1. technology and computing

THZ-SYSTEMS
How to Perform a Security Check
without Undressing People
Can Terahertz Technology Answer the Conflict
Between Security and Privacy?
 People have a desire for security, especially when they are about to enter an
aircraft. Nobody wants to sit next to a terrorist flying 30,000 ft above ground. So,
modern technology is expected to provide a solution to screen anybody willing to
enter an aircraft for hidden threats. Obviously the actual solution using metal detectors and manual control is secure only
to a certain level; moreover it is inefficient
and offending. Security screening and respecting privacy at the same time constitute a conflict which cannot be easily solved. Characteristic for that dilemma is the
actual public discussion regarding the so
called “strip scanner” [1], where a technology with high security standard obviously violates privacy when in a figurative
sense it undresses people and even worse
illuminates them with arguable radiation
(x-ray or millimetre waves). This article
describes a possible solution, which respects people‘s privacy and complies with
their basic need for security.
According to German standards, technologies which should serve man must meet basic ethic principals. So, the security research
program of the German ministry of education and research (BMBF) is accompanied an
ethical review board [2]. The ongoing project “THz-Videocam” [3] pursues a concept,
where a security screening is done passively
by tracing the shadow of suspicious objects
on the terahertz emission from the human
body. It intentionally eliminates two major
concerns in public acceptance: the active
­illumination and the ‘naked appearance’ of
the recorded images.
A Terahertz Security Camera:
How Should it Perform?
At first sight and above all technical consideration there are some basic criteria, which
make the use of a security camera reasonable. The gain of security should be noteworthy, meaning that it must have advanta-
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim The AutHors
TORSTEN MAY
HANS-GEORG MEYER
Torsten May was born
on 28th of February 1971
in Gera (Thuringia). In
1997 he completed his
degree in physics at the Friedrich Schiller
University in Jena.
Since then he has been working for the
department “Quantum Detection” at the
Institute of Photonic Technology. His subject is the development of detectors based
on superconducting technology, in particular for sensing electromagnetic radia­
tion. Since 2007 he is leading the group
“Quantum Radiometry” as part of the named department.
Hans-Georg Meyer was
born on 19th of August
1949 in Plauen (Saxonia).
In 1981 he received a
doctorate in physics at the Friedrich Schiller
University in Jena and habilitated there in
1991. Since 1974 he has been working on
the subject of weak superconductivity, in
particular on the development of ultra-sensitive magnetometers. In 1993 he became
the head of the department “Quantum Detection“ at the Institute of Photonic Technology. Here he is responsible for the development of superconducting electronics
and systems and their application in sophisticated measuring techniques.
●●
●●
Torsten May
Institute of Photonic Technology
Albert-Einstein-Str. 9
07745 Jena, Germany
Tel.: +49 (0)3641 206123
Fax: +49 (0)3641 206199
E-Mail: [email protected]
Website: www.ipht-jena.de
ges over existing technologies. To be more
precise: it has to detect and localize not
only hidden metallic weapons, but also
­ceramic ones, plastic explosives, liquids and
so on. At the same time it has to do it
­without risking the health of a person under
test and the operating personnel.
Further, a device which represents a
more “camera-like” imaging style ought to
have advantages over the traditional portaloriented check scenario. Besides the obvious benefit of flexible installation a camera
can image a suspicious person already at a
secure distance, which protects not only
the check point but also the personnel from
suicide bombing attacks.
Dr. Hans-Georg Meyer
Institute of Photonic Technologies
Albert-Einstein-Str. 9
07745 Jena, Germany
Tel.: +49 (0)3641 206116
Fax: +49 (0)3641 206199
E-Mail: [email protected]
Website: www.ipht-jena.de
The last point on the “to do” list is a definite demand for moving pictures. A video
camera is not only a nice feature – it will gain
in detection probability because a hidden
object might be invisible from a certain angle of view, leaving it undetected on a still
image. A video will show the person under
test moving, giving a multitude of angles
and hereby enhanced chances of positive
detection.
To summarize, a promising candidate for
a next generation security tool would
be a device which combines the following
features: the ability to scan from a secure distance with sufficient spatial resolution, passive operation (preferable in all environments,
www.optik-photonik.de 31
THZ-SYSTEMS
without the ‘outdoor trick’ to use natural
contrast amplification), and video frame rate.
The following paragraphs will demonstrate
the technical challenges on the way.
entrance
aperture
D
Stand-Off Detection of Threats
Passive Terahertz Imaging
For reasons derived from the scenario in the
paragraph above, the task remains to detect
terahertz light in a narrow band around 800
µm to 900 µm. Every object with a temperature above zero Kelvin emits electromagnetic waves in a broad band, with a maximum
α min
diffraction
pattern
λ
1.22 ·
D
α min =
Figure 1: Diffraction-limited resolution.
10–10
max @ 18 THz
10–11
spectral emission (Wm–2Hz–1)
Although the term terahertz camera is used
very often, most of the proposed solutions
safely go below the one terahertz limit. There are two reasons for this: firstly, the atte­
nuation of a humid atmosphere becomes
­interfering above 1 THz, and secondly, the
ability to find something hidden under cloth
vanishes likewise. Unfortunately this constitutes a trade-off: using lower terahertz frequencies has to be paid by a decreasing
spatial resolution due to diffraction.
Obviously one needs to find a compro­
mise. It is known from astrophysics, that a
humid atmosphere provides a few narrow
windows, where the attenuation might be
acceptable for the propagation of light. Promising candidates are windows at 0.35, 0.6
and 0.85 THz. From clothing materials it is
known that transparency becomes insufficient above 0.6 THz. Therefore the approach
described in this paper has chosen the
0.35 THz window, corresponding to wavelengths between 800 µm and 900 µm.
In comparison to traditional optics these
wavelengths are still interferingly large. Since the diffraction limited resolution of any
optics depends on the ratio between the
wavelength λ and the diameter of the entrance aperture D (see Fig. 1), for a sufficiently high resolution sizable optical components are required.
This is even more important for the design of a stand-off camera. For example an
aperture of 1 m diameter designed for a frequency of 0.35 THz can discriminate object
points separated by an angle of 1 mrad.
This would allow pinpointing objects with
dimensions in the order of one centimetre,
which constitutes the minimum specification for a security camera.
To demonstrate the abilities at IPHT a
prototype with a 40 cm telescope was built
for a 5 meter scenario, mainly to test it in
the restricted space of a lab environment.
Hereby a smaller entrance pupil was adequate, cases with larger distances are almost a pure upscale.
image
plane
10–12
10–13
10–14
band of interest:
335 to 375 GHz
10–15
10–16
10–11
10–12
10–13
frequency (Hz)
The INstitute
Institute of Photonic Technology
Jena, Germany
The Institute of Photonic Technology
(IPHT) in Jena is an application oriented
research facility institutionally funded by
the Free State of Thuringia. About 270
IPHT employees are working within two
research divisions: Photonic Instrumen­
tation and Optical Fibers & Fiber Appli­
cations on custom made solutions for
practical applications. These are achieved
by developing new scientific concepts to
overcome technological boundaries. Furthermore, these concepts are implemented in close collaboration with numerous
industrial and academic partners during
the development of new components or
devices. The IPHT combines its innovative
qualities in order to develop an idea with
basic research into a prototype. To reach
this goal, the institute possesses e. g. an
area of about 300 m2 for clean room laboratories with e-beam lithography and
equipment for micro-fabrication.
www.ipht-jena.de
32 Optik & Photonik December 2008 No. 4
10–14
Figure 2: Spectral
emission of a black
body at 310 Kelvin,
per frequency unit.
depending on its actual temperature. Figure
2 shows the spectral emission of a black
body with a temperature of 310 K (keep in
mind: at terahertz, the human body can be
seen as an almost perfect black body), calculated by Planck’s equation. The emission is
maximal at a frequency of about 18 THz and
drops dramatically to both sides. In the band
of interest it is already almost three orders of
magnitude lower than the maximum.
One way to detect such weak signals is
to use cooled power detectors, which is a
proven method known from infrared technology. By cooling the detector close to absolute zero the smallest detectable power
(referred to as “Noise Equivalent Power”
NEP), can be as small as one requires to detect extremely faint terahertz signatures
from astronomical objects [4].
In the variety of suggested detectors for
terahertz light the so called Transition Edge
Sensor (TES) has emerged as the so far most
effective combination of high sensitivity on
one hand and ease of operation on the other.
Basically a TES transforms the incoming radiation to thermal energy by absorbing it in an
appropriate antenna. Due to this it is possible to measure the temperature increase. For
this purpose a TES uses a superconductor
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
THZ-SYSTEMS
Figure 3: detector chip with seven TES
bolometers.
Figure 4: Mechanical scanner to perform
a spiral sampling.
rence as Watts per square root of bandwidth,
meaning that if the integration time becomes longer, the smallest detectable power
also decreases. However, a long integration
is inappropriate because the final goal will be
a video camera, where the integration times
can be as short as milliseconds. This demand
is intensified in the case that the detector receives the signal after modulation by a fast
mechanical scanner (see next paragraph).
The IPHT camera uses a detector with a
time constant of 100 µs. At the bandwidth of
10 kHz the NEP has to be as low as 2 x 10-16
W/Hz1/2, which can be achieved by cooling
the detector close to absolute zero. Using
modern cooling technologies like pulse tube
refrigeration this poses no particular effort.
The TES is manufactured using sophisticated
micro-technology, based on freestanding
SiN membranes, as can be seen in Figure 3.
Video Frame Rate:
Array Size vs. Mapping Speed
Figure 5: Camera prototype as it was tested in lab and in real scenarios.
being operated in its transition. This implies
two consequences: Firstly, one needs a superconducting material with a transition
point at the chosen working temperature,
and secondly, this working temperature has
to be extremely stable. In practice this would
make the concept almost infeasible.
The idea of the voltage biased bolometer [5] smartly deals with both terms. It utilizes a bias voltage which drives the thermometer to heat up itself to its transition
point. The power of this self heating is inverse proportional to the electrical resistance at the respective working point. Since an incident radiation would heat up the
detector, the increasing resistance of the
thermometer would intrinsically decrease
the self heating. This effect self-stabilizes
the operating temperature.
The IPHT (Institute of Photonic Technology in Jena, Germany) prototype uses such
a TES. At the example of the imaging of a
typical scenario to be expected for a security camera one can estimate the needed
performance. As a model lets assume a ‘human black body’ (T1 = 37 °C) in front of a
background at T2 = 20 °C. Using Planck’s
equation, one can calculate the difference
in thermal power for these two tempera-
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim tures. This measure has to be related to the
emitting area and the frequency band of
interest. Assuming the optical layout of the
prototype, at 5 meter distance the field of
view of a single detector is focused on a
small spot of 1 cm diameter, so the detector
will only sense energy coming from that
area. For reasons mentioned above the
bandwidth is restricted to be around
40 GHz, so you will end with a power difference between a spot at 37 °C and a spot at
20 °C which is as small as 6 x 10–9 W!
Unfortunately, the 1 cm spot does not
collimate its energy into the direction of the
detector, rather than evenly distributing it to
a hemisphere. The fraction which ends up
on the surface of the 40 cm telescope mirror
is as small as 0.08 %. And that is not the end
of the story: a detector able to resolve such
difference would yield only in a black-andwhite image. If one wants to create a
greyscale image with 256 shades, the difference to be sensed is as small 2 x 10-14 W!
The figure of merit of the power detector
in use is its intrinsic noise caused by random
motion of electrons and phonons. These
noise sources can be cancelled out by integration over time. This fact is accounted by
defining the smallest detectable power diffe-
In the case of such a relatively new and
therefore still expensive detector technology
it is necessary to find a good compromise
between the complexity of the imaging array and the required imaging performance.
The TES in use can achieve a rather short
time constant of 100 µs. At first approximation, even at full video rate (25 Hz = 40 ms)
one TES could possibly record 400 separated
values. Therefore, it is feasible to combine
only a few detectors with a quasi-optical
scanner, which is comparable to the situation which occurred 30 years ago with the
emergence of the first detector arrays for IR.
The actual prototype uses only one detector to record images at 1 Hz frame rate,
corresponding to approximately 10,000
data points or a 100 x 100 pixel map. The
remaining challenge is to design a scanner.
In contrast to scanning mechanisms used at
visible or infrared wavelength, here the
scanner has to move optical components
with a notable inertial mass. Standard techniques like piezoelectric actuators fail to accelerate such large objects.
This was the reason to build the unusual
spiral scanner used in the prototype (see Fig.
4). Here, the secondary mirror is rotated
around the optical axis of the telescope,
whereas it is tilted around one orthogonal
axis. The advantage of that idea is that the
whole field of view can be scanned by controlling only one degree of freedom: the tilting angle. The only trade-off is the unusual
pattern of the recorded data points, corresponding to a polar grid, which also causes
the recorded image to be circular. However,
a modern software algorithm can handle it
easily, so the pros outweigh the cons.
www.optik-photonik.de 33
THZ-SYSTEMS
Proof of Principle
To conclude the three paragraphs above,
one can say that the needed components
for a passive terahertz video camera are
forthcoming. What is left is to demonstrate,
that such an approach can supply relevant
security information without being mistrusted and accused of violation of privacy.
Hence, a demonstrator was built at IPHT
combining an ultra sensitive TES detector
and optics for 5 m stand-off imaging. Figure 5 shows the device as it is realised in lab.
This prototype is in use now for almost
one year. It has been tested under various
conditions, in the heated lab during summer time where the radiometric contrast
was as low as 5 °C as well as in harsh outdoor conditions of a military camp during a
manoeuvre of the German army.
The design parameters (optical resolution, thermal sensitivity) have been achieved.
Within the first year, the image acquisition
time was decreased from 25 seconds down
to 1 second, using only one TES detector.
The capability to upscale the concept to a
video-rate camera is obvious: naively calculated an array of 25 detectors could possibly yield in a 25 Hz frame rate. Of cause, there will be redundancies if an array is used in
combination with a spiral scanner, however, the number of pixels needed for video
rate remains manageable.
Figure 6 shows a typical image recorded
by the prototype (on the left). Remarkable at
first sight is the ‘ghostlike’ appearance of the
human figure: no anatomic features catch
the eye of the viewer. Due to the lack of an
external illumination there are no shadows
which would outline curves hidden underneath clothing, as it is typical for actively
­recorded images as published recently in the
press [6]. What remains are ‘characteristic
shadows’ on the figure, corresponding to
objects which block the terahertz light coming from the human whereas they do emit
or reflect comparatively less power.
It now depends on the actual material of
the object which level of power, expressed in
shade of grey, can be expected in contrast to
the almost perfect black body radiation of
the watched person. An object of interest
could possibly reflect, absorb or let pass this
radiation, whereas it also can emit radiation
itself. A metallic object is an almost perfect
reflector, meaning that it will block the human body signal and at the same time reflect
the ambient environment at about 20 °C. So
it will appear at the same shade of grey as the
background behind the person.
An object with a reflectivity smaller than
one could emit some amount of terahertz
light by itself, so the contrast to the human
Figure 6: Comparison between a passive terahertz image (left) and a typical millimetrewave image [6].
body is lower. At extreme, an object with the
same emission characteristics as the human
body will become invisible if it has reached
the exact temperature of its bearer. The other
extreme would be an object which is completely transparent for terahertz light; obviously it also will be invisible to the camera.
Using the prototype, various objects, ranging from a handgun with a fibre reinforced
plastic frame to a ceramic kitchen knife have
been detected without problems. Nevertheless this is only a proof of concept. In reality,
a camera which creates only still images can
easily miss objects if the angle of view is unfavourable. Moreover, it cannot look through
a person, so at least two images, a frontal
one and one from the back side have to be
recorded. This trouble disappears with a video camera. Imaging persons in motion is
highly beneficial; the throughput increases,
detection probability becomes larger and
moreover, it is easy to create a scenario
where persons under test are walking on a
u-turn way and in doing so naturally become
visible subsequently from front and back.
So, this shows the ways to be followed
for the next generation, currently under
34 Optik & Photonik December 2008 No. 4
construction in lab: implement a manageable array (some 10 pixels), combine it with
an optic for 10–20 meters variable focus
and increase the frame rate to at least
10 Hz, later on to 25 Hz. Such a tool can be
expected to operate at airports in the near
future. And hopefully travellers will acknowledge the coexistent gain in privacy and
security, which is currently not achievable
by any other existing technology.
References
[1]„Die nackte Kanone“, Frankfurter Allgemeine
Zeitung, 24.10.2008.
[2]www.izew.uni-tuebingen.de/kultur/theben.
html
[3]www.bmbf.de/de/12917.php
[4]E. Kreysa et al., “Bolometer array development
at the Max-Planck-Institut für Radioastronomie”, Infrared Phys., 40 191-197, (1999)
[5]K. D. Irwin, “An application of electrothermal
feedback for high resolution cryogenic particle
detection”, Applied Physics Letters Vol. 66,
No. 15, 1998 (1995).
[6]“Politiker entsetzt über geplante Nacktscanner”, Spiegel Online, 23.10.2008
© 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim