1. Introduction

In low energy physical systems it is often necessary to cool a sample to avoid drowning the phenomenon of interest in thermal noise. This occurs when the thermal energy kBT becomes equal to or greater than the system's energy scale eV. Therefore the requirements on the amplifier becomes more demanding in low temperature experiments. Such experiments demanding temperatures at 4K and below are usually carried out in a cryostat. The sample is mounted inside the cryostat and then connected to a room temperature preamplifier. The preamplifier has an important function; to match the sample to the measurement electronics without adding noise to the signal. The connection between the sample and the preamplifier is done with coaxial cables and/or DC-lines. These cables inevitably acts as distributed capacitors which together with the samples equivalent output resistance makes up a low pass filter limiting the system bandwidth. The cables also pick up ever present ambient noise. By moving the preamplifier inside the cryostat one can reduce these unwanted effects since the cable lengths are reduced. The preamplifier acts as a buffer and match the cables to the sample. It also provides gain to the signal making it less sensitive to noise pick up. Another benefit of a cooled amplifier is the reduction of its internal noise [1] . One type of noise stems from charges moving randomly which modulates the current in the device. By lowering the temperature this random motion is lowered resulting in a lower noise. Other types of noise are shot noise and telegraph noise which are also affected by the temperature of the surroundings.

When choosing the technology for such an amplifier one is limited to GaAs metal-on-semiconductor field effect transistors (MESFET). This is very unfortunate since all commercially available operational amplifiers and almost all discrete transistors are made in Si. Due to its large donor bandgap, Si suffers from carrier freeze-out and consequently its conductance approaches zero at 4.2K and below whereas GaAs can be doped so that a substantial amount of charge carriers are excited at 4.2K [2] . Most op-amps use bipolar junction transistors (BJT). Such transistors rely on minority charge carriers in their base region. Since the minority charge carrier concentration drops at low temperature, these devices have a current gain near one when cooled, compared to 100 and greater at room temperature. Majority charge carriers do not show this strong temperature dependence and devices using them as charge transport mechanism, such as FETs, can be used at low temperatures [2] . As a consequence of the material and device related properties, the GaAs FET family is a candidate for cryogenic usage.

Due to their high mobility, GaAs-MESFETs for UHF and microwave applications are today commercially available. Unfortunately most FETs don't work at 4.2K although there have been certain types, currently not in production, that have been reported to work well. For example, the Plessey GAT 1/010 GaAs MESFET had a spot noise level of 1.5 nV/(Hz)½ at 6 MHz and 4.2K [3] while a UHF amplifier using this device with spot noise of 0.19 nV/(Hz)½ at 430 MHz and 4.2K has been reported [4] . This device is now discontinued. The 3SK166 is a UHF GaAs MESFET today commercially available from Sony Electronics which has been reported to work at cryogenic temperatures [5] . One problem with microwave FETs is their small gate area. This lowers the input capacitance but requires higher drain currents to get as high transconductance as a large gate FET. The small gate also makes the device sensitive to single impurities giving a peaked low frequency noise spectrum instead of the usual 1/f noise. In general, microwave devices are not optimized for 1/f noise at frequencies below some 100 MHz which makes them hard to use in DC amplifiers.

The MESFET should neither be confused with the MOSFET (Metal-Oxide-Semiconductor FET) which is used in all microprocessors today (CMOS technology), nor the HEMT (High Electron Mobility Transistor) which rely on a two dimensional electron gas (2DEG). The MESFET uses a reverse biased Schottky diode whereas a MOSFET uses an insulating oxide barrier. Furthermore the MESFET creates a channel by gate-controlled material inversion as is the case in a MOSFET, but not in the HEMT which uses the 2DEG as channel.

To Top
To Table of Contents
Next Chapter