Atomic Emission Spectroscopy Atomic absorption spectroscopy was limited in usefulness by poor sensitivity, and the inability to perform multi-element analysis. Emission spectroscopy has advantages in these areas. A flame, even the hottest, will only reach a temperature of about 3000K. This will only excite elements in the first two groups. For high temperatures, electrical discharges must be used. A DC arc will reach up to 10,000K. These can be used with solid samples, and are used during the analysis of manufactured steel. Spark sources can also be used. However, these electrical discharge sources are relatively unstable. An atomisation source was needed that would give a detection limit of less than 1ppm for multi-element solutions. In 1965, the inductively coupled plasma (ICP) was developed. This is the most efficient atomiser known. It was initially used in ICP-AES, and more lately in ICP-MS.

nk excited atoms per cubic metre. The intensity of the emission is related to the Einstein Transition Probability for spontaneous emission (Aki) - the probability per second that an excited atom will emit a photon and return to the ground state. Aki is equal to the inverse of the half-life of the excited state. Single atom detection is possible, but of no use. So, photons per second per cubic metre = nkAki E = hν, so P = nkAkihν J s-1 ( W m-3) Better to relate to number of atoms per cubic metre - closer to chemical concentration. Let N = number of atoms per cubic metre.

Ek = energy of excited state, J = statistical weight.

Define partition function, Z(T), a function of temperature.

So,

This suggests a linear relationship (through origin) of concentration and power.

This equation gives watts into all space which we cannot measure. Amount of light is determined by the area of the slit and the solid angle.

We want watts per metre per solid angle.

, but ΔV = ΔA x L

so,

Z(T) is dependent on temperature, so power does not have a normal exponential curve.

Also, at high temperatures, electrons are ionised off and lost. The resulting M+ ion cannot make a contribution to the spectral line as it is a different species. Thermodynamic Equilibrium Gas temperature for a monatomic gas: (a) electronic temperature Te (b) gas temperature Tg (c) excitation temperature Tex

ik (d) ionisation temperature Ti If molecules are present: (e) dissociation temperature Tdis (f) vibrational temperature Tex

vib (g) rotational temperature Tex

rot For LTE: Te = Tg = Tex

ik = Ti = Tdis = Texvib = Tex

rot Self-Absorption The emitted photon may be absorbed by atoms of the same kind in the source. The higher the density of atoms in the source, the process of self-absorption is more likely. Consequences

(a)

WHMH stays the same to a limit. The peak intensity gets stuck at the black body limit. P is now a function of temperature only. This limit is called the Planck temperature, same as an incandescent metal bar, P ∝ c. The line also gets broader. Peaks are Gaussian - they extend to ±∞. Although the middle limit has been reached, the “wings” continue to increase in intensity.

B’(ν) = B0[1-exp(-kνL)] Aν = 1-exp(-kνL) absorptivity From Kirchoffs law :

B(ν) = AνBB(ν) measured expected from black body.

Obtains BB(ν) from Plancks Law

The total absorption factor

(b) As long as the bandpass is sufficient to see everything, the slope changes from 1 to 1/2 as P ∝ c → P ∝ √c. If narrow bandpass :

(c) If we fire a pulse of electrons or laser light into excited atoms.

τ is the average time that an atom stays in the excited state, typically 1-10ns. A higher density means there is more chance that a different atom will be excited by emission of a photon, delaying the excited state by another 1-10ns. This is called radiation trapping. i) LINE BROADENING ii) LOSS OF LINEARITY iii) RADIATION TRAPPING Optically thin - all photons get out. Optically thick - some photons are trapped. Depends on concentration N, and viewing depth L. Self-Absorption and Self-Reversal

The increased velocity of atoms in the hottest part of the sources causes the frequency to be doppler-shifted. Self-reversal arises since only the hottest part significantly emits but all, especially cool, atoms will absorb. Hot atoms emit photons which are absorbed by cool atoms. This is bad news. Extreme example :

Inductively Coupled Plasma (ICP) A torroidal plasma.

Need a higher frequency, 27MHz - “donut” plasma.

If the particles are gently wafted into the plasma, they go around instead.

Particles are injected at about 2 m s-1, and they cannot move sideways because it is too hot. Any compound with a diameter less than 10µm is blown into free atoms. ICP is the best ever source for atomisation. It is also a pretty good excitation source - will excite atomic spectra and ionic spectra. Injection will not interrupt power coupling as the injection zone is only indirectly heated. Can inject gases, liquids and solids. Can even insert a carbon rod with the sample on the end into the central channel. So ICP is: 1. best atomisation source 2. very good excitation source 3. very good ionisation source 4. very robust. The temperature distribution is the analytical region is pretty flat. Flat temperature field, so no self-reversal. The optical depth, L, is small.

Atoms expand out a bit, but not a lot. Dynamic range is large - 6 orders of magnitude. Ionisation Equilibrium Electrons reach ionisation energy of Ar (15.76eV), then avalanche breakdown occurs and rate of ionisation = rate of recombination + rate of diffusional loss. • Ionisation mechanism Ar + e* ⎯→ Ar+ + 2e- (one step) Ar + e* ⎯→ Ar* + e- (+ e*) ⎯→ Ar+ + e- (two step) • Recombination Ar+ + e- ⎯→ Ar + hνcont (dominant background radiation below 500nm) • Three Body Recombination (less common) Ar+ + e- + e- ⎯→ Ar + e* Above 500nm the main source of background continuum is Bremstrahlung. Macroscopic View Of Plasma Heating - Plasma may be considered a 1-turn coil of finite resistance. - Current flowing in induction coil induces an eddy current in the plasma. - Resistance to the eddy current results in heating. - Because RF is 27.50MHz, eddy currents only flow in the outer regions of the plasma (“Skin Depth Effect”) Power Requirements

Cp = specific heat at constant pressure = 520 J kg-1 K-1 Power for =2.77 x 10-4 kg s-1 (10 l min-1 Ar) and ΔT = 5000K, P = 720W assuming 100% coupling efficiency. Plasma Zones

PHZ : Pre-heating zone. - desolvation, vapourisation, partial atomisation. IRZ : Initial radiation zone. - atomic emission first observed, e.g. Na- yellow, Y-red. NAZ : Normal analytical zone - strong emission from high energy atomic and ionic lines, e.g. Y-blue. - about 14mm above load coils. Tailflame - reappearance of low energy atomic lines and band emission from atmospheric entrainment. Hard and Soft Lines Soft lines, Eexc ≤ 5.5eV e.g. Li, Na, K, CaI, BaI, SrI, CuI up to Cd 288.8nm (atomic line) Hard lines, Eexc (or Eion + Eexc) > 5.5eV e.g. Zn 213.8nm (atomic line), Cd(II) 226.5nm (emission line). Summary Soft lines - peak low in plasma, close to IRZ - position of maximum strongly dependent on transition and operating parameters - peak shifts higher in plasma as injector flow increases - peak shifts lower as power increases - maximum emission intensity increases with power and decreases with injector flow rate - position of emission peak correlates with “normal” temperatures for transition Hard lines - exhibit peak emission higher in plasma (in NAZ) - position of peak emission largely unaffected by operating parameters or species - increase in power increases the intensity - increase in the injector flow shifts the emission peak upwards slightly and reduces the emission intensity

Effect of Operating Parameters on Plasma Zones - as power goes up, zones shift downwards and NAZ expands. - as injector flow goes up, zones shift upwards. - as outer-flow increased, slight downward shift and expansion of zones. - as intermediate flow increases, body of plasma moves upwards. Instrumentation Comprises: Source unit Sample introduction system Spectrometer Computer for control and data manipulation Source Unit RF Generator (i) Two types :- - free running oscillator (frequency alters as the plasma does - load coil is part of the generator) - crystal oscillator and power amp (more common, can be miniaturised, and can now be separated) Type does not affect analytical performance. (ii) Frequency : 27-50MHz. Well screened. 50MHz is more common. As f↑, Texc

↑ but Tthermal↑.

Texc = excitation temperature (T is Boltzmann temperature). (In plasma, temperatures are not in thermal equilibrium (but nearly). Above 100MHz Tthermal is not much hotter than a flame - start to get interferences). ∴ Spectral intensity↑, but atomisation efficiency↓. 40MHz used for OES, 27MHz for ICP-MS. (iii) Power (argon plasma) Min. 0.8kW - aqueous sample, low matrix (typically 1kW today) Max. 2.0kW - organic solvents, high matrix. Power increment for organic solvents = +400W. (iv) Stability and Regulation

≈0.1% required to avoid frequent recalibration (today all generators are good) (v) Screening to National Standards r.f. radiation leakage < 0.1mW cm-2 Torches (i) Standard : 18mm i.d. outer tube 1mm annular gap 1.5mm i.d. injector tube (max. 3, 1.5-2 common) Material : fused silica Injector : alumina for HF (used in geochemical analysis for silicates -attacks silica) Gas flow rates outer = 10-15 l min-1 intermediate = 0.5-1.5 l min-1 injector = 0.5-1.0 l min-1 (ii) Gas consumption can be reduced using smaller annular gap, but cost of the torch increases, or, use mini-torch, 13mm i.d. outer tube (5-10 l min-1). (iii) Molecular gas plasma can be operated at slightly higher powers, e.g. Ar + 10% N2 (outer flow), gives increased performance but not used commercially. Air plasma has been used for process control - start with argon. Sample Introduction System ICP is very robust with respect to sample introduction and can accept samples as solids, liquids or gases. However, particle size d ≤ 6µm. Plasma loading ≈ 20mg min-1. (i) Nebulisation 99% of ICP analysis is carried out on aqueous samples using a nebuliser/spray chamber combination to convert the sample to a fine aerosol. Pneumatic nebuliser operating at 1 l min-1 Ar (injector flow) and 1-2ml min-1 sample → polydispersers and aerosol., 0.1 < d < 100µm. Spray chamber removes large (and small) particles, allowing particles d < 6µm to the plasma. Transport efficiency ≈ 1%. Types of nebuliser:

Concentric flow and cross flow - good for low matrix samples. Babbington - essential for slurries. Slurries are aqueous suspensions of powdered (d < 6µm) solids and surfactant, usually < 4g / 100 conc. (ii) Electrothermal Vapourisation (ETV) Uses an adapted AA furnace atomiser. Advantages : - detection limit improved by x10 - small sample volumes can be handled (5-50µl) - analyte and matrix emission can be separated in time. Disadvantages : - slow, 2-3 mins per sample. - interferences in vapourisation stage. - transient signals have to be measured. (iii) Hydride Generation HG Volatile elements, e.g. As, Se, Sb, Pb, Te : not good detection limits using nebulisation. Can be reduced to their hydride using acidic NaBH4. M + NaBH4 + HCl ⎯→ MnHm e.g. AsH3, H2Se, PbH4 (not very soluble in water) A gas liquid separator is used to separate the hydride from the aqueous phase and the hydride swept into the plasma using the injector flow. Advantages: - 100% of analyte is transported into plasma (⇒ 10x (+) improvement in sensitivity) - separation of analyte and matrix ⇒ some (spectral) interferences are reduced. Disadvantages: - slow, 2mins per sample. - H2 is liberated, must control to prevent extinction. - transition metals, e.g. Cu, cause interferences. (iv) Solid Sampling Techniques Arc and Spark

- conducting materials, primarily metals. Laser Ablation - non-conducting materials, e.g. rocks, ceramics, biological materials. Can sample very small areas with very high powers. NB. Solid sampling techniques require standards that are closely matched in chemical content and physical form to the sample. A major element in the sample whose concentration is reasonably constant is often used as an internal standard - individual line intensities are expressed as a ratio to the internal standard. Spectrometers (i) Single Channel Scanning Spectrometer Linear dispersion typically 0.4 nm mm-1. Slits ≈ 20µm Atomic lines are very narrow, e.g. 0.003nm. Many thousands of lines in spectrum so can get many lines in 1nm of the spectrum, so need bandpass about 0.01nm. Can get up to 4800 lines per millimetre on modern gratings. Advantages: (a) Flexibility of line selection, 200nm, flushing with e.g. nitrogen can get lower. Up to 1600nm (sulfur / phosphorus / aluminium lines are found below 200nm). (b) High resolution (c) Ease of automatic, off-peak background correction. (d) Low cost, £50,000 up to £100,000 (e) Survey analysis Disadvantages: (a) Speed of analysis is slow, e.g. spray sample into chamber, move to e.g. Cd line for 20 secs, move to next line etc. Must keep spraying sample - 10/15 ml for 10 elements. (b) Higher sample consumption (c) Inability to implement internal standardisation (using In or Sc). Cannot measure two lines simultaneously. Need two spectrometers. PE used three : (i) background

(ii) internal standard (iii) analyte Uses : moderate throughput of samples (e.g. 12-20 per day), type of analysis not known in advance, high resolution required. (ii) Multi-Channel Spectrometer Based on curved grating (part of Rowland circle). Costs about £1000 to install each line. Scan spectrum at each exit slit - place on stepper motor slide, move back and forward ≈ 1mm to ensure no overlapping lines. Advantages: (a) Speed of analysis - spray 20secs to stabilise signal, then 20secs to measure all lines. (b) Minimal sample consumption - spray once. Powerful system for small samples if coupled with ETV. Can couple to HPLC in speciation studies (or CE), e.g. HPLC for vitamin B12, see large Co peak. (c) Accuracy and stability of λ setting. (d) Flexibility of background correction on-peak and off-peak. (e) Ease of applying matrix corrections - account from interferences from sample. (f) Ease of implementing internal standardisation. Disadvantages: (a) High cost. (b) Inflexibility in line selection, must specify to manufacturer. (c) Moderate resolution, 20-25µm entrance, 50µm exit. Uses: high throughput (e.g. 200 samples per day), routine samples, not too severe spectral interferences. (iii) Less Common Spectrometers - SIM-SEQ : combination simultaneous-sequential system, flexible and fast but expensive. - Echelle Grating Spectrometer : uses coarsely ruled grating at high order to achieve high resolution, need 2nd disperser to sort orders, expensive, low light throughput.

- Fourier Transform : very expensive (£230k), research only. - Spectrometers with solid state detection PDA - photo-diode array (poor in UV below 250nm) CCD - charge-coupled detector, high res. chip 1000x1000 CID - charge injection detector CCD, 10mm x 10mm chip and large spectrum, need advanced optics (available). PE developed CCD chips, have 1,000,000 points and takes a computer a long time to read all the points, so have 1200 regions of interest. Cost ≈ £100,000, but CCD/CID are very powerful. Modern choice : Scanning £60-£70k or PCD/CCD/CID £100k, measure line and background and can measure all elements in periodic table. Contract labs usually buy scanners. Interfernces Two general types: (i) Translational Interferences - Sometimes called Addition Interferences.

- Translational interferences are proportional to the concentration of the interferrent and independent of the concentration of the analyte. - The effect is to add an increment to the analyte emission signal. The increment is constant provided that the concentration of the interferrent is constant. - Translational interferences arise from :

-Spectral Interferences a) Spectral Overlap

This is a physical problem, get a merging of the two peaks to give a double hump in the scanned peak. b) Line Interference

Get overlapping peaks. c) Background Interferences

Large interfering line (e.g. Al). Observed peak is larger than true peak. Techniques for Background Correction Line Overlap

- use an alternative line. Always best, but sometimes can’t because of sensitivity problems. - use ON PEAK BACKGROUND CORRECTION On-peak background correction involves preparing a cross-calibration curve of the contribution of the interferrent to the emission intensity at the anlayte wavelength. The inteferrent concentration is then measured using a line free from interference and the approximate intensity (or concentration equivalent) is subtracted from the intensity observed from the analyte. The success of this approach depends on the overlap and relative magnitude of the two lines. The interfering line contributing noise and CL increases by 0.1x magnitude of the correction. Line Interference - use an alternative line - use ON PEAK BACKGROUND CORRECTION - improve the resolution - narrow slits give S/N problems. Can go down to 3µm. Lose too much light. - work in a different order. - use a mathematical deconvolution procedure. Background Interferences - use another line - use ON PEAK BACKGROUND CORRECTION But often don’t know the problem, so don't know what to measure. - use OFF PEAK BACKGROUND CORRECTION Off-peak background correction

Involves scanning the line and estimating the contribution of the background at the analyte line wavelength from a measurement away from the line centre. (ii) Rotational Interferences - Sometimes called multiplicative interferences.

- Rotational interferences occur when the effect of a given concentration of the interferrent is to multiply the analyte signal by a constant factor. The factor is proportional to the concentration of the interferrent and, to a first approximation, independent of the concentration of the analyte. - Usually suppresses peaks, but can enhance. - The effect of multiplication by a constant factor is to rotate the analytical curve about the origin. - Use matrix matched standards. - If this doesn’t work, separate out the interferrent. Inductively Coupled Plasma - Mass Spectrometry (ICP-MS)

Principle ICP used as an ion source for a quadrapole mass spectrometer ( a tuneable mass filter with a resolution of 0.5-1.0 amu). A three stage interface is used to couple the plasma at atmospheric pressure to the MS at 10-7 torr.

Ion lenses are cylinders with different positive voltages on them. They re-focus the ion beam. Analytical Advantages 1. Very high sensitivity < 1ppt (pg ml-1) for some elements - partially due to the low background, ≈ 10cps or better compared with analyte count rates of 5x106cps per ppm. 2. Wide element coverage, nearly all periodic table except inert gases can be determined (quadrapole can scan complete mass range in ≈ 50µs - appears to be simultaneous). 3. Very large dynamic range. 4. Capable of isotopic analysis. 5. Simple spectra, only 211 isotopes for all the periodic table. 6. Uniform sensitivity - sensitivity varies by a factor of 2-3 across the periodic table (heavy elements are more sensitive). The mass-bias can be calculated and then semi-quantitative (±10%) “what is in it” analysis carried out following single element calibration, e.g. indium. Disadvantages 1. Interferences are more severe than ICP-OES.

2. Dissolved solids have to be kept to < 1000ppm to avoid blocking the sampling cone. 3. Sample enters the interface and some deposits in the machine, ∴ regular cleaning is required (particular problem for radioactive materials). Try not to have to clean quadrapole - sensitive to alignment. 4. Purchase price (£200k) and running costs are higher than ICP-OES. Why Buy One ? 1. Will do isotopes, not just elements (Nuclear industry) 2. Ultra-trace, multi-element in water (NRA/ British Gas) 3. Rare-earth elements - optical very complex, overlaps, lose sensitivity (geochemistry) 4. Geochemists radioisotope ratios to give age of rocks, e.g. 204Pb is the only natural isotope of lead. 5. Metabolism studies, diets with isotopically enhanced, check urine, blood etc. Interferences 2 types : - spectral - matrix Spectral Interferences 2 main types : Isobaric Interferences - caused by isotopes having very similar m/z ratios that are not resolved by the quadrapole. e.g. 46Ti-46Ca ; 54Fe-54Cr Most serious example is 40Ar-40Ca. 40Ca is the most abundant isotope (96.97%) of Ca. Can use higher resolution (organic) MS, pushes costs up to £300k. - solve by choosing another isotope. Molecular Ion Interferences

3 main causes: (i) at m/z < 80, ions derived from plasma gas Ar, O, N, H e.g. 16O2

+ - 32S Worst is 40Ar16O+ - 56Fe (91.66% abundant). Iron is very important. (ii) Molecular ions derived from the solvent HCl , e.g. 51ClO - 54V (99.76%) 75ArCl - 75As (100%) As mostly of biological importance (poison) - lots of Cl in biological samples. H2SO4 48SO - 48Tl 64SO2 - 64Zn / 64Ni 65SO2 - 65Cu HNO3 (preferred whenever possible) 54ArN - 54Fe+ (iii) Oxide Molecular Ions - Rare earths give rise to very strong oxide bonds, e.g. CeO, but optimising operating conditions reduce them to 1-3% of element signal. e.g. 156CeO - 156Gd Matrix Interferences (i) Matrix Derived Changes in Ion Transmission - high matrix levels → high ion currents, because of space charge effects, ions are ejected from the ion beam. - effect is most serious for a heavy element matrix, e.g. Pb and a light element analyte, e.g. B. (ii) Ionisation Interference - low ionisation potential elements, e.g. Na, K, Cs suppresses the ionisation, particularly at high ionisation potential elements and cause interference.