This paper presents a narrow linewidth, high resolution, and high quantum efficiency imaging transmission filter based on optical trapping of resonance radiation in potassium vapor. The filter can be used to image radiation over a bandwidth narrow enough to fall within a Fraunhofer dark zone in the solar spectrum, and it can be applied to the imaging of flames, plumes or discharges containing potassium. It may also be applicable to the imaging of Raman scattering from a tunable laser. The spectral and imaging properties of the filter are demonstrated with a 1 cm aperture optically thick potassium cell illuminated by a narrow linewidth tunable laser. The spectral width at the potassium D2 line wavelength, 766.5 nm, is shown to be 1 to 2 GHz (.002 nm). At the line center, the quantum efficiency is better than 60% and the imaging resolution is better than 30 line pairs per mm. By employing a 200 micron “thin” potassium vapor cell, it is also shown that the filter maintains the high quantum efficiency (~50%) and good imaging capability (~20 lines per mm) across the 2 GHz spectral bandwidth of the cell. The “thin” cell has an out-of-band rejection of better than 1000. Its operation is demonstrated with a tunable laser as well as with broad band light from a potassium lamp and from a potassium chloride seeded flame.
© 2006 Optical Society of America
Very narrow linewidth imaging filters cannot be constructed using conventional dielectric coated optics because the necessary variation in light propagation vector angles broadens the transmission linewidth. In order to overcome this limitation, atomic vapor filters and atomic vapor Faraday filters for narrow band light detection have been studied extensively.[1,2]. Due to their very narrow linewidth, high out-of-band suppression, stable center wavelength and very high damage threshold, atomic vapor cells are attractive for applications in spectroscopy. The use of atomic filters for narrow linewidth imaging was proposed in 1961 by Langberg et al.  through resonant fluorescence and, more recently, by Temirov et al.  through resonant ionization. Langberg proposed sodium vapor as the resonant fluorescence material; the filter operated at the sodium D lines with a predicted resolution of 10 lines per millimeter and 70% efficiency. The resonance ionization imaging detector demonstrated by Temirov required a high-power laser source for photoionization and used a microchannel plate array to achieve a spatial resolution of 200 microns. The work done in our laboratory at Princeton focused on the use of mercury, rubidium,  and potassium  for atomic vapor resonant filters. Narrow linewidth imaging was demonstrated with mercury vapor at 253.7 nm and a mercury filter was employed to resolve the rotational Raman scattering of nitrogen and oxygen in air. 
The concept of the refluorescent atomic filter is shown in Fig. 1. Only photons near the resonant frequency of the atomic transition line are absorbed; all other photons pass though the cell without interference. As the density of atoms increases, the absorption through the cell increases until the cell becomes optically thick, where the e-1 optical depth is a small fraction of the length of the cell. In this case the extinction at the center of the resonance is many orders of magnitude. If the excited atoms can only can radiate back to the ground state, then photons absorbed by the atoms are trapped within the optical depth near the entrance window of the cell. This trapping occurs at the same wavelength independent of the angle at which the photons arrive. In the absence of quenching, these photons undergo a random walk as they are absorbed and emitted by the atoms. Since the mean free path is only a few microns, the trapped photons finally exit the cell in the backward direction, through the entrance window. In this manner, the cell appears as a Lambertian scatterer for light within the absorption bandwidth. As with any Lambertian scatterer, the illumination can come from any direction, so the atomic filter has an acceptance angle of up to 2π steradians. For the work reported here, potassium atomic vapor is chosen.
2. Potassium cell absorption
If a beam of monochromatic light propagates through an atomic gas with length L, the unscattered transmitted irradiance (power per unit area) is given by Beer’s Law:
where Iout and Iin are the transmitted and input light irradiance respectively, and α is the absorption coefficient at frequency ν. The absorption coefficient is the summation of all hyperfine structure transition contributions at the frequency, ν:
where Ni is the number density of a specific isotope, fi is the oscillator strength of a particular hyperfine transition, and V(ν,νi) is the normalized Voigt line shape function for that specific hyperfine transition with center frequency, νi.
Potassium has two well known resonance lines to the ground state, the D2 line: 4P3/2→4S1/2 at 766.5 nm and the D1 line: 4P1/2→4S1/2 at 769.9 nm. The transition probabilities of these two lines are approximately the same: AD2=AD1=3.87×107 s-1, but the oscillator strengths are not. The oscillator strength f can be calculated from the following equation:
Since the degeneracy of the ground state, gS1/2, =2, and the degeneracies of the two upper states, gP3/2 and gP1/2, are 4 and 2, respectively, f is 2/3 for the D2 line and 1/3 for the D1 line. Consequently, the absorption coefficient of the D2 transition is twice as large as that of the D1 transition. Since the absorption coefficient is proportional to the product of the atom number density and oscillator strength, the two D lines behave similarly in the process of absorption and fluorescence. In this paper the modeling and experiments address the D2 transition manifold of the potassium atom. Those results can be extended to D1 line manifold in a straightforward manner.
There are three naturally occurring isotopes of potassium: 39K, 41K and 40K. Their natural abundances are 39K 93.26%, 41K 6.73%, 40K 0.012%. If we use the transitions of 39K as the reference, the isotopic shifts for other isotopes are: 41K: 236.15 MHz, and 40K: 126.43 MHz . Different isotopes also have different nuclear spins: 39K: I=3/2, 41K: I=3/2, 40K: I=4. Because the natural abundance of 40K is comparably small, it is neglected in the modeling.
The hyperfine energy shift due to the nuclear spin orbital angular momentum coupling is given by:
where C=F(F+1)-I(I+1)-J(J+1), F is the total angular momentum, h is Planck’s constant, I is the nuclear spin, and A and B are the hyperfine interaction constants for each state. The two terms in the equation originate from the magnetic dipole and the electrical quadrupole interaction respectively. The value of the coupling constants A and B are given in reference . The relative wavelength and strength of all the lines in the D2 transition manifold included in the model are plotted in Fig. 2. It can be seen from Fig. 2 that all the hyperfine lines of the two dominant isotopes fall within 0.6 GHz of each other, which is quite different from rubidium (~10 GHz) and mercury (~ 30 GHz). This small isotopic shift and hyperfine splitting makes potassium a particularly attractive atom for the narrow linewidth filter.
The absorption profile is found from the summation of the broadened hyperfine lines. The three broadening mechanism considered in the model are thermal broadening, natural broadening and collisional self broadening. The broadening due to atomic thermal motion contributes an inhomogeneous Gaussian component:
is the Gaussian-component width. At the experimental conditions, the body of the K cell is kept at T=200 °C, which gives us the Δνg~700 MHz. Because the hyperfine splitting is on the order of 100 ~ 200 MHz, with 700 MHz thermal broadening, hyperfine lines will not be seen separately in the measured absorption profiles.
The natural broadening and collisional self broadening both combine to form a homogeneous Lorentzian profile.
where ΔνL=ΔνN+ΔνS. Natural broadening ΔνN=2π/Aij=6.16 MHz. Collisional self broadening is Δνs=KsbN, where N is the number density of K atoms. The self-broadening coefficient Ksb can be found in reference .
The combination of the Gaussian component and Lorentzian component is a Voigt function:
For simplification of the calculation of the Voigt profile, the Whiting approximation is used . The essence of the Whiting approximation is an empirical approximation by a single equation, and the error is predicted to be -5%~+2% within 10 Δνg. The result of the modeling is plotted in Fig. 3 together with the experimental results.
In this work, the wavelength, given by a WA-10 Burleigh Wavemeter, is the vacuum wavelength of the laser beam. The conversion from vacuum wavelength to air wavelength is given by Morton :
where λ 0 is the vacuum wavelength in Angstroms, and λair is the air wavelength in Angstroms. For the K D2 line, the air wavelength is 766.4900 nm, and the vacuum wavelength is 766.7015 nm. 
In absorption experiments, a Sacher TEC-500-765-10 External Cavity Diode Laser was used as the light source. The Sacher laser has an output power of approximately 10 mW and a linewidth of less than 1 MHz. It has an electronically controllable frequency tuning range of approximately 250 GHz that is achieved by changing the voltage on a piezo-electric actuator. Continuous frequency tuning occurs over about 10 GHz. Beyond this range there is a discontinuity in the frequency due to laser mode hopping. The laser also has a coarse tuning range of approximately 10 nm that is achieved by turning a screw on the laser head.
The potassium vapor cell used in the initial experiments reported here consists of a cylindrical cell body and a cold tip. The length of the cell body is 5 cm. Both parts are heated, the body temperature is kept at 200 °C, and the tip temperature of the cell is used to control the potassium vapor pressure.
The absorption profiles are measured at different tip temperatures. The potassium vapor pressure can be calculated for the various tip temperatures from the expression if Equation 10 .
The results of the temperature calculation are shown in Table 1 for reference. In the following part of this paper, only the cold tip temperature will be shown in the experimental results.
The measurement results are plotted in Fig. 3. The modeling matches the experimental result almost perfectly. Note that the Full width at Half maximum absorption bandwidth of the absorption profile increases with the increase of potassium vapor pressure, from 0.5 GHz (0.001 nm) at 30 °C to 3 GHz (0.003 nm) at 100 °C.
3. The quantum efficiency of the refluorescent filter
The efficiency of the refluorescent filter is one of the most important parameter for its application. The overall efficiency of the system will depend on the reflection efficiency of the potassium vapor filter, the collection efficiency of the entrance lens system, and the collection efficiency of the camera imaging lens. Only the reflection efficiency of the filter is studied here. To estimate the fundamental quantum efficiency of the potassium vapor cell, the fluorescence signal was compared to scattering from a Lambertian surface. The collimated Sacher laser beam was sent to the front window of the K vapor cell after passing through an optical chopper and a neutral density filter. The size of the laser beam spot on the cell was roughly 3~5 mm2. The fluorescent light was collected by a lens, and imaged onto a PMT detector as a spot with similar size. An aperture was used in front of the PMT to ensure that only the fluorescent photons at the potassium vapor cell front surface were detected by the PMT. The PMT signal was passed through a lock-in amplifier and sent to the computer to be analyzed. The wavelength of the Sacher laser was tuned across the resonant line and the fluorescence signal as a function of frequency was recorded for cell cold tip temperatures ranging from 65 °C to 110 °C.
Subsequently the potassium cell was replaced with a Lambertian surface made from fresh white MgO smoke, with an estimated scattering efficiency of 98% . The Lambertian surface was located at the same position as the potassium cell’s front window and the signal as a function of laser frequency was recorded. The Lambertian surface is used as the reference for our efficiency estimation. Finally, to assure that background was not significant, the laser was blocked and another scan taken. Fig. 4 shows the results of these experiments
From the figure it is evident that the fluorescence cell has its highest efficiency at 90 °C cold tip temperature where it reflects nearly 47% compared to the MgO Lambertian surface, leading to a reflectivity of the potassium cell of approximately 48%. The measured reflectivity does not take into account the reflection losses at the two windows. Light passed in and out, so it must pass through eight surfaces compared to the Lambertian scattered light. At 766.5 nm, each surface has reflection losses of 3.7%, leaving 96.3% transmission. That to the eighth power is 74%, so with antireflection coatings, it may be possible to increase the reflectivity by a factor of 1.35 to 65%.
4. Imaging capability of the refluorescent filter
The imaging capability of an imaging system can be characterized by the modulation transfer function (MTF). The modulation of image is defined as:
where T max is the maximum intensity of the image, T min is the minimum intensity. Since most images have multi spatial frequency components, the modulation for each spatial frequency component can be measured separately. The MTF is defined for an imaging system as the capability of transferring each different spatial frequency component from the object to the image:
where Mi(f) is the modulation of the image at spatial frequency f, and Mo(f) is the modulation of the object at f.
In order to measure the MTF of the potassium vapor cell as a passband filter, an experiment was set up as shown in Fig. 5. The beam from a narrow bandwidth laser was expanded and sent though a diffuser as the light source. Various Ronchi Rulings with numbers of different lines per inch (lpi) were used as the object. A Nikon camera lens set was used to image the Ronchi Ruling onto the face of the refluorescent cell through a Pellicle beamsplitter. The camera lens located the image of the Ronchi Ruling at the position of potassium vapor and glass interface. The refluorescent pattern was recorded using a camera with a microscopic lens that observed the face of the cell by reflection from the beam splitter. Interference due to reflections from the cell windows was avoided by tilting the cell at a small angle (<5°). The loss of resolution introduced by the Nikon lens set and the microscopic lens set is negligible at the center of the field of view compared to the resolution limitation of the refluorescence cell.
The laser was tuned across the potassium D2 line, and the image on the refluorescent cell only appeared when the laser frequency was at the resonant line. Different Ronchi Rulings were used and set at the same position to collect a set of refluorescent images. Refluorescent images from a 100 lpi and 200 lpi Ronchi Rulling are shown in Fig. 6.
The refluorescent images are digitized and transferred into the frequency space by a Fast Fourier Transformation (FFT). The result of the 110 µm line spacing image is plotted in Fig. 7 (a). As a reference, a black and white bar pattern with the same spatial frequency as the refluorescent image is generated and transferred to the frequency space as shown in Fig. 7 (b).
The FFT of a square-wave has discrete peaks in frequency space. The peak at zero frequency is the average intensity of the signal. The first peak at 9 cycles/mm is the intensity of the fundamental frequency, then the 3rd harmonic, 5th harmonic and so on. The MTF of the filter at each of these frequencies is the amplitude measured in Fig. 7(a) divided by the ideal amplitude of the square wave given in 7 (b), normalized by the DC component.
Due to the limitation of the spatial frequency component of the Ronchi ruling, only two or three points of the MTF can be generated from each Ronchi ruling. In the experiment, Ronchi rulings with 50 lpi, 100 lpi, 150 lpi and 200 lpi were used. All data are combined into one graph and shown in Fig. 8.
From the plot, it can be seen that 50% resolution point for the potassium refluorescent cell is roughly 33 cycles per mm, corresponding to ~30 µm. The potassium vapor pressure can be deduced from the cell’s tip temperature. Using the perfect gas law, the number density of potassium atoms in the potassium vapor cell can be estimated. Using the absorption model in section 2, the optical depth of potassium vapor at Ttip=100 °C is calculated to be around 35 µm. From Beer’s absorption law, it means that 70% of the on-resonant photons are trapped in the first 35 µm after they entered the K vapor. Since the reemission point is random within approximately this area, any image components with finer resolution than this length scale will be blurred. This is consistent with our MTF measurements.
Since the optical density is inversely proportional to the potassium number density, from the reason above, higher cell tip temperature will produce higher potassium number density, and higher values of the MTF can be achieved. But at higher potassium number density, the quantum efficiency of the refluorescent filter appears to be reduced, presumably by the increase of the collisional quenching with the window. Some reduction in resolution is expected if the D1 transition manifold rather that the D2 manifold is used because of the factor of two lower oscillator strength.
5. Ultra-thin potassium imaging cell
When the laser frequency is tuned off the central portion of the absorption curve, the resolution of the thick cell drops, along with the quantum efficiency. This change is caused by the decrease of the absorption coefficient of the atomic vapor, and the corresponding increase of the penetration of the photons in to the atomic vapor. The linewidth of potassium radiation from a flame, a plume, or a discharge lamp is normally much broader than the few GHz of the imaging filter due to optical thickness, thermal broadening, and broadening by collisions with the ambient gas molecules. Even over the few GHz band of the filter, there is significant variation in the absorption constant, so blurring and out of focus background noise will occur when these sources are imaged with a thick cell, such as the one used for the work presented above.
In order to improve the resolution for off-resonant photons, a thin cell was constructed with 1 cm diameter entrance and exit windows. With this configuration, off-resonance photons pass through the cell and out the other side. Those that are absorbed and subsequently fluoresce stay within the focal plane of the imaging system. The challenge of constructing a thin cell is to control the gap spacing between two optical windows while keeping an outlet for vacuum pumping and for the potassium reservoir.
In order to construct this thin glass cell, two different diameter glass tubes were used. Each tube was fitted with a window on one end. The small diameter tube was inserted into the larger diameter tube, and a 200 µm thick copper woven wire mesh was used as a spacer to control the gap spacing. After adding a reservoir tip and sealing the two glass tubes at the end opposite the windows, the copper mesh was dissolved with HCl and the reservoir filled with potassium under vacuum. Fig. 9 shows the schematic design of the 200 µm K cell. This method of construction allows for different gap distances to be specified by selecting the size of the spacer material. The same procedure as that used for the thick cell can be used to analyze the imaging capability of the 200 µm K cell at different wavelengths near the potassium resonance.
Fig. 10 shows the fluorescent images on the thick K cell and thin K cell as the laser wavelength is tuned to a slightly off resonant wavelength at 766.699 nm. In these images, it can be seen that, though the photons are slightly off resonance, some of them are still absorbed, producing a fluorescent image. The difference between the two cells is that in the thick cell, the off-resonance photons are trapped deep in the vapor and the fluorescence back out the entrance window leads to a significant background, while in the thin cell case, these off-resonance photons pass through the vapor cell and the background is eliminated.
6. Out-band suppression
One major advantage of atomic resonant filter is the out-band suppression of near resonant light. The performance of the potassium imaging filter for near resonant out-band suppression can be tested by observing a broad band potassium light source that has the central few GHz photons removed. In the experiment, this is achieved by placing a thick (3 cm) atomic potassium absorption cell
(blocking cell) between a potassium hollow cathode lamp source and the collection optics for the atomic filter.
The potassium hollow cathode lamp has 10 torr Argon as the buffer gas. Although commonly used as a narrow emission light source for atomic absorption spectroscopy, due to the collisional broadening and high temperature in the cathode region, the linewidth of the hollow cathode lamp radiation is several times broader than the potassium vapor cell absorption linewidth. The spectrum of the hollow cathode also has many other radiation lines.
A 10 nm passband filter is used to eliminate the far out of band light, but transmits the emission in the vicinity of 766–770 nm, thus light both in-band and out-of-band potassium D2 and D1 light falls on the thin potassium imaging cell. The cell responds to the in-band light at both the D1 and D2 lines, and rejects the out of band light. The rejection is observed by removing the in-band light from the lamp, leaving only the out-of-band light and measuring how much of that light gets through. The in-band light is removed by the 3 cm potassium blocking cell that is placed in front of the lamp. As the tip temperature of the blocking cell is increased, the absorption strength and linewidth increases. In the optically thick regime corresponding to tip temperatures above about 70°C, the strong absorption removes the central portion of the lamp output, but leaves the remaining out-of-band light. As the blocking cell tip temperature is further increased, the portion removed increases in width until all of the central few GHz in the vicinity of the D1 and D2 potassium manifolds are removed. At that point, a perfect narrow line width filter will see nothing, even though there is significant light entering the collection optics. The total light entering the system is calibrated with a Lambertian scatterer replacing the thin potassium imaging cell. In the experiments, the thin potassium imaging cell is kept at constant conditions with a tip temperature of 130°C and a body temperature of 180 °C. The same set up is used to take the hollow cathode lamp images with the Lambertian scatterer. The temperature of the blocking cell was changed, and images from the thin cell and from the Lambertian scatterer were taken and analyzed.
Figure 11 shows the plots of the binned pixel intensities across the center of the spot in the fluorescent images and the image from a Lambertian surface as the tip temperature of the blocking cell is varied from room temperature 24 °C to 120 °C. Over this range the absorption of the central portion of the lamp line becomes greater until there is no more light that falls within the pass band of the thin imaging filter, and the brightness of the peak intensity on the thin cell image falls by a factor of 1500, taking into account the different integration times. By comparison, there is only a factor of 13 change in the light irradiance for the Lambertian scatterer.
At the point where the blocking cell tip temperature reaches 80°C, the ratio of the peak Lambertian image to the peak thin cell image reaches the order of 1200, and that ratio only increases slightly for the higher tip temperatures, indicating that by 80°C the thick potassium cell in front of the lamp has removed essentially all of the in-band light. The higher two blocking cell temperatures, 100 °C and 120 °C primarily lead to reduced out of band light entering the collection system. This indicates that the thin cell can suppress approximately a factor of 1000–2000 of near resonant out of band light. A close look at the image pattern in the highly suppressed cases indicates that the residual scattering is primarily from some dirt that is deposited on the window of the cell. That deposit may be associated with the thin cell fabrication process, and further advances in that technology will certainly improve rejection capability of the filter. For application to Raman imaging, a tunable laser would be used and tuned such that the Raman scattering coincided with the filter transmission wavelength. In that case, out of band rejection is particularly important to select against Rayleigh and Mie scattering and other Raman lines, and rejection factors of better than 1000 may be required.
6. The demonstration of the potassium imaging filter with potassium lamp and flame
To provide a simulation of the atomic potassium radiation from plumes or flames, a methane flame seeded with KCl was set up. Although the potassium D lines have much higher peaks, the color of the flame is still dominated by the continuum spectrum background, and looks yellow-white.
In order to determine the line width of the potassium line emitted by the flame, a laser beam was collimated and sent through the center of the K seeded flame, and then the collected by a photomultiplier in the far field. A chopper and lock-in amplifier were used to suppress the room light background and the flame radiation. The laser wavelength was monitored by a WA-10 Burleigh Wavemeter with a resolution of 0.001 nm (0.5 GHz). The D2 absorption spectrum is shown in Fig. 12.
From Fig. 12, it can be seen the Full Width Half Maximum (FWHM) of the optically thick D2 line absorption is around 0.05 nm, corresponding to 25 GHz. Since the flame is optically thick, this provides a good estimate of the expected emission line width, and it indicates that the flame is a good test source for quantifying the performance of the filter. Note that the 1.5–2GHz filter will only capture 5 to 10 % of the emitted potassium light from the flame assuming no line shape inversion from optical thickness effects, but since that narrow band of light falls within the Fraunhofer dark band, the observability of the potassium emission in daylight is increased by orders of magnitude. This may be important for imaging missile plumes or other potassium containing combustion sources that may otherwise not be visible due to solar background.
Fig. 13 shows several images taken with the flame illumination. Fig. 13 (a) is taken with the thin K cell tip temperature kept at 135°C. Note that the filter captures both the K D1 line and D2 line radiation from the flame. The tip temperature of the thin cell is higher than it is in the laser measurement to ensure the vapor is optically thick for both D1 line and D2 line. No other interference filters are required in the experiment to separate out D2 line from D1 line radiation. Fig. 13(b) is a reference image taken with the tip temperature kept at room temperature. Fig. 13(c) is taken with the K cell replaced by a Lambertian surface. A knife edge is put in front of the flame to provide contrast. The average intensity of the flame image in Fig. 13(a) is about 4 % of the average intensity in Fig. 13(c), which is comparable to the expected ~50% response of the potassium cell to the portion of K flame radiation spectrum that falls within its 2 GHz passband. When the vapor cell is at room temperature, the potassium vapor pressure is so low that no significant absorption occurs, so no fluorescence is present in the reference image (b). The bright spot in Fig. 13(b) is caused by the scattering from dirt on the front window of the K vapor cell and is also seen in Fig. 13(a). Fig. 13(c) is at much reduced sensitivity and shows the image of the flame seen on a Lambertian screen. Separate experiments also confirm that when the potassium seed material is removed from the flame, the filter captures images that are similar to what is seen with the cell at room temperature, which essentially is only scattering from the dirt.
In order to quantitatively measure the optimum resolution of the refluorescent image of the flame, a Ronchi ruling is put in front of it. The image of the illuminated Ronchi ruling rather than the shape of the flame is then captured by the filter, so the radiative depth of the flame doesn’t affect the quality of the image. The digitalized vertically binned row from an image of the flame back-illuminated Ronchi ruling on the thin potassium cell is shown in Fig. 14. By comparison with the Fourier transform of the square wave, it can be concluded that the thin K vapor cell gives at least 35% MTF at the 20 lines/mm spatial frequency. This reduced resolution compared to the laser illuminated images may be in part due to the presence of the D1 line manifold in the flame spectrum. There is also some residual background scattering. Part of that background arises from secondary light scattering from the light traps and from optical elements. In this experiment, the light traps are black felt and the optics are not antireflection coated. Since these elements are out of the focal plane, they contribute background diffuse light, much of which can be removed if higher performance is required.
The capability of the imaging fluorescent filter is demonstrated with a Princeton shield, back-illuminated by a hollow cathode potassium lamp. The fluorescent image of the Shield is shown in Fig. 15.
For this image, a Princeton shield mask was placed in front of a potassium hollow cathode lamp, and the thin potassium imaging filter was used with a tip temperature of. 135 °C. The size of the Princeton shield on the surface of the imaging cell is about 1 mm wide, as indicated on the figure.
It has been shown in this paper that an optically thick potassium cell can be used as a passband filter with good imaging capability. The imaging cell has a spectral resolution of between 1 and 2 GHz for the D2 line, and the highest spatial resolution measured is 33 lines/mm (30 µm). By comparing the cell to a Lambertian surface with known reflectivity, the quantum efficiency of is estimated to be >60%. Unlike the conventional interference filters, the refluorescent filter has a theoretical acceptance solid angle of 2π. It is also shown that a 200 µm path length thin cell has significantly better out-of-band spatial resolution for broadband sources than does a thick cell. This thin cell filter maintains the high quantum efficiency (~60 %) and high quality spatial imaging capability (20 lines per mm), and has a transmission linewidth ranging from 1.5 to 2.0 GHz (0.003~0.004 nm), depending on the operating temperature. This performance was demonstrated using a narrow line width laser, a KCl seeded flame and a potassium lamp.
This work was supported by Princeton Scientific Instruments Co. through an SBIR program under the Office of Naval Research.
1. D. Pappas, T. L. Correll, N. C. Pixley, B. W. Smith, and J. D. Winefordner, “Detection of Mie Scattering Using a Resonance Fluorescence Monochromator,” Appl. Spectrosc. 56, 1237–1240 (2002). [CrossRef]
2. J.A. Gelbwachs, “Passive Fraunhofer-wavelength atomic filter at 460.7 nm,” IEEE J. Quantum Electron. 28, 2577–2581, (1992). [CrossRef]
3. Langberg, Naylor, and Hechtsher, “An Image-Forming, Resonance Scatter Filter,” Conference on Optical Instruments and Techniques, 229–237, (1961).
4. J.P. Temirov, N.V. Chigarev, and O.I. Matveev, et al. “A resonance ionization imaging detector based on cesium atomic vapor,” Spectrochemica Acta, Part B , 59(5), 677–687, (2004). [CrossRef]
5. R.B. Miles, A. Yalin, Z. Tang, S.H. Zaidi, and J. Forkey, “Flow field imaging through sharp-edged atomic and molecular ‘notch’ filters,” J. Meas. Sci. Technol. , 12, 442–451, (2001). [CrossRef]
6. T. Tang, “Infra-red rubidium atomic resonant filters for low wavenumber scattering,” Phd Thesis, Department of Mechanical and Aerospace Engineering, Pricneton University, (2001).
7. L. Qian, S.H. Zaidi, and R.B. Miles, “Narrow Linewidth Potassium Imaging Filter for Near Infrared Detection of Missile Plumes,” 43rd AIAA Aerospace Sciences Meeting and Exhibit, 2005–0825, (2005)
9. N. Bendali and H.T. Duong et al., “Optical resonance detection by field-ionization of Rydberg state in collinear laser spectroscopy,” J. of Phys. B: Atomic, Molecular and Optics Physics 11, 4231–4240, (1981). [CrossRef]
10. D.S. Hughes and P.E. Lloyd, “Pressure effects of homogeneous K vapor in absorption,” Phys. Rev. , 52, 1215–1220, (1937). [CrossRef]
11. E.E. Whiting, “An Empirical Approximation of the Voigt Profile,” J. Quant. Spectrosc. Radiat. Transfer , 8, 1379–1384, (1968). [CrossRef]
12. D.C. Morton, “Atomic Data for Resonance Absorption Lines III,” Astrophysical Journal Supplement Series , 149, 205, (2003). [CrossRef]
13. CRC Handbook of Chemistry and Physics, 4–129, (1999).
14. F. Grum and G. W. Luckey, “Optical sphere paint and a working standard of reflectance,” App. Opt. 7, 11, 2289, (1968). [CrossRef]