«SANDIA REPORT SAND2016-0109 Printed January 2016 Electrical Breakdown Physics in Photoconductive Semiconductor Switches (PCSS) Alan Mar, Fred ...»
5.2.1. Vertical PCSS Design As discussed in Section 4, a new configuration of the PCSS was designed and implemented that results total current handling scaling with a 2D array of current-sharing filaments. To operate this switch, we use deeply-penetrating sub-bandgap 1064nm laser light to seed filaments through entire wafer thickness. Another important feature of the switch design is the use of a solid backside heatsinking interface to improve the properties of the damage-prone p-contact.
High current p-contacts are more difficult due to the roughly order-of-magnitude lower mobility of holes in GaAs compared to electrons (n-contact).
As shown in Figure 5-8 an array of filaments is induced through the bulk GaAs from the front surface to the backside contact via an array of apertures in the top contact. With sufficient standoff between the top and bottom contacts and edges of the device, the switched field is thus heldoff across the thickness of the bulk GaAs substrate, and high-voltage operation in air is achieved without the need for immersion in high-breakdown fluids such as oil (Figure 5-9) as is the case with the conventional surface gap device.
% '( ( ) &+ &,-. / 012 3 & *"$) * 4* 5617"57 4* 5617"57 ! "#$ ! "#$% '() '& "+,Figure 5-8: Vertical PCSS design results in 2-D array of filaments across the bulk GaAs
Figure 5-9: Vertical PCSS installed in test fixture for electrical biasing and laser triggering 5.2.2. Vertical PCSS Electromagnetic and Circuit Modeling The vertical PCSS design necessary still has interfaces in its structure that results in localized Internal E-field enhancements that will limit high-voltage operation. This was modeled using the ELECTRO EM simulation software tool. The resulting field distribution shows that the field is highest at n-contact edges (corners) triple-point boundary by a factor of ~2x compared to the bulk field, as shown in Figure 5-10. The experimentally achieved breakdown voltages correspond very closely to the ‘textbook’ breakdown field in GaAs (~100-200kV/cm). This amount of field enhancement is likely very difficult to substantially improve upon in the design.
Figure 5-10: Electromagnet modeling of the fields in the Vertical PCSS device and fixture
Since the sudden change in carrier generation when the switch voltage crossed the avalanche voltage caused rapid numerical oscillations, equations 2-3 were replaced with the smooth function of voltage, Vp, which turns on the avalanche term with adjustable slope, Vs.
Rp’/Rp = [tanh((|Vp|-Va)/Vs)+1)/2]/Ta + 1/Tr = -n’/n 
The differential circuit equation, L Q’’ + R(t) Q’ + Q/C = 0  was numerically solved with the Matlab function “ode15s” that finds slowly changing solutions compared to the time step (stiff). At each time step of the calculation, the subsequent value for the PCSS resistance was calculated from equation 5.
The results from this model are shown for a range of fixed circuit resistances in Figure 5-11. All other parameters were kept constant at values that have been implied from experimental results.
Figure 5-11. Solutions to a model for a capacitive discharge circuit with a high gain PCSS, as a function of fixed circuit resistance.
It is difficult to see the difference in the current produced with a PCSS L-R-C circuit compared to the current produced with an ideal switch. To look at this more carefully, Figure 5-12 shows the model results for one fixed circuit resistance with the dynamic resistance of a PCSS and the fixed resistance of an ideal (constant “on” resistance) closing switch.
Figure 5-12. A comparison of the high gain PCSS model solution to a the solution for and ideal closing switch with constant “on” resistance.
The circuit used to test vertical PCSS was a low inductance, low resistance, capacitive discharge circuit. Estimates of the circuit parameters which gave a reasonable fit to the measured current are shown in Figure 5-13.
Figure 5-13. The solution with the parameters that made the best fit to the current measured in a vertical PCSS test.
5.2.3. Vertical PCSS Lifetime Testing Figure 5-14 depicts the experimental setup for testing the new multifilament PCSS. An IRsensitive video camera images filaments emitting recombination light at ~880nm by imaging through a notch filter that reflects at this wavelength but transmits the trigger light at 1064nm.
The vertical switch is constructed from standard 600µm thickness GaAs wafer material. At the applied voltage of 2.2kV, this corresponds to 35kV/cm internal field without accounting for enhancement due to the structure of the internal interfaces in the switch structure. The resulting current limited by stray inductance, resistance is 0.4kA. Computer acquisition of data for lifetesting was utilized for this test as well as video imaging of the filaments generated.
Figure 5-14: Configuration for testing multifilament vertical PCSS Figure 5-15: Image of filaments in vertical PCSS as seen through the apertures used to inject trigger light. A filter is used to image recombination radiation from the filaments.
Approximately 120 filaments are visible on each shot as shown in Figure 5-15, changing position on each firing of the switch. There possibly may be more filaments beneath the metal and the filament brightness may reflect either the proximity to the aperture, or the current density associated with a particular filament, as the filaments will tend to terminate on the field metal and their visibility at the aperture will be masked to varying degrees. This phenomena is Figure 5-16: Filament path in a vertical PCSS, terminating on the triggering apertures on top depicted in Figure 5-16, where absorbed e-h pairs seed the filament in the centers of the apertures. The internal electric fields guide filament termination towards the metal outside the aperture at top contact. The recombination light that makes filaments visible is strongly absorbed in the GaAs at the surface in the aperture, so the absence of a visible filament is not necessarily an indication of the lack of filament formation at such an aperture.
The switch performance under these conditions is shown in Figure 5-17. The test was terminated at ~1e5 shots after several days of testing. The actual final number of firings was 84125 due to pulse charger malfunctioning, as determined later from the recorded waveforms (including an anomalous high-bias firing at ~60000 shots. No discernible performance degradation was detected throughout this test, and the contact metal looked pristine under visual examination. This is testament to the robust design of the heat-sinked p-contact and the effective
Figure 5-17: 1E5 shot lifetime test of vertical PCSS at 0.4kA / 35kV/cm
sharing of current amongst many (~120) filaments on every shot to prevent contact degradation.
This result could likely be achieved at higher currents with thicker switches so higher voltage can be applied.
The same switch was then retested, charged to ~150kV/cm (4.5kV) to achieve 1kA total current. The imaging showed signs of breakdown due to the excessive field and the switch failed due to high-voltage breakdown, tracking at n-contact edge where the predicted field enhancement would occur. There was little discernible performance degradation until failure at 1400 shots, as shown in Figure 5-18. This illustrates the point that thicker substrates are needed to operate at lower fields for given voltage to achieve higher total currents in the external circuit.
At the end of this LDRD project, before tests could be carried out, 2mm thick GaAs wafers were procured, and growths planned when follow-on work can be pursued. 5mm material procurement RFQ was also received from American Crystal Technologies, so it is established that wafer thickness procurement is not a fundamental limitation of this PCSS design for highvoltage operation, in addition to stacking switches in series modules.
Figure 5-18: 1.0kA lifetest of vertical PCSS. Switch breakdown failure at 1400 shots due to excess field in thin (600µm) wafer structure.
6. PCSS TRIGGER GENERATOR TESTING AND FOLLOW-ON
DEVELOPMENTAnother important application pursued in this LDRD is the use of PCSS for trigger generator applications. Conventional in-plane PCSS have achieved triggering of a 100kV sparkgap (Kinetech-type) switch of the type similar to switches being considered for accelerator upgrades (Figure 6-1, Figure 6-2). This switch is used to recharge a fast Marx generator. The advantage provided is optical triggering in an electrically noisy environment. Compared to conventional commercial trigger generators, PCSS triggering offers the possibility of faster risetime (sub-ns.) and co-location at the spark gap switch to eliminate 50Ω cabling due to minimal volume requirements of the compact PCSS device.
The trigger is also being developed for pulsed power for HPM applications that require miniaturization and robust performance in noisy compact environments using diode laser triggers (Figure 6-4). This has spawned new programs for developing this technology, including an STTR for VCSEL trigger laser integration, also pursuing other follow-on applications (Figure 6-5).
Figure 6-1: PCSS triggering of a 100kV Kinetech-type sparkgap switch Figure 6-2: Circuit for PCSS trigger generator applied to 100kV sparkgap switch.
Figure 6-3: Resulting waveforms from PCSS trigger generator and 100kV spark gap.
Figure 6-4: Characteristics of compact laser-diode based trigger laser system for mobile platforms.
Many key advances in developing PCSS for high-voltage/current pulsed-power applications have been achieved in this LDRD project. High-current density GaAs contacts designs have achieved 10x shotlife improvement compared to previous best designs. Such contacts have achieved 3.7e4 shots, switching 100A in a single filament.
High-Voltage multi-filament triggering experiments have been done to explore using microlens arrays, rib-etched gaps at 75kV, demonstrating multifilament confinement achieved in both etched rib structures and cylindrical lens arrays. These experiments reveal an important relationship and evidence for faster laser pulses to overcome beam nonuniformity.
This LDRD project has succeeded in developing new optically-triggered photoconductive semiconductor switch (PCSS) designs that show great promise for scaling to modules capable of 200kV (DC) and 5kA current that can be stacked in parallel to achieve 100's of kA with 10e5 shot lifetime. The new vertical switch design configuration generates parallel filaments in the bulk GaAs (as opposed to just beneath the surface as in previous designs) to achieve breakdown fields close to the maximum for the bulk GaAs (~100kV/cm) while operating in air, and with 2-D scalability of the number of current-sharing filaments. This design also may be highly compatible with 2-D VCSEL arrays for optical triggering. The demonstration of this design in this LDRD utilized standard thickness wafers to trigger 0.4kA at 35kV/cm (limited by 0.6mm wafer thickness), tested to ~1e5 shots with no detectable degradation of switch performance. Higher fields, total current, and multi-kA/kV switching voltages would be achievable with thicker GaAs wafers.
Another important application pursued in this LDRD is the use of PCSS for trigger generator applications. Conventional in-plane PCSS have achieved triggering of a 100kV sparkgap (Kinetech-type) switch of the type similar to switches being considered for accelerator upgrades.
The trigger is also being developed for pulsed power for HPM applications that require miniaturization and robust performance in noisy compact environments. This has spawned new programs for developing this technology, including an STTR for VCSEL trigger laser integration, also pursuing other follow-on applications.
BIBLIOGRAPHY D. H. Auston, “Picosecond optoelectronic switching and gating in silicon,” Appl. Phys. Lett., vol. 26, p.
 P. LeFur and D. H. Auston, “A kilovolt picosecond optoelectronic switch and Pockels cell,” Appl. Phys.
Lett., vol. 28, p. 21, 1976.
 C. H. Lee, “Picosecond optoelectronic switching in GaAs,” Appl. Phys. Lett., vol. 30, p. 84, 1977.
 W. C. Nunnally and R. B. Hammond, “80 MW photoconductor power switch,” Appl. Phys. Lett., vol. 44, p. 980, 1984.
 G. M. Loubriel, M. W. O’Malley, and F. J. Zutavern, “Toward pulsed power uses for photoconductive semiconductor switches: Closing switches,” in Proc. 6th IEEE Pulsed Power Conf., Arlington, VA, (New York), p. 145, Institute of Electrical and Electronics Engineers, Inc., 1987.
 F. J. Zutavern, M. W. O’Malley, and G. M. Loubriel in Proc. 6th IEEE Pulsed Power Conf., Arlington, VA, (New York), p. 577, Institute of Electrical and Electronics Engineers, Inc., 1987.
 F. J. Zutavern, G. M. Loubriel, H. P. Hjalmarson, A. Mar, W. D. Helgeson, M. W. O’Malley, M. H.
Ruebush, and R. A. Falk, “Properties of high gain GaAs switches for pulsed power applications,” in 1997 Pulse Power Conference, 1997.
 F. J. Zutavern and G. M. Loubriel, High-Power Optically Activated Solid State Switches, ch. 4, p. 61.
Boston: Artech House, 1994.
 H. W. Thim and S. Knight, “Carrier generation and switching phenomena in n-GaAs devices,” Appl.
Phys. Lett., vol. 11, p. 83, 1967.
18 S. G. Liu, “Infrared and microwave radiation associated with a current-controlled instability in GaAs,” Appl. Phys. Lett., vol. 9, pp. 79—81, 1966.
 J. A. Copeland, “Switching and low-field breakdown in n-GaAs diodes,” Appl. Phys. Lett., vol. 9, pp.
 K. K. N. Chang, S. G. Liu, and H. J. Prager, “Infrared radiation from bulk GaAs,” Appl. Phys. Lett., vol.
8, pp. 196—198, 1966.
 E. M. Conwell, “Electron-hole generation in GaAs,” Appl. Phys. Lett., vol. 9, pp. 383—385, 1966.
 E. M. Conwell and M. O. Vassell, “High-field transport in n-type GaAs,” Phys. Rev., vol. 166, pp.