IDA. GAMMA RAY SPECTROMETRIC METHODS IN URANIUM EXPLORAnON­AIRBORNE INSTRUMENTAnON

Q. BristowGeological Survey of Canada

Bristow, Q., Gamma ray spectrometric methods in uranium exploration: airborne instrumentation; inGeophysics and Geochemistry in the Search for Metallic Ores; Peter J. Hood, editor; GeologicQlSurvey of Canada, Economic Geology Report 31, p. 135-146, 1979.

Abstract

Gamma ray spectrometry is used in uranium exploration to determine as precisely as possiblethe crustal distributions of potassium, uranium and thorium by quantitative measurements of thegamma radiation from the natural radioisotopes associated with these elements. This applies toairborne and ground surveys and to borehole logging.

In recent years and particularly since 1972, the commercially available instrumentation for suchsurveys has achieved a level of sophistication found only in research laboratories. Technicalspecifications generated by various national agencies involved in uranium resource assessmentprograms have hastened this process in the case of airborne surveys by requiring multichannel analysisusing analogue-to-digital converters, complete sectral recording, atmospheric radon monitoring andby setting sensitivity and overall detector performance standards for the equipment.

The advent of 4 x 4 x 16 inch (l00 x 100 x 406 mm) prismatic-shaped sodium iodide detectors,with a volume equivalent to a conventional 9 x 4 inch (230 x 100 mm) detector, but with only onephotomultiplier tube, has greatly simplified the packaging of multidetector arrays and associatedelectronics, allowing lighter and more compact spectrometers with lower power consumption. It isnow possible, for example, to operate a system with a 2000 cubic inch (32.8 L) volume, complete withfUll spectral recording using a helicopter.

Most airborne gamma ray spectrometry instrumentation is now designed as one component ofmultisensor integrated data-acquisition systems based on microprocessors or minicomputers. This newfound flexibility has encouraged the development of a variety of features such as real-time on-linecorrections for Compton scatter, background radiation, detector gain shift and verification of tapeddata by read-back-and-compare, all of which can be implemented by software.

Electromechanical methods of mass storage such as magnetic tapes and discs are the mostfailure-prone components of any field system. Recent advances in memory technology may bring solidstate equivalents such as bubble memories, into use as practical and economically viablereplacements before 1985.

Resume

Dans la recherche de l'uranium, la spectrometrie de detection de rayons gamma permet dedeterminer de fa on aussi precise que possible la repartition de potassium, d'uranium et de thoriumdans la croute terrestre par des mesures quantitatives de la radiation gamma emise par les radioisotopes naturels de ces elements. Cette technique s'applique aux leves aeriens et terrestres ainsi qu'ala diagraphie par trou de sonde.

Au cours des derna~res annees, particuli(wement depuis cinq ans, les instruments que l'on trouvesur Ie marche pour de tels leves ont atteint des niveaux de perfectionnement que l'on retrouveseulement dans les laboratoires de recherche. Les specifications techniques exigees par diversorganismes nationaux qui s'occupent de programme devaluation des ressources en uranium ontaccelere ce processus dans Ie cas des leves aeriens, en exigeant l'utilisation danalyses multi canalfaites au moyen de convertisseurs numeriques-analogiques, des techniques completes denregistrementspectral, Ie contr61e du radon atmospherique et l'etablissement des normes de sensibilite et derendement global des detecteurs pour Ie materiel.

La mise au point de detecteurs prismatiques a iodure de sodium, de 4 x 4 x 16 po(l00 x 100 x 406 mm), dun volume equivalent a celui d'un detecteur classique de 9 x 4 po(230 x 100 mm), mais avec un seul tube multiplicateur, a grandement simplifie l'assemblaged'ensembles de multi detecteurs et des composantes connexes, permettant ainsi l'utilisation despectrometres plus legers et plus compacts, d'une consommation energetique moindre. nestmaintenant possible, par exemple, de faire fonctionner un reseau complet de 2000 po cubes(32,8 litres), avec tout l'enregistrement spectral, abord dun helicoptere.

La plupart des instruments aeroportes de spectrometrie a rayons gamma se retrouventmaintenant sous forme de composantes de systemes integres de saisie de donnees a multicapteursbases sur des microprocesseurs ou des miniordinateurs. Ce nouvel apport de flexibilite a favorisel'elaboration dun ensemble de particularites comme les corrections en direct et en temps reel pour ladiffUSion Compton, Ie rayonnement de la zone de fond, Ie decalage de rendement du detecteur et laverification des donnees sur bandes par relecture/comparaison, operations qui peuvent toutes etreeffectuees par logiciel.

136 Q. Bristow

Les methodes electromecaniques de memoires de grandes capacites comme les bandes et lesdisques magnetiques constituent les composantes les plus sujettes au defaillances de tout systemed'exploration sur place. Des progres recents dans les techniques de mise en memoire peuvent favoriserl'utilisation, avant 1985, d'equivalents transistorises comme les memoires abuZZes, comme moyen deremplacement pratique et economiquement rentable.

HISTORICAL DEVELOPMENT OF AIRBORNEGAMMA RAY SPECTROMETRY

4 counting channels, each with a single channel analyzerand analogue rate meter

4 channel strip chart recorder

The block diagram is shown in Figure 10A.3. Three ofthe single channel analyzers were set to cover the pulseamplitudes (i.e. energy bands) corresponding to those ofpotassium, uranium and thorium, while the fourth was set tocover an energy range encompassing all three, the "totalcount".

MgO reflective powder

Nal(TI) Crystal

Photomultiplier tube(P.M.T.)

Charge sensitivepreamplifier

Gamma ray scintillation detector.Figure lOA.l.

Photo Optical coupling greasesensitive --rl~~~~~~~:::::::::cathode Optical window

Highvoltagesupply

Various systems based on the above concept wereproduced in the years following and they found wideacceptance in the exploration industry. A lack of standards,and in many cases a lack of understanding of the principles ofthe new technique, caused difficulties in the collection andinterpretation of spectrometric data. It also became evidentthat a gap had developed between the level of instrumentaltechnology in the equipment used for uranium prospecting andsimilar equipment which was available in the laboratory atthat time.

Since then and particularly during the last five yearsthe situation has changed dramatically with the disseminationof more information and the development of more sophisti­cated techniques and instrumentation, the latter beingcovered in some detail below.

No attempt is made in this article to describe availablecommercial instrumentation. For this information the readeris referred to an excellent annual review which appears in theCanadian Mining Journal (P .J. Hood, 1967, et seq). Thepurpose here is to acquaint the reader with currently usedtechniques, their advantages and their limitations.

While the radiation physics aspects of gamma rayspectrometry are covered in detail elsewhere in this threepart article, the spectrum from a standard calibration pad(Fig. 10A.4) is included here as a convenient aid for discussingthe instrumental aspects of the technique.

FUNDAMENTAL CONSIDERAnONS

150 x 100 mmoneusuallydetector,Scintillation(6 x 4 inch).

There are three radioelements in the terrestrial crustwhich are readily detectable by the gamma radiation emittedby the natural radioisotopes associated with them, potassium,uranium and thorium. However, until the early 1950s therewas no method of distinguishing between different gammaradiation energies. The standard radiation detector, theGeiger-Muller counter, provided only a total count indication.Thus it was not possible to distinguish between potassium,uranium and thorium. The advent of the thallium-activatedsodium iodide scintillation detector was a major advance fortwo reasons, it had a stopping power (i.e. detectionefficiency) several hundred times that of the Geiger-Mullercounter for the energy range of interest in this applicationand the output pulses it produced had amplitudes proportionalto the energy of the gamma radiation which caused them. Byadding electronic circuitry capable of sorting and countingpulses according to their amplitudes it was possible toseparate the contributions from the three radioelementsaccording to their energies, which is by definition gamma rayspectrometry.

The essential details of a scintillation counter and itsmethod of operation can be understood by reference toFigure lOA.l. Gamma rays which interact in the crystal causetiny flashes of light (scintillations) the intensities of whichare proportional to the energy deposited in the crystal by thegamma rays. The photomultiplier tube which is opticallycoupled to the crystal, usually wi th transparent grease,converts the scintillations to corresponding electrical signalswhich can be amplified and sorted in the subsequentelectronic circuitry.

The technique of sorting pulses according to theiramplitudes is known as pulse height analysis. The mostelementary method is to use a single discriminator circuitwhich allows all pulses above a certain preset amplitude(i.e. energy) to be counted and rejects all others, or viceversa. A slightly more sophisticated arrangement uses twosuch discriminators which can be preset to different levels.Pulses having amplitudes which fall between the two levelsare counted while all others are rejected. This arrangementis known as a Single Channel Analyser (SCA) and is shown inFigure 10A.2.

Until the mid 1960s airborne gamma radiation surveyswere largely confined to "total count" measurements. Thefirst gamma ray spectrometry systems for airborne use whichbecame available at that time consisted broadly speaking ofthe following items:-

Gamma-Ray Instrumentation 137

Figure lOA.3. Simple four-channel spectrometer, widelyused for airborne surveys until about 1975.

Inaccurate or varying window width settings.

Excessive noise in the pulse amplifying and shaping chaincausing an additional statistical variation in pulse heights.

Spectrum distortion at high counting rates.

Pulse height distortion due to detector overload by highenergy events from cosmic sources.

It should be noted that even if these pitfalls are avoidedand the objective indicated above is achieved, the measure­ments still require correction for the following factors: -

Spectral interferences; due to the physical processesinvolved in the interaction of gamma radiation with thedetector material and to the introduction of statisticalnoise by the photomultiplier tube, there is a mutualoverspill or crosstalk between the closely adjacentpotassium ('OK) and uranium eI'Bi) windows and a con­tribution to both of them from the thorium CZ 0 8 Tl) window."Stripping ratios" are experimentally determined andapplied to recorded data to correct for this.

Background radiation due to sources other than the crustalradioelement content. These include atmospheric radon,cosmic sources and radioactive material in either theequipment itself or the aircraft used to carry it.

Atmospheric attenuation.

The theory behind these corrections and the methods ofapplying them are dealt with in more detail elsewhere in thisarticle.

The instrumental techniques and hardware which arenow coming into use in modern exploration gamma rayspectrometers are discussed here with reference to thefundamental considerations outlined above For conveniencethey are separated into groups as follows: -

Detectors

Signal-conditioning electronics

Pulse height analysis

Spectrum stabilization

Data display and recording

Detectors

The scintillation detector is still the most widely usedtype for airborne and ground follow-up equipment and appearslikely to remain so for some time. Figure lOA.l shows thecrystal material labelled as NaI(Tl), which is an abbreviationfor sodium iodide (thallium-activated). While this is thematerial most commonly used in scintillation detectorsdesigned for uranium exploration applications, it is by nomeans the only one. Other inorganic crystals sometimes usedin borehole logging applications are cesium iodide, activatedwith either sodium or thallium [CsI(Na), CsI(TI)] and bismuthgerminate (Bi ,Ge 30 12). The cesium iodide crystals have adensity which is approximately 25 per cent greater than thatof sodium iodide with a correspondingly higher efficiency orstopping power for sensing gamma radiation; however theability to resolve spectral peaks (resolution) is somewhat lessthan for sodium iodide. Bismuth germinate has a density verynearly double that of sodium iodide, but the resolution ofdetectors made from this material is barely adequate atpresent for spectrometry. An entirely different class ofscintillators is that based on the use of organic plastics.These materials have low atomic numbers and hence very lowefficiencies for the complete absorbtion of gamma rays in the

CURRENT LEVEL OF TECHNOLOGY FORGAMMA RAY SPECTROMETERS

------? To counter or ratemeter

Single channel analyzer.Figure lOA.2.

Kev

0 400 0 1360 0 1660 0 2420Single

channel0 0 0 0analysers 2820 1560 1860 2820

::;; K U Th

Analogue CiJ IT] m f;:]ratemeters

LEetector

rs;;-intillation - IPreamplifier

From the strictly instrumental point of view thepurpose of a radiometric survey is to record accurately theradiation intensities (count rates) at the detector in predeter­mined energy bands or "windows". Instrumental factors whichdegrade or cause errors in this process can be summarized asfollows: -

Inadequate counting statistics for reliable quanti tati vedata.

Inadequate detector resolution (i.e. inability to resolveenergy peaks sufficiently well for satisfactory pulseheight analysis) due to poor quality crystals or defectiveassembly.

Detector gain drift, i.e. systematic variations in the pulseheight corresponding to scintillations generated by a givenenergy.

Detector pulsesIn ----;.

138 Q. Bristow

0.5-3.0 MeV region which is the useful range in uraniumexploration applications. While they are widely used inlaboratory work as active shields and in related applicationswhere the requirement is for detecting and timing radiationevents as opposed to energy measurements of such events,their very limited ability to resolve spectral peaks coupledwith their poor efficiency renders them unsuitable for use inuranium survey work. Detectors made from plasticscintillators are much less costly than their inorganic crystalcounterparts and some workers have conducted experimentsto evaluate them as a less expensive alternative for airbornegamma radiation surveys (see e.g. Duval et aI., 1972).

Experiments and studies have been conducted toevaluate the potential for airborne work of intrinsicgermanium solid state detectors which have .a resolution farsuperior to that of scintillation detectors. These indicatethat the advantage of high resolution does not offset the costof an array of sufficient volume to match the detectionsensitivity of currently used scintillation detectors. Therequirement for cryogenic operation with liquid nitrogen alsopresents an operational problem with solid state detectors. Itseems probable however that solid state detectors will beused in borehole gamma ray spectrometry. In this applicationthe detector volume is limited by the borehole tool diameterwhich in turn severely limits detector stopping power orefficiency for energies much above 1.0 Mev. Consequently itwould be an advantage to have a detector capable ofresolving the closely spaced lower energy peaks alsoassociated with uranium and thorium.

For portable spectrometers single scintillationdetectors with one photomultiplier tube (PMT) and a76 x 76 mm Ox3 inch) crystal are still the standard unit. Inairborne systems arrays of large detectors are in commonuse, and usually these are offered as add-on-modules of16.4 L (1000 cu. in.) pre-packaged with suitable thermalinsulation and shock mounting. Until recently the individualdetectors have been of the traditional cylindrical type 100 to130 mm thick by 150, 179, 203, 229 or 279 mm indiameter (4 or 5 inches thick by 6, 7, 8, 9 or 11 inches indiameter). The number of PMTs per crystal vary from threeon the smaller ones to four on the 9 x 4 inch to seven on the11 x 4 inch crystal. The reason for having more than onePMT on large crystals is to ensure more nearly constant lightcollection no matter where in the crystal a scintillation ofgiven intensity occurs. This requires that each PMT be fittedwith a separate gain control and that they be equalized byexperiment to optimize the detector resolution, a tedious andtime consuming task.

During the last three years a detector has becomeavailable in a prismatic configuration with a 100 x 100 mm(4 x 4 inch) square cross-section by 406 mm (16 inch) longwith a single PMT mounted on one end. This has become verypopular for airborne systems as it lends itself to compactpackaging with an array of four or six side by side forming a"slab". Figure 10A.5 is a photograph showing such anarrangement.

43039

o

>....c

1000900BOO

(208Tt){2615 Kev

700600

Na I detector volume 3,054 ins3(Twelve 9" x 4" detectors): 50,046 cm3

Counting time: 10 live minutes

Calibration pad dimensions: 25 It. x 25 It.7.62 mx 7.62 m

Approximate pad-detector distance: 2 It.0.6m

Calibration pad composition: K 2.21%U3.0 ppmTH 26.1 ppm

SOD

CHANNEL NO.

400300zoo100

zce:I:U......VlI­z:::>ou

'":>

~......VlI-Z:::>ouooo~

Figure lOA.4. Typical natural radioactivity spectrum.

Gamma-Ray Instrumentation 139

Standard method of measuring detector

Energy-

W= Full width at halfmaximum height

(FW.H.M.)

137Cs single energypeak at 661 Kev

Detector resolution given by:

(~X100) %

Figure IOA.6.resolution.

Figure IOA.7. The PHOSW1CH (phosphor-sandwich) Detector.

Different scintillation decay times in Csl and Nal enable onlygamma rays which are absorbed in the Nal core (A in figure)to be recognized and accepted on basis of pulse shape. Thisgreatly reduces the Compton Scatter interference normallygenerated by analyzing and counting partially absorbedgamma rays.

The partially absorbed photons which appear in lowerenergy windows after the pulse sorting process, are reducedin a new type of detector known as the "Phoswich" detector,the name being a mnemonic for "phosphor-sandwich". Theprinciple is illustrated in Figure IDA. 7. The crystal is in aform similar to that of an iced cake with the "cake" beingmade from sodium iodide and the "icing" from cesium iodide.The decay time for scintillations in the two materials isdifferent by about 4:1, so that it is possible to differentiatebetween photons which have been completely absorbed in thesodium iodide central portion, (the active part of thedetector) and those which have been absorbed by both parts,i.e. those which have been partially absorbed and scattered.

Figure IOA.5. Typical array of six 100 x 100 x 406 mm(4 x 4 x 16 inch) Nal detectors installed in a containerdesiflned to provide both thermal and electrical shielding.esc 203492-K

More recently an actually slab-shaped detector hasappeared measuring 279 x 279 mm by 100 mm(11 x 11 x 4 inches) thick with four PMTs mounted on theupper surface. It remains to be seen how widely this type willbe used.

One of the standard measures of performance of ascintillation detector is the resolution of the 661KeV singleenergy peak of the isotope 137 Cs. This is a measure of thesharpness of the photo peak and is by implication a measureof the ability of the detector to resolve two closely spacedpeaks. Figure 10A.6 illustrates the way that detectorresolution is calculated and specified as a percentage.Individual detectors used in an airborne array such as the100 x 100 x 406 mm (4 x 4 x 16 inch) size and others of com­parable volume should have resolutions of 9.5 per cent orbetter (i.e. less).

A gamma ray in the 0.5 - 3.0 MeV range interacts withsodium iodide by a series of interactions with electrons in thematerial, during which most of the energy is converted to aseries of virtually simultaneous scintillations. If all of theenergy of an incident gamma ray is absorbed by the detectorin this way then the resulting composite scintillation willproduce a pulse height corresponding to the full energy.However if after a number of energy absorbing collisions thegamma ray, now reduced in energy, escapes from thedetector, then the pulse height produced by the event will beidentified as being of a lower energy and will not be countedin the proper energy window. The process is analagous todropping a bead of mercury into a beaker; it shatters into alarge number of small droplets which if all remain in thebeaker will add up to the mass of the original bead. If someare "scattered" over the top then the mass will be less thanthat of the original bead.

AC

B

Scintillation decay times

A~ Accept

B~ Reject

C~Reject

140 Q. Bristow

Pulse shape discrimination circuitry is used in the subsequentelectronics to reject all events which have occurred eitherwholly or partly in the caesium iodide "icing" and accept onlythose which have been completely absorbed in the sodiumiodide. The "Phoswich" detector is still in the experimentalstage and it remains to be seen whether the advantages willoffset the additional complication and expense.

A new type of photomultiplier tube, presently at anearly stage of development, makes use of a silicon diode toreplace the dynodes used in the conventional photo multipliertube. Scintillations liberate photoelectrons from a photo­cathode deposited on glass in the usual way. These are thenaccelerated by a potential of the order of 10 kilovolts andfocused onto a silicon photodiode which produces charge at aneffective rate of approximately one electron per 3.3 eV(electron volts) of energy of each of the acceleratedelectrons from the photocathode. Since each of these has anenergy of about 10 keV, approximately 3000 times as muchcharge is produced by the silicon diode as was liberated fromthe photocathode. This effective multiplication, whilenowhere near as large as with a conventional multi dynodearrangement, is sufficient and is produced in one stepcompared to the ten or more steps using a conventional tube.Consequently the statistical variation in the number ofelectrons produced per photoelectron liberated from thephotocathode is less, offering the potential of improvedenergy resolution. However a high-gain low-noise charge

preamplifier is then required to boost the signal to the levelnormally produced by a conventional photomultiplier tube.Early indications are that this technique is capable ofproviding a useful improvement in scintillation detectorresolution over what can now be obtained with ordinaryphotomultiplier tubes. Additional advantages are a reduceddependence of the overall detector gain on high voltage andambient temperature variations.

Signal-Conditioning Electronics Circuitry

This is the term generally used to describe the processof extracting, filtering or amplifying analogue signals so thatthe information they contain can either be digitized or feddirectly to some form of visual or audio device with theminimum of unwanted noise. Examples would be thedemodulation of stereo or colour TV signals, or the removalof d.c. levels from strain gauge signals to obtain an outputsuitable for driving a strip chart recorder. In this case theobject is to convert the tiny charge signals which appear atthe anode of the PMT of a scintillation detector into voltagepulses whose amplitudes are proportional to the energies ofthe gamma ray events which caused them.

Well-designed systems have a charge sensitive pre­amplifier for each detector, usually located in the detectorpackage, which converts the charge pulses to voltage pulsescapable of driving a reasonable length of line. This isfollowed by a main ampli fier containing special filter circuitswhich shape the pulses to give an optimum signal to noiseratio. Figure lOA.S shows the arrangement and how the zerolevel or base line between pulses following the preamplifier ispoorly defined and noisy, but is considerably improved afterthe filtering and pulse shaping in the main amplifier. Thereare a number of excellent main amplifiers available in thepopular "NIM" (Nuclear Instrument Module) package fromlaboratory instrumentation manufacturers. The generallypreferred kind are those which produce Gaussian-shapedpulses. Most main amp Ii fiers designed for laboratory work inthe NIM package are more than adequate for explorationspectrometers and have provision for "base line restoration",an important feature when moderately high counting rates(greater than 20 000 cis) are likely to be involved.

The need for base line restoration can be understood byreference to Figure lOA.9. At some point in the signalconditioning chain the pulse train will be passed through acapacitor to remove a d.c. level, following which the averaged.c. level will always be zero. This means that at high countrates with a large number of pulses causing a net positiveshift, the base line between pulses settles down to a slightlynegati ve level to maintain the overall average at zero. Thesubsequent pulse-height analysis circuitry however does notrecognize the negative shift of the base line, but sees all thepulses apparently reduced in amplitude and sorts them intolower energy slots than they should be. In other words at highcount rates the entire spectrum is shi fted to the left causingpeaks to move out of their windows by an amount which isdependent on the count rate - a very undesirable effectindeed.

This effect can be compensated for in a variety of wayswhich are collectively termed base line restoration and asindicated earlier most good quality main amplifiers now havethis feature incorporated.

The mul tidetector arrays used for airborne surveysrequire that the pulse signals from each of the individualdetectors be adjusted so that the pulse amplitudes for a gi venenergy are the same from each one. The outputs from all thedetectors are then summed together at some point betweenthe preamplifiers and the main amplifier. The circuitry forthis is normally in a separate module with separate gaincontrols and switches for each detector to facilitate gainmatching.

A modern airborne survey system with a detectorvolume of 50 L 0000 cu. in.) should have an overall resolutionat the 137CS energy of at least 12 per cent (FWHM);although with good quality detectors and carefully designedsignal conditioning electronics much better resolution isachievable.

Pulse-Height Analysis

The simplest form of pulse height measurement device,the Single Channel Analyzer (SCA) described earlier, acceptsdetector pulses from the main amplifier and puts out a logicsignal whenever a detector pulse falls between amplitudelimits set on two front panel controls. The most seriousproblem with SCAs is the difficulty in determining whatenergy limits the calibrated front panel controls actuallycorrespond to. In the case of portable four-channel spec­trometers for exploration work, a fifth channel is usuallyincorporated preset to cover the energy peak from an isotopesuch as the 661 KeV peak of 137CS, to allow for manual cali­bration. The other window limits also have preset controls notaccessible on the instrument front panel. Initial settings ofthese is virtually a factory adjustment, requiring a reliablenuclear pulse generator with calibrated controls and provisionfor varying the pulse shape to match that of the detectorbeing used. Most SCA discriminator circuits are sensitive topulse shape to some degree as well as pulse height, so thatinitial pulse generator calibrations by the manufacturer canbe in error if this point is not appreciated.

Once the initial calibration is done the accuracy of thesubsequent measurements by the user (perhaps over manyseasons), depends entirely on all of these preset controlsmaintaining their relative positions corresponding to therequired window limits.

For many years so called "multichannel analyzers" havebeen available for making much more accurate pulse heightmeasurements. These require an analogue-to-digitalconverter (ADC) which is similar to a very high speed digitalvoltmeter but with no visual display of the numbers itproduces. The amplitude of each incoming detector pulse is

Gamma-Ray Instrumentation 141

Preamplifier

Xlal

Pulse shaping Amplifier

Figure lOA.B. Signal conditioning.

Well-designed signal conditioning electronics are essential for accurate pulse-height analysis.

Low count rate

Oetector pulse heights and gamma ray energies(~~~~------"A,---------,A,,----Zeroo Negligible base line shift

Hi9hCou'-"nt"-raL./te..LL/C./LL./'-£J-Lo..-. (1)1S;l~~~_ ::::::;:Base line =Zero

Base Ii ne =Zero

I

Figure lOA.9. Baseline shift.

If the detector pulse train is passed through a resistor-capacitor network (centre), the baseline shiftsdownward to maintain equal areas above and below the true zero level. Since pulse heights aremeasured from the true zero, this shift can cause serious errors at high count rates causing the entirespectrum to be shifted to the dashed line position (bottom).

142 Q. Bristow

ADC

By way of example suppose that it was only required tosort the incoming detector pulse into one of sixteen channelsrather than 256. Suppose further the particular pulse beinganalyzed had an amplitude of 13 on this scale. The ADCrequires only four yardsticks to sort any pulse into one ofsixteen channels, their lengths (in arbitrary units) being 1, 2,4 and 8 units. Beginning with the longest (8 units) it wouldfind that the pulse was larger and subtract the equivalent of 8units. It would then compare the remainder with the 4 unityardstick and still find a mismatch. Having subtracted off 4more units it would test the remainder against the 2 yardstickwhich would be too large. Finally the 1 yardstick would beused to find a match. The height of the pulse would thus becomputed as 8 + 4 + 1 = 13 units.

As to the ADCs themselves there are two distinctprinciples of operation for converting the pulse height into anumber proportional to it. One is known as the "WilkinsonRamp" method and the other as the method of "Successi veApproximations". Since there has been some controversy asto which is most suitable, their relative merits are worthsome discussion.

The Wilkinson ramp principle is illustrated inFigure 10A.ll and is analogous to the "hour glass" principle.Imagine an hour glass being filled with sand to a level whichis equal to the height of a detector pulse. If the hour glass isthen inverted, the time the sand takes to run out is thenproportional to the pulse height. The number of seconds ofelapsed time could be interpreted as the channel number forthat pulse height (gamma photon energy). Note that if asecond pulse arrives during the time the hour glass is runningfor the first one, the second one is ignored and neitheranalyzed nor counted. This is the ADC "dead time" and in theWilkinson ramp type it is proportional to the height of thepulse being analyzed, plus a fixed time for some logicaloperations in the ADC

In electrical terms a capacitor is charged to the heightof each detector pulse and then discharged linearly to zerowhile pulses from a high frequency oscillator (50 to 200 MHz)are counted. The number in the counter at the end of thisprocess is then taken as the channel number of that detectorpulse.

or plotted on hard copy and provides complete information onthe sources of radiation which were detected (within thelimi ts of the detector resolution).

The second method is fairly straightforward and can beimplemented without any memory, with the necessarycircuitry being duplicated to produce as many "digitalwindows" as required. The advantage over the equivalentSingle Channel Analyser arrangement is that all the windowlimits are set by specifying channel numbers generated by theADC, and these have a fixed and predictable relationship witheach other, since each channel spans a known energyincrement.

The successi ve approximations method can be under­stood by reference to Figure lOA.12. Essentially it works bycomparing the incoming detector pulse height to a series ofyardsticks of varying lengths on a trial-and-error basisstarting with the longest yardstick. If the pulse amplitude isgreater than this then an amount equal to this yardstick issubtracted and the next shortest one is compared to theremainder; if this is too long the next shortest is tried. Theprocess continues until a match is found. The circuitry theninfers from the particular yardsticks which were used tosubtract pieces from the incoming pulse, what the channelnumber is.

Channel number(Core memory locations)

1 3 7 9 13 15

~~~I Digital windows I

CoUNT

l40;2-JJjt;±;atl~~So 1 2 3 4

I

measured by the ADC which assigns it a number between 1and some maximum. In laboratory systems this could be ashigh as 8192 if high resolution solid-state detectors are beingused; in airborne survey systems 256 is usually the maximum.The ADC thus has the capability of grading or sortingincoming detector pulses according to amplitude (i.e. energy),into one of 256 or more slots or channels. This is equivalentto 256 Single Channel Analysers in one box, except that tomake use of all this information we now need 256 counters tokeep track of the number of detector pulses that areaccumulating in each channel. From this point there are twoalternative methods for handling these data as indicated inFigure 10A.IO.

Use 256 locations of a minicomputer memory orequivalent device and increment each location when itsnumber is "called" by the ADC

Use digital circuitry that will increment a single counterfor all numbers generated by the ADC between certainpreset limits corresponding to a desired energy window.

The first method produces a representation of thegamma ray energy spectrum in the memory which is simply ahistogram of energy divided into 256 discrete steps, showingthe number of gamma photons which fell into each channelduring the counting period. This can be displayed on a CRT,

ADC measures height of each incoming pulse, in this case ona scale of 16 and increments contents of correspondingmemory locations by one to build up a 16 channel spectrum orhistogram. By using more channels a smoother curve can beobtained corresponding to the dotted line. If only countingwindows are required then counters can be connected to theADC via a simple interface and arranged to count all pulsesin preset ranges of channel numbers - no computer ormemory is required.

Figure lOA.10. Advantages of an Analogue/Digital Converter(ADC).

Gamma-Ray Instrumentation 143

the successive approximations type the dead time is a fixedlength, whereas it is proportional to the amplitude of thepulse being analyzed for the Wilkinson-ramp type. Acorrection can be made in the former case by multiplying thetotal number of pulses counted in say one second by theknown conversion time (a few microseconds) and increasingthe contents of all channels pro rata to compensate for thelost time. A hardware method applicable to both fixed andvariable dead time ADCs involves two time counters, one ofwhich is interrupted for all intervals when the ADC is busy(i.e. "dead"). At the end of each sample time the differencebetween the contents of the two counters is a measure of thedead time for that record and can be used to make an on-linecorrection to the data just acquired, or simply recorded ontape along with other data for use in off-line data processing.

This is a much quicker process than the clock pulsecounting required by the Wilkinson ramp type. For examplein order to sort a pulse into one of 256 slots, (including zero)only 8 yardsticks are required with at most only 8 compar­isons. The difficulty arises in making accurate comparisonsand subtractions of the yardsticks, (voltages developed acrossprecision resistors in fact) with and from the incomingdetector pulses.

In the case of a 256 channel successive approximationsADC, a detector pulse having a channel number of 128 will beanalyzed after applying the very first yardstick, although theremaining seven comparisons will be made anyway becausethe logic is simplified by so doing, but one having a value of127 will require the successive subtraction of voltagescorresponding to all of the 8 yardsticks before it isdetermined. This places very stringent conditions on theaccuracy of both the yardsticks and thesubtraction/comparison circuitry used if the effective widthsin terms of energy increments for the pairs of channels whichoccur at "yardstick change" points are to be equal.

The advantages and disadvantages of the two types ofanalogue-to-digital converter can be summarized asfollows: -

Successive Approximations Type

rapid conversion time which is the same for all detectorpulses regardless of ampli tude

high probability of sharp discontinuities in adjacentchannel widths particularly at 1/4, 1/2 and 3/4 of fullscale due to the nature of the conversion process

Hour glass

(l)[ ! , I

10 15

Time20

QI

24

Wilkinson Ramp Type

Conversion time which is proportional to the amplitude ofthe pulse being analyzed plus a small constant and whichis of the order of twice as long as for the successiveapproximations ADC at full scale.

Differential nonlinearity, i.e. variation in channel widthsacross the range, far less than for the successi veapproximations type due to the inherently linear "hourglass" principle of conversion.

Universally used by nuclear laboratories where highprecision is required and where the constraints of highcount rate and limited available counting time are not aproblem.

As a result of many years of refinement, well-designedADCs of this type are available off-the-shelf in one ortwo width NIM (Nuclear Instrument Module) packages atrelatively modest cost from a number of sources.

So long as the scintillation detector with its com­paratively poor resolution is being used with only256 channels to cover the spectral region of interest, thenthere is probably little to choose between the two typesunless the data are to be used for very precise spectralanalysis. However if solid-state detectors come into vogue asthey probably will for borehole logging, requiring 2048 or4096 channels for adequate spectral information, then thesuccessive approximations ADC will require considerablymore development to achieve the necessary differentiallinearity.

ADC "dead time" is a factor which requires some formof correction no matter which type is used. The dead time isthe interval during which the ADC is making a conversion andis unable to accept any other detector pulses. In the case of

Imagine an hour glass filled to the height of an incomingdetector pulse. The time for the sand to run out would thenbe proportional to the energy of the corresponding gammaray. A measurement of the time in appropriate units(microseconds in practice) is then a digital representation ofthe energy.

Figure lOA.ll. Principle of the Wilkinson Ramp type ADC.

4

Pulse height~8+4+1= 13

8

The amplitude of an incoming pulse is digitized by comparingit with a series of precision voltage "yardsticks". In the caseshown four comparisons only are required to determine theamplitude as a number on a scale of 1-15. A more precisedetermination on a scale of 0-255 would require eightyardsticks with relative values 1:2:4:8:16:32:64: 128.

Figure lOA.12. Principle of Successive Approximation ADC.

144 Q. Bristow

Error signal

Thorium

AMP

Potassium Uranium

\

\~

\ I I~, I I

\ I I\~, I I

, I

',-J (214Bi) (208rl)

/1760Kev ~2615Kev

_/~--t---+t-t--.. ~ =>J

(description to be published), the 4°K peak position ismonitored and the counting window positions are recomputedperiodically to correct for any dri ft.

Datu Acquisition: Display and Recording

More OftRIl than not the gamm8 ray spectrometer is butone component of a multiparameter system in airborneapplications and some complexity is necessary in handling theincoming data in order to display and or record it in anorderly manner. Until recently this function was performedby hardwired lugic controllers with operator entries beingmade via thumbwheel switches.

When count. rates in windows A and B are equal the errorsignal is zero. When the count rates differ a positive ornegative error signal is generated according as A>B or A<Bwhich alters the high voltage in such a way as to move thepeak back and restore symmetry. A stabiliZing peak mustalways be present for such an arrangement and can begenerated by a pulsed light source or radioisotope implantedin lhe detector, or by using an external source.

Figure IOA.14. Spectrum Stabilization.

Figure IOA.13. The dashed line shows what happens to thespectrum if the detector gain or subsequent amplifier gainsdrop. [n this extreme case the 2 I 4 Bi peak has moved into the4 °K window. Unless drifts of this sort are quickly spotted bythe equipment operator, useless and misleading data may beused in good faith to complete radiometric maps or makeexploraton decisions.

Spectrum Stabilization

All spectrDmeters ore vulnerable tD spectrum drift, i.e.the pDssibility that the detectDr pulse height correspDnding toa given energy changes over a period Df SDme hDurs. Thismanifests itself as a unifDrm scale change on the horizontalaxis of the spectrum. Figure 10A.13 shows the effect of areductiDn in pulse height causing the spectral peaks to shifttD the left and off the centres of the counting windows.

The major sources of spectrum dri ft are the gains of thePMTs which are nDtoriously temperature dependent and notparticularly stable even at constant temperature, and changesin the high voltage supply. A change of 0.1 per cent in thehigh voltage applied to the I'MTs changes the pulse heights byabout 1 per cent. MDdern high voltage supplies designedspeci fically for this purpDse are now more than adequate toreduce this source of spectrum drift to negligible prDportions.The PMT gain variotion can be reduced to manageable limitsby packoging the dctectors in a thermally cDntrolled environ­ment. Virtually all airbDrne spectrometers currently on themarket have h.eated detector packages for this reason.

There are some spectrometers available whichincorporate active spectrum stabilization. This involvesmonitoring detector pulses which are known to be due toscintillations of a certain energy generated in the detectoreither by 3 specially implanted radioisotope, or an arti ficiallight source. The principle is illustrated in Figurc IDA.14.Two adjacent counting wi ndows are set up to centrc on thespectral peak generoted by the implanted source. Thecircui try is so arranged that when their counting rotes are,"qual no output is obtoined, but when they differ a sui tableunalogue error signal is generated, positive say if the peak isoff centre to the left ,.md negati ve if it is off centre to theright. This error signal can then be used to change the gainof the system somewhere in the chain to restore the properrelation between spectral peaks and counting windDws. Theusual technique for applying corrective action is through thehifjh voltage supply, although challging the conversion fjuin ofthc ADC or the gain of the main amplifier (more difficult)would achieve the same result.

Spectrum stablization is not di fficult to implement andmany variations on the general technique have been utilizedat one time or another, particularly in portable spec­trometers. The problem is to avoid interfering with thespectral information which is being sought in the first place.This constraint means that the implanted source shouldnot generate a peak in the region of interesti.e. approx. 0.5-3.0 MeV. In the case of an implanted lightsource such as a PIN diode, it can be pulsed on duringpredetermined short intervals when data ucquisition isdisabled to avoid any such interference. However a radio­isotope cannot be so controlled other than by a mechanicalshutter which is cumbersome.

To avoid interference with the datCl and theuncertainties of the PIN diDde light source (still in theexperimental sta~e), some manufacturers use low energysources such as I 3 Ba which generates a peak ut 356 KeV.At this low energy even a small base-line shift at high countrates equivalent to say 20 KeV, would be interpreted by :1

spectrum stabilized using 13 3 Ba as a reduction in gain of20/356 or about 6 per cent. It would then call for an overallgain increase of 6 per cent to correct for a minor offsetwhich would not Df itself have caused a significant error inthe results. The 6 per cent gain increase would however inthis case result in a very significant error.

The problem can be circumvented in airborne systemswith arrays of large detectors by using thepotassium 40K peak itself as the reference, sincc it is areliably prominent one under most conditiDns. In a mini­computer-based experimental system desiCJned by the author

Gamma-Ray Instrumentation 145

Nav. data

Multichannel stripchart recorder

9 track W' mag. tape unit

Mini computeror

~ processor

iSignal conditioning

electronics

1Shield

t

DetectorArray

Detector

10 0 0 0 0 I

C.R.T. display forspectrum withcounting windowsidentified and foralphanumericinformation

~,~

(~)Hard copy terminal and keyboard

Such a system will have full spectral recording, means for atmospheric radon correction, aspectrum display with windows identified and a strip chart recorder on which a variety of profilescan be plotted. Such a system would normally be minicomputer or microprocessor based.

Figure lOA.15. Modern airborne gamma-ray spectrometer

Most equipment now incorporates programmable hard­ware based on minicomputers or their microprocesserequivalents. The advent of this technology in a form suitablefor field and airborne instrumentation has suddenly opened upthe possibility for sophisticated on-line data correction, CRTdisplays, keyboard entry of commands etc., to an extentlimited only by the imagination of the designers.

Manufacturers have been understandably cautious inexploiting these possibilities, however most have thecapability for storing complete gamma ray spectra and mostei ther have or are adding CRTs for displaying the fullspectrum with provision for identifying the counting windows.This is a well nigh indispensable feature in the author's viewfor detector gain matching and other performance checks.Window counts and other alphanumeric data are displayedeither on CRT terminals or LED readouts; somemanufacturers offer hard copy terminals rather than CRTdisplays.

Multichannel strip chart recorders are still universallyused to display inflight profiles of window counts and otherdata such as ratios and radar altimeter readings. Somesystems also use the same recorder for generating plots ofcomplete spectra for system performance records.

Digital recording of airborne survey data was for manyyears done on 0.5 inch magnetic tape at a density of 200 or300 bits per inch (bpi), using incremental seven-track tapetransports. The years 1976 and 1977 saw an almost universalchange over to the newer nine-track tape transports whichare compatible with data centres all over the world and with16 bit minicomputer systems. They also offer much higherrecording rates than the older seven-track systems and moreeconomical information storage as the bit density is 800 bpi,about three times that of the older standard. One dis­advantage is that this high bit density makes factors such astape condition, cleanliness of the recording heads and capstandrive etc., very much more critical than they were with theolder systems, so that more attention must be paid to tapestorage and handling practices and preventive maintenanceschedules for the tape transports during survey operations.

The higher recording rates now available make itpossible to record complete gamma ray spectra at the end ofeach counting period if they are stored in a sui table memoryby the ADC as described earlier. Several survey specifica­tions drawn up by national agencies now call for completespectral recording from two separate detector arrays, onewith a downward field of view and a smaller one shieldedfrom direct terrestial radiation which is used to monitoratmospheric radon. The recorded spectra extend to 6.0 MeV,allowing cosmic background to be monitored so that correc­tions can be made continuously for this contribution to thepotassium, uranium and thorium derived signals. Thearrangement is fairly elaborate and requires two ADCs andbuffer memories, so that one pair of spectra can be acquiredwhile the previous ones are being written to tape.Figure 10A.15 shows the block diagram of a typical modernairborne spectrometer.

FUTURE TRENDS IN AIRBORNEGAMMA RAY SPECTROMETRY

One thing which has bedevilled electronic dataprocessing in general and field recording systems in particularhas been the need for electromechanical mass storagesystems such as magnetic discs, drums, and tapes large andsmall. It now seems likely that "magnetic bubble" memoriesand similar technology will replace these devices in theforseeable future. Bubble memories can be considered (verybroadly indeed) as the solid state analogue of the magneticdisc memory. Instead of the disc physically rotating past asensing head, the tiny magnetic anomalies (bubbles) in a pieceof otherwise uniform material are moved in the materialitself. Storage and readout are not yet as fast as with core orsolid state memory devices, but even now are orders ofmagnitude faster than for a disc. There seems little doubtthat by 1985 or even sooner, the cost and the developmentwill have reached the point where bubble memories couldreplace magnetic tape for in-flight storage of survey data.

146 Q. Bristow

As microprocessors become faster and moresophisticated there will be no reason not to do on-line dataprocessing, including the deconvolution of spectra to correctfor the known and imperfect response of detectors to thegamma radi3tion they receive.

Analogue-to-digital converters are already available insingle width Nuclear Instrument Modules (e.g. Nuclear Datamodel 575). As large scale integration becomes moresophisticated, their sizes will shrink to the point whcre itwould be possible to consider having one ADC for eachdetector in an airborne array. If at the same time the PIN,diode li~ht source or an implanted alpha emittersuch as 2 lAm becomes a reliable method of providing areference source, then the necessary signal conditioningelectronics circuitry an ADC and a high voltage power supplycould be designed to be integrnl with the detector as a singlepackage. The detector "output" would then be a digitaladdress rather than an analogue pulse height, with the encrgyand <Jddress relation internally stabilized by the referencescintillation source using appropriate windows all the ADC asdescribed earlier. This would avoid the analogue problemswhich develop at high count rates since the detector"outputs" would be "summed" digitally. The superimpositionof analogue pulses from different detectors would also beavoided by pushing the digital processing one stage fartherback in this way.

Detectors with internally stabilized digital outputscould then be assembled in very large arrays indeed withoutany loss in overall resolution, thereby allowing higher surveyspeeds.

REFERENCES

The literature cited below is a selected bibliographywhere additional discussions of the instrumental aspects ofgamma ray spectrometry can be found. These are notnecessarily in the context of uranium exploration as thisrepresents a fairly limited and specialized application of thescience. For references to earlier work in this field thereader is referred to the proceedings of the Symposium onMining and Groundwater Geophysics held at Niagara Falls,Ontario 1967 (see below for citation).

Breiner, S., Lindow, J.T., and Kaldenbach, R.J.1976: Gamma-ray measurement and data reduction con­

siderations for airborne radiometric surveys; i!lI.A.L.A. Proc., International SymposiumExploration of Uranium ore deposits, Vienna 1976.

Bristow, Q.1974: Gamma-Ray; sensors; in geoscience instrumento­

tion; LA. Wolff, E.P. Mercanti, ed., John V'liley &Sons, New York.

1975: Gamma-ray spectrometry instrumentation; inReport of Activities, Part C, Geol. Surv. Can.,Paper 75-1C, p. 213-219.

1977: A system for the offline processing of boreholegamma-ray spectrometry data on a NOVA mini­computer; in Report of Acti vi ties, Part A, Geol.Surv. Can.;-Paper 77-1A, p. 87-89.

1978: The application of airborne gamma-rayspectrometry in the search for radioactive debrisfrom the Russian satellite COSMOS 954(Operation "Morning Light"); in Current P.esearch,Part B, Geol. Surv. Can., Paper78-1B, p. 151-162.

Chase, R.L.1961: Nuclear pulse spectrometry, McGraw Hill, New

York, 221 p.

Crouthamel, C.E.1960: Applied gamma-ray spectrometry, Pergamon

Press, New York, 443 p.

Duval, J.S., Worden, J.M., Clarke, R.B., and Adams, J.A.S.1972: Experimental comparison of Na(Tl) and solid

organic scintillation detectors for use in remotesensing of terrestrial gamma-rays; Geophysics,v. :57(5), p. 879-888.

F airstein, E. and Hahn, .1.1965/1966: Nuclear pulse amplifiers - fundamentals and

design practice, Part.s 1-5, Nucleonics: v. 23(7),p. 56 (july 1965); v. 23(9), p. 81, (Sept. 1965)v. 23(1l), p. 50 (Nov. 1965); v. 24(1), p. 54 (Jan.1966); v. 24(3), p. 68 (Mar. 1966).

Foote, R.S.1970: Radioactive methods in mineral exploratIOn; In

Mining and Groundwater Geophysics, ed.,L.W. Morley; Geo!. SLJrv. Can., Econ. Geol. Rept.No. 26, p. 177-190.

I.A.E.A.1974: Recommended instrumentation for uranium and

thorium exploration; I.A.LA. Technical Rept.No. 158, Vienna.

Killeen, P.G. and Bristow, Q.197':>: Uranium exploration by borehole gamma-ray

spectrometry using off-the-shelf instrumentation;I.A.E .A. Proc. International Symposium onExploration of Uranium Ore Deposits, Vienna.

Nicholson, P.W.1974: Nuclear electronics; John Wiley & Sons, London.

Reed, J.H. and Reynolds, G.M.1977: Gamma-ray spectrum enhancement; U.S.,

LR.D.A. Rept. No. GJBX-25(77).

Orphan, V., Polichar, R., and Ginaven, R.1978: A "solid state" photomultiplier tube for improved

gamma counting techniques; Uranium ExplorationTechnology Ann. Amer. Nuc!. Soc. Meet., SanDiego, Calif, June, 1978. (Abs.)

Pemberton, R.H.1970: Airborne radiometric surveying of mineral

deposits; !!l Mining and Groundwater Geophysics,ed., L.W. Morley; Geol. Surv. Can., Econ. GeoJ.Rept. No. 26, p. 416-424.

Schneid, E.J., Swanson, F.R., Kamykowski, E.A., andMendelsohn, A.

1977: Airborne detector improvement; U.S. E.R.D.A.report No. GJBX-40(77).