Zoek in de site...

Photoacoustic Spectroscopy (PAS)

The photoacoustic effect

PAS_1A gaseous molecule that absorbs electromagnetic radiation is excited to a higher electronic, vibrational or rotational quantum state. Generally, depopulation of this quantum state to lower lying states occurs either via fluorescence or collisions, the latter giving rise to a temperature increase of the gas due to energy transfer to translation. This non-radiative relaxation process occurs when the relaxation time can compete with the radiative lifetime of the excited energy levels. Radiative decay has a characteristic lifetime of 10-7 seconds at visible wavelengths as compared to 10-2 seconds at 10 µm. For non-radiative decay these values depend on the pressure (decay time t inversely proportional to the pressure) and can vary strongly at atmospheric pressures (10-3-10-8 s).

By modulating the radiation source at an acoustic frequency the temperature changes periodically, giving rise to a periodical pressure change which can be observed as an acoustic signal; in the gas phase the effect can be detected with a sensitive microphone.

Laser-based photoacoustic detectors are able to monitor trace gas concentrations at atmospheric conditions with orders of magnitude better sensitivity as compared to conventional scientific instrumentation; in addition they are able to monitor non-invasively and on-line under dynamic conditions.

PAS_2

History

The photoacoustic effect was first reported by Alexander Graham Bell in 1880; he discovered that thin discs emitted sound when exposed to a rapidly interrupted beam of sunlight. In a later experiment, he removed the eye-piece of a commercial spectroscope and placed absorbing substances at the focal point of the instrument. The substances were put in contact with the ear by means of a hearing tube (Figure 1a and b) and he found 'good' sounds in all parts of the visible and invisible electromagnetic spectrum of the sun.

PAS_3

Bell's 'photophone' : The eye piece of a spectroscope
is removed and substances are placed in the focal
point of the instrument behind a slit. These substances
are put in contact with the ear by means of a hearing tube.

Other publications on this phenomenon followed this first work; we mention here the works of Röntgen, Tyndall and Preece. However, due to the lack of a quantitative description and the lack of a sensitive microphone, the interest in the photoacoustic effect soon declined.

In 1938 Viegerov refined the photoacoustic technique for the first spectroscopic gas analysis; hereafter Luft measured trace gas absorption spectra with an infrared broadband light source down to the part per million level.

By the end of the 1960's, after the invention of the laser, the scientific interest expanded again. In 1968, Kerr and Atwood utilized laser photoacoustic detection to obtain the absorption spectrum of small gaseous molecules. Due to the high spectral brightness of lasers and improved phase-sensitive lock-in techniques to amplify the acoustic signal, they were able to determine low concentrations of air pollutants. Kreuzer demonstrated that it was possible to detect concentrations of e.g. 10-8 (10 part per billion) of methane in nitrogen , using an intensity modulated infrared (3 µm) He-Ne laser. Patel demonstrated the potential of the technique by measuring the NO and H2O concentrations at an altitude of 28 kilometers with a balloon-borne spin-flip Raman laser.

From here on the photoacoustic effect was introduced into the field of trace gas detection with environmental, biological and medical applications.

Light sourcesLamps

Lasers are not essential to operate photoacoustic gas detection systems. Although the spectral power density of broad band infrared light sources is orders of magnitude lower as compared to lasers, their advantages are reliability and cost effectiveness. Infrared light sources in combination with various photoacoustic detection schemes are commercially available for trace gas detection at ppm levels. Spectral selectivity is achieved by using FTIR (Fourier Transformed Infrared) spectroscopy in combination with spectral band filters in front of the photoacoustic cell; thus typically 7 molecular gases can be detected simultaneously at 1-100 part per million (ppm) level (Bruel & Kjaer).

PAS_4

Infrared gas analyzer with photoacoustic detection scheme. Light from the infrared source is split into two paths. The chopper modulates the intensity for both paths. A filter volume in each path serves to filter out light of wavelengths not needed for the detection process; they can be filled with gases of which the absorption spectra do not overlap with those of the species under scrutiny. M1 and M2 serve as measuring cell and reference cell, respectively. With the help of the equalizer both light intensities become equal before entering the last cell. The last cell consists of two compartments with a membrane in between. Both compartments are filled with the gas under investigation so that all wavelength characteristics for this gas contribute to the signal. If the attenuation differs in M1 from that in M2 the membrane starts to oscillate with the frequency of the chopper. This oscillation (typically a few Hz) is detected capacitively.

The gas analyzer of Hartmann & Braun uses a photoacoustic detection scheme which is able to detect a specific gas out of a multi-component gas mixture avoiding cross interferences. In this instrument selectivity is achieved by comparing the direct absorption in a sample cell to that in a reference cell. After passing the sampling cells, each attenuated light beam enters a second detection cell, filled only with the gas of interest the detection cells are interconnected via a membrane connected to a capacitance. Since the dual beam is modulated the difference in acoustic energy reflects the difference in absorption and thus the concentration difference between sample cell and reference cell. Selection of wavelength occurs by the species under investigation itself in such a way that all wavelengths at which absorption occurs are simultaneously active. When there is no spectral overlap from other gases, additional absorptions in the sample cell will not contribute to the acoustic signal; the light passes the detection cell unattenuated. When a specific compound, e.g. H2O, causes spectral overlap, an extra cell can be placed in the light path filled with the interfering gas. This cell attenuates completely the wavelengths where this interfering molecule absorbs including the spectral overlap regions. Thus, these wavelengths cannot contribute to the photoacoustic signal. Thus, a single component can be detected out of complex multi-component gas mixtures.

High spectral brightness renders cw (continuous wave) laser sources ideally suitable for photoacoustic trace gas detection. In contrast to direct absorption techniques, the photoacoustic signal is proportional to the laser power; from the Lambert-Beer law one finds for small absorptions

P = P0 exp(-sigma N L) --> (P0 - P) ~ P0 sigma N L

with P0 and P the laser power before and after the photoacoustic cell, respectively, sigma the absorption cross-section per molecule (in cm2), N the number of absorbing molecules per cm3 and L the absorption path length (in cm). The absorbed energy (P0 - P) is converted into acoustic energy which is recorded by the microphone.

Lasers

10AS_5Laser light sources achieve besides sensitivity, high selectivity. The spectral selectivity is only limited by the pressure broadened absorption profiles of the gases under investigation. The first practical lasers which were used to detect trace gases were CO2 lasers. These line tunable lasers cover the infrared 9-11 µm wavelength region with a laser line spacing of 0.5 - 2 cm-1. Fingerprint absorption spectra can be achieved if we compare the laser line spacing to the pressure broadened absorption lines for the trace gas molecules (typically a few GHz at atmospheric pressures). Additionally, they are able to deliver high laser powers (1-100 W) from a relatively small gas discharge tube.

A new development is the application of the CO laser in the 5.0-7.6 µm wavelength region and the 2.5-3.8 µm region. Although less powerful (typically 1 W) its performance can be improved by applying an intracavity setup. CO lasers are line tunable with line spacing between 0.5 - 1 cm-1. Other cw lasers in the visible and infrared have been applied to photoacoustic trace gas detection such as a Spin Flip Raman laser, diode lasers and dye lasers. Although dye lasers and Titanium Sapphire do not cover the ideal wavelength region for trace gas detection they are very well suited for photoacoustic spectroscopy of weak absorption bands with their continuous tunability and a typical cw laser power of 1 W and they have proved their potential in molecular spectroscopy of highly vibrationally excited molecules. However, for trace gas detection they are less applicable since the overtone molecular absorption cross sections are weak thereby raising the detection limits. For the same reason infrared diode lasers derived from telecommunication research (0.8 and 1.5 µm) are not very well suited.

From the recently developed periodically poled non-linear crystals, Periodically Poled Lithium Niobate (PPLN) is probably most well-known. Lithium Niobate (LiNbO3) can be used in combination with pulsed lasers to generate mid-infrared radiation (2-5 µm) by parametric oscillation. Due to phase matching problems cw operation was limited to a few µW of laser power. Quasi phase matching with periodically poled materials overcomes this problem resulting in laser powers up to a few Watt in the infrared.

To perform trace gas detection the ideal photoacoustic cell should amplify the generated sound originating from the molecular gas absorption meanwhile rejecting acoustic (and electric) noise and in-phase infrared absorption from other materials. Interfering gases should be distinguished by spectroscopic or physical methods.

For gas phase measurements, mainly resonant cells are combined with modulated cw lasers and lock-in amplifiers; pulsed lasers are combined with piezo-electric detectors and boxcars. These selective amplifiers arise from the necessity to lower acoustic and electric noise levels thus improving the signal-to-noise ratio.

Other requirements for photoacoustic cells are low gas consumption or a fast response; for this the active volume of the cell should be small so that no dilution can take place when the trace gas and its carrier flow through the acoustic cell.

If we consider a non-resonant, cylindrical cell, its performance can be expressed as its efficiency to convert absorbed photon energy into acoustic energy; this cell constant F (Pa.cm/W) is proportional to:

Fnon resonant=(G (cp/cv - 1) L) / (w V)

with L and V the length and the volume of the cell, respectively, cp/cv the specific heat constant, w the modulation frequency and G a geometrical factor in the order of one. Within a non resonant cell the gas absorption signal is independent of the cell length, but inversely proportional to its diameter. However, photoacoustic signals are also generated by infrared window absorption. To reduce these signals inside the cell its length should be as long as possible to spread this locally generated sound over the total cell volume.

For a resonant cell one has to amplify the above cell constant F with the quality factor Q of the generated acoustic resonance

Fresonant = Q Fnon resonant

Q is expressed by the ratio of the energy stored in the acoustical standing wave divided by the energy losses per cycle. This acoustical amplification process is limited by various dissipation processes which can be divided into surface and volume effects. Main surface losses are viscous and thermal losses at the resonator surface, microphone losses and acoustic wave scattering losses at obstacles in the cell. Volumetric losses are not as important as surface losses and are mainly due to free space viscous and thermal losses and V-V, V-R, V-T relaxation losses of poly-atomic gases.

PAS_6

Resonant acoustic modes of a cylindrical closed chamber; the fundamental longitudinal, azimuthal and radial modes.

A cylindrical cavity can be a resonant cavity for sound waves. The resonance frequencies of such a cavity are given by

fmnp = vs( ( amn / 2R)2 + (p/2L)2 )½

with vs the sound velocity in the gas filling the cavity, R the radius of the resonator, L the length, p= 0,1,2,3... axial mode numbers and am a suitable solution of Bessel equations with m= radial mode number and n= azimuthal mode number.

The cell constant F for all types of cylindrical resonant cells is proportional to L½ /R. Due to their larger diameters resonances in the radial or azimuthal acoustic mode have high Q-values and high resonance frequencies but low F-values. A longitudinally excited resonator will have a low Q value but, due to its small diameter, a high F-value.

PAS_7

Resonant photoacoustic cell with resonator and buffer volumes. BW: ZnSe Brewster window; TAC: adjustable quarter wave notch filter (tunable air column) to suppress window signal; Gi: Gas inlet flow; NF: half wave notch filter to suppress flow noise; Go: gas outlet flow; M: Knowless microphone.

Photoacoustic cells

PAS_8In order to obtain an optimum signal-to-noise ratio, one has to take into account noise control and interfering signals. Theoretically, the ultimately lowest acoustical noise results from random density fluctuations, i.e. Brownian motion, in the gas which is distributed over all frequencies in the sound spectrum. The total power of these density fluctuations is constant, but the frequency distribution is dependent upon local resonances and their Q-values. Therefore, using a resonant cell the signal-to-noise ratio will not be improved considering only these random thermal noise fluctuations.

However, these noise levels are far below other noise sources such as amplifier noise and acoustic disturbances from outside the cell. The power of the amplifier noise varies as 1/f, where f is the modulation frequency of the light beam. Therefore, in contrast to Brownian noise, it is more advantageous to operate at a cell resonance to increase the generated acoustical signal above the 1/f noise. A prerequisite is that external acoustic disturbances are shielded from the microphone by a proper cell wall construction, material choice and proper design of inlet and outlet ports.

Noise fluctuations do not have a fixed phase relation with the modulation of the light intensity. Other disturbing factors limiting the sensitivity do have a fixed phase relation. Directly generated coherent acoustical background signals, caused by the modulation process (e.g. chopper, current modulation) can be suppressed in the same way as external acoustical disturbances.

Other, more serious interferences are coherent photoacoustic background signals which are caused by absorption of the light beam in the window material or light scattered or reflected off the resonator wall. It is generated at the same frequency and in-phase with the modulated light beam. In resonant cells, window absorption signals can be suppressed by using large buffer volumes and quarter wave tubes next to the windows. These one-end-open tubes, placed near the window, act as a notch filter for the window signal at the resonance frequency. The influence of the scattered light on the photoacoustic background signal can be minimized by using, for the resonator wall, a highly reflective polished material, with a thermally well conducting material as substrate, e.g. in the case of the CO2 laser a polished gold coated copper tube.

In the past special designs have been developed for longitudinally, azimuthally and radially resonant photoacoustic cells, even without windows to improve sensitivity.

In order to improve the selectivity the combination of the Stark- or Zeeman effect with photoacoustic detection represents an interesting solution for specific molecules such as ammonia (NH3) and nitric oxide (NO). The change in absorption at a specific laser frequency depends on the shift and splitting of the molecular absorption lines of the species under investigation. Although the method does not suffer from interference problems within multicomponent gas mixtures there are some limitations.