Electrical Pulsing of a Laser
Diode for Usage in
Fluorescence Microscopy
Karin Jerner
Karin Jerner LiTH-ISY-EX--17/5023--SE Supervisor: Andreas Forsberg
Syntronic
Examiner: Atila Alvandpour
isy, Linköpings universitet
Division of Integrated Circuits and Systems Department of Electrical Engineering
Linköping University SE-581 83 Linköping, Sweden Copyright © 2017 Karin Jerner
En relativt ny applikation för lasern är fluorescensmikroskop. Fluorescensmikro-skopet behöver en lampkälla med hög effekt. Användandet av en laserkälla för-bättrar precisionen hos mikroskopet. En pulsad laserkälla ökar prestandan hos fluorescensmikroskopet och en laser diod kan köras på högre effekt utan att ta skada. Denna uppsats undersöker vilka egenskaper laserpulser behöver ha angå-ende pulsbredd, period och spänningsamplitud. Uppsatsen undersöker även hur dessa pulser kan genereras genom användning av elektriska komponenter. Den önskade laserpulsen bör ha en pulsbredd på 100 ps och en pulsperiod på 50 ns. Laserpulsen bör även ha en väldefinierad våglängd, stabil effekt och bör snabbt kunna slås av och på. För att uppnå denna laserpuls, bör insignalen till laserdio-den vara en spänning på 5 V, en ström på 250 mA, en pulsbredd på 100 ps och en pulsperiod på 50 ns. För att genera denna puls borde SRD:n ha låg övergångska-pacitans, låg emballagekapacitans och låg emballageinduktans. MESFET:en öns-kar låg drain ström och borde ha hög transkonduktans och en stor negativ trös-kelspänning.
A relatively new application for the laser is in fluorescence microscopes. The fluo-rescence microscope needs a high power light source input. Using a laser source improves the precision of the microscope. A pulsed laser source enhances the performance of the fluorescence microscope and a laser diode can be overdriven without being damaged. The thesis investigates which properties of the laser pulses are needed regarding pulse width, pulse period and waveform. The thesis also investigates which properties are desired for the electrical pulses driving the laser, and how they can be generated using electrical components. The desired laser pulse should have a pulse width of 100 ps and a pulse period of 50 ns. The laser pulse should also have a well-defined wavelength, stable output power and it should be able to quickly turn on and off. To achieve this laser pulse, the desired input to the laser diode should have an input voltage of 5 V, an input current of 250 mA, a pulse width of 100 ps and a pulse period of 50 ns. For generating this pulse the chosen pulse generator, an SRD, should have low junction capacitance, low package capacitance and low package inductance. The chosen amplifier, a MESFET, desires low drain current and should have high transconductance and a large negative threshold voltage.
1 Introduction 1 1.1 Background . . . 2 1.2 Previous Work . . . 2 1.3 Method . . . 2 1.4 Restrictions . . . 3 1.5 Outline . . . 3
2 Part I - Investigation of the Laser Waveform 5 2.1 Fluorescence Microscopy . . . 5
2.1.1 Light Sources . . . 6
2.2 Laser . . . 8
2.2.1 Light Emission . . . 9
2.2.2 Required Pulse Waveform . . . 9
2.2.3 Laser Diode . . . 13
2.3 Summary . . . 15
3 Part II - Investigation of Pulse Generation 17 3.1 Pulsing Method Choice . . . 17
3.1.1 Desired Electrical Pulse . . . 18
3.1.2 Average Power and Peak Power . . . 18
3.2 Circuit Components . . . 19
3.2.1 Pulse Generation . . . 19
3.2.2 Amplification and Narrowing . . . 22
3.3 MESFET Amplification and Narrowing of SRD Impulses . . . 24
3.4 Summary . . . 24 4 Implementation 27 4.1 LTspice . . . 27 4.2 Laser Diode . . . 28 4.2.1 Input Voltage . . . 29 4.2.2 Input Current . . . 29 4.3 SRD . . . 30 4.3.1 Junction Capacitance, Cj0 . . . 31 ix
4.3.2 Package Capacitance, Cp . . . 32 4.3.3 Package Inductance, Lp . . . 33 4.4 MESFET . . . 34 4.4.1 Transconductance . . . 35 4.4.2 Threshold Voltage . . . 35 4.4.3 Drain Current . . . 36 5 Discussion of Implementation 39 5.1 Laser Diode . . . 39 5.2 SRD . . . 40 5.3 MESFET . . . 41 5.4 Summary . . . 41 5.4.1 Future Work . . . 43 6 Conclusions 45 A Simulation graphs 49 Bibliography 65
List of figures
Figure 1.1 Overview of the pulsed laser 1 Figure 2.1 Overview of a a fluorescence microscope 6 Figure 2.2 Singlet and triplet states 8 Figure 2.3 Stimulated emission by inserted photon 9 Figure 2.4 Overview of a waveform og an ideal pulse 10 Figure 2.5 Overview of a waveform of a generated pulse 11 Figure 2.6 PN-junction and band gap of laser diode 14 Figure 3.1 Overview of the electrical approach 17 Figure 3.2 Overview of approach affecting the laser beam 17 Figure 3.3 Step recovery diode, impulse output from sinus input 20 Figure 3.4 Equivalent circuit of a step recovery diode 21 Figure 3.5 Equivalent circuit of a step recovery diode with a diode 21 Figure 3.6 Overview of an GaAs nMESFET transistor 22 Figure 3.7 The MESFET schematic symbol 23
Figure 3.8 Effect of MESFET 23
Figure 3.9 The pulse path through an SRD and a MESFET 25 Figure 4.1 Simulation circuit of diode 28 Figure 4.2 Simulation circuit of SRD 31 Figure 4.3 Simulation circuit of MESFET 35
List of tables
Table 4.1 Table over diode output voltage dependence on input voltage 29 Table 4.2 Table over diode output voltage dependence on input current 30 Table 4.3 Table over SRD output voltage dependence on junction capacitance 31 Table 4.4 Table over SRD pulse width dependence on junction capacitance 32 Table 4.5 Table over SRD output voltage dependence on package capacitance 32 Table 4.6 Table over SRD pulse width dependence on package capacitance 33 Table 4.7 Table over SRD output voltage dependence on package inductance 34 Table 4.8 Table over SRD pulse width dependence on package inductance 34 Table 4.9 Table over MESFET input pulse properties 35 Table 4.10 Table over MESFET output voltage dependence on transconductance 36 Table 4.11 Table over MESFET output voltage dependence on threshold voltage 36 Table 4.12 Table over MESFET output voltage dependence on input current 37 Table 4.13 Table over MESFET pulse width dependence on input current 37 Table 5.1 Table over preferred properties of inputs and component parameters 42
1
Introduction
This thesis focuses on the topic of using a pulsed laser diode as light source of a fluorescence microscope. Mainly regarding which pulses are required, and how they can be generated. Pulsed laser diodes are commonly used in research and clinical diagnosis in the medical area, e.g. in fluorescence microscopy, which require large power laser emissions. The alternative, a constant high power input, can cause damage to the laser and the specimen that is being analyzed. Instead, if the laser is only active during a fraction of the time the fluorescence microscope performance can be enhanced. This can be done by pulsing the laser with very short pulses meanwhile using a slow repetition rate, relative to the pulse width. An overview of the whole system is presented in Fig. 1.1, where the block called ’Pulse generation’, and the ’laser beam’ are investigated in this thesis. The output signal from the pulse generation block pulses the laser, which in turn produces light that is sent to the fluorescence microscope. These pulses are not trivial to create, but the technique is desired for practical applications. The purpose of this master thesis is to investigate which pulses are required and techniques that can be used to produce these pulses.
Figure 1.1:Overview of a pulsed laser.
1.1
Background
Ever since the laser was first invented in 1960 the usage has increased more and more every decade.
At present, lasers are often used in analytical applications, such as for re-search and for clinical diagnosis in the medical area. Applications also exist for materials science, particle analysis and for quality control in the semiconductor and pharmaceutical industry.
In the medical field, a relatively new application for the laser is in fluorescence microscopes. The sample being analyzed in the microscope, absorbs energy caus-ing its atoms to become excited. When the electron drops back to the ground state it emits a photon and the atom becomes fluorescing. The emitted light is de-tected and a viewable picture is produced, AB [2014]. By using a laser as the light source the precision is improved, De and Goswami [2016], which is the reason for using the laser.
The main reasons for using a pulsed laser source is that the laser diode can be overdriven without damage and that the fluorescence microscope performance is enhanced by using a pulsed source. Due to the importance of this, this thesis focuses on an electrical approach to pulse the laser diode.
Generating subnanosecond pulses is a new field, and generating pulses with large amplitude is even more in its starting tracks, Huiskamp et al. [2015], Chen-gou et al. [2015]. For the thesis, the first problem is to investigate what properties of the laser pulses are needed regarding pulse width, pulse period and waveform. The second problem is to generate electrical pulses, to create these laser pulses, using electrical components. Because of the complexity of similar problems, and lack of previous reports, this master thesis aims to investigate and resolve this issue.
1.2
Previous Work
In the article by Aswani et al. [2012], several light sources for the fluorescence mi-croscope are investigated. The comparison is between the long-lived high energy focused LEDs and the relatively short-lived broad spectrum arc lamps. The arti-cle’s conclusion is that the LEDs are a good investment if the microscope is heav-ily used. The article by De and Goswami [2016] uses a pulsed laser light source to enhance the performance of the fluorescence microscope, decreasing photo-damage. Regarding pulse generation the article by Miao and Nguyen [2003], uses a SRD and MESFET to create an impulse with 2.3 V amplitude and a half-power pulse width of 115 ps.
1.3
Method
This thesis was written at one of Syntronic’s offices. It is based on literature on fluorescence microscopy, pulse generation, amplification and basic electronics. Simulations of a laser diode and chosen circuit components have been carried
out in respect to parameters with impact of the devices. The simulations was performed in LTspice using SPICE models and equivalent circuits.
1.4
Restrictions
Circuit implementation is a very broad area with a lot of different aspects to consider. There are several simulation programs and analysis techniques which can be used. Since this master thesis aims to investigate what pulses are needed for fluorescence microscopes and to find a technique to generate these pulses, the main focus is on finding that information and technique and performing a qualitative simulation. The thesis work and writing was done at Syntronic and the external supervisors helped with setting the limitations.
The pulse generation will only be implemented using models in this thesis and evaluated using a simulation tool.
1.5
Outline
Following this chapter:
• Chapter 2 presents the investigation of the desired laser waveform, includ-ing discussion.
• Chapter 3 presents the investigation of pulse generation, including discus-sion.
• Chapter 4 describes the implementation that has been made. This part in-cludes the results that have been obtained.
• Chapter 5 discusses the results of this work as well as future improvement and development.
2
Part I - Investigation of the Laser
Waveform
This chapter gives an overview of the fluorescence microscopy and physical quantities of the laser waveform. The purpose of this part of the thesis is to investigate what properties of the laser pulses is required for the fluores-cence microscopy application.
2.1
Fluorescence Microscopy
As mentioned in the previous chapter, the laser technique is becoming more common in the medical area of the fluorescence microscopes, which by be-ing awarded with the Nobel price in Chemistry 2014 proves to be a tech-nique with considerable potential.
The basis of the fluorescence technique is to excite atoms in a specimen and analysing the emitted light. The specimen can, for instance, be a cell membrane and the technique is often used in medical research. Spring and Davidson [1999] An overview of a fluorescence microscopy is given in Fig. 2.1. Fluorophores are added to a specimen binding to certain molecules of interest, with a corresponding wavelength desired as light input. The light goes into the microscope through a filter, which makes sure that only the desired frequencies enter the microscope, effectively keeping only the desired energy levels. The incoming light is reflected down on the specimen that is being analysed using a beam splitting mirror. Photons from the light source excites electrons in the specimen which become excited and jump up to a higher energy state. The electrons cool down and jump back to their ground state. While returning to the ground state the atoms emit characteristic light, by sending out photons with a characteristic amount of
energy, related to the material in the specimen. The emitted light is emitted spherically and a part of the light travels up through the beam splitting mirror. This mirror is designed in such a way that the wanted frequencies is not reflected. The light then passes through another filter, making sure no undesired frequencies reach the final step. The light continues up to the observer or detector creating a picture of the specimen, where it can be analysed. Spring and Davidson [1999]
Figure 2.1:Overview of a fluorescence microscope.
2.1.1
Light Sources
There are multiple light sources applicable to the fluorescence microscope. For a broad spectrum there are several white light sources available, i.e arc lamps. These are high-powered light sources generating intense bands for fluorescence excitations covering the entire UV-visible light spectrum. These lamps have a spectral peak where the lamp produces a lot of power. This is also a disadvantage for these lamps due to the fact that they have a non-uniform illuminating across the microscope field of view. Another disadvantage is that the bulb has to be replaced since the lamp intensity decreases over time.
emit-ting diode, LED. The LED is positioned to take over the market as the light source of choice due to several reasons. LEDs are much smaller than the alternatives and can be built into microscope stands. The LEDs generate a discrete excitation peak at a certain wavelength, making the LEDs limited at a narrow spectrum of light emission. The LEDs can be switched on and off in an instant, and have a much longer lifetime than the alternatives. A number of LEDs can be used in combination to mimic a white light source, to broaden the LEDs spectrum. Aswani et al. [2012] The fluorescence mi-croscopy lit by a LED uses fluorophores matched for certain wavelengths. Bosse et al. [2015]
Laser diodes combine the benefits of the arc lamps and the LED. They pro-vide intense light, as for arc lamps, but focus the light on a specific point, instead of across the whole field, making the intensity less damaging to a living cell. The lasers are long-lasting, just as the LEDs, providing light at specific wavelengths, decreasing the need for filters. Adding these prop-erties together probably makes the laser diodes the best light source for fluorescence microscopes. Lasers can be used in, so called, superresolution microscopes with nanoscale resolution, able to show individual molecules in unprecedented detail. Although, a disadvantage of the laser is the higher price. This makes the laser diode still used primarily when higher resolu-tion is needed. Bushwick [2012]
Pulsed Light Source
The traditional light source in a fluorescence microscope is a continuous wave. The continuous wave can lead to excited state absorption causing photo-damage. Instead of this continuous wave a pulsed light source can be used, achieving significant fluorescence enhancement, due to reduction of the photo-damage. There are two main reasons for the pulsed light source to be advantageous over the continuous wave. The atoms in the specimen are ensured to go back to ground level before the next light pulse goes in to the microscope using a pulse width less than the lifetime of the excited-state. De and Goswami [2016] An ordinary excitation excites an electron from the ground state, S0, to a singlet state, e.g. S1, and back to the ground
state again. The excited state molecule may convert into a state where the electron has changed it spin, called triplet state, T1. The triplet state has
gotten its name due to the fact that it corresponds to three states of equal energy. Valeur [2013] The excitation and de-exitation of electrons passing through a triplet state can be seen in Fig. 2.2. According to De and Goswami [2016], the pulse period shall also be longer than the excite-state triplet lifetime due to the fact that the slowest de-excitation pathway is the triplet state relaxation.
The fluorescence lifetime is the time for which the electron in the fluo-rophore is excited. In the study, Berezin and Achilefu [2010], and book, by Valeur [2013], the fluorescence lifetime goes from around hundred
picosec-Figure 2.2:Singlet and triplet states
onds to a few tens of nanoseconds. According to the study De and Goswami [2016], a shorter pulse width than the fluorescence lifetime is desired, giv-ing that a pulse around 100 ps could be used for most fluorophores. Also according to De and Goswami [2016], the period of the pulses has to be adapted to the triplet lifetime, giving that a period of around 50 ns should also be appropriate. These requirements are applicable to a wide range of possible fluorophores that might be used.
2.2
Laser
Laser is an acronym for: Light Amplification by Stimulated Emission Radia-tion. It was first discovered by Theodore Maiman in 1960, and has since become a crucial device in many applications, both in our every day life, such as in computers, and in highly advanced technical devices used in in-dustries, medical treatment and research. Diels and Arissian [2011]
2.2.1
Light Emission
The gain medium of the laser consists of atoms well prepared to be excited. The excitation is performed by a pump mechanism. The efficiency is of-ten determined by this mechanism, which is a very important part of the laser. The excited atoms can emit photons either spontaneously, which is done in any direction, or stimulated in a predefined direction. The stimu-lated emission, see 2.3, is initiated by the insertion of a few photons which contribute to other electrons jumping back to ground level, emitting addi-tional photons of energy Eg, corresponding to the band gap. In an instance
there are double the amount of photons able to stimulate other electrons similarly. All the photons in stimulated emission follow the first photon’s direction, creating a sharp and strong light beam, characteristic for lasers. Therefore, the stimulated emission is crucial for the laser to work properly. Diels and Arissian [2011] The time for which this event takes place is called the propagation time for the laser, being the time for which the laser turns on. The propagation time creates the lower bound for pulsing of the laser, but is much lower than the pulse around 100 ps, appropriate for the fluo-rophores.
The band gap energy corresponds to a certain wavelength, due to Eg = hf·c,
where h is Planck’s constant c is the speed of light and f is the wavelength of the light. A smaller wavelength corresponds to a larger band gap and energy. Diels and Arissian [2011]
Figure 2.3:Stimulated emission by inserted photon.
2.2.2
Required Pulse Waveform
Laser pulses suited for the fluorescence microscopy technique should be focused, having a well-defined wavelength, to provide light adjusted for a
specific fluorophore. A laser with a specific wavelength removes much of the need for filtering, required for wide-spectrum light sources. It also re-duces photobleaching, compared to wide-spectrum light. The laser pulses should also be of high intensity, to stimulate as much emission as possible in the analysed specimen, providing a more intense image of the specimen. The pulse power should be kept stable to provide an even input to the mi-croscope. To make imaging of live cells possible the light source should be able to quickly turn on and off, due to the fact that continous high intensity light may cause damage, such as phototoxicity and cell death. The light has to be used efficiently, focusing the laser pulses only on the parts of the spec-imen being analysed, and only during a required time frame. The pulse widths and repetition rate has to be suited for the fluorophores used for the analysis. Bushwick [2012], Aswani et al. [2012], Bosse et al. [2015]. An ideal pulse has zero rise and fall time, and the output is constant, as can be seen in Fig. 2.4. The desired pulse have several similarities to an ideal pulse. According to Markettech [2015], aspects such as rise time, fall time, overshoot, undershoot, ringing and reflections can not be neglected for a generated pulse waveform. Some of these aspects can be seen in Fig. 2.5.
Figure 2.4:Overview of a waveform of an ideal pulse.
To obtain a pulse suited for use in the fluorescence microscopy application, several aspects of the generated pulse has to be regarded. According to Maxim [2000], there are several different distortions caused by the setup of the laser that can be solved easily. The impact of the different distortions are taken under consideration, while regarding the importance to resolve them.
Figure 2.5:Overview of a waveform of a generated pulse.
Overshoot
Overshoot of a pulse is when the pulse exceeds the intended amplitude of the pulse, e.g. 100 %, as can be seen in Fig. 2.5. Overshoot can be caused by both the rising edge and a low bias current. If the rising edge of the pulse is too fast, the rising edge may overshoot the digital one level. This can be re-solved by using a low-pass filter with a frequency cutoff at 75 % of the data rate. This decreases the overshoot by slowing down the rising and falling edge. If the digital zero level is below the threshold of the laser is may cause a delayed rising edge. The delayed edge results in a build-up in potential causing the laser to overshoot whilst the threshold is reached. This can be resolved by increasing the bias current above the threshold value. Maxim [2000]
Undershoot
Undershoot of a pulse is when the pulse goes below the intended amplitude, e.g. 100 %, as can be seen in Fig. 2.5. Undershoot can be caused by the rising and/or falling edge not reaching their high or low level within the first half or the unit interval. This may be resolved by decreasing the damping of the output circuit. Maxim [2000]
Ringing
Ringing of a pulse is when the pulse signal oscillates, as can be seen in Fig. 2.5. Ringing on rising and/or falling edges may be caused by impedance
discontinuities, large inductances in the circuit or resonance effects of cir-cuit components. For resolving this issue these has to be decreased or re-moved. Maxim [2000]
Reflections
Reflections can appear as overshoot, undershoot, ringing or other distor-tions and are caused by transmission-line impedance discontinuities. For re-solving this issue the discontinuities has to be decreased or removed. Maxim [2000]
Discussion Regarding Distortions
As stated, the desired laser pulse should have a stable output power, fast switching between on and off, and high intensity for a narrow wavelength-spectrum. These properties has to be kept in order for the fluorescence microscopy to work as intended.
Overshooting of the laser pulse creates a higher intensity output for a cer-tain part of the laser pulse. This may result in an output pulse concer-taining too much energy. This issue needs to be taken care of so that the laser pulse does not damage the specimen being analysed.
Undershooting of the laser pulse creates a lower pulse than desired. This may result in that the output pulse does not contain the required amount of energy to excite the specimen in the fluorescence microscope. This issue has to be taken care of in order for the fluorescence microscope to be able to create an image of the specimen being analysed.
Ringing can, as for overshoot and undershoot, create differences in power output. Ringing can also cause the output to act like several pulses, if the ringing is large. This may cause additional laser pulses to go in to the flu-orescence microscope, causing undesired excitation. Ringing in the form of overshoot and undershoot has the same effect on the fluorescence mi-croscopy analysis as mentioned above, and needs to be reduced accordingly. As for the ringing resulting in several pulses being created, the specimen be-ing analysed in the fluorescence microscope might get excited from several photons. This could create additional sources of emission, making the im-age less sharp.
Reflections cause the same problems as the respective distortion it appears as, and needs to be solved for the same reasons.
An ideal pulse should be the most fit pulse to send in to the fluorescence mi-croscope, but small distortions of the laser pulse should not cause problems for the fluorescence microscope. As stated previously, the most common light source for the fluorescence microscope has been the arc lamp. The arc lamp, which decreases in intensity over time and has a continous flow
of light. The fluorescence microscope is proven to work under these condi-tions, at least with respect to undershoot and ringing. But, this is also the reason why the laser has been proven to be a supreme solution for the flu-orescence microscope, due to the longer endurance of the light source and sharpness of the image. Based on that the lasers are much more expensive than the alternatives, these problems should indeed be solved so that the fluorescence microscope can work as good as possible, creating really sharp images of the analysed specimens.
2.2.3
Laser Diode
The laser diode is a semiconductor device consisting of a crystal that has been doped differently on either side, see Fig. 2.6. The PN-junction laser is relegated to history but serves well to illustrate some of the main principles, according to Sands [2004]. The PN-junction laser is n-doped with donors on one side, which produce an abundance of electrons, and p-doped with ac-ceptors on the other side, with an abundance of holes. This gives the crystal different physical properties on the different sides. The currect flows easily through the semiconductor from the p- to the n-side while there is great resistance if positive voltage is applied to the n-side while negative voltage is applied to the p-side. This is why it is called a ’diode’. In the semiconduc-tor the conduction band, containing the exited electrons, and the valence band, containing the corresponding holes, are separated by a band gap, be-tween which the electrons can jump during excitation/emission. Diels and Arissian [2001]
The voltage drop, V , over the laser diode is, according to Vanzi [2008], ap-proximately equal to the band gap energy, Eg. The relationship between
the voltage drop, measured voltage, VE, current, I, and series resistance,
RS, can be seen in Eq. (2.1).
VE= V + RS· I (2.1)
When high current density is going through the laser diode the temperature rises. The temperature is an important factor for the performance of the laser diode and the laser diode needs to be cooled down or alternatively operated under pulsing conditions. According to Sands [2004], pulsed PN-junction laser diodes have, at room temperature, been made operating as well as constantly turned on diodes.
According to Andrews et al. [2013], the total forward current, IF going
through the PN-junction diode can be calculated as in Eq. (2.2). IS is the
saturation current, VF is the forward voltage and, VT is the thermal
volt-age. At room temperature the thermal voltage, VT, is approximately 0.026
V, as can be seen in Eq. (2.3), where k is Boltzmann’s constant, T is the tem-perature in Kelvin and q is the elementary charge. Together these give the relationship given in Eq. (2.4).
Figure 2.6:PN-junction and band gap of a laser diode. IF = IS· (e VF VT −1) (2.2) VT ≡ k · T |q | = 1.380 · 10−23· 300 |1.602 · 10−19| ≈0.026 (2.3) IF = IS· (e VF 0.026 −1) (2.4)
Desired Laser Pulse and Laser Diode
According to Hodgson et al. [2004] the pulsed laser diode receives lower threshold current and a slight increase in slope efficiency mainly due to the decrease in junction temperature. As noted earlier the temperature is crucial for the laser diode and according to Hodgson et al. [2004], pulsing at a duty cycle of less than one percent, the heating effects are insignificant. The stated pulse width of 100 ps and pulse period of 50 ns has a duty cycle of 0.2 %, implying that the heat is not an issue.
The pulse stated above is applicable for the fluorescence microscope ap-plication based on properties within the laser diode. This gives that the required pulse is 100 ps wide, with a period of 50 ns.
In the study, Lopez et al. [2000], the fluorescent dye SYPRO Ruby Pro-tein Gel Stain is used for proPro-tein analysis. The SYPRO Ruby ProPro-tein Gel Stain from Sigma, product number S 4942, has optimal excitation at 302 and 470 nm. This is a well-suited fluorescent for the analysis made in this thesis. The laser diode determines the requirements on the pulse shape in terms of current and voltage. The fluorescent dye SYPRO Ruby Protein Gel Stain desires an incoming light source with wavelength 302 or 470 nm. A laser diode able to excite an 470 nm output is theSMLS14BET from ROHM Semiconductor. The forward voltage of the laser diode is typically 3.2 V with a maximum value of 5 V. The forward current is typically 20 mA with max-imum value of 30 mA. The maxmax-imum power dissipation is 117 mW. This laser diode is suited for the fluorescence microscope application based on the application’s requirements on the output of the laser diode.
2.3
Summary
A fluorescence microscope analyses specimens using a high energy light source producing light corresponding to an added fluorophore. The per-formance of the fluorescence microscope is enhanced using a pulsed laser diode as light source. The waveform of the laser pulse should have fast rise and fall times, a stable power output and a narrow wavelength-spectrum. Distortions, such as overshoot, undershoot, ringing and reflection, should be decreased or removed for the fluorescence microscope to work properly. The laser pulse should have a pulse width less than 100 ps and a period of more than 50 ns. The laser diode’s wavelength should correspond to a cer-tain fluorophore. The laser diodeSMLS14BET is a suitable light source for the fluorescence microscopy, corresponding to the fluorescent dye SYPRO Ruby Protein Gel Stain used in protein analysis.
3
Part II - Investigation of Pulse
Generation
This chapter gives an overview of the pulsing of lasers. The purpose of this part of the thesis is to investigate how the desired laser pulse stated in the previous chapter can be generated.
3.1
Pulsing Method Choice
Pulsing the laser light has two fundamentally different approaches. The pulsing can either be electrical, turning the laser on and off by targeting the laser input chain, or by effecting the laser beam itself. The two alternatives are illustrated in Fig. 3.1 and Fig. 3.2, respectively.
Figure 3.1:Overview of the electrical approach.
Figure 3.2:Overview of the approach affecting the laser beam.
For the electrical approach, a circuit that produces pulses is needed. The
propagation time of the laser is the time it takes for the laser to turn on. The rise time of the output can not go below the propagation time of the laser, which sets the maximum frequency possible of the pulsing. The ultimate challenge for this alternative is to reach, but not exceed, the speed of the propagation time. The propagation time is well below the rise time inves-tigated in this thesis, so this will not be a restriction. Furthermore, higher energy pulses can be sent through the laser without damaging it due to the fact that the laser is off periodically, reducing the temperature of the laser. Generating the pulse can be done by using a simple pulse generator, which can be enough for many applications. An oscillator can, e.g, generate a si-nusoidal wave with 10 V amplitude by itself. The generated pulses might need amplification and/or narrowing to reach given requirements, this can be done by some sort of amplifying circuit. Kilpela [2004], Chengou et al. [2015], Huiskamp et al. [2015] and Zhou et al. [2015].
The second approach is based on adding a device located after the laser beam, for which the rise time of the outgoing signal is totally dependent on the implementation properties, which can vary between the techniques, and get very short. There are several options available for affecting the laser beam, such as fiber, mechanical, using mirrors etc.
For the implementations in the second approach the laser is constantly turned on, which is uneconomical and decreases the lifetime of the laser. The electrical approach can pulse the laser with high energy without dam-aging the laser. Therefore, it should be able to solve the problem by reduc-ing the pulse width, meanwhile bereduc-ing the more energy efficient alternative. Arrigoni et al. [2012], Lin and Lin [2014] and Shaozhen et al. [2012].
3.1.1
Desired Electrical Pulse
The laser pulse stated in the previous part can be achieved by electrically pulsing the laser diode. As stated previously the laser diode is chosen be-cause of its narrow wavelength and stable power output, which is not signif-icantly affected by the input signal. The rise and fall times of the electrical pulses could be considered equal to the laser pulse, due to the laser diode’s quick response to the input signal. This also makes the requirement on the pulse width and pulse period equal to the desired laser pulse. As a conse-quence, the desired electrical pulse has a pulse width of 100 ps and a pulse period of 50 ns, and the rise and fall times should be short.
3.1.2
Average Power and Peak Power
In order to keep a laser undamaged the average power has to be kept low. The average power of the pulses can be calculated using the period of the pulses, the width of the pulses and the peak power of the pulses. For easier calculations the pulse is assumed to be an ideal square wave, with straight
edges. The period of a signal, T , is the duration of time of one cycle in a repeating sequence. This is, for a pulsing signal, the time between e.g. the rising edges of two consecutive pulses. The pulse width, Twidth, is the time
duration of the pulse, from rise to fall. The peak power, Ppeak, is the highest
power generated at some point throughout the signal. The peak power of an ideal pulse signal can be seen as the amount of energy generated during a single pulse, divided by the pulse width. This is due to the fact that the power between the pulses is ideally zero. These relationships can be seen in Eq. (3.1).
The average power, Paverage, is the amount of energy generated on a
sus-tained basis, over a longer period of time, regardless of pulse duration. Av-erage power is equal to the total energy measured, Etotal, divided by the
total time in seconds during which the measured amount of energy was ac-cumulated, Ttotal. The average power is, by definition, always equal or less
than the peak power of a signal. In an ideal case the total energy during a period is equal to the energy generated during one pulse, Epulse. The
aver-age power from the signal is, approximately, the energy generated during a single pulse. This is the peak power multiplied with the pulse duration, divided by the period time. These relationships can be seen in Eq. (3.2). A laser has a limit for the average power going through it, and suffers dam-age above that limit. The peak power needs to be ’high’ so that the defined performance is maintained. Therefore, the average power can be kept low either by keeping the pulse width narrow, the period long, or a combination of the two bringing the average power down to a safe level. Newport [2016] Epulse= Ppeak· Twidth (3.1)
Paverage= Etotal Ttotal = Eperiod T = Epulse T = Ppeak· Twidth T (3.2)
3.2
Circuit Components
A circuit that, from a DC voltage input, produces a pulse train can be im-plemented using a few circuit components. From the DC voltage a pulse is generated, amplified to the right level and narrowed to the right width using the circuit parts presented in Sec. 3.1. The circuit parts, will be ex-plained in the following sections.
3.2.1
Pulse Generation
Many pulse generators are oscillators, which produce periodical output sig-nals from a DC input. The output signal can be a square wave, sine wave, triangular wave etc. Kazimierczuk [2015]. Although, a very commonly
used device for pulse generation is the step recovery diode, SRD. Miao and Nguyen [2003]
Step Recovery Diode
The SRD can be used as an pulse generator by converting a sinusoidal in-put signal into narrow pulses, down to <150 ps. The repetition rate of the pulses is the same as for the sinusoidal signal, due to the fact that the SRD outputs one pulse per period. An illustration of this is presented in Fig. 3.3.
Figure 3.3:Step recovery diode, impulse output from sinus input. It is an advantage for the SRD to output one impulse per period of the sinewave, due to the fact that the frequency of the pulses is easily controlled. An SRD is a two-terminal PN-junction with similar DC characteristics as a usual PN-junction diode, but with different dynamic characteristics, used in switching. Hewlett-Packard [1968]. During positive bias, minority car-riers are inserted on both sides of the PN-junction. The SRD allows the minority carriers to concentrate to a narrow region near the junctions. The minority carriers have a long lifetime, which means in that they cannot re-combine during positive bias. When positive bias is turned into a negative bias these minority carriers flow in the opposite direction from the junction, which create a strong backward current. When all the minority carriers are extracted the current is quickly reduced to a low level, cutting the diode off, and thereby creating a step voltage. Zhou et al. [2016]
A real SRD experiences impact from package parasitics which may result in ringing, overshoot and ramping. An equivalent circuit of an SRD can be seen in Fig. 3.4. The SRD can be seen as a diode, a capacitance and an inductance, seen in Fig. 3.5.
The average value of the output voltage has to be equal to zero during a cycle, which restricts the pulse height and the output during cut-off. The impulse width, W , is dependent on the SRDs , Cj0 and inductance, Lp,
ac-cording to Eq. (3.3), Hewlett-Packard [1968]
Figure 3.4:Equivalent circuit of a step recovery diode.
Figure 3.5:Equivalent circuit of a step recovery diode as a diode with pack-age parasitics.
To generate a narrow pulse width with high amplitude and low ringing an SRD with small junction capacitance should be used. Zhou et al. [2016] According to Hewlett-Packard [1968], an SRD can be used to generate an impulse at below 150 ps, which is close to the required pulse stated in the previous chapter, but not close enough. To further reduce the pulse width and to increase the amplitude of the impulse generated by the SRD a nar-rowing and amplifying circuit is needed, to simultaneously reach below the required 100 ps and to keep the voltage relatively high. The fact that the SRD can make an impulse from a sinusoidal signal with the same repeti-tion frequency is a clear advantage for the SRD to be a good candidate as a pulse generator, especially due to the fact that a sinusoidal signal is easily generated by an oscillator.
The SRD model used for implementation is taken from Hewlett-Packard [1968] and the specific SRD was chosen for its low junction capacitance in accordance to results in Zhou et al. [2016].
3.2.2
Amplification and Narrowing
The main purpose of an amplifier circuit is to increase the power level of an incoming signal. Kazimierczuk [2015] Sometimes they can also be used as narrowing circuit components.
Metal-semiconductor Field Effect Transistor
A metal-semiconductor field effect transistor, MESFET, is a semiconductor device which can be used to narrow and amplify pulses. The MESFET con-sists of a conducting channel between a source and drain contact region. A Schottky barrier diode is used to isolate the device’s metal gate from the channel and is used to control the carrier flow in the channel. This is done by varying the depletion layer width, modulating the thickness of the con-ducting channel. Zeghbroeck [2004] An overview of a Gallium Arsenide, GaAs, nMESFET, n-doped and constructed with a GaAs-semiconductor ma-terial, can be seen in Fig. 3.6, and the schematic symbol can be seen in Fig. 3.7, where S represents the source, D the drain and G the gate. The metal gate-control electrode is connected to the channel, Hudson [2014]. This can be compared with the more common MOSFET, abbreviation for metal-oxide semiconductor field effect transistor, which uses an oxide layer to separate the gate and the channel.
Figure 3.6:Overview of an GaAs nMESFET transistor.
Compared to a MOSFET the MESFET has one key advantage, which is its higher carrier mobility. The higher carrier mobility leads to higher current
Figure 3.7:The MESFET schematic symbol.
and transconductance of the device. The transconductance of the device can be calculated according to Eq. (3.4). A disadvantage of the MESFET is that the Schottky metal gate limits the forward bias voltage, so that the threshold voltage must be lower than the turn-on voltage off the Schottky diode. Although this is easily tolerated, and the advantages exceed the dis-advantages. For the material used GaAs is advantageous over silicon, due to its 5 times higher carrier mobility, with a twice as high peak electron velocity. Zeghbroeck [2004]
gm=
dID,sat
dVGS
(3.4)
Figure 3.8:Effect of MESFET.
By varying the DC voltage at the gate of the transistor an impulse can be voltage amplified and compressed, in order to significantly reduce the
broader pulse foot and the strongly nonlinear variation of the transconduc-tance. An overview of the effect of the MESFET is given in Fig. 3.8.
3.3
MESFET Amplification and Narrowing of SRD
Impulses
In the article, Lee and Nguyen [2001], an impulse is generated by an SRD and shaped by a MESFET network, including a Schottky diode, a capaci-tor and resiscapaci-tors, at a frequency of 10 MHz. The article claims that the frequency on such a circuit is limited by the performance of the SRD, be-ing able to manage several hundred MHz. The output pulse from the SRD reaches a width of 300 ps with an amplitude of 2 V, peak-to-peak. The fre-quency limitation by the SRD, which the article suggests, does not seem to place any limitation for our construction, where a frequency of only 20 MHz is required.
According to Miao and Nguyen [2003] a circuit containing an SRD and two GaAs MESFET transistors can create a narrow impulse of 115 ps with an amplitude of 2.3 V. An overview of the pulse path can be seen in Fig. 3.9. The SRD is used as pulse generator generating a 140 ps impulse with an amplitude of 1.1 V. To amplify this pulse a MESFET is used, inverting the signal as well as amplifying it. The pulse then reaches 2 V at 130 ps, so the MESFET is also narrowing the pulse slightly. The second MESFET is used primary for narrowing of the pulse, but is also amplifying the pulse slightly. After this stage the impulse of 115 ps with an amplitude of 2.3 V is reached. The pulse path going through the SRD and MESFETs can be seen in Fig. 3.9. The SRD and MESFET devices should be implemented for analysis, to see if they are fit for providing the desired electrical pulse.
3.4
Summary
The desired pulses can be generated by electrically pulsing the laser using a circuit implementation. To achieve high energy emission the laser can be overdriven for a certain time, as long as the average power is low. The needed pulses are produced using a pulse generator circuit and an amplifi-cation circuit, which together need to fulfill functionality requirements on the desired pulse. An SRD could be used to generate a narrow pulse and a MESFET circuit could be used to amplify it. Both the requirements from the fluorescence microscopy and the laser diode itself have to be considered when creating a circuit implementation.
4
Implementation
This chapter includes the implementation of the laser diode and circuit com-ponents chosen to generate the desired pulse from Part I. The implementa-tion in this thesis consisted of several simulaimplementa-tions of the chosen components regarding input parameters and device properties detected as important in Chapter 3, analyzed by simulation over a wide range of values.
Something worth noticing is that the pulse width is the time from 50 % power level of the rising edge to 50 % power level of the following falling edge. Rise and fall time can have great impact on the pulse width. The efficiency described in this chapter is the measure of output voltage in rela-tions to the input voltage. The efficiency is given in percent.
Figures of simulations results and graphs can be found in the Appendix.
4.1
LTspice
The simulations are performed in the circuit simulation program LTspice. LTspice is a freeware computer software implementing a SPICE simula-tor. SPICE, Simulation Program with Integrated Circuit Emphasis, is used broadly to simulate electrical circuits. Some components have been mod-eled by SPICE models, and some are modmod-eled by an equivalent circuit, as in the case with the SRD. LTspice is not a perfect program and it has its limitations. Parameters not stated in the models are set to default values by LTspice. Due to the fact that the analyses are only considering the depen-dencies on parameters the models does not have to be exact. Circuit para-sitics are not taken into consideration in simulations, due to the fact that the master thesis is only an initial study of the topic. The simulations that
are run are transient analysis simulations. These simulations are run over a range of time. The simulated parameters are run over a set of 6-9 different values. These are based on the parameter values of the real devices that the simulated models shall correspond to. Values both above and under the actual value is simulated. The values were selected to get a broad spectrum of parameter values. Additional values were simulated, which followed the presented lines, not needed to be stated in the tables and graphs.
4.2
Laser Diode
The SPICE model of the laser diode is provided fromROHM Semiconductor, given below. These parameter values are used in the laser diode model. .MODEL SMLS14BET
+ IS=215.94E-21 N=2.9257 RS=13.21 IKF=3.4242E-3 EG=3.5 CJO=41E-12 + M=.38596 VJ=5.6484 ISR=20.977E-12 NR=10 BV=5 TT=17.4n
The diode implementation was carried out using a circuit schematic shown in Fig. 4.1. The pulse coming in to the laser is assumed to be the desired pulse stated earlier to have a pulse width of 100ps and a pulse period of 50 ns.
Figure 4.1:Simulation circuit of diode.
Properties determining the output of the laser diode is believed to be the current and voltage, and their impact was simulated and analyzed.
To analyze the impact of the input voltage on the chosen diode the current was kept constant. To analyze the impact of the input current on the cho-sen diode the voltage was kept constant. This was done by changing the value of the connected resistance in series, Rconnected according to Ohm’s
law, which can be seen as R1 in Fig. 4.1. For example, at a voltage of 10 V
and a current of 50 mA, the connected resistance was set to 186.79Ω, ac-cording to Eq. (4.1) and Eq. (4.2). The efficiency of the laser diode regarding
Input volt. (V) Output volt. (V) Volt. drop (V) Efficiency [%] 1 0.3 0.7 70.0 3 2.3 0.7 23.3 5 4.3 0.7 14.0 7 6.3 0.7 10.0 10 9.3 0.7 7.0 15 14.3 0.7 4.6 30 29.3 0.7 2.3
Table 4.1:Table over diode output voltage dependence on input voltage
voltage was calculated by dividing the voltage drop by the input voltage, to see how much voltage was used efficiently by the laser diode.
V = R · I = (Rconnected+ RS) · I (4.1) Rconnected= V I −RS= 10 V 0.05 A−13.21Ω = 186.79 Ω (4.2)
4.2.1
Input Voltage
During the simulation the current was kept constant at 50 mA by varying the connected resistance, to match the voltage. E.g, at the 10 V input the resistance was set to 200Ω, according to:
R = V I =
10 V
50 mA = 200Ω (4.3)
Simulation results from a low, 5 V, and a high, 30 V, input voltage can be seen in Fig. A.1 and Fig. A.2.
The interesting result is the voltage drop over the laser diode and the effi-ciency of the laser diode at different voltages, given in percent. This can be seen in Tab. 4.1.
The corresponding graph can be seen in Fig. A.3
4.2.2
Input Current
During the simulation the voltage was kept constant at 10 V by keeping the circuit’s voltage source constant.
Simulation results from a low, 10 mA, and a high, 0.5 A, input current can be seen in Fig. A.4 and Fig. A.5.
Input curr. (mA) Output volt. (V) Volt. drop (V) Efficiency [%] 10 9.8 0.2 2 30 9.6 0.4 4 50 9.3 0.7 7 100 8.7 1.3 13 150 8.0 2.0 20 250 6.7 3.3 33 500 3.4 6.6 66
Table 4.2:Table over diode output voltage dependence on input current.
The interesting result is the voltage drop over the laser diode and the effi-ciency of the laser diode at different voltages, given in percent. This can be seen in Tab. 4.2.
The corresponding graph can be seen in Fig. A.6
4.3
SRD
Due to the fact that the SRD has the ability to generate a narrow pulse from a sinusoidal input, the SRD is fed with such a signal. The period of the signal is 50 ns, according to the desired pulse, due to the fact that the SRD is able to generate one pulse per period of the sinusoidal input signal. The SRD simulation was carried out using a circuit schematic shown in Fig. 4.2. The supporting devices in the schematic are taken from the analyse made in the article by Zhou et al. making the circuit able to analyse in a sim-ilar matter, creating a simsim-ilar environment for the device. The inductance creates a low-inductance path to ground for low-frequency signals, keeping the output node drained. When the SRD fires, a very high-frequency pulse is created. This signal sees the inductance as a high-impedance path, caus-ing a sharp rise in the output node voltage. The output resistance dampen the circuit and the capacitance is used as a DC blocker.
The diode in the equivalent circuit was modeled by the following SPICE model:
.MODEL DMOD D
+ IS=2.3e-15 CJO=0.89e-12 VJ=0.4 FC=0.5 + BV=42.6 M=0.2 RS=0.8 IBV=10e-6
The impact of the junction capacitance, parasitic package capacitance and parasitic package inductance of the SRD was simulated and analyzed. For all simulations of the SRD the input voltage sinusoidal wave is much larger than the output signal. The input signal might look a bit strange because it
Figure 4.2:Simulation circuit of SRD.
Junction cap. (pF) Output voltage (V) Efficiency [%]
0.30 3.40 34 0.89 4.10 41 2.00 4.60 46 3.50 4.85 49 5.00 5.00 50 10.00 5.20 52
Table 4.3: Table over SRD output voltage dependence on junction capaci-tance.
is zoomed in, this is done to be able to see the output better. The efficiency of the SRD regarding voltage was calculated by dividing the output voltage with the input voltage.
4.3.1
Junction Capacitance,
C
j0The junction capacitance was simulated to see its dependence on output voltage and pulse width.
A low, 0.3 pF, and a high, 5 pF, junction capacitance and their respective results can be seen in Fig. A.7 and Fig. A.8.
Output Voltage
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given in percent. This can be seen in Tab. 4.3. The corresponding graph can be seen in Fig. A.9
Junction cap. (pF) Pulse width (ps) 0.30 150 0.89 300 2.00 500 3.50 600 5.00 700 10.00 1000
Table 4.4: Table over SRD output pulse width dependence on junction ca-pacitance.
Package cap. (pF) Voltage output (V) Efficiency [%]
0.01 4.00 40.0 0.10 4.05 40.5 0.37 4.11 41.1 0.50 4.12 41.2 0.70 4.20 42.0 1.00 4.25 42.5
Table 4.5: Table over SRD output voltage dependence on package capaci-tance.
Pulse Width
The junction capacitance’s impact on the pulse width can be seen in Tab. 4.4. The corresponding graph can be seen in Fig. A.10
4.3.2
Package Capacitance,
C
pThe package capacitance was simulated to see its dependence on output voltage and pulse width.
A low, 0.1 pF, and a high, 1 pF, package capacitance and their respective results can be seen in Fig. A.11 and Fig. A.12.
Output Voltage
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given in percent. This can be seen in Tab. 4.5. The corresponding graph can be seen in Fig. A.13
Package cap. (pF) Pulse width (ps) 0.10 140 0.37 150 0.70 175 1.00 190 1.50 205 2.00 215
Table 4.6:Table over SRD pulse width dependence on package capacitance.
Pulse Width
The impact of the the package capacitance on the pulse width can be seen in Tab. 4.6. The corresponding graph can be seen in Fig. A.14.
4.3.3
Package Inductance,
L
pThe package inductance was simulated to see its dependence on output volt-age and pulse width.
A low, 0.5 nH, a middle, 3 nH and a high, 10 nH, package inductance and their respective results can be seen in Fig. A.15, Fig. A.16 and Fig. A.17. To be clearer with the difference of these signals, the input signal has been cut from the simulation figures. The input signals for these cases are similar as in previous SRD simulations.
Output Voltage
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given in percent. This can be seen in Tab. 4.7. The corresponding graph can be seen in Fig. A.18.
Pulse Width
The impact of the package inductance on the pulse width can be seen in Tab. 4.8. The corresponding graph can be seen in Fig. A.19.
Package ind. (nH) Voltage output (V) Efficiency [%] 0.5 3.72 37 1.5 4.12 41 3.0 4.29 43 5.0 3.97 40 7.0 3.62 36 10.0 3.18 32
Table 4.7: Table over SRD output voltage dependence on package induc-tance.
Package ind. (nH) Pulse width (ps)
0.5 295 1.5 300 3.0 320 5.0 365 7.0 410 10.0 460
Table 4.8:Table over SRD pulse width dependence on package inductance.
4.4
MESFET
The MESFET model used for implementation is taken from the Miao and Nguyen [2003]. The supporting devices in the schematic are taken from the analysis in the article making the circuit able to analyse in a similar matter, by creating a similar environment for the device. The capacitance is used as a DC blocker, the resistance dampens the circuit and the inductance and capacitance form the DC bias current.
Simulations of the MESFET was carried out using a circuit schematic shown in Fig. 4.3.
The impact of transconductance, threshold voltage and drain current for the MESFET was simulated and analyzed. The efficiency of the MESFET regarding voltage was calculated by dividing the output voltage with the input voltage.
The input to the MESFET was a pulse with properties approximated from the SRD output according to Miao and Nguyen [2003]. These properties can be seen in Tab. 4.9.
Figure 4.3:Simulation circuit of MESFET. Rise and fall time 10 ps Pulse duration 200 ps Initial voltage 0 V Pulse voltage -4 V
Table 4.9:Table of MESFET input pulse properties.
4.4.1
Transconductance
The transconductance was simulated to see its dependence on output volt-age.
A low, 5 mS, and a high, 100 mS, transconductance and their respective results can be seen in Fig. A.21 and Fig. A.22.
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given by percent. This can be seen in Tab. 4.10.
The corresponding graph can be seen in Fig. A.23.
4.4.2
Threshold Voltage
The threshold voltage was simulated to see its dependence on output volt-age.
A low negative, -0.5 V, and a high negative, -10 V, threshold voltage and their respective results can be seen in Fig. A.24 and Fig. A.25.
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given by percent. This can be seen in Tab. 4.11.
Transcond. (mS) Output volt. (V) Efficiency [%] 0.1 0 0 5.0 0.7 19 10.0 1.5 38 25.0 3.7 93 50.0 7.5 188 75.0 11.2 280 100.0 15.0 375
Table 4.10:Table over MESFET output voltage dependence on transconduc-tance.
Threshold volt. (V) Output volt. (V) Efficiency [%]
0 0 0 -0.5 0.15 4 -1.2 0.78 20 -3.0 3.50 88 -4.0 5.40 135 -5.0 6.90 173 -7.0 8.20 205 -10.0 9.00 225
Table 4.11: Table over MESFET output voltage dependence on threshold voltage.
The corresponding graph can be seen in Fig. A.26.
4.4.3
Drain Current
A low, 10 mA, and a high, 150 mA, drain current and their respective results can be seen in Fig. A.27 and Fig. A.28.
Output Voltage
The interesting result is, besides the actual output voltage, the efficiency at different voltages, given by percent. This can be seen in Tab. 4.12.
Drain current (mA) Output volt. (V) Efficiency [%] 5 37.5 938 10 19.8 495 30 6.9 173 50 4.2 105 100 2.1 53 150 1.4 35
Table 4.12:Table over MESFET output voltage dependence on drain current.
Drain current (mA) Pulse width (ps)
5 21 10 40 30 110 50 180 100 219 150 219
Table 4.13:Table over MESFET pulse width dependence on drain current.
Pulse Width
The impact of the drain current on the pulse width can be seen in Tab. 4.13. The corresponding graph can be seen in Fig. A.30.
5
Discussion of Implementation
This chapter discusses the results from the previous chapter.
5.1
Laser Diode
According to the simulations a higher input voltage does not increase the voltage drop over the laser diode. This contributes to the fact that the ef-ficiency decreases significantly with higher voltage applied. This could be expected due to the fact that the voltage drop is dependent on the band gap of the pn-junction, which can be seen as constant. The input voltage should be kept low due to efficiency reasons, and due to the fact that higher voltage consumes more power.
The simulations show that a higher current gives a higher efficiency of the laser diode, creating a higher voltage drop over the laser. The required pulse of 100 ps with a period of 50 ns has a duty cycle of 0.2 %, implying that the laser diode can be overdriven without damage. The simulations show an equal result. For example, while applying a 0.5 A current to the laser diode model, using a 6.79 Ω resistance in series with the 13.21 Ω laser diode, the laser is overdriven. This seems to only increase the laser diode efficiency, which reaches 66 %.
A higher input voltage makes it possible to have a higher input current, according to Ohm’s law, due to the fact that the resistance is limited by the series resistance of the laser diode.
Combining that a higher voltage result in lower efficiency and high energy consumption with the fact that a higher voltage leads to possibly higher input current makes it worth using a bit higher input voltage. Looking at
the laser diode the maximum forward voltage and current during normal operation are VF=5 V and IF=30 mA. With a 5 V input voltage the laser
diode can be overdriven with a input current of 250 mA using a 6.79Ω resistance in series with the laser diode. The input voltage is chosen to be 5 V, with an input current of 250 mA. During normal operation the output power is 1.25 W, which is much higher than the laser diode’s maximum power of 117 mW.
At an applied input voltage of 5 V, a pulse width of 100 ps and a pulse period of 50 ns the average power is calculated using Eq. (3.2). For our pulsed laser diode the average power is 2.5 mW, according to Eq. (5.1). This is far below the maximum value.
Paverage=
5 · 0.25 · 100 ps
50 ns = 2.5 mW (5.1)
5.2
SRD
According to the results the SRD produces a narrow negative pulse from the sinewave input.
The simulations shows that a higher junction capacitance gives higher ef-ficiency to the SRD. This is not what the study, Miao and Nguyen [2003], says, where a lower junction capacitance gave a higher amplitude of the generated pulse. A reason for this may be that the higher junction capaci-tance makes the SRD able to accumulate more charge. Due to the fact that a more charged capacitance generates a higher amplitude of the output pulse. Also, the SRDs used are not the same and the models are not equal, which may impact the results. The simulations also show that a lower junction capacitance makes the pulse width shorter, as stated in the study, Miao and Nguyen [2003]. The junction capacitance can not get infinitely small, so as stated in the theory the SRD might not get much smaller than the given 150 ps. The higher junction capacitance produce more oscillations on the output pulse, seen in the simulations. These are highly undesired and in addition to the shorter pulse widths the choice for the SRD is to have low junction capacitance.
The package capacitance is barely giving a dependency on the output volt-age. The simulations show a slightly higher efficiency using a higher capac-itance. This might be due to the fact that the SRD accumulates more charge before generating the pulse. Using a lower package capacitance generates a pulse with shorter pulse width. A reason for this may be that the discharge can be made quicker due to less accumulated charge. The capacitance does not seem to have any noticable affect on the oscillations. The output voltage is barely affected, and due to the fact that a smaller pulse width is desired, the choice for the SRD fall on a low package capacitance.
The package inductance shows a different appearance where a lower effi-ciency is achieved for both high and low inductances. A highest effieffi-ciency is found at around 3 pF. Using a lower package inductance generates a pulse with shorter pulse width. Higher package inductance generates more oscil-lations to the output pulse. On similar grounds as for the junction capaci-tance, the choice for the SRD falls on a low package inductance.
5.3
MESFET
According to the simulations a higher transconductance gives a higher effi-ciency of the MESFET. The effieffi-ciency is approximately linearly dependent on the transconductance, which according to Eq. (3.4) is to be expected. The choice for the MESFET falls on having a high transconductance.
A larger negative threshold voltage gives a more effective MESFET. The choice for the MESFET falls on having a large negative threshold voltage. The MESFET device is highly dependent on the drain current added to it, with a lower current making the device more efficient. A lower current also makes the pulse width smaller. The drawback is that a lower current also creates a negative pulse which has to be filtered out using another cir-cuit component. This could be done using another MESFET which is only switched on during the positive pulse. Due to the fact that the negative pulse can be filtered out using another component the choice for the MES-FET falls on a lower current.
5.4
Summary
As stated in Chapter 2, the requirements from the fluorescence microscope on the laser diode is that the pulse width is around 100 ps with a pulse period of 50 ns. These pulses should increase the performance of the fluo-rescence microscope.
According to the laser diode the input voltage should be around 5 V and the input current around 250 mA. The average power going through the laser diode under these conditions is around 2.5 mW. The laser diode should not suffer any damage from this. In contrary this should increase the lifetime of the laser and lower the energy used.
The simulated SRD can produce a narrow pulse from a sinusoidal wave. To produce a pulse as close to the desired pulse as possible a SRD with low junction capacitance, low package capacitance and low package inductance should be used. The pulses can get as narrow as 150 ps with an amplitude of 3.4 V. This is close to the desired pulse, but it needs some additional amplification and narrowing.
Input/Component parameter Preferred property
Diode input voltage Low
Diode input current High
SRD junction capacitance Low SRD package capacitance Low SRD package inductance Low MESFET transconductance High MESFET negative threshold voltage High
MESFET drain current Low
Table 5.1:Table over preferred properties of inputs and component param-eters.
The simulated MESFET proved to be able to both amplify and narrow the pulse. To produce an high voltage and narrow pulse a MESFET with high transconductance and a large negative transconductance should be used, fed with low drain current. The simulation results show examples on pulses as low as 40 ps with an amplitude of almost 20 V, for the MESFET with an drain current of 10 mA. The desired pulse should be able to achieve by mul-tiple variations of transconductance, threshold voltage and input current. The input current has big impact on the MESFET and is easy to change in a circuit, compared to changing parameters of a circuit component. This gives great trust in that the desired pulse should be possible to generate using available MESFETs. The negative pulse generated by the MESFET should be able to filter out and should not be a problem for the solution. An additional current source might be needed due to the fact that the MES-FET needs much lower current than the laser diode.
Connecting the SRD and the MESFET should create the desired pulse in the matter of pulse width, pulse period and voltage. It remains to connect these, filter out the unwanted pulse from the MESFET and to match the current with the desired value from the laser diode. For the circuit to work as intended, there has to be additional analyses regarding the connection of components, especially when dealing with these high-frequency pulses. There also has to be considerations taken for the actual circuit, with trans-mission line effect etc. Also, the MESFET has to be analysed specifically to take in to consideration the effect of parasitics, which will probably have great effect on these high-frequency pulses.
5.4.1
Future Work
To continue the work that has been done in this thesis, the complete circuit could be simulated and tested. A filter should be added together with a current source. The next step would be to produce the circuit for further testing, and finally to put the circuit, including the laser diode, as light source in a fluorescence microscope.
6
Conclusions
According to this study the desired light source should be a laser diode. The waveform provided to the fluorescence microscope should have a wave-length corresponding to the used fluorophore and be fast switching with a stable output power. Distortions of the laser pulse should be decreased or removed in order for the fluorescence microscope to work properly. To create these laser pulses, the electronic pulse provided to the laser diode should be 5 V and 250 mA with a pulse width of 100 ps and a pulse period of 50 ns. This pulse can be achieved by the chosen laser diode by using an oscillator together with an SRD as pulse generator and a MESFET for amplifying and narrowing the pulse.
The SRD used should have low junction capacitance, low package capaci-tance and low package induccapaci-tance. The MESFET used should have high transconductance, large negative threshold voltage and a low drain current. The circuit is in need of filtering to remove the unwanted negative pulse generated from the MESFET. An additional current source might be used to raise the current before the pulse enters the laser diode.
A
Simulation graphs
Figure A.1:Diode simulation with input of 5 V.
Figure A.2:Diode simulation with input of 30 V.
Figure A.4:Diode simulation with input of 10 mA.
Figure A.6:Graph over diode output voltage dependence on input current.
Figure A.8:SRD simulation with junction capacitance 5 pF.
Figure A.9:Graph over SRD output voltage dependence on junction capaci-tance.
Figure A.10: Graph over SRD output pulse width dependence on junction capacitance.
