X-ray Microanalysis zThe fluorescent production of X-rays by electrons is one of the most important interactions available in the SEM because it permits chemical (atomic) identification and quantitative analysis to be performed zAbout 60% of all SEMs are now equipped for X-ray microanalysis

Characteristic X-rays zCharacteristic X-rays are formed by ionization of inner shell electrons. The inner shell electron is ejected and an outer shell electron replaces it. The energy difference is released as an X-ray

X-ray peaks zThe characteristic X-ray signals appear as peaks (lines) superimposed on the continuum, These peaks have fixed energies

Mosleys law zMosley showed that the wavelength of the characteristic X-rays is unique to the atom from which they come zThis is the basis of microanalysis

Wavelength and Energy X-rays can be identified by either their wavelength or by their energy E zThese two quantities are related by this relationship, so either can be used

Mosleys law zK-lines come from 1st shell (1s) zL-lines come from 2nd shell (2s) zM-lines come from 3rd shell (2p) zEach family of lines obeys Mosleys law

K-lines zK-lines are the easiest to identify and highest in energy zGaussian shape K and K come together as a pair

L-lines zOften occur in groups of three or four lines so shape can vary zCan overlap K-lines zImportant for analysis of elements Z>40 Silver L-line cluster

M-lines z... and N- and O - lines are very complex zNot all lines are shown on all analyzer systems so check with standards if in doubt zAvoid use if at all possible ! However at low energies they must be used. Lead and gold are best analyzed with the M lines

Fluorescent Yield zNot all ionizations produce X-rays The fractional yield (the fluorescent yield) is called varies rapidly with atomic number Z and is low for low Z

Measuring X-rays Wavelength Dispersive Spectrometers measure by diffraction from a crystal. Accurate but slow and low sensitivity z Energy Dispersive Spectrometers measure photon energy. Fast, convenient, good sensitivity but has limitations in energy resolution

The Energy Dispersive Spectrometer zA solid state device - Si(Li) P-I-N diode zConverts X-ray energy to charge. The output voltage step is exactly proportional to the deposited X-ray energy zMeasures the photon in about 100microseconds so can process 1000 or more photons/second Bias Window PIN diode Capacitor C Voltage=Q/CXray generates electron/hole pairs (3.6eV / pair) Charge ~ Xray energy

The EDS detector zThe cryostat cools the pre-amp electronics and detector diode zThe window protects the detector from the SEM vacuum, BSE, and visible light zBeware of ground loops, noise (TV monitors), lights in the chamber (the ChamberScope !)

System peaks zX-rays are also produced by electrons hitting the lens, the aperture and the chamber walls. zTo keep these system peaks to an acceptable level a collimator must look at the point where the beam hits the surface. EDS sample Lens aperture Chamber wall

Detector position zThe working distance must be set to the correct value in order to maximize count rate and minimize the systems background z12 mm in the S4700

Deadtime Processing and displaying pulse takes some finite time zMCAs (multi-channel analyzer) only handle one pulse at a time so some pulses will be missed zThis deadtime must be allowed for in quantitative analysis

How much deadtime? zDeadtime increases with count rate (beam current and energy) and process time (set by operator) zValues greater than 25% may allow 2 or more pulses to hit detector at same time giving sum peak. zValues >50% waste time and may cause artifacts

MCA parameters zDuring spectrum acquisition the operator has control of a variety of parameters zThe most important of these are the beam current, which controls the input count rate, and the pulse processing time zThe processing time must be set with care to achieve optimum results

Count throughput zFor spectra choose a low count rate, and a long process time to give best resolution zFor x-ray mapping choose the highest beam current and the shortest process time to give highest throughput

Resolution zThe spatial resolution and depth penetration of a microanalysis is set by beam energy and material zTypically of order of 1 micron but can be much less if E is close to E crit z Monte Carlo models are a valuable aid in understanding the lateral and depth resolution of X-ray microanalysis

Reading the spectrum zGOLDEN RULE - identify the highest energy peaks first zThen find all other family members of this peak i.e the L,M lines zThen identify the next highest energy peak

If a peak cannot be identified.. zIs it a sum peak ? (look for dominant peaks at lower energies, one half of the energy.) zIs it an escape peak ? (look for a strong peak 1.8keV higher in energy) zIs the system calibrated properly? zIs it really a line? - is it of the right width, does it have the right shape, are there enough counts to be sure ? How would we know?

Detectable limits zFor an X-ray line to be statistically valid it must exceed the noise (randomness) in the corresponding background region of the spectrum by a suitably large factor zRule of thumb the peak should be twice the background to be considered valid 2x 5x 10x Visibility and peak height 10x 1x?

Counting statistics zThe signal is equal to the peak integral - background zPoisson statistics apply to the data so the noise estimate = (background) 1/2 zFor 95% confidence we need the signal to be 3 standard deviations over noise zPeak>3.(background) 1/2 Bkg =25 Peak>15 Bkg =100 Peak>30 Bkg =250 Peak>50 i.e minimum size of peak falls as fraction of BKG as count rises

Detection limits zThis statistical limit determines the lowest concentration of an element that might be detectable (MDL - the minimum detectable limit) zFor an EDS system this is typically in the range 1-5% depending on the overall count acquired in the spectrum and on the actual elements involved

Optimizing MDL zCount for as long as possible zSince P/B (peak to background) rises with beam energy use the highest keV possible zSet MCA process time for highest detector energy resolution zMaximize take-off angle where possible zMinimize system peaks, spurious signal

Trace detection ? zEDS is not a trace detection technique - needs a 10x improvement to achieve even parts per thousand level zBut minimum detectable mass (MDM) is very good ( to grams) for this technique zBest with inhomogeneous samples

Low Energy Microanalysis zThe reduction in interaction volume makes possible high spatial resolution microanalysis even from solid samples zLower cps and lower dead times X-ray generation in silicon at 3keV

Microanalytical Performance zCount rates are lower than at conventional beam energies zK lines are better than M lines. L lines are lowest in yield zBeam energy will determine which elements can be analyzed

Elements accessible to X-ray Microanalysis at 10keV

Elements accessible to X-ray microanalysis at 5keV

Practical Problems for Low Energy EDS zAll available lines are in 0-3keV range zThere are more than 60 elemental lines between 0 and 2keV, and more than 30 between 2 and 4keV zSpectrometers with better than 30eV resolution are needed! Distribution of X-ray lines as a function of spectral energy