FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Cold Electron Bolometer (CEB) made by Shadow Evaporation Technique Ihab Sinno1 , Alimujiang Fulati1 1

Nanoscale Science and Technology IMP Student, Chalmers University of Technology

Abstract According to the Big Bang theory, universe has emerged from an enormously dense and hot state of matter and energy some 13.7 (±2%) billion years ago. Since then, universe has been continuously expanding (according to Hubble’s law) and evolving from one state into another. Accordingly, George Gamow has qualitatively predicted (1948) the existence of cosmic microwave background radiation (CMB); a thing that was first verified and observed in 1964 by Arno Penzias and Robert Wilson who were awarded the Noble Prize for such a discovery (1978). The CMB is composed of photons emitted during baryogenesis; a set of hypothetical physical processes that caused an asymmetry between baryons (matter) and anti-baryons (anti-matter) in the very early universe –according to the Big Bang theory-. At such stage, universe was in thermal equilibrium where radiation was constantly being absorbed and reemitted in a process called “Compton scattering”. However, universe’s expansion has caused the average temperature to drop below 3000K, a point at which nuclei and electrons have combined to form atoms while the primordial plasma has turned into a neutral gas (photon decoupling). Interestingly, the CMB has a blackbody spectrum (of about 2.726K), and it is freely streamed to every single point of the universe until today. Thus, being able to measure and study this radiation is extremely important for knowing more about the properties and the origin of the universe. Today, many ground and space devices have already been investigating the CMB. For instance, NASA has launched the Cosmic Background Explorer satellite (COBE) in 1989 where it has determined that the CMB is isotropic to about 10ppm. In 2003, the Wilkinson Microwave Anisotropy spacecraft (WMAP) results were released to show the most accurate values for some cosmological parameters. In this paper, we will investigate a possible accurate sensing element for the CMB; the Cold Electron Bolometer (CEB) with superconductor-insulator-normal metal tunnel junction proposed by L. Kuzmin. The physical theory behind the operation of the CEB will be discussed first. Then, a suggested approach for the fabrication of CEBs will be described thoroughly (shadow evaporation technique). Finally, some experimental results will be compared to the corresponding theoretical models for characterization purposes.

1 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Introduction A bolometer is a device for measuring incident electromagnetic radiation. It was invented in 1878 by the American astronomer Samuel Pierpont Langley. It consists of an "absorber", which is connected to a heat sink (volume of constant temperature) through an insulating link. The result is that any radiation absorbed by the absorber raises its temperature above that of the heat sink—the higher the power absorbed, the higher the temperature will be. A thermometer of some kind, attached to the absorber, is used to measure the temperature, from which the absorbed power can be calculated. In some designs the thermometer is also the absorber; in others the absorber and thermometer are separate; this is known as "composite design". While usual bolometers can be used to measure radiation of any frequency, more sensitive methods of detection can be used for a given narrow bandwidth. However, for sub-mm wavelengths (from around 200 µm to 1 mm wavelength), the bolometer is one of the most sensitive detectors; hence a direct application for bolometers lies in astronomy. Nevertheless, for achieving a premium sensitivity, a bolometer must be cooled down to a fraction of a degree above absolute zero (typically from 50mK to 300mK); a thing that makes the operation technically challenging. The term bolometer is also used in high-energy physics (particle physics) to designate an unconventional particle detector. The same principle described above is used, with sensitivity not only constricted to light but rather to any other form of energy. More conventional particle detectors are often sensitive to the ionization effect of ionizing particles. Bolometers on the other hand are almost directly sensitive to the energy left inside the absorber; for this reason, they can be used not only for ionized particles and photons, but also for non-ionized particles for any sort of radiation and even to search for unknown forms of mass or energy (like dark matter). Thus, compared to more conventional particle detectors, bolometers are extremely efficient in energy resolution and sensitivity, where they can be used to test very high radio-purity (they are also known as thermal detectors). It should be noted that the use of bolometers as particle-detectors is still at the development stage, where the first regular use, even if in a pioneering way, was only in the 1980s because of the difficulty associated with having a system at cryogenic temperature. In the last decade, superconducting detectors have become the most sensitive radiation detectors of the sub-mm, infrared, and optical radiation with an estimated ultimate sensitivity down to 10-20 W/Hz1/2 [1]. A few modest imaging arrays for ground-based sub-mm observations are already operational, and plans for building significant larger arrays are being approved. As it has been selected by Science [2], the breakthrough of the Year has been “Illuminating the Dark Universe”. The portraits of the earliest universe made in microwaves by the Wilkinson Microwave Anisotropy Probe (WMAP) confirm that universe is made up mainly of mysterious dark energy and dark matter.

2 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Fig.1 Understanding the nature of dark energy and dark matter on a more precise level is getting more and more necessary. To understand the nature of them, the future cosmology needs to get a more detailed picture of the cosmic microwave background radiation (CMB). The proposed NASA missions, SPIRE, SPIRIT and SPECS, will determine the highest level of requirements for bolometers for the near future. The detector’s goal is to provide a noise equivalent power of less than 10-20 W/Hz1/2 [3] over the 40–500μm wavelength range in a 100×100 pixels detector-array. For the moment, the most developed superconducting sensor is the TES (transition-edge sensor) with strong electro-thermal feedback [4]. However, the TES has some problems with excess noise, saturation, and the most drastic problem of artificial overheating by dc power for the feedback. In contrast to this overheating, the novel concept of a “Cold-Electron” Bolometer (CEB) with direct electron cooling (Fig.2) has been proposed by Kuzmin et al. [5, 6]. The CEB is the only concept suggesting the effective removal of the incoming background power from the supersensitive region of absorber. This concept has good prospects because it returns the system to a low temperature (noise) state with a high responsivity to the signal. All signal power is used for measurement (almost without losses associated with the electron-phonon leakage). The time constant is determined by the tunneling time, and it is at 2-3 orders of magnitude shorter than the electron-phonon mean interaction time, where it is estimated to be as low as 10 ns [7]. The CEB can be especially effective for operation in the presence of a real background power load. So far, the optimal realization of this sensor proved to be a two SIN-junctions cold-electron bolometer with a capacitive coupling to the antenna [8]. Theoretical estimations and preliminary experiments show that it is possible to realize the necessary sensitivity of better than 10-18 W/Hz1/2 with the antenna coupled to the nanosized-bolometers at a temperature ≤0.3 K. Additional advantages of such a detector are the easy array integration and the possibility of polarization measurements.

3 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Fig.2 Capacitive-coupled Cold-Electron Bolometer (CEB) with SIN tunnel junctions for direct electron cooling and power measurement

The Cold-Electron Bolometer Coupling between thermodynamic sub-systems The basic idea of the absorber is to use the electronic temperature of the absorber material to determine the amount of radiation detected. Thus, electrons have to be coupled thermally from the phonons in order to define an electronic temperature, Te, which is different from the phonon temperature, Tph. The phonon temperature is almost equal to the temperature of the substrate, T0.

Fig.3 Coupling between the thermodynamic subsystems

4 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

In steady state, the amount of energy per unit time injected into the electron subsystem is the same amount of energy flowing per unit time from the electron subsystem to the phonon subsystem. This is written down as: P = Σ ν (Te5 - Tph5)

(2.1)

Where Σ is a material-dependent parameter and ν is the volume of the absorber. In the non-equilibrium case, the flow of energy between the electron subsystem and the phonon subsystem is described by the electron-phonon thermal conductance, Ge-ph, which can be derived from equation 2.1: Ge-ph = dP/dT = 5 Σ ν Te4

(2.2)

From the normal metal phonon subsystem, energy can flow to the phonons in the substrate quiet quickly through a relatively big thermal conductance denoted by Gk, which is the proportional to the third power of the phonon temperature. The subscript ‘K’ refers to P.L. Kapitza who’s discovered the resistance to heat flow across an interface, Rk (called the Kapitza resistance). Therefore, the normal metal’s phonon temperature is almost in equilibrium with the substrate’s phonon temperature (rather than the electrons’ temperature below 1K where the electron-phonon thermal resistance dominates over the Kapitza resistance). This is ideal for the bolometer application, since one can use electrons’ temperature -which is independent of the two phonon temperatures- to measure the incident radiated power [9].

The SIN Tunnel Junction The current through an SIN tunnel junction is proportional to the different in the tunneling rates between tunneling from the normal metal to the superconductor, ГN→S and tunneling from the superconductor to the normal metal ГS→N, integrated over all the energies.

Fig.4 5 / 22

Göteborg - Sverige

CHALMERS UNIVERSITY OF TECHNOLOGY

FKA 190

MAY 2006

It can be written as:

I (V , Te ) = −e ∫ dE [ΓN → S ( E ) − ΓS → N ( E )]

(2.3)

Where tunneling rates are given by the following expressions:

ΓN → S ( E ) =

N S ( E − eV ) n( E , Te )[1 − n( E − eV , Ts )] e2 R

(2.4.a)

ΓS → N ( E ) =

N S ( E − eV ) n( E − eV , Ts )[1 − n( E , Te )] e2 R

(2.4.b)

Where Ns is the normalized density-of-states of the superconductor and Δ is the superconducting energy gap. As it can be inferred from the last set of equations, tunneling rates are proportional to the number of corresponding available states in the superconductor and in the normal metal. The probability of occupation of a certain energy state at a temperature T is given by:

n( E , T ) =

1 ⎞ +1 exp⎛⎜ E ⎟ ⎝ k BT ⎠

(2.5)

Thus, if the applied potential eV across the tunnel barrier is comparable to the energy gap Δ with temperature much less than the gap energy, the current can be approximated by the exponential function:

I≈

πΔk B Te

⎡ Δ − eV ⎤ exp ⎢− ⎥ 2eR ⎣ k B Te ⎦

(2.6)

As the potential is increased further, the current rises relatively quickly and then begins to approach a constant rate of increase with voltage. This constant rate is equal to the resistance of the tunnel junction when the superconductor is at the normal state. In reality, in the sub-gap region, there is some finite current owing to physical irregularities in the tunnel barrier (the oxide) such as non-uniformities in thickness across the area of the junction.

6 / 22

Göteborg - Sverige

CHALMERS UNIVERSITY OF TECHNOLOGY

FKA 190

MAY 2006

CEB and Responsivity, Noise

The heat balance equation is given by [8, 10]:

P (V , Te , Ts ) + Σν (Te − T ph ) = I 2 R + 5

5

V2 I + β Δ + P0 Rs e

(2.7)

The left-hand side describes the cooling terms (cooling power and thermal conductance power), while the right-hand side describes the heating terms (Joule heating, sub-gap leakage current, retunneling and external power load in the sequence written in equation 2.7). This equation then gives us the electron temperature in the normal metal, and depending on which of the six processes is the dominant process, we would either have heating or cooling of the absorber. The two merits that characterize any bolometer would be the responsivity and the noise equivalent power (NEP). The following expression describes responsivity in the current-biased mode:

SV (ω , I ) =

δ Vω = δ Pω

− (∂I

∂T

)(∂I

∂V

) (∂I

) 4 ∂T ∂P − iω cV ν + 5ΣνTe + (∂P ) − ∂T (∂I ) ∂V ∂V

(2.8)

While the NEP is given in the following expression:

2

NEPtotal =

δ Vω 2 2

amp

SV (0, I )

+ 10k B Σν (Te + T ph ) + NEPSIN 6

6

2

(2.9)

The idea is that we measure some noise, which corresponds to the noise of the bolometer. This noise then translated in terms of minimum detectable power by dividing it by the bolometer responsivity on power.

7 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Fabrication In the following pages, the fabrication technique will be presented, noting that sketches are just used for clarification reasons rather than being accurate real structures. Three main layers are to be constructed on the oxidized (for insulation purposes) silicon wafer: the contact-pads, the gold traps and the bolometer-elements:

Thick Gold (contact pads)

Photoresist layers:

1. clean the oxidized silicon wafer inside the ultrasonic cleaner 2. put it on the spinner while adding acetone then isopropanol for cleaning

3. spin a 200nm layer of LOL2000 (at 3000rpm for 1min, with an acceleration/deceleration time of 3000ms) 4. bake on a hot plate (at 180C for 10min)

5. spin a 630nm layer of S1813 (at 4000rmp for 1min, with an acceleration/deceleration time of 3000ms) 6. bake on a hot plate (at 110C for 2min 5sec)

8 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Exposure and development:

1. place the sample and the mask on their suitable pieces 2. align the sample carefully to its correct position (noticing the ripples when contact is achieved with the mask) 3. expose for 30sec in DUV (it was a bit over-exposed) –another sample was exposed for 36sec and it was much over-exposed-

Round edges showing a sign of over-exposure

4. 5. 6. 7.

develop for 2min 35sec (using MF319 developer) rinse with deionized water inspect with the microscope to check the development we found that the sample needs 10sec more of development

8. ash (at 50W with oxygen plasma at a rate of 10 sccm) for 30 sec under a pressure of 250mTorr 9 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Evaporation and lift-off:

1. 2. 3. 4. 5.

6. 7. 8.

ash the sample (at 50W with oxygen plasma) for 15 sec under a pressure of 250mTorr vent the chamber to the atmospheric pressure place the sample in the evaporation chamber place the boats carrying the metals (preferably in the same sequence of evaporation to minimize rotation complexity) in our case, the sample should have these layers evaporated: - 10nm of chromium (1st layer) - 130nm of gold (2nd layer) - 10nm of palladium (3rd layer) set the parameters for each layer on the machine (density, acoustic impedance and terminal thickness) close the chamber and press cycle to start roughing, then, use the turbo-pump to reach 8E6mbar evaporate the chromium layer at 0.1nm/sec (note that chromium will first cause an increase in the pressure (releasing gases), but then it will reach a limit where it’ll start reducing the pressure back (consuming oxygen); hence, monitoring the pressure is required during chromium evaporation)

9. evaporate gold at 4E-5mbar, at a rate of 0.3nm/sec (60A) 10. in our case, the first boat of gold ran-out at 84.7nm; therefore, we had to adjust the thickness required of gold (to achieve the desired 130nm)

11. evaporate palladium at a rate of 0.3nm/sec (45A) and a pressure of 1.4E-5mbar; note that its harder to melt palladium than gold and chromium

10 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

12. vent the chamber, take your sample, then lift-off in Shipley-1165 Remover at 65C with a slight agitation (as to the cross-mark used for e-beam alignment, we obtained a width of 2.6μm with a central diameter of 5.3μm)

Part of a Cross-Mark used for e-beam positioning and alignment

Thin Gold (normal-metal traps) E-beam resist:

1. setup the spinning table for 5000rpm for 1min rotation time, at an acceleration/deceleration time of 3000ms (those parameters are used for both layers of the e-beam resist) 2. clean up the wafer (spinning while pouring acetone then isopropanol) 3. spin the bottom layer copolymer -MMA(8.5) MMA El(14)- to get a thickness of 600nm 4. bake at 170C for 10min on a hot plate

11 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

5. spin the top layer (950 PMMA A4) at the same configuration to obtain a thickness of 170nm 6. bake at 170C for 10min on a hot plate

E-beam exposure and development:

1. place the sample inside the chamber, and make sure that the corresponding LEDs are active 2. calibrate the machine so that 40pA of current (peak value) are obtained at the sensing element’s (faraday cup) position, with a dose of 250μC/cm2 (the smallest element is about 100nm in dimension; therefore, using the divide by 5 convention, this dose was required according to the graph) 3. after making sure that the sample is perfectly aligned and in position (using the SEM), start the exposure; note that it’s taken 13 hours to be done in our case

4. develop the exposed sample with already prepared PMMA developer (Toluene : IPA = 1:3) for 55sec (30sec + 15sec + 10sec)

12 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

5. rinse with isopropanol, then inspect using the microscope (it was noticed that the bottom chips of our (8x8) wafer are lacking some exposure and development, while the middle chips (like 6-4) are well developed) 6. develop the copolymer layer with ECA : Ethanol = 1:5 for 2min

7. rinse with isopropanol, then inspect on the microscope (an undercut of 0.35μm was observed)

Note the shadow that corresponds to the under-cut

Evaporation and lift-off:

1. ash (at 50W with oxygen plasma at a rate of 10 sccm) for 30 sec under a pressure of 250mTorr 2. vent the chamber to the atmospheric pressure and place the sample 3. place the boats carrying the metals (amounts should be enough for): - 10nm of chromium (1st layer) 13 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

- 30nm of gold (2nd layer) - 10nm of palladium (3rd layer) 4. set the parameters for each layer on the machine (density, acoustic impedance and terminal thickness) 5. close the chamber and press cycle to start roughing, then, use the turbo-pump to reach 8E6mbar 6. deposit the chromium layer at a rate of 0.1nm/sec (45A), with pressure maintained around 1.6E-5mbar

7. evaporate gold at 1.9E-5mbar, at a rate of 0.1nm/sec (50A)

8. evaporate palladium at a rate of 0.2nm/sec (50A) and a pressure of 1.7E-5mbar

9. vent the chamber, take your sample, then lift-off in acetone at 55C

14 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Cutting:

1. spin a layer of S1813 to protect your sample elements against cutting damages (use spinning at 4000rpm) 2. bake the wafer for 2min at 110C on a hot plate

3. cut the wafer into 2x2 sets of chips 4. take back your chips, and lift-off the S1813 layer

Main Bolometer Structures E-beam resist:

1. setup the spinning table for 5000rpm for 1min rotation time, at an acceleration/deceleration time of 3000ms (those parameters are used for both layers of the e-beam resist) 2. take a 2x2 set of chips (in our case it was the one containing 3-QPA2-3) and clean it up on the spinning table while pouring acetone then isopropanol 3. spin the bottom layer copolymer -MMA(8.5) MMA El(14)- to get a thickness of 600nm 4. bake at 170C for 10min on a hot plate

5. spin the top layer (950 PMMA A4) at the same configuration to obtain a thickness of 170nm 6. bake at 170C for 10min on a hot plate

15 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

E-beam exposure and development:

1. place the sample inside the chamber, and make sure that the corresponding LEDs are active 2. calibrate the machine so that 20pA of peak current are obtained at the beam sensor, and set the base dose to 370μC/cm2 3. after making sure that the sample is perfectly aligned and in position (using the SEM), start the exposure; note that it’s taken about 20min to be done in our case (per set of 2x2 chips)

4. develop the exposed sample with already prepared PMMA developer (Toluene : IPA = 1:3) for 25sec

5. rinse with isopropanol, inspect using the microscope, then break the sample to get the 4 chips apart 6. develop the copolymer layer with ECA : Ethanol = 1:5, where different times required by each chip are: - 1min 20sec for COOL4 - 2min 10sec for QPA2 - 1min 20sec for COOL5 - 2min 10sec for T1

16 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

7. rinse each chip directly with isopropanol (to stop development), then inspect on the microscope

Evaporation and lift-off:

1. ash the chips (at 50W with oxygen plasma at a rate of 10 sccm) for 30 sec under a pressure of 250mTorr 2. vent the chamber to the atmospheric pressure and place the samples on the extra chamberpiece (to enable rotation) 3. place the boats carrying the metals (amounts should be enough for): - 31.1nm of aluminum(1st layer) - 3nm of chromium (3rd layer) - 30.1nm of silver (4th layer) 4. set the parameters for each layer on the machine (density, acoustic impedance and terminal thickness) 5. close the chamber and press cycle to start roughing, then, use the turbo-pump to reach 1.8E-6mbar 6. as for the aluminum layer, its required to have a thickness of 20nm (since we are using angle evaporation at 50deg, this is equivalent to 31.1nm of normal evaporation); deposition was carried out at a rate of 0.1nm/sec (20A), with pressure maintained around 1.2E-5mbar (2 boats were needed in our case)

7. oxidize the aluminum, where typical parameters used where 5E-2mbar for 1min

17 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

8. evaporate chromium at 2.5E-6mbar, at a rate of 0.1nm/sec (50A) and with a 50deg angle inclination

9. evaporate silver at a rate of 0.1nm/sec (30A) and a pressure of 1.3E-5mbar with an inclination angle of -5.5deg (to obtain a 30nm thick layer)

10. vent the chamber, take your chips, then lift-off in acetone at 55C (after a microscopic inspection, we found some shadows of the aluminum layer on the substrate, a thing that indicates the overdevelopment of the copolymer layer)

18 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

A picture showing the contact-pads layer (yellow arrow), the gold-traps layer (red arrow), and elements layer (black arrow)

An AFM picture showing 4 SIN junctions

Measurements and Conclusion: As for now, measurements of the I-V characteristics of different bolometer-SIN-junctions have been carried out, so that sensitivity estimates can be done by comparing the slope of the

19 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

superconducting region with that of the normal-state region. It should be noted that the system was current biased in all the following measurements with temperature around 295mK.

It can be seen from the figure that the superconducting-gap spans over a region of 300 microvolts. The dotted line corresponds to the measured I-V characteristics; however, after subtracting the wiring resistances as well as the absorber resistance, the solid black curve is obtained (which is the junctions’ resistance). This measurement is a four point measurement, and it shows a very good fit with theory (the red curve) in the normal-state region. It should be noted that a superconducting gap voltage of 280 microvolts was used in the theoretical simulation along with a normal resistance of 1.48 KΩ and a temperature of 290mK. An additional linear resistance can be added in the theoretical model to obtain a perfect fit with the measured values. As to our best experimental result, a ratio of 750 was obtained between the superconducting and the normal-state resistance in a 10-junctions structure (shown in the next figure). 20 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

Considering future measurements to be carried-out, measuring the variation of the voltage with respect to temperature will be done as well as trying to measure the electron temperature versus voltage; moreover, different parameters such that the total NEP and power responsivity will be evaluated. In the end, we’d like to thank Ian Agulo for his direct supervision and relentless effort while trying to explain the whole process and theory, Leonid Kuzmin for his general coordination and valuable comments, and the BOLO group at Chalmers University which is continuously trying to optimize its sensor for a better understanding of the origins of life.

21 / 22

Göteborg - Sverige

FKA 190

CHALMERS UNIVERSITY OF TECHNOLOGY

MAY 2006

References [1] A. Bitterman, Superconductors and Cryoelectronics, Vol. 12, No. 2, 17 (1999) [2] BREAKTHROUGH OF THE YEAR 2003: “Illuminating the Dark Universe”, Science, 302, p.2038 (2003) [3] Leisawitz D et al, Scientific motivation and technology requirements for the SPIRIT and SPECS farinfrared/submillimeter space interferometers Proc. SPIE 4013 36 (2000) [4] A. Lee, P. Richards, S. Nam, B, Cabrera, K. Irwin, Applied Physics Letters, 69, (1996) [5] L. Kuzmin, I. Devyatov and D. Golubev, “Cold-electron bolometer with electronic microrefrigeration”, Proc. SPIE, v. 3465, pp. 193-199 (1998) [6] L. Kuzmin, “On the Concept of a Hot-Electron Microbolometer with Capacitive Coupling to the Antenna” Physica B: Condensed Matter, 284-288, (2000) 2129 [7] D.V. Anghel and L. Kuzmin, Applied Physics Letter, 82, N2, 293-295 (2003) [8] L. Kuzmin and D. Golubev, Physica C 372-376, pp 378-382 (2002) [9] I.J. Agulo, “Cold-electron Bolometer with Superconductor-Insulator-Normal Metal Tunnel Junction” Thesis work report for Licentiate (2006) [10] D. Golubev and L. Kuzmin, J. Appl. Phys. 89 6464–72 (2001)

22 / 22

Göteborg - Sverige