analysis, Microchemical Journal. 132 (2017) 15–19. doi:10.1016/j.microc.2016.12.017. A.G. Althoff, C.B. Williams, T. McSweeney, D.A. Gonçalves, G.L. Donati, Microwave-Induced Plasma Optical Emission Spectrometry (MIP OES) and Standard Dilution Analysis to Determine Trace Elements in Pharmaceutical Samples, Applied Spectroscopy. 71 (2017) 2692–2698. doi:10.1177/0003702817721750. C.B. Williams, G.L. Donati, Multispecies calibration: a novel application for inductively coupled plasma tandem mass spectrometry, J. Anal. At. Spectrom. 33 (2018) 762–767. doi:10.1039/C8JA00034D. C.B. Williams, B.T. Jones, G.L. Donati, Naturally occurring molecular species used for plasma diagnostics and signal correction in microwave-induced plasma optical emission spectrometry, Journal of Analytical Atomic Spectrometry. 33 (2018) 1224–1232. doi:10.1039/C8JA00086G. H. Li, Q. Li, P. Wen, T.B. Williams, S. Adhikari, C. Dun, C. Lu, D. Itanze, L. Jiang, D.L. Carroll, G.L. Donati, P.M. Lundin, Y. Qiu, S.M. Geyer, Colloidal Cobalt Phosphide Nanocrystals as Trifunctional Electrocatalysts for Overall Water Splitting Powered by a Zinc-Air Battery, Advanced Materials. 30 (2018) 1705796. doi:10.1002/adma.201705796. D.R. Onken, S. Gridin, R.T. Williams, C.B. Williams, G.L. Donati, V. Gayshan, S. Vasyukov, A. Gektin, Investigating the origins of double photopeaks in CsI:Tl samples through activator mapping, Nuclear Instruments and Methods in Physics
138
Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment. 893 (2018) 151–156. doi:10.1016/j.nima.2018.03.028. C.B. Williams, T. McSweeney, B.T. Jones, G.L. Donati, Determination of Ca, K and Na in Soft Drinks Using MP-AES, (2018). C.B. Williams, B.T. Jones, G.L. Donati, Multi-flow calibration applied to microwave-induced plasma optical emission spectrometry, Journal of Analytical Atomic Spectrometry. (2019). doi:10.1039/C9JA00091G.
CONFERENCE PRESENTATIONS
“Use of N2+ Emission Intensity To Estimate Plasma Robustness in
Microwave-Induced Plasma Optical Emission Spectrometry (MIP OES),” Oral
Presentation, Winter Conference on Plasma Spectrochemistry, Amelia Island,
Florida, January 13, 2018
“Multi-Flow Calibration as a Novel Strategy in MIP OES,” Poster
Presentation, FACSS SCIX, Atlanta, GA, October 24, 2018