In this chapter we will introduce the FIR laser and measurement systems. The
first section explains why FIR is used and the methods of production. The second
section concentrates on the two FIR laser systems that have been used. These two
facilities are: the nanosecond FIR laser in the Max Planck Institute Grenoble,
and the CW laser at the University of Nottingham. Both lasers are described and
critical comparisons are made between gas flow and sealed-off cavities. The
final part of the section deals with the novel pulse slicing technique used in
the nanosecond laser system.
The next section explains the whole CW system at Nottingham, and describes the method of measurement, and the instruments used. The chapter concludes with a description of the measurement system of the nanosecond facility.
Far infrared (FIR) is usually, and loosely, defined as the region of the
electromagnetic spectrum between about 300 cm-1 and 10 cm-1
(about 30 to 1000 um). The energy of a typical FIR photon corresponds to the
typical energy of many of the electric dipole transitions in IIIV semiconductors
in a reasonably attainable magnetic field (B < 10 T). There are two ways of
investigating such transitions. The first is to keep the magnetic field constant
(ie to fix the separation of the energy levels) and to vary the energy of the
radiation. When the radiation energy, hn, equals that
of the transition an absorption, or a change in conduction, can be detected.
This is shown schematically in figure 1.
Figure1. Shows the method of exciting transitions with interferometer sources. The energy level separation is determined by the fixed magnetic field. The incident frequency is varied over a range and at a resonant frequency the photoresponse of the sample changes.
The other method is to use incident radiation of a constant energy and vary the magnetic field. This has the effect of varying the splittings between the energy levels and at some resonant magnetic field the energy of the radiation equals that of the transition. The resonance, again, manifests itself as a change in absorption or photoconduction. This method is illustrated in figure 2.
Figure 2 Shows the method of exciting transitions when a fixed frequency source is used. The magnetic field is swept, increasing the separation of the energy levels and at a particular field a resonance is observed.
The FIR can be from two sources: a monochromatic source (such as a laser), or an interferometer. An interferometer takes two beams of a light from a broad-band mercury source and combines them: one beam has a fixed path length, and the other has a path length that varies with time (see Kimmitt, 1970, pp 93 128). The consequence is that FIR is produced that has a frequency variable with time. The sample response is Fourier transformed to convert the interferomogram into the frequency domain. Some resolution is necessarily lost in the Fourier transform procedure.
The source of monochromatic FIR is usually a laser. Over the past twenty years or so the FIR molecular laser has been successfully used as a source of over one thousand lines of coherent, high power radiation (Knight, 1982). The power of the radiation produced is several magnitudes higher than that obtainable by other techniques. Button et al (1984) have extensively described the background theory of molecular gas lasers, which therefore, will not be covered in this thesis, save for a cursory description of the present equipment and a few fine points that experience has shown will help in the production of laser lines of greater intensity.
The "interferometric" method is characterised by the low power of the FIR produced, but also by the wide range of wavelengths. The low power necessitates the careful construction of the optics that guide the FIR onto the sample. A laser has the advantage of high powers, but one is restricted in the range of available wavelengths.
Figure 3 shows, schematically, the present FIR opto-magnetic laser system. The magnet is a 10 T Oxford Instruments superconducting magnet. The laser is an Edinburgh Instruments FIR laser system comprising two parts: a PL4 CO2 infrared laser, and a 295 FIR laser cavity.
3.1 The CO2 Laser.
The Edinburgh Instruments PL4 laser produces infrared of 9 or 10 um from the
change in the CO2 molecule vibration states. Further changes in the
rotational state finely tune the infrared produced, and it is possible to
produce some fifteen or so of each of the so-called P and R branches for both
the 9 and 10 um lines.
The laser gas is supplied ready mixed and is composed of 7% CO2, 18% N2, and 75% He: the CO2 is the lasing gas but it is the N2 molecules that are excited by the electric discharge. The population inversion of the CO2 molecules is obtained by collisions with these excited N2 molecules. Since N2 is homonuclear, radiative transitions are forbidden and so this process is quite efficient. Thus the N2 molecules are excited to an energy level of 2329.66 cm-1 and then excite the CO2 molecules to the higher laser level of 2349.3 cm-1. Transitions to the lower vibrational states at 1285.5 and 1388.3 cm-1 give laser emissions at 10.4 and 9.4 um. The He gas is used to depopulate these lower CO2 laser levels, and since He has a high thermal conductivity it helps to prevent over-heating of the discharge tube.
Most of the pump lines can be obtained by using the CO2/N2/He gas mixture at a pressure of 25 mbar and at a discharge current of 22 mA. To obtain some of the lower power pump lines it is necessary to change these parameters. (For example to obtain 9R4 it is necessary to use a pressure of 25 mbar and a current of 15 mA.) The pump line is tuned coarsely with a diffraction grating whilst monitoring the power of the infrared produced and comparing the grating drive reading with the factory test results. Fine tuning is then carried out by adjusting the CO2 laser tube length with a piezo-electric translator. Again the power produced is monitored. To improve further on the power of the pump-line, the gas pressure and discharge current can be finely adjusted.
The PL4 CO2 laser can produce CW or pulsed infrared. For the measurements carried out the conductivity was measured with a phase-locked technique and so pulsed infrared was used. It has been found that a pulse frequency of 70 Hz usually produced the best photoconductive signal.
Figure 3 The opto-magnetic system used in Nottingham. The figure also shows the positioning of the sample and the detector in the magnet
3.2 FIR Laser Cavity General Description.
3.2.1 FIR Gases.
There are many possible molecular gases that are suitable for the production
of FIR. Methanol (CH3OH denoted MA) and formic acid (CHOOH denoted
FA) are used for the CW work and ammonia (NH3) in the nanosecond
work. The molecule is excited into a higher vibro-rotational state by the pump
infrared. The molecule then relaxes to a lower state emitting a FIR photon.
Careful adjustments of the laser cavity length and gas pressure are important
for the production of the desired wavelengths. Different lines can be obtained
if different isotopic forms of the gases are used: by using the heavier
deuterium instead of hydrogen the resonant frequencies of the methanol molecule
For example trideuterated (CD3OH or DMA), Odeuterated (CH3OD or MAD), or fully deuterated methanol (CD3OD or DME) produce ranges of lines not seen with the H-isotopic methanol. Lists of FIR lines can be found in Knight (1982), and the relatively limited number of FIR lines obtained with the Nottingham CW laser are given in appendix 1.
3.2.2 FIR Cavity.
The laser cavity is essentially a Fabry-Perot resonant cavity. That is, it
consists of two parallel mirrors each at the end of a waveguide. The cavity
length can be adjusted by altering the mirror separation. The output coupler
must be chosen carefully to allow only FIR, and not the pump infrared, to leave
the laser. A similar method has to be applied to the input coupler to prevent
the FIR escaping in the wrong direction. The Edinburgh Instruments 295 FIR
cavity (figure 4) uses a quartz window as its output
coupler (quartz attenuates radiation between 4 and 40 um [Kimmitt, 1970, p21]),
and a ZnSe Brewster window as the input coupler. A Brewster window fully
transmits transverse magnetic polarised (TM) light, whilst reflecting part of
the transverse electric polarised (TE) radiation. This ensures that the FIR
produced is TM polarised.
Figure 4 The 295 FIR cavity.
Two systems are possible: sealed off and gas flow. The difference is the method of filling the cavity with the FIR gas: the former requires just one fill, the latter requires the gas flowing through continuously. The sealed off method is used in the Nottingham system, whilst the gas flow method is used as part of the nanosecond system in Grenoble. A brief description of each follows along with their relative advantages, and disadvantages.
3.3 FIR Laser Cavity Sealed off mode.
The Edinburgh Instruments 295 FIR cavity employs the sealed off method. The
cavity is filled at the chosen pressure with the FIR gas and then sealed off
with a high vacuum valve. Inspection of figure 4 shows
that the output end of the cavity has a bellows arrangement this allows the
output coupler mirror to be moved a few millimetres to alter the length of the
cavity and to tune the cavity to be an exact number of half wavelengths of the
In theory the gas pressure cannot be adjusted whilst the laser is in action. In practice, however, it is possible to raise, or lower the pressure by utilising a rather rough, but effective method. This involves opening the high vacuum valve whilst pump radiation is incident on the cavity, to enable the cavity to be pumped out, or refilled, with FIR gas. This procedure is not as careful as that recommended by the manufacturers for initial cavity fill, and naturally allows impurities to enter the cavity, but, as will be discussed below, the contamination so introduced is insignificant compared to contaminants inherent to the system. The control over the FIR gas pressure is not completely satisfactory, but it does allow optimum pressures to be found and thus noted for future use.
The CW pump radiation of about 50W produces CW FIR of above 100mW. Such a power is very high compared to the other methods of interferometers or monochromators.
3.4 FIR Laser Cavity Gas Flow Mode.
The nanosecond facility at Grenoble uses a gas flow FIR cavity. In this
system the FIR gas is constantly flowing through the cavity. Adjustments can be
made to the gas pressure whilst the FIR is being produced, a feed-back mechanism
adjusts the pressure to keep it constant with time. The waveguide and optics
(quartz output window and a ZnSe input mirror) of this particular system are
designed for easy removal, which is facilitated by the absence of high vacuum
seals. This means that the waveguide, or the optical couplers, can be changed in
a matter of half an hour or so. The entire system is more extensively described
elsewhere (Rikken, 1987).
The gas flow cavity could be used with a CW CO2 laser to provide CW FIR. However, since the equipment is required to produce short pulses of FIR, a CO2 TEA (Transverse Electric Atmospheric) laser is used to provide the pump radiation. This produces 100 ns infrared pulses of about 1 MW. The FIR produced is shortened in length using a pulse slice technique, described later.
3.5 A Comparison of the Two Types of Cavity.
The Grenoble cavity can be used to produce CW radiation if a CW CO2
pump laser is used, thus the two types of cavity are compatible. The Grenoble
cavity is "home made", and so perhaps the comparison with a
commercially produced system is unfair. However there are several design points
that should be emphasized.
The sealed off system relies on high vacuum parts to allow the cavity to be free from contaminants for an extended period. The filling manifold parts are joined together with knife-edge steel joints into copper gaskets. The vacuum valves are teflon, except for the one sealing off the cavity which is of the knife-edge high vacuum type. A high vacuum seal at such joints is difficult to obtain and it is best to avoid them when possible. Considerable time was spent in leak testing all the vacuum seals to the point that the cavity, once filled, produces FIR of a reasonable power for about four or five days. After that time the contaminants in the cavity are sufficient to reduce the power. These impurities are introduced through a leak on the cavity itself, rather than in the fill system. Such a leak is difficult to find and seal.
The gas flow system does not require such good seals. All the joints use normal rubber "O" rings. Such joints are easy to make, and are easy to break and then remake. They are more economical: a copper gasket has to be thrown away after one use: a rubber "O" ring is reusable. The system is surprisingly tolerant to changing the FIR gas: requiring pumping out for a few minutes and then flushing a few times. The sealed off system requires at least pumping overnight, and if a gas like formic acid is used, the pumping time may be extended to a day or more.
These two differences are inherent in the cavity type. There are several problems that are specific to the 295 and are design faults. The most important is the gas filling system. The recommended procedure is to pump out the cavity with a rotary pump and then, when a base pressure is reached, to pump out with a vac-ion pump. The ion pump can obtain and hold a pressure of 107 mbar in the cavity. The FIR chemicals used are liquid at atmospheric pressure, and their glass containers are attached to the high vacuum manifold with a rubber "O" ring seal. However, the clip that holds the joint together exerts very little pressure, and consequently the joint tends to leak very badly. There is no possibility of improving on this seal as supplied, and it must be regarded as a definite weak point in the system. It is not possible to fill the cavity without introducing some impurities through this joint.
A further unsatisfactory feature concerns the output coupler of the system which is attached to the bellows arrangement (figure 4), the whole section being held in a cradle. The orientation of the mirror is adjusted with three micrometers mounted on the end plate of the cradle. The whole cradle is moved by the translation micrometer to adjust the cavity length. The problem with this arrangement is that when the cavity length is changed, so is the mirror alignment. Tuning the cavity has the effect of disturbing the optics.
Furthermore, the output mirror is positioned in such a way that whilst one side of the mirror is at atmospheric pressure the other side is at the pressure of the FIR gas (typically 0.1 mbar ie one ten thousandth of an atmosphere). This has the effect of bending the mirror, and hence if the mirror is aligned with one pressure in the cavity, and this pressure is adjusted, then the alignment will also need to be adjusted.
The problem can be overcome by using separate filters as output or input couplers. This method is used on the input of the 295 cavity, where a separate input window (a ZnSe Brewster window) and mirror (of silver coated steel) are used. The FIR cavity in the nanosecond laser system has separate windows and mirrors at each end. The advantage of this is that since the FIR gas is allowed to flow on both sides of the mirror the pressure will be the same on each side, so the mirror would not be bent. The output window will have unequal pressures on each side and may be bent but this is now not so critical since it is not used to reflect the beam. (To a certain extent the focusing power of the output filter is affected by this bending, but this is not critical as the laser produces a diffraction-limited divergent beam of FIR.)
To sum up: the sealed off system allows economical use of bizarre and expensive gases. In practice these are usually avoided, and in any case the cost of a day's supply of FIR gas for a gas flow system is far less than for the cryogenic liquids used in the whole system. In general, however, the gas flow system is the better method of producing FIR. It is easier to use, the construction is simpler, and it is more readily adaptable.
The Nottingham system is used to carry out the CW laser measurements. The system is described below.
4.1 FIR Optics.
In the wavelength range used (between 40 um and 600 um) electromagnetic
radiation behaves in such a manner that light-pipes can be used to direct the
radiation. The usual procedure is to use internally polished brass tubes for
light-pipes since the material is cheap and easy to use. In the cryostat, where
we wish to prevent heat conduction down the walls of the light-pipe, a thin
walled stainless tube is used instead. The long wavelength radiation suffers
multiple reflections off the sides of the tube, and although it can be focused
at the end with a cone attachment, the emerging beam is essentially unpolarised.
FIR is strongly absorbed by water vapour, and so light-pipes are usually
evacuated to prevent attenuation of the beam.
FIR is transmitted well through TPX (Poly-4-methyl-pentene-1) (Kimmitt, 1970, p24), and lenses can be constructed of this material to focus the radiation. Such lenses are not used in this system since the laser power produced is sufficient for the present application.
In the experiment the laser radiation is carried along an evacuated brass light-pipe and into the magnet insert (figure 5). The FIR enters the insert via a black polythene window which excludes room temperature radiation (Kimmitt, 1970, p30), it is then reflected into the cryostat using a front surface evaporated aluminium mirror. The FIR is directed down the steel light-pipe and focused onto the sample with an electro-formed copper cone which has a linearly decreasing diameter and a solid angle of 0.017 sr. Parabolic cones have been shown to focus the FIR better than a straight cone (Winston, 1970), but the FIR power obtained is sufficient with the cone used.
FIR is transmitted well through helium and so it is unnecessary to evacuate the part of the light-pipe that is immersed in the cryostat. Indeed, the helium is encouraged to fill the light-pipe through holes drilled periodically along its length. Mylar and polythene also transmit FIR (see Kimmitt, 1970, p22) and these are used as windows, the former at the bottom of the cryostat to allow the mounting of a detector in the vacuum space, and the latter at the laser end of the light-pipe.
Figure 5 The magnet insert. FIR is directed along the light-pipe on to the sample which is held in the centre of the magnetic field
4.2 Magnetooptical System.
The equipment is shown schematically in figure 3. The
magnetic field is supplied by a 10 T Oxford Instruments superconducting
solenoidal magnet. This can produce up to 9 T when the magnet coil is at 4.2K,
and up to 10 T when the coil is at 1.8K. A lambda plate 'fridge is used to lower
the helium bath to the lambda point by pumping on a copper coil filled with
liquid helium. The pumped helium drops in temperature to the lambda point,
cooling down the coil. The coil then cools the helium at the top of the helium
bath to the lambda point. This helium is now a superfluid and has a very high
thermal conductivity, thus the whole bath soon becomes superfluid.
Within the magnet coil is a 26 mm bore variable temperature insert. This allows the temperature of the sample space to be at any temperature between 1.8 and 60K. The apparatus consists of a non-inductively wound heater, a RhFe temperature sensor and a needle valve to control the helium flow. A digital temperature controller alters the power supplied to the heater in response to the temperature change measured by the RhFe sensor in the sample space. For temperatures above 4.2K the helium flows through the sample space. For temperatures below 4.2K the sample space is filled with helium, the needle valve is closed, and the helium is "pumped".
The FIR is directed into the cryostat with a light-pipe (described in the previous section). The insert light-pipe serves as a mechanical support to hold the sample in the centre of the magnetic field and also to hold the electrical connections to the sample. It has a facility to adjust the angle at which the sample is held with respect to the magnetic field. This is achieved with a gear mechanism driven by a worm screw (figure 5).
Transmission measurements are possible using a calibrated Ga doped Ge detector (Kimmitt, 1987). The detector is very sensitive to magnetic fields and so we employ two methods to reduce the stray field. The first is to increase the distance from the magnet. The sample space extends below the magnet coil and so the detector is mounted at the bottom of the sample space in the inner vacuum space. A mylar window, which transmits FIR, is used to separate the sample space and vacuum space. Mylar is slightly porous to helium but only at temperatures above liquid Nitrogen. The detector is firmly anchored onto a copper block which is thermally in contact with the helium jacket, thus it is held at 4.2K. The second method uses cancellation coils which produce a field proportional and opposite to that of the main coils. The magnetic field at this position is guaranteed by the manufacturers to be less than 0.2 T with the main coil at full field.
Figure 6 The Electrical and control system. A BBC Master microcomputer fitted with an Acorn IEEE interface is used to control the Stanford SR510 lock-in amplifier and the Oxford Instruments power supply unit.
The Ge detector may be removed and replaced with another window to allow light-pipes to transmit the FIR through the base of the magnet and to an external detector. The option exists for a sensitive silicon bolometer to be used. This is housed in a separate cryostat, and has its own integral amplifier.
All the magnet controls can be set via an IEEE interface. The magnet current, and hence the magnetic field, can be read via this interface. The interface facilitates the use of a microcomputer to control the experiment.
5.1 Production of Nanosecond FIR Pulses.
There are two general methods of producing very short pulses of FIR with an
optically pumped molecular laser. The first is to use very short pump pulses of
infrared, and the second is to produce FIR with CW or long pulses of infrared
and then shorten it.
To produce short pulses of infrared it is possible to use a transversely excited atmospheric pressure (TEA) CO2 laser. Pulses of about 100 ns can be produced with this laser, and shorter pulses may be made by a variety of methods such as cavity dumping (Keilmann et al, 1980) or pulse slicing (Corkum, 1985). However, producing FIR from these short, high intensity, infrared pulses is not trivial since the lifetimes of the excited-states of the FIR molecules are relatively long (typically 100 ns).
Two methods of shortening long pulse FIR are cavity dumping (de Bekker et al, 1989) and pulse slicing. Pulse slicing involves a technique of optically switching the reflectivity of semiconductor slices external to the laser in order to define the start and end of the FIR pulse. Cavity dumping involves the use of an intra-cavity slab of semiconductor which is suddenly made highly reflective by an intense YAG laser burst.
5.2 Pulse Slicing.
This technique uses the property of an electron plasma in a semiconductor to
fully reflect FIR. If an intrinsic semiconductor is held at the Brewster angle
to the incident radiation there will be 100% transmission for vertically
polarised (TM) photons which have an energy that is below the band-gap. If the
incident radiation has an energy above the band-gap the energy will be absorbed
and an electron hole plasma will be produced very quickly (Saltzmann et al,
In the Grenoble nanosecond laser system Si slabs are used as the optical switches and a Nd:YAG laser provides the above band-gap controlling radiation. This laser can produce very intense visible light (10 J in 8 ns pulses is possible). By careful synchronisation with the CO2 pump laser the pulse slicing can produce FIR pulses of less than 20 ns duration with very sharp cut-offs.
Figure 7 The nanosecond FIR system used at MPI, Grenoble. The inset A depicts the pulse slice method. Inset B is the polarisation flipper shown in figure 8.
The slicing arrangement is depicted in inset "A" in figure 7. A beam splitter is used to provide two control beams from the YAG laser: the transmitted beam controls the "switching on" of the FIR pulse; whilst the reflected beam travels an extra path length and controls the "switching off" of the FIR. The physical length of this extra path determines the duration of the sliced FIR pulse.
The first Si slab is arranged so that without the YAG control pulse the FIR is transmitted to a point away from the experiment. When the slab is illuminated with the YAG an electron plasma is rapidly formed (Saltzmann et al, 1983), and the slab becomes reflective to the FIR. The FIR is thus reflected to the second Si slab. This slab is held at the Brewster angle to the incident FIR and so without a YAG control pulse the FIR is transmitted to the experiment. Once the delayed YAG pulse illuminates this second slab, it becomes reflective and reflects the FIR away from the experiment.
By simple geometric considerations of the maximum delay path length possible, the maximum pulse length can be calculated to be 20 ns. For some measurements a longer pulse is required (to allow equilibrium conditions to be obtained), and in these circumstances only the second slab is used.
5.3 Nanosecond FIR Laser System.
The entire laser system is shown in figure 7. The
laser cavity uses a gas flow system which utilises a feedback mechanism (a
diaphragm on the inflow tube) to stabilise the gas pressure. Different
waveguides may be used to optimise the intensity of the wavelength required (Sigg,
1985). A "cavity dump" waveguide may be used to produce very short FIR
pulses (de Bekker, 1989a). There are many molecular gases that can be used to
produce pulsed FIR, but the main gas used was ammonia which gave the strong
lines of 292 um, 148 um, and 90.9 um.
Figure 8 Shows the method of "flipping" the infrared polarisation by 90 degrees. It is constructed from a plexiglass block in which are seven Au-Cr mirrors.
The polarisation of the FIR depends upon the polarisation of the pump infrared used. The FIR may be polarised parallel or perpendicular to the pump radiation. However, the FIR switches are orientated at the Brewster angle and require that the polarisation of the incident FIR is vertical. To ensure that this is the case the polarisation of the pump infrared may need to be changed. This is carried out by the polarisation flipper marked "B" in figure 7. The device contains seven mirrors and works on the principle that light of a certain polarisation has its polarisation changed by 90o upon reflection. Figure 8 shows how the polarisation flipper produces this change of polarisation without changing the optical axis.
5.4 Magneto-optical System.
An Oxford Instruments superconducting magnet is used to provide magnetic
fields up to 10 T. The magnet cryostat hold the sample at 4.2K or, if the sample
space is pumped, at 1.8K. Light-pipes are used to direct the FIR in the
cryostat, and the insert is essentially the same as that used in the Nottingham
CW equipment (figure 5).
Transmission measurements can be carried out using a GaAs detector which has a very fast response time (de Bekker et al, 1989b). A fast response necessarily means a low responsivity (Kruse,1970) and hence the detector is mounted directly beneath the sample.
5.5 Biasing and Measurement Systems.
Photoconduction measurements are made by biasing the sample in the constant
voltage mode. In this situation the photocurrent is measured as the change in
potential across a series 50W resistor. However, it
was found that at low temperatures (around 4K) and low fields (<0.5 T) the
resistance of the sample was near that of the resistor so this field range was
ignored. Above this range the sample resistance was more than a kilo-Ohm and it
was assumed that the voltage measured gave a fairly accurate indication of the
Three types of measurement were possible: Shubnikov de Haas; time integrated photoconductivity (long pulse stimulus); and time resolved photoconductivity. For Shubnikov de Haas measurements the conductivity was measured directly from the sample. For photoconduction measurements the signal was first amplified by an AC amplifier which also filtered out the DC component (which is measured in SdH measurements). The long pulse PC signals are detected and averaged by a Box Car and then displayed on an XY plotter. The time resolved photoconduction signals (and TR transmission signals) were obtained by either displaying on a storage oscilloscope (and photographing it) or by averaging with a transient analyser and displaying it with a computer.
6.1 Biasing Techniques.
There are two biasing methods possible either keeping the current constant or
keeping the voltage constant, which method is used depends upon the sample's
characteristic resistance and the instrumentation available.
In the constant current mode the sample is biased in series with a large resistor and the signal is measured as the potential change across the sample. The large resistor serves to limit the current flowing in the circuit, so the current remains approximately constant. The change in sample resistance due to the application of FIR can be calculated from the circuit current and the change in sample voltage.
The constant voltage condition is achieved by using a small resistor in series with the sample, and the signal is measured as the change in potential across this resistor under the action of FIR. A small capacitor across the voltage supply will ensure that it remains constant. In this situation the change in the resistance of the sample will cause a change in the circuit current, and since the series resistor is constant the changing current will change the voltage across it (the small size of the resistor ensures that the voltage change is small compared to the sample voltage). These measurements are used to calculate the photocurrent.
Both systems have advantages and disadvantages, but in the experiments the constant voltage configuration was used for the following reasons. Firstly, we can directly measure the photocurrent, and hence obtain a value for conductivity. This gives a strong feeling for the number and mobility of the carriers involved in photoconduction. Secondly, electric fields have a strong heating effect upon the carriers, which is utilised in some of the phenomena seen in bulk nGaAs and in GaAs/AlGaAs 2DEGs in chapters 3 and 4. Wherever possible, it is necessary to keep the electric field (and hence the sample voltage) constant throughout the measurement.
The measurement procedure used is, in effect, an AC measurement, and so the sample is connected to the measurement circuit with standard 50W coaxial cable. For the optimum measurement the components of the system need to be impedance matched. This, ideally, would mean that the series resistance (in constant voltage), or the sample resistance (in constant current) would need to be near 50W. All the samples used had resistances at 4.2K in excess of several kW, and in most cases of several MW, this meant that only the constant voltage method could be used. In some of the bulk GaAs and MQW measurements the samples had resistances in excess of 100 MW, and any change in photocurrent would result in a very small voltage change across a 50W resistor. In these cases a 100 kW resistor was used to obtain a reasonable noise-free signal. The interdigitated samples produced for the MQW experiments (see chapter 5) had resistances of a few kW, and so a 50W series resistor could be used. As a general rule of thumb, the series resistor was chosen to be around 1/100 of the sample 4.2K resistance.
6.2 Measurement Technique.
It is necessary to measure the sample conductivity both with and without FIR.
Two methods have been employed to do this: the first is a phase-locked
measurement using a lock-in amplifier; and the second is for pulsed measurements
using a standard box car system.
The (quasi) CW FIR system in Nottingham uses a lock-in amplifier to measure the sample photoconductivity. Mechanical chopping of the FIR can be carried out but it was found that the FIR laser produces a higher power if the CO2 pump laser was used in the pulsed mode. The discharge current is applied as a square wave, a sample of which is used as the reference for the lock-in amplifier. Several chopping frequencies have been tried, and it has been found that about 70 Hz produced the best results, both in terms of the FIR power and in signal response. The chopping frequency is not a harmonic of the mains electrical frequency and so we have no problems from that as a source of noise. In some cases, where the photosignal was small, higher frequencies were used (up to 600 Hz) which helped reduce the low power, higher frequency noise. Even in this case, however, the chopping frequencies were very low compared to the response of the samples, so effectively we have a CW system.
6.3 Measurement Configuration.
The electrical photoresponse is measured with the sample biased in the
constant voltage mode, and the current flowing through the sample is monitored
by measuring the change in potential difference across a load resistor using a
phase-locked technique. The arrangement is shown in figure 6.
The load resistor R is chosen to be insignificant compared to the resistance of
the sample. In this situation the bias voltage across the sample is effectively
constant, although the current changes when FIR is applied. The load resistor is
mounted near the sample and so is also at 4.2K. This helps to reduce electrical
noise. A lock-in amplifier measures the voltage change across the sample in
response to the FIR which is pulsed at 70 Hz.
The measurements are usually carried out with the sample in the Faraday configuration, that is with the electric field of the laser perpendicular to the applied magnetic field. This is the easiest configuration, since the laser beam is directed straight down the sample space. Some impurity transitions are only possible if the laser electric field is parallel to the applied magnetic field (the Voight configuration). This configuration has not been used in the present measurements.
The lock-in amplifier is connected via an IEEE standard interface so that measurements can be sent to the computer for storage. The sample bias is provided by a battery and can be varied between 0 and 18 V.
6.4 Computer Control System.
Both the magnet power supply unit and the lock-in amplifier are controlled by a BBC Master microcomputer via the IEEE interface. All data is stored on disc for future reference and for ease of manipulation, for example smoothing and differentiation. Further details of the computer programs can be found in appendix 2.
Button K J, Inguscio M, Strumia F, eds., 1984, "Reviews of Infrared and
Millimetre Waves", Vol. 2: "Optically Pumped Far Infrared
Lasers", Plenum Press: New York.
de Bekker R E M, Claessen L M, Wyder P, 1989a, Appl. Phys. Lett., to be published.
de Bekker R E M, Chamberlain J M, Claessen L M, Wyder P, Stanaway M B, Grimes R T, Henini M, Hughes O H, Hill G, 1989b, submitted to Applied Physics Letters.
Corkum P B, 1985, IEEE J. Quantum Electronics, QE21, 216.
Keilmann F, Koteles E S, 1980, Opt. Quantum Electronics, 12, 347.
Kimmitt M F, 1970, "Far Infrared Techniques", (Psion, London).
Kimmitt M F, 1987, private communication.
Knight D J E, 1982, NPL Report Qu 45.
Kruse P W, 1970, in "Semiconductors and Semimetals" Vol 5, eds. R K Willardson and A C Beer, New York: Academic Press, p15.
Rikken G L J A, 1987, Ph.D. thesis, University of Nijmegan, Netherlands.
Saltzmann H, Vogel T, Dodel G, 1983, Optics Commun., 47, 340.
Sigg H, 1985, Ph.D. thesis, University of Nijmegan, Netherlands.
Winston R, 1970, J. Opt. Soc. Am., 60,245.