Investigations into Modeling and Further Estimation of Detection Limits

Article pubs.acs.org/ac

Investigations into Modeling and Further Estimation of Detection Limits of the Liquid Electrode Dielectric Barrier Discharge Tobias Kraḧ ling,*,†,§ Antje Michels,† Sebastian Geisler,‡ Stefan Florek,‡ and Joachim Franzke*,† †

Leibniz-Institut für Analytische Wissenschaften - ISAS - e. V., Bunsen-Kirchhoff-Straβe 11, 44137 Dortmund, Germany Leibniz-Institut für Analytische Wissenschaften - ISAS - e. V., Albert-Einstein-Straβe 9, 12489 Berlin, Germany

‡

S Supporting Information *

ABSTRACT: The liquid electrode dielectric barrier discharge (LE-DBD) is a miniaturized atmospheric pressure plasma as emission excitation source for elemental determination with pulsed behavior. Metals dissolved in liquids are detectable in flow systems with low flow rates of 20 μL min−1 by means of optical emission spectrometry using a simple portable spectrometer. Time-resolved determination of the hydrogen excitation temperature Tαβ indicates that the LE-DBD does not reach a stable state during a burning phase, whereat the maximum and minimum Tαβ is independent of the flow rate. Adding dissolved metals to the liquid electrode does not influence the minimum Tαβ at the end of a burning phase. With the help of measured doubly charged lanthanum lines and spatially resolved measurements, the mechanism of the liquid transfer into the plasma will be clarified. Emissions from metal oxides indicate a thermal evaporation transfer mechanism, but only an additional electrospray-like transfer mechanism can explain the observed La III emissions and nonhomogeneous spatial distribution of exited species. The reaction pathways for electrosprayed hydrated metal ions are discussed for triply and doubly charged ions. The analytical performance is evaluated for 23 elements from the categories of alkali, alkaline earth, transition, and poor metals. The achieved detection limits are between 0.016 mg L−1 for Li and 41 mg L−1 for Bi.

I

atomic emission lines of several metals dissolved in the liquid electrode were easily exited for excitation energies below 6 eV. The detection limits were between 2 and 4 μg L−1 for K, Li, and Na and between 80 and 100 μg L−1 for Ca, Cu, Mg, and Zn. A significant improvement is achieved by addition of nonionic surfactants. Greda et al. used a 0.6 mol L−1 HCl electrolyte solution with additional 4-(1,1,3,3-tetra-methylbutyl)phenylpoly(ethylene glycol)s.16 Using Triton X-405 with a concentration corresponding to 5 times of the critical micelle concentration (cmc) the intensities of molecular bands and the overall background were reduced and the intensities of atomic lines were strengthened. Achieved improvements of the detection limits were between 2 and 25. A discharge of the ELCAD type operated with alternating current was investigated by Huang et al. and called alternating current electrolyte atmospheric liquid discharge (ac-EALD).3,17 Compared to the ELCAD the ac-EALD operated at a lower flow rate in the range of 0.1−0.8 mL min−1 and achieved detection limits of 40 and 90 μg L−1 for Na and Cd, respectively. György et al. combined an ELCAD with atomic absorption spectrometry (AAS) and reported that the spatial distribution of atomic species along the plasma length is highly nonhomogeneous and element-dependent.18 Compared to the

n the recent years different miniaturized atmospheric pressure glow discharges (APGD) as excitation sources for optical emission spectrometry (OES) of dissolved analytes in liquids have been developed and investigated for analytical chemistry purposes.1−8 Advantages of these discharges are small sample sizes and low power consumption as well as reduced construction and operation costs. They are operating at atmospheric air pressure without need for any vacuum technique and often within ambient air without need of additional working gases. Cserfalvi et al. first reported on an APGD in contact with a liquid cathode, called electrolyte as a cathode discharge (ELCAD), applicable for analytical measurements.1 This discharge has been investigated in recent years by different groups.9−12 A modified version was reported by Webb et al. and called solution-cathode glow discharge (SCGD).2,13−15 Between a flowing electrolyte of some milliliters per minute as cathode and a metallic anode a dc voltage plasma is ignited and analyzed by optical emission spectrometry. Using a flow rate of 2.46 mL min−1 Webb et al. measured with their SCGD detection limits of heavy metals to be up to 3 orders of magnitude higher than those of alkali and alkaline earth metals with 0.06 and 0.2 μg L−1 for Li and Mg and 6 and 22 μg L−1 for Pb and Hg, respectively.13 A dc-APGD with 0.6 mol L−1 HCl electrolyte solution as liquid electrode and a flow rate of 0.6 mL min−1 was investigated by Jamróz et al.12 They reported that strong © 2014 American Chemical Society

Received: February 11, 2014 Accepted: May 15, 2014 Published: May 15, 2014 5822

dx.doi.org/10.1021/ac500583h | Anal. Chem. 2014, 86, 5822−5828

Analytical Chemistry

Article

of the liquid is performed by acidification with HNO3 to a concentration of 1 mol L−1. The tungsten electrode is connected to ground with a current limiting resistor of 100 Ω. Pulsed high voltage, provided by a homemade peak generator in the kilohertz range, is capacitively coupled into the liquid over a copper coating on the outside of the capillary. The wall of the capillary acts as dielectric barrier. Inside the capillary, the discharge is ignited between the liquid meniscus and the tungsten wire, whereby the liquid and tungsten wire form both electrodes of the discharge. For spectroscopic measurements the polyimide covering of the capillary is removed in the plasma burning region to form an observation window. The discharge ignition starts independently once the distance between the liquid meniscus and tungsten electrode is small enough. Shortly after ignition, the plasma is established with a typical trapezoidal shape and fills almost the entire space between the liquid meniscus and the tungsten electrode. While the discharge is ignited, more liquid is removed from the liquid boundary layer and transferred to the plasma as the constant flow rate of the liquid electrode delivers. This increases the space between the liquid meniscus and tungsten electrode, which also increases the plasma volume. Once the gap between the liquid and tungsten electrode gets too large, the plasma extinguishes. Through the constant flow rate of the liquid the gap between both electrodes decreases until the next ignition starts. This discharge pulsation results in a stabilization of the system, as gas bubbles in the liquid that arise during the burning time of the plasma are collected by the reverse flow of the liquid meniscus and do not cause a short circuit between both electrodes. If not specified otherwise the LE-DBD was operated with a high voltage of 3.2 kVpp, a generator frequency of 87 kHz, and a flow rate of 20 μL min−1. Under these conditions the length of a pulsation cycle is about 1.7 s, while the plasma burning time within a pulsation cycle is in the range of 0.5 s (cf. Figure 2 in ref 5). Spectroscopic Measurements. A portable spectrometer (Ocean Optics USB4000, 200−900 nm) was used successfully for time-resolved and analytical performance measurements. The emitted light from the plasma was collected by an optical fiber (P-600-SR, 200−1100 nm, Ocean Optics, FL, U.S.A.), where the bare end of the optical fiber was adjusted perpendicular to the capillary (see Figure 1). However, the spectral performance of compact spectrometers using linear detectors is strongly limited due to the low number of available

ELCAD-OES, a lower sensitivity of the ELCAD-AAS was observed and detection limits between 0.9 and 9.2 mg L−1 for Zn, Cd, Cu, and Na were achieved. Approaches with nonflowing liquid electrodes were investigated by several research groups.6−8,19 The liquid electrode spectral emission chip (LEd-SpEC)6 produces a glow discharge in air or moderate vacuum with liquid electrodes. Another approach is the liquid electrode plasma atomic emission spectrometry (LEP-AES)7 where both electrodes are the analyzed liquid solution. It is operated with pulsed dc voltage. Recently, Van Khoai et al. integrated a solid-phase extraction (SPE) column on an LEP chip for sample preconcentration and improved the sensitivity for lead detection by a factor of 2 compared to the LEP method.19 He et al. introduced a novel liquid film dielectric barrier discharge (LFDBD) for microsample elemental determination.8 A glass slide acts as dielectric barrier as well as sample plate between two electrodes, where a small amount of sample solution (≤80 μL) forms a thin liquid film on the glass surface. Achieved detection limits were between 7 and 79 μg L−1 for Na, K, Cu, Cd, and Zn. In our previous paper on the liquid electrode dielectric barrier discharge (LE-DBD) the analytical performance was demonstrated by calibration measurements of several metals, in particular, the alkali and alkaline earth metals.5 Generally, atomic emission lines of neutral elements are observed, whereas for alkaline earth metals additional emissions of single-ionized metal ions as well as metal oxides are detectable. The observed shapes of the calibration curves of metals as well as for Sr oxides are nonlinear. One explanation could be that at high analyte concentration the available energy for dissociation of metal oxides and excitation of analyte atoms is not sufficient. A more complex explanation for this behavior is that there are at least two mechanisms for the transfer of liquid into the plasma: a thermal evaporation and an electrospray-like process. On one hand, the appearance of alkaline earth oxides can be explained by thermal evaporation with the decomposition of nitrates reaction 2M(NO3)2 → 2MO + 4NO2 + O2. On the other hand, the observed emission from single-ionized alkaline metal ions can be caused by an electrospray-like process, where the sprayed alkaline earth metal ions M2+ capture in several steps an electron and excited M+ ions are formed. De-excitation of excited M+ ions occurs through radiative relaxation. The purpose of this work is to further characterize the LEDBD and to validate the proposed complex liquid transfer mechanism. A time-dependent hydrogen excitation temperature is investigated for different flows and concentrations. The hydrogen excitation temperature is calculated by the twotransition method for the Balmer lines. With the help of measured La III lines and spatially resolved measurements, we try to explain the complex mechanism of the liquid transfer into the plasma. Finally, the analytical performance of the present method is evaluated for 23 elements.

■

EXPERIMENTAL SETUP Plasma Discharge System. The used LE-DBD was described in detail elsewhere.5 We explain it briefly here. A polyimide-covered fused-silica capillary (i.d. 700 μm, o.d. 850 μm) is connected with a pulsation free syringe pump system (neMESYS, Centoni GmbH, Germany) which provides a constant liquid flow. At the downstream port of the capillary a tungsten wire with 400 μm diameter is inserted into the capillary to about a length of ∼900 μm; otherwise the port is open to ambient air. Adjustment of the electrical conductivity

Figure 1. Experimental arrangement for time-resolved and analytical performance measurements. 5823



Article

pixels. A novel echelle spectrograph developed at the Optical Spectrometry Group, ISAS Berlin, was used to investigate extended wavelength intervals with high spectral resolution. Echelle spectrographs focus a two-dimensional pattern of high diffraction orders on array detectors with millions of pixels. The special design of the compact ISAS echelle spectrograph is characterized by simple switching between two different spectral channels, one for the spectral range from 191 to 459 nm (UV−vis) and the other from 357 to 830 nm (vis−NIR). For optimum exploitation of the detector array, the shift from one channel to the other is realized only by rotation of a pivotable off-axis paraboloid working both as collimator and camera mirror. All other optical elements, the entrance slit, and detector position are fixed. In each channel spectra of more than 300 mm total length are detected by an integrated 1024 × 1024 pixel back-illuminated CCD camera (greateyes GmbH, Berlin, Germany). More technical details of the echelle spectrograph are given in the Supporting Information, Figure S2, and in general in refs 20 and 21. For the plasma investigations with high spectral and spatial resolution the LE-DBD was mounted on a xyz micrometer translation stage holder. An unmagnified (1:1) image of the discharge was collimated by means of a toroidal mirror (f = 300 mm) onto the echelle entrance slit (50 μm), cf. Supporting Information Figure S-1. Reagents and Samples. Single-element AAS stock solutions with an element concentration of 1 g L−1 purchased from Carl Roth (Germany) were used, except for lanthanum where a lanthanum stock solution (5 mg L−1) was prepared from lanthanum nitrate salt (Johnson Matthey). Samples were made by suitable dilution with deionized water (Millipore system) and acidified with concentrated nitric acid (Carl Roth, Germany). The final concentration of the samples were adjusted to 1.0 mol L−1 HNO3.

Figure 2. Typical evolution of the Hαβ excitation temperature Tαβ during a burning phase. A solution of 1.0 mol L−1 HNO3 with a flow rate of 40 μL min−1 was used as the liquid electrode. The dashed line is an exponential decay fit of the excitation temperature of the form T = T0 + A exp[−(t − t0)/τ].

conditions, the plasma burning time within a pulsation cycle was approximately 1.4 s. Shortly after ignition the maximum Tαβ was reached. After that, Tαβ dropped continuously and before a stable state was reached the plasma extinguished. The developing of Tαβ can be approximated with an exponential decay fit of the form T = T0 + A exp[−(t − t0)/τ] with a good agreement. In the case of using a flow rate of 40 μL min−1 the decay constant τ has a value of (0.5 ± 0.1) s and shows also that the plasma does not reach a steady state at the end of the burning time. The minimum and maximum T αβ are independent from the flow rate of the liquid electrode as Supporting Information Figure S-3a indicates. Due to the fact that the burning time is proportional to the flow rate of the liquid electrode and the maximum gap size at the end of the burning phase is independent of the flow rate, the decay is faster when the flow rate is lower. The decay constant τ can thus be interpreted as characteristic time of the system. For physical and analytical measurements this has the consequence that the LE-DBD with various flow rates are comparable if the characteristic time is taken into account for all measurement parameters which are time-dependent. A variation of the composition of the liquid electrode by adding diluted metals shows only minor effects to Tαβ. For this study silver and palladium, each with two different concentrations, were selected, because the emission spectra of both elements do not interfere with both hydrogen emission lines Hα and Hβ. The maximum Tαβ is reduced at higher concentrations of diluted metals in the liquid electrode, whereas the minimum Tαβ is not affected by the metals dissolved in the liquid electrode, cf. Supporting Information Figure S-3b. This means that the perturbation of the plasma through additional dissolved elements in the liquid electrode at the beginning of the burning phase is present. Nevertheless, at the end of the burning phase, i.e., at maximum extension of the plasma, the impact on the characteristics of the plasma is negligible. For analytical applications, this means that, despite of different compositions of the liquid electrode through additional analytes, the established plasmas are comparable and applicable for concentration determination when emission spectra are used which are measured at the end of the burning phase. High Spectral Resolution and Spatially Resolved Measurements. For spatially resolved and high spectral resolution measurements, an echelle spectrograph was used

■

RESULTS AND DISCUSSION Hydrogen Excitation Temperature. To describe the population of atomic excited states the hydrogen excitation temperature was estimated by means of the two-transition method. For the calculation the hydrogen Balmer lines at 656.46 nm (Hα) and 486.27 nm (Hβ) with the line data given in Table 1 were used, in the following denoted as Hαβ excitation temperature Tαβ. Further details on the calculation of the excitation temperature are given in the Supporting Information. Figure 2 shows the typical evolution of Tαβ during a burning phase, where a 1 mol L−1 HNO3 solution as liquid electrode with a flow rate of 40 μL min−1 and an integration time of 25 ms was used. Due to the higher flow rate as for standard Table 1. Used Emission Line Data for Calculating the Excitation Temperature of Hydrogen, Taken from Wiese and Fuhr (Ref 22)a transition λvac/nm Aul/108 s−1 gu Eu/eV

Hα (n: 2 ← 3)

Hβ (n: 2 ← 4)

656.464 4.410 × 10−1 18 12.0875

486.270 8.419 × 10−2 32 12.7485

a λvac denotes the vacuum wavelength, Aul is the transition probability (Einstein coefficient), gu is the statistical weight, and Eu is the energy of the upper state of the transition.

5824



Article

Figure 3. High-resolution spectra of 100 mg L−1 Sr in 1.0 mol L−1 HNO3 for both measurement ranges of the echelle spectrograph. Further parameters, 3.6 kVpp, 12 μL min−1, observing position 200 μm from the tungsten electrode in direction to the liquid. (a) Spectrum of the UV−vis range with the OH (A−X) band and the strontium ion lines. The inset shows that there is no emission from the N2 (C−B) band. (b) Spectrum of the vis−NIR range where neutral and single-ionized strontium, hydrogen, oxygen, and natrium impurities are detectable. The inset shows fully resolved neutral oxygen triplet at 777.2 nm.

Figure 4. Spatially resolved measurements with a high-resolution echelle spectrograph. Operating conditions: 87 kHz, 3.6 kVpp, 12 μL min−1, liquid electrode with 100 mg L−1 Sr in 1.0 mol L−1 HNO3, 60 s integration time. Origin of the distance is the end of the tungsten electrode; negative values are distances in direction to the liquid. (a) Spatial distribution of the most dominate emission lines, where the peak area is used as line signal and for each line the peak areas are normalized to the maximum line peak area. (b) Spatial distribution of several peak area ratios.

transition 5s2 1S0 ← 5s5p 1P°1 of the neutral strontium at 460.73 nm, the hydrogen Hα Balmer line at 656.28 nm, and the oxygen triplet at 777.2 nm (3s 5S2° ← 3p 5P{1,2,3} transitions). The slight increase of the ground line around 610 and 675 nm is from strontium oxide emissions. Other than by measurements with the low-resolution USB4000 spectrometer no emission from the Hβ Balmer line at 486.14 nm was observable. The wavelength range between 340 and 365 nm is shown in the inset in Figure 3a where usually emissions of the molecular nitrogen second positive system [transition N2 (B 3Πg+) ← N2 (C 3Πu+)] are observable if plasmas are operated under ambient air. In the case of the LE-DBD, no emission from molecular nitrogen was detectable. This behavior was also mentioned in the previous work.5,23 Due to the transfer of liquid from the liquid boundary layer into the plasma the ambient air in the capillary is extruded from the inner capillary shortly after ignition. The plasma is running in a self-generated water vapor and OH atmosphere (cf. Mezei and Cserfalvi24 for the same behavior as for the ELCAD).

since it gives more information about the processes in different regions of the discharge. Because the intensity emitted by the discharge is too less for spatially resolved measurements with small integration time, an integration time in the range of 1 min was chosen. With this the emission intensity is high enough for evaluation. Due to the pulsation of the discharge (cf. Figure 2 in ref 5) it is clear that this method has the disadvantage that only integrated line signals of plasma developing and several discharge pulsations were received. Figure 3 shows high-resolution spectra in the near tungsten electrode region, where as liquid electrode a 100 mg L−1 Sr in 1.0 mol L−1 HNO3 solution was used. The UV spectrum in Figure 3a shows emissions from the OH (A−X) band between 280 and 330 nm and strontium ionic lines, mainly the transition 5s 2S1/2 ← 5p 2P°3/2 and 5s 2S1/2 ← 5p 2P°1/2 at 407.77 and 421.55 nm, respectively. The line at 347.49 nm belongs also to ° ← 4p65d single-ionized strontium (transition 4p65d 2P3/2 2 D3/2). In the visible range shown in Figure 3b the prominent lines are both strontium ionic transitions, emission from the 5825



Article

the transitions is at 1.685 eV; the upper levels are at 5.594 and 5.210 eV, respectively.25 Evaporation of lanthanum follows in forming the monoxide LaO, not the sesquioxide La2O3.26 A pathway for exciting La2+ over thermal evaporation is given in Figure 6. Due to the

The spatial distribution of several lines in Figure 4a shows that the plasma composition changes over the plasma length. Indeed due to the pulsation of the discharge, the effective measuring time of emissions from the plasma decreased for measuring points that were further away from the tungsten electrode. However, this contribution to the measured spatial distribution was identical for all emission lines and could not cause the different distributions. Starting from the liquid electrode, the order of maximum relative contribution to the whole emission intensity will be from hydroxyl radical, then oxygen and hydrogen, followed by neutral metals and ionized metals if applicable. Figure 4b shows the spatial distribution of several line signal ratios. The ratio between neutral strontium and hydrogen emission shows a linear increase in direction to the tungsten electrode. Ratios between single-ionized strontium and hydrogen or neutral strontium emission, respectively, follow a nonlinear distribution. First, the increase is in both cases higher than between neutral strontium and hydrogen, and second, the increase is overproportional and follows more an exponential or power distribution. A discussion of this is given in the next section. Mechanism for Transfer of Liquid into the Plasma. In our previous paper5 we proposed a preliminary model for the transfer of solution from the liquid electrode into the discharge over two pathways: a thermal evaporation and an electrospraylike process. For discharges of the ELCAD type the proposed transfer mechanisms are cathodic sputtering, thermal desolvation, and electrospray (cf. Schwartz et al.15). In contrast to the ELCAD system, cathodic sputtering cannot be taken as a primary contributor for the transfer mechanism in the LE-DBD system due to the positive polarity of the liquid electrode. In order to confirm the electrospray-like explanation,5 highresolution measurements with lanthanum solution were performed. In addition to emissions from neutral and singly ionized lanthanum, emission lines from doubly ionized lanthanum were visible, see Figure 5. The both La III lines at ° and at 317.2 nm are from the transition 6s 2S1/2 ← 6p 2P3/2 ° . The lower energy level for 351.7 nm from 6s 2S1/2 ← 6p 2P1/2

Figure 6. Pathway for exciting La2+ over thermal evaporation (modified from ref 27; energies in eV from refs 27 and 28).

relatively high energy needed for dissociation of LaO and LaO+ as well as the high ionization energy for La+ the probability for production of doubly ionized lanthanum ions from thermal evaporation is negligible in a plasma burning in water vapor and hydroxyl atmospheres (quenching through energy transfer to excite the OH (A−X) band). Only direct transfer of ionized atoms or molecules from the liquid into the plasma, namely, electrospray-like, can explain the observed La III emission lines. Blades et al. showed that with electrospray and mass spectrometry triple-charged ions M3+ (M = Y, La, Ce, Nd, Sm) in water solutions led only to charge reduced ions MOH(H2O)n2+ in the gas phase.29 In addition to the charge reduced ions Bush et al. found for the trivalent metal ions La, Ce, and Eu formation of triple-charged ion clusters M(H2O)n3+, but only for n ≥ 16.30 Subsequent dehydration (eqs 1a and 1c) and charge reduction (eq 1b) lead to MOH2+ which is further dissociated to a doubly ionized metal ion M2+ (eq 1d) and exited. For lanthanum, the dissociation energy for the reaction LaOH2+ → La2+ + OH is given by Schröder et al.31 with D0 = 5.7 eV. M(H 2O)n ≥ 16 + k 3 + → M(H 2O)n − k 3 + + (H 2O)k

(1a)

M(H 2O)n ≥ 16 3 + → MOH(H 2O)n − (k + 2)2 + + H3O(H 2O)k + (1b)

MOH(H 2O)n 2 + → MOH(H 2O)n − k 2 + + (H 2O)k

(1c)

MOH2 + → M2 + + OH

(1d)

A similar pathway for doubly charged alkaline earth and transition metal ions in aqueous solutions is possible32 and given in eqs 2a−2d. Thereby, the hydration number r, where charge reduction reactions (eq 2b) through collision-induced dissociation (CID) occur, depends on the doubly charged metal ion hydrate. These hydration numbers were given by Peschke et al. for several elements with Mg (3), Ca (2), Sr (2), Ba (0), Mn (3), Co (4), and Zn (5).33 M(H 2O)n 2 + → M(H 2O)n − k 2 + + (H 2O)k

(2a)

M(H 2O)r 2 + → MOH(H 2O)r − (k + 2)+ + H3O(H 2O)k + (2b) +

+

MOH(H 2O)n → MOH(H 2O)n − k + (H 2O)k +

+

MOH → M + OH

(2c) (2d)

The electrospray-like contribution of the transfer mechanism explains also the spatial distribution and line intensity ratios in Figure 4. Positively charged particles, which are transferred directly from the liquid in the plasma, are accelerated to a

Figure 5. Lanthanum emission lines measured with a high-resolution echelle spectrograph. Further parameters, 3.2 kVpp, 20 μL min−1, observing position 200 μm from the tungsten electrode in direction to the liquid, 32 s integration time. 5826



Article

greater extend by the electric field in direction to the tungsten electrode, as compared to charged particles, which are produced in the plasma. This results in a density gradient of charged metal ions in direction to the tungsten electrode and in an increase of the emission line ratios between singly ionized and neutral strontium or hydrogen. The increase of the ratio between neutral strontium and hydrogen is explainable by recombination of ionized strontium. Analytical Performance. The analytical performance of the LE-DBD was evaluated for 23 dissolved elements under optimal operating conditions (flow rate, 20 μL min−1; solution, 1 mol L−1 HNO3; high voltage, 3.2 kVpp; generator frequency, 87 kHz) (Table 2). For each element 10 concentrations were

atomic emission lines. The very intense single-ionized emission line from Mg at 279.5 nm could not be used due to overlapping with emission from OH. For estimation of the detection limits (LOD) the equation IL(λ) = IB(λ) + kσB(λ) was used, where λ denotes the emission wavelength, IL is the emission intensity at the detection limit, IB is the intensity of the blank emission, σB is the standard deviation of the emission, and k is a constant factor which is set to k = 3.34,35 The calibration curve for each elemental emission line was fitted with a power function model of the form I = IL + A|c − cL|p, where c and cL denote the concentration and the detection limit, respectively. The fit parameter p lay between 0.58 for Ag and 1.16 for In, where for 17 elements 0.8 < p < 1.1 was determined. Values of p > 1 mean that the response of the LE-DBD on concentration changes is overproportional. A reason for this can be preconcentration of the analyte in the plasma and accumulation on the inner capillary or a too strong perturbation of the plasma at high loads which impacts excitation mechanisms, energy transfers, and plasma composition.

Table 2. Analytical Performance of the LE-DBD for Several Metals Dissolved in 1 mol L−1 Nitric Acid and a Flow Rate of the Liquid Electrode of 20 μL min−1a detection limit element

emission wavelength/nm

measd concn range/mg L−1

no. of concns

mg L−1

ppm

Ag Au Ba Bi Ca Cd Co Cr Cs Cu Fe Hg In K Li Mg Mn Ni Pb Pd Sr Tl Zn

338.3 267.6 493.4 472.3 393.4 508.6 345.4 359.0 852.1 521.8 374.6 253.7 451.1 766.5 670.8 518.4 403.1 352.5 405.8 363.5 407.8 535.1 481.1

5−100 2−900 5−900 40−950 1−200 1−950 5−950 5−500 1−70 5−950 2−200 20−200 0.2−10 0.01−10 0.02−3 1−200 5−500 2−500 5−200 2−400 2−200 0.2−40 20−950

10 9 9 10 8 6 6 9 9 7 7 8 6 7 10 8 7 10 6 9 9 7 10

0.39 34 9.6 41 0.83 13 31 5.7 1.18 18 5.4 11 0.16 0.044 0.016 0.94 6.5 2.8 2.6 1.5 2.3 0.24 18

0.072 3.5 1.4 3.9 0.41 2.3 11 2.2 0.18 5.7 1.9 1.1 0.028 0.023 0.046 0.77 2.4 1.0 0.25 0.28 0.52 0.023 5.5

■

CONCLUSION Measurements of the hydrogen excitation temperature Tαβ indicate that the LE-DBD does not reach a steady state during the burning phase. The time dependency of Tαβ can be described well by an exponential decay function with a characteristic time τ, which depends on the flow rate of and analytes dissolved in the liquid electrode. The LE-DBD with various flow rates is only comparable for physical and analytical measurements if the characteristic time is taken into account for time-dependent parameters. However, Tαβ at the end of a burning phase is independent of the flow rate and additional analytes dissolved in the liquid electrode. For comparable conditions by analytical measurements emission spectra should be used which are measured at the end of the burning phase. High-resolution measurements confirm that emissions from molecular nitrogen are not observable, even though operating under ambient air. The LE-DBD is running in a self-generated water vapor and OH atmosphere and shows the same behavior as Mezei and Cserfalvi24 reported for the ELCAD. The spatial distribution of the composition of different species over the length of the LE-DBD plasma is highly nonhomogeneous and analyte-dependent. From the liquid electrode in direction to the tungsten electrode, the order of maximum relative contribution to the whole emission intensity is from hydroxyl radical, oxygen and hydrogen, neutral metals, and if applicable, ionized metals. In addition to thermal evaporation5,23 it can be confirmed that an electrospray-like process is the second main pathway for the transfer mechanism of liquid from the liquid electrode into the plasma. Cathodic sputtering, as discussed as transfer mechanism for the discharges of the ELCAD type, cannot be contributing primarily to the transfer mechanism for the LEDBD through the positive polarity of the liquid electrode. The LE-DBD is found to be a well applicable technique for detection of a width range of dissolved elements in flow systems. Analytical performance measurements were performed for 23 elements from the categories of alkali, alkaline earth, transition, and poor metals. The achieved detection limits are between 0.016 mg L−1 for Li and 41 mg L−1 for Bi. These are 2−3 orders of magnitude higher compared to systems of the ELCAD type. However, the LE-DBD operates with a 1−2 order lower flow rate than other systems with a flowing liquid electrode and requires a smaller sample volume. Taking into

a

The detection limit in ppm means the mole fraction of the analyte in the liquid.

prepared, whereby the measurement of the calibration curve for an element started with a blank sample (only 1 mol L−1 HNO3) and in ascending concentration order. A new capillary was used for each element. Emission spectra were collected continuously for 45 s with an integration time of 100 ms, except for Ag, Co, and Ni with 25 ms and for Pd with 50 ms. A self-written software determined all intensity maxima per cycle for a certain emission line and concentration. Afterward, outliers were detected through a Hampel test34 and eliminated. The averaged intensity maxima, normalized to the maximum scale of the spectrometer, were used as intensity signal for a certain emission wavelength. Normally, an emission line from neutral atoms was used to evaluate the analytical performance. For the alkaline earth metals Ba, Ca, and Sr the strongest emission lines from single-ionized atoms were used, because by using these lines a lower detection limit was reached as by using neutral 5827



Article

account the flow rate and measurement time, the absolute detection limits of the LE-DBD and ELCAD systems are in the same order of magnitude, e.g., for Li 0.7 versus 0.3 ng and for Hg 17 versus 92 ng, respectively.13 Additionally, the analytical performance measurements are performed with a portable lowresolution spectrometer. Advantages of the LE-DBD compared to most other systems with a flowing liquid electrode are the small sample volume, the simple and compact arrangement, and no liquid waste production. Together with the simple spectrometer, this reduces costs and allows the application as or integration into portable analysis systems, coupling with microseparation devices or the usage as a simple monitoring device in technical flow systems. An improvement of the LEDBD sensitivity is possible by using a spectrometer with higher spectral resolution. The pulsation of the plasma is a limiting factor of the LE-DBD. On the one side it makes the measurement process more complex and limits the usage of high-resolution spectrometers. Otherwise it reduced the time resolution by utilization as monitoring device.

■

(9) Mezei, P.; Cserfalvi, T.; Jánossy, M. J. Anal. At. Spectrom. 1997, 12, 1203−1208. (10) Mottaleb, M. A.; Woo, Y.-A.; Kim, H.-J. Microchem. J. 2001, 69, 219−230. (11) Jenkins, G.; Franzke, J.; Manz, A. Lab Chip 2005, 5, 711. (12) Jamróz, P.; Pohl, P.; Ż yrnicki, W. J. Anal. At. Spectrom. 2012, 27, 1032−1037. (13) Webb, M. R.; Andrade, F. J.; Hieftje, G. M. Anal. Chem. 2007, 79, 7899−7905. (14) Webb, M. R.; Hieftje, G. M. Anal. Chem. 2009, 81, 862−867. (15) Schwartz, A. J.; Ray, S. J.; Elish, E.; Storey, A. P.; Rubinshtein, A. A.; Chan, G. C.-Y.; Pfeuffer, K. P.; Hieftje, G. M. Talanta 2012, 102, 26−33. (16) Greda, K.; Jamróz, P.; Pohl, P. Talanta 2013, 108, 74−82. (17) Xiao, Q.; Zhu, Z.; Zheng, H.; He, H.; Huang, C.; Hu, S. Talanta 2013, 106, 144−149. (18) György, K.; Bencs, L.; Mezei, P.; Cserfalvi, T. Spectrochim. Acta, Part B 2012, 77, 52−57. (19) Van Khoai, D.; Kitano, A.; Yamamoto, T.; Ukita, Y.; Takamura, Y. Microelectron. Eng. 2013, 111, 343−347. (20) Becker-Ross, H.; Florek, S. V. Spectrochim. Acta, Part B 1997, 52, 1367−1375. (21) Becker-Ross, H.; Florek, S.; Okruss, M. Spektrometeranordnung. German Patent DE 10 2007 028 505 A1, 2008. (22) Wiese, W. L.; Fuhr, J. R. J. Phys. Chem. Ref. Data 2009, 38, 565− 719. (23) Krähling, T.; Müller, S.; Groh, S.; Franzke, J. Detection of dissolved metals using a liquid electrode dielectric barrier discharge. Presented at the 30th International Conference on Phenomena in Ionized Gases (ICPIG), Belfast, Northern Ireland, U.K., 2011. (24) Mezei, P.; Cserfalvi, T. Sensors 2012, 12, 6576−6586. (25) Odabasi, H. J. Opt. Soc. Am. 1967, 57, 1459−1463. (26) Patterson, P. M.; Goodings, J. M. Int. J. Mass Spectrom. Ion Processes 1995, 148, 55−76. (27) Ackermann, R. J.; Rauh, E. G.; Thorn, R. J. J. Chem. Phys. 1976, 65, 1027−1031. (28) Kramida, A.; Ralchenko, Y.; Reader, J.; NIST ASD Team. NIST Atomic Spectra Database, ver. 5.1 [Online]; National Institute of Standards and Technology: Gaithersburg, MD, 2013; http://physics. nist.gov/asd. (29) Blades, A. T.; Jayaweera, P.; Ikonomou, M. G.; Kebarle, P. Int. J. Mass Spectrom. Ion Processes 1990, 101, 325−336. (30) Bush, M. F.; Saykally, R. J.; Williams, E. R. Int. J. Mass Spectrom. 2006, 253, 256−262. (31) Schröder, D.; Schwarz, H.; Harvey, J. N. J. Phys. Chem. A 2000, 104, 11257−11260. (32) Blades, A. T.; Jayaweera, P.; Ikonomou, M. G.; Kebarle, P. Int. J. Mass Spectrom. Ion Processes 1990, 102, 251−267. (33) Peschke, M.; Blades, A. T.; Kebarle, P. Int. J. Mass Spectrom. 1999, 185−187, 685−699. (34) Mermet, J.-M. Limit of Quantitation in Atomic Spectrometry Concepts and Definitions. In Encyclopedia of Analytical Chemistry; Meyers, R. A., Ed.; John Wiley & Sons, Ltd.: Hoboken, NJ, 2009. (35) Kaiser, H. Fresenius’ Z. Anal. Chem. 1965, 209, 1−18.

ASSOCIATED CONTENT

S Supporting Information *

Experimental setup for spectral high-resolution measurements, calculation method of the hydrogen excitation temperature, and further details of the Tαβ measurements by liquid flow and composition variation. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Present Address §

T.K.: University of Mü nster, Department of Clinical Radiology, Albert-Schweitzer-Campus 1, D-48149 Münster, Germany.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The financial support by the Ministerium für Innovation, Wissenschaft, Forschung und Technologie des Landes Nordrhein-Westfalen, by the Bundesministerium für Bildung und Forschung, and by the Deutsche Forschungsgemeinschaft is gratefully acknowledged.

■

REFERENCES

(1) Cserfalvi, T.; Mezei, P.; Apai, P. J. Phys. D: Appl. Phys. 1993, 26, 2184−2188. (2) Webb, M. R.; Andrade, F. J.; Hieftje, G. M. Anal. Chem. 2007, 79, 7807−7812. (3) Huang, R.; Zhu, Z.; Zheng, H.; Liu, Z.; Zhang, S.; Hu, S. J. Anal. At. Spectrom. 2011, 26, 1178−1182. (4) Tombrink, S.; Müller, S.; Heming, R.; Michels, A.; Lampen, P.; Franzke, J. Anal. Bioanal. Chem. 2010, 397, 2917−2922. (5) Krähling, T.; Müller, S.; Meyer, C.; Stark, A.-K.; Franzke, J. J. Anal. At. Spectrom. 2011, 26, 1974−1978. (6) Wilson, C. G.; Gianchandani, Y. B. IEEE Trans. Electron Devices 2002, 49, 2317−2322. (7) Banno, M.; Tamiya, E.; Takamura, Y. Anal. Chim. Acta 2009, 634, 153−157. (8) He, Q.; Zhu, Z.; Hu, S.; Zheng, H.; Jin, L. Anal. Chem. 2012, 84, 4179−4184. 5828


Investigations into Modeling and Further Estimation of Detection Limits

Recommend Documents