Research Article Vol. 57, No.

23 / 10 August 2018 / Applied Optics 6729

Detection of micro-toxic elements in commercial

coffee brands using optimized dual-pulsed
laser-induced spectral analysis spectrometry
AHMED ASAAD I. KHALIL1,* AND OSAMA A. LABIB2
1
Department of Laser Sciences and Interactions, National Institute of Laser Enhanced Sciences (NILES), Cairo University, Giza 12613, Egypt
2
Department of Environmental Health, College of Public Health, Imam Abdulrahman Bin Faisal University, Dammam, Saudi Arabia
*Corresponding author: ahmedasaad68@yahoo.com

Received 27 March 2018; revised 18 June 2018; accepted 29 June 2018; posted 9 July 2018 (Doc. ID 326830); published 6 August 2018

A fast detection system based on dual-pulsed laser-induced breakdown spectroscopy (DP-LIBS) was successfully
developed and optimized for the estimation of micro-toxic element contents in commercial coffee brands to mon-
itor pollution and ensure food safety. A series of 13 various coffee brands were quantitatively analyzed in this
study utilizing DP-LIBS and a standard analytical technique like inductively coupled plasma/optical emission
spectrometry (ICP-OES). The micro-toxic elements, such as aluminum (Al), lead (Pb), zinc (Zn), and chromium
(Cr), that exist in the coffee brands were exactly identified. We prepared standard matrices in a known concen-
tration in the coffee sample to draw the standard calibration curves for each element, as well as by utilizing a tactic
based on the intense line emission of the element of interest as a quantitative analysis. The analytical routines were
approved under the expectations that the plasma created by the dual-pulsed lasers was in local thermodynamic
equilibrium (LTE) and was optically thin. We investigated the influence of different parametric dependence stud-
ies to enhance our DP-LIBS detection sensitivity. Furthermore, the precision of our DP-LIBS data for determin-
ing the concentration of micro-toxic elements present in coffee samples was validated via the ICP-OES technique.
The results achieved by the DP-LIBS technique were in full agreement with the ICP-OES results. In addition, the
estimated limit of detection of our DP-LIB spectrometer for Al, Pb, Zn, and Cr were 105.13
07, 90.17
12,
83.58
15, and 68.78
09 μg L−1 , respectively. The suggested protocols demonstrated the excellent benefits of
the DP-LIBS for the detection of micro-toxic elements existing in coffee and for checking the purity and quality of
food products. © 2018 Optical Society of America
OCIS codes: (300.6365) Spectroscopy, laser induced breakdown; (140.3440) Laser-induced breakdown.

https://doi.org/10.1364/AO.57.006729

1. INTRODUCTION sensitive technique such as dual-pulsed laser-induced break-

Coffee is one of the greatest main agricultural harvests in the
world. In terms of international industry, coffee is second in
value only to petroleum, placing into motion around US$35
billion per year [1]. Currently, the coffee chaff (spermoderm)
is used as an absorption material and as sound insulation for the
building industry [2]. The determination of micro-toxic sub-
stances in coffee requires a great deal of attention, owing to its
large intake by millions of people around the world. The micro-
toxic elements in coffee are dangerous bio-accumulate metals,
which accumulate in the fatty tissues of human bodies and in
the food chain [3]. The micro-toxic elements in coffee become
harmful to human health and well-being when their concen-
tration surpasses a definite level as a function of the sensitivity
of the consuming organism [4,5]. Strict public environmental
and protection legislation concerns resulted in the search for a
down spectroscopy (DP-LIBS) to identify these micro-toxic
elements. There are many types of coffees, such as Arabic,
Robusta (canephora), and Luwak (civet), which are widely con-
sumed beverages all over the world [6,7]. These types of coffees
depend on the complex composition and quality of the beans as
well as testing of the coffee. Therefore, it is important to evalu-
ate and characterize the contents of the coffee by developing
a quality control system using a high-sensitivity analytical
DP-LIBS spectrometer and performing cross-validation by
inductively coupled plasma/optical emission spectrometry
(ICP-OES) as a standard spectroscopic technique.
The DP-LIBS technique makes it possible to perform rapid
quantitative and qualitative analysis of trace, minor, and
major elements in coffee samples [8–10]. In addition, DP-LIBS
analysis in coffee offers several advantages in comparison with

1559-128X/18/236729-13 Journal © 2018 Optical Society of America

 6730 Vol. 57, No. 23 / 10 August 2018 / Applied Optics
conventional methods of analysis like atomic absorption
spectroscopy (AAS) and X-ray fluorescence (XRF) spectros-
copy. In this study, the micro-toxic elements were aluminum
(Al), lead (Pb), zinc (Zn), and chromium (Cr), while the
macro-nutrient elements were calcium (Ca), magnesium (Mg),
potassium (K), sodium (Na), and manganese (Mn). Calcium is
an important element for building strong bones and teeth in
human beings. It is essential for the functioning of the muscles
and heart and to protect the body against diabetes, high
blood pressure, and cancers. It acts as a protector against heart
disease, promotes healthy growth, and is a natural antioxidant
element [11].
Aluminum has been linked with neurological disorders [12],
and the maximum dose of 1 mg Al/kg of bodyweight/day is
considered a tolerable daily intake [13]. Lead is one of the most
common micro-toxic elements; it causes harmful health effects
in young children and adults, including permanent adverse
health effects such as high blood pressure, kidney failure,
and weakness of the brain and nervous system [14]. Zinc tox-
icity can occur in both chronic and acute forms. Higher zinc
intake may cause vomiting, nausea, itching, stinging, abdomi-
nal cramps, kidney damage, headaches, and diarrhea [15].
Chromium is a micro-toxic metallic element, an essential part
of metabolic processes, and it acts as a human mutagen and
carcinogen. It causes havoc in the human body. Therefore,
it is vitally important to define the concentration of micro-toxic
heavy elements in coffee brands for the safety of human beings
who drink coffee.
In this study, we engaged an improved DP-LIBS spectrom-
eter for the assessment and detection of micro-toxic elements in
13 brands of coffee in Egyptian food stores. In addition, we
thereby estimated the potential health hazard associated with
coffee consumption in view of the amount the body can tol-
erate and calculated the average daily intake of these elements.
The novelty of this paper is to emphasize that the DP-LIBS is a
powerful technique capable of detecting micro-toxic trace ele-
ments in different coffee brands. The results of both techniques
(DP-LIBS and ICP-OES) were compared to identify the qual-
ity differentiation of 13 brands of coffee with different compo-
sitional contents in terms of their origin, aroma, and taste. Our
focus is on detecting the concentrations of micro-toxic heavy-
weight elements such as Al, Pb, Zn, and Cr, which are detri-
mental to human health, even at low absorption intake, using
the DP-LIBS technique. The plasma emissions contain all the
analytical information about the heavy metal compositions of
the coffee brands utilizing the spectral strengths and the finger-
print wavelengths [16]. In addition, the calibration curves for
some heavy metals relative to Al, Pb, Zn, and Cr were done,
and the slope of each curve was estimated as the sensitivity of
the DP-LIBS technique. We optimized the experimental con-
ditions for the precise evaluation of Al, Pb, Zn, and Cr, under
verification of the assumption of local thermodynamic equilib-
rium (LTE) and the creation of optically thin plasma through
the typical plasma features like electron number density (N e )
and electron temperature (T e ). The quantification of the Al,
Pb, Zn, and Cr outcomes demonstrate a fine harmony and
strong consistency between the DP-LIBS and the ICP-OES
results.
Research Article
2. ANALYTICAL PROCEDURE FOR COFFEE
BRANDS
A series of 13 various coffee samples were carefully collected
from different Egyptian markets in different cities such as
Cairo, Helwan, Bhera, Beni-Suef, and Kafr El Sheikh. The
samples were obtained from different batches. The collected
coffee samples represent the most commonly consumed coffees
in Egypt. For ethical reasons, the brand names of the product
are kept anonymous, and instead, the samples are labeled with
numbers. We analyzed three samples from the same batch of
each coffee branch (i.e., three samples per coffee branch) with
different concentrations according to the degree of the purity of
the coffee using both techniques (ICP-OES and DP-LIBS).
The current study explored the level and source of micro-toxic
elements from dissimilar types of coffee in Egypt. Coffee sam-
ples were examined for heavy element content via the ICP-OES
technique. The concentration detected with our setup for the
heavy elements was within the standards set by the Egyptian
commission responsible for environmental standards in Egypt.
The types of coffee brands used had different shapes, compo-
sitions, and colors. The physical properties of the coffee samples
offered the determination of the health effects of these param-
eters on the human body [17,18]. In the ICP-OES technique,
we prepared the sample as a standard sample by following steps:
(1) Filtration—all solutions or digested samples introduced
to the instrument were carefully filtered to avoid a partially
or completely clogged nebulizer. This is most efficiently
performed through Acrodisc filters that fit over a syringe.
(2) Matrix matching—such as nitric and hydrochloric acid be-
side perchloric acid; also, the calibration standards and samples
were matrix-matched in terms of composition, total dissolved
solids (TDS), and acid concentration (in %) of the solution.
For analysis of solid materials, we used a standard solution
to measure the heavy metal concentration relative to the
standards. The samples have been compositionally matrix-
matched. Acid concentration is used during a particular
run, such that it is the same for all standards and samples in
that run. Typical concentrations of the coffee samples are
5% or 10% in HNO3 . For interstitial water analysis, the cal-
ibration standards have been prepared with the same back-
ground salt concentration. (3) Cleaning—we kept all the
used sample bottles and ICP glassware for the analysis of differ-
ent matrices separate. Because Li and B are of particular interest
to interstitial water chemists, all sample bottles, auto sampler
vials, tubes, Teflon tubing, peristaltic pump tubing, and torch
assembly glassware were used during the analysis of sediments.
Tremendous amounts of Li and B in the LiBO2 flux were kept
separate from those items used in the preparation and analysis
of interstitial waters. It is important that bottles and glassware
that have been used in the preparation and analysis of LiBO2
flux solutions are cleaned separately from the interstitial water
apparatus. The dilution of the interstitial water sample was 10×.
Depending on the analysis speed and the number of elements
analyzed, we generated enough analyte with 0.5 ml of intersti-
tial water. Approximately 1–1.5 g sample
5 ml digested re-
agent and 50 ml de-ionized water solution. Maximum dilution
allows for replicate analysis. (4) An analytical blank was pre-
pared identically by analyzing pure de-ionized water acidified
 Research Article Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6731

to matrix match the samples. Then, we measured the standards

and samples.
The coffee samples were first digested in 5 ml of digesting
reagent (2 hydrochloric acid: 1 nitric acid), followed by the ad-
dition of 5 ml of nitric acid that was about 99.99% pure (Fisher
Scientific, USA), and boiled with stirring for 10 min. We
cleaned the residue using cotton wool, ventilated at room tem-
perature and diluted to 50 ml with purified water. Thus,
the ICP-OES technique was used to measure the different types
of heavy metals according to their toxicity. For the analysis pro-
cedure, the polyethylene and glassware containers were
splashed with tap water, then drenched overnight in a 6N
HNO3 solution and washed a few times with ultra-pure water
to exclude the absorbing results from the cleaner. Precisely
balanced (1 g) vegetable samples were relocated into a silica
crucible and reserved in a muffle furnace for baking at 450°C.
The last residue was softened in 0.1 M HNO3 solution and
completed up to 50 ml. Standard solutions were arranged fol-
lowing the Shimazdu Perkin Elmer Clean Atomic Spectroscopy
Standards procedures (Perkin Elmer Corporation, NIST trace-
able CRM, USA). The prepared standard solutions were
arranged by attenuating the normal solution with 0.1 M nitric
acid for proving the linearity. One gram (1 g) of coffee was
digested using 12 cm3 of a mixture (5 ml v/v) of concentrated Fig. 1. (a) Schematic diagram of orthogonal double-pulse LIBS
HCl and HNO3 acids. Grade reagents were utilized for the setup in re-heating mode, (b) arrangements of laser focuses, collection
preparation of the standard solutions of those elements, con- optics, and a coffee sample surface.
suming their nitrate salts (Al, Pb, Zn, Cu, and Cr). The
ICP-OES equipped with direct intake via a nebulizer and a
cyclonic chamber was used for the determination of the pres- (IR) Nd:YAG laser running at 1064 nm was positioned parallel
ence of Al, Pb, Zn, Cu, and Cr. The element concentrations in to the coffee pellet surface and focused onto the intersection of
the coffee brands have been calculated from regular curves by the first laser beam; it elicited a plasma spark and reheating
extrapolation. The coffee powders were mixed well for homo- with an enhanced emission in atmospheric pressure as shown
geneity using a blending system (model Moulinex, France) and in Fig. 1(a). The irradiated sample area became vaporized and
were utilized in a digestion Gallenkamp Kjeldahl block (made the ablation started due to shockwave pressure and high tem-
in England) device. Finally, the powder quantities were injected perature. Thus, the plasma was generated as well as the basic
into the ICP center with argon filling gas. source of optical radiations. The longer-wavelength (1064 nm)
For DP-LIBS measurements, we mixed the coffee samples laser had a higher photon energy and concluded the mecha-
with 2 g of a potassium bromide (KBr) matrix and then pal- nisms of inverse bremsstrahlung absorption, thermal excitation,
letized them in sizes of 2 cm (diameter) and 2 mm (thickness) reduction of the fraction of matrices, and interaction with rar-
in a cylindrical dye. The pressing process was achieved under a efied ambient conditions [8]. Thus, further motivated species,
hydraulic press (Carver, Inc. Wabash, Indiana) with an applied ions, and free electrons were formed, which cooperated with
pressure of about 8 × 106 Pa for 30 min. We have been assured the arriving DP laser photons and revealed further heating
that no pollution comes from the press. The coffee pellets were and ionization, which increased the absorption of further laser
refined smoothly and delicately to provide symmetrical and photons. The incoming energy absorbed by the plasma was
uniform surfaces in parallel. Figure 1(a) depicts a schematic ar- recycled to raise the core plasma energy and accordingly raise
rangement of the DP-LIB system, where a pair of Q-switched, the ion velocity. In addition, it is possible to provide the spectral
pulsed Nd:YAG lasers were fitted with an orthogonal re-heating signatures of chemical composition in different coffee samples.
mode. The first pulsed Nd:YAG laser is a Big Sky laser, and the The laser pulse energies were measured utilizing a power
second pulsed Nd:YAG laser is a INDI-50 Spectra Physics meter, and the intensity of the laser was optimized to be
model. We fixed the maximum of the pulse repetition rate 1.15 × 1010 W∕cm2 . The advantages of the orthogonal scheme
of both lasers at 20 Hz and the pulse duration of both lasers in re-heating mode are a larger atomized mass with a larger
at 6 ns. In the orthogonal DP-LIBS scheme, the primary (first) emitting volume and higher plasma temperature.
pulsed ultraviolet (UV) Nd:YAG laser beam running at 266 nm We varied the voltage on the flash lamps in order to opti-
was focused onto various distances relative to the coffee pellet mize the pulse energy of both lasers with 20 mJ/pulse of the
surface for providing better laser–sample coupling, resulting in first laser and 100 mJ/pulse of the second laser. The ablative
raising mass removal. Hence, electrons near the coffee surface first laser pulse (266 nm) incident to the coffee pellet was
become energized to produce free excited atoms, ions, and blasted at −2.5 mm beneath the surface, producing a spot
plasma emission. The secondary (second) pulsed near-infrared diameter of about 100 μm by a quartz lens of focal length
 6732 Vol. 57, No. 23 / 10 August 2018 / Applied Optics
f  150 mm, as depicted in Fig. 1(b), while the pre-ablation
second laser beam (1064 nm) parallel to the sample surface was
focused at a certain distance (1 mm) by a quartz lens of focal
length f  100 mm above the surface for forming air break-
down and plasma. We also used a fast digital oscilloscope
(500 MHz) and fast photo-diode (rise time 1 ns) for controlling
the timing reliability of the inter-pulse delay between both laser
pulses that was generated using a Stanford digital pulse delay
generator DG535 (Stanford Research Systems, INC, Photonic
Solutions, UK) of 1 ns timing jitter. The inter-pulse delay be-
tween the first and second laser pulses has been varied from 0 to
500 ns in order to maximize the signal-to-noise ratio (SNR) of
the Al II 624.336 nm, Pb I 368.346 nm, Zn I 334.502 nm,
and Cr I 428.040 nm emission lines. It was found that the
optimized inter-pulse delay was at 260 ns for maximizing
the SNR. Here we want to optimize the delay between the two
laser shots using the SNR. The energies of our lasers are well
adjusted constantly with a 3-Sigma energy meter (Coherent,
USA). Different combinations of harmonic nanosecond Nd:
YAG lasers were investigated to obtain a maximal SNR as a
good adjustment. Many pairs of harmonics and 1064 nm of
Nd:YAG for the first and second lasers, respectively, were tried,
and we remarked that 266 and 1064 nm produced the finest
SNR. Each laser has a Gaussian intensity distribution profile,
i.e., TEM00 mode, which was designated by the Gaussian fit
function of a correlation of 0.90 with the beam quality factor of
M 2  1.54. The beam quality factor (M 2 ) represents the de-
gree of variation of a beam from an ideal Gaussian beam. We
adjusted the energy of each laser continuously to improve the
tentative requisite. The best of the combined 266
1064 nm
DP lasers shows the essential role of developing reproducible
and effective ablation on the coffee surface. The influence of
the gap separation (d) between the laser’s blast point and the
pellet surface (−2.5, −1 mm in the opposite direction of
the surface, 0 mm on the surface, and 1, 2.5 mm overdue
the surface) was examined as illustrated in Fig. 1(b).
The sample was rotated on a rotary disc with enough speed
utilizing a motorized X -Y translation holder stage to deliver a
new surface position for each laser blast and suppress the ex-
cessive ablation. We controlled the translation holder by using
a software code to prevent the development of any crust. The
emission plasma of the coffee sample was accumulated by two
quartz lenses with equal f  100 mm, located at 45 deg rec-
ognition angle regarding the direction of the plasma expansion
and directed to a high-resolution spectrometer (PI-Echelle
spectrometer, SE200PI-OH high order, Princeton, USA).
The PI-Echelle has a numerical aperture (f -number):10, focal
length: 20 cm, and fixed spectral resolution λ∕N λ  3100,
where λ is the wavelength over a 190–1100 nm range display-
able in a sole spectrum, and N λ is the full width at half-
maximum (FWHM) peak width. A gated-intensified charge
coupled device (ICCD) camera, (DiCAM-Pro-12 bit, UV en-
hanced, 43000 channels, PCO Computer Optics, Germany)
coupled to the spectrometer was used for detection of the dis-
persed light. The overall linear dispersion of the spectrometer
camera system ranges from 0.006 nm/pixel (at 200 nm) to
0.033 nm/pixel (at 1000 nm) and was utilized to allow simul-
taneous spectral analysis. A personal computer was utilized for
Research Article
spectroscopic data analysis, storage, and processing (data analy-
sis software program code, GRAMS version 8.0, Galactic
Industrial Salem, New Hampshire). An Echelle diffraction gra-
ting has 1 mm number of lines, which is much less than a con-
ventional spectrometer. A quartz prism is located in front of the
grating, whereas altered orders are scribbled vertically in the
focal plane of the Echelle spectrometer. In order to maximize
the spectral line intensity, the detector was synchronized with
the Q-switch of a 266 nm (main ablative) laser with a gate pulse
width (ICCD shutter opening time) of 10,000 ns. The gate
delay (the period between the recording gated pulse and the
pulse from the main ablative first laser) was optimized to be
t d  200 ns. The software program code was used to correct
the plasma emission signal from the chip by deducting the dark
signal of the detector. The wavelength calibration and spectral
response of the detection device were achieved by a reference
mercury lamp. It is of great importance to adjust the various
experimental parameters of the DP-LIBS technique before fir-
ing the single-pulse (SP)- and DP-laser shocks. These param-
eters include the ICCD gate delay (periodic interval between
the excitation laser beat and the data acquisition), number
of accumulation, laser pulse energy ratio (LPER), schematic
configurations, and the gap between the laser blast position
and the pellet surface. Sample #1 was analyzed throughout
the optimum conditions in order to obtain a highly sensitive
DP-LIBS spectrometer.
3. RESULTS AND DISCUSSION
The DP-LIBS spectrum was registered using orthogonal Nd:
YAG lasers as the source of double-pulse excitation at a fixed
260 ns inter-pulse delay (the delay between both laser pulses)
for certain spectral regions (260–490 nm), (495–670 nm), and
(700–950 nm), as depicted in Figs. 2(a)–2(c). This spectrum is
of great benefit for both quantitative and qualitative analyses of
the coffee brands. The elements, Mg, Ca, Na, K, Al, Pb, Zn,
and Cr were identified through the NIST database. The ele-
ments were found with different emission line intensities of
all coffee brands, which indicated that each brand had different
element contents (i.e., different concentrations). To quantify
the Al, Pb, Zn, and Cr contents, we first studied the correlation
between the recorded spectral line intensities of Al (II)
624.336 nm, Pb I (368.346 nm), Zn I (334.502 nm), and
Cr I (428.040 nm) and their concentrations. Different emis-
sion lines, such as Al I (308.215, and 394.400 nm), Al II
(466.305, 559.330, and 624.336 nm), Pb I (266.315, and
368.346 nm), Pb II (504.258, and 666.020 nm), Zn I
(334.502, 472.215, and 481.053 nm), and Cr I (357.870,
360.534, and 428.040 nm), recorded from coffee brands with
known Al, Pb, Zn, and Cr concentrations were checked. The
DP-LIBS emissions and sensitivity depended on the physical-
chemical properties of the brand.
A. Influence of Gate Delay Time on DP-LIBS
Intensity
Figure 3 elucidates the plot between the distinction of normal-
ized signal intensities of the marker atomic transition Pb (I)
368.346 nm line (fingerprint wavelength) against the ICCD
gate delay by directing the DP lasers at altered distances (d)
 Research Article Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6733

(a) 1.0
6000

DP-LIBS Intensity (a. u.)

0.8

Normalized intensity
5000

4000
0.6
3000
0.4
2000 d = -2.5 mm
d = -1 mm
1000 0.2 d = 0 mm
d = 1 mm
0 d = 2 mm
280 320 360 400 440 480 0.0
Wavelength, (nm) -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Gate delay time ( s)
(b)
Fig. 3. Gate delay dependence of the normalized emission inten-
4500
sities of the Pb (I) 368.346 nm line.
DP-LIBS Intensity (a. u.)

4000
3500
3000
2500
2000
averaged. It is evident that the release intensities of the Pb (I)
1500 368.346 nm line increase gradually to reach an extreme value at
1000 around 0.6 μs gate delay and then reduce. Thus, the finest gate
500 delay time is designated to be 0.6 μs due to its nearness to the
0
500 525 550 575 600 625 650
normal t d of the other components under study. In addition,
Wavelength, (nm) the maximum intensity of the Pb (I) was attained by directing
the DP laser rays at 2 mm beneath the pellet surface.
K I 766.489 nm

(c)
N I 868.028 nm
O I 777.539 nmK I 769.896 nm

B. Effect of Laser Pulse Energy Ratio on the

4500
Intensity of the DP-LIBS
DP-LIBS Intensity (a. u.)

Na I 818.325 nm

O I 844.636 nm

4000
Al II 704.206 nm

The influence of LPER on the DP-LIBS signal intensity of Pb

N I 822.314 nm

3500
N I 746.831 nm

at 368.346 nm was investigated, and the data output is eluci-

Pb I 905.082 nm

3000
N I 939.279 nm
O I 926.601 nm

2500
dated in Fig. 4, which shows that the DP-LIBS emission in-
2000
1500
tensity strongly varies with LPER from 0.4 to 12 at various
1000 gaps (d) between the DP lasers’ blast location and pellet surface:
500 −2.5, −1 mm in the anterior of the pellet surface; 0 on the
0 pellet surface; and 1, 2 mm beneath the pellet surface. Both
700 750 800 850 900 950
Wavelength, (nm) Nd:YAG lasers (266 nm and 1064 nm, each of energy 25 mJ,
Δt  5 μs) were applied, and the laser pulse energies were
Fig. 2. Typical DP-LIBS spectra of coffee brand sample #1 were
recorded by double-pulse excitation, orthogonal Nd:YAG lasers at
2 mm behind the coffee surface, and fixed 0.2 μs delay time for some
selected spectral regions (a) from 260 to 490 nm, (b) from 495 to 5000
670 nm, and (c) from 700 to 950 nm wavelength. d = 2 mm
DP-LIBS intensity (a. u.)

d = 1 mm
4000 d = 0 mm
d = -1 mm
d = -2.5 mm
3000
between the DP lasers’ focus position and the pellet surface:
−2.5, −1 mm in the anterior of the pellet surface; 0 on the pellet 2000
surface; and 1, 2 mm beneath the pellet surface. Each laser’s
pulse energy has 25 mJ with the inter-pulse delay between 1000
the two lasers Δt  5 μs. One of the research group observed
maximum enhancement of inter-pulse delay, Δt  8 μs be- 0
0 2 4 6 8 10 12 14
tween laser beats, during the detection of Na in water [19].
Laser 2 energy/ Laser 1 energy
The plasma in our case is hot after laser excitation at the early
stage, and it takes a period of time to cool down. Hence, the Fig. 4. Variations of the DP-LIBS signal intensity of the Pb (I)
ICCD gate delay time (t d ) should be optimized because it may 368.346 nm line [transition probability, Aki  1.37 × 108 s−1 ] with
the laser pulse energy ratio E 2 ∕E 1 by focusing the DP lasers at various
vary from one substance to another. Hereby, the t d changes
distances relative to the coffee surface (d  −2 mm, −1 mm front of the
from 0 to 1 μs (Fig. 3). The first Q-switched Nd:YAG laser surface; 0 mm, 1 mm, and 2 mm behind the surface) recorded in atmos-
pulse trigger, the delay generator, and the acquisition jump pheric pressure. The maximum signal intensity [4297. 120 (a. u.)]
after a fixed delay to sweep the ICCD gate camera. In order of the Pb (I) 368.346 nm line at 2 mm beneath the pellet surface
to suppress the background noise to an acceptable level relative when the second laser pulse energy (E 2 ) is nearly 1.75 times the first
to the double laser pulse spectrum, the registered spectra were laser pulse energy (E 1 ).
 6734 Vol. 57, No. 23 / 10 August 2018 / Applied Optics
organized by neutral density filters. The optimization of the
LPER was obtained within the 260–480 nm wavelength range.
We perceived maximum signal intensity [4297. 120 (a. u.)] on
the Pb (I) 368.346 nm line at 2 mm beneath the pellet surface
when the second laser’s pulse energy (E 2 ) is nearly 1.75 times
the first laser’s pulse energy (E 1 ). Hence, the pulse energies
were carefully selected to be 25 mJ/pulse and 43.75 mJ/pulse
for the 266 nm UV laser and 1064 nm NIR laser, respectively.
One of the research groups reported that the enhancement
factor in the dual-pulse scheme for a copper line was varied
from 4 to 10 [20]. Depending on the above-mentioned results,
it can be established that the dual-pulse LIB spectrometer is
superior, if the finest value is kept as E 2 ≥ 1.75 times E 1 .
C. Determination of Plasma Parameters
The velocities of ions and atoms in plasmas are lower than the
velocity of electrons because the masses of both ions and atoms
are higher than the mass of electrons. Therefore, the emission
spectrum is highly related to the velocity and energy distribu-
tion of the scattering electrons. The incident energy absorbed
by the plasma is used to increase the internal energy of the
plasma. Therefore, an increase of the N e and T e could be due
to the increase of plasma species (electrons, ions, and atoms)
that are produced due to ionization and higher the ablation rate
of the target at higher laser intensity [21]. The pulse energy
density affects the plasma temperature, which, in turn, controls
the velocity gradient during plasma expansion. The electron
temperature (T e ) and density (N e ) and the density of the emit-
ting species in plasma are quite significant for both quantitative
and qualitative analysis of our pellet in order to know the con-
dition of the laser-induced plasma. Hence, the definition of
these plasma parameters is mandatory for evaluating the pres-
ence of local thermodynamic equilibrium (LTE) and satisfying
the optically thin plasma state. To validate the condition of the
optically thin plasma state [22], the corrected line intensities of
various emission lines of Al, Pb, Zn, and Cr were compared
with the ratio of their transition probabilities under the exper-
imental terms [23], which are in agreement within 10%. The
plasma temperature was determined to improve the DP-LIBS
sensitivity and to understand the complicated plasma creation
processes such as excitation, atomization, ionization, and dis-
sociation. The DP-LIBS emission was characterized in the basis
of collisional excitation, photo-ionization, three-body recombi-
nation, collisional ionization, inverse bremsstrahlung, the
de-excitation process, and radiation decay.
In this work, the generated plasma is assumed to be in
the LTE region, i.e., the free electron distributions are
Maxwellian. Thus, the electron can be defined by the Maxwell
distribution, Saha’s equation, and the distribution of population
in various energy states and the radiation density. Whereas the
Maxwell distribution describes the plasma species Boltzmann’s
statistics, Saha’s equation designates the collisional ionization,
and the emission density follows Plank’s law. By knowing the
T e and N e , one could conclude whether the LTE condition
is applicable or not by using the McWhirter criterion [24].
The electron energy distribution function (EEDF) and the
McWhirter circumstance are the most commonly used methods
to estimate LTE. If the free electron distribution in the plasma is
Maxwellian, the created plasma will be in the LTE condition.
Research Article
The electron velocity distribution (EVD) of comparatively
dense and low-temperature plasma (N e > 1016 cm−3 and
kT < 5 eV) was roughly Maxwellian. The fulfillment of the
McWhirter criterion is a necessary but not sufficient condition.
The LTE condition was checked by selecting a line such as Pb I at
368.346 nm to validate a criterion proposed by the McWhriter
circumstance for the SP- and DP-LIBS schemes used (i.e., SP at
266 nm and DP at (266 + 1064) nm as follows [24–26]:
N e cm−3  ≥ 1.6 × 1012 T e K1∕2 ΔEeV3 , (1)
where ΔE is the gap energy between the higher and lesser states
of the spectroscopic alteration lines, which are perceived to be in
the LTE condition. In this demonstration, the N e lower limit
value is 1 × 1014 cm−3 [Eq. (1)]. In addition, the Boltzmann plot
and Stark broadening have been utilized to evaluate the T e and
N e , respectively. Thus, the populations of the greater excited
stages follow a Boltzmann distribution, and their relative
emissivity (εjk ) tends to the equivalence [25]



λjk εjk Ej M T 
ln −
ln , (2)
Ajk g j kT e U T 
where Ajk , g j , E j , and λjk are the transition probability of the
excited state, the statistical weight, the energy of higher stage,
and the wavelength, respectively. T e is the plasma temperature;
k is the Boltzmann constant; U T  is the partition function, and
M T  is the number of ion density. By drawing E j versus the
left-hand side of Eq. (2), the plasma electronic temperature is
able to be estimated from the slope of the straight line, i.e.,
the slope  −1∕kT.
In this work, the plasma temperature exists in LTE, i.e., all
temperatures are assumed to be equal, T e ≈ T ion ≈ T plasma for
all elements in the coffee samples. Thus, the calculated plasma
temperature for Mg I lines represents the plasma temperature
for any element that appears in the spectrum. We selected
the Mg lines for calculating the plasma temperature since
the concentration of magnesium as a nutritional element in
coffee samples was very high compared to the concentration
of toxic elements Al, Zn, Pb, and Cr. In addition, the DP-
LIBS intensity of all magnesium lines (Mg II 279.799 nm,
Mg I 285.212 nm, Mg II 293.650 nm, and Mg I
383.230 nm) was higher than the intensity of Al, Zn, Pb,
and Cr lines.
The neutral spectral lines of the Mg I wavelength, the stat-
istical weight, the energy of the upper stages, and the transition
probability were engaged from the National Institute of
Standards and Technology’s (NIST’s) data base [11] and tabu-
lated in Table 1. Hence, the Boltzmann plot method was applied
to evaluate the electron temperature, T e  6785  425 K for
UVSP-LIBS and 9625  523 K for (UV
IR) DP-LIBS. The
inset of Fig. 5 elucidates the linearity in the Boltzmann plot
method utilizing four fingerprint wavelengths of atomic magne-
sium (Mg I) transition lines at 277.9 nm (3s 3p 3 P2 − 3p2 3 P2 ),
284.8 nm (3s 3p 3 P1 − 3s 5d 3 D2 ), 765.9 nm (3 s 4 s 3 S1 − 3s 5p
3 P ), and 821.3 nm (3s 3d 1 D –3s 6f 1 F ), respectively, for
1 2 3
which their transition probabilities are also known. These tran-
sition lines of atomic magnesium are important for calculating
the T e (i.e., plasma temperature) through the Boltzmann plot
method. Hence, the populations of the excited stages track the
 Research Article Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6735

Table 1. Various Atomic Magnesium (Mg I) Lines Used in the Boltzmann Plot for Calculating the DP-LIBS
Plasma Temperature
State of Wavelength Transition Lower Energy Upper Energy Statistical Statistical
Lines λik (nm) Probability Ajk s−1  Level E i (eV) level E k (eV) Weight g i Weight g k
Mg I 277.9 4.09 × 108 2.72 7.18 5 5
Mg I 284.8 1.77 × 107 2.71 7.06 3 5
Mg I 765.9 1.23 × 106 5.10 6.72 3 3
Mg I 821.3 4.38 × 106 5.75 7.26 5 7

Boltzmann law for plasma with good regression (∼0.98), and an
LTE postulate is satisfied in plasma by confirmation of the lin-
earity. In order to calculate the plasma temperature, the spectro-
scopic factors for Mg atomic lines, which are stated in the NIST
database, are used (Table 1). Thus, the T e can be estimated from
the obtained slope as depicted in the inset of Fig. 5. This plot was
employed to clarify the dependence of the plasma temperature
along the plasma plume on the gap from the pellet surface for the
plasma formed by the 266 nm SP laser with 25 mJ/pulse total
laser energy and (266
1064) nm orthogonal DP lasers with
2 × 25 mJ∕pulse total laser energy in re-heating schemes as elu-
cidated in Fig. 5. The data depicted that the T e changes from
7213 to 930 K in the distance (0–1.5 mm) range of the excited
plasma under a 266 nm SP laser and from 10,732 to 2770 K in
the distance (0–2.75 mm) region by DP-LIBS operating at
(266
1064) nm. It can be remarked that the T e s are hotter
in the case of DP-LIBS by nearly 1.5 times as related to the
UVSP-LIBS spectrum. The T e s are hotter near the target surface
and quickly decrease in the UVSP-LIBS scheme, while they
slowly decrease in the UVDP-LIBS scheme along the plume
expansion axis. The error bars revealed the standard deviation
of 10 gathered spectra.
In plasmas, the spectral line broadening is disturbed by the
pressure and Doppler broadening. Due to low temperatures
and greater densities in this plasma, the donations by reso-
nance, Doppler, and Van der Waals broadening is minor for
the Pb I at 368.346 nm, which had been recycled to estimate
the N e . Two spectroscopic methods are commonly used to de-
termine the electron number density; the first method requires
a measure of the Stark broadening of the spectral lines, and
the second method requires the measure of the population ratio
of two successive ionization states of the same element. The
14000
SP (266 nm)
DP (266 + 1064) nm
12000 -2
ln (I /gA)
emission spectral lines of the laser-produced plasma are notice-
ably broadened; thus, N e can be extracted from the widths of
the spectral lines. The Stark line broadening, induced by
collisions of emitting atoms with charged species, is the dom-
inant mechanism manipulating these emission spectra in com-
parison with the Doppler broadening [27,28]. In the present
work, the Doppler width is estimated to be about few picom-
eters (pm) for the Pb transition. This width is too small to be
recorded, and hence, it is ignored. There is also another broad-
ening mechanism is likely to contribute significantly to the line-
width observed in the produced plasma, namely, resonance
pressure broadening, and it is proportional to the ground state
number density of the corresponding species and transition os-
cillator strength. Only a small transition oscillator strength has
been found for Pb I emission lines. Therefore, the resonance
broadening part can also be safely neglected. The N e could
be calculated from the FWHM of the neutral Pb line (Pb I
at 368.346 nm). If the Stark width of the line (FWHM) Δλ1∕2
is determined experimentally, and the theoretical Stark
width parameter ω for neutral atoms is calculated or tabulated
in the literature, therefore, the N e can be obtained from the
relation [28]
Δλ1∕2 A


  

Ne N e 1∕4 3 −13 Ne
 2ω
3.5A x 1 − N ω ,
1016 1016 4 D 1016
(3)
where A is the ion-broadening parameter, and N D is the num-
ber of particles in the Debye sphere. Both ω and A are weak
functions of temperature. The contribution is almost due to
electron impact, and the half-width of the Stark-broadened
transition can be estimated to a good approximation [29]. The
impact approximation is used for electrons and the quasi-static
approximation is used for ions due to their much larger mass.

10000
-4
Comparing the electron and ion mass, it can be concluded that
-6
the contribution of ion broadening is much smaller in compari-
T e (K)

8000 DP (266 + 1064 nm)

-8
SP (266 nm)
Fitting line
son with electron broadening in plasma. Hence, the velocity of
6000 -10
6.7 6.8 6.9 7.0 7.1 7.2 7.3 the electron is much larger than the velocity of the ion.
E (eV)
4000 Therefore, the second term of Eq. (4), which describes the
2000 ion-broadening contribution, can be neglected, whereas the im-
pression of the width is nearly resulting from electron impact
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 (collision). We selected a well-isolated and Stark-broadened
Distance (mm) transition of lead, Pb I at 368.346 nm and hydrogen H α at
Fig. 5. Dependence of T e predicted from the Boltzmann plot of the 656.285 nm. The widths of these lines was determined. The
plasma on the plume expansion distance using SP and DP geometries, N e (cm−3 ) is evaluated from the Stark broadening of the
namely, 266 nm SP and (266
1064) nm DP laser schemes. spectral line from the following relation:
 6736 Vol. 57, No. 23 / 10 August 2018 / Applied Optics Research Article


Ne
Δλ1∕2 A  2ω
1016
where the right linewidth Δλ1∕2 of the spectral line Pb I at
368.346 nm is considered by subtracting the instrumental line-
width of the observed linewidth,
Δλtrue  Δλobserved − Δλinstrument :
The instrumental linewidth of the LIBS spectrometer is
inferred to be 0.05  0.01 nm by means of a dye laser of a
narrow linewidth. The equivalent prorated fault is sacrificial
to 10%, and the width of the Pb I at the 368.346 nm line
was mathematically fitted by a Lorentzian shape as illustrated
in Fig. 6(a). The LIBS intensity of the well-isolated Pb I at the
368.346 nm spectral line was selected from other spectral lines
to measure Δλ1∕2 . The comparative value of the Stark width
(a)
4000
DP-LIBS Intensity (a. u.)
factor (ω) is equal to 0.0131 nm [27], and substituting the
, (4) value of Δλ1∕2 in Eq. (4), the N e values can be determined.
In this work, to evaluate the N e from Eq. (4), the Stark-
broadened profile of atomic lead, Pb I at 368.346 nm, has been
utilized as illustrated in Fig. 6(a). The FWHM of the Pb (I)
profile is (0.5532  0.01) nm. Hence, the calculated N e for
(5) Pb I is 1.07  0.90 × 1017 cm−3 . The evaluated N e encoun-
ters the electron energy distribution function (EEDF) criterion
for the LTE with N e > 1016 ; this value of N e is larger than the
critical electron density N e critical  1.80  0.01 × 1016 cm−3.
Calculations of the FWHM of the H α line at 656.285 nm are
offered in Fig. 6(b), depicting the experimentally observed
line profile bounded by two vertical lines representing
FWHM as (0.84  0.08) nm. The N e can be calculated via the
relation [30]

0.67965
Ne
FWHM  0.549 nm × : (6)
1023 m−3
Experimental result
Fit line The N e is deduced as 1.89  0.20 × 1017 cm−3 . A good

Pb I (368.346 nm) agreement between the number densities derived from the
3000 Stark-broadening parameter of Pb I at 368.346 nm and the
H α line at 656.285 nm is obtained. Average values of the N e
2000 were used to determine the compositions of the targets using
calibration-free LIBS (CF-LIBS). Figure 6(c) depicts the varia-
1000 tion of N e for the UVSP-laser as a function of the coupling of
the laser beam to the pellet surface. The N e close to the
0 pellet surface in 266 nm SP-LIBS and (266
1064) nm
367.0 367.5 368.0 368.5 369.0 369.5 370.0
Wavelength, (nm)
DP-LIBS are 1.51  0.50 × 1017 cm−3 and 2.99  0.50 ×
1017 cm−3 , and thus, reduce nonlinearity to 1.39  0.50 ×
(b) 500 1016 cm−3 and 1.75  0.50 × 1016 cm−3 , respectively, which
Experimental result
DP-LIBS Intensity (a. u.)

Fit line includes the errors associated with the extracted FWHM, start-
400 ing at 0.1 mm from the target surface. It is worth mentioning
that the N e has a peak up to 0.2 mm and a reduction up to
300 3 mm. During the start-up of the plasma, it grows directly by
the photo-ionization (PI) process of the excited atoms. In
200 addition, the mass ablation is greater for the DP-LIBS than that
attained with UVSP-LIBS, as elucidated in Fig. 6(c). The laser
100 pulse energy (LPE) was adjusted until the laser saturated at a
650 652 654 656 658 660 662 664 certain level of the pulse energy as follows: 25 mJ/pulse from
Wavelength, (nm) UVSP and 2 × 25 mJ∕pulse for DP at (266
1064) nm.
(c) 18
Above the definite value of LPE, the free negative species in
10
the warm plasma absorb more laser energy, and thus the inverse
bremsstrahlung (IB) effects are considered.
SP (266 nm)
N e ( cm )

D. Quantification of Micro-Toxic and Nutrition

-3

DP (266 + 1064) nm
17
Concentration in the Test Sample
10
Atomic and ionic emission intensities are utilized for quantifi-
cation via the DP-LIBS spectrometer of various elements in
different coffee brands. We used two methods for the analysis,
a standard calibration curve method (SCCM) as the first
10
16
method and the CF-LIBS as the second method. We used the
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Distance (mm) SCCM method to estimate the level of Al, Pb, Zn, and Cr in
the coffee brands. In order to build the curves, we used standard
Fig. 6. (a) Stark-broadened line profile of the Pb (I) line at
samples of known concentrations of Al (3050, 1500, 1000,
368.346 nm. (b) Calculation of full width at half-area of the hydrogen
H α line at 656.285 nm at 50 mJ laser energy. (c) Dependence of N e 500, 250, 100 μg L−1 ), Pb (1050, 800, 600, 400, 200,
estimated from the typical spectral line of Pb (I) 368.346 nm, the 100 μg L−1 ), Zn (1050, 800, 600, 400, 200, 100 μg L−1 )
Boltzmann plot of the plasma on the plume expansion distance using and Cr (1050, 400, 300, 200, 100 μg L−1 ) and then mixed the
SP and DP geometries, namely, 266 nm SP and (266
1064) nm DP samples with a high purity of these elements (99.99%) in pow-
laser schemes. der form added to a KBr blinder in a ball mill to obtain sample
 Research Article
uniformity. Sometimes the intense spectral lines seem to have
either a dip at the central frequency (self-absorption) or a flat-
topped profile (i.e., saturation of the DP-LIBS signal with laser
fluence) [31]. The self-absorption depends on the level of en-
ergy degeneracy, the oscillator strength, the plasma parameters,
and the broadening parameters. In this study, we did not find
any dip at the central frequency or saturation of the obtained
spectral lines. The self-absorption happens due to atoms in the
plasma having higher energies relative to atoms on its surface.
Hence, when the atoms in the plasma de-excite to create res-
onance lines, the outer atoms absorb the energy, leading to less
intense spectral lines. In addition, the self-absorption can affect
the accuracy of the calibration curves. Since self-absorption is
often seen in intense lines emitted from ions and atoms that
relax back to their ground state, the lines that terminate at
higher energy levels can be chosen as analytical lines since
the effects of self-absorption are reduced. It is known that both
self-absorption and matrix effects can disturb the linear relation
between peak intensities and the concentrations of a certain
element, while calibration curves can be applied to matrix-
matched samples.
In the DP-LIBS system, the calibrated standard pellets of Al,
Pb, Zn, and Cr at different concentrations were prepared by
adding exactly known masses of Al (II) sulfate, Pb (II) sulfate,
Fig. 7. Superimposed DP-LIBS spectra of standard samples having 100, 200, 400, 600, 800, and 1050 μg/L lead concentrations for plotting the
calibration curve.
Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6737
Zn (II) sulfate, and Cr (II) sulfate in the coffee sample matrix
(sample #1). Subsequently, we used Al (II) at 624.336 nm, Pb I
at 368.346 nm, Zn I at 334.502 nm, and Cr I at 428.040 nm
emission lines as the marker lines for drawing the calibration
curve. The added concentrations of aluminum sulfate, lead sul-
fate, zinc sulfate, and chromium sulfate were 100, 250, 500,
1000, 1500, and 3050 μg/l; 100, 200, 400, 600, 800, and
1050 μg/l; 100, 200, 400, 600, 800, and 1050 μg/l; and 50,
100, 200, 300, 400, and 1050 μg/l, respectively. The respective
DP-LIBS signal intensity spectra for each Al, Pb, Zn, and Cr
concentration added to the sample #1 matrix are illustrated in
Fig. 7, which depicts a linearity in the growth of the DP-LIBS
signal intensity as a function of Al, Pb, Zn, and Cr concentra-
tion. The results presented in Fig. 7 were used to plot the linear
calibration curve, as illustrated in Fig. 8 for the detection of Al,
Pb, Zn, and Cr in various coffee pellets. The calibration curves
were used to quantify the Al, Pb, Zn, and Cr concentrations
existing in the coffee pellets.
The typical DP-LIBS calibration curves were constructed by
drawing the DP-LIBS line intensity of special species against
the concentrations of Al, Pb, Zn, and Cr in the standard sample
as depicted in Figs. 8(a)–8(d), respectively. From these curves,
we can easily determine the concentrations of Al, Pb, Zn, and
Cr in the unknown coffee by the process described in previous
 6738 Vol. 57, No. 23 / 10 August 2018 / Applied Optics Research Article

Fig. 8. Calibration curves for DP-LIBS measurement of Al, Pb, Zn, and Cr elements.

publications [32,33]. We used the intense emission lines Al (II) following conditions: LTE and no self-absorption. By using the
624.336 nm, Pb I (368.346 nm), Zn I (334.502 nm), and Cr I Boltzmann equation, the concentration of the neutral atoms
(428.040 nm) as marker wavelengths for calibrations by the can be determined as follows [34]:
DP-LIBS system to quantify the level of Al, Pb, Zn, and Cr  
U z T  kE kT
existing in the coffee samples. These specific atomic transition FCz  I k e B , (7)
lines were chosen because they are isolated, strongly intense, Ak g k
and do not interfere with other atomic transition lines in this
where C z is the concentration of atoms, the F factor is asso-
spectral region. The intensities of the Al (II) 624.336 nm, Pb I
ciated with the ablated mass, I k is the line intensity, g k is the
(368.346 nm), Zn I (334.502 nm), and Cr I (428.040 nm)
statistical weight of the higher state, Ak is the transition prob-
emission lines were perceived to increase linearly versus the
ability, U T  is the partition function, kB is the Boltzmann
concentration of Al, Pb, Zn, and Cr, having a correlation
constant, E k is the energy of the higher state, and T eV is
coefficient (R 2 ) value of 0.9965, 0.9892, 0.9796, and
the electron temperature. The wholly spectroscopic factors
0.9930, respectively, as illustrated in Figs. 8(a)–8(d). It can
utilized for the compositional analysis were engaged from
be seen from the calibration curve that the entire concentration
the NIST database [11].
is linear, and the good linear least-squares fit of the DP-LIBS
The average values of the plasma parameters (T e and N e )
calibration curve with a small band error (7%) was verified by
were utilized in our analysis. We used the following equation to
the great standards of the correlation coefficient value
obtain the concentrations of the ionized species [35–37]:
(R 2  0.9965). That means that both self-absorption and ma-
trix effects have no effect on the linear relation between peak 3  
C z
1 2me k B T 2 2U z
1 E ion
intensities and the concentrations of Al, Pb, Zn, and Cr. Thus,  exp − cm−3 : (8)
Cz ne  h3 Uz kB T
the emission lines of these elements could be used for further
analytical study. The sum of the concentrations of the ionized and neutral atoms
The CF-LIBS method was used to estimate the composition of any element in a coffee brand represents the total concen-
of substances in different coffee brands after knowing the tration as follows:
plasma parameters (T e and N e ). The advantage of this method C st  C sz
C sz
1 : (9)
is that this method does not require standard samples and
involves self-calibration as well. This method is based on the Approving this method, we detected 11 elements with various
creation of emission spectra of an unknown sample for every concentrations in all the coffee brands. Therefore, the concen-
element existing in the sample and then the calculation of every trations of the elemental contents in our all coffee brands can
emission line area. In addition, this procedure obliges the be computed, and then the compositional analysis of the
 Research Article Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6739

ICP-OES results can be matched with the DP-LIBS results. In
the CF-LIBS, we used the Boltzmann relation to estimate the
concentration of Al, Pb, Zn, and Cr in the coffee samples
[Eq. (7)]. All other nutrient elements existing in the coffee
brands were utilized by choosing their neutral lines Mg (I)
285.212 nm, Na (I) 588.995 nm, K (I) 535.957 nm, and
Ca (I) 445.478 nm for completion of the quantitative analysis
of the coffee sample. In the recorded spectrum, we selected Mg
(I) 285.212 nm and Pb (I) 368.346 nm for calculating the T e
and N e of the plasma, respectively.
In this study, the concentration of each of the micro-toxic
substances Al, Pb, Zn, and Cr in all coffee brands was
determined using DP-LIBS, the CF-LIBS, and the standard ana-
lytical system, ICP-OES. Figures 9(a)–9(d) depict the variations
in the concentrations of Al, Pb, Zn, and Cr versus the type of
coffee brands using the DP-LIBS, ICP-OES, and CF-LIBS tech-
niques. The concentrations of Al, Pb, Zn, and Cr estimated in all
the coffee brands are between 515–2561 μg/L, 121–1610 μg/L,
305–1780 μg/L, and 495–1350 μg/L, respectively. The concen-
tration of Pb in the 13 coffee brands was determined using the
DP-LIBS, ICP-OES, and CF-LIBS techniques, as shown in
Table 2. It can be seen from Figs. 9(a)–9(d) and Table 2 that
the estimated concentrations of Al (2561 μg/L), Pb
(1610 μg/L), Zn (1780 μg/L), and Cr (1350 μg/L) were found
to be most abundant in the test samples #2, #4, #6, and #4, re-
spectively. It is evident that the Al, Pb, Zn, and Cr compositions
found by both the DP-LIBS and CF-LIBS techniques show ex-
cellent agreement with those of the ICP-OES as a standard tech-
nique. About (2–4)% deviation from the standard values has
been remarked in these brands. The benefit of investigating la-
ser-produced plasma in atmospheric pressure is that the plasma
remains confined. In addition, some lines of hydrogen, specif-
iccally the H α line, manifest in the recorded spectrum. Since the
line profile of the H α line is free from self-absorption, it can be
used to presume a credible value of the N e . The limit of detec-
tion is basically referring to the degree of detection capability of
any DP-LIBS scheme. It is highly important for our samples to
compute the figure of merit of our analytical device. We
processed an SNR of 10 laser blasts for estimating the limit
of detection (LOD). This procedure is regularly used for the
spectroscopic analysis that demonstrates reference noise lines.
Principally, LOD is considered as 3 times the ratio of the
standard deviation (SD) of the stray levels (σ B ) to the slope
of the calibration curve (S, sensitivity of calibration), as [36,37]
LOD  3σ B ∕S, (10)
where S is the slope of the calibration curve. The LOD is
counted on the stray light and on residual background noises.
The SD of the typical value was figured out from our DP-
LIBS data. Hence, the LOD for Al, Pb, Zn, and Cr in coffee
were found to be about 105.13, 90.17, 83.58, and 68.78 μg/
L, respectively. It is worth mentioning that the coffee composi-
tions have a large impact on LOD. In addition, the maximum
intensity of the spectral line depends on the elemental compo-
sitions, the baseline, and the input laser influence. Therefore,
the DP-LIBS result can be achieved by knowing the area of a
neutral line (ANL) for a special emission line distributed by
an illustrative baseline area (BA), (i.e., ANL/BA). The ANL/
BA is proportional to the micro-toxic element concentration
in a given target at constant laser intensity. The auto-integrator
of the instrument was employed to determine the magnitude of
the noise. Commonly, the SNR value is a tolerable value for
LOD estimation. A comparison of the accuracy of the DP-
LIBS data and the ICP-OES outcomes by means of relative ac-
curacy (RA) was carried out. It was found that there was agree-
ment between the results obtained via the DP-LIBS and the
ICP-OES techniques of enacting the statistical correlation test
on the developed data. The RA practice was estimated by the
following equation [10]:
pffiffi
d
SDx t0.975
n
RA  × 100, (11)
M
where M is the measurement from the standard ICP-OES tech-
nique; d is the alteration between the results from DP-LIBS and
ICP-OES; SD is the standard deviation of the DP-LIBS gauges;
n is the measurement number, and t 0.975 is the t-value at 2.7%
error assurance. The t-statistic is the departure ratio of an
estimated issue from its typical error and its hypothetical mag-
nitude. The linear correlation factor for our SCCM demon-
strates a great linear correlation of about 0.9933 between the
strongest DP-LIBS intensity and ICP. The mutual relative stan-
dard deviation is within the 1%–7% range for the DP-LIBS
geometry [38]. The RAs of Al, Pb, Zn, and Cr in each coffee
brand for each pointed-up concentrations are 5.82%, 6.20%,

Table 2. Comparison of DP-LIBS and ICP-OES for Lead Recorded in the Brand Samples
Brand Sample Fingerprint Wavelength DP-LIBS (μg/L) (SCCM) ICP-OES (μg/L) CF-LIBS (μg/L)
1 368.346 1361  130 1254  110 1390  140
2 368.346 623  100 634  80 650  130
3 368.346 1398  140 1455  120 1510  140
4 368.346 1541  160 1592  130 1610  110
5 368.346 234  80 246  70 260  120
6 368.346 623  90 654  80 665  120
7 368.346 878  120 921  100 910  140
8 368.346 125  60 123  40 120  80
9 368.346 121  70 112  50 115  80
10 368.346 125  80 110  60 90  90
11 368.346 291  100 302  80 320  110
12 368.346 153  110 164  90 180  130
13 368.346 1398  140 1432  120 1460  150
 6740 Vol. 57, No. 23 / 10 August 2018 / Applied Optics
(a)
(b)
(c)
(d)
Fig. 9. Comparison among the concentrations in μg/l of various
micro-toxic elements, (a) Al, (b) Pb, (c) Zn, (d) Cr, in all coffee brands
obtained from the DP-LIBS standard calibration curve method (green
column), CF-LIBS (red column), and ICP-OES (black column)
techniques.
3.91%, and 3.76% relative to the DP-LIBS and ICP-OES
concentrations. This work is important to detect the micro-toxic
elements in some coffee brands, which have an effect on the
health and well-being of coffee drinkers. Hence, in order to ob-
tain food safety and protection, there is an urgent requirement
for strong governmental inspection.
Research Article
4. CONCLUSIONS
We report on the measurement of micro-toxic concentrations
in commercial coffee brands taken from Egyptian stores under
optimized (266
1064) nm dual-pulse Nd:YAG laser ablation
spectroscopy and inductively coupled plasma-optical emission
spectroscopy systems. We used the intense emission lines Al
(II) 624.336 nm, Pb I (368.346 nm), Zn I (334.502 nm),
and Cr I (428.040 nm) as markers, and the intensity of those
lines depicted a great linear growth with Al (II) 624.336 nm, Pb
I (368.346 nm), Zn I (334.502 nm), and Cr I (428.040 nm)
concentrations, respectively. The results show that the signal
intensity of the spectral lines improved as opposed to the in-
crease of the laser pulse energy from single to dual pulses and
the decrease with the increase of the gate delay. The finest gate
delay (t d ) and the gap (d ) between the DP-laser blasts and the
pellet surface are crucial settings for peak growing in a DP-LIBS
scheme. Such experimental parameters, like lasers pulse energy
ratio (LPER), distance (d ), and the gate delay (t d ) should be
adjusted. Our DP-LIBS setup was investigated to enhance the
DP-LIBS sensitivity for determination of micro-toxic concen-
tration. We used the SCCM method as well the CF-LIBS
method to estimate the Al, Pb, Zn, and Cr contents in coffee
brands. In DP-LIBS performance, the greatest peak improve-
ment was achieved when the second laser had a pulse energy
1.75 times greater than the primary laser pulse. The measured
concentrations of Al, Pb, Zn, and Cr by DP-LIBS were verified
via ICP-OES as a standard technique. We observed that the
measurements for estimating the Al, Pb, Zn, and Cr contents
in coffee brands by both techniques (DP-LIBS and ICP-OES)
are in fine matching, i.e., there is a good linear correlation for
both spectroscopic methods, and the results are more reliable
via SCCM. Besides, with further optimization and calibration
to a fast system like DP-LIBS, it can be exploited for coffee
classification and identification of coffee contents on the
market.
Acknowledgment. The authors acknowledge the con-
tinuous support equipped for this work by the National Institute
of Laser Enhanced Sciences (NILES), Cairo University, Giza,
Egypt, the Laser Laboratories at University of Central Florida
(UCF), USA, and the Department of Environmental Health,
College of Public Health, Imam Abdulrahman Bin Faisal
University, Dammam, Saudi Arabia.
REFERENCES
1. P. Ricciardi, F. Torchia, E. Belloni, E. Lascaro, and C. Buratti,
“Environmental characterization of coffee chaff, a new recycled
material for building applications,” Constr. Build. Mater. 147, 185–193
(2017).
2. N. D. Freedman, Y. Park, C. C. Abnet, A. R. Hollenbeck, and R. Sinha,
“Association of coffee drinking with total and cause-specific mortality,”
N. Engl. J. Med. 366, 1891–1904 (2012).
3. M. You, Y. Huang, J. Lu, and C. Li, “Environmental implications of
heavy metals in soil from Huainan,”Anal. Lett. 48, 1802–1814 (2015).
4. M. S. Kim, H. G. Min, N. Koo, J. Park, S. H. Lee, G. I. Bak, and J. G.
Kim, “The effectiveness of spent coffee grounds and its biochar on the
amelioration of heavy metals contaminated water and soil using
chemical and biological assessments,” J. Environ. Manage. 146,
124–130 (2014).
 Research Article
5. X. S. Wei, Y. W. Wu, and L. J. Han, “Determination of lead and cad-
mium in water and pharmaceutical products by inductively coupled
plasma optical emission spectrometry with pre-concentration by
Thiourea immobilized silica,” Anal. Lett. 48, 996–1008 (2015).
6. M. F. Marcone, “Composition and properties of Indonesian palm civet
coffee (Kopi Luwak) and Ethiopian civet coffee,” Food Res. Int. 37,
901–912 (2004).
7. J. Serrano-Gómez, H. López-González, M. T. Olguín, and S.
Bulbulian, “Carbonaceous material obtained from exhausted coffee
by an aqueous solution combustion process and used for cobalt
(II) and cadmium (II) sorption,” J. Environ. Manage. 156, 121–127
(2015).
8. N. Ahmed, Z. A. Umar, R. Ahmed, and M. A. Baig, “On the elemental
analysis of different cigarette brands using laser induced breakdown
spectroscopy and laser-ablation time of flight mass spectrometry,”
Spectrochim. Acta B 136, 39–44 (2017).
9. N. Ahmed, R. Ahmed, and M. A. Baig, “Analytical analysis of different
karats of gold using laser induced breakdown spectroscopy (LIBS)
and laser-ablation time of flight mass spectrometer (LA-TOF-MS),”
Plasma Chem. Plasma Process. 38, 207–222 (2018).
10. A. A. I. Khalil, M. A. Gondal, M. Shemis, and I. Khan, “Detection of
carcinogenic metals in kidney stones using ultraviolet laser-induced
breakdown spectroscopy,” Appl. Opt. 54, 2123–2131 (2015).
11. NIST, “Atomic spectra database,” 2017, https://www.nist.gov/pml/
atomic-spectra-database.
12. N. N. Al-Hassan, “The prevalence of iron deficiency anemia in a Saudi
University female Students,” J. Microsc. Ultrastruct. 3, 25–28 (2015).
13. J. A. T. Pennington, “Aluminum content of foods and diets,” Food
Addit. Contam. 5, 161–232 (1988).
14. FAO/WHO, Summary Evaluation Performed by the Joint FAO/WHO
Expert Committee on Food Additives (JECFA) Toxicological
Evaluation of Certain Veterinary Drug Residues in Food Geneva,
WHO Food Additives Series (1994), pp. 33–51.
15. A. A. I. Khalil, “Spectroscopic studies of UV lead plasmas produced by
single and double-pulse laser excitation,” Laser Phys. 23, 015701
(2013).
16. A. A. I. Khalil and M. A. Morsy, “Quantitative determination of copper in
a glass matrix using double pulse laser induced breakdown and
electron paramagnetic resonance spectroscopic techniques,” Talanta
154, 109–118 (2016).
17. S. J. Genuis, “Elimination of persistent toxicants from the human
body,” Human Exp. Toxicol. 30, 3–18 (2011).
18. W. C. Winkelmayer, M. J. Stampfer, W. C. Willett, and G. C. Curhan,
“Habitual caffeine intake and the risk of hypertension in women,”
J. Am. Med. Assoc. 294, 2330–2335 (2005).
19. A. Kuwako, Y. Uchida, and K. Maeda, “Supersensitive detection of
sodium in water with use of dual-pulse laser-induced breakdown
spectroscopy,” Appl. Opt. 42, 6052–6056 (2003).
20. V. Pinon, C. Fotakis, G. Nicolas, and D. Anglos, “Double-pulse laser-
induced breakdown spectroscopy with femtosecond laser pulses,”
Spectrochim. Acta B 63, 1006–1010 (2008).
21. V. Piscitelli, M. A. Martinezl, A. J. Fernandez, J. J. Gonzalez, X. L.
Mao, and R. E. Russo, “Double pulse laser induced breakdown spec-
troscopy: experimental study of lead emission intensity dependence
on the wavelengths and sample matrix,” Spectrochim. Acta B 64,
147–154 (2009).
Vol. 57, No. 23 / 10 August 2018 / Applied Optics 6741
22. R. Noll, Laser-Induced Breakdown Spectroscopy—Fundamentals
and Applications (Springer, 2012).
23. L. Radziemski and D. Cremers, “A brief history of laser-induced break-
down spectroscopy: From the concept of atoms to LIBS 2012,”
Spectrochim. Acta B 87, 3–10 (2013).
24. G. S. Senesi, M. Dell’Aglio, R. Gaudiuso, A. D. Giacomo, C. Zaccone,
O. D. Pascale, T. M. Miano, and M. Capitelli, “Heavy metal concen-
trations in soils as determined by laser induced breakdown spectros-
copy (LIBS), with special emphasis on chromium,” Environ. Res. 109,
413–420 (2009).
25. A. A. I. Khalil, W. O. Younis, and M. A. Gondal, “A comparative study
of highly-ionized Al plasma based on dual pulse laser-induced break-
down spectroscopy,” Indian J. Phys. 91, 327–336 (2017).
26. R. W. P. McWhirter, Plasma Diagnostic Techniques, R. H. Huddlestone
and S. L. Leonard, eds. (Academic, 1965), Chap. 5, pp. 206–465.
27. A. Alonso-Medina, “Experimental determination of the Stark widths of
Pb I spectral lines in a laser-induced plasma,” Spectrochim. Acta B 63,
598–602 (2008).
28. C. Colon, A. Alonso-Medina, and C. Herran-Martinez, “Spectroscopic
study of a laser-produced lead plasma: experimental atomic transition
probabilities for Pb III lines,” J. Phys. B 32, 3887–3897 (1999).
29. S. Maurmann, V. A. Kadetov, A. A. I. Khalil, H.-J. Kunze, and U.
Czarnetzki, “Thomson scattering in low temperature helium plasmas
of a magnetic multipole plasma source,” J. Phys. D 37, 2677–2685
(2004).
30. B. Praher, V. Palleschi, R. Viskup, J. Heitz, and J. D. Pedarnig,
“Calibration free laser-induced breakdown spectroscopy of oxide
materials,” Spectrochim. Acta B 65, 671–679 (2010).
31. T. Takahashi and B. Thornton, “Quantitative methods for compensa-
tion of matrix effect and self-absorption in laser induced breakdown
spectroscopy signals of solids,” Spectrochim. Acta B 138, 31–42
(2017).
32. S. Kyuseok, C. Hyungki, and L. Yong, “Investigation of the line broad-
ening mechanism for laser-induced copper plasma by time resolved
laser induced breakdown spectroscopy,” Microchem. J. 63, 53–60
(1999).
33. N. K. Rai and A. K. Rai, “LIBS—An efficient approach for the
determination of Cr in industrial wastewater,” J. Hazard. Mater.
150, 835–838 (2008).
34. G. Cristoforetti, E. Tognoni, and L. A. Gizzi, “Thermodynamic equilib-
rium states in laser-induced plasmas: from the general case to laser-
induced breakdown spectroscopy plasmas,” Spectrochim. Acta B 90,
1–22 (2013).
35. E. D. Andrea, S. Pagnotta, E. Grifoni, S. Legnaioli, G. Lorenzetti, G.
Palleschi, and V. Lazzerini, “A hybrid calibration-free/artificial neural
networks approach to the quantitative analysis of LIBS spectra,”
Appl. Phys. B 118, 353–360 (2015).
36. E. Tognoni, G. Cristoforetti, S. Legnaioli, and V. Palleschi,
“Calibration-free laser-induced breakdown spectroscopy: state of
the art,” Spectrochim. Acta B 65, 1–14 (2010).
37. A. Shrivastava and V. B. Gupta, “Methods for the determination of limit
of detection and limit of quantification of the analytical methods,”
Chron. Young Sci. 2, 21–25 (2011).
38. A. Miziolek, V. Palleschi, and I. Schechter, Laser-Induced Breakdown
Spectroscopy (LIBS): Fundamental and Applications (Cambridge
University, 2006), pp. 1–614.