Cooled and uncooled infrared detectors based on yttrium barium copper oxide
Roman Sobolewski*
Department of Electrical and Computer Engineering and Laboratory for Laser Energetics
University of Rochester, Rochester, NY 14627-0231
Donald P. Butler and Zeynep Çelik-Butler
Electrical Engineering Department, Southern Methodist University
Dallas, TX 75275-0338
ABSTRACT
We review performance and physical characteristics of yttrium barium copper oxide (YBCO) compound as an infrared (IR)
photodetector. YBCO has been used as the IR detector material in both superconducting (oxygen-rich) and semiconducting
(oxygen-depleted) phases. YBCO in its crystalline, YBa2Cu3O6+x phase with x > 0.95 is a high-temperature superconducting
material with the superconducting transition Tc ≈ 90 K. The superconducting YBCO IR detectors operate as either nonequilibrium
(quantum) or bolometric (thermal) devices. The nonequilibrium devices are characterized by very short, single-picosecond
photoresponse times and are expected to find applications in optoelectronics and imaging, as well as ultrafast optical-to-electrical
transducers for digital input applications. The bolometric mechanism results in relatively slow but very sensitive detectors with
possible applications in astronomy. In addition to superconducting IR sensors, interest in uncooled YBCO devices is growing
very rapidly. Despite somewhat lower sensitivity and significantly reduced speed of response, as compared to the superconducting counterpartners, the uncooled IR detectors are characterized by much lower operating cost and weight due to lack of cooling
cryogens and are compatible with existing silicon-based processing and fabrication. The last point is of paramount importance
if the IR-sensitive pixels are to be integrated with CMOS read-out circuitry for monolithic focal plane arrays and infrared
cameras. Amorphous uncooled YBCO photodetectors operate as either photoconductive bolometers or unbiased pyroelectric
devices.
Keywords: yttrium-barium-copper-oxide compound, infrared photodetectors, high-temperature superconductors, bolometers,
pyroelectric effect.
1. INTRODUCTION
The discovery of high-temperature superconducting (HTS) materials has opened new opportunities for applications of
superconductors in electronics and optoelectronics. The HTS perovskites represent a new class of solid-state materials,
exhibiting many, very interesting electronic, optical, and electro-optic properties. It is well known that electrical and optical
properties of HTS crystals are very sensitive to small variations of the crystallographic structure and atomic composition in the
HTS elementary cell.1 The most important is oxygen doping, which plays a similar role to impurity doping in semiconductors
and controls the material’s characteristics.
Yttrium barium copper oxide (YBCO) compound is the most-studied and best-characterized material among CuO-based HTS
materials. It exists in several stable phases, with the most widely known being the so-called 123 phase, i.e., YBa2Cu3O6+x.
Figure 1 shows the YBCO unit cell. Like other copper oxide superconductors, YBCO is based on a perovskite (calcium titanate)
structure tripled along the c axis. The result is an orthorhombic unit cell with dimensions of a = 3.00 Å, b = 3.84 Å, and c
= 11.63 Å with cell volume equal to about 173 Å.3 Two CuO2 layers (conduction II and III layers), formed by Cu2, O3, and O4,
surround the solitary yttrium ion. The layers with oxygen ions surrounding the barium ions are called the intercalated layers.
According to charge transfer models, II and III layers are responsible for carrier conduction while the intercalated layers serve
as charge reservoirs.2 The number of carriers is determined by the amount of charge transferred from the intercalated layers to
the conduction layers. The oxygen atoms in the CuO chains, formed by Cu1 and O1, have an important role to play since they
provide carriers (holes) to the conducting CuO2 layers. Changing the oxygen doping level, one can completely change material
electrical properties ranging from a superconductor (x ≈ 1) to an antiferromagnetic insulator (x ≈ 0). Decreasing the oxygen
content, the YBCO crystal structure simultaneously undergoes a transition from orthorhombic (a < b) to tetragonal (a = b). In
*Electronic address: sobolewski@ece.rochester.edu. Also at the Institute of Physics, Polish Academy of Sciences, PL-02668 Warszawa, Poland
more detail, for x > 0.4, the YBCO compound exhibits a metallic behavior, with the increasing superconducting transition
temperature Tc, reaching 91 K for x ≈ 0.97. For 0.2 < x < 0.4, the material shows the electronic characteristics of a Fermi glass
with a negative temperature coefficient of resistance, leading to semiconductor-like characteristics. As x is decreased below 0.2,
the compound behaves as an antiferromagnet and is insulating at low temperatures. Figure 2 presents the resistance versus
temperature characteristics for several YBCO samples with various oxygen doping contents. As we can see, electrical properties
of the samples range from metal-like with Tc ≈ 90 K to semiconductor-like with thermally activated transport.
Cu1
I
O1
1.0
× 105
0.8
Y
O2
O3
Cu
R(T)/R(300 K)
Ba
II
O4
Cu2
O
III
c
0.6
5.0
× 104
0.4
0.2
b
a
0.0
IV
Z2451
0
40
80 120 160
Temperature (K)
0.0
200
Z2460
Fig. 1. The YBCO unit cell.
Fig. 2. Resistance versus temperature characteristics for
YBCO thin films containing various oxygen contents.
Our paper is arranged into five sections. The second section gives a brief review of the optical detection basics. Next, in
Sec. 3, superconducting YBCO is discussed with its implementation in both the quantum and bolometric IR detection schemes.
Section 4 is devoted to semiconducting YBCO detectors, operating either as micromachined microbolometers or pyroelectric
devices. Finally, conclusions are drawn in Sec. 5.
2. OPTICAL DETECTION BASICS
A radiation detector is a device that produces an output signal that depends on the amount of radiation hitting the active region
of the detector. In general, infrared (IR) sensors can be classified as photon (quantum) or thermal type of detectors. Photon
detectors operate using the photovoltaic effect or photoconductivity. In either case, a photon detector relies on the absorption
of a quantum of light by an electron. In semiconducting quantum devices, the electron is either excited to a state where it can
be transported over a barrier, or, more commonly, photon absorption results in the generation of the electron-hole pair, transiently
increasing conductivity (photoconductivity) or introducing a shift in the quasi-Fermi levels (photovoltaic effect). To minimize
the dark current and simultaneously maximize detectivity, IR photon detectors are generally cooled to cryogenic temperatures.
The necessity of cryocooling significantly increases the system cost and the complexity of the devices, making them applicable
only in the highest-performance military or civilian (mainly astronomic) IR detectors. Most IR semiconductor photon cameras
are based on HgCdTe devices, Si Schottky barrier diodes, or InGaAs avalanche photodiodes.
In superconductor nonthermal devices, the photoresponse mechanism is different from that characteristic for semiconductors
and is based on the so-called nonequilibrium hot-electron effect. After the material is excited by light quanta, the photon energy
is used to break Cooper pairs and is absorbed in the electron subsystem only, which leads to electron thermalization and the
establishment of electron temperature Te, somewhat higher than the equilibrium operating temperature. This thermalization
process causes an avalanche multiplication of quasiparticles (broken Cooper pairs) with the energy values kBTe, close but above
either the superconductor gap energy or the lattice (phonon) energy, depending on the device operating conditions. The
possibility of generating hundreds of quasiparticles by a single photon leads to a very high quantum efficiency and responsivity
of hot-electron photodetectors (HEP’s). In addition, the hot-electron phenomena in thin superconducting films reveal no
frequency restrictions in the IR spectrum and, as we show later, are characterized by picosecond response times. Providing that
extrinsic noise floor is sufficiently low, the HEP can be sensitive to single quanta of the incoming radiation.
All thermal detectors convert incident radiation into heat, thereby raising the temperature of the detector element. This change
in temperature leads to a change in the measurable property of the material, such as, e.g., the electrical resistance (bolometers)
or spontaneous polarization (pyroelectric detectors). Bolometers are discrete semiconductor or superconductor elements used
to measure temperature. Their performance is characterized by the temperature coefficient of resistance β or TCR, defined as
the relative magnitude of the change in the electrical resistance R and given by
TCR = β =
1 dR
.
R dT
(1)
The voltage response of the IR bolometric detector directly depends on TCR, which is regarded as the main figure of merit. From
Fig. 2, one can immediately realize that for other parameters being constant, superconducting bolometers operating at the
superconducting transition temperature Tc are the most sensitive devices, because of their strongly nonlinear R versus T
dependence near Tc and resulting extremely high TCR. They can, however, be easily saturated because of the very limited
temperature range of the superconducting transition. On the other hand, the detectors based on the oxygen-poor, semiconducting
YBCO exhibit substantial TCR in a wide range of operating temperatures, starting with the room temperature. Pyroelectric
YBCO devices are based on the thermal pyroelectric effect that refers to the temperature dependence of the spontaneous electric
polarization. Pyroelectric detectors do not need to be externally voltage or current biased since the change in the polarization
varies the surface charge and results in the spontaneous flow of a pyroelectric current. Thermal detectors are capable of
responding to a very wide range of wavelengths without appreciable variation in responsivity, and they display sufficient
sensitivity at room temperature to permit imaging applications. The most-advanced devices use Si substrates and micromachining
techniques that allow producing the sensing element as a thin-bridge structure with extremely low thermal capacitance and
thermally isolated from the outside world.
3. SUPERCONDUCTING YBCO PHOTODETECTORS
The discovery of HTS materials has prompted a series of photoresponse studies, with optical excitation signals from both the
visible and IR ranges. Direct measurements of the transient photoimpedance response in superconducting YBCO, using a pulsed
laser source and a high-speed oscilloscope were performed, and resulted in a very rich collection of experimental results.3–14
The measurements have demonstrated the existence of an ultrafast optical response, but some early results, e.g., obtained on
granular samples, must be taken now with strong reservations since the data were dependent on the sample fabrication details
and did not reflect the intrinsic properties of the HTS material. An additional common disadvantage of direct, voltage-recording
methods was the limited bandwidth of all experimental arrangements referenced above. A complicated circuit analysis had to
be used in order to analyze the collected experimental data, but it always resulted in relatively large errors and difficulty in
properly distinguishing between the nonequilibrium and thermal responses.
In the above context, the femtosecond optoelectronic technique, described in this section, is the most advantageous for HTS
photoresponse experiments since it ensures above-1-THz bandwidth (below-1-ps time resolution) and below-1-mV signal
sensitivity. Thus, it enables subpicosecond time-resolved studies and results in determination of the intrinsic thermalization and
relaxation time constants of the photoresponse process.
3.1 Superconducting Hot-Electron Photodetectors
The time-resolved optoelectronic studies of quasiparticle dynamics in YBCO thin films were performed using as an excitation
100-fs-wide laser pulses generated by a Ti:sapphire laser.15–17 A typical experimental test structure is shown in Fig. 3. Our sample
consisted of a 5-µm-wide, 7-µm-long microbridge placed in the middle of the coplanar waveguide (CPW) center line and
patterned in a high-quality, 100-nm-thick epitaxial YBCO film, grown either on a LaAlO3 or MgO substrate using pulsed laser
deposition. After processing, the microbridge typically exhibited Tc = 90.5 K with the superconducting transition width <1 K,
and a critical current density jc > 2 × 106 A/cm2 at 77 K. The detection system was based on a cryogenic electro-optic (EO)
sampling system, which can be regarded as an ultrafast (<200-fs temporal resolution) and ultrasensitive (<150-µV voltage
sensitivity) sampling oscilloscope. The entire experimental setup has been presented in detail in Ref. 16. Briefly, a commercial
Ti:sapphire laser, pumped by an Ar-ion laser, provided ~100-fs-wide optical pulses with 780-nm wavelength and 76-MHz
repetition rate. The train of pulses from the laser was split into two paths by a beamsplitter. The first (excitation) beam was
frequency doubled in a nonlinear crystal, intensity modulated by an acousto-optic modulator, and focused by a microscope
objective to a 10-µm-diam spot on the YBCO microbridge. The second (sampling) beam acted as a subpicosecond sampling gate
since it sensed the birefringence induced in the LiTaO3 crystal by the electric photoresponse transient generated by the bridge
and propagated in CPW underneath LiTaO3. As it is schematically shown in Fig. 3, the sampling beam was focused at the gap
between the CPW center line and ground plane, only ~20 µm away from the microbridge. The beam was reflected by a highreflectivity coating at the bottom of LiTaO3 and directed to a compensator/analyzer detection section (not shown), where the
polarization change due to the transient-induced birefringence was converted into an intensity change and measured
differentially using a lock-in amplifier. The computer recorded the time-domain evolution of the electric field at the sampling
point by controlling an optical delay between the excitation and sampling beams and by measuring the output of the lock-in
amplifier. The time-resolved mapping of the microbridge transient photoresponse was stored, averaged over many collected
waveforms, and displayed on a monitor. By introducing a known voltage on the CPW and measuring the resulting samplingbeam intensity change, the system could be calibrated so that the computer could directly display the measured signal magnitude
in millivolts.
Excitation beam
Sampling beam
YBCO coplanar
waveguide
LiTaO3 crystal
Fig. 3. Schematic representation of the YBCO test structure mounted for EO measurements.
Z2200
The current-voltage characteristics of our bridges were typical for a long constriction,18 and one could clearly distinguish the
two different voltage states—the superconducting/flux-flow state with zero/low voltage across the bridge and the switched
(resistive) state where the current was almost constant while the voltage across the bridge increased rapidly. As we have recently
demonstrated,19,20 in the superconducting state the observed photoresponse was a fast, oscillatory transient, characteristic for
the nonequilibrium kinetic-inductive effect.10 The secondary oscillations are the signature of the quantum coherent interaction
between the quasiparticle and the phonon exchange fields.21 The positive parts of the waveform represent the Cooper-pair
breaking process, while the negative parts correspond to the pair recombination. The main transient lasted less than 2 ps with
the recombination part essentially as fast as the pair breaking; thus, there is no phonon contribution effect present in the
photoresponse of YBCO. This observation is contrary to the standard response of a low-Tc superconductor, but is in direct
agreement with the theoretical prediction22 that in YBCO acoustical phonons do not participate in the secondary Cooper-pair
breaking, the electron–electron thermalization process is dominant, and the quasiparticle lifetime is the real recombination time.
The above observation is also in good agreement with the fact that quasiparticle relaxation in YBCO does not depend on the film
thickness15 and is only very weakly dependent on the sample temperature.20
When the bridge was biased in the switched state, we observed a single-spike response with a low-noise baseline.19,20 The
switched-state photoresponse was due to a rise in the electron temperature in the film and was very successfully modeled by the
two-temperature (2-T) model.2–4 Fitting the leading edge of the photoresponse signal to the 2-T model, we obtained the
quasiparticle thermalization time τet = 0.56 ps, while the falling edge of the transient enabled us to directly extract the electron–
phonon scattering time τe–ph = 1.1 ps. The hot-electron response was superimposed on a bolometric signal, which on the
picosecond scale corresponded to a ~250-µV constant-level plateau after the pulse and was equivalent of the 0.3-K increase of
the phonon (lattice) temperature Tph. The comparison between the hot-electron and bolometric portions of the voltage response
and the fitted Te and Tph, allowed to determine the ratio between the phonon and electron specific heats Cph/Ce to be 38.20 The
bolometric signal could be directly observed on the fast oscilloscope.
It is interesting to note that by varying the intensity of optical excitation one can obtain the YBCO photoresponse, which is
characteristic for both the kinetic-inductive and hot-electron effects. Figure 4 shows a series of photoresponse waveforms
obtained at different levels of incident optical power. The graph shows a clear evolution of the purely kinetic-inductance response
at the optical input power 0.5 mW, to the onset (0.75 mW) and the development (1.5 mW) of a second peak and the slow falling
edge of the pulse at the increasingly higher optical intensities. Ultimately, the waveform obtained at the 2-mW illumination
power shows a broad pulse with an almost flat top. This transformation of the single-picosecond bipolar waveform to a >10-pswide pulse at high optical powers was observed earlier by some of us,15 but no satisfactory explanation was provided. Here, we
demonstrate that it can be explained in terms of a superposition of the kinetic-inductive and the resistive hot-electron
photoresponse mechanisms. At low optical intensities, light causes only a small nonequilibrium concentration of excess
quasiparticles, leading to the purely kinetic-inductive bridge response. At high optical intensities, however, the incident photons
do not only break Cooper pairs, but also are directly absorbed by quasiparticles, significantly rising Te, and eventually pushing
the superconducting bridge into the resistive state. Thus, the hot-electron resistive mechanism adds to the kinetic-inductive
photoresponse at higher optical input powers. Figure 5 presents the comparison of the waveform with 1-mW optical power from
Fig. 4 to the combined response of the 0.5-mW kinetic-inductive signal and the simulated hot-electron resistive response. The
excellent fit between the simulated combined response and the experimental waveform verifies the simultaneous occurrence of
the kinetic-inductive and hot-electron photoresponse mechanisms. Similar comparisons (not shown) were also obtained for the
other waveforms presented in Fig. 4.
2.0 mW
1.5 mW
1.0 mW
0.75 mW
Photoresponse signal (mV)
Normalized voltage response
20
1.0-mW optical power
15
10
5
0
0.5 mW
–5
Z2518
Z2521
Experiment
Resistive response
Kinetic-inductive response
Combined response
Time (5 ps/div)
Time (10 ps/div)
Fig. 4. Sequence of time-resolved YBCO photoresponse
waveforms for different incident optical powers.
Fig. 5. The kinetic-inductive response with low optical input power is added
to the simulation of the hot-electron resistive response to obtain the combined
response, which is being compared with the YBCO experimental photoresponse
(Fig. 4) measured at 1-mW optical input power.
The experiments presented above demonstrated the intrinsic photoresponse dynamics in YBCO material and allowed
extracting the intrinsic characteristic response times. For experimental convenience the measurements were performed with the
visible light, but since the hot-electron effect does not depend on the frequency of the excitation signal, the same conclusions
regarding both the physics and applications can be drawn for the IR response of YBCO. Besides providing new insight on the
physics of energy relaxation in HTS under nonequilibrium conditions, our studies also demonstrate the potential of YBCO for
ultrafast detection of optical transients. Optoelectronic YBCO devices should be ideally suited for photodetection applications
because of their very high absorption coefficient in the entire wavelength range from the ultraviolet to >10 µm and ultrafast
response. Simple circuits consisting of a YBCO microbridge and Josephson junction are ideal as optical-to-electrical
transducers. As demonstrated very recently, such structures can transform to the electrical domain the input information coded
in the form of a train of ultrafast optical pulses and feed into the digital circuitry as a train of single-flux-quantum pulses.23
3.2 Superconducting Bolometers
As we have mentioned in Sec. 2, a bolometer measures the incident power deposited by radiation, by sensing the temperature
rise DT in the detector element. This implicitly assumes thermal equilibrium, since we are interested in ∆T between the sensor
and the outside world. If the output signal of a bolometer is voltage, the output voltage ∆V per input radiant power is
∆V = Ib βR∆T ,
(2)
where Ib is the bias current and β is given by Eq. (1). The bolometric process is a general property of conducting solids, but
superconducting sensors tend to be some of the most-sensitive thermometers among bolometric devices since they exhibit very
large values of TCR in the vicinity of Tc. The major limitation of a transition-edge superconducting bolometer, however, is that
its performance is determined by the width of the superconducting transition ∆Tc—the sharper the transition, the more sensitive
the bolometer element, but, simultaneously, the narrower the sensor’s dynamical range. In addition, one would like to maximize
the magnitude of Ib [see Eq. (2)], but it should be small enough to avoid excessive self-heating of the element and the heat-related
broadening of the transition. The above illustrates the difficult trade-offs that must be made in designing the transition-edge
superconducting bolometer.
Low-Tc, thin-film bolometers have been developed for ultrasensitive infrared detection (mainly for astronomical applications) for many years. With the advent of HTS, there has been renewed interest in the development of sensors that can operate
at 90 K and above.17 The best reported values for detectivity D* for YBCO bolometers are of the order of 1010 cmHz1/2/W.
Finally, YBCO bolometers are characterized by a few-ns response time,17 which is quite remarkable for bolometers. It is,
however, still three orders of magnitude slower that the response of a YBCO superconducting HEP, presented in Sec. 3.1.
4. ROOM-TEMPERATURE YBCO PHOTODETECTORS
Depending on its oxygen content, YBCO changes its electrical properties from a metallic phase for 0.4 < x < 1 to an insulating
(or semiconducting) phase for x < 0.4. Several investigations25,26 into the YBCO semiconducting phase have confirmed the
existence of thermally activated transport above 100 K and variable range hopping at lower temperatures. We have found that
thermally activated transport characteristics are caused by the activation of hole-like carriers from localized states around the
Fermi level to extended states. For x = 0.4, epitaxial samples showed activation energies EA of the order of 34 meV above
160 K. Below that temperature, local activation energy monotonically decreased exhibiting transition to hopping conduction.
When the oxygen content was decreased further to x = 0.3, EA was found to increase to 148 meV for room-temperature
conduction. Again, transition to hopping was observed. This time, however, transition occurred at a higher temperature. The
amorphous semiconducting YBCO samples remained within the thermally activated transport regime in the temperature range
of 253 K to 318 K, with EA = 0.2 eV.26
4.1 Semiconducting YBCO as a Room-Temperature Bolometer
The necessity of thermal isolation of bolometer sensors requires that the YBCO device must be compatible with
micromachining technology in a CMOS environment. This allows for the fabrication of focal plane arrays utilizing CMOS
readout circuitry. Suspending the thermometer above the substrate also allows some of the readout circuitry to be placed under
the sensor. Our YBCO microbolometers were designed in the form of floating bridge structures of linear arrays and initially
fabricated using bulk-Si micromachining techniques. The 1 × 10 arrays were fabricated with 60 × 60- and 40 × 40-µm2 pixels
with a 10-µm separation. Early structures were fabricated by growing a 15,000-Å-thick SiO2 membrane on 4-in. <111> n-type
Si wafers.27–30 Trench cuts were patterned by photolithography, and the exposed Si was etched from underneath the oxide
membrane to produce suspended bridges. A 2000-Å-thick YBCO layer was rf magnetron sputtered at room temperature on a
MgO buffer layer. The semiconducting YBCO deposited in this manner was an oxygen-deficient, amorphous (or very fine
polycrystalline) material with an elemental ratio in terms of Y:Ba:Cu:O varying from 1:0.5:2.0:4.5 to 1:0.5:2.4:4.8.31 After the
YBCO/MgO deposition, the sensor areas were defined by wet etching and Au electrodes were sputtered to complete the
microbolometers. We stress that SiO2 bridges were thick enough to withstand all of the above processing steps.
Recently, our YBCO microbolometers were surface micromachined with MgO32 and, most recently, with polyamide
sacrificial layers.33 The devices utilize a thin-film Si3N4 as the supporting membrane. The current fabrication procedure is much
closer to that employed in commercial microbolometer focal plane arrays. The structures are produced in the same basic
geometry as for the bulk-Si micromachined devices, with 1 × 10 arrays. A schematics and SEM micrograph of the structure that
utilized the MgO sacrificial layer are shown in Fig. 6. The fabrication started with the rf magnetron deposition of the 1-µm-thick
MgO sacrificial layer on <111> n-type Si wafers. Next the Si3N4 membrane was deposited, followed by metallization layers.
A two-metal electrode process was adopted utilizing a thin Nb layer (about 300 Å) for a low heat conductivity lead, and a thicker
Au layer (3000 Å) to make electrical contact to the YBCO and form bond pads. Device patterning was performed using a standard
photolithographic technique. The active YBCO layer was approximately 1000 Å thick and was deposited and patterned using
the same procedure as for bulk-Si micromachined structures.
Both the Si and MgO sacrificial layer microbolometers were tested under identical experimental conditions. The dc
measurements lead to the TCR values, extracted from experimental R–T curves, and found to vary from −2.8 to −4.0%/K in the
282 K to 312 K range with a room-temperature (295 K) value of −3.3%/K for the bulk-Si device and −2.8%/K for the MgO-based
detector. The average TCR measured across the wafer varied less than 10%. The I–V curves were linear at all temperatures tested
up to about 1-µA bias current, above which Joule heating caused some nonlinearity. The ac optical response of the YBCO
microbolometer arrays was measured by using a mechanically modulated thermal blackbody source with average output power
density of 1 mW/cm2 and a usable wavelength range from ~0.6 µm to 12 µm. The optical response was measured both in air
and in an evacuated cryostat. Two different biasing schemes were used: current biasing and voltage biasing. For the case where
the devices were current biased, the voltage response across the device was directly measured, while a voltage bias was used
where the current responsivity RI was measured across a resistor Rs in series with the detector.
The RV and detectivity D* measured as a function of chopper frequency for a 40 × 40-µm2 pixel fabricated using a MgO
sacrificial layer are shown in Figs. 7(a) and 7(b), respectively. The highest RV was approximately 107 V/W for Ib = 1 µA
[Fig. 7(a)]. The cutoff frequency of the responsivity was 29 to 34 Hz, yielding a thermal constant τth = 4.7–5.5 ms. The thermal
conductivity measured directly using resistive heating method revealed Gth ≈ 5 × 10−7 W/K in vacuum at room temperature.28
The thermal capacity calculated from the thermal time constant and conductance was Cth = 2.3–2.9 × 10−9 J/K. This latter value
was much lower than that measured for the bulk-Si micromachined microbolometers with SiO2 membranes. The highest
measured D* was 108 cm Hz1/2/W [Fig. 7(b)]. Its current-bias independence and the decline at lower chopper frequencies are
indications of the 1/f noise. In general, the voltage-bias technique where RI was measured was found be less noisy than the
arrangement where our detector was current biased. This is due to the high resistance of the detector (~8M Ω), which serves as
a good source of voltage fluctuations due to Johnson noise ∆Vnj ∆f = 4 kTR , but does not create a high level of current
fluctuations ∆Inj ∆f = 4 kT (1 R) . Finally, the noise equivalent temperature difference (NETD), calculated for the 8 to
14 µm IR spectrum, was 57 mK for vacuum operation. A summary and comparison of the performance of the Si and MgO
sacrificial-layer YBCO microbolometers is presented in Table 1.
(
)
(
)
109
108
Detectivity (cm Hz1/2/W)
Air gap
Responsivity (V/W)
(a)
107
106
Ib = 0.5 µA
Ib = 1.0 µA
Si
105
YBCO
1
Si3N4
Z2472
Au
100
10
Chopper frequency (Hz)
1000
(b)
108
107
106
Ib = 0.5 µA
Ib = 1.0 µA
105
104
1
100
10
Chopper frequency (Hz)
1000
MgO
Nb
Z2471
Fig. 6. Schematics and SEM micrograph of a YBCO microbolometer
fabricated with MgO sacrificial layer. The pixel size is 40 × 40 µm2.
Fig. 7. (a) Voltage responsivity and (b) detectivity as a function of chopper
frequency at different current biases for a MgO sacrificial layer micromachined
YBCO microbolometer pixel (40 × 40 µm2). The measurements were performed
in vacuum at room temperature.
4.2 Semiconducting YBCO as a Room-Temperature Pyroelectric Detector
The thermoelectric behavior of crystalline metallic (x ~ 1) and semiconducting (oxygen deficient, x ≤ 0.5) YBCO compounds
in the normal state has been studied extensively in recent years.34–44 Very recently, we have investigated the thermoelectric
behavior in amorphous, semiconducting YBCO thin films.28,45–47 The work on amorphous, semiconducting YBCO was
motivated by its application as uncooled thermal detectors for IR detection and novel solar cells. The physics of the pyroelectric
effect in amorphous YBCO was studied using pulsed laser excitations.47,48 The samples were prepared by rf magnetron
sputtering onto Si wafers with a MgO buffer layer, under conditions very similar to amorphous YBCO for bolometric
photodetectors (Sec. 4.1). The YBCO films were naturally strain-poled during the fabrication process. To allow for the
propagation of high-frequency electrical signals, the 50-Ω Ag microwave microstrip line (MSL) was fabricated on top of the
YBCO. The MSL had a 20-µm-wide gap to allow for the direct optical illumination of the YBCO film and was connected to a
400-MHz-bandwidth digital oscilloscope. The laser had a repetition rate of 80 kHz, pulse duration of 150 ps, wavelength of
1064 nm, and beam spot diameter of less than 50 µm at the sample surface. Typical transients of an unbiased YBCO,
corresponding to the change in the voltage across the gap in the MSL, are shown in Fig. 8(a). We note that each response is
composed of a two-step rising edge of the pulse, followed by a much slower, millisecond decay. The initial fast-risetime
Table 1: Summary of performance parameters for YBCO micromachined bolometers.
Micromachined under SiO2
bridge using isotropic etch,
single metallization layer
Micromachined under Si3N4
bridge using MgO scrifical layer,
double metallization layer
Pixel area (µm × µm)
40 × 40
40 × 40
(K−1)
−3.5%
Description
TCR, β
10−6
−2.8%
10−7
Thermal conductance, G (W/K)
8×
Thermal capacitance, C (J/K)
4.2–4.6 × 10−9
2.3–2.9 × 10−9
Thermal time constant, τth (ms)
0.53–0.58 in air
4.7–5.5 in vacuum
0.26–0.27 in air
9.45 × 107 at 1.5 µA
1.08 × 108 at 1 µA
(4.4 × 1013)/4 kT for Ib – 1.5 µA
Not applicable
76
57
Maximum detectivity, D* (cmHz1/2/W)
Power normalized 1/f noise corner frequency (J−1)
Estimated NETD (mK)
5×
in vacuum
−6
9 × 10 in air
in air
component, time resolved in Fig. 8(b), equal to ~50 ns, is followed by the slow-risetime component of ~20 µs. The amplitudes
of both components appeared to retain a constant ratio with increasing laser pulse energy. In both cases, they exhibit a rapid
increase for laser pulse energies up to E/Emax ≈ 0.2 and subsequent saturation for higher energies. The fast-risetime component
is due to the initial heating of the film by the laser pulse, when a rapid reorientation of the polarization occurs. On the other hand,
the slow-risetime component is, apparently, due to the redistribution of heat in the film. Although the two-step leading edge is
not predicted by the pyroelectric voltage response47
V (t ) =

 −t 
 −t  
exp   − exp
 ,
τ th − τ e 
 τ th  
 τe 
Kτ e
(3)
where K is the material constant and τe and τth are the electrical and thermal time constants, the overall results of the laser
excitation experiments are consistent with pyroelectricity.47 Fitting Eq. (3) to the measured pulse data (E = 0.005 Emax,
Fig. 8) resulted in a very good match with τe = 5 µs and τth = 0.16 ms. We note that the fitted τe possesses a significant thermal
contribution due to the heat redistribution process; thus, the fast-rise-time component is a better measure of the intrinsic electrical
time constant of the YBCO film. The falltime of the voltage response (τth = 0.16 ms) is limited by slow thermal relaxation due
to acoustic mismatch between YBCO and the Si substrate. Finally, the saturation of the voltage response with increasing pulse
energy occurs due to the saturation of the spontaneous polarization with temperature.
50
(a)
Voltage (mV)
(b) Emax = 25 µJ
0.4 Emax
Emax = 25 µJ
0.4 Emax
0.1 Emax
0.025 Emax
0.005 Emax
40
30
20
0.1 Emax
0.025 Emax
0.005 Emax
10
0
–10
0.0
Z2475
0.2
0.6
0.4
Time (ms)
0.8
1.0 0
10
30
20
Time (µs)
40
50
Fig. 8. (a) Typical pyroelectric voltage response of an unbiased YBCO under pulse laser excitation at different laser pulse energies. (b) The signal leading
edge expanded.
Three different types of micromachined pyroelectric detectors have been investigated: suspended parallel plate capacitors and
lateral detectors on SiO2 membranes, as well as lateral self-supporting YBCO detectors.31,45,46 The lateral geometry was initially
intended for bolometer applications, but it was observed to provide a large zero-bias response, interpreted as pyroelectric.
Typically, the YBCO was strain-poled as deposited, and only in some cases, additional poling was attempted by applying an
external electric field.
The optical ac response of our pyroelectric detectors was measured without any device bias, using the broadband optical
source and experimental arrangement the same as discussed in Sec. 4.2, concerning bolometric detectors. RV of the YBCO
detectors was calibrated against a calibrated pyroelectric detector. Figures 9(a) and 9(b) show the room temperature RV and D*,
respectively, for three different samples (#3 best, #1 average, and #2 poor) as a function of the modulation frequency. We note
the large differences in both RV and D* of our samples, with the sample #2 being orders of magnitude below #3. In all cases,
D* increased with increasing modulation frequency. The performance variations are most likely, a result of the different poling
strength in our samples and varying degree of thermal isolation between the device and the Si substrate. For the best sample (#3)
tested, RV reached almost 104 V/W [Fig. 9(a)] and D* exceeded 109 cmHz1/2/W [Fig. 9(b)], which are the excellent values for
any pyroelectric IR detectors.
104
1010
Detectivity (cm Hz1/2/W)
Responsivity (V/W)
(a)
103
102
101
100
Sample 1
Sample 2
Sample 3
101
102
103
104
(b)
109
108
107
105 0
10
101
102
103
104
Frequency (Hz)
Frequency (Hz)
Z2478
Sample 1
Sample 2
Sample 3
106
Fig. 9. Room temperature voltage responsivity (a) and detectivity (b) of three different micromachined YBCO pyroelectric detectors.
4.3 Comparison of Semiconducting YBCO to Other Uncooled IR Materials
Currently, the most commonly used microbolometer material for IR detection is vanadium oxide (VOx), which has several
phases, the best known among them being V2O5 and VO2.49–52 Preparation of VOx in both a bulk monocrystalline form and a
thin-film form is a difficult task, given the narrowness of the thermodynamic stability range of the oxides. Several types of nightvision cameras based on VOx photodetectors are currently being produced for both defense and commercial use, i.e., focal-plane
arrays consisting of 320 × 240 VOx pixels of 50 × 50-µm2 dimensions that exhibit 14-mK NETD with F/1 optics near room
temperature.52 Another possible bolometer material is amorphous silicon, a:Si.53 Although it shows a reasonable TCR of up to
2.8%/K and excellent compatibility with the CMOS technology, it has been shown to exhibit a very large 1/f noise and, therefore,
low sensitivity. A summary of the performance of VOx, a:Si, and semiconducting YBCO is presented in Table 2. The comparison
uses the TCR, which determines the relative responsivity of the material and the power normalized 1/f-noise corner frequency,
which is a measure of the inherent volume-normalized noise of the material.
Table 2: Comparison of figures-of-merit for competing uncooled bolometer materials.
Material
TCR (%/K)
Power normalized 1/f-noise
corner frequency
Semiconducting YBCO
From 2.8 to 4
Typical 3.5
4.4 × 10−13
Vox52
From 1.5 to 2
1.2 × 10−13
Typical 2.8
4.0 × 10−12
Amorphous Si53
A list of commonly used thin-film pyroelectrics and their pyroelectric coefficient p include lithium tantalate (LiTaO3) p
= 20 nC/cm2K, lead zirconium titanate (PbZrTiO3) p = 70 nC/cm2K, lead scandium tantalate [Pb(ScxTa1–x)2O6] p
= 500 nC/cm2K, and lead titanate (PbTiO3) p = 95 nC/cm2K.54 In the above context, as-deposited, amorphous semiconducting
YBCO films compare favorably with p measured to be in the range between 50 nC/cm2K to 20 µC/cm2K, with the latter value
obtained by external poling the YBCO thin film that initially exhibited weak pyroelectric behavior.45,49 Contrary to most other
pyroelectrics, YBCO films are also compatible with CMOS fabrication. Despite, however, the impressive performance of our
best YBCO structures, it must be noted that amorphous semiconducting YBCO films displayed a difficult to predict pyroelectric
behavior. The inconsistent behavior of our YBCO films may be a result of their fine grain structure and is the subject of our current
research.
5. CONCLUSION
We reviewed the principles and performance of superconducting and amorphous semiconducting YBCO structures, intended
as optical detectors for operation at IR wavelengths. Superconducting YBCO can be operated as an ultrafast (single picosecond)
detector, which displays strong optical absorption from far infrared to ultraviolet wavelengths. Depending on the operating
conditions, the superconducting YBCO photoresponse can be understood based on either hot-electron resistive or kinetic
inductive mechanism. Superconductive YBCO microbridges, operated as photodetectors, exhibit intrinsic bit rates greater than
300 Gbit/s, making them both simple and one of the fastest optical devices. Such devices should find applications as opticalto-electrical transducers for input interface of ultrafast digital superconducting electronics. YBCO detectors can also be operated
in a bolometric mode, near Tc, utilizing the highly nonlinear resistive transition. The devices possess very large TCR, but,
simultaneously, a limited dynamical range, restricting their practical applications.
Amorphous semiconducting YBCO has demonstrated potential as an uncooled IR detector operating with zero bias as a
pyroelectric detector and voltage or current biased as a bolometer. Uncooled YBCO photodetectors are expected to find their
applications as low-cost, high-performance thermal IR detectors for both the civilian and military market, including night vision
and thermal imaging in transportation, manufacturing, radiometry, medicine, law enforcement, security, pollution monitoring,
and spectroscopy. The room-temperature fabrication of semiconducting YBCO devices is simple and compatible with CMOS
technology. Full compatibility is also achieved with the use of micromachining techniques and various sacrificial layers to
produce thermal isolation structures. As a bolometer, semiconducting YBCO possesses a relatively high TCR (>3%/K) over a
wide temperature range near room temperature and low intrinsic noise as measured by the power-normalized 1/f-noise corner
frequency. The pyroelectric behavior of semiconducting YBCO is less understood. It displays a wide range of pyroelectric
effects, with pyroelectric coefficients as large as 20 µC/cm2K. Semiconducting YBCO will have a definitive advantage over
other thin-film pyroelectrics, providing that one will be able to reproducibly fabricate samples with consistent pyroelectric
behavior.
6. ACKNOWLEDGMENT
This material is based in part upon work supported by the National Science Foundation under grant ECS-9800062 and the
Army Research Office ARO-38673PH (SMU). The photolithography masks used to pattern the device structures were fabricated
at the Cornell Nanofabrication Facility, a member of the National Nanofabrication Users Network, which is supported by the
National Science Foundation under grant no. ECS-9319009, Cornell University, and its industrial affiliates. DPB and ZCB
acknowledge the contributions and support from past and present members of their research group. RS thanks all the members
of the University of Rochester Femtosecond Optoelectronics Laboratory for their help and contributions. The Rochester research
was supported by the U.S. Office of Naval Research under grant ONR N00014-00-1-0237 and by the NYS Center for Electronic
Imaging Systems.
7. REFERENCES
1.
2.
3.
4.
5.
J. W. Lynn, ed., High Temperature Superconductivity: Graduate Texts in Contemporary Physics, Springer-Verlag, Berlin, 1990; C. P. Poole, Jr., H. A.
Farach, and R. J. Creswick, Superconductivity, Academic Press, San Diego, 1995.
J. D. Jorgensen, Phys. Today 44, 34 (1991).
A. D. Semenov, R. S. Nebosis, Yu. P. Gousev, M. A. Heusinger, and K. F. Renk, Phys. Rev. B 52, 581 (1995).
M. A. Heusinger, A. D. Semenov, Y. P. Gousev, O. Kus, and K. F. Renk, in Applied Superconductivity, edited by D. Dew-Hughes, Institute of Physics
Conference Series No. 148, Vol. 2, pp. 749–752, IOP Publishing Ltd., Bristol, England, 1995.
A. D. Semenov, M. A. Heusinger, J. H. Hoffmann, and K. F. Renk, J. Low Temp. Phys. 105, 305 (1996).
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
A. Frenkel, M. A. Saifi, T. Venkatesan, L. Chinlon, X. D. Wu, and A. Inam, Appl. Phys. Lett. 54, 1594 (1989).
H. S. Kwok, J. P. Zheng, Q. Y. Ying, and R. Rao, Appl. Phys. Lett. 54, 2473 (1989).
F. A. Hegmann and J. S. Preston, Phys. Rev. B 48, 16,023 (1993).
G. L. Carr, M. Quijada, D. B. Tanner, C. J. Hirschmugl, G. P. Williams, S. Etemad, B. Dutta, F. DeRosa, A. Inam, T. Venkatesan, and X. Xi, Appl. Phys.
Lett. 57, 2725 (1990).
N. Bluzer, Phys. Rev. B 44, 10,222 (1991).
A. Ghis, J. C. Villegier, S. Pfister, M. Nail, and Ph. Gibert, Appl. Phys. Lett. 63, 551 (1993).
F. A. Hegmann, R. A. Hughes, and J. S. Preston, Appl. Phys. Lett. 64, 3172 (1994).
W. R. Donaldson, A. M. Kadin, P. H. Ballentine, and R. Sobolewski, Appl. Phys. Lett. 54, 2470 (1989).
M. Johnson, Appl. Phys. Lett. 59, 1371 (1991).
F. A. Hegmann, D. Jacobs-Perkins, C.-C. Wang, S. H. Moffat, R. A. Hughes, J. S. Preston, M. Currie, P. M. Fauchet, T. Y. Hsiang, and R. Sobolewski,
Appl. Phys. Lett. 67, 285 (1995).
M. Lindgren, M. Currie, C. A. Williams, T. Y. Hsiang, P. M. Fauchet, R. Sobolewski, S. H. Moffat, R. A. Hughes, J. S. Preston, and F. A. Hegmann, IEEE
J. Sel. Top. Quantum Electron. 2, 668 (1996).
M. Lindgren, M. Currie, C. Williams, T. Y. Hsiang, P. M. Fauchet, R. Sobolewski, S. H. Moffat, R. A. Hughes, J. S. Preston, and F. A. Hegmann, IEEE
Trans. Appl. Supercond. 7, 3422 (1997).
G. D. Poulin, J. Lachapelle, S. H. Moffat, F. A. Hegmann, and J. S. Preston, Appl. Phys. Lett. 66, 2576 (1995).
R. Sobolewski, Proc. SPIE 3481, 480 (1998).
M. Lindgren, M. Currie, C. Williams, T. Y. Hsiang, P. M. Fauchet, R. Sobolewski, S. H. Moffat, R. A. Hughes, J. S. Preston, and F. A. Hegmann, Appl.
Phys. Lett. 74, 853 (1999).
R. Sobolewski, Y. Xu, C. Williams, and J. Mostowski, “Analysis of the nonequilibrium kinetic-inductive photoresponse of YBCO by use of the quantum
coherence model,” to be published.
A. V. Sergeev and M. Yu. Reizer, Int. J. Mod. Phys. B 10, 635 (1996) and references therein.
R. Adam, M. Currie, C. Williams, O. Harnack, M. Darula, and R. Sobolewski, Appl. Phys. Lett. 76, 469 (2000).
See, e.g., . L. Richards, J. Appl. Phys. 76, 1 (1994) and references therein.
G. Yu, C. H. Lee, A. J. Heeger, N. Herron, E. M. McCarron, L. Cong, G. C. Spalding, C. A. Nordman, and A. M. Goldman, Phys. Rev. B 45, 4964 (1992).
Z. Çelik-Butler, P. C. Shan, D. P. Butler, A. Jahanzeb, C. M. Travers, W. Kula, and R. Sobolewski, Solid-State Electron. 41, 895 (1997).
D. P. Butler, A. Jahanzeb, P.-C. Shan, C. Travers, and Z. Çelik-Butler, Proc. SPIE 2816, 46 (1996).
A. Jahanzeb, C. M. Travers, Z. Çelik-Butler, D. P. Butler, and S. G. Tan, IEEE Trans. Electron Devices ED-44, 1795 (1997).
C. M. Travers, A. Jahanzeb, D. P. Butler, and Z. Çelik-Butler, IEEE/ASME Journal of Microelectromechanical Systems 6, 271 (1997).
J. E. Gray, Z. Çelik-Butler, D. P. Butler, and M. Almasri, Proc. SPIE 3436, 555 (1998).
P. C. Shan, Z. Çelik-Butler, D. P. Butler, A. Jahanzeb, C. Travers, W. Kula, and R. Sobolewski, J. Appl. Phys. 80, 7118 (1996).
J. E. Gray, Z. Çelik-Butler, and D. P. Butler, IEEE/ASME Journal of Microelectromechanical Systems 8, 192 (1999).
M. Almasri, D. P. Butler, and Z. Çelik-Butler, SPIE’s 44th Annual Meeting, The International Symposium on Optical Science, Engineering, and
Instrumentation, Growth and Characterization of Materials for IR Detectors III, 18–23 July 1999, Denver, CO.
C. L. Chang, A. Kleinhammers, W. G. Moulton, and L. R. Testardi, Phys. Rev. B 41, 11564 (1990).
D. Mihailovic and A. J. Heeger, Solid State Commun. 75, 319 (1990).
H. Lengfellner, G. Kremb, A. Schnellbögl, J. Betz, K. F. Renk, and W. Prettl, Appl. Phys. Lett. 60, 501 (1992).
D. Mihailovic and I. Poberaj, Physica C 185-189, 781 (1991).
I. Poberaj and D. Mihailovic, Ferroelectrics 128, 197 (1992).
S. Zeuner, W. Prettl, and H. Lengfellner, Appl. Phys. Lett. 66, 1833 (1995).
H. S. Kwok, J. P. Zheng, and S. Y. Dong, Phys. Rev. B 43, 6270 (1991).
A. Palti, V. Svetchnikov, V. Pan, and A. Snarski, J. Appl. Phys. 77, 2814 (1995).