Atomic Absorption Spectroscopy: Principle and Uses

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-09-20

Atomic absorption spectroscopy (AAS) is a quantitative elemental technique. You dissolve or digest a sample, convert the analyte into free ground state atoms, pass light from an element specific lamp through that atom cloud, and measure how much light the atoms absorb. The absorbance is proportional to concentration over a limited range, so you convert it to a concentration with a calibration curve.

By the end of this guide you will be able to trace the optical path of a bench AAS instrument, explain why the hollow cathode lamp must match the analyte element, build and validate a calibration curve, calculate a result in mg/L (which equals ppm) from an absorbance reading, recognize the three main interference classes, and decide whether a flame or graphite furnace method fits your sample. You will also know why AAS measures one element at a time and when an inductively coupled plasma mass spectrometer (ICP-MS) is the better tool.

On hand before you start: a validated AAS instrument with the correct hollow cathode lamp, high purity acetylene or argon supply as required by the atomizer, matrix matched calibration standards, trace metal grade acids for digestion, and a dilution scheme that keeps every reading inside the linear range.

What AAS Measures and Why It Matters in the Microbiology Laboratory

AAS quantifies an element, not a molecule. In a clinical or research microbiology setting this matters for several recurring jobs. Zinc is a cofactor in many mammalian proteins and acts in cell signaling, and graphite furnace AAS (GF-AAS) is a sensitive way to measure intracellular zinc pools in cultured cells [1]. Platinum measurement in plasma, plasma ultrafiltrate, and whole blood supported pharmacokinetic studies of the anticancer agent enloplatin, using a graphite furnace with a platinum hollow cathode lamp and Zeeman background correction [2]. Copper, iron, zinc, magnesium, and calcium in soil extracts have been measured by AAS to assess available nutrients [3].

The technique is also used well outside microbiology. Lead and cadmium in cereals are measured by graphite furnace AAS, with detection limits of 0.012 mg/kg for Pb and 0.0014 mg/kg for Cd in one automated method [4]. Palladium in soil extracts has been measured by flame AAS after preconcentration, with a detection limit of 6.4 ng/mL [5]. The same physics underlies all of these applications.

The unifying idea is that every element has a unique set of electron energy levels. When a ground state atom absorbs a photon of exactly the right energy, an outer electron jumps to a higher orbital. That absorption happens only at discrete wavelengths called resonance lines. AAS exploits the strongest resonance line for the element of interest.

The Four Core Components of an AAS Instrument

Every AAS instrument, from a basic flame unit to a graphite furnace workhorse, contains the same four functional blocks in the same order. Understanding them in sequence explains nearly every practical problem you will encounter.

1. The Hollow Cathode Lamp Source

The hollow cathode lamp (HCL) is a sealed glass tube with a cathode made of, or containing, the element to be measured, plus an inert fill gas such as neon or argon. A voltage across the electrodes ionizes the fill gas. The ions bombard the cathode and sputter atoms of the cathode element into the gas phase, where collisions excite them. As the excited atoms relax, they emit light at the element's characteristic wavelengths.

The key property is line narrowing. The emission lines from a hollow cathode lamp are extremely narrow, far narrower than the absorption linewidth of atoms in a flame or furnace. That near perfect overlap is what makes AAS selective.

Because the cathode is made of the analyte element, one lamp measures one element. Multi-element lamps exist and combine several elements in a single cathode. A multi-element lamp covering copper, iron, zinc, magnesium, and calcium gave available nutrient results statistically equivalent to single-element lamps for Cu, Fe, Zn, and Ca, with a linear correlation coefficient of 0.86 and a slope of 0.85 for Mg, and cut analysis time by 50 to 60 percent [3]. A composite lead-cadmium hollow cathode lamp allowed simultaneous determination of Pb and Cd in cereals, reaching up to 240 measurements in 8 hours [4]. Even so, multi-element lamps trade some sensitivity for throughput. A simplex optimization study of a simultaneous multi-element flame spectrometer found that multi-element sensitivity was reduced by no more than a factor of two compared with single-element determination [6].

Lamp current is a real tuning parameter. Higher current gives more emission intensity but also broader, self-absorbed lines and shorter lamp life. In one urine method, the calcium lamp was run at 3 mA and the magnesium lamp at 4 mA [7]. In a graphite furnace method for high purity alumina, hollow cathode lamp current was one of the optimized variables alongside pyrolysis and atomization temperatures [8].

2. Atomization: Flame or Graphite Furnace

Atomization is the step that makes AAS possible. The sample must be converted into free, unexcited, ground state atoms in the optical path. Two atomizers dominate.

Flame atomization. A nebulizer pulls liquid sample into a spray chamber, where it mixes with oxidant and fuel gas. The mixture burns at a slot burner head, typically air-acetylene or nitrous oxide-acetylene. The flame desolvates, vaporizes, and dissociates the sample into atoms. Flame AAS is fast, robust, and forgiving of matrix, but the residence time of atoms in the beam is short and the dilution by flame gases is large, so detection limits are modest.

Flame conditions are element specific. In a urine method for calcium and magnesium, the burner height was 8 mm and the air to acetylene ratio was 6 to 1 for both elements [7]. Phosphorus in wine can be determined indirectly by flame AAS using either nitrous oxide-acetylene or air-acetylene, by measuring the molybdenum that complexes with phosphorus [9].

Graphite furnace atomization. Also called electrothermal atomization, this approach places a small aliquot of sample into a graphite tube. The tube is heated in programmed stages: drying, pyrolysis (ashing), atomization, and cleaning. Atoms are generated in a confined volume and remain in the optical path for a much longer time, which drives detection limits down by orders of magnitude and allows very small sample volumes.

The trade-off is speed and matrix tolerance. Graphite furnace runs are slower, one element and one aliquot at a time, and the matrix must be controlled with modifiers and background correction. A graphite furnace method for platinum in biological fluids used Triton X-100 and Antifoam-B as matrix modifiers and Zeeman background correction, and was linear from 0.05 to 10.0 micrograms Pt per mL [2]. A slurry introduction furnace method for copper, iron, and sodium in high purity alumina used Smith-Hieftje background correction, with detection limits of 0.66, 2.5, and 0.13 ng per g respectively and relative standard deviation not more than 5.2 percent [8].

3. The Monochromator

The monochromator isolates the resonance line from everything else the hollow cathode lamp emits. It is a diffraction grating with an entrance and exit slit. You select the wavelength, and you set the spectral bandpass with the slit width.

The wavelength must match the element. Calcium is measured at 422.8 nm and magnesium at 285.2 nm in the urine method described above [7]. Choosing the wrong wavelength gives no signal at all, or a signal from the wrong transition with poor sensitivity.

Slit width controls how much light reaches the detector and how much neighboring emission is admitted. Narrow slits improve selectivity but reduce light throughput and worsen the signal to noise ratio. In the multi-element flame optimization study, slit width was one of seven factors tuned for the best overall response across copper, iron, manganese, and zinc [6].

Modern instruments also apply background correction, which is not part of the classic four component description but is essential in practice. Deuterium lamp correction, Zeeman correction, and Smith-Hieftje correction all serve the same purpose: subtract the broadband absorbance from matrix species so that only the narrow atomic line remains. The alumina method used Smith-Hieftje correction [8] and the platinum method used Zeeman correction [2].

4. The Detector

The detector is a photomultiplier tube. It converts the transmitted light intensity into an electrical signal, which the instrument electronics convert into absorbance.

Absorbance is defined as the negative base ten logarithm of transmittance:

absorbance = log10(I0 / I)

where I0 is the intensity with a blank (no analyte atoms) and I is the intensity with the sample. Absorbance is a dimensionless ratio, which is why it carries no units.

The instrument alternates between measuring the lamp signal and the sample signal, often with a chopper or pulsed lamp, so that flame emission and ambient light are subtracted. This is why flame AAS works at all: the flame emits light, but the detector only counts the modulated lamp signal.

The Measurement Sequence

The optical and physical sequence in a single AAS measurement follows a fixed order. Each step exists to solve a specific problem.

flowchart TD
    [Hollow cathode lamp emits narrow line] --> [Chopper modulates the beam]
    [Chopper modulates the beam] --> [Sample enters atomizer]
    [Sample enters atomizer] --> [Flame or furnace makes ground state atoms]
    [Flame or furnace makes ground state atoms] --> [Atoms absorb part of the beam]
    [Atoms absorb part of the beam] --> [Monochromator isolates resonance line]
    [Monochromator isolates resonance line] --> [Detector measures transmitted intensity]
    [Detector measures transmitted intensity] --> [Electronics compute absorbance]
    [Electronics compute absorbance] --> [Calibration curve converts to concentration]

Beer-Lambert Quantification and the Linear Range

The Beer-Lambert law states that absorbance is proportional to the concentration of the absorbing species and the path length:

A = epsilon times b times c

where A is absorbance, epsilon is the molar absorptivity (a constant for a given transition), b is the path length through the atom cloud, and c is concentration.

In AAS the path length is fixed by the burner or tube geometry, so absorbance becomes directly proportional to the number of ground state atoms in the beam, which is proportional to analyte concentration in the sample.

The critical practical caveat is that Beer-Lambert linearity fails at high absorbance. The law assumes that every absorbing species sees the same light intensity and that the absorbing line is narrower than the source line. Neither holds perfectly. At high atom density, the source line is partially absorbed across its full width, stray light becomes a larger fraction of the transmitted signal, and the calibration curve bends toward the concentration axis. In practice, most AAS methods are linear only up to roughly 1 to 2 absorbance units, and many are linear only well below 1.

The Beer-Lambert law has documented limitations in real measurement conditions, including scattering, chemical equilibria, interference, spectral bandwidth disagreements, stray radiation, and instrumental effects, all of which can cause mistaken or misinterpreted results if ignored [10]. In an optically thick potassium flame, the absorbance became so high that the sample was opaque at the line center, and the authors had to fit full absorption profiles rather than rely on a simple linear relation [11].

The operational rule is simple. If your sample reads above the top calibration standard, dilute it and re-run. Never extrapolate a calibration curve. Dilute into the linear range, then multiply the result by the dilution factor.

Building a Calibration Curve

A calibration curve converts absorbance into concentration. Build it with matrix matched standards, meaning the standards contain the same acid, salt, and modifier composition as your samples.

A typical workflow:

1. Prepare a blank (the matrix without analyte) and at least five standards spanning the expected range.
2. Set the instrument to the correct wavelength, slit width, and lamp current for the element.
3. Run the blank to zero the instrument.
4. Run standards from low to high, recording absorbance for each.
5. Fit a linear regression of absorbance against concentration.
6. Check the coefficient of determination. Values above 0.999 are routine for well behaved methods, as seen in the alumina slurry method where linear regression coefficients were better than 0.999 [8].
7. Verify with a certified reference material or a spiked recovery sample.

Recovery is the standard accuracy check. The urine method reported recovery of 96 to 104 percent with a relative standard deviation of 1.8 percent [7]. The palladium soil method reported recovery of 90 to 101 percent [5]. The cereal method reported recoveries of 80 to 110 percent across eight laboratories compared with ICP-MS [4].

The calibration slope carries units of absorbance per concentration unit, for example absorbance per mg/L. That slope is the conversion factor for every subsequent sample.

Worked Example: Converting Absorbance to ppm

Suppose you are measuring copper in a digested water sample by flame AAS. Your calibration standards were prepared in 2 percent nitric acid, and the regression of absorbance against concentration gave the following fit.

Calibration standards (Cu, mg/L) and measured absorbance:

  0.00 mg/L   A = 0.000
  0.50 mg/L   A = 0.061
  1.00 mg/L   A = 0.122
  2.00 mg/L   A = 0.243
  5.00 mg/L   A = 0.607

Linear regression:  A = 0.1214 * C + 0.0009
Coefficient of determination R^2 = 0.9998

Your unknown sample reads an absorbance of 0.412. You also note that you diluted the digest 1 in 10 with 2 percent nitric acid before running it.

Step 1. Confirm the reading is inside the calibration range. The highest standard was 5.00 mg/L with absorbance 0.607. Your reading of 0.412 sits between the 2.00 and 5.00 mg/L standards, so it is inside the linear range and no further dilution is needed.

Step 2. Rearrange the regression equation to solve for concentration.

A = 0.1214 * C + 0.0009

C = (A - 0.0009) / 0.1214

C = (0.412 - 0.0009) / 0.1214
C = 0.4111 / 0.1214
C = 3.386 mg/L

Step 3. Correct for the dilution factor. The instrument measured the diluted digest, so the original digest concentration is higher by the dilution factor of 10.

Concentration in original digest = 3.386 mg/L * 10
                                = 33.86 mg/L

Step 4. State the result in the correct units. One milligram per liter equals one part per million in a dilute aqueous solution, because 1 L of dilute aqueous solution weighs approximately 1 kg, and 1 mg divided by 1 kg is 1 part per million.

Result: 33.9 mg/L Cu  =  33.9 ppm Cu

Step 5. Sanity check. The result is roughly 6.8 times the highest standard before dilution correction, which is consistent with a 10 fold dilution bringing it into range. A spike recovery on a duplicate aliquot should fall between about 90 and 110 percent for a well controlled flame method.

If the absorbance had been 0.950 instead, that would sit above the 5.00 mg/L standard and the sample would need a further dilution before re-analysis. Extrapolating the slope to 0.950 would give about 7.8 mg/L, but that value would be unreliable because linearity is not guaranteed above the calibrated range.

Matrix Interferences and How to Control Them

AAS is selective but not immune to matrix effects. Interferences fall into three classes, and each has a specific fix.

Chemical Interference

A chemical interference occurs when the analyte forms a thermally stable compound in the atomizer that does not fully dissociate into free atoms. The classic example is calcium in the presence of phosphate, sulfate, or silicate, which form refractory calcium salts in the flame and depress the calcium signal.

The fix is a releasing agent or a protective chelating agent. Lanthanum chloride was tested as a releasing agent for calcium in urine and performed better than strontium chloride [7]. Lanthanum preferentially binds the phosphate, freeing calcium to atomize. In graphite furnace work, a chemical modifier such as palladium or magnesium nitrate serves a similar purpose by stabilizing the analyte until the atomization step.

Ionization Interference

An ionization interference occurs when the flame or furnace is hot enough to strip an electron from the analyte atom, producing an ion that no longer absorbs at the resonance line of the neutral atom. This is most common for elements with low ionization energy, such as the alkali metals, in hot nitrous oxide-acetylene flames.

The fix is an ionization suppressant, typically an excess of an easily ionized element such as potassium or cesium. The suppressant floods the flame with electrons and shifts the equilibrium back toward neutral atoms.

Physical Interference

A physical interference arises from differences in viscosity, surface tension, or density between samples and standards, which change the rate at which the nebulizer aspirates and delivers sample to the flame. High dissolved solids, high acid concentration, or high organic content all cause this.

The fix is matrix matching. Prepare standards in the same acid and salt background as the samples. If matrix matching is impossible, use standard additions, in which known amounts of analyte are spiked into the sample itself and the slope of the resulting response is used to calculate the original concentration.

Background Absorption

Background absorption is a fourth effect that is often grouped with the others. It comes from molecular species, smoke, or scattering particles in the atomizer that attenuate light across a broad wavelength range. It is worst in graphite furnace work with complex biological or environmental matrices.

The fix is background correction. Zeeman correction, deuterium lamp correction, and Smith-Hieftje correction all subtract the broadband component. The platinum method in biological fluids relied on Zeeman correction [2], and the alumina method relied on Smith-Hieftje correction [8].

Modifiers in Graphite Furnace Work

Graphite furnace methods depend heavily on chemical modifiers, which are reagents added to the sample or the standard before atomization. They serve three purposes: they stabilize the analyte during pyrolysis so that higher ashing temperatures can be used to remove matrix, they convert the matrix into a more volatile form that leaves before atomization, and they improve the consistency of atomization.

Common modifiers include palladium nitrate, magnesium nitrate, ammonium dihydrogen phosphate, and mixtures of these. In biological matrices, Triton X-100 is added to lyse cells and reduce surface tension, and an antifoaming agent prevents foaming during the drying stage. The platinum method used exactly this combination, Triton X-100 plus Antifoam-B, in plasma, plasma ultrafiltrate, and whole blood [2].

Flame Versus Graphite Furnace: Choosing the Right Atomizer

The two atomizers serve different niches. The table below summarizes the practical differences.

The detection limit figures in the table are not directly comparable because they come from different methods with different preconcentration steps. The palladium figure of 6.4 ng/mL was achieved by flame AAS only after dispersive solid phase microextraction with magnetic nanoparticles, plus a slotted quartz tube to increase atom residence time in the beam [5]. Without preconcentration, flame detection limits for palladium would be far higher. This illustrates a general point: flame AAS can reach trace levels when combined with a preconcentration step, but the extra chemistry adds time and error.

Graphite furnace methods reach low detection limits directly. The cereal method achieved 0.012 mg/kg for Pb and 0.0014 mg/kg for Cd with a diluted acid extraction and no preconcentration column [4]. The alumina slurry method achieved 0.66, 2.5, and 0.13 ng per g for copper, iron, and sodium respectively [8].

A slotted quartz tube fitted over a flame burner head is a middle path. It traps atoms in the optical path longer, boosting absorbance without the full complexity of a graphite furnace. It has been used for palladium [5] and for cobalt in a vitamin B12 method coupled to liquid chromatography [12].

Practical Method Development Notes

Wavelength and Lamp Selection

Match the lamp to the element. Calcium at 422.8 nm and magnesium at 285.2 nm are standard choices [7]. The lamp current should be set to the manufacturer's recommendation for the element and adjusted during optimization if sensitivity or noise is poor [8].

Background Correction

Choose the correction mode based on matrix complexity. Deuterium correction is adequate for simple flame methods. Zeeman correction is preferred for graphite furnace work with heavy matrices because it corrects structured background more effectively [2]. Smith-Hieftje correction is an alternative that uses the lamp itself to generate the correction signal [8].

Slurry Introduction

Slurry introduction avoids the acid digestion step for solid samples. The alumina method directly determined copper, iron, and sodium in high purity alumina by slurrying the powder and injecting it into the furnace, with calibration against aqueous standards [8]. This saves time and reduces contamination risk, but it demands that the slurry be homogeneous and that particle size be small enough to atomize reproducibly.

Hyphenation

AAS can be coupled to separation techniques. A nebulizer interface linked high performance liquid chromatography to flame AAS for cobalt determination in vitamin B12, with a linear range of 4.7 to 92 mg/kg as Co and a detection limit of 1.6 mg/kg [12]. This kind of hyphenation adds speciation capability, which is valuable when total element concentration is less informative than the chemical form.

Automation

Automated graphite furnace systems can process large batches with minimal operator input. The cereal method automated the entire workflow from sample weighing to result output, including diluted acid extraction, and reached up to 240 measurements in 8 hours [4].

Common Mistakes and Limitations

Extrapolating above the calibration range. Beer-Lambert linearity fails above roughly 1 to 2 absorbance units, and often earlier. A sample reading above the top standard must be diluted and re-run. Never multiply the slope by an out of range absorbance.

Forgetting the dilution factor. The single most common calculation error. Always record every dilution and multiply the final result by the cumulative factor.

Using the wrong lamp. A copper lamp will not measure zinc. Multi-element lamps exist but trade sensitivity for convenience [3][6].

Ignoring matrix mismatch. Standards prepared in water and samples prepared in strong acid will give different nebulization efficiency and different absorbance for the same concentration. Match the matrix or use standard additions.

Skipping background correction in graphite furnace work. Complex matrices produce broadband absorption that inflates the apparent analyte signal. Use Zeeman, deuterium, or Smith-Hieftje correction as appropriate [8][2].

Overlooking chemical interference. Calcium in the presence of phosphate is the textbook case. Add a releasing agent such as lanthanum chloride [7].

Assuming AAS is multi-element like ICP-MS. A standard AAS instrument measures one element at a time. Multi-element lamps and simultaneous spectrometers exist but are specialized configurations [4][3][6].

Neglecting lamp warm-up. Hollow cathode lamps drift during the first minutes of operation. Allow the recommended warm-up time before calibrating.

Contamination from labware and reagents. Trace metal analysis at ppb levels demands acid washed glassware and trace metal grade acids. A dirty tube or a contaminated blank will ruin the calibration.

Forgetting that AAS measures elements, not molecules. AAS cannot distinguish oxidation states or chemical forms without a separation step coupled upstream [12].

When AAS Is the Right Tool and When It Is Not

AAS is the right tool when you need quantitative elemental analysis of one or a few elements at ppm to ppb levels, when the instrument budget is limited, when the method is well established, and when sample throughput requirements are moderate.

AAS is the wrong tool when you need dozens of elements in a single run, when you need isotope ratios, when you need speciation without a coupled separation, or when you need the lowest possible detection limits across many elements simultaneously. In those cases ICP-MS is the standard alternative. The cereal method for lead and cadmium was validated against ICP-MS and showed no significant difference by t-test, with relative standard deviations under 5 percent for both elements and recoveries of 80 to 110 percent across eight laboratories [4]. That comparison shows that a well optimized AAS method can match ICP-MS for a small, targeted element panel, at lower cost and with simpler instrumentation.

For a microbiology laboratory measuring zinc in cell lysates, copper in media, or platinum in pharmacokinetic samples, AAS remains a practical and defensible choice. For a laboratory screening twenty elements in a single digest, ICP-MS is the better fit.

Frequently Asked Questions

What is atomic absorption spectroscopy in simple terms?

AAS measures how much light a cloud of free atoms absorbs at a wavelength specific to one element. The more atoms present, the more light is absorbed, and the absorbance converts to concentration through a calibration curve.

Why does the hollow cathode lamp have to match the element being measured?

The lamp cathode is made of the analyte element, so it emits that element's narrow resonance lines. Only those lines are absorbed efficiently by the sample atoms, which is what gives AAS its selectivity.

What is the difference between flame and graphite furnace AAS?

Flame AAS aspirates liquid sample into a burner flame and is fast and robust with ppm detection limits. Graphite furnace AAS heats a small aliquot in a graphite tube and reaches ppb or lower detection limits with much smaller sample volumes but slower runs.

Why can I not read a sample above the calibration range?

Beer-Lambert linearity breaks down at high absorbance, typically above about 1 to 2 absorbance units. Above that point the curve bends and the slope no longer predicts concentration accurately, so the sample must be diluted.

What is the difference between mg/L and ppm?

For dilute aqueous solutions, 1 mg/L equals 1 ppm. This is because 1 liter of dilute aqueous solution weighs approximately 1 kilogram, and 1 mg divided by 1 kg is one part per million.

What are the three main types of interference in AAS?

Chemical interference from refractory compound formation, ionization interference from loss of electrons in hot flames, and physical interference from differences in viscosity or surface tension between samples and standards. Background absorption is a fourth effect controlled by correction systems.

What is a chemical modifier used for in graphite furnace AAS?

A modifier stabilizes the analyte during the pyrolysis stage, helps volatilize the matrix before atomization, and improves reproducibility. Common examples include palladium nitrate, magnesium nitrate, and ammonium dihydrogen phosphate.

Can AAS measure more than one element at a time?

A standard AAS instrument measures one element per run. Multi-element hollow cathode lamps and simultaneous spectrometers exist and can measure a few elements together, but with some loss of sensitivity compared with single-element determination.

Related Articles

• Single-Use Bioreactors: Principles, Applications, and Best Practices
• Nano-300 Nanodrop: Principle, Use, and Pitfalls for Students
• Ab Initio Protein Structure Prediction: Principles, Challenges, and When to Use It
• Canine Gastrointestinal Physiology: Motility, Secretion, and Absorption
• Integrating Raman Spectroscopy into Routine Microbial Identification Workflows
• Biosafety Cabinet Principles: How They Work and How to Use Them Effectively
• Hardy-Weinberg Law: Principle, Equation, and Uses
• Meclizine Side Effects and Uses in Pets
• Danio Rerio: Zebrafish Biology and Research Use

Sources

1. Atomic Absorbance Spectroscopy to Measure Intracellular Zinc Pools in Mammalian Cells.
2. Analysis of enloplatin by liquid chromatography and of platinum by atomic absorption spectrometry in various biological fluids.
3. [[Determination of soil available nutrient contents using multi-element hollow cathode lamp].](https://pubmed.ncbi.nlm.nih.gov/21942054/)
4. Rapid, Simultaneous, and Automatic Determination of Lead and Cadmium in Cereals with a New High Performance Composite Hollow Cathode Lamp Coupled to Graphite Furnace Atomic Absorption Spectrometry.
5. Determination of palladium in soil samples by slotted quartz tube-flame atomic absorption spectrophotometry after vortex-assisted ligandless preconcentration with magnetic nanoparticle-based dispersive solid-phase microextraction.
6. Optimization of a simultaneous multi-element atomic absorption spectrometer.
7. [[Simultaneous determination of calcium and magnesium in urines by flame atomic absorption spectrometry].](https://pubmed.ncbi.nlm.nih.gov/15766134/)
8. [[Direct determination of copper, iron and sodium in high-purity alumina by slurry introduction furnace atomic absorption spectrometry with Smith-Hieftje background correction].](https://pubmed.ncbi.nlm.nih.gov/21428097/)
9. Phosphorus in wine: comparison of atomic absorption spectrometry methods.
10. Beer-Lambert law for optical tissue diagnostics: current state of the art and the main limitations.
11. Tunable Diode Laser Atomic Absorption Spectroscopy for Detection of Potassium under Optically Thick Conditions.
12. Combination of high performance liquid chromatography and flame atomic absorption spectrophotometry using a novel nebulizer interface supported T shaped slotted quartz tube for the determination of Vitamin B12.