Imaging Open-Path Fourier Transform Infrared Spectrometer For 3D Cloud Profiling Julia Rentz Dupuis, David J. Mansur, Robert Vaillancourt, David Carlson, Thomas Evans, and Elizabeth Schundler OPTRA, Inc. 461 Boston St., Topsfield, MA 01983 phone: (978) 887-6600 fax: (978) 887-0022
[email protected] www.optra.com Lori Todd, Kathleen Mottus University of North Carolina 121 Rosenau Hall, Chapel Hill, NC 27599 ABSTRACT OPTRA is developing an imaging open-path Fourier transform infrared (I-OP-FTIR) spectrometer for 3D profiling of chemical and biological agent simulant plumes released into test ranges and chambers. An array of I-OP-FTIR instruments positioned around the perimeter of the test site, in concert with advanced spectroscopic algorithms, enables real time tomographic reconstruction of the plume. The approach is intended as a referee measurement for test ranges and chambers. This Small Business Technology Transfer (STTR) effort combines the instrumentation and spectroscopic capabilities of OPTRA, Inc. with the computed tomographic expertise of the University of North Carolina, Chapel Hill. In this paper, we detail the design and build of a prototype I-OP-FTIR instrument. Key Words: FTIR spectrometer, cloud profiling, tomography
1
INTRODUCTION
Under a U.S. Army Small Business Technology Transfer (STTR) solicitation, a need was identified for a near real-time 3D plume concentration profiler system for chemical and biological agent simulants released into test ranges and chambers. The technology developed under this opportunity will be considered as a candidate referee sensor for the test range or chamber. The requirements of the application include the ability to generate time-dependent concentration information with sensitivity better than that of the sensors under test at the range. The referee sensor must be responsive to a number of different chemical and biological agent simulants, and it must be able to operate in the zero-temperature contrast scenario between the plume and the background against which it is being viewed. In response to this opportunity, OPTRA is developing an imaging open-path Fourier transform infrared (I-OP-FTIR) spectrometer which is used with an array of plastic injection molded retroreflector arrays to simultaneously monitor an array of open paths (OPs) for chemical content and concentration by infrared spectroscopy. Via the optical design presented in this paper, from a single FTIR modulator the I-OP-FTIR emits an array of discrete OP beams to respective retroreflector arrays which return the beams to the instrument for multiplexed detection and processing. Our present design emits an array of eleven OP beams spread over a 30° fan (i.e. spaced at 3° increments). All beams operate over 7 to 14 μm (1428 to 714 cm-1) with a spectral resolution of 4 cm-1. Unlike previous work employing a single channel OPFTIR with a scanner1, the OPTRA approach simultaneously monitors an angularly resolved region thereby eliminating errors due to the movement of the plume over the duration of the single channel OP-FTIR scan. Unlike previous work using passive hyperspectral imaging2, the OPTRA approach is quantitative without the need for background, plume, and air column radiance estimation; the OPTRA approach is also valid in the zero temperature contrast scenario where passive IR spectrometers, radiometers, and imagers cannot detect the plume. With a number of I-OP-FTIR systems strategically positioned around the test range or chamber (Figure 1) in concert with spectroscopic multicomponent algorithms, an accurate tomographic reconstruction of the plume concentration can Copyright 2009 Society of Photo-Optical Instrumentation Engineers. This paper will be published in The Proceedings of Chemical, Biological, Radiological, Nuclear, and Explosives (CBRNE) Sensing X and is made available as an electronic preprint with permission of SPIE. One print or electronic copy may be made for personal use only. Systematic or multiple reproduction, distribution to multiple locations via electronic or other means, duplication of any -1material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.
be made. Under this STTR effort, OPTRA has teamed with University of North Carolina, Chapel Hill, which brings tomographic expertise for this IR spectroscopic monitoring application. Table 1 details the specifications for the I-OP-FTIR system. The initial application of this work will be a point sensor test range with the associated requirements listed in Table 1. Future work may scale the range of our instrument to on the order of 1 km test grids for standoff sensor test ranges. Figure 1: System Concept
Table 1: System Performance Specifications
OPTRA’s I-OP-FTIR (x3) OP IR beams
plume Plastic Retroreflector Arrays
Test range
REQUIREMENT Noise equivalent concentration Maximum concentration Concentration accuracy Update rate Test grid size Grid spatial resolution
VALUE < 0.1 mg/m3 20 mg/m3 ± 10% < 5 sec 100x100x10 m 2.5 m
Figure 1 (left) shows the system concept. Each I-OPFTIR emits an array of discrete IR beams which are projected through the plume to remotely located retroreflector arrays and back again through the plume to the instrument for multiplexed detection and processing. This approach allows for simultaneous interrogation of the plume unlike scanner-based systems. This approach also supports detection in the zero temperature contrast scenario unlike passive IR spectrometers.
The following contains the details of our prototype I-OP-FTIR design and build.
2
DESCRIPTION OF TECHNICAL APPROACH
2.1 Multichannel Concept Our approach is similar to a traditional monostatic OP-FTIR in that we employ a single IR source from which IR radiation is coupled by a set of lenses and directed through a Michelson interferometer before being expanded and collimated by a projector optic and directed to a remotely located retroreflector array. The retroreflector array then returns the OP beam to the sensor for detection. However, in this case, we couple the radiation exiting the interferometer into an IR fiber optic assembly composed of eleven IR fibers bundled at the input end and free at the output end. The free/output ends of these fibers are then positioned along the focal plane of the projector optic. The result is eleven OP beams whose divergence is determined by a throughput match between the fiber diameter and numerical aperture (NA) with the projector aperture. The pointing of the OP beams is set by the off-axis location of each fiber relative to the optical axis of the projector lens. Each beam returned by the respective retroreflector array is then focused by the projector lens and folded onto a respective detector using an intensity beamsplitter. 2.2 Radiometric Projections The radiometric model is based on a per-channel noise equivalent spectral radiance (NESR) calculation along with projected signal to noise (SNR) and a noise equivalent concentration (NEC) for a short list of compounds of interest to the sponsor. The radiometric model takes into account the off axis performance of the projector lens and associated radiometric efficiency due to overfilling the retroreflector arrays which is a function of range. The model also takes into account the performance of the retroreflector arrays as well as atmospheric attenuation. NESR is given by: Ad W [ =] (1) NESR = 2 D * ⋅η ⋅ Θ ⋅ Δσ ⋅ Δt cm ⋅ ster ⋅ cm −1 where each of these values are detailed in Table 2. Radiometric signal to noise (SNR) is given by
-2-
ε source ⋅ N(σ, Tsource ) [=] unitless (2) NESR where N(σ,Tsource) is the Planck equation for IR source temperature, Tsource, and εsource is the source emissivity (unitless). Noise equivalent concentration is given by 1 1 ⎞ mg ⎛ NEC = − ⋅ ln⎜1 − (3) ⎟ [ =] α ⋅ L ⎝ SNR ⎠ m 3 where α and L are the line strength (m2/mg) and pathlength (m) respectively. SNR =
Figures 2a and 2b show the projected per-channel SNR and NEC for m-cresol which is the weakest absorber on the simulant list as a function of range assuming 1 m retroreflector arrays are used. Note that the other simulants of interest to the army are sulfur hexafluoride, acetic acid, triethyl phosphate, and methyl salicylate. Values for εsource and Tsource are 0.9 and 1450°C (1723 K), respectively. We assume an absorption line strength of 4×10-4 m2/mg for m-cresol. We assume the plume fills half of the standoff range (i.e. pathlength = range owing to double pass). Table 2: Radiometric Projection Variables VARIABLE -1
Δσ (cm ) dint (cm) uint (rad)* Ωint (ster)** Θ (cm2ster)** D* (cm√Hz/W)*** dtel (cm) η (unitless)**** AD (cm2)*** Δt (s)
DESCRIPTION Spectral resolution Interferometer mirror diameter Interferometer ½ angle Interferometer solid angle Etendue Detectivity Telescope aperture diameter Radiometric efficiency Detector area Integration time
VALUE 4 1.27 .016 7.85×10-4 10×10-4 5×108 15 η(R) 7.85×10-3 2.5
Notes for Table 2: *
This angle is set by the throughput of the fibers (900 μm core, 0.25 NA) matched to that of the interferometer aperture diameter (1.27 cm) and is below the obliquity limit associated with 4 cm-1 spectral resolution and the highest optical frequency of 1428 cm-1 (7 μm). ∗∗
Ωint = 2π(1-cos(uint)) and Θint = Aint·Ωint where Aint = 1.27 cm2. Eleven 1 mm uncooled lithium tantalite (LiTaO3) single element detectors are used. **** Radiometric efficiency is based on a surface and vignetting loss analysis of the I-OP-FTIR as well as an analysis of the range dependent losses of the retroreflector arrays. Figure 2c shows the results of a model predicting the retroreflector performance as a function of range taking into account the following: ***
1. Geometric efficiency (geo) – the efficiency associated with each point emitted at the sensor aperture returning on itself with a diameter equal to twice the retroreflector element diameter of 1.9 cm. 2. Edge efficiency (edge) – the efficiency associated with edge and bevel losses (assumes .001” edges and bevels) 3. Angular accuracy efficiency (ang. acc.) – the efficiency associated with the angular accuracy of the retroreflector elements. Note that this effect dominates diffraction (at 10 μm) for the plastic injection molded retroreflector arrays. 4. Fill efficiency – the efficiency associated with the geometry of the retroreflector elements which have triangular facets and an associated 66% radiometric efficiency. 5. Atmospheric losses – the efficiency associated with water vapor and carbon dioxide absorption in the atmosphere. This projection uses Hitran3 reference spectra and integrates over the 7 to 14 μm spectral range. 6. Overfill efficiency – the efficiency associated with overfilling the retroreflector array with the open path beam.
-3-
Figure 2a: SNR vs. Range
Figure 2b: NEC vs. Range for M-Cresol
Figure 2c: Retroreflector Array Efficiency vs. Range
According to our radiometric model and based on our pathlength assumptions, all but the two outside channels (± 15°) meet the NEC requirement of 0.1 mg/m3 for m-cresol (the weakest absorber) over the full range of standoffs between 0 and 100 m. The requirement is fully met for standoffs up to 80 m and for all of the other chemical agent simulants of interest over the full range of standoffs between 0 and 100 m.
2.3 Optical Design Figure 3a shows the optical layout for the prototype I-OP-FTIR. We show one arm of the interferometer where the second arm is opto-mechanically identical. Radiation emitted by the IR source is coupled into the interferometer using a doublet lens to form an image of the source on the interferometer mirrors. The radiation traverses and exits the interferometer and is focused to a field point at the common end of the fiber optic bundle (also called the fiber optic image transformer). Each fiber therefore samples radiation from every point on the interferometer mirrors over an angular subtense which is below the obliquity limit associated with the spectral resolution and highest optical frequency of this system. Each fiber then transmits its respective portion of the total radiation to a point along the curved focal plane of the projection lens where it passes through an intensity beamsplitter prior to illuminating the projector lens. The result is an array of quasi-collimated beams transmitted to the remotely located retroreflector arrays where the divergence of each beam is dictated by the throughput match between the 900 μm, f/2 fibers and the 15 cm diameter projector lens. In other words, the divergence is about dfiber·ufiber/dproj = (900 μm)·(0.223 rad)/(15 cm) = 1.3 mrad where 0.223 rad is the half angle associated with the f/2 fibers. The pointing of the beams (i.e. the ± 15° spread) is dictated by the off-axis location of each fiber. All of the lenses shown in Figure 3a are germanium with a broadband antireflective (AR) coating for 7-14 μm. The secondary beamsplitters are also germanium with a 50% intensity beamsplitter coating for 45° incidence (with a broadband AR coating on the other side). The interferometer mirrors are diamond turned aluminum, and the interferometer beamsplitter/compensator pair are zinc selenide.
-4-
Figure 3a: Optical System Layout IR Source Source Lenses
11x Secondary Beamsplitters
Exit Lens
Interferometer Mirror
Beamsplitter/ Compensator Fiber Optic Image Transformer
Projection Lens
The fiber optic image transformer (Figure 3b) is composed of eleven silver halide IR fibers which are bundled at one end and free at the other. The image shows all fiber ends covered because exposure to ultraviolet (UV) light will degrade the IR transmission. The fiber cores are 900 μm, and the NA is 0.25 (approximately f/2). The purpose of the image transformer is to evenly distribute the modulated radiation exiting the interferometer into an array of (modulated) sources spread along the curved focal plane of the projector lens. The off-axis location of each “source” relative to the optical axis of the projector lens determines the pointing of each OP beam within the fan. Figure 3b: Fiber Optic Image Transformer Free Output End
Common Input End
Figure 3c: Spot Diagram at Retroreflector Arrays 12° off-axis
9° off-axis
1m
1m
15° off-axis
1m
The projector lens is a germanium asphere designed for optimal off-axis collimation (±15˚) at f/2 set by the fibers. Figure 3c shows the spot sizes at a range of 100 m for on-axis and 3˚, 6˚, 9˚, 12˚, and 15˚ off-axis where the scale bar is 1 m. We extracted radiometric efficiency predictions used in the radiometric projections (Section 2.2) from these spot diagrams. 2.4 Mechanical Design
1m
-5-
on-axis
1m
3° off-axis
6° off-axis
1m
Figure 4a is a solid model showing the mechanical design of the I-OP-FTIR. Figures 4b and 4c show close-ups of the interferometer and mounting for the fiber optic image transformer from two views.
Figure 4a: Solid Model of I-OP-FTIR Sighting Scope
Cover
Projector Lens Mount
Image Transformer Mount
DAQ Terminal Board
Interferometer Heat fins for IR source
Figure 4a: Solid Model of I-OP-FTIR Figure 4a is a solid model of the I-OP-FTIR mechanical system. The interferometer, fiber optic image transformer mount, projector lens, detectors with amplifiers, and terminal board for the data acquisition are all housed in the sensor head. The fiber optic bundle is not shown in this model. Electronics to control the interferometer as well as the data acquisition are located on PCI boards housed by a “lunchbox” PC. The lunchbox PC also contains the power supply for the IR source. The mechanical design allows for adequate (passive) heat dissipation for the IR source.
-6-
Figure 4b: Solid Model – Interferometer View
Figure 4c: Solid Model – Image Transformer View IR Source with Heat Sink
Fiber Optic Image Transformer Mount
Fiber Bundle Launch Mount
11x LiTaO3 Detectors
Detector Preamp Boards (x2)
Moving Mirror
Interferometer
Fiber Bundle Launch Mount
Encoder Reference
Figure 4b: Solid Model – Interferometer View Figure 4b shows a close-up of the interferometer which employs an encoder-based reference system. The encoder-based interferometer is a variant of our J-Series modulator4 developed for the Joint Services Lightweight Standoff Chemical Agent Detector (JSLSCAD). The IR portion of the interferometer is identical to the J-Series with diamond turned aluminum mirrors and a zinc selenide beamsplitter/compensator pair.
Figure 4b: Solid Model – Image Transformer View Figure 4c shows a close-up of the image transformer mount which holds the eleven free/output fiber ends, eleven secondary beamsplitters, and eleven LiTaO3 detectors. The free fiber ends are plane ferrules which can be adjusted in z within the mount (relative to the projector lens) to set the collimation of each OP beam. Not shown is an initialization detector opposite the center LiTaO3 detector. The initialization detector samples the interferogram from the first secondary beamsplitter reflection and does not require the retroreflector to be in place. This detector is used to both initialize the scan as well as check the alignment of the interferometer.
2.5 System Build and Integration
Figure 5a: Integrated I-OP-FTIR – Front View
Figure 5a is a photograph of the partially-integrated IOP-FTIR system (without detectors) showing the germanium projector lens and a laser sighting system that aids in the alignment of the retroreflector arrays. The laser sighting system rotates and snaps into eleven fixed positions every 3° (corresponding to the pointing of the eleven OP beams). The fiber optic image transformer mount is shown in the background.
Laser Sight
Germanium Projector Lens
Figure 5b shows a side view of the I-OP-FTIR. This shows the relative position of the projector lens to the fiber optic image transformer (operating at f/2). The fiber optic image transformer is shown on the right where the common/input end is held at the field point at the exit port of the interferometer by a tip/tilt mount with x, y, and z adjustment. The free/output ends are held in their bores by set screws and ultimately a UV epoxy. The eleven secondary beamsplitters are germanium and therefore seal off the free/output ends from all visible and UV light which would otherwise damage the fibers. The eleven LiTaO3 signal detectors
Fiber Optic Image Transformer Mount
-7-
are mounted above the secondary beamsplitters, and the single initialization detector (not shown) is mounted below the secondary beamsplitter opposite the center channel signal detector. This photo also shows the two detector preamplifier boards and the IR source heat sink. Figure 5b: Integrated I-OP-FTIR – Side View Laser Sight Germanium Projector Lens
11x LiTaO3 Signal Detectors
Fiber Optic Image Transformer Mount
Fiber Optic Image Transformer
Detector Preamp Boards
IR Source Heat Sink Fiber Optic Image Transformer Launch Mount
Figure 5c is a close-up of the fiber optic image transformer and mount before the detectors are mounted. This figure shows the curved profile of the mount which effectively allows us to negate the effects of field curvature of the projector lens on the collimation of the OP beams. Figure 5c: Fiber Optic Image Transformer Close-up Fiber Optic Image Transformer Mount
Interferometer
11x LiTaO3 Signal Detectors
IR Source Heat Sink
Fiber Optic Image Transformer Fiber Optic Image Transformer Launch Mount
Figure 5d is a close up of the interferometer showing the encoder and moving and stationary mirrors. The lens assemblies which image the source onto the interferometer mirrors and focus the interferometer output to a field point are enclosed in aluminum mounts. The IR source is a miniature igniter manufactured by St. Gobain.5
-8-
Figure 5d: Interferometer Close-up IR Source
Source Imaging Lenses
Encoder Reference Fiber Optic Image Transformer Launch Mount
Moving Mirror
Stationary Mirror
Exit Lens
The fiber optic image transformer was aligned as follows. With the IR source aligned to the interferometer and the interferometer itself aligned, the common/input end of the fiber optic image transformer was aligned to the field point by serially monitoring the peak-to-peak interferogram amplitude (on an oscilloscope) of the energy exiting each fiber and adjusting the common end in x, y, z, θx, and θy until all were optimized. Note that the alignment of the common end is re-optimized as a final step once all detectors are in place and we have the ability to simultaneously monitor all eleven channels. The free/output fiber ends were then positioned for collimation relative to the projector lens using an IR camera focused at infinity to look back through the projector lens at each “source”. The fiber ends were each in turn adjusted in z to minimize the spot imaged by the IR camera. Figures 6a through 6k show these images for the eleven channels. The image of the 900 μm fiber formed by the 150 mm focal length IR camera lens should be about dimage = dfiber·FIR camera/Fproj = (900 μm)·(150 mm)/(360 mm) = 375 μm or about seven or eight, 50 μm IR camera pixels (the camera format is 120×160 pixels). Following this step, a retroreflector array was then positioned in fronts of the I-OPFTIR, and each detector was brought into focus by adjusting each in x, y, and z to optimize the peak-to-peak interferogram amplitude (Figure 7). The I-OP-FTIR was rotated to direct each beam in turn onto the retroreflector array for each detector alignment. Note the interferogram shown in Figure 7 was acquired at a much higher mirror velocity and associated modulation frequency than will be used in operation (to provide a constant signal); because of the detector bandwidth, a much cleaner interferogram will result at the operating mirror velocity. The interferogram shown in Figure 7 is also acquired with open-loop scanning. Once all eleven channels were aligned and the detectors wired to the data acquisition system, a final optimization was made by adjusting the common end of the fiber optic image transformer while simultaneously monitoring all eleven detector outputs.
-9-
Figure 6a: -15º OP Beam Figure 6b: -12º OP Beam Figure 6c: -9º OP Beam Figure 6d: -6º OP Beam
Figure 6e: -3º OP Beam Figure 6f: 0º OP Beam Figure 6g: 3º OP Beam Figure 6h: 6º OP Beam
Figure 6i: 9º OP Beam Figure 6j: 12º OP Beam Figure 6k: 15º OP Beam
Figure 7: Interferogram Used for Alignment
2.6 Next Steps At this point in time the I-OP-FTIR is fully integrated and is awaiting test. System characterization will include SNR measurements over a series of standoff ranges up to 100 m as well as spectral resolution using a 10.6 μm carbon dioxide laser. The interferometer itself has already been previously characterized for sampling errors, etc. Following system characterization, a library or background spectra will be acquired for the purpose of training the automated multicomponent algorithms6 which will be applied to identify and quantify the chemical agent simulants of interest to the customer as well as a short list of compounds we will test at OPTRA. We will then perform a short series of evaluation to verify the multicomponent algorithms are working properly prior to shipping the instrument to University of North Carolina for a full tomographic demonstration
- 10 -
3
SUMMARY AND FUTURE PLANS
In summary, we have presented system level, optical, and mechanical designs of the I-OP-FTIR system for 3D cloud profiling of chemical agent simulant plumes at test ranges and chambers. We have completed the build and integration of an I-OP-FTIR prototype and will begin testing shortly. Test plans include characterization of SNR at a series of standoff ranges as well as spectral resolution. We will perform a series of spectroscopic multicomponent demonstrations at OPTRA before sending the prototype to University of North Carolina for a full tomographic demonstration. University of North Carolina will demonstrate the tomography using a small array of Teflon balloons they have developed specifically for this purpose to serve as stationary plumes which can be filled to known concentrations with various chemical agent simulants. The I-OP-FTIR prototype will be serially positioned around the balloons and spectra will be collected at each position (simulating multiple instruments). The results along with sensor geometry will then be used to tomographically reconstruct the stationary plumes for each of multiple compounds. Future direction for this work may include extended standoff range operation as well as processing upgrades to handle the real time requirements which were not addressed during the Phase II.
Acknowledgements This research was conducted under a Small Business Technology Transfer Phase II contract funded by the U.S. Army. The technical monitor is Dr. Alan Samuels, Edgewood Chemical Biological Center.
References 1
L. A. Todd, S. K. Farhat, K. M. Mottus, and G. J. Mihlan, “Experimental Evaluation of an Environmental CAT Scanning System for Mapping Chemicals in Air in Real-Time,” Appl. Occupational and Environmental Hygiene, Vol. 16, No. 1, (2001). 2
B. R. Cosofret, D. Konno, A. Faghfouri, H. S. Kindle, C. M. Gittins, M. L. Finson, T. E. Janov, M. J. Levreault, R. K. Miyashiro, and W. J. Marinelli, “Imaging Sensor Constellation for Tomographic Chemical Cloud Mapping,” Appl. Opt. Vol. 48, No. 10, (2009).
3
Ontar Corporation, North Andover, MA.
4
J. R. Engel, J. H. Rentz, and D. L. Carlson, “FTIR modulator for first responder applications,” Proc. SPIE Vol. 5268, Chemical and Biological Standoff Detection, Photonics East, Providence, RI, (2003). 5
World Wide Web
6
J. R. Dupuis, D. J. Mansur, J. R. Engel, R. Vaillancourt, L. Todd, and K. Mottus “Imaging Open-Path Fourier Transform Spectrometer for 3D Cloud Profiling,” Proc. SPIE Vol. 6954, Chemical, Biological, Radiological, Nuclear, and Explosives (CBRNE) Sensing IX, Defense and Security, Orlando, FL, (2008).
- 11 -
