A Measurement Platform for Characterization of Quantum Cascade Lasers

(1)

DEGREE PROJECT IN ENGINEERING PHYSICS, SECOND CYCLE, 30 CREDITS

STOCKHOLM, SWEDEN 2016

A Measurement Platform for Characterization of Quantum Cascade Lasers

JOAKIM STORCK

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF INFORMATION AND COMMUNICATION TECHNOLOGY

(2)

Department of Materials and Nanophysics (MNF) Supervisor: Richard Schatz

Examiner: Sebastian Lourdudoss

(3)

(4)

Abstract

A measurement platform has been built to characterize quantum cascade lasers regrown at KTH by measuring their spectral and electrical properties at room temperature. Since the current source used in this work had a compliance voltage which was too low to get above threshold, an adapter was made to put an external voltage source in series with it, raising the voltage over the laser. The adapter was first simulated using OrCAD PSPICE and then soldered together and put inside a plastic box with connections for the voltage source, current source and laser mount.

A software was made using LabView to automate the electrical characterization. It ramps the current over a specified range, records current, voltage and output power of the laser and saves the data in a file.

The platform was tested using a QCL sample borrowed from III-V Lab, France. The L-I-V curve and the spectrum of the sample was measured at different optical output power levels and temperatures.

From the L-I-V curves the slope efficiency, η, and the threshold cur- rent, Ithwere extracted and in turn used to calculate the characteristic temperatures T0= 40K and T1= 10K. This led to the conclusion that either the sample had degraded or the thermal dissipation was not efficient enough, since typical values for these temperatures lie around 200K.

From the spectral measurement results, a qualitative analysis was made that indicated a slight increase of mode wavelength with rising temperature as well as mode jumps to longer wavelength modes. It could also be seen that the laser showed single-mode behaviour that became more unstable as temperature or current or both were increased.

Unstable multi-mode behaviour was seen at an optical output power of 60mW at 20^oC.

After tests and analysis, suggestions on possible future improvements of the platform were made.

(5)

Referat

En mätplattform har konstruerats för att karaktärisera kvantkaskadlas- rar som återodlats på KTH genom att mäta deras spektrala och elektriska egenskaper vid rumstemperatur. Strömkällan som användes i detta arbete hade en spänningsgräns som var för låg för att kunna driva en kvantkaskadlaser över tröskeln. Därför byggdes en adapter för att sätta en extern spänningskälla i serie med denna och därmed höja spänningen över lasern. Adaptern simulerades först i OrCAD PSPICE och löddes sedan ihop och monterades i en plastlåda med kontakter för ström- och spänningskälla samt laserfästet.

Automationsmjukvara gjordes i LabView för att automatisera den elektriska karaktäriseringen. Den ökar gradvis strömmen över ett de- finierat intervall och mäter ström och spänning över lasern samt dess optiska uteffekt och sparar alla data i en fil. Plattformen testades på ett en laserchip lånat från III-V Lab i Frankrike. Laserns L-I-V-kurva och spektrum mättes vid olika uteffekter och temperaturnivåer.

Från L-I-V-kurvorna kunde verkningsgraden η och strömmens trös- kelvärde Ithbestämmas och användas för att räkna ut de karakteristiska temperaturerna T0 = 40K och T1= 10K. Detta resultat ledde till slut- satsen att antingen var lasern skadad eller så fungerade värmeledningen bort från lasern dåligt. Typiska värden på dessa temperaturer är i stor- leksordningen 200K.

En kvalitativ analys gjordes av de spektrala mätresulteten, vilken indikerade en liten ökning av modernas våglängder och även modhopp till moder med längre våglängd vid ökning av temperaturen. Lasern var enkelmod men blev instabilare ju mer effekten eller temperaturen eller båda två ökade. Instabilt multimodbeteende sågs vid en uteffekt på 60mW vid 20^oC.

Efter mätningar och analyser framlades förslag på möjliga framtida förbättringar av plattformen.

(6)

Acknowledgments

This project was funded by the Linnaeus Center in Advanced Optics and Photonics (ADOPT).

I would like to thank Prof. Fredrik Laurell and the Laser Physics group at Alba Nova who collaborated by providing laboratory space and equipment. A special thanks goes to Riaan Coetzee who aligned the spectrometer setup and helped to perform measurements, it was greatly appreciated.

I would like to thank my supervisor Richard Schatz for granting me the opportunity to write this thesis and for always being encouraging and enthusiastic. I also would like to thank Prof. Sebastian Lourdu- doss, Prof. Srinivasan Anand, Ghiriprasanth Omanakuttan, Bikash Dev Choudhury and everyone else at the Semiconductor Materials group for making me feel very welcome as a part of the group. I especially want to thank Giriprasanth and Bikash for the excellent lunch company and for making me understand cricket. And last, but not least, I would like to thank my wonderful girlfriend Liz and my family for all their encouragement and support.

(7)

Introduction

Semiconductor lasers have always been limited by the choice of material to achieve a direct band gap that has the desired energy, and thereby wavelength, of the emitted light. Using the quantum confinement effect in thin material layers it is possible to create discrete energy levels in the conduction band, so-called sub-bands, tailored to practically any wavelength in the mid to far infrared by varying the thickness of these layers. Utilizing the energy transitions in these sub-bands when making a laser, the need for a direct band gap material is circumvented and even indirect band gap materials, such as silicon, become possible laser materials. Lasers made using this technique are called quantum cascade lasers (QCL).

The main applications for the QCL lie in spectroscopy. When investigating the vibrational energies of a molecule, typically only harmonics of the fundamental mode can be detected using lasers in the visible and near infrared spectrum. However, the fundamental vibrational modes are easier to detect due to their higher oscilla- tor strength compared to the harmonics, meaning a higher absorption coefficient.

Therefore, since many molecules have their fundamental vibrational modes in the mid and far infrared, a laser operating in this region is a very useful spectroscopic tool. This means that, for example, lower concentrations of gases can be detected.

In the table below are some examples of fundamental vibrational modes of some simple molecules taken from NIST(National Institue of Standards and Technology) tables.

Molecule Fundamental vibrational modes (µm) CO

2

4.26, 15.0

N

2

O 4.64, 7.90, 17.5 H

2

O 2.66, 2.73, 6.27

The QCL technology is the core of a European project called MIRIFISENS, Mid InfraRed Innovative lasers for Improved SENSor of hazardous substances. The goal of MIRIFISENS is to develop better mid-infrared light sources and sensors for a range of different applications, for example gas spectroscopy and remote sensing.[1].

As part of this project, KTH is performing the regrowth process on QCLs which

(11)

CHAPTER 1. INTRODUCTION

means growing a cladding layer of Indium Phosphide(InP) around the active region of the laser. At KTH Electrum there is a reactor optimized for InP growth which is used for this purpose, but KTH has had no way of characterizing and testing these regrown lasers without sending them to another facility. The purpose of this thesis has been to build a characterization and measurement set-up for these regrown QCLs to be used at KTH and to test it. This set-up measures the electrical characteristics and the emitted spectral profile of these samples, allowing for further analysis, such as slope efficiency, temperature dependencies etc.

2

(12)

Chapter 2

Quantum Cascade Lasers

The way a QCL works can be simply described as electrons flowing down a staircase potential, emitting photons at every drop. The energy of these emitted photons correspond to the quantized electron energy levels in the quantum wells formed by the thin layers in a QCL. Since all transitions occur in the conduction band, each electron emits a number of photons cascading down the staircase, reusing the electron many times. To compare, in a classic diode laser each electron is used once at the junction when it recombines with a hole. This reduces the threshold current density and increases slope efficiency[8]. The frequency range of QCLs is wide and ranges from roughly 1 − 100THz (3 − 300µm). The wavelength relies primarily on the layer thickness, but also on the choice of material.

2.1 History

The quantum cascade laser (QCL) has a history that goes back to 1970. It begun with the idea of using a semiconductor superlattice to create negative differential resistance proposed by L. Esaki and R. Tsu [7]. This negative differential resistance can be utilized to create Bloch oscillations of electrons, and thereby emitting mi- crowave radiation.

The idea of using the superlattice to amplify electromagnetic waves was proposed the year after by R. F. Kazarinov and R. A. Suris. They suggested a staircase of quantum wells which electrons would cascade down and emit photons every step, which is the principle behind the QCL[12].

It wasn’t until 1994 that the QCL was realised, and it relied upon the fact that molecular beam epitaxy (MBE) had matured enough to make it possible to pre- cisely create the nanometer-thin layers of material needed for the QCL to work.

The first QCL also had to be cooled to cryogenic temperatures to work, while today

there are room temperature devices available.

(13)

CHAPTER 2. QUANTUM CASCADE LASERS

2.2 Principle of operation

The active region of a QCL is made up of alternating layers of emission regions, where the radiative transition takes place, and superlattice relaxation/injection lay- ers, where the electrons are quickly relaxed after emitting a photon and injected into the next emission region. The emission region is basically a three level system where the lifetimes of the different levels depend on the interaction between states in neighbouring quantum wells.

2.2.1 Superlattice

The foundation of the QCL builds on a superlattice of thin layers shorter than the mean free path of electrons, typically in the nanometer range, in order to facilitate cascading, scattering between layers and other quantum effects. A superlattice is a one-dimensional periodic potential in mono-crystalline semiconductors made for example by varying thin n- and p-doped layers or different alloys during epitaxial growth.

Figure 2.1. A superlattice with the periodicity d, well width L, barrier thickness h and the conduction band offset Vb[8]. z is the growth direction.

In an actual QCL, the relaxation region is a superlattice of gradually doped lay- ers, causing the band diagram to resemble sawtooth shapes, see Figure 2.2. This is to prevent space-charge effects and will at the threshold voltage become completely flat, allowing electrons free passage down the staircase.

4

(14)

2.2. PRINCIPLE OF OPERATION

Figure 2.2. A simplified picture of how an electron travels through a QCL superlat- tice. The gradual n-doping determines the threshold bias. q is the elementary charge, V is the threshold voltage, N is the number of steps, hν is the photon energy and 3kT is the energy of the optical phonon released during the E2 to E1 transition, see Figure 2.3.[8]

Figure 2.3. A superlattice with an applied electrical field creating a staircase po- tential. The ground state of the nth well is aligned slightly above the first excited state of the (n + 1)th well. Here shown with photon emissions predicted by Kazarinov and Suris[12]. E1−3represent energy levels, d is the period length of the superlattice,

 the energy difference between E3(n + 1) and E1(n). hω is the photon energy and eF d represent the energy an electron acquires from the external electric field over the distance d (e is the elementary charge and F the electric field strength).

(15)

CHAPTER 2. QUANTUM CASCADE LASERS

2.2.2 Active region

Figure 2.4. a) shows the cascading action of the laser, with one photon emitted per period. b) shows a schematic of the 3-level laser system in the active region. τ32> τ2

required for population inversion. [8]

When designing a laser, one of the first things to consider is how it is pumped, i.e how population inversion is achieved. In a QCL, population inversion is created in the active region by electrons that are injected via tunneling into an excited energy state with a relatively long lifetime. The lifetime of the excited state needs to be longer than in the lower state. Therefore fast relaxation from the lower level is crucial. The active region is based on a coupled well(CQW) system, with the laser transition being diagonal in real space from the upper state in the left well to the lower state in the right well. From this lower state, there is a very fast transition to the relaxation region, where the electron is tunneled through and then injected into the next active region.

2.2.3 Cladding

Since there is significant heat development in QCLs, the cladding layer needs to work as a good heat conductor, carrying heat away from the active region. At the same time it needs to be a conductor under and on top of the active region and an electrical isolator at sides, see Figure 2.5. For these reasons, InP is a good cladding material. The type of QCLs used in this work are so-called buried heterostructure QCLs, which means that the cladding is grown around the active region, making it completely encased in InP except for the end facets.

6

(16)

2.2. PRINCIPLE OF OPERATION

On the sides the cladding is made of InP:Fe and sandwiching the active region are two layers of InP:Si. InP is a good heat conductor and can be doped to be either an electrical isolator or conductor. InP:Si is n-type, making it a good electrical conductor and InP:Fe is semi-insulating, making it an electrical isolator. This design makes for good heat conductivity for more effective cooling of the active region while also preventing leakage currents past it. The growth of InP is a mature technique that can be done fast and with high quality in Electrum’s unique reactor.

Figure 2.5. Cross-section of BH-QCL. The active region is surrounded by InP on all sides for heat dissipation.

(17)

(18)

Chapter 3

Platform setup

The platform was designed for both electrical characterization and spectral charac- terization, both largely using the same equipment. In this section, the two different setups (see Figure 3.4 and Figure 3.6) and the equipment are described. The design and assembly of the QCL driver adapter as well as the automation software are also described here.

Since the electrical characterization was to be carried out at Electrum in Kista, and the spectral characterization at Alba Nova near KTH main campus, the set- up would have to be transportable as well, meaning all equipment needed to be portable.

3.1 Equipment

Current supply

The current supply used in this project has a current limit of 4A and a 8V compliance voltage. It is able to run in both continuous wave (CW) mode and quasi-continuous wave (QCW) mode. In the QCW mode it can run at up to 1kHz frequency with a duty cycle range of 0.1 − 90%, pulse length of 0.1 − 600ms, rise time < 20µs, fall time < 1µs. When using external triggering it can support up to 50kHz repetition frequency.[15]

Voltage supply

The voltage supply is able to handle up to 5.5V at 7A current. It is manually

adjusted by turning a knob and has an accuracy of about 0.01V. Using the adapter

seen in Figures 3.4 and 3.6 it is put in series with the current supply and thereby

reaching a voltage over the QCL of up to 13.5V.

(19)

CHAPTER 3. PLATFORM SETUP

TEC & TEC controller

The TEC in the laser mount is designed to be water cooled, and is capable of handling 10W of heat being generated by the laser at low temperatures.[6] As long as a pump can provide a flow of at least 1.5 liters per minute, the TEC can keep up with this heat development. The TEC controller has a maximum effect of 120W and an temperature control range of −99

^o

C to 250

^o

C.[14]

Water cooler and pump

The water cooler and pump unit is capable of maintaining a flow of at least 2 liters per minute and is cooled with a Peltier cooler. The hot side of this cooler is cooled with a large heat sink and a fan. It has a cooling effect of 200W at 20

^o

C.[20]

Infrared light power sensor

The power sensor used in this work is a thermopile sensor with a beam tracking functionality. The thermopile ring is divided into four so called quads, and by comparing the readings between them the beam center can be determined with an accuracy of 0.15mm and a resolution of 0.02mm. This beam tracking ability

Figure 3.1.

makes alignment of the laser a much easier task. It has broadband absorption, 0.19 − 20µm, and a 100µW −3W power range[10]. A more in-depth description of the thermoelectric effect can be found in Appendix B.

Power meter

The power meter is connected to the sensor and interprets the information received from it. In this work it was used for beam tracking and logging of the power output

10

(20)

3.1. EQUIPMENT

impinging on the sensor, but it can also be used to measure energy in one-shot pulses.[9]

QCL mount

The QCL mount is a custom made mount designed to house chips from III-V Lab.

After the laser has been mounted, the front containing an aspheric molded chalco- genide glass lens is mounted on the front and it is screwed shut. It is possible to vacuum seal or purge with nitrogen, but this was not done in this work. It has connections for the current supply, the TEC controller and for the cooling water.

The mount includes a XYZ translation stage for the lens so it can be aligned for optimal output. The lens is made from molded chalcogenide glass and is AR-coated for a 2−6µm wavelength range, giving it 95% transmission. It is 6.5mm in diameter and has a focal length of 4.0mm. It has a numerical aperture of 0.56.[6]

Figure 3.2. The part of the mount where the QCL chip is placed. The screws are used to clamp it down and two metal pegs connect the laser to the circuit.

Spectrometer

The spectrometer is a Czerny-Turner type grating spectrometer and was used to-

gether with a DSP lock-in amplifier and a computer. The slit size used for the mea-

surements was 0.1mm. The grating used in this spectrometer had a groove density

of 300mm

⁻¹

. This grating combined with the slit size of 0.1mm gave a wavelength

resolution of < 0.36nm. The detector head is a removable and separate part of the

spectrometer, and it contains one silicon detector for shorter wavelengths and a lead

selenide detector for wavelengths between 2 and 6 micrometers. The detectors can

be turned on and off separately.

(21)

Figure 3.3. Schematic of a Czerny-Turner monochromator. The rotating grating is used to both select a single wavelength and scan across a spectrum.[17]

QCL driver adapter

To bypass the low compliance voltage of the current source, a custom adapter was made to connect a voltage source in series with it. This is described in detail in its own section in this chapter.

3.2 Electrical characterization

For the electrical characterization, the laser and sensor were mounted on an optical table with the laser simply pointing straight into the power meter while a custom software measured both current and voltage over the laser and its output power.

The equipment used was everything listed above except for the spectrometer. How the equipment was connected and how it was operated is described in the platform operation chapter.

12

(22)

3.2. ELECTRICAL CHARACTERIZATION

Figure 3.4. The platform for measuring the L-I-V curve of the QCLs with all equipment and connections.

3.2.1 Measurement automation software

In order to perform the measurements and store the data in a time-efficient manner, the process was automatised by making a simple LabView interface that uses a loop that ramps the current over time and measures current and voltage of the current source and the optical power received by the sensor. The LabView interface allows one to toggle between which current levels the ramp goes, how many measurement points to record, and how long to wait between each step for the thermopile sensor to respond to the increased output. The sensor head used in this set-up has a response time of 1.8 seconds [10], which would be the minimum time to wait between measurements. After the program has finished running, it presents an L-I-V curve and saves the data in a text file for further analysis.

The water cooler and pump and the TEC controller are not automated. There- fore, prior to making any measurements these would have to be switched on and handled manually. However, there should be no reason to change these settings during an ongoing measurement loop.

In more detail, the program starts by initializing communication with the con- nected equipment, which is the power meter and the current source. If communi- cation can not be established with either instrument, it stops and returns an error.

When connection has been established with the instruments, the start current is

set in the source and output is turned on. After waiting for 3 seconds, which is the

turn-on delay of the current source, the measuring loop starts by taking a first mea-

surement of the current, voltage and measured power, storing the data in separate

arrays for each data type and then increases the current by one step. The step size

is equal to the interval between starting and ending current divided by the number

of data points wanted. When this is done, the program waits for 1.8 seconds before

(23)

Figure 3.5. The virtual interface as it looks on screen.

starting the next iteration of the measurement loop.

When the number of iterations reaches the number of data points specified, the program moves on, recording the last data point before merging the data arrays into a matrix together with headers and comments and saving everything in a text file at the specified location. If a file already exists with the same name it is overwritten, so it is important to either move the data file or change the file path in the program if the previous data is to be kept. Lastly, the recorded L-I-V graph is displayed in the graph on the VI.

3.3 Spectral characterization

The measurement and characterization of the spectrum was carried out at Alba Nova, using the same setup as for the L-I-V measurements, apart from switching out the power meter and power sensor for a spectrometer and borrowing a He-Ne laser from the Laser Physics department along with lab equipment such as mirrors etc. The power meter and sensor were still useful for aligning the laser but were not used during measurements. The He-Ne laser was instrumental in aligning the setup, its beam overlapping with that of the QCL. The alignment and operation of the setup is more thoroughly described in the platform operation chapter.

14

(24)

3.4. DESIGN AND ASSEMBLY OF THE QCL DRIVER ADAPTER

Figure 3.6. The platform for measuring the spectrum of the QCLs with all equip- ment and connections.

3.4 Design and assembly of the QCL driver adapter

Since the current supply had a compliance voltage of 8V, and the QCLs that will be characterized will have a threshold voltage of over 8V [18], finding a way to circumvent the compliance voltage was necessary. This was done by building a custom circuit making it possible to connect a voltage supply in series with the current supply, thereby increasing the voltage over the laser without going over the compliance voltage over the current supply.

Building the circuit was done by first sketching a possible circuit that could solve this problem by connecting a voltage supply in series with the current supply and the QCL, and then simulating it using OrCAD PSPICE. After simulations showed it was viable, the components were soldered on to an experiment card and put in a box with connections for the current supply, the voltage supply and the laser mount.

The circuit would have to be able to handle DC current and up to MHz-range pulses up to 4A, which is the limit of the current supply [15]. A possible drawback of this solution is that, because of the constant voltage, there will always be a small current flowing through the QCL. The heat produced by such a small electrical power will most likely not be significant enough to impact the laser performance much, though.

3.4.1 Circuit design

First, a rudimentary idea sketch was made to have something as a starting point. It

was then simulated using OrCAD PSPICE to see how it would work and to modify

it before deciding what components to use. As seen in Figure 3.7, the voltage supply

V4 and current supply I3 are modeled with their Thévenin and Norton equivalents,

respectively. The QCL model used here was custom made, its impedance based on

the I-V curve of a photonic crystal QCL regrown at KTH.[18]

(25)

Figure 3.7. OrCAD idea sketch of the circuit.

The simulated current supply is in Figure 3.7 run in pulsed mode, with pulse data taken from the data sheet of the current supply unit [15]. The idea was to have a way avoid possibly damaging the current supply when it is turned off by using shunt-connected diodes to bypass it. In total, four diodes D1, D2, D3 and D4 with low forward voltage drop were used for this purpose to reduce the electric effect on each diode. To protect the laser from transient effects when pulsing the current, two capacitors were shunt connected over the voltage source. One with high capacitance, 6800µF, to handle, for example, switching off DC current and one with low capacitance, 220nF, to handle rapid pulses. The RC constant of the small capacitor is 0.1 · 220 · 10

⁻⁹

= 22ns, which is fast enough to protect against back surges between the 0.1ms long pulses generated in the current supply. The reason for choosing these capacitances was simply that they were the capacitors readily available.

To make the circuit isolated, easy to connect and disconnect, and more robust for transport, it was mounted inside a plastic box, with connection for the current supply, laser mount and voltage supply. The connections for the current supply and the laser mount are DB9 connections, male on the supply side and female on the mount side. To connect the voltage supply there are two 4mm socket plugs mounted on the same side as the current supply connection. The circuit was designed for a 4A maximum current, which meant that the connections for the power carrying lines were divided onto several rails on the card, since each rail is dimensioned for a maximum of 2A.

16

(26)

Figure 3.8. The card on which all components was to be mounted is made of epoxy and has thin copper rails with equidistant holes along their length.

3.4.2 Circuit simulation

Simulations were run in both DC and pulsed mode. For the DC simulations, the current was ramped from 0-4A at an applied voltage from the voltage source ranging from 0V to 7V with a 1V step size. The step size for the current was 0.01A.

Figure 3.9. The final version of the simulated circuit.

The pulsed simulation had pulse data taken from [15]. The pulses went from 0A

to 4A and the applied voltage was 4V.

(27)

Figure 3.10. The voltage across the QCL as a function of the current output of the current source.

Figure 3.11. The voltage across the QCL and the pulsed current output of the current source in the time domain. The green curve is the current and the red curve is the voltage.

3.4.3 Assembling the adapter

After the circuit was modeled and simulated, holes were drilled in the housing box for the connections and the components were soldered onto the experiment card.

Before any soldering or drilling was done, a drawing was made to plan where on the experiment card the different connections would be soldered and to plan the

18

(28)

physical placement of the different components. Then holes were made in the box

Figure 3.12. The circuit map as it would look on an experiment card.

to fit the connections for the instruments and the experiment card was cut to fit inside the box.

Isolated copper wires were taken from a power cord and soldered to the power plugs for the voltage source connection and the laser power pins on the DB9 con- nector. They were then soldered to the experiment card as shown in the drawing, divided onto four parallel rails to evenly distribute the current in order to avoid overloading/overheating.

Soldering the other diodes, capacitors and the smaller wires from the DB9 con- nector was straight-forward. To make sure the diodes and capacitors were properly shunted across all four strips on each side, a small piece of copper wiring was laid across them touching the connection pin and soldered in place.

In order to test that the circuit was working, a halogen lamp was connected by soldering its bare connector wires onto the experiment card divided onto four rails, making the lamp shunted across the Lasermount+ and Lasermount- connections.

The testing was done by simply shorting the voltage source connection with a cable and connecting the current source and then driving current through the circuit. The testing confirmed that the circuit was indeed working, and that the current source shut off when the voltage went above its compliance voltage.

When the soldering was done, the card was fastened with screws using the screw

holes integrated in the box. The lid was then screwed shut and the adapter was

ready for use.

(29)

Figure 3.13. An inside look of the finished adapter. The two white cables are connected to the halogen lamp mount used to do a first function test. On the right is the "input" side where voltage and current sources are connected and on the left is the "output" side where the laser mount is connected.

20

(30)

Chapter 4

Platform operation

In this chapter, the handling and measurement procedure is explained. This part can be used as a manual for anyone who wants to use the characterization platform in their experiments.

4.1 Electrical characterization

4.1.1 Setting up and connecting

The equipment used in the electrical characterization consists of all instruments described in the previous chapter except the spectrometer.

• The first thing to do is to mount the QCL submount in the QCL mount. Make sure the submount is facing the right way so that the anode and cathode are correctly oriented. Then remove it from the mount and apply cooling paste to its underside. Then gently place it on the mount and use the mount screws to clamp it down. Attach the XYZ-stage to the mount and fasten it with its screws.

• Make sure that the water cooling is properly set up and connected. On the pump side, the hoses are easily disconnected and connected without leakage from the pump but one should be careful when disconnecting since there may still be water in the hoses. Before sliding the hoses onto their connections on the QCL mount, slide a clamping ring onto each hose so the connection can be securely fastened and tightened. When this is done, the pump is filled with more water, if necessary, through the hole on top.

• The pump can now be plugged into the outlet and turned on. The water

temperature is controlled via the front panel on the pump. Once it is confirmed

that the pump and cooler are working, it can be turned off since it is quite

noisy. At this point, fasten the QCL mount to the optical table where you are

working.

(31)

CHAPTER 4. PLATFORM OPERATION

• Now, the TEC controller, the current source and the voltage source can be plugged into a wall outlet. The current source is then connected to the QCL driver adapter with a male-female DB9 cable and the voltage source is con- nected to the adapter with two standard lab cables. Make sure to use connec- tion on the voltage source with a fuse dimensioned for at least 4A since this is the maximum current output from the current source. These connections are on the same side of the box, making it the "input"-side of the adapter.

The QCL driver adapter is then connected with a male-male DB9 cable from the "output"-side to the QCL mount. The TEC controller is connected to the QCL mount with a DB25-7W2 cable.

• When this has been done, the power sensor is connected to the power meter which in turn is connected to a computer with a RS232-USB cable. It is possible to connect to a serial port on the computer as well, using a different cable. Then the current source is also connected to the computer with a USB cable.

• In order to be able to use the power meter and current source with LabView, the correct set of drivers need to be installed. They can be downloaded from the Ophir Photonics and Arroyo Instruments websites, respectively. In the measurement automation software in LabView, there are adresses assigned to the power meter and current source. Use VISA viewer or something similar to find their logical adresses and put them into the appropriate box. In the

"usb channel 2" box, out in the USB channel adress of the current source. In the "VISA resource name 2" box on the back panel, first slide, put in the full VISA adress of the power meter. In the "Path to data file" box, use the browse button to specify where to save the output data file. In the "Comments" box, put in information about the measurement deemed relevant, for example TEC temperature, current range, QCL submount number, voltage of the voltage source. Date and time are added automatically into the comments.

• When all of the above has been done, the final step is to align the lens in the XYZ-stage on the front side of the mount. Place the power sensor directly in front of the exit pupil of the mount and turn on the laser. Adjust in the X and Y dimensions to reach a maximum. Then place the power sensor at

"infinity", which can be a meter or so away from the laser mount and use the Z-knob to collimate the beam, reaching a maximum. Then place it closer to the mount again and fasten it to the table, making sure the beam is centered in the sensor.

4.1.2 Performing measurements

When performing measurements, the first thing to do is to turn on the cooler pump and set it to an appropriate temperature above the ambient dew point, for example 16-17

^o

C.

22

(32)

4.2. SPECTRAL CHARACTERIZATION

Figure 4.1. The electrical characterization setup. The laser points straight into the power meter. The XYZ-stage has been attached to the front.

Turn on the TEC controller and set it to the temperature you want using the turning knob and then press the "Output" button. A blue light will indicate that the instrument is active and the screen will show the set and current temperature as well as voltage and current going through the TEC in the QCL mount. Once this is done, the voltage source is set to the voltage you need to add to the 8V compliance of the current source in order to be able to reach the threshold of the laser. For the experiments done in this work, 4V was used. Now turn on the voltage source. At this point, turn on the current source. Set it manually to at least 30mA before turning the output on. If the starting current is too low the instrument will turn off the output and give an error message. This happens because the current source will try to go to negative voltages in order to keep the current down if the set starting current is too low. Since it can not go to negative voltages, it will turn off the output and return an error message. Turn the current up to a point above threshold, make sure the output is on and align the power sensor. It is handy to use its beam tracking function at this stage.

4.2 Spectral characterization

The setting up of the equipment is the same as for the electrical characterization, but without aligning and fastening the power sensor. When the equipment is set up, one also needs a laser in the visible spectrum, for example a He-Ne laser which was used for this thesis. As can be seen in Figure 3.6, the He-Ne laser was pointed away from the spectrometer. The reason for this was that it was also used in another experiment at the time and moving it would disrupt that experiment’s alignment.

The extra mirrors required to redirect the beam also helped when adjusting the

beam’s angle and height.

(33)

CHAPTER 4. PLATFORM OPERATION

4.2.1 Aligning the QCL and spectrometer

What was used for the alignment was 5 silver mirrors and a CaF

2

window, which is transparent at 4.7µm, for combining the beams. Since the light going into the spectrometer has to enter parallel to the table, height adjustment of the beams was necessary. This was made possible by using two mirrors to adjust each laser.

See Figure 3.6, mirrors 2 and 3 adjust the He-Ne laser and mirrors 4 and 5 adjust the QCL/combined beam. Another reason for height adjustment of the beams was the fact that the lasers were situated on one optical table and the spectrometer on another and these tables had slightly different heights.

First the QCL mount was mounted at approximately the same height as the spectrometer entrance slit using metal slabs since no height adjustable mount was available. Then the CaF

2

window was placed on the spectrometer table at a 45 de- gree angle to allow the QCL beam to pass through and to reflect the He-Ne beam.

Mirrors 1 and 2 (Figure 3.6) were used to point the He-Ne beam toward the spec- trometer table and mirror 3 was used to direct that beam onto the CaF

2

window.

After the CaF

2

window, two more mirrors were used to direct the beams onto the concave mirror (mirror 6). Two apertures were used between mirror 5 and mirror 6 in order to align the beams parallel to the table and at the same height as the entrance slit of the spectrometer. The apertures were placed between mirrors 5 and 6 as far from each other as possible while leaving some room for the power meter.

First the QCL beam was aligned with the help of the power sensor placed behind the last aperture in front of the concave mirror. Mirrors 4 and 5 were adjusted to maximize the power to the sensor. After this, the He-Ne beam was aligned through the apertures by adjusting mirrors 2 and 3 to get the beams to overlap.

Before starting to align the beams with the entrance slit, the PbSe detector was turned on. The beams were directed into the entrance slit by adjusting the concave mirror tilt and position until the a signal was picked up by the detector.

After adjusting the concave mirror to reach a signal maximum, the detector head position was adjusted until another maximum as reached. Then the concave mirror and detector position were adjusted alternately until the signal was deemed strong enough, around 2.5mV detector response. Then the alignment was finished and measurements could be performed.

4.2.2 Performing measurements

When the setup had been aligned, measurements could be performed and the shutter connected to the lock-in amplifier was placed in front of the concave mirror. Before beginning it was decided to measure the spectrum at 30mW output power at 16, 18, 20, 22 and 24

^o

C and at 20

^o

C at 5, 10, 20, 30, 40, 50 and 60mW output power.

The time that was allocated to these measurements were not enough to do more.

The reason for this was that the alignment of the setup took much longer than

24

(34)

4.2. SPECTRAL CHARACTERIZATION

Figure 4.2. The aligned setup for spectral characterization. A are silver mirrors, B is the CaF2plate, C is the shutter, D is the concave mirror focusing the light into E, the entrance slit of the spectrometer.

anticipated, leaving only a couple of hours for measurements. Ideally, measurements at more power levels and more temperatures would have been done in order to have access to more data.

The output power was measured before the measurements by placing the power sensor directly in front of the QCL. The current was then adjusted manually until the power reached the desired level. Since the QCL seemed to take at least a few minutes to stabilize, these power drifted while making these measurements and they therefore have an accuracy on the order of 0.1-1mW. The output power drifted a couple of mW, but waiting for the laser to stabilize would have taken too much time.

Between each adjustment of current and temperature, the temperature was al-

lowed to stabilize for a minute or two and then measurements were done. The

measurements were done by using a software, in this case Origin, to sweep a spec-

ified spectrum at a specified resolution. Each measurements took between five to

ten minutes.

(35)

(36)

Chapter 5

Measurement results

In this chapter the results from L-I-V and spectral measurements are presented.

The measurements were done on a QCL sample borrowed from III-V Labs in Paris, France. It was already characterized and those results are presented in the table below. The discussion about the results in this chapter are in relation to that.

Center wavelength 4.7µm Maximum power 240mW Threshold voltage 10.5V Threshold current 709mA Maximum current 1473mA

5.1 L-I-V continuous wave measurements

The L-I-V curve was measured for different temperatures, from 13

^o

C to 27

^o

C. From these curves, the temperature dependency of the threshold current, the slope effi- ciency and the differential resistance were calculated in order to have parameters for comparison with other QCL samples in the future.

The slope efficiency and threshold at each temperature was calculated by per- forming a linear fit, y = ax + b, over the fifteen first data points after the threshold.

The slope efficiency η = a. The threshold current I

th

=

b a

. The slope efficiency and threshold current were calculated by linearizing the widely used empirical ex- ponential temperature dependencies by Taylor expansion around room temperature

T_R

= 293K and comparing them to linear fits made to the data.

η

(T ) = η

0e⁻

T

T1

≈ η

₀e⁻^TR^T1

−

η₀ T1

e⁻^TR^T1

(T − T

R

) (5.1)

I_th

(T ) = I

0e

T

T0

≈ I

₀e^TR^T0

+

I0

T₀e^TR^T0

(T − T

R

) (5.2)

The characteristic slope efficiency temperature, T

1

, and the characteristic thresh-

old temperature, T

0

, were calculated. T

0

and T

1

indicates how temperature sensitive

(37)

CHAPTER 5. MEASUREMENT RESULTS

Figure 5.1. The L-I-V curves for different temperatures. Threshold increases and slope efficiency decreases with rising temperature. The voltage is not affected signif- icantly by a change in temperature.

Figure 5.2. The linear fit to calculate T1.

η

and I

th

are. A higher value means less sensitivity.

Using 5.1 to calculate T

1

, we get T

1

= 10K and using 5.2, we get T

0

= 40K, meaning

28

(38)

5.2. SPECTRAL MEASUREMENTS

Figure 5.3. The linear fit to calculate T0.

that for every 10K the temperature changes, η changes by a factor of e and for every 40K the temperature changes, I

th

changes by a factor of e. This means that both the threshold and the slope efficiency are highly temperature dependent around room temperature. These figures are quite low compared to what has been achieved by similar QCLs, for example T

0

of 168K and a T

1

of 295[16]. The threshold was higher than one would expect from the characterization data received from III-V Lab and the optical power was lower. III-V Lab had at 20

^o

C measured a threshold current of 709mA and a maximum power of 240mW.

These results could depend either on the laser having degraded or on insufficient heat conduction from the active region due to too high thermal resistance in the laser mount or in the thermal paste used, meaning that the active region temper- ature was mush higher than the TEC temperature. The exact cause or causes for this remains to be investigated and is a topic for the future.

5.2 Spectral measurements

When the setup had been aligned, measurements were done and results were an- alyzed to see how the spectrum of the laser depended on temperature and output power.

Due to time constraints it was not possible to do as many measurements as was

planned, but there was still enough data to draw some conclusions about the QCL

sample’s behaviour.

(39)

CHAPTER 5. MEASUREMENT RESULTS

5.2.1 Temperature dependence

For looking at the temperature dependency of the spectrum, measurements were done at a constant output effect of 30mW at different temperatures(16

^o

C, 18

^o

C, 20

^o

C, 22

^o

C and 24

^o

C). When keeping a constant output effect of 30mW and varying

Figure 5.4. The emission spectra at different temperatures.

the temperature, there was stable single mode operation with the exception of the double-mode operation at 16

^o

C. This double-mode operation might also be an effect of mode-hopping during a scan due to the laser not having stabilized completely before measuring. In Figure 5.4 one can see a tendency toward multi-mode operation at 24

^o

C. Fact is, also when the optical effect reaches its peak (see Figure 5.1) the laser becomes unstable and mode competition occurs. When this occurs seemingly depends on both temperature and laser current.

5.2.2 Current dependence

When making a qualitative analysis of the current dependence of the spectrum, it was more intuitive to plot it as a function of output power, which in turn is directly dependent on the current at constant temperatures. In Figure 5.5 we see that the wavelength increases with increased output power (increased current). As the power increased, all the modes shifted slightly toward longer wavelengths, and the lasing mode tended to jump to a longer wavelength mode. In Figure 5.5, the wavelength at 30mW output power is shorter than at 20mW, for example.

At high output power the QCL no longer operated in a single mode. There was mode competition and the spectrum was unstable. The spectrum in Figure 5.6

30

(40)

5.2. SPECTRAL MEASUREMENTS

Figure 5.5. Emission spectra at 20^oC at different output powers.

is shown in dB-scale. The spectrum at this high power, 60mW, would produce a different spectrum each measurement, with the spectrum likely changing during the measurement due to mode competition.

Figure 5.6. Multi-mode behaviour at high laser current. In this case it is 1300mA.

Logarithmic scale.

(41)

(42)

Chapter 6

Conclusion

A portable platform to characterize quantum cascade lasers has been designed, assembled and tested. Using this set-up, the L-I-V characteristics as well as the spectrum of a QCL sample has been measured. Using these results, temperature and current dependent effects could be seen and were analyzed, such as slope effi- ciency η and threshold current I

th

. By having done the this, the aim of this thesis has been fulfilled.

Since the compliance voltage of the current supply was too low at 8V, a custom adapter circuit was made and mounted in a box. This adapter puts an external voltage source in series with the current supply, effectively raising the compliance voltage. If the external voltage is 4V, then the range in which the current supply can operate is 4 − 12V. This box also protects the laser from transient effects, such as back surges of current which can easily harm QCLs.

To automate the L-I-V measurements, a LabView software was made to au- tomatically scan over a specified current range, record the voltage, current and measured optical power, store the data in a text file and present a curve plotted from that data.

The set-up was straight-forward to assemble when measuring the L-I-V curve,

but as it turned out, the aspheric lens in the mount does not compensate for the

inherent astigmatism of the beam enough to completely collimate it. When putting

the sensor on different distances from the lens, two focal planes can be observed,

seen as maxima of the measured effect by the sensor. However, as long as the

distance was within a couple of meters it did not have a detrimental effect for the

purpose of this work, it does not seem to be too divergent. The actual divergence

was not measured in this work, though it is something that might be interesting in

the future for beam profile characterization and, for example, long range tests.

(43)

CHAPTER 6. CONCLUSION

6.1 Future work

While the set-up works well as it is described in this work, there are some things that can be done in order to improve it further. These improvements concern primarily cooling, alignment for spectral measurements and the automated measurements.

Since the set-up is intended for characterizing QCLs quickly, it needs to be a relatively quick and easy procedure to change the sample. For this, using a graphite sheet instead of cooling paste is likely a good alternative. It removes the need for cleaning and extra manual handling of the samples when changing them. This would also eliminate the issues that come with having either too much or too little cooling paste.

Aligning the laser with the spectrometer was very time consuming, and should be handled differently when measuring on more than one sample. One thing that would eliminate much of the alignment procedure is a MIR-transmissive optical fiber, for example a ZnSe-fiber. If one wants to characterize a number of samples at a reasonable pace, eliminating the time consuming alignment step is crucial. Such a fiber would also be useful during the electrical characterization, though the amount of time saved would most likely not be high.

One might in the future want to measure the L-I-V curve at many more temper- ature levels. In that case it would be convenient to also automate the temperature adjustment between measurements. That way all measurements for one sample could be done without any manual handling with just one cycle of the program.

Other types of measurements that will be useful to do are pulsed and modula- tion response measurements. These would explore the high frequency dynamics of QCLs and would be interesting for applications such as spectroscopy and telecom- munications.

34

(44)

Bibliography

[1] Mirifisens presentation kit, 2012. [Online; accessed 14-March- 2016] http://www.mirifisens-project.eu/files-pub/MIRIFISENS_

Presentation_Kit.pdf .

[2] Robert M. Besançon. The Encyclopedia of Physics 3rd ed. SpringerLink, 1990.

[3] Ken Brazier. Thermoelectric cooler diagram, 2008. [Online;

accessed 10-November-2015] https://upload.wikimedia.org/

wikipedia/commons/thumb/3/3b/Thermoelectric_Cooler_Diagram.svg/

277px-Thermoelectric_Cooler_Diagram.svg.png .

[4] Ken Brazier. Thermoelectric cooler diagram, 2008. [Online; ac- cessed 1-February-2016] https://upload.wikimedia.org/wikipedia/

commons/thumb/8/8b/Thermoelectric_Generator_Diagram.svg/

220px-Thermoelectric_Generator_Diagram.svg.png .

[5] D.R. Brown, T.B. Stout, J.A. Dirks, and N. Fernandez. "The Prospects of Alternatives to Vapor Compression Technology for Space Cooling and Food Refrigeration Applications". Energy Engineering, 109(6):7–20, 2012.

[6] Newport Corporation. Ldm 4872 series, 2013. [Online; accessed 5-February-2016] http://assets.newport.com/webDocuments-EN/images/

4872_QCL_mount.pdf .

[7] Leo Esaki and Rafael Tsu. "Superlattice and Negative Differential Conductivity in Semiconductors". IBM Journal of Research and Development, 14(1):61–65, 1970.

[8] Jérôme Faist. Quantum Cascade Lasers. Oxford University Press - Special, 2013.

[9] Ophir Photonics Group. Vega User Manual, 2014. [Online; accessed 16- November-2015] http://www.ophiropt.com/laser--measurement/sites/

default/files/vega-user-manual.pdf .

[10] Ophir Photonics Group. 3A-QUAD_3A-P-QUAD_10A-PPS_0,

2015. [Online; accessed 26-January-2016] http://www.ophiropt.

(45)

BIBLIOGRAPHY

com/laser--measurement/sites/default/files/3A-QUAD_3A-P-QUAD_

10A-PPS_0.pdf .

[11] Ophir Photonics Group. Laser power & energy measurement laser beam analysis, 2015. [Online; accessed 1-February-2016] http:

//www.ophiropt.com/laser--measurement/sites/default/files/laser_

measurement_instruments_catalog_1.pdf .

[12] R. F. Kazarinov and R. A. Suris. "Possibility of the amplification of electro- magnetic waves in a semiconductor with a superlattice". Sov Phys Semicond, 5(4):707–709, 1971.

[13] Charles Kittel. Introduction to Solid State Physics 8th ed. John Wiley & Sons, Inc., 2004.

[14] Arroyo Instruments LLC. Tecsource 5300 series user’s guide, 2011. [Online;

accessed 15-March-2016]http://www.arroyoinstruments.com/manuals/

Arroyo5300TECSourceUsersManual.pdf .

[15] Arroyo Instruments LLC. Lasersource 4300 series user’s guide, 2012. [Online;

accessed 6 November 2015] http://www.arroyoinstruments.com/manuals/

Arroyo4300LaserSourceUsersManual.pdf .

[16] A. Lyakh, R. Maulini, A. Tsekoun, R. Go, S. Von der Porten, C. PflÃ¼gl, L. Diehl, Federico Capasso, and C. Kumar N. Patel. "High-performance continuous-wave room temperature 4.0-Î¼m quantum cascade lasers with single-facet optical emission exceeding 2 W". Proceedings of the National

Academy of Sciences

, 107(44):18799–18802, 2010.

[17] Bob Mellish. Diagram of a czernyturner monochromator, 2014. [Online;

accessed 15-March-2016]https://en.wikipedia.org/wiki/Monochromator#

/media/File:Czerny-Turner_Monochromator.svg .

[18] R. Peretti, V. Liverini, J. Wolf, C. Bonzon, S. Lourdudoss, W. Metaferia, M. Beck, and J. Faist. Room temperature operation of a photonic crystal quantum cascade laser, 2015.

[19] Vincenzo Spagnolo, Antonia Lops, Gaetano Scamarcio, Miriam S. Vitiello, and Cinzia Di Franco. "Improved thermal management of mid-IR quantum cascade lasers". Journal of Applied Physics, 103(4), 2008.

[20] AMS Technologies. Thermocube_200-400_ProductSheet_2014_NEW, 2014.

[Online; accessed 11-November-2015] http://www.amstechnologies.

com/fileadmin/amsmedia/downloads/5054_thermocube_200400_

productsheet_2014_new.pdf .

36

(46)

Appendix A

LabView Virtual Interface back panel

Figure A.1. The initialization and establishing of connection with the instruments.

Figure A.2. Starting the current output and waiting before the firs measurements.

(47)

APPENDIX A. LABVIEW VIRTUAL INTERFACE BACK PANEL

Figure A.3. The measurement loop.

Figure A.4. The last data point is collected.

38

(48)

Figure A.5. Data is saved in a file, presented on screen, and communication shut down after turning the current output off.

(49)

(50)

Appendix B

Sensor and cooling technology

B.1 Thermoelectric Cooling

The high effect and high thermal resistance of QCLs give rise to significant heat development[19]. This is something that need to be handled in order to ensure stable operation of the laser and to prevent damage. In this work, with the laser operating at or around room temperature, this is done by thermoelectric cooling with Peltier elements inside the laser mount, close to where the laser is. Thermo- electric coolers can reach about 10% to 15% of Carnot efficiency. They are not as efficient as vapour-compression systems, which can reach a Carnot efficiency of 60% [5], but they can be made much smaller, they contain no moving mechanical parts, and they are easily tailored in shape and form to fit its application. Because of these reasons they are the cooling system of choice when it comes to QCLs and other cases where vapor-compression coolers are not an option. There is still the practical issue of where the dissipated heat goes, and when ambient cooling is not enough, a second cooler is used to carry the heat away from the TEC. In this work a thermoelectric water cooler was used to cool water that was then circulated through the laser mount, thereby carrying excess heat away from the TEC in the mount.

B.1.1 The Peltier effect

The thermoelectric effect, also called the Peltier effect after its discoverer Jean Charles Athanase Peltier, is heat being generated or absorbed at a junction be- tween two different conductors due to a current flowing through it. In this case the difference in the Helmholtz free energy, F = U − T S, between the two conductors causes the charge carriers to carry heat as well as charge. F is the Helmholtz free energy, U[J] is the system’s internal energy [J], T [K] is the absolute temperature and S [J/K] is the entropy. This creates a solid state heat pump. The effect is stronger between an n- and a p-doped semiconductor due to the large difference of the Peltier coefficient between the two (see below).

The Peltier coefficient, Π is an intrinsic material property that can be defined as

(51)

APPENDIX B. SENSOR AND COOLING TECHNOLOGY

Figure B.1. A schematic of how a semiconductor thermo-electric cooler (TEC) works[3]. The charge carriers carry heat from the cold side to the hot side.

Π

e

= − (E

c

− µ +

³₂k_BT

)

e

(B.1)

for electrons, and

Π

h

= (µ − E

v

+

³₂k_BT

)

e

(B.2)

for holes [13](p.215). E

c

is the conduction band energy, E

v

the valence band energy,

µ

the Fermi level, e the elementary particle charge, k

B

the Boltzmann constant and

T

the temperature in Kelvin.

So, for an n type material it will be negative and for a p type material it will be positive, indicating the energy flux follows the charge carriers, electrons and holes.

When an n type and a p type material are used together in a so called thermocouple, larger difference in Π can be achieved than if just using two different conductors.

For a thermocouple with two materials A and B, with a current density J flowing from A to B, the energy Q absorbed at the junction will be

Q

= Π

ABJ

(B.3)

with Π

AB

= Π

B

− Π

A

[2](p.1236). If the direction of the current is reversed, the heat flux is also reversed, and a cooler becomes a heater and vice versa. A TEC is typically designed for a specific ambient temperature range and a specific temperature difference between its hot and cold sides.

A practical TEC consists of many junctions in series distributed over an area that is adjusted to whatever application it is used for. In many cases several of these layers are stacked on top of each other to increase the cooling effect(ref). They are therefore very flexible and are used when form factor and reliability are more important than efficiency.

42

(52)

B.2. THERMOPILE SENSORS

B.2 Thermopile sensors

When measuring output of relatively low power IR sources, meaning less than 3W in this case, thermopile sensors provide accurate measurements. However, they are only useful for CW or single-shot measurements, because of their long response time.

The thermopile sensor used in this work is in the form of a disc with high thermal conductivity surrounded by a ring of thermocouples connected in series. When a laser beam impinges on the disc, heat is produced, which in turn heats up the inner part of the thermopile ring. This, due to the Seebeck effect, induces a voltage over the thermopile which can in turn be measured. Then, knowing the Seebeck coefficient of the thermocouples used, the power can be calculated.

Figure B.2. Drawing of a thermopile sensor used for low power applications.[11](p.18)

B.2.1 The Seebeck effect

The Seebeck effect is basically the Peltier effect in reverse; heat induces voltage

instead of vice versa. The Peltier effect has already been described above and the

principle is the same for the Seebeck effect.

(53)

APPENDIX B. SENSOR AND COOLING TECHNOLOGY

Figure B.3. Schematic of how the Seebeck effect works. Heat generates current at a junction between different conductors.[4]

Every conductor has an absolute thermopower, S, and it is in general temper- ature dependent. So, the voltage observed over a thermocouple can be described as

V

=

^Z ^T²

T1

(S

B

− S

_A

)dT (B.4)

with T

2

≥ T

₁

. S

A

and S

B

are the absolute thermopowers for the different conduc- tors. [2](p.1236)

44

(54)

TRITA ICT-EX-2016:20

A Measurement Platform for Characterization of Quantum Cascade Lasers