|Ellen V. Miseo and Norman A. Wright
|Developing a chemical-imaging camera
Infrared (IR) imaging, the technology for
using an IR camera to generate images of
hot sources, has become an important tool in
the optoelectronics industry, and it is used
routinely in industrial sensing, security, and
firefighting. Major developments in detector
technology have made IR imagers and focalplane
arrays available to industry and in
technical areas such as quality control, where
the cost was previously prohibitive.
|Figure 1. A
chemical mix on a plate (visible image at top left) is represented
by a characteristic cube where the front face is an infrared
image at one wavelength, succeeding slices are images at increasing
wavelengths, and on the right face is traced the absorbance
intensity for various species at increasing wavelength.
These advances have also laid the foundation
for developing a chemical-specific imaging
camera, one dependent on IR technology
and capable of capturing the chemical composition
and distribution of a sample in seconds.
When fully developed to the size of a
camcorder with almost instantaneous imaging,
this technology will have applications in
research, industrial, security, military, and
first-responder operations. It will, for example,
enable a biohazards team to immediately
determine whether a clear liquid leaking
from a truck is water or a toxic fluid.
|Figure 2. A
visible image of a styrene acrylonitrile copolymer sample with
a defect (a) does not show nearly as much information as an
infrared image at a wavelength of 1735 cm–1 (b).
Infrared detectors use two types of materials,
either photon detectors, such as indium
antimonide (InSb) and mercury cadmium
telluride (MCT), or thermal detectors
such as deuterated triglycine sulfate (DTGS).
Both classes of detectors have proven technologies,
and they are used as single-detector
elements in many applications, including
An IR imaging system takes the technology
of the detector and packages it to generate
images. In a mechanical scanning system,
one or more mirrors scan IR radiation
across a linear array of these detectors and
slowly build up an image of the sample.
Focal-plane systems are two-dimensional
detector arrays located at the focal point of
the optical system and act as the film in a
Definitions of various IR wavelength ranges
differ, especially in detector technology,
between chemists and physicists (Table 1).
Initial development of focal-plane arrays
in the mid-IR (2–15 µm) was funded by the
astronomy community and the U.S. Department
of Defense. Defense spending drove
the technology toward military technical
specifications and applications that included
heat-seeking missiles and night-vision
heads-up displays for aircraft and ground
vehicles. Recently, however, emphasis has
shifted from military uses to the development
of commercial applications by private
The technology for using focal-plane array
detectors to sense IR radiation and generate
thermal pictures has developed and matured
over the past 15 years. Today, the technology
is divided into two technical areas, uncooled
and cooled. Table 2 lists some materials used
for array detectors, each of which has parameters
that make it appropriate for some
applications and not others.
Uncooled technologies currently receive
the most development funding. Often, however,
either their wavelength response is inadequate to determine
chemical spectral signatures,
or the optical
requirements of the detector
material are too stringent
for a spectrophotometer
focal-plane arrays are used
in some commercial
security systems and
industrial inspection, so
prices should fall as the
technology matures. The most consumer-oriented
application is called Night
which General Motors’ Cadillac Division
began incorporating into its DeVille models
in 2000. Night Vision relies on an uncooled
camera, which projects an image of the road
ahead onto the windshield and allows the
driver to see objects before the headlights
Chemical information derived from the
fundamental vibrations of molecules is available
in the 2- to 20-µm region of the spectrum.
A subset of that region, called the fingerprint
region, lies between 5 and 10 µm
and is the richest in chemical data. Chemists
have used the fingerprint region since the
1950s to determine the chemical identity of
a compound by measuring frequency-dependent
absorbances. Initially, the instruments
used gratings or prisms and scanned across
the wavelengths of interest.
|Figure 3. An
infrared reflective image of the inside liner of a soda can
shows considerable variation in thickness and composition (a).
A typical spectrum from the liner shows an excellent signal-to-noise
ratio in a 5-s data collect (b).
In the late 1960s, chemists took a cue
from physicists and began using Fourier
transform IR (FTIR) instruments for chemical
analyses. The first commercial FTIR unit
entered the marketplace in 1969, and by the
mid-1980s, almost all laboratory-based general-
purpose IR instruments were FTIRs.
Today, FTIRs can capture the entire IR spectrum
in a single scan of the interferometer
and obtain a high-quality spectrum in less
than 1 s, but without
Coupling an FTIR to a
detector with a broadrange
gives both a spectral
and spatial picture.
This coupling idea
dates back to 1972,
when A. E. Potter of the National Aeronautics
and Space Administration patented a
concept for a multispectral imaging system
using a Michelson interferometer. Practical
execution of this system took until 1979,
when Potter and R. J. Huppi of the Stewart
Radiance Laboratory (Bedford, MA) independently
implemented the idea for atmospheric
and astronomical measurements.
Again, chemists were slow to notice the
technology, and it took twenty-five
years for them to use it in a
|Figure 4. An
infrared image at 1,675 cm–1 of a metal surface that
was contaminated with methyl salicylate and then cleaned still
shows hot spots (red) where the contaminant was not removed
(a). An image at another characteristic frequency (1,090 cm–1)
confirms the contamination (b).
The first commercial spectrochemical
used in a chemical laboratory was
introduced in 1995. It incorporated
an InSb detector, which
gave chemists some chemical
information. But because of the
detector’s long-wavelength IR
cutoff at about 5 µm, it had limited
uses. To overcome this problem,
Digilab introduced a commercial
MCT-based spectrochemical imaging
system in 1997. MCT focal-plane arrays
available at that time typically had large formats
(256 × 256 pixels or larger) and were
custom-designed for military or astronomical
applications. The new analytical instruments
used a smaller-format device (64 × 64
pixels), the only cost-effective MCT array
readily available at the time, which was
developed for the U.S. Army’s heat-seeking
Javelin missile. Most of the early spectroscopy
studies of chemical systems published
in the chemical literature since 1995
have used a 64 × 64-pixel MCT array, commonly
called the Javelin detector.
The Javelin detector’s essential technology
did not change much between 1997
and 2000 because of the limited number
of low-cost MCT focal-plane arrays commercially
available. The Javelin detector
has several design parameters that limit its
application in laboratory spectroscopy.
First, the device’s rolling-mode readout, in
which it reads one column of data as it
integrates the next, is slow and requires the interferometer to
scan at slow speeds. As
a result, the amount of
data gathered is only
about 3% of that possible,
because the pixels
are sampled sequentially
and large amounts
of usable information
|Figure 5. Infrared
images of an eggshell contaminated with salmonella show lower
concentrations of calcium carbonate, the major component of
eggshell, where the salmonella attenuates the 1,460-cm–1
carbonate absorbance (top right in a) and high concentrations
of salmonella at the characteristic 1,630 cm–1 frequency
Matching the flux levels
from the spectrometers
is another serious
issue with Javelin detectors. In a spectroscopic
experiment, the spectrometer typically
provides high flux levels. To use all the
energy provided to the detector most efficiently,
the focal-plane array must
be capable of dealing with the
high flux levels and remain linear,
which the Javelin cannot do
because its design does not allow
independent adjustment of the
integration time regardless of
frame rate. To couple efficiently to
the interferometer, the Javelin
detector’s limitations in a spectroscopic
application can be overcome
by using a snapshot readout,
which increases the detector’s efficiency
from about 3% to more
The potential for instruments with wavelength
specificity and spatial resolution to
deliver images in the IR opened the door for
a camera that could quickly pick up a specific
chemical signature and generate a threedimensional
data cube of spectral, spatial,
and intensity information (Figure 1, above).
With the experience gained with the
Javelin detector, Digilab developed a second
generation of focal-plane array systems
for use with FTIR spectroscopy. This family
of mid-IR focal-plane arrays is available in
formats from 16 ×16 to 128 ×128 pixels.
Arrays of 64 ×64 pixels can be read out
rapidly, allowing their use with a conventional
rapid-scan FTIR spectrometer and
providing fast data-acquisition times. Even
the 128 × 128 format array can read out
faster than the older technology using a
scan-data collect rate of 800 Hz. These two
advantages led to the ability to create a
chemical-specific imaging camera, which in
the laboratory can produce a spectral and
spatial image in 2 s.
The new camera system consists of a
commercial FTIR instrument, the newly
designed MCT array, and a PC with a framegrabber
card, and it runs Windows 2000.
In some applications described here, the
camera was put on a microscope optical
system designed to work in the IR to obtain
approximately a 15× magnification of the
sample. The measurement of chemical heterogeneity,
which drove the initial development
and applications of this system, typically
happens at the microscale level.
Coupling the focal-plane array to a microscope
that can view in the visible and collect
in the IR gives a system with a spatial
resolution of approximately 5 µm. This resolution,
for example, can reveal whether a
time-release drug coated on an implantable
polymeric device has been heterogeneously
distributed so that it will deliver the
Combining an FTIR spectrometer
with an array detector enables the collection
of all data and spatial information
at once. Using FTIR technology
enables collecting the full range of IR
spectra (4,000 to 950 cm–1) from each
pixel in the detector when the system
is used for spectrochemical imaging.
This technology can deliver a picture
of the chemicals that make up a sample
and their distribution in it.
Figure 2a shows a visible image of a
styrene acrylonitrile copolymer with a
defect, and Figure 2b, obtained in the IR,
illuminates the defect. Data on this sample
were collected with 32 scans of the interferometer
at 5 kHz and 8-cm–1 resolution in
about 30 s. Acrylonitrile has a functional
group that contains a carbon with a triple
bond to a nitrogen, which has an IR
absorbance at 2238 cm–1. Other species in
the sample had absorbances in the spectral
regions characteristic of styrene. The defect
area showed up strongly at 1735 cm–1. IR
absorbance at this wavelength is due to a
carbonyl species, which has a carbon double-
bonded to an oxygen atom. Further
examination of the spectra showed a speck
of stearate material.
In a spectroscopic experiment, it is
important to determine the signal-to-noise
ratio in order to differentiate chemistries.
Figure 3a shows an image taken in reflection
on the inside of a soda can. At the spatial
resolution achieved by the optics
(5 µm/pixel), the inside liner shows considerable
variations in thickness and composition. Figure 3b shows a typical
from the can liner illustrating an excellent
signal-to-noise ratio in a 5-s data collect.
In a chemically contaminated system,
decontamination efficiency is an important
parameter to measure. As an example, we
applied methyl salicylate to a metal surface.
The compound has several IR absorptions in
the mid-IR region, which can be used to
illustrate the power of the spectrochemicalimaging
technique. Figure 4a shows the
metal surface after contamination with
methyl salicylate and cleaning with a swab.
The image—generated at 1,675 cm–1, a
strong absorption in the chemical—shows
hot spots (red) where the contaminant was
not removed. This data set was collected in
approximately 80 s. A chemist trying to confirm
the identity of a chemical compound
first looks for a characteristic absorbance
and then looks for a confirming absorbance
at a different frequency. At 1,090 cm–1, a
similar image confirmed the contamination
in the lower right corner (Figure 4b).
We also examined the surface of an
eggshell for biological contamination. The
eggshell surface was inoculated with nonpathogenic
salmonella. Figure 5a shows an
image constructed at the absorption for calcium
carbonate, the major component in
eggshell. In it, the area of salmonella contamination
shows up as a lower concentration
of carbonate because the overlaid salmonella
attenuates the absorbance from
the carbonate species. The image in Figure
5b clearly shows the location of the salmonella
on the right-hand side of the image.
Preliminary data demonstrates the feasibility
of developing a chemical-specific
camera with important applications in
industry, research, and national security. As
the new technology matures, the speed and
ruggedness of the equipment will improve.
The ultimate goal is to develop a camera
that provides images and chemical identification
in real time.
Crocombe, R. A.; Wright, N.; et al. FT-IR
spectroscopic imaging in the infrared “fingerprint”
region using an MCT array detector.
Microscopy and Microanalysis, Vol. 3,
Supplement 2, Proceedings; Springer-Verlag:
New York 1997; pp. 863–864.
Huppi, R. J.; Shipley R. B.; Huppi, R. E.
Balloon-borne Fourier spectrometer using a
focal plane detector array. In Multiplex
and/or High Throughput Spectroscopy, Proc.
SPIE 1979, 191, 26–32.
Lewis, E. N.; et al.
imaging using an infrared focal-plane array
detector. Anal. Chem. 1995, 67, 3377.
Potter, A. E. Multispectral Imaging System;
U.S. patent no. 3,702,735.
Wells, W.; Potter, A. E.; Morgan, T. H.
Near-infrared spectral imaging Michelson
interferometer for astronomical applications.
In Infrared Imaging Systems Technology,
Proc. SPIE 1980, 226, 61–64.
Ellen V. Miseo is
imaging products manager and Norman
A. Wright is worldwide applications
manager at Digilab, LLC, in