Chapter 20 Atomic Spectroscopy Problems: 1, 2, 6, 8, 9, 10 20-1 What is Atomic Spectroscopy Atomic Spectroscopy is principal tool for measuring metallic elements at ppm (parts per million level - :g/ml). A block diagram is shown in figure 20-1. -Liquid sample is aspirated (sucked) through a plastic tube into a flame -Flame evaporates all liquid, breaks all molecules into atoms, and excites many atoms into high energy states -Concentration of elements measured by absorption or emission of light from atoms in flame (figure 20-2) Atomic Absorption - measure concentration by absorption of light Requires a light source of proper wavelength Atomic Emission - measures concentration by emission of light No light source required Both require monochrometer to tune to proper wavelength -Compare figures 18-5 or 18-6 with figure 20-3 spectra of gaseous atoms incredibly sharp when compared to molecules in solution (line width 100 nm in solution, .01-.001 nm in atomic vapors) Makes these methods very selective, can often analyze for many different elements in one sample with no interference -Measures concentrations down to ppm so very sensitive (In fact you may have to dilute a sample down to get it acceptable for analysis) -precision is only about 2% so not terribly precise 19-2 Atomization: Flames, Furnaces, and Plasmas Atomization - Process of breaking analyte into gaseous atoms Older or inexpensive machines (like ours) use a combustion flame Newer, more expensive instruments use either a electrically heated graphite furnace or inductively coupled argon plasma to atomize samples Flame (our machine) Typical premix burner shown in figure 20-4 Sample, oxidant, fuel all mixed together before hits the flame Flow of oxidant gas draws liquid through small opening and makes into a mist (this part called a nebulizer) similar to air brush for painting or old fashioned perfume bottle Mist directed at a glass bead to further break droplets apart 2 Then mist directed down a series of baffles Mixes gases and mist together Catches larger droplets and pulls them out of the gas Only about 5% of original sample actually makes it into the flame! In flame, solvent first evaporates from around each mist droplet Leaves individual atoms in flame Many metal atoms (M) oxidized to oxides (MO) or hydroxides (MOH) In MO or MOH form, no longer atomic, so no longer absorbs or emits at same wavelength, so signal decreases ‘Rich’ flames (yellow)- rich in fuel - presence of excess C in fuel tend to reduce MO and MOH to M so can get better signal. However, rich flames are cooler, so not a many M atoms are excited state ‘Lean’ Flames (Blue) - low in fuel, high in oxidant, more tendency to oxidize metals, but hotter flame so more M in excited state Choosing right flame condition is often a trade-off, will use different flames for different elements or for absorbtion vs. emission Common flame gases and temperatures shown in table 20-1 Our machine set up for Acetylene/air (2,400-2,700K) or acetylene/nitrous (2,900-3,100K) Other variables include flow of gases to flame, height above the flame that you do your measurement, ratio fuel to oxidant. So lots of variable to tweak if you want to get optimum conditions for a particular experiment Furnace Typical electrically heated graphite furnace shown in More sensitivity than flame Less sample used than in flame 1-100 ul of sample injected in hole in middle Windows at ends keep sample from escaping Furnace heated to 2,250K for <7 sec (Furnace heats up at rate of 2000K/sec!) Furnace made of graphite - so kept in inert atmosphere so doesn’t burn up! Why more sensitive? 3 Sample constrained to light path for several seconds (In flame passes through light in a fraction of a second) Less sample - here on 1-100 ul of sample in furnace - In flame may aspirate 10 ml through flame to get a good reading Look a bit more on construction. Sample actually sits on a platform called a L’vov platform. This allows to heat more uniformly and reproducibly Look a bit more at heating. Not simply heating the bejasus out of it. Typically in 3 phases (similar to broccoli lab) Heat for 125 fo 20 s to evaporate water Heat to 1400 for 60 s to char (burn all organic matter) Heat to 2,100 for 10 sec to atomize - during this time you turn on the detector and integrate the reading over the entire 10 seconds Heat to 2,500 for 3 sec to burn off any residue and start over So you can see there are some variables in optimizing heating Also variables in what the sample is dissolved in This is called the matrix Can add different compounds to matric called matric modifiers to change evaporation characteristics of sample Can also have things in the sample itself that affect sample evaporation. Bottom line - not a simple and just stick it in and heat it up Inductively Coupled Plasma (ICP) Photo at start of chapter on page 410 and figure 19-6 Here we are going to have a ’flame’ at about 10,000K 3-4X normal flame To get it we are going to use an inert gas, usually Ar Will explain in a minute Of all techniques has best signal because of high temperature and stability. However is last technique you will see at a place like this because very expensive to buy and to run. How does it work Start with inert gas, Ar Put some coils of wire around it Run radio frequency AC current through the wire Current running in wire makes AC electric field outside the coil (this is the inductively coupled part) AC field rips electrons off of Ar atoms to make Ar+ ions and 4 e-. (Now have a gas of Ar+ ions so this is the plasma ) Electrons ramming back and forth in AC field heat up to 10,000 K So hot coils will melts so need to have the coils cooled with water By convection the hot plasma will move up, so need to replace Ar in induction area with fresh Ar Also blow Ar past to help cool! That is why looks like a flame Can use a nebulizer system similar to that used in a flame to introduce sample into ICP flame. But can do even more high tech with a ultrasonic nebulizer. Spray sample onto a crystal vibrating at 1 MHz. Gases then mixed, and heated to evaporate solvent. Get sensitivity up by a factor of 10 Picture of apparatus figure 20-7 20-3 How temperature affects Atomic Spectroscopy Have mentioned temperature several time as important. Why? Has to do with number of atoms in excited state Need to look at Boltzmann distribution N* is number of atoms in upper (excited) state N0 is number of atoms in lower (ground) state g* is number of upper states with the same energy (>1 if degenerate and more than one state have the same energy) g0 is the number of lower states with the same energy (again is >1 if the state is degenerate) )E is the energy difference between the state k is the Boltzmann constant (1.381x10-23 J/K) T is the temperature in K. Let’s calculate the Boltzmann distribution for the lowest state of a sodium atom at 298K, 2500K (a typical air/acetylene flame) and 10,000K (a typical ICP ‘flame’) The )E for this transition is 3.371x10-19 J/atom (Homework problem from previous chapters. What is the wavelength of 5 this transition?) Assume the excited state has a degeneracy of 2, and the ground state is nondegenerate (degeneracies should always be given in a problem) 6 298K So you would need about 1036 atoms (that is 1013 moles) before you would have a single atom in the excited state at room temperature 2,500K So at 2,500K only .01% of our atoms are in the excited state. Still not much, but it is a lot better than at RT 10,000K What does this mean for absorption or emission signals? In absorption you start with the ground state, so the more atoms in the ground state, the more signal. Absorption will work well at either RT or even 2,500 because most atoms are in the ground state In emission you start in the excited state. The more atoms in the excited state the more signal, and yo can see that here the 10,000K of the ICP gives you literally .174/.000115 about 1,500 time more atoms in the 7 excited state or 1,500 times more signal. Also note, if you follow the presentation in the book, it shows that at 2,600 K just a change of 10K in the temperature will change the number of atoms in the excited state by 4%, so if you use flame for emission, small changes in the flame temperature can change your answer be several %, so you need a very stable flame. Bottom line the flame is OK for atomic absorbtion, the ICP is best for emission (But we will still do emission with a flame because it does work, it just isn’t as sensitive or as accurate as it could be with the fancy expensive instrument) 20-4 Instrumentation Since we are dealing with spectroscopy, our instrument has many of the same elements as any other spectrometer (light source, sample holder, monochrometer, detector etc.) The main places we will see difference are 1. Light source 2. Sample holder 3. How we subtract the background Linewidth Problem One of the things you should have picked up on in the previous chapters is that to accurately measure an absorbance you need to be at an absorbance maximum in the spectrum. Do you remember why? (This is the only region where Beer’s Law applies...If you are on a shoulder then the absorbance varies as a function of wavelength over the region and Beer’s Law doesn’t work) In our typical UV/Vis spectrometer the width of the overall transition was 20-100 nm, and the bandwidth (region that the machine could be tuned to) was on the order of 1 nm or so. In Atomic spectroscopy we have already seen that the transition lines are much sharper, on the order of .01-.001 nm, so the bandwidth of our spectrometer need to be at least 10x smaller! In theory the bandwidth of atomic absorbtions is actually about .0001 nm, but there are two effect that broaden the transitions into the .01-.001nm range Doppler effects (figure 20-11) Just as a train whistle has a different pitch when the train is coming toward you or away from you, and atom ‘sees’ a slightly different 8 wavelength depending on whether it is traveling toward a source or away from it, to this spread the frequencies out slightly. Pressure Broadening As the pressure inside the flame increases, the number of collisions of atoms in the flame increases. The energy of these collisions can be added or subtracted to the energy of the transition itself, and this also can broaden out the overall range of transitions observed. Hollow-Cathode Lamps (figure 20-12) This is the light source used in Atomic absorbtion Tube filled with inert gas (Ne or Ar) Hollow cathode (negative) made with metal we want to detect Run a high voltage between anode and cathode This makes Ne or Ar ionize Ne+ or Ar+ attracted to hollow metal cathode As these ions hit the metal, atoms of metal are ejected into the gas As metal atom interact with energetic electrons atoms are excited, so generate light at that metals wavelength Since excitation is not by flame the linewidth is extra sharp Thus we don’t need a monochrometer! It only emits the wavelength we need It emits is in such a narrow line (compared to the absorbtion spectrum) that we are right at the ‘sweet-spot’ of absorbance max Only disadvantage is that we need a different lamp for each element Background Correction Along with the signal from the metal you are interested in, there is always a signal from the background, that is a signal from absorption, emission and/or optical scattering from the sample matrix, the flame plasma, or the white hot furnace itelf. This is especially true in the graphite ovens, where the oven itself holds in smoke that can affect your readings. A typical signal from a furnace/oven is not shown in text. Note that the overall background it about .3, while the signals are from .4 to 1, so the baseline correction, or a subtraction of .3 from all values is very significant. With an older machine like ours, the background correction is fairly simple, run a blank with no metal in it as a control on the baseline for 9 each experiment With the fancier newer machines there are fancier correction techniques Zeeman Background correction (Zay-man) Used in absorption Apply strong magnetic field parallel to light path Magnetic field changes energy of atoms Some with higher E, some lower, some no change One with lower E emit at longer wavelength Ones with higher E emit at shorter wavelength Ones with same E don’t interact with light because is is in the wrong polarization Thus when turn magnetic field on, signal from sample disappears (higher and lower E atoms cannot absorb the light at original frequency, atoms at same E cannot interact with light due to polarization) So turn on magnetic field to get background, then turn off to get signal. Kind of like the chopper we had on a double beam spectrophotometer. This text is a bit weak here. Look as Skoog, Holler & Neiman Multielement Analysis with ICP In our low end Flame AA/emission machine, you have to manually change the monochrometer to the correct wavelength for each element you want to analyze. With the more expensive ICP machines you can have it scan the entire wavelength range and see many elements in a single sample, but it takes time to complete a scan, so sometimes the machines are designed with the diffraction grating fixed, and detectors set at the correct angles for the elements you think you want to analyze for. This is makes an individual analysis much faster, but the machine then can’t analyze any different element. The best solution is to match an ICP instrument with a photodiode array detector so you can analyze many element at once without scanning. However, the resolution in a single, linear, photodiode array is not good enough to separate all the lines of all the element. Instead, as shown in color plate 20, the light dispersed in 1 dimension using a grating and in a second dimension using a prism, so the spectrum get spread out in two dimensions, and read of using a 2 dimensional charge injection device 10 (CID) One of the more sensitive alternative is to take the output of atoms from an ICP flame and send it directly to a mass spectrometer, that detects atomic masses and separates elements by their atomic mass. This give you even more sensitivity, and retains the ability to identify all elements. This hybrid machine the ICP/MS is about the ultimate machine for elemental analysis at this time 11 Detection Limits Figure 20-16 Detection limit is defined as the amount of material needed to give a signal that is 2x the peak to peak noise of the baseline (see figure 20-15) The table in figure 20-16 summarizes the detection limits for various elements using methods described in this chapter In general Flame AA is least sensitive with detection limits in the range of 1 - 100 ng/ml (1-100 ppb) Furnace detection limits are about 100X better (do you remember why?) ICP emission is in between, but can approach the furnace levels ICP/MS is even 100-1000x better, with detection limits in the ppt-parts per trillion range! 20-5 Interference Interference is defined as any effect that changes the signal of an analyte while its concentration remains the same (note can increase or decrease the signal) Type of interference Spectral Interference - analyte signal overlaps signals from wither element in sample, flame or furnace. Typically increases signal, so need to subtract a background correction. Best solution is to choose a different wavelength for analysis where interference doesn’t occur Chemical Interference - Chemical reactions between analyte and matrix that prevent the atomization of the analyte. One common one is that Sulfate or phosphates anions will form nonvolitile salts with Ca+2, so its signal decreases. Can be treated in two ways 1. Releasing agents - can add to metal to prevent formation of interference complexs. This this example EDTA or 8hydroxyquinoline. Can also add to anion - in this case La+3 2. Use a fuel rich flame to reduce oxides or use a hotter flame Ionization interference - mostly for alkali metals Ionization E is lowest for these metals (remember from Gen Chem?) . Thus the E of the flame provides enough E that a sizable portion of atoms are in ionic instead of atomic state. Usual fix is to add an ion suppressor, an element that ionizes more 12 easily than the analyte This gives you a high concentration of the ion of the suppressor in the flame, and this then pushes the equilibrium of the analyte to favor the un-ionized state Method of Standard Addition There are all kinds of interferences. Often can take weeks to identify and fix each different interference that might be occurring in a sample. The method of standard addition is a way of doing an experiment that corrects for many interferences, so you don’t have to worry about them See section 5-3 Essentially a different kind of calibration curve Usually make calibration curve by dissolving a standard in water In method of standard addition make standard curve by adding measured amounts of analyte to sample (see figure 20-17), and then extrapolate back to the zero point, where you haven’t added anything Why does this work? When we make a standard curve you want the standard to resemble the sample in every way. Complex, real world samples contain all sorts of junk, and it is hard to make a standard that exactly matches it. Thus your artificial standard may easily leave out some trace element that is causing an interference and changing the signal in your analysis. In the method of standard addition, since each sample you analyze was made from the original sample + a little more of the standard, you know that is has all the original components, so all the interferences are there. When you make your standard cure with the interferences intact, the interferences cancel out from the analysis so you don’t have top worry about them! Virtues of ICP Once again ICP is superior. At the high T of the ICP flame, most common interferences disappear. Add to that the better signal, atomization, lack of oxide formation, and low background, and you have a superior method