Geoscience and Remote Sensing Society

Abbreviation: GRSS, S Code 29


IFT Home
Technical Briefs
IGARSS 2000 IFT Papers
Technical Challenges
Meeting Announcements
GRSS Homepage


Chung-Chi Lin
European Space Agency - Earth Observation Group
Noordwijk, Netherlands

I. Introduction

Natural emissions of gas molecules are best measured in the sub-mm-wave and far IR spectral range where the Planck distribution exhibits its maximum. Each gas molecule can be recognized by its spectral signature which varies with temperature and pressure, and its intensity is proportional to the specie concentration. Due to fine spectral lines, which typically constitute those signatures, accurate spectroscopic measurement calls for a heterodyne detection. Radio-astronomers have long been exploiting this technique in order to study the composition of the Universe.

II. Limb Sounding of Earth's Atmosphere in the Sub-mm and Far IR Spectral Range

Limb sounding of the Earth's atmosphere from an orbiting spacecraft has several key-advantages over other measurement methods due to its particular sensing geometry (tangent to the atmospheric limb):

(a) it maximizes the reception of the signal emitted by the atmospheric layer at the viewed tangent height provided that the receiver antenna has a very narrow beam;

(b) the background temperature (that of the deep space) is much colder than that of the atmosphere, which guarantees very low biasing of the measured atmospheric emissions;

(c) vertical profiles with a high resolution can be obtained by limb scanning with a very narrow antenna beam;

(d) a global coverage is achieved with a low Earth polar orbiting satellite.

The major disadvantages of this technique are:

(i) low horizontal resolution due to the measurement geometry (long path-length) and high speed of the spacecraft (signal integration along the moving tangent point)

(ii) high cost of developing and launching a spacecraft.

The lower part of the Earth's atmosphere, the so-called troposphere, is essentially opaque in the sub-mm wave and far IR spectral range because of its higher water vapor content. Therefore, measurements in this frequency range are restricted to the higher part of the atmosphere, namely the stratosphere and mesosphere.

Special requirements are placed on satellite-borne instruments as they are subject to difficult environmental conditions in space:

(i) the thermal control of the instrument is restricted to some range of temperature when no special active control is provided;

(ii) the instrument must have a high reliability in harsh environmental conditions during launch and over several years in orbit;

(iii) the above required reliability must be demonstrated during the development and testing of the instrument before the launch (space qualification aspects);

(iv) the mass and required DC power of the instrument must remain within the limits imposed by the spacecraft. Consequently, those requirements strongly restrict the types of usable technologies for such instruments.

These considerations inevitably exclude some well-known technologies used in the traditional radio-astronomy such as use of SIS (Superconductor-Insulator-Superconductor) detectors. Active cooling of such devices are indeed out of reach for spaceborne instruments from both the on-board resources and reliability points of view, although they represent by far the most sensitive devices in mm and sub-mm-wave radiometry.

III. Technology Status for Heterodyne Receivers and Spectrometers

The current generation of limb sounders (MLS on UARS and MAS on Shuttle) use technologies inherited from the radio-astronomy in the mm-wave frequency range. Typically, GaAs Schottky diode mixers are utilized in receiver front-ends which enable room temperature operations. Best performances have so far been achieved by the combination of a corrugated horn feed and a fundamentally pumped whisker contacted diode mixer mounted inside a waveguide block. There are a number of problems associated with such front-ends as the whole structure is scaled down for sub-mm wave operations. First, the miniaturized corrugated horn and waveguide block are becoming increasingly difficult to manufacture. The size of the mixer diode is accordingly reduced together with its anode diameter, rendering its mounting and the contacting of the whisker rather difficult to achieve. The reproducibility of good whisker contacts is greatly reduced as well as their reliability during the launch of the satellite and operation in orbit. The qualification of such front-ends are also quite involved since no tests of the mixer can be performed before the complete assembly of the waveguide mixer block.

The local oscillator signal is also typically generated with the use of a Gunn diode followed by a multiplier chain that consists of whisker contacted varactor diodes. All the above mentioned disadvantages concerning the receiver front-ends also apply here. Presently, the highest frequency so far achieved using this technology stops at 650 GHz, where an airborne operation has been demonstrated.

An alternative to whisker contacted diode mixer exists in the form of planar Schottky diodes, which are fabricated using integrated circuit processing techniques. Such diodes offer comparable performance up to about 300 GHz beyond which its conversion loss quickly increases due to high parasitic capacitances and series resistance.

The surface-channel / air-bridge technique has successfully been developed at the University of Virginia in order to reduce these parasitics and a good performance has been demonstrated up to 600 GHz.. Those planar diodes do offer several advantages since they are reproducible, structurally rugged, and can be DC-tested on wafer A pair of matched diodes can be fabricated on the same wafer in order to be used for anti-parallel sub-harmonic mixers.

Spectral signatures of target molecules are measured at the receiver back-end using spectrometers. Both MLS and MAS utilize filter banks to isolate narrow frequency channels of interest. Filter banks are well-suited if only a small number of frequency points are to be measured. As the number of frequency points are increased, they become prohibitively heavy and consume more power. More recent alternatives being developed elsewhere are the chirp transform spectrometer (CTS) for narrower band operations, and the acousto-optic spectrometer (AOS) for broader band operations. The digital autocorrelation spectrometer (DAS) is the most recent alternative which makes a maximum use of digital signal processing techniques. It offers excellent stability and accuracy, and narrower band spectrometers have been demonstrated in radio-astronomy. Unfortunately for broadband operations, the required high speed digital operations lead to high power consumption.

IV. Future Needs of Sub-mm / Far IR Spaceborne Radiometers / Spectrometers

The next generation of limb sounders will operate at higher frequency ranges as compared to the existing ones, starting from the upper mm-wave range (approx. 200 GHz) up to the far IR spectral range (a few THz). This upscaling of the lowest sensing frequency would improve the vertical resolution, since the maximum antenna aperture is physically limited on a given satellite. The primary motivation for extending the upper sensing frequency is driven by the desire to measure the OH radical, which is believed to play a key role in the ozone destruction process. This is detectable at 1.8, 2.5, 3.5 THz and above. There are indications that the most appropriate frequency for sensing OH is at 3.5 THz due to lesser contamination of the spectral line emission by the emissions of other species.

The total instantaneous bandwidth of the receivers will also grow in order to enable simultaneous measurements of a large number of target trace gases. This would allow more complete analyses of atmospheric chemistry processes. Several channels with their bandwidth exceeding several GHz are desirable, resulting in a total bandwidth of a few tens of GHz. Consequently, the receiver back-end will need to process very large spectral bands.

(a) Mixer Diodes As the implementation of whisker diodes is increasingly difficult at higher frequency ranges, innovative diode designs must be sought. Typically, integrated circuit processing techniques are most suitable in order to fabricate complete diodes. The on-going efforts at the University of Virginia on planar diodes are concentrating on the reduction of parasitic capacitances and series resistance through appropriate geometrical designs, as well as on the optimization of the doping profile in order to achieve low noise performance. An alternative concept is being pursued at the Technical University of Darmstadt in Germany. A so-called quasi-vertical structure is nearly identical to whisker contacted diodes, except that the actual whisker contact is made of a micro air-bridge. Therefore, one could expect a similar performance to that of whisker contacted diodes. Indeed one can reasonably assume that the geometry of whisker contacted diodes is almost ideal from the parasitics point of view, i.e the parasitic elements are minimized. One could therefore fabricate a complete diode using micro-machining techniques, which opens the door to the concept of integrated receiver front-ends. As compared to the conventional machining of feed horn, waveguide block, and implantation of diode mixers together with IF matching circuits, the complete front-end could be fabricated by several micro-machining steps. Notable examples are the works performed at the University of Michigan in the States, and the more recent TINT IN (Terahertz INtegrated Technology INitiative) consortium lead by the University of Bath in the UK. Since tight tolerances could be maintained in such fabrication techniques, more reliable and reproducible terahertz front-ends could be manufactured. (b) Local Oscillator Sources In order to achieve the lowest conversion loss with Schottky diode mixers, a sufficient local oscillator power is needed. Presently the conventional approach using a Gunn oscillator followed by a frequency multiplier chain (multiple multiplications) quickly becomes inefficient above around 600 GHz.

The following solutions could be envisioned:

1. The fundamental signal of the Gunn oscillator should be optimized in terms of power and frequency.

2. The varactor multiplier diodes should be improved to reduce the conversion losses.

3. Use of multiplier diodes in parallel in order to increase the output power.

Currently, the multipliers of choice for lower sub-mm-wave region is again the whisker contacted diodes. Thus, the suggestions which have been made to the mixer diodes also apply to the diodes used in multiplier chains.

For frequencies above 1 THz, only laser oscillators can provide sufficient power today. Laser oscillators are not very desirable for space-based applications due to their low reliability, short life-time, high input power (a few hundred Watts) and a limited set of available frequencies (set by the choice of gas). More recently, quantum-well l hetero-structure devices such as resonant tunnelling diode oscillators have been investigated for direct signal generation in sub-mm-wave range. Unfortunately, the level of output power is still way below the required one (e.g. 0.7 mW at 750 GHz.).

(c) Spectrometers

The need to measure simultaneously spectral signatures across a bandwidth of a few tens of GHz with a resolution on the order of a few MHz obviously leads to a bulky and power hungry spectrometer unit. Thus, there is a need for very wide-band spectrometers with low mass and low power requirements. The acousto optic spectrometers, which have lately seen extensive developments, have desirable features such as large bandwidth (best result = 1.5 GHz.), compactness and low power consumption. These spectrometers have to be put in parallel by means of a simple filter bank. Space qualification aspects, reliability and stability of such spectrometers are still unknown factors since they are essentially analog devices. Luckily, widespread activities in this technology world-wide will certainly bring an answer to the question of whether they are suitable for space-based applications. A great potential for further improvement can be expected for digital autocorrelation spectrometers (DAS). Currently available DAS's have limited bandwidth and high power consumption. Nevertheless, they possess a number of desirable features such as high accuracy, high stability, flexibility and compactness. Therefore major efforts should be put into increasing the bandwidth and reducing the power consumption.

The most demanding part of such spectrometers is the correlator circuit which has to operate at the clock rate. It has been shown that the signal quantization can be done with low resolution, e.g. 2 or 3 bits, without appreciably degrading the performance. A logical technological solution is offered by implementation in the form of a fully custom ASIC chip, incorporating A/D converter, correlator circuit, registers and FFT circuits. Power consumption could be reduced by decreasing the gate size of the transistors. The first satellite to fly such spectrometer chips is called ODIN (a joint Swedish / French project) which intends to perform both astronomical and limb sounding measurements.