Metal Analysis by Optical Emission Spectroscopy
Metal Analysis by Optical Emission Spectroscopy
Optical emission spectroscopic techniques originated in experiments performed in the mid 1800s, yet they remain some of the most useful and flexible means of performing elemental analysis. Free atoms emit light at a series of narrow wavelength intervals, when placed in an energetic environment. These intervals, termed emission lines, form a pattern known as the emission spectrum, which is the characteristic of the atom producing it. The intensities of the lines are normally proportional to the number of atoms producing them. The presence of an element in a sample is indicated by the presence in light from the excitation source of one or more of its characteristic lines. The concentration of this element can be determined by measuring line intensities. Thus, the characteristic emission spectrum forms the basis for qualitative elemental analysis, and the measurement of intensities of the emission lines forms the basis of quantitative elemental analysis. The emission spectrums for iron ad lithium are shown in Fig 1.
Fig 1 Emission spectrum for iron and lithium
Optical emission spectroscope also known as optical emission spectrometer is used normally for (i) quantitative determination of major and trace-elemental constituents in various sample types, and (ii) qualitative elemental analysis. Examples of application include (i) rapid determination of concentrations of alloying elements in steels and other alloys, (ii) elemental analysis of geological materials, (iii) determination of trace impurity concentrations in semi-conductor materials, (iv) wear metals analysis in oils, (v) determination of alkali and alkaline earth concentrations in aqueous samples, and (vi) determination of calcium in cement.
Samples are in the form of conducting solids (arcs, sparks, and glow discharges), powders (arcs), and solutions (flames). The sample size depends on specific technique varying from around 0.000001 grams to several grams. Sample preparation is done by machining or grinding (metals), dissolution (for flames), and digestion or ashing (organic samples).
Limitations of the optical emission spectroscopic techniques are (i) some elements are difficult or impossible to determine, such as nitrogen, oxygen, hydrogen, halogens, and noble gases, (ii) sample form is to be compatible with specific technique, and (iii) all methods provide matrix-dependent responses. Estimated analysis time ranges from 30 seconds to several hours, depending on sample preparation requirements.
Capabilities of related techniques include (i) x-ray fluorescence is for bulk and minor constituent elemental analysis and needs sophisticated data reduction for quantitative analysis and is not useful for light elements (atomic number 9), (ii) inductively coupled plasma (ICP) emission spectroscopy is for rapid quantitative elemental analysis with parts per billion detection limits, with samples to be in solution and it is not useful for hydrogen, nitrogen, oxygen, halides, and noble gases, (iii) direct-current plasma emission spectroscopy is similar in performance to ICP emission spectroscopy, and (iv) atomic absorption spectroscopy is a single-channel technique, inefficient for multi-element analysis but has favourable sensitivity and precision for most elements.
In a broader sense, the optical emission spectroscopy includes ICP optical emission spectroscopy, which uses an ICP as the excitation source. The terms ‘optical emission spectroscopy’ and ‘ optical emission spectroscopy’, however, normally refer to optical emission spectroscopy using spark discharge, direct current arc discharge, glow discharge, or flame source for generating the excitation discharge. In this article, optical spectroscopy with spark discharge is discussed, since it is being used in the steel industry.
Many optical emission spectroscope feature ‘Pulse Distribution Analysis’ (PDA) to enhance the measurement reproducibility (accuracy). This method involves statistical processing of the spark pulse-generated emission spectra obtained from spark discharges in an argon atmosphere. The optical emission spectroscope offers rapid elemental analysis of solid metal samples, making it indispensable for quality control in the steelmaking processes
The characteristic spectrum an atom produces reflects the electronic structure of the atom. Changes in the energy of the valence or outer shell electrons result in the atomic lines used in emission spectroscopy. Each atom has a ground state in which all of its electrons occupy positions of minimum potential energy. As an atom absorbs energy, one or more of the outer electrons can be promoted to higher energies, producing an excited state. The energy of an atomic state is a function of the energies of the individual electrons and of energy changes resulting from interactions among the electrons.Each possible combination of electron configurations produces a spectroscopic term which describes the state of the atom. Fig 2 shows the principle of emission spectrum taking an example of a lithium atom.
Fig 2 Principle of emission spectrum
Electron energy levels – The simplest atoms, such as hydrogen and the alkali metals, have only one electron outside a filled shell. The simple electron configurations of these atoms produce several possible terms. Atomic emission lines result when the atom undergoes a spontaneous transition from one excited state to another lower energy state. Not all possible combinations of states produce emission lines. Only transitions obeying quantum mechanically derived selection rules occur spontaneously. Diverse factors control the relative intensities of the lines. Those transitions between a low excited state and the ground state, termed resonance transitions, generally yield the most intense emission.
The energy of the excited electron increases with decreasing spacing between excited states until it reaches an ionization limit. At this point, the electron is no longer bound to the atom and can assume a continuous range of energies. Such unbound electrons can undergo transitions to bound states. Since the upper state of the transition is not limited to discrete values, the light from such transitions is spread continuously over a range of wavelengths.
The ionization limit for the atom corresponds to the ground state of the singly charged ion. Excitation of the remaining bound electrons yields a new term system and a new set of lines. Ionization and excitation can continue until an atom is completely stripped of its electrons. In practical emission sources, ionization rarely proceeds beyond removal of two electrons, and in most cases, only the first stage of ionization need be considered. However, a line from the first ion spectrum is normally used in analysis instead of a neutral atomic line.
Spectral overlap – The use of atomic emission for elemental analysis needs measurability of the emission intensity from a line of interest independent of overlapping emission from other species in the sample. The probability of undesired overlap depends on the number of lines in the spectrum and on the wavelength spread or line width of each transition. If all atomic term systems are as simple as that shown for lithium in Fig 2, the probability of spectral overlap is to be low. However, lithium is one of the simplest atoms.
Atoms with more complex electronic structures produce correspondingly complex emission spectra. The iron spectrum (shown in Fig 1), exemplifies such spectral complexity. The spectrum from one ionization stage of a single element can, given sufficient excitation energy, consist of hundreds of emission lines. The complexity is compounded when several elements are present in a sample, each generating neutral and ionic spectra.
Line broadening – Spectral complexity is not to be a problem if, in practice, each emission line are strictly monochromatic and instruments are available with infinite spectral resolution. The energy associated with an electronic term is not defined exactly, but spread over a range of values. The uncertainty in the energy levels appears in the emission spectrum as wavelength broadening of the emission lines. Several factors dictate the magnitude of the energy spread. The most important for emission spectroscopy are frequent collisions of the emitting atom or ion with other species in the excitation source and placement of the emitter in an inhomogeneous electric field.
The first type of line broadening is collisional broadening while the second is the Stark broadening. A third type, Doppler broadening, results from motion of the emitting species relative to the device detecting the emission. For fixed transition energy, the emission recorded from an atom moving toward the detector is at shorter wavelengths than that recorded from an atom at rest. The emission from an atom moving away from the detector is at longer wavelengths. The relative magnitude of these three line-broadening contributions depends strongly on the type of source exciting the emission. The collisional contribution to line width is mainly a function of source pressure. The Doppler contribution for a given element depends on source temperature. The magnitude of the Stark contribution depends on the density of the charged species near the emitter.
Self-absorption – Atomic line profiles produced by any of the above effects can be altered by self-absorption. At high concentrations of atoms in the spectroscopic source, the probability is reasonable which the radiation an atom emits is absorbed by another atom of the same type. The probability of absorption is greater at wavelengths near the centre of the line profile than at wavelengths near the wings. The emission profiles observed under such conditions are flatter and broader than those observed in the absence of self-absorption. If the absorbing atoms are at lower temperatures than the emitting atoms, a line profile is similar to that shown in Fig 3. The Doppler absorption profile of low temperature absorbers is narrower than the emission profile of the hotter emitters. This is called self-reversal.
Fig 3 Emission profile of a self absorbed line and partial cross section of a plane diffraction grating
Molecular emission – The energetic emitting volume of a spectroscopic source can contain small molecules in addition to free atoms. Like the atoms, the molecules produce optical emission which reflects change in the energies of the outer electrons of the molecule. Unlike the atoms, the molecules have numerous vibrational and rotational levels associated with each electronic state. Each electronic transition in the molecule produces an emission band composed of individual lines reflecting the vibrational and rotational structure of the electronic states involved in the transition.
Molecular bands appear in a recorded spectrum as intense edges, out of which develop at higher or lower wavelengths less intense lines with a spacing which increases with distance from the edge. The edge is the band head. Composed of many closely spaced lines, molecular bands can dominate a region of the spectrum, complicating detection of emission from other species in that region. Emission sources are frequently designed to minimize molecular emission. Less often, band intensities are used in place of atomic line intensities to measure concentration.
Atomic emission is analytically useful only to the extent that the emission from one atomic species can be measured and its intensity recorded independent of emission from other sources. This detection and quantification needs high resolution wavelength-sorting instrumentation. Further, before the light can be sorted, it is to be collected efficiently, sometimes only from an isolated region in a spatially heterogeneous emission source.
Wavelength sorting instruments – The key element in modern wavelength-sorting instruments is the diffraction grating, a precisely shaped reflective surface having many closely spaced parallel grooves. A partial cross section of a diffraction grating is shown in Fig 3. Parallel rays of light strike adjacent grooves on the grating. The incident rays are in phase with each other. The rays scattered from the grating have traversed different paths. The difference in the path lengths is AB + BC.
At angles producing a path difference that is an integral number of wavelengths, the exiting rays are in phase, and light is diffracted at that angle. At other angles, the exiting rays are out of phase, and destructive interference occurs. The angles at which diffraction takes place for a given wavelength can be determined by noting that AB = d sin x and BC = d sin y where d is the diffraction grating groove spacing, x is the angle of incidence, and y is the angle of diffraction. The diffraction condition is given by equation m.lambda = d.(sin x +/- sin y). The minus sign enters when the incident and diffracted beams are on opposite sides of the grating normal.
Two types of wavelength-sorting devices (Fig 4) are normally used for the emission spectroscopy. The first, the grating monochromator, is used for single-channel detection of radiation. Fig 4 shows the light path through a Czerny-Turner monochromator, a typical configuration. Light enters the monochromator through the entrance slit and passes to the collimating mirror. The collimated light strikes the plane diffraction grating and is diffracted at an angle dependent on its wavelength. Some of the light is diffracted at angles such that it strikes the focusing mirror. It is then focused to form an array of entrance slit images in the focal plane of the monochromator. The position in the array of a slit image depends on the angle at which the light that forms it exited the grating. The wavelength of the image centered on the exit slit is given by the equation m. lambda = 2d.sin q.cos p where q is the angle through which the grating is rotated, and p is the instrument angle and is the angle which a line through the centre of the grating and the centre of either mirror makes with the centerline of the instrument. The relationships between q and p and the angles x and y used in first equation are shown in Fig 4. As the grating is rotated, images from different wavelengths pass sequentially through the exit slit and are detected by a photomultiplier tube.
Fig 4 Wavelength sorting device
The second general type of wavelength sorter is the polychromatic. Most polychromators are variations on the Rowland circle mount (Fig 4). The diffraction grating is concave, with a radius of curvature R. If an entrance slit is located on a circle of radius R/2 tangent to the grating face, the diffracted images of the slit are focused around the circle. Exit slits and photomultiplier tubes can be placed at positions on the focal curve corresponding to wavelengths of lines from various elements. Line intensities from 40 to more than 60 elements, depending on instrument capability, can be determined simultaneously.
Alternatively, a strip of film or a photographic plate can be positioned in the focal curve in place of the slits and photomultiplier tubes, converting the polychromator into a spectrograph. An entire emission spectrum can be recorded in a short time on a plate or piece of film. Photographic detection allows more flexibility in line selection and provides more information than the combination of fixed slits and photomultiplier tubes. However, the time needed to process the photographic medium, locate the lines of interest, and record their intensities makes the use of photographic instruments tedious. Advancements in the data acquisition and processing capabilities of computerized polychromators are edging spectrographic instruments into disuse.
Collection optics for a spectroscopic instrument transfers radiant power from the source to the detector with maximum efficiency and resolve or, in some cases, scramble spatial heterogeneities in the emission from the source. The first requirement is met if radiation from the source fills the entrance slit and the collimating optic of the spectrometer. A simple lens of suitable size can be used to image the source on the entrance slit with sufficient magnification to fill it. The size of the lens is selected such that radiation passing through the slit just fills the collimating optic. The entrance slit, then, defines the area of the source viewed by the system, and any source non-uniformity within that area is transferred to the detector. Photographic detection frequently needs spatial uniformity of the slit images. The desired uniformity is achieved if the source is imaged onto the collimating optic by a lens near the slit. Other lenses are then used to generate an intermediate image of the source at an aperture to provide spatial resolution.
An emission light source is to decompose the sample from some easily prepared form into an atomic vapour, and then excite the vapour with sufficient efficiency to produce a measurable emission signal from the sample components of interest. Each of the four types of emission sources (arcs, high-voltage sparks, glow discharges, and flames) has a set of physical characteristics with accompanying analytical assets and liabilities.
Excitation mechanisms – The property of an emission source most closely linked to its excitation characteristics is temperature. Temperature indicates the amount of accessible energy in the source. Since energy can be partitioned variously among different species, different temperatures can reflect that partitioning. Gas kinetic temperature and electron temperature indicate the kinetic energies of heavy particles and electrons, respectively. Excitation and ionization temperatures reflect the electronic energy content of atomic and molecular species.
In addition, molecules store energy in rotational and vibrational modes, which is expressed as vibrational and rotational temperatures. In many source environments, excess energy in one mode is rapidly exchanged, or transferred, to another. In such cases, all the above temperatures are equal, and the source is in local thermodynamic equilibrium (LTE). When LTE exists, excitation conditions can be described without an understanding of the microscopic mechanisms of energy transfer. The population distribution among the possible excited states for a given species is given by the Boltzmann equation.
When LTE does not exist, then a complete description of excitation in such cases is to account for microscopic collisional processes which can excite or de-excite a given energy level with an efficiency far different from which predicted using LTE. For example, in low-pressure discharges, a small portion of the electron population can have a temperature far higher than the gas temperature in the discharge. These fast electrons can produce highly excited atoms or ions in much greater numbers than generated under LTE conditions. The excitation efficiency in non-LTE sources frequently depends on close matches in the kinetic or internal energy of colliding species and hence displays sharp variations as the chemical composition of the excitation region changes.
Ideal emission source – The ideal emission source samples all materials efficiently regardless of form and delivers vapour to the excitation zone with a composition directly proportional to the sample composition. Excitation is uniformly efficient for all elements. It produces simple spectra, with all the excitation energy concentrated in a few excited states. The source generates no background spectrum. Hence, the analytical results for equal concentrations of elements in two samples are identical, regardless of differences in the concentration of other sample constituents. That is, sampling and excitation have no matrix dependence.
Spark sources – The high-voltage spark is an intermittent electrical discharge characterized by operating voltages sufficient to cause spontaneous breakdown of an analytical gap and high currents resulting from capacitively stored energy in the discharge circuit. Fig 5 shows a controlled waveform spark source consisting of a high-voltage charging circuit, an inductor capacitor tank circuit with a high-voltage switch, and wave shaping circuitry incorporating the analytical gap.
Fig 5 Controlled waveform spark source
The circuit generates series of identical spark discharges with precise control over current magnitude and direction as well as discharge duration. In practice, the charging section of the circuit is simply a high-voltage transformer and full-wave rectifier. The spark is triggered at times delayed from the zero-crossing of the alternating current (AC) charging waveform selected to produce the same capacitor voltage at the beginning of each discharge. The trigger is usually a hydrogen thyratron or a high-voltage silicon-controlled rectifier. For a given discharge voltage, the relative values of the inductances and capacitance dictate the shape and amplitude of the current waveform in the tank and waveshaping sections of the circuit. For analytical operation, the component values are normally selected to provide a unidirectional discharge current with peak amplitudes from 50 A to 200 A and durations from 50 microseconds to 150 microseconds.
Analysis limitations – The analytical spark gap consists typically of a tungsten pin anode and a cathode of the material to be analyzed. Since the sample forms one of the electrodes, analysis by spark emission spectroscopy is limited to samples which are conductive or can be made so. The analysis is normally carried out in an inert atmosphere, which is provided in a closed chamber or as a flowing sheath of gas. Unless stabilized, the individual sparks in a train strike different locations on the sample electrode, generating a several millimeter wide burn pattern on a planar sample. With a sheath of argon flowing from the anode to the cathode, the sparks strike much more reproducibly, and the burn area is reduced by a factor of ten.
When a spark strikes the sample electrode, rapid local heating ejects electrode material into the spark gap. In an un-stabilized spark, the trajectory of the ejected material is random. In a stabilized spark, the material propagates upward through the gap as an expanding cylinder about the inter-electrode axis. In either case, the vapour is subjected to various excitation conditions during a single spark. It first passes through the energetic cathode spot, where it can undergo several stages of ionization. As it continues upward, the vapour in the current-conducting spark channel remains highly excited the vapour removed from the inter-electrode axis experiences much less energetic conditions. The current in the discharge increases and decreases, altering the excitation conditions markedly, coincident in time with the movement of sample vapour through the gap.
The temporal and spatial inhomogeneity of the spark precludes its characterization in terms of an excitation temperature. Emission is generated at different times and places from several stages of ionization of the sample material and the atmospheric gas. The dominant form of emission is frequently the first ion spectrum. Lines from the singly charged ion are traditionally termed spark lines.
Changes in spark emission occur within micro-seconds. The emission from a spark train also varies within minutes. This long-term change in emission intensity, the sparking-off effect is mainly a reflection of change in the sample electrode caused by repeated sparking at its surface. Chemical and physical changes in the electrode contribute to the sparking-off effect. Hence, the exact nature of the sparking-off curves depends strongly on experimental conditions.
Spark source parameters (capacitance, voltage, inductance, and repetition rate), sample composition, sample phase structure, sample surface condition, sparking atmosphere, and burn area are to be considered in describing sparking-off behaviour. Of particular importance are the dependencies on sample composition and phase structure, indicating that the emission results for an element are strongly matrix dependent. This is important when the spark is used as an analytical emission source.
Minimizing undesirable characteristics – Diverse procedures have been adopted in spark analysis to minimize the effect of source non idealities. Sparking-off effects are traditionally dealt with by recording emission only after the largest intensity changes associated with a burn have occurred. Light to the detector is blocked during a preburn period typically lasting around 1 minute, during which the spark conditions the fresh electrode surface. If the spark is unstable positionally and samples a large area, the emission following the preburn period, for most elements, remains fairly constant for the 30 seconds needed to record an emission spectrum.
Positionally stable sparks generate sparking-off curves compressed in time compared to those produced by unstable discharges. Instead of increasing to a steady state value, the emission intensity increases to a maximum within a few seconds, then decreases to a relatively low value. The emission becomes progressively more erratic as the burn continues. The emission peak during the first 2 minutes of the burn contains information on the concentration of an element in a sample and the type of matrix in which the element is found.
An additional compensation for the nonideality of the spark source is use of intensity ratios rather than unmodified emission intensities to indicate elemental concentration. The intensities of lines from minor constituents are ratioed with intensities from a major matrix constituent. For example, in steel analysis, the line intensities from the alloying elements are ratioed with the intensity of an iron line. This procedure compensates somewhat for variations in sampling and excitation efficiency from one sample to another. It involves the implicit assumption that sampling and excitation of the reference component represent the same processes for the minor constituents. This cannot always be the case, especially if the sample contains inclusions which differ significantly in concentration from the bulk of the sample.
The above measures do not produce satisfactory analytical results unless the spark spectrometer is calibrated using standards which match the unknown sample closely in chemical composition and physical form. Laboratories using the spark for analysis are to have sets of standards for each type of material to be analyzed. Standards for spark investigations are not easily produced and generally are to be purchased.
Sampling and sample preparation
Spark source excitation is the most rapid method of obtaining an elemental analysis of a metal or alloy. Samples are taken from liquid steel, thin sheet, semi-finished or finished products. Speed is critical in the steel industry, for which a melt from a production furnace is to be sampled and analyzed and the concentrations of alloying elements adjusted to within predetermined ranges.
When sampling liquid steel either a sampling probe is used or small quantities of liquid steel collected by a sampling spoon is poured into a casting mould. The sample is cooled. Unnecessary parts of the sample are removed using a grinder or a cutter. If trace elements are not being investigated, a grinder or belt sander can be used. A flat surface is ground or sanded on the cooled sample then placed with no additional preparation into a spark source. However, to minimize contamination, the grinding wheel or belt is to be changed for each sample. The analysis is performed, and the results are sent immediately to the furnace, where appropriate adjustments are made in the composition of the heat. Sampling and analysis needs at the most a few minutes. Samples taken from the semi-finished or finished products are to be at least 12 mm in diameter. Small rod-shaped samples can be analyzed using special equipment. Fig 6 shows schematically sampling, sample preparation and analysis by the optical emission spectroscope.
Fig 6 Sample preparation and analysis by OES