MHE - Mega Impulse, Ltd.

Home Optics Electronics Tech Transfer Contact MHE MHE News

Mega Impulse, Ltd.

MHE represents Mega Impulse, Ltd., a Russian research organization associated with St. Petersburg's prestigious Ioffe Physico-Technical Institute. Mega Impulse is currently conducting research in several very exciting areas of physics and electronics. They have achieved a number of significant breakthroughs including several in cryoelectronics and superconductivity. Mega Impulse also produces microsecond pulse generators (both production and prototype models with applications in optics as well as electronics) and novel semiconductor devices.

To learn more about Mega Impulse's products and technologies, click on the appropriate button below.



Novel Semiconductor Devices

Microsecond Pulse Generators



Field Effect Transistors with Superconducting Channel (SuFET)

High temperature superconductors (HTSC)are very promising materials for electronics mainly because of their near zero DC-resistivity at relatively high temperatures (higher than liquid nitrogen -- 77ēK). Very good passive electronic components are designed on the basis of HTSC, for example -- power cables, high quality resonators, striplines with extremely low attenuation and dispersion, etc. Success in the area of active components has been considerably less impressive. The brilliant physical idea of Josephson junctions has not yet been realized in practice because of its very complicated manufacturing processes. Switches on the basis of simpler ideas -- destruction of superconductivity by short pulse of extra-current or by pulsed magnetic field -- have rather poor basic parameters. It has been shown that superconducting field effect switches cannot be designed on the basis of low temperature superconductors, because of very high concentrations of free carriers (approx. 1023cm-3). In HTSC this concentration is two orders lower (approx.1021cm-3) which is the reason for intensive investigation of field effect in HTSC. One of the main stimuli for investigation is the inherent speed limitation in present computer technology.

The speed and performance of today's computers is determined to a large extent by on-chip and chip-to-chip interconnect properties: significantly more than half of the total delay is due to interconnect delay time. (Chip-to-chip interconnects operate in time-of-flight mode and on-chip interconnects in RC mode.)

Existing metallic interconnects have high dispersion and attenuation in time-of-flight mode because of high resistivity of the metal plate. HTSC interconnects can dramatically improve the situation, but cannot be matched with MOSFET transistors because the impedance of such interconnects is, as a rule, 5 to 10 Ohms, while the impedance of microelectronic MOSFET channel is about 1 kOhm. (Interconnects should be matched at one side with the MOSFET channel, the other side should be open.)

The delay time of RC-interconnects can be reduced dramatically in an HTSC version because of the reduced resistivity. However in this case the operating current would increase by orders of magnitude -- but this is not possible for MOSFET switches because of high resistance of the channel. In addition the materials and processing techniques for HTSC and for normal silicon devices are absolutely different and would be extremely difficult to combine.

Field Effect Transistors with a superconducting channel (SuFET) would overcome all of these disadvantages. Because of very low resistance of the channel, it can be easily matched with the time-of-flight HTSC interconnects and have low losses at the high current RC-interconnects. Additionally SuFETs and HTSC interconnects will be made using identical techniques and materials, so there would be no manufacturing incompatibility. In this case very fast, all-HTSC integrated circuits can be designed and manufactured. Also it is very important that the main technical solution used in MOSFET-based integrated circuits can be in HTSC-circuits. Therefore SuFET now is a key problem of HTSC-microelectronics.

The operating principles of SuFET are based on experimental data which shows that the critical temperature Tc decreases with a decrease in the free carriers in HTSC-film.

The SuFET gate structure consists of interchanging HTSC-Insulator-Metal layers. The input voltage applied between the metal and HTSC results in changes in the concentration of carriers in the HTSC channel, and hence changes the critical temperature, Tc, of the superconducting transition. Experiments show that D Tc/Tc @ D N/N. This means that the SuFET can be switched on and off at a constant external temperature by means of gate voltage variations.

Generally speaking the SuFET is not a simple device for design and manufacture. Calculations show that the screening length in HTSC is rather small and the electric field can penetrate HTSC film made of three elementary cells or less: approximately 3.5 nanometers in the case of Yba2Cu3O7. This means that the HTSC channel must be approximately the same thickness, and that the Tc of the first HTSC layer near the HTSC insulator interface must not be lower than the Tc of the other layers. The quality of the insulator layer must also be very high -- the product of e EBu must be >1.7 x 109 cm/s, where e is the dielectric constant, EB is the breakdown field, and u is the field effect mobility in HTSC.

The metal-insulator-HTSC structure for investigation of the field effect. Superconducting and insulating layers are made with laser ablation technique, metal contacts with thermal evaporation of Ag.

Our laser ablation system consists of two neodymium:yttrium aluminum garnet lasers synchronously operated in Q-switched mode with pulse energy of about 0.25J. Laser beams are concentrated on two rotating cylindrical targets. A working plasma torch is formed due to interaction of intersecting plasma jets passing from the target. At the top of this torch, a special shield is placed to separate the substrate from the direct flow of erosion plasma so that films grow only from lateral streams of diffusion plasma. This technique allows us to produce an extremely smooth, thin film.

It is well known that lattice mismatch and surface defects result in a reduction of superconducting parameters of ultrathin YBaCuO channel layers with less than a six-cell thickness. The situation can be improved by a special buffer layer placed between substrate and ultrathin channel film. Material of this layer is normally the semiconducting ceramic PrBa2Cu3O7. Lattice parameters of PrBaCuO are identical to YBaCuO but the proximity effect prevents a major improvement of channel layer superconducting parameters. Namely due to these phenomena the critical temperature of ultrthin channel layers is far less than 77ēK.

We have developed a new material for the buffer layer -- YBa2Cu3-xNbxO7. Our investigations show that this material is a mixture (Fig. 4) of superconducting YBCO and dielectric YBa2NbO6 grains. Proportion between grains is determined by the Nb content. At x<0.21 material was found to be superconducting at Tc~ 90ēK. At x~ 0.21 there is a superconductor-insulator transition. As it turned out, insulating YBCNO material is a perfect buffer layer for growing ultrathin YBCO films. YBCO grains of this material (Fig. 4) play the role of the perfect nuclei for the first cells of film. Because of the high velocity of the film growing in the "ab" plane, these perfect regions reach each other very quickly, forming the perfect ultrathin film. Typical SEM image in the back scattering electron mode of buffer layer with x=0.24 is shown on the right side of Fig. 4. The size of superconducting (dark) and insulating (white) grains is 10 to 100nm. X-ray diffraction shows an orthorhombic structure of YBCO grains with c-axis normal to the substrate plane and cubic structure of YBa2NbO6 insulating grains grown in (100) direction.

Dependencies of normalized resistance R(T)/R(100K) on temperature for ultrathin Yba2Cu3O7 films are shown in Fig. 5. One can see that the YBa2Cu2.7Nb0.3O9 buffer layer gave us an opportunity to make ultrathin (3 - 4 cells) with superconducting transition at a temperature far higher than 77ēK. To our knowledge this is the best results obtained worldwide to date.

The other important aspect of SuFET is the gate insulator. We grow this layer using the same laser ablation technique which is used for YBaCuO. We investigated many materials: Y2O3, SrTiO3, etc. Ferroelectric Pb(Zr0.5Ti0.5)O3 seems to be the best material. At 150nm thick, this material blocks 20v in both directions (E @ 1.3 x 106 V/cm-1) with less than 1m A of leakage current. Fig. 6 shows the SuFET structure with a thin film insulator. The dimensions of the active (thin) part of the channel structure are 200 x 300m m2. All layers were made by laser ablation using special silicon masks.

The thickness of the active part of the channel is 5nm and the thickness of the contact part is approximately 15nm; both parts are made simultaneously during one ablation process. Fig. 6 shows the 30% modulation of channel resistivity (in resistive state) by electric field at 77ēK and 6V at the gate. It is also the best result obtained worldwide to date, but this result is not sufficient for normal operation of a SuFET.

Our next step will be to increase the insulator's dielectric constant e to approximately 2,000 (currently e @ 200), which we predict should be sufficient to achieve switch-on and switch-off in a SuFET at 77ēK. This would be a real breakthrough in HTSC-electronics.



Cryogenic (1.6-300K) Operation of a Bipolar Transistor

with a Tunnel MOS Emitter and an Induced Base

Cryogenic applications of conventional silicon-based bipolar transistors with p-n junctions is quite difficult even at around 77ēK and becomes unrealistic at liquid helium temperature. However for certain applications in cryoelectronics, just as in "ordinary" electronics, bipolar devices are the most efficient components. Therefore is it worth seeking alternatives to conventional bipolar transistor structures for cryogenic applications.

A promising candidate is the tunnel metal-oxide-semiconductor (MOS) emitter transistor with an induced base which behaves satisfactorily down to 1.6ēK, and, in particular, does not exhibit much reduction of current gain (~20% in the temperature range from 300 down to 1.6K).

This transistor is based on the reversely biased Al/tunnel-thin (~2.5 nm) oxide/ n-Si/ n+ -contact structure ("+" to Si). Unlike well-known MOS capacitors, the application of a noticeable (several Volts) reverse voltage (which is, in transistor's notation, a collector voltage) normally will not invert the Si/oxide interface, as the holes formed due to thermal generation are tunnelling into a metal without gathering near the interface. Instead the application of a collector voltage leads only to the depletion of the underlying semiconductor, so that the tunnel-thin insulator remains almost unbiased (oxide voltage is less than 0.1V), unless there is an external resupplying of holes. The latter may be accomplished by a base current (created by applying a positive voltage to an adjacent p+ base contact used as a source). This results in the formation of rather narrow (2-3 nm) inversion hole layer that acts as a base of the transistor and causes electron injection from the emitter.

There are no majority carriers within the "active" fraction of the device (except the p+ contact) and, moreover, a strong electric field (~1 MV/cm) exists just near the tunnel barrier (near the tunnel emitter). Therefore injected electrons are drifting in this field toward the collector just after their appearance in the semiconductor material. Thus there are no problems with carrier transport, which otherwise one occur at cryogenic temperatures if there is a quasi-neutral base section as in conventional bipolar transistors.

The amplification of the modified device is not expected to be strongly temperature-dependent because the tunnelling currents are almost insensitive to cooling. Therefore the modified transistor structures should prove suitable for low-temperature electronics.

Additionally, tunnel MOS emitter transistors are promising devices because:


Novel Semiconductor Devices

Fast Ionization Dynistor

A Novel Superfast Silicon Switch, the Fast Ionization Dynistor (FID)

Semiconductor research in Russia has lead to the observance of a new physical phenomenon in semiconductors -- wave breakdown. A new closing switch, known as the Fast Ionization Dynistor, has been developed on the basis of this phenomenon. Its chief characteristics are superfast switching (< 1ns current risetime), very low resistance in the early stages of turn-on, and excellent reliability due to uniform device junction operation. Among several promising applications are pumping pulses for copper- and gold-vapor lasers.

The physical process of FID superfast switching proceeds as follows. Initially the dynistor n+pNp+-structure is forward biased and the collector p-n junction blocks the forward DC voltage. Then a positive overvoltage pulse is applied to the dynistor, at a pulse rise rate (dV/dt) of more than 1012 V/sec). Overvoltage reaches a level at least twice that of the DC breakdown voltage because of the very low concentration of free carriers in the space charge region near the pN junction. At the same time, a high displacement current passing through the device produces in the neutral part of the N-layer a rather high electric field which causes lattice impact ionization by majority free carriers (electrons). Holes produced here reach the super high field region near the pN junction for 2 - 3 nsec causing extremely intensive and fast breakdown. High density neutral electron-hole plasma fills this region; the field here is reduced, but increases in the nearby region of the N-layer causing intensive breakdown.

The resulting breakdown wave moves very rapidly to the p+N junction, creating plasma behind the over-voltage front. When this front reaches the p+ region, the turn on process is completed. This entire switching process usually takes less than 1 ns. This turn-on state is maintained by thyristor-type positive feedback, with resistance typically less than 0.1 ohms. This provides a unique opportunity to combine fast rise time with very long pulse duration and very low losses. Because the maximum PRF is usually limited by heat, FIDs can operate at hundreds of Khz in burst mode and dozens of KHz continuously, with good frequency response.

With proper triggering circuits one can construct a pulse generator with low jitter, < 100 ps. Other advantages include compact size and light weight. FIDs exhibit very long lifetimes typical of semiconductor devices, as the wave breakdown process does not lead to filamentation.

These characteristics make the FID an excellent replacement for thyratrons or other tube- or gap-type switches, including in laser applications such as copper-vapor and gold-vapor lasers.


Inverse Recovery Diode

MegaPulse, Ltd. has successfully developed a new semiconductor opening switch with much higher current density than any similar, previously known device:

3000 Amps/cm2 or nearly two orders of magnitude greater than step recovery diodes (SRDs).

The new device is called an Inverse Recovery Diode (IRD), whose physical principle of operation is that a much larger amount of plasma is pumped into the semiconductor structure, during a pumping pulse of 3-5 microseconds, and then the reversal process is initiated (i.e. the plasma is extracted).

This leads to a large voltage multiplying effect: more than 20:1 in an early prototype unit. This unit generates a pulse of 31 kV, with a risetime of 3 ns and a pulse width of about 15 ns (FWHM). This unit operates off a 220VAC wall plug and is capable of 1 kHz PRF without special cooling. This lab-style unit is quite small -- approximately the size of two shoeboxes -- and could be designed to be much more compact if necessary. Another advantage of IRD pulsers is their high efficiency, up to 75-80%.

Pulse generating units of several gigawatts power should be readily feasible; Grekhov's team offers to build a unit of 100 kV into 12.5 ohms, 8kA pulse current, 3-4 ns risetime, 10-15 ns pulse width, and 100-300 Hz rep rate. Please note that this is not the upper limit of IRD technology.

Potential applications include accelerators, chemical processing, and environmental remediation, including water purification, diesel exhaust cleaning, and pulse energization of electrostatic precipitators.

We are actively seeking one or more US industrial partners, plus novel project applications with US government organizations such as the national laboratories.

We would welcome your inquiry and the opportunity to quote on pulse generators to meet your needs.


Reversely Switched Dynistor

RSD-Based Pulsers for Excimer Laser Pumping

Usually pulsed power supplies for power excimer lasers should be capable of forming pulses within 30 - 100KV, hundreds of nanosecond pulse width, units to tens energy, 102 to 103 Hz repetition rate. The promising solution for these power supplies is RSD-based pulsers.

The Reversely Switching Dynistor (RSD) is a novel thyristor-like device with radically improved switching characteristics which are a result of uniform semiconductor structure operation. RSDs are triggered by short (1-2m s) reverse voltage from the main (power) circuit. During this reverse, the short reverse current pulse passes through the semiconductor structure forming the thin electron-hole plasma layer in the collector junction plane. Then the applied voltage polarity returns to the initial state where the plasma layer uniformly distributes the gate electron, which injects majority carriers into the base layers of the RSD's structure. When properly triggered, the RSD can switch very high current, for example -- an RSD with silicon wafers (56mm diameter) is capable of switching ~ 100KA with di/dt=40KA/m s, 50m s pulse width.

RSD-based pulsers for excimer laser pumping usually consist of very reliable low voltage (~ 2KV) high current pulsers (comparable to one RSD), a high voltage pulse transformer and one or two magnetic cells for pulse compression.

In comparison with SCR-based pulsers, the RSD pulser is more reliable because of much better switching characteristics of RSDs. This improved switching allows the use of only one device without any series or parallel connections. Additionally, because of the high di/dt capability of RSDs, the pulse width of the low voltage pulses can be relatively short (3-5m s)and the number of magnetic cells required for pulse compression, in this case, is less than in SCR pulsers, thus providing increased efficiency.