Fabrication and Characterization of Schottky Diode

Transcription

1 Fabrication and Characterization of Schottky Diode Arnab Dhabal

2 Acknowledgements I would like to express my greatest gratitude to the people who have helped and supported me in this project. I wish to thank Prof. Damodaran and Prof. Anjan K. Gupta for helping throughout right from giving invaluable suggestions and comments to making arrangements for obtaining the samples. I also thank Prof. Y.N.Mohapatra, for giving advice on our method of fabrication and providing me with the silicon sample. Special thanks go to Mr. Ramesh, whose vast experience with the Vacuum Deposition Unit was of great help, and Mr. Upendra, for helping me in getting the project done smoothly and in time. The data acquisition and characterization part would have been unsatisfactory without the assistance of Mr. Indranuj Dey, who not only helped set up the electric instruments, but also took interest in discussing the results, thereby providing deeper insight as I worked on the project. I am extremely grateful to him. Page 2 of 22

3 Contents Introduction.. 4 Fabrication of a Schottky Diode. 4 Choice of Material 4 High Vacuum System Overview 5 Steps of Operation 5 Electrical Characterization.. 8 Current-Voltage Characteristics 8 Current Variation with Time due to Heating 12 Voltage Characteristics on Application of Square Wave Pulses 13 Discussion and Scopes of Improvement References Page 3 of 22

4 Introduction A Schottky diode is a special type of diode with a very low forward-voltage drop and a very fast switching action. When current flows through a diode, there is a small voltage drop across the diode terminals. A normal diode has between volt drops, while a Schottky diode voltage drop is between approximately this lower voltage drop translates into higher system efficiency. Chemically, a Schottky diode has a Schottky contact between a semiconductor and some appropriate metal. The other end of the semiconductor has an Ohmic contact, with a metal. To ensure that the two ends of the semiconductor form different junctions, a gradient in the doping concentration is required within the semiconductor, such that the end with the Ohmic contact has more carriers than the Schottky contact. Fabrication of a Schottky Diode Choice of material SiC performs best as a semiconductor for Schottky diodes. However, only Si wafers were available within IIT Kanpur. On reading a few papers, it was understood that n-typed Si works better as a diode than p-type Si. Also a Schottky contact is formed on Si by depositing Au, while the Ohmic contact could be of any metal, like Aluminium. The initial plan was to take a glass plate as the base, and to sputter deposit the following components as 2mm x 5mm strips in the order given: Aluminium Heavily doped n- Si (10 18 /cc) Low doped n- Si (10 16 /cc) Gold This had to be done in such a way that later the wires can be soldered to the Gold and Aluminium ends. However this plan was rejected due to two major reasons: i. Unavailability of n-si of two different types of doping concentration. ii. The possibility that on sputter depositing the Si, the doping and the Si might not evaporate and get deposited in the same ratio as in the original wafer sample. So it was decided that I should start with a Si sample that already has a gradient in the doping concentration. Au was to be sputter deposited onto the lowly doped (shiny) Si surface. Page 4 of 22

5 The Ohmic contact was to be made of Indium solder, while the wire to be drawn from the Gold side was to be attached to it by Silver paste. High Vacuum System Overview The HINDIVAC system at IIT Kanpur Modern Physics Lab utilizes 3 pumping devices in stages: i. The rotating Mechanical Pump - Primary source for creating vacuum. Can reach only up to 10-3 mbar ii. The Diffusion Pump - It uses hot oil and has the advantage of reaching up to 10-7 mbar but must be backed by a rotary pump. iii. The cold trap - It reduces pressure by condensing, or freezing out condensable vapours that may exist in the system. It also prevents oil vapour in diffusion pump from back streaming into the system. Liquid N2 is used for the purpose. Other system components include valves and baffles to aid the control of action of these pumps. The valves allow Roughing and Backing modes of operation. In the roughing mode, only a rough vacuum is obtained by the rotary pump. The Foreline valve and the Hi-Vac valve isolates the diffusion pump, and the cold trap, from the chamber. After completion of roughing (Pressure < mbar), the Foreline Valve is opened to start the Backing mode. Within the Vacuum chamber there are electrodes on which the material to be deposited are kept in boats, and high Power is given to them continuously until the material starts evaporating. The substrate which is placed above the electrodes gets coated. The system also has a Digital Thickness Monitor, which should be placed in the vacuum chamber at the same distance from the material being sputtered as the substrate. The Acoustic Impedance and Density of material are given as input. It displays the thickness of material deposited in Angstroms, and the rate of Deposition. Steps of Operation Initially the vacuum chamber is tested for leaks. This is done by just running the Rotary pump, with the Combination Valve in Roughing. Possible sources of leaks are identified and attended to. The substrate is prepared by cleaning it using acetone, and placing it appropriately on a plate, using cleaned blades for support. The Aluminium boat is placed on the electrode with the gold after rinsing it in acetone as well. Page 5 of 22

6 After closing the chamber properly, the vacuuming process is started: i. The power supply and the main MCB is switched on. All valves (High-Vacuum valve - HV and Combination Valve- CV) are closed. ii. The Rotary Pump is switched on. iii. CV is put to roughing iv. When vacuum reaches ~0.005 mbar in the roughing line, CV is changed back to Backing. v. Water supply is turned on, and then the Diffusion Pump is switched on. vi. In around 20 minutes, the pressure in the backing region also falls to mbar. vii. CV is changed to Roughing for 1-2 minutes and again brought back to Backing. viii. When the pressure reaches 10-4 mbar, HV is opened. ix. Liquid N 2 is poured into the cold trap. x. The pressure is allowed to stabilize at <10-5 mbar for 30 minutes. Figure 2 Gold sputtering in process in 10-5 mbar pressure Figure 1 - High Vacuum Generator and Thin Film Depositor Now current is passed through the electrode for heating up the Au to evaporating temperatures. The current is increased at the rate of 2 Amp/minute, till deposition temperature is reached. The DTM was not functioning properly, so it was used only for getting a rough estimate, by seeing the rate of deposition from time to time. This is carried out over a period of 30 minutes, and the current rate was reduced again at a steady rate of 2 Amp/minute. It was estimated that 800nm 1200 nm of Gold had been deposited on the Silicon surface. Page 6 of 22

7 After 10 minutes, HV is closed, and the DP is switched off. After around 20 more minutes, CV is closed and Rotary pump and water supply are also switched off. The air admittance valve is opened for fast release of the vacuum. The sample is collected out. The gold coating on it is visible. It is found that gold had been deposited at the edges of the Silicon piece, which can cause unexpected characteristics. So the edges with depositions are chipped off. Indium is used to solder a thin copper wire onto the rough surface of the Silicon (for the Ohmic contact). On the other surface, Silver paste is used to attach a thin copper wire on to the Gold. For ease of use, the sample thus produced is mounted on a PCB. The Au coated Si piece is vertically attached to the PCB using insulating glue. Thicker wires are soldered on-to the PCB and connected to the two thin wires of the diode. These two terminals can now be directly plugged into bread-boards for electrical characterization. Figure 4 Design schematic of Fabricated diode Figure 3 The fabricated diode mounted on a PCB with connections Page 7 of 22

8 Electrical Characterization 1) Current Voltage Characteristics The following circuit was implemented for the normal Voltage Current characteristics: Figure 5 - Circuit Diagram A The circuit with the diode A was made on a breadboard. A Voltage source, 2 Multimeters, and Wires were the other requirements Figure 6 - Circuit Set-up A Page 8 of 22

9 Readings for Forward Bias: Voltage in mv Current in µa Voltage in mv Current in µa Voltage in mv Current in µa Readings for Reverse Bias: Voltage in mv Current in µa Voltage in mv Current in µa Voltage in mv Current in µa Page 9 of 22

10 Figure 7. Current vs Voltage characteristics in both forward reverse regions The I vs V characteristic shows that current flow in one direction is definitely favoured over the other. In the reverse bias, there is flow of current, but substantially low (<5 µa at 1 V) However, it can be seen that the positive region does not have a very steep slope as expected from normal diodes. This is indicative of an inbuilt Resistance in series, possibly at one of the junction of the diodes, or also because of the thick layer of Silicon. We can roughly say that sans the series resistance the Voltage drop across the diode would have been 0.19 V. The resistance changes with temperature. So at higher voltages when there is substantial heating, the current reading gradually goes up, for (>5 minutes at 2 V forward bias) indicating that the resistance is lowered. If the current is switched off for some time, the resistance returns to its room temperature value. So when switched on again, the current returns to its low starting value. It was found that above 500 mv, the heating effects caused too unstable current values, and hence the readings taken in that region were not very reliable, and hence has been excluded from the plot (Figure 7). Page 10 of 22

11 The V-I equation for an ideal diode is : I D qv = Io exp 1 nkt Correction for resistance: I (V IR) = Io exp 1 where V0 = V0 i.e. V = +1 + Figure 8 - Plot of Forward biased region after curve fitting Using OriginLab 7.5, Curve fitting was carried out for the above equation and found the coefficients as: R = kω I 0 = µa V 0 = mv It is noted that the resistance of 6200 Ω is quite high in comparison to normal diodes. Page 11 of 22

12 2) Current variation with time due to heating The rate of change of resistance value is quite high as the following dataset comprising of current values noted down at intervals of 10 seconds at a voltage of 2.0V indicates. These characteristics were repeated if the circuit was given sufficient time to cool down. Time in secs Current in ma Time in secs Current in ma Time in secs Current in ma Figure 9 Plot of Diode Current against time at Voltage = 2.0 V The possibility of this effect being caused by a capacitor discharging was also considered, but given the high time period it is highly unlikely of a capacitor of capacitance ~ 0.025F to be present alongside the diode. Besides, the results of the following section show that there is capacitance but of the order of nanofarads. Page 12 of 22

13 3) Voltage characteristics on application of square wave pulse: The following circuit was implemented, with the use of a Function Generator at 1kHz and the characteristics were studied by using a Oscilloscope: Figure 10 - Circuit Diagram B Figure 11 Circuit Set-up B The experiment was once carried out for -3V to 3V and for 0V to 6V using both the sample made and a normal LED. The data were collected on a PC and analyzed separately. Page 13 of 22

14 V in = V to V LED: Figure 12 Plot of V(in) and V(out) for LED and V(in) changing from -3 V to +3V Voltage across LED in forward bias= = 1.80 V Diode A: Figure 13 - Plot of V(in) and V(out) for Diode A and V(in) changing from -3 V to +3V Page 14 of 22

15 Voltage drop across the diode = 2.96 V 0.56 V = 2.4 V. Resistance of Diode in forward bias = (2.4/0.56)*1 = 4.3 kω The sample prepared follows the first requirement of a diode. It allows current to flow in the forward direction and prevents it from flowing in the reverse direction. There is substantial capacitance in the diode, because of which we get a curve like that of a differentiator circuit. The charge stored in the capacitor continues to discharge even after the voltage is reversed. In the reverse direction, the slope of the curve changes twice in the course of the value returning to zero. Initially it was thought that it is indicative of the fact that there are 2 capacitors at work, the effect of one of which changes with time. But on closer inspection it was noticed that the input voltage had developed a kink in that region. Somehow, due to momentary heavy current drawing, the Voltage source fails to keep the same negative Voltage for a period of about 66 µs. It was further found that this effect reduced and became indiscernible, as the Voltage amplitude was brought down to zero. Because of this analysis of the negative regime would largely prove to be futile. Figure 14 Magnified view of region of voltage direction changing It is to be noted that these magnified values were recorded after some heating had already taken place in the circuit and hence the resistance being lower, voltage drop across the diode is lesser and that across the 1kΩ resistor is more. Page 15 of 22

16 The capacitance in the forward direction was obtained by curve fitting of an exponentially decaying curve using OriginLab. In the curve, y = V0 + A*exp(-t/τ), subjected to the constraint V0 + A = 5.9 (theoretical Voltage value from which capacitor starts discharging), the coefficients obtained by Levenberg-Marquardt iterations are as follows: V 0 = V A = V τ = ( )x 10-7 s From equation, at t =, Voltage drop across diode = = 2.00V Thus, diode resistance in forward bias R = (2/0.96)*1 kω = 2.1 kω Hence from the following model I of circuit (in forward biased region), we get By circuit analysis, Equivalent resistance = 3.1 kω Capacitance C = τ/r = 1.6 nf Figure 15 Possible simple equivalent circuit A Another model II of circuit (in forward biased region), is given alongside. From this, By circuit analysis, Equivalent resistance = (1*2.1)/(1+2.1) = 0.68 kω Capacitance C = τ/r = 7.1 nf Figure 16 Possible simple equivalent circuit B Page 16 of 22

17 (i) V in = 0 V to V LED: Figure 17 Plot of V(in) and V(out) for LED and V(in) changing from -3 V to +3V Voltage across LED in forward bias= = 2.00 V Diode A: Figure 18 Plot of V(in) and V(out) for Diode A and V(in) changing from 0 V to +6V Page 17 of 22

18 Voltage drop across diode = ( ) V = 4.40 V Resistance of Diode in forward bias = (4.4/1.52)*1 = 2.9 kω It is evident from the curves that the slope in the forward and reverse biasing are different. Thus it indicates that the Time constant during the 2 cases are different. If we model the diode as having only one capacitance, it must be having a resistance in parallel to it, that becomes infinite on reverse biasing. This might be in addition to another series resistance within the diode. So we may model the diode as follows: Figure 19 - Model for diode circuit that can explain the results of both the forward and reverse biases Figure 20 Magnified view of region of voltage direction changing for forward bias, V(in) changing from 0 V to +6V Page 18 of 22

19 Note: The input voltage is a square pulse from 0.08 V to 5.80 V. The forward bias region of the curve, was fitted on to an exponentially decaying polynomial curve, y = V0 + A*exp(-t/τ), subjected to the constraint V0 + A = 5.8 (theoretical Voltage value from which capacitor starts discharging), the coefficients obtained by Levenberg-Marquardt iterations are as follows: V 0 = V A = V τ 1 = ( )x 10-7 s From equation, at t =, Voltage drop across diode = = 4.4 V Thus, diode resistance in forward bias = (4.4/1.4)*1 kω = 3.1 kω From Figure 19, R1 + R2 = 3.1 kω (1) Equivalent resistance as seen from capacitor = (R1*(R2 +1))/(R1 + R2 +1) Thus Time constant τ 1 = 5.6 x 10-7 = C(R1*(R2 +1))/4.1 (2) Figure 21 Magnified view of region of voltage direction changing for reverse bias, V(in) changing from 0 V to +6V The reverse bias region of the curve, was also fitted on to an exponentially decaying polynomial curve, y = V0 + A*exp(-t/τ), subjected to the constraint V0 + A = 5.8 (theoretical Voltage value from which capacitor starts discharging), the coefficients obtained by Levenberg-Marquardt iterations are as follows: Page 19 of 22

20 V 0 = V A = V τ 2 = ( )x 10-7 s Now from Figure 19, the circuit is only a discharging circuit having a capacitor C and resistors R2 and 1 kω. So τ 2 = 2.7 x 10-6 = C*(R2 +1) (3) From (1), (2) and (3), R1 = 0.85 kω, R2 = 2.25 kω, C = 0.8 nf Figure 22 - Model for diode circuit that can explain the results of both the forward and reverse biases Page 20 of 22

21 Discussion and Scopes of Improvement The diode that was fabricated acted like a diode but it had many defects most visibly the high resistance and capacitance. None of the results regarding the calculation of the resistances and capacitances are very conclusive as no two results exactly support each other. However we can say that that the diode has a resistance of the order of kω and a capacitance of the order of nf. A better analysis could have been possible had the diode been studied for a few more ranges of square waves. The capacitance and high resistance can be arising from : a. Formation of oxide layer on the silicon wafer. This possibility could have been removed by treating the Silicon wafer with HF before using it as a substrate. b. An intrinsic property of the silicon wafer, caused during the formation of the gradient in the doping concentration Also there is substantial effect of temperature on the resistance of the diode, which again makes the results dependent on how long the experimentation is on. This effect can be reduced by using some mechanism of dissipating away the heat generated, in the thick Si layer. Since the voltage sources are not very sharp in the transition from +ve to ve Voltage, it was not possible to exactly measure the reverse recovery time, which is supposed to zero for Schottky diodes. Page 21 of 22

22 References Papers: Metal-semiconductor Contacts for Schottky Diode Fabrication Mark D. Barlow Comparison of Current-Voltage Characteristics of n- and p-type 6H-SiC Schottky Diodes Q. Zhang, V. Madangarli, M. Tarplee, and T.S. Sudarshan Schottky Contact Barrier Height Enhancement on p-type Silicon by Wet Chemical Etching G. A. Adegboyega, A. Poggi, E. Susi, A. Castaldini, and A. Cavallini Web-sites: Page 22 of 22