Radiation effects on wide bandgap semiconductor devices

IN

DEGREE PROJECT ENGINEERING PHYSICS, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020,

Radiation effects on wide bandgap semiconductor devices

For possible applications in biomedical instruments

PATRIC ELF

KTH ROYAL INSTITUTE OF TECHNOLOGY

Radiation effects on wide bandgap semiconductor devices

For possible applications in biomedical instruments

Patric Elf

SK202X Degree Project in Applied Physics, Second Cycle

Department of Applied Physics Royal Institute of Technology (KTH)

September 2020

Abstract Svenska:

Gallium Nitrid (GaN) baserade high electron mobility transistors (HEMTs) anv¨ands inom m˚anga olika omr˚aden, s˚asom 5G, bil-industrin, flyg/rymd och i sensorer f¨or kemiska, mekaniska, biologiska och optiska applikationer. Tack vare dess goda materialegenskaper s˚a ¨ar GaN baserade HEMTs s¨arskilt anv¨andbara i h˚arda milj¨oer, som till exempel i f¨orbr¨anningsmotorer, avgaser, i rymden, samt till medicinska instrument d¨ar p˚alitlighet och t˚alighet ¨ar efterstr¨avat. I det h¨ar examensarbetet s˚a unders¨oks effekten av protonbestr˚alning p˚a GaN HEMTs samt m¨ojligheten till anv¨andning av dem inom biomedicin och diagnostik. Ar- betet ¨ar uppdelat i tv˚a delar: den ena behandlar den teoretiska bakgrunden av GaN HEMTs och den andra presenterar de experiment/simuleringar som utf¨orts f¨or att se effekterna p˚a komponenterna f¨ore och efter protonbestr˚alning.

I bakgrunds-sektionen s˚a beskrivs hur HEMTs fungerar, tillverkningstekniker och mekanismerna f¨or hur defekter uppkommer under olika former av protonbe- str˚alning. D¨arefter s˚a karakt¨ariseras HEMT komponenterna och relaterade test- strukturer f¨ore och efter protonbestr˚alning, med ett fokus p˚a doser mellan 10¹¹ to 10¹⁵ protoner/cm², samt en j¨amf¨orelse med resultat som f˚atts fr˚an simu- leringar med SRIM/TRIM-program. Ut¨over detta s˚a beskrivs och diskuteras

¨

aven biokompatibiliteten och applikationer inom biomedicin av GaN komponen- ter vid protonbestr˚alnings-scenarion i arbetet.

English:

Gallium nitride (GaN) based high-electron-mobility transistors (HEMTs) are used in a wide variety of areas, such as 5G, automotive, aeronautics/astronautics and sensing fields ranging from chemical, mechanical, biological to optical ap- plications. Owing superior material properties, the GaN based HEMTs are es- pecially useful in harsh operation environments e.g. in the combustion engine, exhaust, space, and medical instruments where the reliability and resilience are highly demanded. In this thesis the effect of proton irradiation on the GaN HEMTs as well as the possible incorporation of them in biomedicine and diag- nostics are investigated. The thesis includes mainly two parts: one is on theoretic background of GaN HEMTs, and another presents the experiment/simulation de- tails of the devices before and after proton radiation. In the background section, the HEMTs function, manufacture technique and defect formation mechanism in the device under different proton radiation conditions are introduced. Then, the characterizations of the HEMT devices and related test structures before and after the proton radiation with dose range from 10¹¹ to 10¹⁵ protons/cm² are emphasized, as well as the comparison with simulation results obtained using SRIM/TRIM program. In addition, the biocompatibility of GaN devices and their biomedicine applications in proton radiation scenarios are also described and discussed in this thesis.

Acknowledgements I want to thank...

Qin Wang for all the support and expertise in regards to the project, the fika meetings and progress report meetings where we always got great feedback and encouragement, I am truly grateful.

Anders Hall´en for the help with the irradiation at Uppsala, discussions regarding radiation doses and effects, without your expertise this project would not have been possible.

RISE and everyone working there, for letting me do my master thesis at the company and the use of their equipment and materials.

˚Angstr¨om Laboratory for the use of their facilities, and the expertise of the people there.

Somewhere, something incredible is waiting to be known.

- Carl Sagan

Contents

1 Introduction 6

1.1 Scope and Objective . . . . 6

1.2 Gallium Nitride . . . . 7

1.3 HEMT & 2DEG . . . . 8

1.4 Production of GaN HEMTs . . . . 10

1.5 Radiation damage in GaN-based devices and HEMTs . . . . 11

1.6 Radiation in biomedicine & Biocompatability of GaN . . . . 13

2 Materials & Method 14 2.1 Test structures . . . . 14

2.1.1 Transmission Line Measurements . . . . 15

2.1.2 Circular Transmission Length Measurements . . . . 16

2.2 Device setup & Characteristics . . . . 16

2.3 DC measurements . . . . 17

2.4 Irradiation . . . . 17

2.4.1 Shielding . . . . 18

2.4.2 Irradiation process & accelerator . . . . 18

2.5 Calculated damage events . . . . 20

3 Results & Analysis 23 3.1 TLM . . . . 23

3.2 Effects of radiation . . . . 24

4 Summary 26

1 Introduction

There is a constant demand when it comes to electronics for smaller, faster and more durable components. This of course requires innovation in material sciences for materi- als and components which are up to the task and tailored for these different challenges.

One material that shows promise and has proven itself is Gallium Nitride (GaN), which due to its wide bandgap and robustness make for an ideal semiconductor device as an high electron mobility transistor (HEMT).

These semiconductor devices made out of GaN can be of use in a variety of dif- ferent fields, such as highly sensitive biosensors to detect different analytes and their concentrations[1] or in satellite and space applications due to its low sensitivity to ra- diation damage.[2] This makes for an interesting argument, that it can be suitable in sensors and equipment used for people undergoing radiation therapy, and in treating and sensing cancerous cells. In this project however, the main focus is on the radiation effects on the GaN devices.

1.1 Scope and Objective

The goal of this project is to investigate the effects of radiation on wide bandgap semiconductor devices, more particularly, the effects of proton irradiation on GaN HEMTs, and the possible applications of GaN HEMTs in life-science. This is done partly through a literature survey to familiarize with the subject as well as to find reference material of what has previously been done and other useful applications, planning and performing irradiation of GaN devices, characterization and comparison between different doses of radiation, as well as analysis of the results. Hopefully this will help us better understand how radiation affects the GaN devices, and show us that they have a place in radiation therapy and life-science.

The literature study consisted of researching the topic, finding relevant studies which deals with similar subjects, as well as act as reference material to what results that could be expected, which would be the baseline for the project.

The planning and performing of the irradiation of the devices is the main part of the project, where the focus is the effect of different dosages. This was done using multiple chips with HEMT devices from a wafer which was provided by RISE. The characterization, comparison and analysis was done at RISE in Kista, while the actual irradiation was done at the ˚Angstr¨om Laboratory in Uppsala.

1.2 Gallium Nitride

Gallium Nitride or GaN is a wide bandgap semiconductor material which has gained a lot of interest due to its applications in high-power and high-frequency electronic devices. GaN has a direct bandgap of 3.4eV, which with the introduction of Aluminium to form the alloy Al_xGa_1−xN , can be tailored from 3.4eV (x = 0) to 6.2eV (x=1). [3]

The AlGaN will also have a a direct bandgap, which gives beneficial properties when it comes to optoelectronics compared to semiconductor materials with a indirect bandgap, which can be the case in other comparable materials such as Silicon Carbide, SiC. GaN has a Wurtzie crystalline structure and is both very hard, high chemical stability, has a high thermal conductivity, and has shown resilience to radiation damage, which thus makes it ideal for hazardous environments, such as for space applications.[4][5]

A comparison between GaN and comparable semiconductor materials can be seen in table 1.

Unit \ Material GaN Si GaAs AlN

Band gap [eV] 3.4 1.2 1.42 6.2

Electron Mobility [cm²/(V · s)] 1400 1440 9400 450 Thermal conductivity [W/(m · K)] 130 149 56 285

Permitivity [F/m] 8.9 11.7 12.9 8.5

Table 1: Comparison of properties between semiconductor materials used in HEMTs.[6]

GaN is also has a high biocompatibility due to its chemical stability, it only leaches negligable amounts even in the presence of hydrogen peroxide, as well as not hinder-

ing cell growth, and promoting cell adhesion to silicon surfaces.[14] This makes it an excellent candidate for use in vivo biosensors.

1.3 HEMT & 2DEG

A 2-dimensional electron gas (2DEG) is a model of for the motion of electrons when they are confined in one axis and free to move in the other two, which occurs in HEMT devices. This is due to the heterojunction between the two semiconductor materials that make up the HEMT-structure, which occurs due to the difference in their bandgaps when they are in contact with each other. Away from the junction, the semiconductor materials must be electrically neutral in composition, but the two Fermi levels for the different semiconductor materials must be the same for there to no net electron transport (in the absence of a external voltage bias). For this to be possible, the bands need to bend near the interface, which it does due to space charges consequent to the transfer of electrons between the two semiconductor materials, which thus causes the heterojunction.[7]

When accounting for this, as in the cause of a AlGaN-GaN HEMT, what will happen, since the fermi level of AlGaN is higher than that of GaN, the electrons will amass from the AlGaN to the GaN, until the Fermi levels will even out, as seen in figure 1.[8]

The electrons are thus confined in the axis between the two semiconductor materials, but free to move in the other two, which is the main principle that makes a HEMT.[8]

The HEMT is also known as a heterostructure field effect transistor, or HFET, which as the name implies is due to the heterojunction between the two semiconductor materials which creates the 2DEG, as can be seen in figure 2. It uses three terminals, source, gate and drain. These work by applying a voltage to the gate, which then affects the conductivity, or flow of electrons, between the source and the drain terminals.

For AlGaN-GaN HEMTs, the AlGaN- and GaN-layers are usually quite thin, only

≈ 10 − 100 nanometers, which means that the 2DEG is close to the surface making it

Figure 1: Schematic of the energy bands for AlGaN-GaN which causes a 2DEG. The flow of electrons from AlGaN to GaN, causing a 2DEG, and shift in the energy bands, where Ec is the conduction band energy, Ef is the fermi level, and Ev is the valence band energy. The yellow region is where the electrons have amassed, thus forming the region of the 2DEG.

quite susceptible to surface charge changes.

The 2DEG, and thus the HEMT as a whole is just as other electronic devices affected by outside magnetic fields and temperature, which makes it important to have materials that mitigate this, dependant on the environment the HEMT is expected to perform in.

There are many possible variants for the substrate in HEMTs, as this is the medium the semiconductor materials are grown upon. It can often be used as a fourth terminal, also known as the body or bulk which serves as a bias for the transistor. It can however, be made of an inert material, and thus not affecting the conductivity or electron mobility as well.

Figure 2: Schematic of HEMT with 2DEG layer marked between AlGaN and GaN layers.

1.4 Production of GaN HEMTs

GaN HEMTs can be manufactured using epitaxial growth in which a crystalline layer is deposited upon a substrate. Of the typical substrates used, the most common is Sapphire, Si and SiC, which all have their merits, where Sapphire and Si are cheaper options while SiC have a higher thermal conductivity but is much more expensive.

Sapphire and Si can also cause issues while annealing, as parasitic conductance can occur from the substrate, which may damage the HEMT structure. GaN can itself also be used as a substrate in bulk form.[9]

To produce a GaN HEMT layer on top of a substrate, molecular beam epitaxy (MBE) or metalorganic vapour phase epitaxy (MOVPE) is commonly used, in which a GaN layer is deposited or grown on top of a substrate, and then the AlGaN layer on top of the GaN.

For the case of MBE, it can be done at lower temperatures than MOVPE and is done in a vacuum. The benefit of this is the growth rate, which can be 3000 nm/h, and also that it can achieve a great degree of precision in the deposition, as well as a low degree of impurities.

The MOVPE however is done at low pressure (less than 1 atm), and at a higher temperature, typically over 1000^◦C for GaN. Precursor gases are injected into a reactor with a non-reactive carrier gas, where they create reactants and possible byproducts,

after which the reactants are then transported to the substrate surface where they react forming the GaN layer. After the GaN layer has formed, Al can be introduced to the precursor creating a AlGaN layer on top of the GaN, thus forming the HEMT structure.[10]

It is important to note that these techniques can also be used to deposit additional

”substrate” or isolation layers between a substrate and the HEMT structure, in order to make better or more high quality HEMT devices from cheaper materials. In any case, this deposition is the most important part of the HEMT, as it is in this heterojunction between the GaN and AlGaN which forms the 2DEG.

The contacts, meaning the source, gate and drain then need to be formed. The source and drain uses ohmic contacts, while the gate uses a schottky contact. These can be made using different types of lithography, and is important to be of high quality to avoid leakage. The length of the gate is also important to keep track of, as this de- termines the threshold voltage, which may be important to have a good high frequency performance.[11]

Isolation between different components is made via mesa isolation, meaning it’s the 2DEG of different component isn’t in contact with one another. This can be obtained through etching. This is important to protect different components, since if one were to malfunction, not all of them need to be destroyed.

On top of everything, there may be a passivation layer. This is to protect and shield the material via a microcoating.

1.5 Radiation damage in GaN-based devices and HEMTs

GaN components are known to have a high resilience to radiation, this is due to the high binding energy between the bonds which means it takes a lot of energy to break them, and thus causing damage. This energy to cause damage comes from charged particles interacting with the GaN and transfering its energy. The two main ways of these interactions is that of ionization and nuclear collisions causing displacement.

Ionization occurs when the energy is transferred to the GaN molecule causing an electron from the valence band to jump to the conduction band and creating a hole in the valence band. For this to happen, the energy required is that of the bandgap.

This means that radiation hardness is also improved by larger bandgaps, as this leads to less ionization.

Ionization is primary way that protons lose energy when travelling through a mate- rial. While travelling through a material, the stopping power for a high energy charged particle increases and reaches it’s peak right before it stops. This stopping power over path length is called a braggs curve, and the peak is called the braggs peak which is caused since the probability of interactions of a charged particle increases as the en- ergy of the charged particle decreases.[12] These interactions are, as mentioned, mainly ionization events, as they are more likely to occur, which means that the braggs curve mainly account for the ionization part. The energy that the charged particle loses is inversely proportional to the velocity squared which is what causes this peak. After this peak the dependence for low energy charged particles, as the particle has lost most of its energy, is proportional to the velocity, causing it to decrease abruptly. This phe- nomenon is used in radiation therapy, and also explains the amount of damage events in a material that one can expect, for a given depth into the material.

Displacements are irregularities in the crystal structure, which occur due to the nuclear collisions when irradiating the GaN. These nucelar collision causes the bonds to break in the GaN, knocking them out of their place causing a vacancy in the crystal lattice as well as a interstitial atom in the structure. The energy required to cause such a vacancy and knock an atom out of position is 22eV for Ga and 25eV for N.[5]

The effects of radiation on GaN HEMTs can occur in different ways, such as the degradation of the saturation current, sheet resistance and transconductance, as well as a shift in the threshold voltage, which often is because of displacements. It has been reported that a proton dose of 2 · 10¹⁴ protons/cm² can cause a degradation in the saturation current of 12.5% when using 5 MeV protons.[5] Similar effects can be seen with irradiation of Neutrons as well, and for low doses (10¹¹neutrons/cm²), GaN-

devices appear to self-annealing and can achieve full recovery at room temperature.[13]

1.6 Radiation in biomedicine & Biocompatability of GaN

When considering materials for in vivo and in vitro applications of biological materi- als, additional consideration has to be taken in order to ensure that they don’t interfer with the biological processes, or are toxic to a host organism. GaN exhibits excel- lent such qualities at a cellular level as they do not appear to hinder cell growth, due to its chemical stability and hardiness only leak negligible amounts in aqueous solu- tions, and low amounts even in the presence of hydrogen peroxide.[14] Gallium oxide dust which can appear during the manufacturing process of GaN-based HEMTs using Metalorganic vapor-phase epitaxy (MOVPE) can be an irritant for the skin, eyes and respiratory system, this however does not pose a problem for already made GaN based components.[15]

Radiation exposure in biomedicine and research can vary, by a wide margin, as for human exposure in medical procedures it can range from 0.1 mSv in a chest x-ray, to 50,000 mSv in the case of destroying a tumor in cancer treatment.[16] A benefit when using protons therapy when it comes to living beings is that due to the Braggs peak, its dose outside the desired region can be significantly lower for deep-tissue.[17][18]

To reach these depths that is often required to reach tumors, high energy protons are required as the penetration depth is energy dependant.

As for radiation in research it can vary from the use of radioactive trace substances which are implemented to track biological processes, to imaging with the commonly used electron microscope. As for protons they can also be used for imaging purposes, where a proton microscope can get a high resolution with fast imaging.[19] It can however be destructive, if done improperly.

It has been shown that GaN-based devices can be used as biosensors. By function- alizing the surface of the device by attaching antibodies or functional groups to the surface of the gate, which then will affect the conductivity once they interact with the

specific compound. This is since the HEMT layer is close to the surface and thin, it will affect the fermi-level of the AlGaN. it has also been shown to be useful as a pH detector.[20]

Figure 3: Multiple chips marked with the radiation dosage which they were exposed to and in which region they were exposed.

Due to the biocompatibility of GaN, where the survival rate of cells is higher on GaN sur- faces than that of silicon, these types of of sen- sors can be used with living cells, and are of great interest in diagnostics. GaN-based sensors also appear to have a low noise-to-signal ratio, which means that even low trace amounts of a toxin might be detectable.[21]

2 Materials & Method

A large variety of HEMTs was available for test- ing, with high quality ones made with SiC sub- strates as well as ones Sapphire substrates. For the sake of simplicity the focus fell on one par- ticular design with a Sapphire wafer substrate that was available with multiple chip fragments, which can be seen in figure 3, where the different doses different sections were exposed to has been marked.

2.1 Test structures

Test structures are implemented on the wafers in order to help characterize the wafers and its internal components, and thus calculate things such as internal resistance or conductivity. Two common test structures are the structures for Transmission line

Measurements and Circular Transmission Length Measurements.

The resistance is calculated, as usual, as the voltage divided by the current:

R = V

I (1)

This is important to remember, when calculating for different types of transmission line measurements.

2.1.1 Transmission Line Measurements

Transmission line Measurements or TLM is used to determine contact resistance in semiconductors, by in turn measure the resistance at different set distances of the semiconductor.[22] We then get contributions to the total resistance, RT, from the metal plates which the electrodes attach to, R_M as well as the contact resistance, R_C and the actual resistance in the semiconductor, RSemi. We thus get:

R_T = 2R_M + 2R_C + R_Semi (2)

Because the resistance in the metal used is in general significantly lower than the contact resistance, R_M << R_C, this can be simplified to:

R_T = 2R_C + R_Semi (3)

As contribution to the resistance of the semiconductor is dependant on the length (L) and inversely to the width (W), we get the following expression:

R_Semi = R_S L

W (4)

We can thus write the total resistance as:

R_T = 2R_M + 2R_C+ R_S L

W (5)

By taking measurements for different lengths, we can extrapolate the resistance for a zero-length resistor, which thus becomes just 2R_C. An issue that we get however is that

of current crowding, which causes the current to move not straight from one contact to the other, which thus affects the distance and measured resistance of the semiconductor.

One way of overcoming this hurdle is with Circular Transmission Length Measurements.

2.1.2 Circular Transmission Length Measurements

Circular Transmission length measurements are similarly to transmission line measure- ments taken at different lengths in order to determine the R_C-value, it does however implement, as the name implies, a circular structure, which overcomes the issue with current crowding, as the current will travel from a ring to another ring of different diameter, and as such no additional isolation for the test structure is required.

Simplified, we get the approximate formulas for the contact resistance ρ_cand total resistance R_{T ,cir}:

ρc = RshL²_t (6)

R_{T ,cir} = R_sh 2πrl

(2L_t) (7)

where R_sh is the sheet resistance, L_t is the transfer length and r_I is the inner ring radius.[23]

2.2 Device setup & Characteristics

The HEMTs used were GaN-based, which had been manufactured on a wafer with a Sapphire substrate, and then broken down into chip fragments. On these chip fragments, there were blocks consisting of 18 different transistors, with varying drain length, gate length and isolation inside/outside the structure. The lengths of the drain was 1.5, 2.0 and 2.5 micrometer, and the lengths of the gate was 0.5, 1.0 and 1.5. With all possible combinations of these in one block, with the two pos- sible configurations of isolation, thus yielding; 3 · 3 · 2 = 18 different combination on each block. Each chip then contained different amounts of blocks, depending

on how they were cut, as can be seen in figure 3. The cross-section of the chips however, was first a Sapphire substrate, after which a 1.2 micron thick GaN (C- comped) which caused it to have a high resistance, thus not contributing to the 2DEG.

Figure 4: Example of tran- sistor configuration.

On top of that there was a 90nm thick undoped GaN channel, and on top of that a 20nm thick undoped AlxGaN1−x layer, where x = 0.25. The passivisa- tion/isolation layer on the transistors was either inside or outside the structure. The setup was somewhat basic, or standard for the transistors of the sort, which can be seen in figure 4, they had however been used for exper- iments previously, which meant that some of them did suffer from some degradation and scratches.

2.3 DC measurements

DC measurements were performed, using a probe-station with 3-point pin-probes, with a Model 4200A-SCS Pa- rameter Analyzer using Clarius software to design record the the measurements. This setup was ideal, as it was easy to switch between different tests, as well as the pos- sibility to do multiple different tests in succession. A

quick-start guide for how to connect and run the simpler measurements was also made for the parameter analyzer.

2.4 Irradiation

The irradiation of the chips was done at the ˚Angstr¨om Laboratory, Uppsala, with a accelerator which was available there. 2 MeV protons were used, at different doses, ranging from 10¹¹to 10¹⁵protons/cm². The set energy of the protons, while varying the dosage was to make it easier to see potential effects from the dose, rather than having

the possibility of having the different energy protons or using other ions interacting with the material differently.

2.4.1 Shielding

As only certain parts of the chip was to be exposed to the irradiation, a simple shield out of aluminium-foil, with a section cut out corresponding to the area that was to be exposed was made. The projected range of protons was calculated using the Stopping Range of Ions in Matter-, SRIM-software,[24] as can be seen in table 2, thus giving an estimate how thick the shielding had to be. Aluminium has a density of 2.7g/cm³, and by using this and weighing a sheet of aluminium, the thickness of the sheet can simply be calculated as:

Z = weight

DensityAl· X · Y (8)

where Z is the thickness, weight is the weight measured by the scale, DensityAl is the density of aluminium, X and Y is the length and width of the sheet. It was concluded that the thickness was 14µm of the particular aluminium foil used, hence folding it 8 times, giving a thickness for the shield of 8 · 14µm = 112µm gave a good margin to be sure that the shielded parts remained free of irradiation.

2.4.2 Irradiation process & accelerator

The irradiation was done using the 5 MV 15SDH-2 Pelletron accelerator at the Tandem Laboratory, which is located in the ˚Angstr¨om Laboratory, Uppsala.[25] Beam line T5 was used, which is used for ion irradiation over an area. The chips were placed on discs, which were then loaded into a magazine and put inside the machine, which was then emptied of air to create a vacuum, which is required unless the ions would collide with the molecules in the air. With the magazine, the different discs with the chips could be loaded for irradiation without having to depressurize in between different runs with irradiation. The accelerator was set to a specific current, and then since we have the

Ion Energy [MeV] Projected Range [µm]

1.00 14.38

1.20 18.83

1.40 23.81

1.60 29.27

1.80 35.22

2.00 41.63

2.25 50.28

2.50 59.64

2.75 69.67

3.00 80.38

Table 2: Projected ranges for protons in aluminium with different energy.

charge of a proton, and a set dosage we want to achieve, the time to have the sample exposed to the beam can be calculated as:

t = dose · Area · C

I (9)

where t is the time in seconds, dose is the protons/cm², the area in cm², C is the charge of a proton (≈ 1.60 · 10⁻¹⁹[A · s]) and I is the current of mono-energetic protons produced by the accelerator. The protons were then accelerated and shot towards the samples, and a gate was opened, which was situated between the accelerator and the sample, for the time needed for the sample to reached the desired dose, after which, it was shut again, and the next sample loaded. It was also possible to set the area of which to scan over with the proton irradiation. In this case, a 5x5cm² area was irradiated in order to create a homogeneous dosage.

2.5 Calculated damage events

Included in the SRIM software previously mentioned, there is a program called Trans- port of of Ions in Matter, TRIM, which uses a Monte-Carlo simulation in order to calculate damage events, the ranges ions will reach, etc. As this is a statistical tool, it of course can’t tell us exactly what will happen, but with enough simulations we can get a picture of what to expect when exposing materials to radiation.

As can be seen in figure 5, statistically, nearly all the protons are expected to reach far down into the substrate for an energy of 2 MeV, and thus will not be close to the AlGaN-GaN 2DEG. We do however see that as it collides with the material it spreads out, which isn’t much of a concern, as there is an area and not a perfect one-line proton beam.

It does however still cause vacancies and damage events while travelling through the material, as can be seen in figure 6. We also see the importance of running simulations for long periods of time with many simulations for these types of statistical programs, as we see how the curve flattens out, nearing linear, which is to be expected in this region close to the surface while irradiating a sample.

(a)

(b)

Figure 5: Simulations of penetration depth of ions, with a setup similar to the chips

(a)

(b)

Figure 6: Simulations of damage events using SRIM for the interval 0-1100 ˚A depth of AlGaN-GaN HEMT structure, with AlGaN being 200˚A and GaN being 900˚A. (a) using 100000 ions, (b) using 1000000 ions. Notice the drop in damage events at 200˚A, due to change from AlGaN to GaN.

3 Results & Analysis

It became clear that lower proton irradiation levels didn’t show a noticeable effect, as well as due to the fact that the chips used, had been used previously in other works, they had some slight differences between them in response levels. Thus, the focus fell on the chip which had the highest irradiation doses, with doses of 10¹¹ and 10¹⁵ protons/cm², as well as an unirradiated section.

3.1 TLM

The zero-length resistance for the wafers was calculated to be 561.52 Ohm, as can be seen in figure 7. If we divide this by a two, as there are two probes attached, we get the contact resistance, 230.76 Ohm. This will also be a factor to consider while attaching the probes on the HEMT-devices. The resistance was calculated, using the average resistance over a C-V-sweep, from 3V to 6V, with a compliance of 0.2A.

Figure 7: Measured resistance at different distances using a TLM test structure.

Unfortunately, a TLM or cTLM test structure wasn’t available on the chip with the highest radiation doses.

3.2 Effects of radiation

It was difficult to see any results caused by the lower doses of protons, but we see a clear effect on the DC characteristics on the HEMT devices for larger doses. A decrease in the saturation current of the HEMTs, as can be seen in figure 8a, b and c, with a decrease to 0.928 for a dose of 10¹¹ protons/cm² and a decrease to 0.370 for a dose of 10¹⁵ protons/cm² of the original saturation current for a gate bias of 0. This is due to, as is suggested from the TRIM calculations, due to the vacancies created.

(a) (b)

(c)

(d)

Figure 8: Averaged transistor responses for (a) no radiation (b) for a radiation dosage of E+11 protons/cm² (c) for a radiation dosage of E+15 protons/cm² (d) Effects of the gate voltage

It is difficult to discern, but there appears to be a slight positive shift in the threshold

voltage, as we see in figure 8d, but the much lower drain current, which is expected as the electron mobility will have been affected by the irradiation.

When considering the saturation voltage, we can see a linear trend for this level of radiation. in figure 9 we see that for a linear fit, we get a R²-value of 0.9893. This linear behavior is since for the Braggs curve for this region can be approximated as linear, since it is such a small distance that makes up the HEMT structure.

Figure 9: Measured saturation voltage for HEMT for different doses of proton irra- diation. Values averaged for measurment points from 9V to 10V and over multiple HEMTs. The R²-value is 0.9893 for the trendline.

When comparing these results to the calculated damaged events using TRIM, we see that they agree with eachother, since we get a linear correlation between damage events in the form of vacancies and the decrease in saturation current.

4 Summary

We have seen that GaN based HEMTs are indeed radiation tolerant, only showing degradation at high levels of proton radiation, meaning that for diagnostic purposes purposes, it shouldn’t pose a problem. The protons used in this thesis were however of much lower energy than those used in cancer treatment, but considering the occur- rence of damage events such as vacancies that form and corresponding interstitials, it shouldn’t necessarily mean that GaN based HEMTs don’t have a role to play in diag- nostics of cancer cells, as they have shown to be biocompatible, with the possibility of detecting toxins through the conjugation of functional groups or antibodies on the surface of the HEMT, which may be in conjuncture with cancer treatment, or imaging using proton irradiation. There is more work to be done on the subject, but I think that this work shows a bit of the promise of GaN-based HEMTs, both in radiation heavy environments and in biomedicine as well. We have also seen that the dosage of protons that causes vacancies and interstitials, which in turn gives a effect on the DC measurments seems to agree with our TRIM calculations, which appears linear for the small interval which forms the HEMT structure even though as a whole, the Braggs curve suggest a inverse squared relation. This inverse squared relation can be seen in TRIM calculations if measuring over a longer range. It’s also important to note, that there are differences in different models of HEMTs, a temperature dependence at high temperatures, so these results may not be applicable in the general sense, but it does show a general promise for GaN devices.

References

[1] RSC Adv., 2019, 9, 15341, Highly sensitive AlGaN/GaN HEMT biosensors using an ethanolamine modification strategy for bioassay applications, Z. Gu, J. Wang, B. Miao, L. Zhao, X. Liu, D. Wu and J. Li

DOI: 10.1039/c9ra02055a

[2] J. Mater. Chem. C, 2013, 1, 877, Radiation effects in GaN materials and devices, A. Y. Polyakov, S. J. Pearton, P. Frenzer, F. Ren, L. Liuc and J. Kimd

DOI: 10.1039/c2tc00039c

[3] Growth and Characterization of Aluminum Gallium Nitride/ Gallium Nitride Ul- traviolet Detectors, p. 8, G. Parish;

Retrieved 2020-09-09 from http://my.ece.ucsb.edu/Mishra/resgroupfiles/parish.pdf [4] T. L. Chu, “Gallium Nitride Films,” J. Electrochem. Soc., vol. 118, no. 7, p. 1200,

1971.

[5] Journal of Vacuum Science Technology B 32, 012201 (2014);

doi: 10.1116/1.4836577

[6] High-Frequency GaN Electronic Devices, p. 9,P. Fay, D. Jena, P. Maki ISBN 978-3-030-20207-1

[7] Introduction to Solid State Physics, Eight edition, p.508-509, Charles Kittel ISBN: 0-471-41526-X

[8] Formation of two-dimensional electron gas at AlGaN/GaN heterostructure and the derivation of its sheet density expression, H Xiao-Guang et al.

DOI: 10.1088/1674-1056/24/6/067301

[9] Comparison of Si, Sapphire, SiC, and GaN substrates for HEMT epitaxy, March 2013 ECS Transactions 50(3):163-171

DOI: 10.1149/05003.0163ecst

[10] Electronic properties of intrinsic defects and impurities in GaN, Tran Thien Duc DOI:10.3384/DISS.DIVA-121710

[11] Fabrication and Characterization of AlGaN/GaN High Electron Mobility Transis- tors p.42-47, P. Javorka

[12] Bragg Curves and Peaks, Brookhaven National Laboratory. Retrieved 2020-09-12 from

https://www.bnl.gov/nsrl/userguide/bragg-curves-and-peaks.php

[13] Neutron irradiationon AlGaN/GaN high electron mobility transistors on SiC sub- strates, B-J. Kim, H-Y. Kim, J. Kim, S. Jang,

doi:10.1016/j.jcrysgro.2011.01.098

[14] Acta Biomaterialia Volume 8, Issue 2, February 2012, Pages 728-733 doi: 10.1016/j.actbio.2011.09.038

[15] Environment, health and safety issues for sources used in MOVPE growth of com- pound semiconductors, D.V. Shenai-Khatkhate et al

doi: 10.1016/j.jcrysgro.2004.09.007 [16] Retrieved 2020-09-12 from

https://www.snmmi.org/ClinicalPractice/content.aspx?ItemNumber=4825

[17] Randomized Trials and Technology Assessment, J. E. Tepper, A. W. Blackstock, doi: 10.7326/0003-4819-151-8-200910200-00146

[18] Radiation biology considerations of proton therapy for gastrointestinal cancers, Beltran C, Schultz HL, Anand A, Merrell K. , J Gastrointest Oncol 2020;11(1):225- 230,

doi: 10.21037/jgo.2019.06.08

[19] High-energy proton imaging for biomedical applications, Prall, M., Durante, M., Berger, T. et al. Sci Rep 6, 27651 (2016)

doi: 10.1038/srep27651

[20] RSC Adv., 2019, 9, 15341; Highly sensitive AlGaN/GaN HEMT biosensors using an ethanolamine modification strategy for bioassay applications, Z. Gu, et al.

doi: 10.1039/c9ra02055a

[21] Electronic Biosensors Based on III-Nitride Semiconductors, R. Kirste, N.

Rohrbaugh, I. Bryan, Z. Bryan, R. Collazo, and A. Ivanisevic, Annu. Rev. Anal.

Chem. 2015. 8:149–69

doi: 10.1146/annurev-anchem-071114-040247

[22] Gary Tuttle, Dept. of Electrical and Computer Engineering, Iowa State Univer- sity, Contact resistance and TLM measurements PDF,

Retrieved 2020-09-03 from

http://tuttle.merc.iastate.edu/ee432/topics/metals/tlm_measurements.pdf

[23] Characterisation of ohmic contacts on n-InP and p-InGaAs based materials van Maasakkers, M. J. C. P. (Author). 28 Feb 2003 Student thesis: Master, p. 23 [24] Stopping Range of Ions in Matter, Retrieved 2020-02-28 from

http://www.srim.org/

[25] Accelerator; Retrieved 2020-09-06

https://www.tandemlab.uu.se/infrastructure/Accelerators/pelletron/