SOLUTION AS CATHODE GLOW DISCHARGE: DESIGN IMPROVEMENTS AND MATRIX MODIFIERS. Todd A. Doroski

Transcription

1 SOLUTION AS CATHODE GLOW DISCHARGE: DESIGN IMPROVEMENTS AND MATRIX MODIFIERS Todd A. Doroski A Thesis Submitted to the University of North Carolina Wilmington in Partial Fulfillment of the Requirements for the Degree of Master of Science Department of Chemistry and Biochemistry University of North Carolina Wilmington 2011 Approved by Advisory Committee S. Bart Jones Stephen A. Skrabal Michael R. Webb Chair Accepted by Dean, Graduate School

2 TABLE OF CONTENTS ABSTRACT...v ACKNOWLEDGMENTS... vi LIST OF FIGURES...vii LIST OF TABLES....viii CHAPTER ONE 1.1 Introduction History Solution as Cathode Glow Discharge General Procedures Reagents Instrumentation Results and discussion Pump Calibration SCGD Cell Delivery System Spectrometer Optimization Interferences Quantification Conclusion...27 CHAPTER TWO 2.1 Introduction General Procedures and Reagents Origins Results and Discussion Mercury Enhancement Factors Detection Limits Conclusion...49 FUTURE WORK...51 iii

3 REFERENCES...52 APPENDIX...54 iv

4 ABSTRACT The solution as cathode glow discharge (SCGD) is emerging as a competitive instrument for the elemental analysis of aqueous samples. A new version of the SCGD described in this thesis does not require a waste reservoir for operation, which allows for a simpler collection of waste and a more compact cell. The delivery system uses a supplemental flow for the in-line addition of reagents. This simplifies the sample and solution preparations. The use of a spectrograph to measure the emission from the discharge plasma of the SCGD allows more accurate corrections and multiple element detection. The design improvements favor a more rugged and portable instrument. Detection limits of Ag, Cu, Fe, Hg, Mg, Ni, Pb, and Se were 0.5, 5, 20, 20, 2, 4, 10, and 3000 parts per billion respectively. These limits are comparable to those of previous SCGD designs. The effect of low molecular weight organic compounds on the emission signal was investigated. The addition of formic acid, acetic acid, or alcohol to the sample matrix yielded major signal effects for certain elements. Two reasons for effects are established and characterized for As, Ag, Ca, Cu, Fe, Hg, Mg, Ni, Pb and Se. There is strong evidence that correlates the enhancement effects to vapor generation. The easily vaporized elements (Ag, Hg, Pb, and Se) showed greater enhancement effects that could be exploited to improve detection limits. The detection limits with a 4% formic acid matrix for Ag, Cu, Fe, Hg, Mg, Ni, Pb, and Se were 0.1, 6, 14, 2, 1.4, 6, 1, and 250 parts per billion respectively. v

5 ACKNOWLEDGMENTS I would like to first thank Dr. Michael Webb, for his constant guidance and insight on any problem I approached with. I would also like to thank Dr. S. Bart Jones and Dr. Stephen Skrabal for their continuous support and serving on my committee, and Dr. Jeremy Morgan who showed me as an undergraduate, that graduate school was the path to take. Thanks to all of the members of the Webb Research Laboratory, who made doing research fun. Thanks to the entire UNCW Chemistry faculty. Thanks to Mom for listening to things you didn t understand and Dad for the different perspective. Scotty for the then do it attitude. Thanks especially to Aly who kept me sane throughout this entire process and putting up with everything. vi

6 Figure LIST OF FIGURES Page 1. Characteristics of direct current glow discharge Schematic diagram of a SCGD cell A schematic representation of the sample delivery system An image of the SCGD cell with a black dot highlighting the analysis area A background spectrum from nm Relative standard deviation as a function of capillary height Emission spectrum from nm Emission spectrum from nm and nm A picture and the negative of the SCGD plasma An irradiation profile of Hg A concentration profile of Hg with formic acid The concentration profiles of Hg with acetic acid or ethanol Normalized concentration profiles for the readily vaporized, transition and alkaline earth metals vii

7 Table LIST OF TABLES Page 1. Elements and the concentration of the standard solutions used to create the calibration curves The wavelengths, integration times, detection limits, R 2 values, and the relative standard deviation for each element Detection limits (in ppb) from a number of sources Elements and standards concentration The enhancement effects for all elements with a 7% formic acid in a HCl or HNO 3 matrix The enhancement effects for all elements with a HNO 3 and 4% LMWOC matrix The standard concentrations, R 2, the relative standard, and enhanced detection limits...48 viii

8 CHAPTER ONE 1.1 Introduction Heavy metals are toxic even in small quantities, so it is important to accurately test for them. There are numerous methods for doing so, including atomic absorption spectroscopy (AAS), atomic fluorescence spectroscopy (AFS), mass spectrometry (MS) and atomic emission spectroscopy (AES). A similarity between all techniques is the use of analytical instruments. When multiple elements are involved, AES and MS allow for the fastest analysis times due to their ability to analyze many elements at once. In general, AES instruments have lower costs associated with their purchase and operation then MS instruments. Although AES instruments generally produce higher detection limits than MS instruments, performance with AES instruments is frequently sufficient to analyze for heavy metals at relevant concentrations. An understanding of the fundamental principles of AES is required to make instrument improvements. Quantized energy levels give every element a unique energy signature. After an atom is excited, it decays by emitting a photon whose wavelength depends on the atom s energy levels. AES measures the light emitted by analyte atoms. The wavelengths of the light are used to identify the elements and the intensities of the light are used to measure the amounts of those elements. AES instruments must atomize and excite analytes to successfully quantify them in a sample. The atomization and excitation phases often require different components, causing instrument complexity and cost to increase. Most commonly, trace metal analyses are performed on a sample with a liquid matrix, or a solid sample that is first dissolved in a liquid matrix. The solution is then atomized. The term atomization is used here to refer to all steps involved in producing an atom freed of its matrix. Often more than one mechanism or step is involved in

9 creating the freed atom. Mechanisms include creating an aerosol, electrothermal atomization, and desolvation of an aerosol. 1 These atoms are transported to an excitation source which pumps energy into the freed atoms. Often, the final steps of atomization occur in the excitation source. The high energy atoms release their energy as light. The emitted light is quantified and the intensity at a particular wavelength is proportional to the amount of atoms of a particular element. Numerous heavy metals are often present at one time in very low quantities. One of the benefits of AES is that it can be used for simultaneous detection of multiple elements. An instrument that can detect multiple metals at a time will cut down on analysis time and has the ability to incorporate an internal standard. A common instrument for the analysis of trace metals is inductively coupled plasma (ICP)-AES. The Environmental Protection Agency (EPA) has a certified method for using an ICP-AES to determine concentrations of metals and trace element in water. ICP-AES has ability to detect concentrations of metals lower then the EPA requirements for safe drinking water for many elements. An ICP-AES instrument typically costs less than seventy thousand dollars, but costs are higher if the instrument is capable of simultaneous multielemental analysis. It has a relatively large footprint, and it requires a constant supply of argon gas and a large amount of power. These factors increase costs and limit ICPs to being immobile lab instruments. An emerging instrument for heavy metal analysis is a solution-cathode glow discharge (SCGD). As an AES source, the SCGD has comparable detection limits to those of the ICP- AES. 2 A SCGD costs about twelve thousand dollars to build and does not require argon to operate. This makes the SCGD-AES a much more economical instrument. Another benefit is the compact size and rugged design. The ruggedness could eventual allow for a portable instrument. 2

10 The first chapter of this thesis describes instrument designs and method developments of the SCGD that are a step forward in creating a useful field optical emission instrument. 1.2 History The idea behind any direct current discharge is that an electric potential is applied across two electrodes. Any direct current discharge can be characterized by its current (i) and voltage (V) relationship. Figure 1 shows the i-v relationship over a range of currents. 4 No units are given because the values vary with geometry and pressure. A normal glow discharge operates at a nearly constant voltage where increasing the current has little effect on the characteristics of the discharge except for the discharge size. At higher currents, the voltage becomes proportional to the current and an abnormal discharge is obtained. The abnormal discharge is used in atomic spectroscopy due to the high current density and the ability to stably control the voltage by adjusting the current. 4 At higher and lower currents, other types of direct-current discharges are obtained. With any glow discharge the difference in the potenital applied across two electrodes causes a split between the positive ions and electrons of a gas which creates the plasma. The cathode draws the positive ions towards it and causes collisions which eject electrons and atoms from the cathode surface. These electrons are drawn towards the anode and must travel through the plasma where two types of collisions occur. 3 The first type of collision is the two-step excitation collision. The first step, the collision between an electron and an atom or an ion, causes an increase in the energy of the atom or ion. The second step is the radiative decay, where the energy is lost due to the emission of a photon, causing the glow of the discharge. The second type of collision is the ionization collision. This 3

11 Figure 1. Characteristics of direct current glow discharge 4

12 collision creates new positive ion-electron pairs, and the positive ions are again drawn towards the cathode. The process repeats itself making the glow discharge a self-sustaining plasma in an electric field. This field causes the electrons to travel towards the anode and the cations to travel towards the cathode. Atoms that are struck by cations and atoms that are formed by an ionelectron recombination will also travel towards the cathode. In a traditional glow discharge, the cathode is often a solid that is also the sample. A process called sputtering occurs when atoms and ions collide with the cathode. These collisions release electrons and atoms of the cathode material into the plasma. The cathode material atoms in the plasma are excited by collisions of the first type and subsequently emit light and can be analyzed by AES. 3 A glow discharge plasma consists of different sections which often include the cathode dark space, negative glow, Faraday dark space, and positive column. The intensities and sizes of the regions depend on the set parameters. Traditional glow discharges operate at low pressure (around 10-3 atm) 5 and relatively low temperature (around 500 K). 6 These conditions do not effectively desolvate aqueous samples, which is a necessary first step in atomizing these samples. 5 In order to analyze aqueous samples using a traditional direct current glow discharge, the sample must be completely desolvated before it enters the plasma. Two methods that have been used for this purpose are drying the liquid atop the cathode surface or using a particle beam device to perform the desolvation on a flowing sample Solution as Cathode Glow Discharge More recently developed glow discharge methods that can directly analyze aqueous samples have been referred to as electrolyte cathode discharges (ELCAD) 7,8,9,10 or solution- 5

13 cathode glow discharges (SCGD). 7 The SCGD is characterized as an abnormal glow discharge, and the basic concept is that the plasma is created between a solid anode and a liquid cathode. The anode is positioned over a capillary with an overflowing electrolyte solution. The overflow is grounded, and the solution acts as the cathode. Successful atomization and excitation of the analyte atoms are required to have a functional AES source. 11 The ICP requires a nebulizer and spray chamber achieve atomization. The nebulizer creates an aerosol, and the spray chamber only allows the finest droplets of the aerosol to reach the plasma. The combination of a nebulizer and spray chamber has poor efficiency and allows less than 2% of the sample to be detected. 11 The equipment needed to transfer elements into the discharge in a SCGD is simple compared to an ICP. The excitation source needs to have a stable background with enough energy to successfully excite all atoms. The variables that affect the SCGD plasma include the flow rate, the ph and conductivity of the solution, the gap size (distance between the sample surface and the anode), and the current applied. 7,8,9,10,12 Under optimum conditions, the SCGD atomizes and excites analyte atoms. Removing the dependence of nebulizers or spray chambers allows for a more compact device and simplifies the atomization process. The following chapter will introduce a more rugged and compact SCGD cell design which has advantages that apply favorably to a portable instrument. It will, for the first time, use a commercially available compact spectrograph for direct analysis of the plasmas emission. The performance of a low-resolution spectrometer that allows for simultaneous element detection will be evaluated. 6

14 1.4 General Procedures Nitrile glove were worn at all times. All containers were rinsed three times with Milli-Q water prior to use. All solutions were stored in HDPE bottles that were washed three times with Milli-Q water and once with the solution to be stored. Because corrosive acids and toxic metals were used, the SCGD was operated in a fume hood to ensure adequate ventilation. 1.5 Reagents Purified water (18.2 MΩ cm) from a Millipore Q-water system (Millipore Corp., Bedford, MA) was used for the preparation of all solutions and samples. Optima nitric acid purchased from Fisher Scientific (Fair Lawn, NJ), was used to make a 0.25 M solution. Separate stock solutions of 1000 ppm Ag, Cu, Fe, Hg, Mg, Ni, Pb, and Se (BDH Aristar Plus, VWR International, West Chester, PA) were used to create all standards. The dilute standards were prepared fresh daily. Solution and Standard preparation 0.25 M nitric acid - About 100 ml of Milli-Q water was added to a 1000 ml volumetric flask ml concentrated Optima nitric acid was added to the flask. The flask was filled to ml with Milli-Q water. Standards Stock solution concentration varied by element. For each element, three different standard solutions were prepared in washed volumetric flasks by diluting the stock solution with Milli-Q water. The three standard solution concentrations for each element used to create the calibration curves can be seen in Table 1. 7

15 Standard Solution Element Concentration Units Ag 50, 250, 500 ppb Cu 250, 500, 1000 ppb Fe 50, 250, 500 ppb Hg 250, 500, 1000 ppb Mg 10, 100, 250 ppb Ni 100, 500, 1000 ppb Pb 250, 500, 1000 ppb Se 10, 25, 50 ppm Table 1. Elements and the concentration of the standard solutions used to create the calibration curves. 8

16 1.6 Instrumentation The SCGD cell is shown in Figure 2. An MR1.5P200L high voltage power supply, from Glassman High Voltage Inc. (High Bridge, NJ), was set to operate with a constant voltage. Constant voltage operation has previously been found to be more resistant to matrix interferences. 18 Two home built digital meters that gave readouts for current and voltage were calibrated. The high voltage power supply was used to apply a positive potential through a 1 kω ballast resistor. The resistor stabilized current that was passed to a 2-mm diameter tungsten anode with a rounded tip. Directly under the anode was a capillary (0.6-mm inner diameter and 1.3-mm outer diameter) that had been inserted radially through a grounded graphite electrode (6- mm diameter). The capillary was attached to a delivery system with a PEEK fitting and PTFE tubing from Cole-Parmer (Vernon Hills, Illinois). Based on previous studies, a discharge gap of 3-mm was chosen. 2,18 Optimum capillary height was when 3-mm protruded above the graphite electrode. For all experiments, operating conditions were 70 milliamps and 1000 volts. These values were chosen based on previous results. 2 The sample delivery system is shown in Figure 3. All solutions were transported using a Perimax-16 antipulse peristaltic pump from Spetec. Two sets of 2-stop pump tubing from Meinhard Glass Products (Golden, CO) were used to create antipulse tubing. The tubes were on offset rollers and their flows were combined into a single tube, which decreased the fluid s pulsing. The individual tubes were calibrated to matching flow rates before combining. The pump had the ability to simultaneously run 3 anti-pulse channels. Two solutions were used to deliver the sample to the discharge cell: a carrier solution used to move sample and an electrolyte solution to provide an electrolyte and maintain a stable discharge. For all experiments, anti-pulse tubing was used in the carrier and electrolyte channels respectively. The carrier solution was pumped into a V-450 two-position six-port valve 9

17 Figure 2. schematic diagram of a SCGD cell consisting of a tungsten anode above a glass capillary where the solution overflow completes the electrical connection to a grounded electrode. 10

18 Figure 3. A schematic representation of the sample delivery system. The carrier and electrolyte solutions are pumped with a peristaltic pump through homemade pulse dampeners towards the SCGD cell. The carrier solution moved through the injecting valve. The sample loop is filled using a syringe. When the valve was switched, solution exited the valve and mixed with the electrolyte solution before entering the SCGD cell. 11

19 from Idex Health and Science (Oak Harbor, WA). The injection valve was set up with a 3 ml PTFE sample loop as seen in Figure 3. All samples were injected with a syringe. A total volume of at least 10 milliliters was used to flush and fill the sample loop. The solution exiting the valve was mixed via a Y-connector with the electrolyte solution. To decrease pulsing, two home-built pulse dampeners consisting of a length of tubing (5 ml volume) knotted in a chain sinnet were used. The first was placed on the carrier flow line before the sample loop and after the anti-pulse tubing, and the second was after the three way valve and before the capillary. The configuration of the dampeners aids effective mixing of the supplemental and carrier flows. Previous studies showed that a 0.1 M HNO 3 matrix yielded optimum discharge conditions. This was matched using a 0.25 M HNO 3 electrolyte solution that made up 40% of the total flow reaching the discharge. This tubing setup yielded a 0.1 M HNO 3 matrix with a flow rate of 4.5 ml/min reaching the SCGD cell at any given time. The solution that flowed over the capillary and ran down the sides completed the electrical connection between the anode and the graphite electrode. To initiate the discharge, the flow rate of the solution was increased causing the surface of the solution to rise close to the anode. The high potential and decreased discharge gap allowed for the discharge to be created. A stable discharge was established when the optimum parameters were set. Two identical 25-mm diameter fused silica plano-convex lens were fitted into a lens tube with a 23-mm clear aperture. The flat sides were pointed outwards to minimize optical aberrations. The lens system achieved a 1:1 magnification having a 26 angle of focused light, when positioned 50-mm away from the discharge. This angle was matched to 0.6-mm core diameter fiber optic cable (Ocean Optics Xtreme Solarization-resistant Optical Fiber) with a 25 angle of the acceptance. This setup allowed for any 0.6-mm area of the discharge to be 12

20 examined. The other end of the fiber optic cable was connected to an Ocean Optics Maya2000 Pro ( cm spectrograph with a CCD detector). SpectraSuite software was used to record the spectra. Operation parameters varied based on experiment. The ideal position to collect light from is the negative glow; this was located approximately 1 mm above the cathode. 2,18 For all experiments, radiation was collected from a 0.6-mm diameter area centered 1 mm above the cathode. Figure 4 is an image of the SCGD cell with a black dot highlighting the analysis area. 1.7 Results and discussion Pump calibration All tubing used was calibrated using 4 different dial settings. Milli-Q water was collected for 5 min. The total volume collected was used to calculate the flow rate in milliliters per minute. A calibration curve fitted with a linear regression line was created to determine a flow rate at any dial setting. The calibration curve for the carrier flow anti-pulse tubing can be seen in the Appendix SCGD Cell An original SCGD 2,7,18 cell design from Webb et. al., consisting of homemade J-shaped capillary with an overflowing electrolyte solution that acted as a cathode was constructed. The overflow ran into a waste pool in a reservoir. The waste in the reservoir was grounded with a graphite electrode. 2,7,18 Graphite was used because of its high conductivity and corrosion resistance. The electrical connection to the cathode was through the waste pool and overflow. When operating in constant voltage mode and under conditions from Webb et al. 2,7,18, the observed emission fluctuated. 13

21 Figure 4. An image of the SCGD cell with a black dot highlighting the analysis area. The gap between the glass capillary and the tungsten electrode is 3 mm. 14

22 As the solution filled the waste reservoir, it caused a change in the capillary height relative to the solution surface. The height change caused changes in the resistance of the system. To achieve a stable background, changes in capillary height need to be minimized. The easiest way to achieve this is removing the waste reservoir that requires a level surface. A SCGD cell design using a platinum wire wrapped around capillary to ground the overflow was investigated based on the design of Shekhar et al. 13 The advantages of this design are that the overflow constantly meets the platinum wire at a set distance, so there is no need for a waste reservoir or the J-shaped capillaries. Platinum wire was not the ideal choice of grounding electrode due to its high cost and, more importantly, low melting point. The compact design allows for arcing to occur. When an arc was formed directly between the electrodes, the high temperature caused the platinum wire to melt. To avoid melting, the current system uses a graphite rod. The rod has a hole just bigger than the outer diameter of the capillary drilled radially through. A straight capillary was inserted through the graphite rod. This allowed for a constant electrode height. The overflow went right into a waste beaker. The graphite rod easily slipped over the capillary, and the capillary height is easily adjusted with spacers. This new design of the capillary inserted directly through the grounding electrode increased the portability of the instrument due to the removal of the waste reservoir Delivery System A delivery system which consisted of two flows was initially investigated. The first flow was the carrier and was used to move the sample through the sample loop and transport it to the discharge. The second flow was a supplemental flow which was used to maintain the discharge 15

23 while injecting a sample. This system used a 0.1 M HNO 3 solution for both flows, requiring all samples to have a matching matrix. The current system uses two different solutions instead; the carrier solution of Milli-Q water and the supplemental solution containing the electrolyte. This allows for safer solution preparation and more accurate matrix matching. All standards and samples can be prepared with Milli-Q water instead of HNO 3. A concentrated nitric acid solution mixes with the carrier flow creating any desired electrolyte concentration in the SCGD cell. The addition of the electrolyte solution after the sample loop allows for a more versatile system. The term robustness has been used to describe how a source will react to slight changes in the matrix composition. The more robust a source, the less the analyte emission intensity changes with slight variations in the sample matrix. 14 Robust conditions for the ICP are high power and low carrier gas flow rate. 14,15 Operating conditions directly affect the ICP characteristics. For the ICP, the robustness is the ability of the plasma to resist changes to the temperature and spatial distribution of the analyte in the plasma. The SCGD emission relies heavily on the conductivity of the matrix. A consistent matrix allows for less variation in the plasma. The current sample introduction system allows for simple and accurate matrix matching. This is important because slight variations between sample and standards will increase error due the variation in emission between matrixes Spectrometer Previous SCGD-AES systems used a Czerny-Turner type monochromator and photomultiplier tube detector to measure emission intensity. The monochromator used in past studies had the benefits of high resolution and wavelength flexibility, but it could only be used to 16

24 monitor one element at a time, had moving parts (including the diffraction grating), was large (60x23x22 cm), and was relatively expensive. A monochromator does not take advantage of one of the key aspects of the SCGD: that all atoms in a sample emit light at the same time. A multi-elemental detector allows for faster analysis times, lower limits of detection through better background correction, and a possibility of better precision and accuracy when coupled with internal standards. For an analytical field instrument simultaneous multielemental measurement is a benefit. The current SCGD-AES uses a compact pixel CCD detector ($5800) that is backthinned and binned vertically to enhance sensitivity. The new spectrometer has no moving parts, making for a simple operation and compact construction (15x11x5 cm) and allowing for better application in a portable device. The spectrograph has a spectral range of nm. The 1200 line/mm grating and 25 μm entrance slit allows for a 0.35 nm Full width at half maximum. One of the disadvantages of using the spectrograph is the larger instrumental bandpass compared to the spectrometers that are typically used with ICP-AES. Studies have shown that detection limits of an ICP-AES are linearly proportional to the effective bandpass. 16,17 Typical ICP-AES instruments have bandpasses less than 10 pm. 20 Therefore, if an ICP were used with a low resolution spectrometer like as the Maya2000 Pro, it would be expected to have detection limits well over an order of magnitude worse than a typical ICP-AES instrument. One of the disadvantages of the CCD is the relatively low sensitivity Optimization It is important to reduce the signal fluctuation in the analysis area. The background spectrum of the SCGD-AES can be seen in Figure 5. The majority of the background emission from a similar 17

25 Figure 5. A background spectrum from the SCGD-AES corrected for dark signal. 18

26 SCGD plasma in the range of 190-nm to 400-nm is due to OH and N 2. 2 The optimum parameters were established to be where the background intensity and signal fluctuations were minimal. The new discharge cell has an additional parameter of capillary height. To investigate the effect the height the capillary, the OH molecular emission (309.2 nm) was observed. Capillary heights of from 2 mm to 5 mm were tested to determine a relationship between the height and signal fluctuation. The capillary height experiment used a 70-mA current, 5-mL flow rate, a final HNO 3 concentration of 0.1 M, 100 ms integration time, and a discharge gap of 3 mm. After achieving a stable discharge, one minute of emission data was taken for each capillary height. The relative standard deviation of the emission intensity was plotted against the capillary height and can be seen in Figure 6. A stable discharge could not be achieved with a 2-mm height due to arcing. The 3-mm capillary height showed the lowest relative standard deviation and was used for all further experiments Interferences Interferences need to be accounted for with any instrument. With the SCGD, a majority of analyte emission is from the atomic lines. The SCGD sees relatively low amounts of analyte ionic lines giving less spectral interference. In an ICP, the Mg nm ionic emission line is typically 10 times larger than the Mg nm atomic emission line. With an SCGD, the ionic line is only 2% of the atomic line. The Mg atomic and ion emissions lines can be seen in Figure 7. Although there is still a chance of overlapping atomic emission lines, spectral interferences are less likely in the SCGD due to the weak or absent ionic emission lines. 19

27 5 4 % RSD Capillary Height Figure 6. Relative standard deviation in OH emission(309.2 nm) as a function of capillary height, operating with a 70-mA current, 5-mL flow rate, a final HNO 3 concentration of 0.1 M, and a discharge gap of 3 mm. 20

28 Figure 7. A spectrum from the SCGD-AES corrected for background emission showing the atomic and ionic emission lines intensity. 21

29 1.7.7 Quantification With the CCD detector, the advantage is the ability to observe multiple wavelengths at once. An emission spectrum showing the simultaneous detection of Cu, Mg, Ni and Pb can be seen in Figure A-2 in the Appendix. The ability to observe multiple wavelengths allows for more accurate background correction and the use of internal standards. For each element analyzed, a concentrated solution was injected to determine the emission peak to be used during analysis. Emission spectra of a magnesium standard (1 ppm), a copper standard (2 ppm), and a blank can be seen in Figure 8. Two emission wavelengths were chosen for each wavelength. The first wavelength was the analyte peak; the second was a nearby wavelength unaffected by the analyte emission. The wavelengths for copper (324.8-nm) and magnesium (285.2 nm) have been highlighted in Figure 8. The emission as a function of time at these wavelengths was recorded, and the data was transferred to Excel for analysis. To account for any fluctuation in the signal, the emission at the nearby wavelength was subtracted from the emission at the analyte wavelength. Two time periods of the corrected data were then used to calculate the signal that would be used in the calibration curve: analyte emission as the injected sample passed through the SCGD and baseline emission after the sample passed through the SCGD. The baseline emission was subtracted from the analyte emission to yield an observed emission signal for any injection. Three-point calibration curves were created for Ag, Cu, Fe, Hg, Mg, Ni, Pb, and Se. For all elements, triplicate standard injections were collected. The averages of each standard s injections were used to create the calibration curve. The detection limit was determined by the standard 3σ/m calculation, where σ is the standard deviation of 10 blank (Milli-Q water) injections and m is the slope of the calibration curve. The wavelengths, integration times, 22

30 Emission Sgnal (kcounts) Emission Signal (kcounts) Wavelength (nm) (a) Wavelength (nm) (b) Figure 8. (a) Emission from nm, the dashed grey line shows a blank spectrum. The solid black line shows the spectrum of a solution containing 1 ppm Mg. Dark signal has been subtracted. (b) Emission from nm, the dashed grey line shows a blank spectrum. The solid black line shows the spectrum of a solution containing 2 ppm Cu. Dark signal has been subtracted. 23

31 detection limits, R 2 value for a linear least squares fit to the calibration curve, and the relative standard deviation of the signal for the highest-concentration standard can be seen in Table 2. All the curves showed good linearity; R 2 values of or better were achieved for all elements. The three major changes in the new SCGD design should affect the achievable detection limits. The removal of the waste reservoir should lower signal fluctuation and lower the detection limit. The delivery system simplifies the preparation and prevents small variations in the electrolyte concentration of sample and standards by an online acidification process which dilutes the sample by a factor of 1.7. The previous SCGD fitted with the new delivery system would expect higher detection limits by a factor of 1.7. No studies have been done to relate the resolution to the detection limit using a SCGD-AES system, but the lower resolution of the current system would be expected to raise the detection limits. This may be offset by the change to a spectrograph, which allows for better background correction. Table 3 shows the detection limits of the current and previous SCGD designs, and an ICP coupled with pneumatic and ultrasonic nebulizers 19. All detection limits were calculated under optimized conditions using the 3σ/m method. Detection limits using an ultrasonic nebulizer are, on average, nearly an order of magnitude better than detection limits using a pneumatic nebulizer. When comparing a SCGD to the ICP, slight differences are most likely due to the different emission lines observed. An ICP typically uses ionic emission lines where the SCGD observes atomic emissions lines. Magnesium is the only element with a significant different between the two SCGD designs. The previous design has a 10-fold better detection limit. It is likely that this difference is due to a combination of the dilution factor and the resolution. The emission spectrum of Mg can 24

32 Element Wavelength (nm) Integration (sec) Detection Limit (ppb) R 2 Relative Standard Deviation (%) Ag Cu Fe Hg Mg Ni Pb Se Table 2. The wavelengths, integration times, detection limits, R 2 value for a linear least squares fit to the calibration curve and the relative standard deviation of the signal for the highestconcentration standard. 25

33 Element Current SCGD Previous SCGD 2 ICP with pneumatic nebulizer 19 ICP with ultrasonic nebulizer 19 Ag Cu Fe Hg Mg Ni Pb Se Table 3 Detection limits (in ppb) from the current SCGD design, a Webb et. al. SCGD design for several elements. Detection limits (in ppb) of an ICP a pneumatic and an ultrasonic nebulizer from Asfaw et. Al. 19 for several elements. 26

34 be seen in Figure 8. The Mg emission overlaps a poorly resolved OH emission band. This overlap in emission will have a greater effect the lower resolution of the spectrometer. The copper emission also overlaps a significant OH emission band, but there is no significant difference in detection limits between the two SCGD designs. This is attributed to the better resolution of the OH emission lines around copper. The detection limits of Cu, Ni, and Pb with the new system are similar to those of the ICP with a pneumatic nebulizer and about an order of magnitude greater than that of the ultrasonic nebulizer. 19 For Fe, both ICP systems 19 have lower detection limits. For Hg, both SCGDs have a 2.5 times lower detection limit than the ICP with a pneumatic nebulizer. With the ultrasonic nebulizer, Hg concentrations could not be measured. 19 Both SCGD systems have at least an order of magnitude better detection limit for Mg over the ICP systems. The SCGD detection limit of Se differs considerably from those of both ICPs. 19 The at least 2-order of magnitude greater detection limit may be due to the difference in excitation temperatures. The Se atomic line observed is the highest energy used in the study, and the SCGD may not have a high enough excitation temperature to sufficiently excite the Se. 1.8 Conclusion The improvements made make for a more compact and rugged design that favors a portable instrument. The current SCGD design gives comparable detection limits to those of the old design. One of the major design changes to the SCGD is the change in the spectrometer. The current SCGD-system used a spectrograph with a resolution of 0.35 nm FWHM and integration times ranging from seconds. The previous SCGD 4 used a monochromator with a 0.04-nm effective bandpass and integration time of 10-seconds. With the exception of Mg, the detection 27

35 limits between the current and previous SCGD designs showed no significant difference. The reason for the differences between the systems for Mg has been described above. Detection limits between the ICP with a pneumatic nebulizer 19 and the current SCGD are all within an order of magnitude with the exception of Se. The new delivery system allows for standard addition and removes variation to the sample matrix; however the dilution decreases the detection limit. A simple and quick improvement that could be made to improve the detection limits would be to use a delivery system that dilutes the sample and standards less. Using a 25% electrolyte flow instead of a 40% flow in theory would improve the detection limit by 25%. Another improvement would be to change to a higher resolution detector, which in theory would lower the detection limit but would increase the cost of the instrument. 28

36 CHAPTER TWO 2.1 Introduction Most atomic spectrometry based analysis is performed on liquids, but these liquids are often produced from solid samples. The digestion, storage, and stabilization of a sample require an assortment of solvents. These solvents affect the excitation and atomization processes of an analyte. It is necessary to understand interferences by the sample matrix so that they can be avoided or compensated for. In some cases, the matrix can enhance the figures of merit (e.g., detection limit and precision) of a method. In such cases the best strategy may be to intentionally add substances to the sample in controlled quantities. For example, Cs at high concentrations, is often added to flame AAS samples to improve the linearity of other elements calibration curves. The ideal excitation source should be of high power. The power of the SCGD as a source is dependent on the flow rate, electrolyte solution, conductivity, current, voltage, discharge gap and capillary height. 7,8,9,10,12 The SCGD is an unusual atomic spectrometry source due to its dependence on the sample matrix for the stabilization of the discharge. Using a sample introduction system to maintain a precise matrix, the conductivity can be precisely controlled. The current sample delivery systems allows for this. One reason the matrix affects the discharge by altering the ionizability of the matrix. A matrix that is easily ionized creates more ions in the plasma thereby increasing the conductivity. 7 The matrix effects must be established for the SCGD because all AE sources are affected differently. These effects can be classified as excitation or transport effects. Effects of inorganic acids have been studied and are vital to the operation of a SCGD source. Increasing that concentration of a matrix increases the emission intensity of the analyte. 7, 8 With higher acid 29

37 concentrations, the same absolute change in acid concentration is a smaller relative change, and the relative change in the emission is also less. The transport process is also affected by inorganic acid concentration. The transport process for the SCGD is not completely understood, and it is likely that multiple processes contribute to the overall analyte flux into the discharge. There is evidence that an electrospraylike process might occur. Images of an illuminated plasma can be seen in Figure.9. The high heat at the junction of the discharge and solution surface creates a turbulent cathode surface. When a portion of the solution surface enters the electric field a droplet is ejected up into the discharge. The droplets are affected by the matrix properties and field strength 36 and can be observed in Figure 9. With the droplet ejection theory, the droplet size and how easily the matrix evaporates would affect the emission of the analyte. Physical properties of the matrix affect the aerosol properties and how efficiently the analyte is atomized. Changes in inorganic acid concentration change the viscosity, volatility, surface tension and density of the matrix. 21 With nebulizer-based sample introduction, lower concentrations of inorganic acid show lower evaporation rates and larger droplet sizes. 22 Matrix modifiers have been used with AES to enhance an analyte signal. The modifiers create smaller nebulized droplets that are more easily vaporized. For example, a 10% acetic acid solution creates a smaller droplet size than a solution without acetic acid under identical ICP nebulizer conditions, which allows more analyte to reach the plasma. 14 Vapor generation (VG) techniques are another way to enhance detection limits. Converting an analyte to a gaseous form more efficiently removes it from a matrix. One of the 30

38 Cathode Glow Figure 9. A picture and the negative of the SCGD plasma. Above the cathode glow droplets can be observed. A green (532 nm) laser pointer was used to make the droplets easier to see. 31

39 most common techniques is using a chemical reaction to create a hydride. This can be achieved chemically (CVG) using reducing agents such as sodium tetrahydroborate(iii) or tin (II) chloride 23,24,25,28,40, electrochemically (EC-VG) 26,27, or photochemicallly (PC-VG). 29,30,31 No matter the method, the analyte is reduced to a lower oxidation state and in this state it can exist as a gaseous species. The purely chemical process requires multiple toxic reagents, some of which are unstable and must be prepared fresh daily. The CVG process also has high interferences. The EC and PC processes use stable solutions and are less affected by interferences, although the EC process does require careful maintenance of the electrode surfaces. With EC and PC, vapors can be generated with the use of a single reagent. 26,27,29,30,31 In photochemical vapor generation, low molecular weight organic compounds (especially acids) are often used. 23,29,30,31 The SCGD is normally used for both atomization and excitation, but it has also been used as a sample introduction system for an ICP-OES instrument. 32,33 The SCGD has been used as a nebulizer or vapor generation system for the detection of Hg. The emission signal from mercury in the ICP was 16 times greater for the SCGD-ICP combination than it was for the nebulizer-icp combination. It was shown that a 1% low molecular weight acid added to the sample solution yielded a further 2-3 fold increase in mercury signal. 32 This enhancement may be due to some form of chemical vapor generation. The discharge is electrochemically active 34 and background spectra show that UV light is emitted, so EC or PC-VG mechanisms may be involved. The SCGD-ICP hybrid produced a detection limit for Hg of 0.7 ppb, which is much lower than typical detection limits of either instrument alone. Typical SCGD detection limits for Hg are 20 ppb, and typical ICP detection limits for Hg are comparably high (see Chapter 1). In order to get this improvement, the advantages of the SCGD, such as small size, low power, and low cost, were sacrificed. Another advantage common to both instruments was also lost. Both 32

40 ICP and SCGD are useful for detecting a wide range of elements and can even be used to detect them simultaneously (see Chapter 1). Despite trying 10 ppm solutions, As, Se, Pb, and Sn could not be detected with the SCGD-ICP hybrid. So far, no investigations have been published studying the effect of low molecular weight organic compounds (LMWOC) on mercury emission from the SCGD itself. Additionally, the effect of LMWOC on other elements emission from an SCGD has not been studied. In this chapter, SCGD vapor generation techniques were applied to conventional hydride forming elements, mercury, transition metals, and noble metals. Unlike previous studies, atomic emission was measured from the SCGD directly rather than a secondary excitation source. This approach has the advantages of simplicity and low cost. The aims of this study were to understand the effects of low molecular weight organic compounds in and SCGD so that they can be compensated for or exploited to produce lower detection limits. 2.2 General Procedures and Reagents All general procedures do not vary from chapter one. In addition to the reagents used in Chapter One, trace metal grade HCl purchased from Fisher Scientific (Fair Lawn, NJ), 99.9% formic acid from Acros Organics (Fair Lawn, NJ), Optima acetic acid from Fisher Scientific (Fair Lawn, NJ), and ethanol from AAPER ( Shelbyville, KY) were used. In addition to the metal standards that were used in Chapter One, 1000 ppm Ca and and 1000 ppm As (BDH Aristar Plus, VWR International, West Chester, PA) solutions were used to prepare standards. All low molecular weight organic solutions were prepared by a volume to volume dilution. 33

41 0.17 M nitric acid - About 100 ml of Milli-Q water was added to a 1000 ml volumetric flask ml concentrated Optima nitric acid was added to the flask. The flask was filled to ml with Milli-Q water M and 0.17 M hydrochloric acid- About 100 ml of Milli-Q water was added to a 1000 ml volumetric flask ml or ml concentrated hydrochloric acid was added to the flask for 0.25 M and 0.17 M respectively. The flask was filled to ml with Milli-Q water. 2.3 Origins The original goal of this chapter s research was to see if PC-VG could be coupled with a SCGD-AES instrument. It has been documented that PC-VG of Hg is achieved in a 15% formic acid matrix with 10 seconds of 254 nm irradiation. 23,37 These conditions can easily be achieved with minor modification to the current sample delivery system. The delivery system was fitted with a 3-mL polyethylene (PE) VG cell in place of the sample loop. The cell consisted of a 10- cm coil of tubing with an Hg pen light inserted into the center, which allowed for maximum irradiation. The pen light primarily emits 254 nm light. The carrier flow was changed to a 15% formic acid solution, and the same matrix was used to make a 400-ppb Hg standard. The electrolyte solution was 0.25 M HNO 3, which yielded a final matrix of 9% formic acid and 0.1 M HNO 3. The loaded cell was irritated for 0, 15, and 60 seconds prior to injection. The irradiation from the SCGD was collected and processed similar to chapter one. There was no significant difference in emission signal between the irradiated injections. The relative standard deviation of all the emission signals was 1.8% suggesting that irradiation had no significant influence on the emission intensity of mercury. 34

42 The efficiency of VG relies heavily on the reaction matrix. 23,29,30,31,37 Along with formic acid, the sample matrix contained HNO 3 which could have an effect on the VG efficiency. Both HNO 3 and HCl lower the efficiency of Hg vapor generation. HCl stabilizes Hg 2+, and HNO 3 suppresses Hg 0 formation. The effects are amplified at greater acid concentration. 38 It is possible that the vapor generation was suppressed by HNO 3. A longer sample irradiation time may be required to achieve vapor generation in the presence of nitric acid. To test sample irradiation time, a system was fitted with a PE 30-ml VG cell. This cell allowed for irradiation times of over 6 min. An irradiation profile was created by changing the carrier flow to a 400 ppb Hg standard in 15% formic acid. The VG cell was loaded with a blank and switched to inject. Once the Hg signal appeared, the irradiation was turned on. Figure 10 shows the initial blank injection with no irradiation, the increase in emission intensity when the Hg reaches the discharge, and a decrease in emission intensity as the sample was irradiated. The signal at a short delay from the irradiation on mark is from Hg that was near the end of the VG cell and was only irradiated for a short time. At longer delays, the Hg has been irradiated for longer times. Although the first irradiation experiment suggested that under known PVG conditions sample irradiation had no effect on the Hg emission intensity, this further experiment shows that Hg emission actually decreases with extended sample irradiation. One explanation of this is that Hg vapor was successfully created but was lost through the PE tubing. This explanation seems reasonable because PE is somewhat permeable to Hg vapor. 39 The longer loop allowed for some of the Hg vapor to escape through the PE tubing. Switching the tubing to PTFE eliminated this effect. With PTFE tubing, the Hg signal from the same solution was not affected by irradiation. In either case, the signal was higher than would be expected based on the results discussed in Chapter One. The addition of formic acid increased 35