Atomic Emission Spectrometry

• A Simple, Low Cost Flame Spectrometer
  1. bgsu chemsitry
  2. paul f. endres

Figure 1 shows a simple flame spectrometer that provides reasonably rapid and reliable measurements for K, Na, Li (Cs, Rb.) For these species, analysis in the 1 ppm (1 mg/liter) range is routine and a student can complete a calibration curve and analyze a sample in about 5 minutes.

We have used this over the past two years in General Chemsitry courses and we have been pleased with the perfromance. It has allowed us to introduce additional instrumentation to this lab and it has sound pedagogic links to atomic structure.We use it for K analysis of beverages and foods. We have students synthesize and analyze K3Fe(C2O4) 3H2O and this has provided a fast way to include a K analysis.

• The source is a modified propane torch, purchased at a local hardware store.
  • The air is drawn into the burner through openings in the torch tip
    • We surround that region with a simple arrangement of 3/4" - 1/2" copper plumbing fixtures
  • A lower tee contains an aspirator
    • Ours was taken from an old AA Spectrometer but simple aspirators should work equally well.
    • This draws in sample, nebulizes it, and the tubing delivers the mist to the burner head
    • A source of compressed air operated the nebulizer. (Even a large aquarium pump would work.)
• The light is captured by an optical fiber oriented towards the flame.
  • We use an inverted 10X microscope objective to improve light collection, but it is not necessary.
• The light is analyzed by a miniature array spectrometer.
  • We routinely use an Ocean Optic PC2000 spectrometer and this report generally refers to data collected with that instrument
  • We have also successfully operated this using a CVI Laser PL290 spectrometer.

• The OO PC2000 spectrometer
  • the detector is a CCD array with 2000 pixels
    • these are monitored with a 12 bit ADC so the raw response is a value from 0-4095
  • with a standard ___ line/inch grating this covers the region about 380 to about 850 nm
    • each pixel covers approximately 0.25 nm
    • the resolution (FWHM) is approximately 0.3 nm.
  • this basically integrates the light signal and the sensitivity can be increased by using longer integration times
    • times typically range from 10-1000 msec
    • Since the flame is actually a relatively weak light source we often use integration times of 2500 msec.

We feel that this apparatus offers a number of benefits

• the spectrometer (instrument) and the spectral display are clearly presented and understood by students
• the spectrometer is relatively low cost ($2000 with educational discounts available)
• the spectrometer can be converted from flame spectrometer to a variety of other applications
• the narrow atomic spectral lines are clearly evident
  • this allows a discussion of selectivity and freedom from interference
  • this allows a clear example of electronic transitions from well defined energy levels
• this allows a discussion of why active metals (like the alkali metals) are easily detected while other elements are not.
  • We usually include Cs and Rb for completeness. While these are expensive, so little is used that it is a minor expense. We have little expectation of finding interesting samples to analyze for those species.
  • the spectrum from a light stick may be used for comparison to show the differences between atomic and molecular spectra.
  • a particularly interesting discussion results when one tries to use other elements that produce strongly colored flames (Cu, Ca, Sr.)
  • these show up as very low intensity bands
  • in a propane flame, these species generally emit from excited states of simple oxides
  • these are molecular species and they form multi-line electronic/vibrational bands
  • these individual states are not resolved and no stante has large, dominant intensity

An estimate of the detection limit for each species is included in the figure.

• The software controls the instrument parameters and the collection of spectra
• The spectrum can be plotted and the scale of the graph can be reset
• A cursor allows a digital value for the intensity at a selected wavelength
• Multiple scans can be averaged to reduce noise level.

• We'd like to include multiple plots
  • we'd like one overall spectrum in the background to clearly show the narrow atomic lines and the relatively clean background.
  • note that we usually see a faint Hg spectral background from room fluorescent lights
  • we'd then like to show 2-4 inset graphs corresponding to the individual atomic lines
• we'd like the screen to display Intensity at several selected wavelengths
  • for efficiency, we'd like that data to be stored and made available for calibration runs and sample analysis
• we also need a more comfortable form for printing results including annotation that identifies individual students.

• First, the nebulizer/flame combination produces a fluctuating signal intensity
  • this can be seen, simply looking at the brightness of a weak Na flame
  • the effect is partially overcome by the relatively long integration times (2-3 sec)
  • if one is patient, averaging multiple scans can further reduce the fluctuations.
• One also observes that the instrument sensitivity undergoes a slow decrease, noticeably so over perhaps 10 minutes.
  • this appears, in large measure, to be due to the cooling of the propane in the tank.
    • as gas is used, propane evaporates to maintain the gas pressure
    • this evaporation slowly cools the liquid, reducing the vapor pressure of the propane
    • this appears to slowly reduce the flame intensity
    • this can be minimized by using the slightly larger propane cylinders and minimizing any thermal insulation of the tank. We could operate the torch from a large gas cylinders, but the apparatus becomes less portable and less intuitive
    • also, the use of smaller flames is recommended. This does slightly reduce the signal.
  • a set of samples (standards and unknowns) should be run over a short period or 5-10 minutes. it is not practical to provide a calibration curve to be used throughout the period.
• We have explored a few alternative
  • we have explored the use of natural gas, but the dynamics of the propane torches make for much more efficient introduction of nebulized sample.
  • we have also tried cylinders of MAPP gas which produces a hotter flame. We found no benefit in this application

• Certainly this is not a burner optimized for atomic spectroscopy. The premix slot burners would clearly give more consistent results. We've also explored some older total consumption burners used in ancient Beckmann DU flame spectrometers.

• The calibration graph in Figure 3 shows an intensity ratio of _____ (Na/Li) and ____ (K/Li.)
• By recording both Li and Na/K intensities, one can compute the concentrations of the latter species. This method compensates for fluctuations in nebulization and minor changes in flame intensity.