CA2245813A1 - Method and apparatus for remote density measurement - Google Patents

Abstract

A Multiphoton Densitometer (MP-Densitometer) for remote density measurement uses special multiphoton sources with very low, licen se-exempt activity to measure the density of samples. For gas/oil/water mixture inside 3 inch steel pipes, the measurement precision, using a 2 microCi source, is substantially equivalent to that obtained by a prior-art gamma densitometer using a 30 milliCi source. Using a 1 microCi source the MP Densitometer measures the density, with uncertainty less than 1 %, of fluids inside pipes with a range of diameters up to 10 inches. The invention achieves very low background (orders of magnitude lower than for prior-art gamma densitometers) through selection of densitometer elements, geometry, electronics, and software.
Reduction of system uncertainties, a temperature compensation scheme, active temperature regulation, real-time statistical analysis, and smoothing of the output signal allowed the elimination of temperature effects and pile-up artifacts. With a very low activity source, the MP-Densitometer has an accuracy in density measurements of 1 % for oil/water laminar flow. The uncertainty is 0.5 % for longer measurement tim es on the order of a few minutes.

Description

CA 0224~813 1998-08-06 WO 97/29356 PCT~US97/02224 l\~ETHOD ~ND APPARATUS
FOR REMOTE DENSITY MEASUREMENT

BACKGROUND OF INVENTION
The invention provides a method and a~paldLus for remote measurement of density of gas, fluids, solids or mi~Lulc;s of the above. The disclosed method of MP-densitometry allows for a densitometer which is transportable, low cost and uses only low activity sources, ~.e., license exempt activity.
The invention relates to improvements in gamma densitometry using sources emitting several high energy photons ~imlllt~n~ously, and det~Gting the signals of these photons in coincidence to reduce the background hundreds of times and provide precision in density measurements better than that of the standard gamma densitometer but with a source with 10,000-fold lower activity than the one used in a standard gamma densitometry (license exempt activity). Such devices are referred to here as Multiphoton den~it-~mPtPrs (MP-densitometers) or enh~nre~l gamma den~ L~l~ (EGD).
In many applications, one needs to measure the density of a material in motion, either gas, liquid, solid, or gr~mll~t~d7 e.g., fluids and gases in pipelines or foils/flats/slabs of metal in the process of milling. Often, the material is either hazardous or can be easily col.~ cl. Examples of hazardous materials are those which are toxic, easily infl~mm~ble or explosive. E;or many biological and agriculture products spoilage or cont~min~fion results from contact with external air. However, the majority of classical methods of density measurement require sample taking or aliquoting and are not practical in disclosed applications.
Gamma den~itom~t~rs are used to measure the parameters of bulk fluids. Especially ol LallL is the use of densitometers in charat;Le~ ion of petroleum and petroleum products, e.g., measurement of relative fractions of petroleum, water and gas. Herein, the main problem is large variability of fluid composition and inhomogeneities. St~ti~tic~l sampling of the properties is made very difficult by the presence of large bubbles of gas, thick layers of water and a combination of all types of small bubbles.
Gamma de~-~ilo,..eters are also used for measurement of density of gr~m-l~t~?-l material.
For example, density is an important parameter when h~n~ling grains, hops and other agricultural products. Typically, it is easier to measure the density of grains in flow than to measure their weight, grains size di~ll ilJuLion, and then calculate the density . For example, the direct measurement of density permits estim~tion of the fraction of water in the sample.
Finally, there are applications when the material itself is solid; typically in form of a thin foil or plate. For example, in many met~ rgic processes the foils and metallic flats of dirr~ thicknesses are produced by a milling from metal at very high ~ LLU1~;. The feedback between the parameters of milling machine and an instrument me~ ring the m~teri~l thickness is required. Similarly gamma densitom~ters can be used in the paper industry to measure the th~ n~s~ of paper or in chemical industries to measure the thickness of plastic foil. In all of these applications the gamma d~n~it-mPt~r is actually used to measure thi~knP~, because the density of solid materials is well known.
De~ eters are often used to measure properties of lUi~LUlC;(S~ of fluids/gases/solids.
The typical example is in the petroleum industry where crude oil is a mixture of petroleum, water and gas. In multiphase problem applications, the densitometers are used not only to measure the density of mixture, but are a part of more complex system that dele,~llilles the composition of the crude oil. In this case, the dependence of absorption of high energy photons on the chemical composition can be used. Herein, a plurality of radiation sources with very different energies are used.
Many types of densitometers use the absorption of radiation (electrom~gn~tic or sound waves) by matter. The ideal densitometer should have * relatively low ~L~llua~ion due to the collLaillel when compared with the ~U~llua~ion in the sample;
* small variation of absorption coefficient with material composition;
* n~g1igihle effects of surface scatLt;lillg.
In existing gamma densitometers, the attenuation of high energy photons, typically with energy larger than 300 keV, is measured in the configuration shown in c~,ll4JaldLiv~ l~igure 1. Photons emitted by the source are ~1~t~cte~1 by a scintillator counter located on at the opposite side of the sample. Unfol Lu.~aL~ly, existing gamma densitometers have two 1imit~ti~)n~:
* high ill~ll5i~y sources are used, typically larger than lO mCi;
* the density resolution is not better than a few percent.
These 1;mit~ti~ns of prior-art gamma densitometers can be traced back to:
* high radioactive background;

CA 0224~813 1998-08-06 W 097/29356 PCTrUS97/02224 * a few percent drifts of the count rate due to temperature sen~ilivi~y of system ~olllpolle * instabilities, mainly due to pileups and dead-time artifacts.

- - SllMMARY OF TH h' INVENTION
The invention relates to a method for d~Lellllillillg the density of a gas, liquid, solid or mixed sample comprising measuring the attenuation and preferably the scatter of at least one of a plurality of photons coll~;ull~;lllly emitted by a radioisotope7 whelc;ill at least two of the photons are (lptected in coincidence in at least two detectors. The method may involve:
(a) placing a radioactive source capable of con~;ulr~llLly emitting at least two photons so that emitted photons pass through the sample, (b) ~1etPcting at least two photons con~;ullell~ly emitted by the radioisotope, in coincidence in at least two detectors, and ~c) mP~llrin~ the ~IlP~ ion in the sample of at least one of the photons CO11~;U11G11lIY
emitted by the radioisotope.
The invention may employ long life positron-gamma radio~ eLs listed in Table 1, long life nuclear cascade radioc;llli~Lt;l~ listed in Table l, and long life electron caphure radio~-niL~ listed in Table 2. The total source activity may be below 10 microCi and background may be rejected in the photon detectors by (i) selecting low nahural radioactive background elements; (ii) analyzing pulse height and pulse shape; or (iii) using coincidence ~ipn~h1re, individual1y or in combination.
The method may further comprise shaping the signals from the detectors to have width of less than 2.0 microseconds, and using a fast coincidence circuit to ~liminich pile-up ~lirdc~. Preferably, the width is shaped to less than 0.5 microseconds and a very fast coincidence circuit is used to ~limini~h the pile-up artifacts.
The signals from the detectors may be on-line or off-line analyzed by DSO to establish the fraction of pile-ups, and appl~",liate software correction is used to estim~tP the true count rate.
Methods can be implemented to e1imin~SP Lt;lll~eldLul~ dependent drifts of count rate including the use of elements with low temperature coefficients, temperature stabilization and temperature colllpen~d~ion. The method may involve placing the scintillator in an a~plo~-iaLe Dewar and stabilizing its temperature electronically with a precision of a few degrees Celsius, preferably with precision of at least lo C. This may involve use of both an appl~lia~ heater (preferably ohmic heater) and cooler ~preferably Peltier element) and -means to homogenize l~ Gl~ture by forced air flow. Temp~,dLule;, of both scintillator and HVS/amplifier module may be sensed with precision of at least 0.2 o C, and preferably with precision of about 0.1o C, by temperature to voltage collvelLel (electronic thermometer) and the equivalent voltage is then measured by al,~r~liate voltage sensor (ADC/DSO) in a r~ central proces~ing unit.
Preferably, an analog voltage signal equivalent to a ternperature is sent to a~lu~lia~
voltage sensor (ADC/DSO) in the central processing unit by means of shielded coaxial cable.
The analog voltage signal equivalent to a temperature is preferably Ll~rolllled in an optocoupler, sent by fiber optics, decoded by light to voltage collvclLcl and then measured by an a~lo~ Le voltage sensor ~ADC/DSO) in the central proces~ing unit. The analog voltage signal equivalent to a lclll~ aLulc is altelllalively transformed in voltage to frequency collv~l~er, sent by shielded cable, preferably coaxial cable, and is measured by a~L,ropliaLc frequenciometer in the central proces~ing unit. Preferably, measurement of Lel"~elature of scintillator and HVS/amplifier module is performed at least 10 times per second.The method may involve implem~nt~ti-)n of a st~ti~tir~l rejection procedure. A preset number of temperature llleasulclllents (N) ~typically a few hundred) is acquired, and lell~el~Lulc values that deviate from running average of temperature by more than 95%
(more than two sigma) are rejected and replaced by weighted average over closest four measurements of temperature.
Preferably, the mea~ùlelllell~ of ~ll~e~dlu~e of scintillator and HVS/amplifier module is used to calculate the true count rate, i.e., the count rate drifts due to temperature changes are accounted for by software. Te~ el~Lu-~ compensation may be performed by means of calibration curve which is a product of three polynomial curves taking in to account temperature sel~iLivi~y of scintillator, PMT and HVS/divider and amplifier, re~e.;Liv~ly.T~ ld~lLc compensation may use 3D look-out tables established in a precedent calibration procedure, or t'ne L~ e~ ul~; dependence of DSO is periodically measured with a system cO~I;.ini~l~ means for signal multiplexing and means to generate pulses of well known and temperature independent shape. Preferably, the t~ d~Ul~ gain calibration for each detector is obtained by a series of measurements in both OR and AND mode ~cqui~itic n A plurality of methods may be used to elimin~fe the long term drifts of count rate, e.g., due to element aging, inrl~ltlin~ the use of external sources of X-ray and gamma-ray photons with known energies with at least two dirrt;~ L Iines, including mrrh~nir~l means for CA 0224~813 1998-08-06 W O 97/29356 PCTnUS97102224 removal or shielding the sources; and the use of two separate modes operation, namely measulGlllellL mode and calibration mode. At least three c~ulll~l~, may be used in measurement mode and at least two channels of DSO in calibration mode.
A calibration mode colll~lises the steps of measuring pulse shapes, calcl-t~ting the energy spectrum, cstim~ting pile-up and dead-time corrections, and m.ozl~llring external electrom~gnrti~ hlLelre~ ce. The results of a calibration run can be compared with previous calibration runs, and the look-up calibration tables are upgraded for subseq~lPnt use for adjustment of raw data.
A plurality of methods can be implemented to compare the data in OR and AND dataacquisition mode, including acquisition of at least 100,000 events with both pulse-height and pulse shape rejection enabled followed by the establishment of detection efficiencies in which the count rate rates in the detectors A and B are compared without the use of the above said rejections.
For fluids such as gases and liquids, and mixtures, steps may be taken to account for the influence of collL~ulel walls, e.g. pipeline walls, including calibrations performed when the container is empty or filled with liquid of known density. Preferably, the liquids are either water or hydrocarbons with well known density.
Where the calibration source has activity lower than 0.5 microCi the energy ~ecLlulll is pl~f~ldbly obtained with DS~ and all rejections enabled for three test configurations, i.e.
with empty pipeline, pipeline filled with hydrocarbon with known density and pipeline filled with water. Where the calibration source has activity lower t'nan 5 microCi the count rates in the selected energy window are preferably obtained with coincidence enabled for three test configurations, i. e. with empty pipeline, pipeline filled with hydrocarbon with known density and pipeline filled with water. Where the source has activity lower than 10 microCi the count rates in the selected energy window are preferably obtained with coincidence enabled.
A st~ti~tit~l rejection procedure may be implemented, wherein a preset number ofcoincidence counts (N) (typically a few hundred counts) is acquired, and the count rates that deviates more than two sqrt~ from the average value N are rejected and replaced by weighted average over closest four mea~,ulclllen~ of count rate.
An apparatus according to the invention colllplises (a) a holder for placing a radioactive source near a sample so that photons from the source pass through the sample, W O 97/29356 PCT~US97/02224 -~b) at least two detectors capable of ~letPcting at least two photons cull~;u~ ly emitted - by a radioisotope and gel~.aLillg a col,~onding signal, at least one of the dt:Le~;~ol~ being placed across the sample from the source and capable of m~curin~ the attenuation and ~ ~ ~ preferably the scatter of at least one of the photons con~ullGllLly emitted by the source, and (c) a data processor for converting the signals to density measul~,lllellL~.
Using radioiso-o~es listed in Table 1, tne detectors may be o~ldt~d in coincidence mode in a symm~otric sandwich configuration as in dldwillg 2; in a modified symmetric sandwich configuration as in drawing 3; in a compensated sandwich configuration as in dldwing 4;
or in a triangular configuration as in drawing 8; and for thin foil or plate samples, in a coln~ellsal~d flat ~yll.. lleLlic sandwich configuration as in drawing 9, in an asymm~tri-~
sandwich with sepaldLo~ configuration as in drawing 10, or in a modified triangular configuration as in ~llawillg 11.
Using radioisoto~es listed in either Table 1 or Table 2, the del~;Lol~ may be operated in coincidence mode in an a~.y.. r~ ~ ic sandwich configuration as in drawing S; in a shifted asymmetric sandwich configuration as in drawing 6; or in a modified asymmetric sandwich configuration as in drawing 7.
The source may be placed inside a spool made of material with a~,~,r~liat~ m.och~nic~l charac~ Lics (e.g. strength and weight) but low atomic number such as beryllium, plastic rolced with dppropliate fiber, ~ll.."i....l~" v~n~-1inm or ~ Preferably, the spool has ec~enti~lly elliptic cross-section, with total cross-section surface close to the cross-sectional surface of two pipes it is joining.
AlLelllaLi~/~ly, the source may be placed inside a bypass made of material with a~plo~liat~
mPrh~nical characteristics but low atomic number such as beryllium, plastic l~ olced with a~ iat~ fiber, ~l~.. i.. , vzln~ m or ~ i.. Preferably, the bypass has esse~Lially elliptical cross-section, with total cross-section surface much smaller than the cross-section of the main pipe.
The ~ us may further comprise a third, anticoincidence detector to (1imini~h background as in dldwillg 12.
Prefera~ly, the detectors are scintillators, and at least one of the detectors is NaI(Tl) with a thi~ nes~ of at least 2" coupled to PMTs. In one embodiment, the detectors are of substantially the same size and consist of the NaI(Tl) scintillators with a thickn~ of at least 2" coupled to PMTs. At least one of the ~letectors may be a BGO scintillator with a thic~n~ of at least 1" coupled to PMT. At least one of the detectors may be a BaF2 CA 0224~813 1998-08-06 W O 97/29356 PCTrUS97/02224 scintillator with a thickness of at least 2" coupled to UV light sensitive P~Ts. Or at least one of the detectors may be either GSO(Ce), CsF or CeF3 scintillator with a thickness of at least 2" coupled to PMT. Where a radioisotope from Family 2 is used, at least one the said detectors is preferably a NaI(Tl), CaF2(Eu) or YAP scintillator with thickness of less than 0.5" coupled to PMTs.
An a~aldLus optimi7Pd for radioisotopes from Family 2 uses hal.lware and software means to reject events due to interaction of CR in PMTs, and may have background due to electrons emitted from PMT ~liminishPd by application of ~pl~p,iate thick optical windows made of quartz or other ultrapure, optically transparent materials. ~lle...,.l iv~ly, r~lio~rtive background is ~limini~hi~ by the use of an essentially cylindrical shield including at least one heavy metal component and placed around at least one of the detectors, and ~lvp~iaLe thick optical windows made of quartz or other ultrapure, optically L,~a,~ materials. The shield preferably c-,l~,ises a few millimeter of Pb/Sn/Cu, the Cu being placed closest to the scintillator and further colllp,i~hlg a shield between the PMT and PMT base. One detector ~5 may be smaller than the other with the shield placed only around the smaller detector.
For radioisotopes from Family 2, the ~ep~aLoL is preferably placed around the source and close to one of the ~letect~ r~, and PMTs are selected to have radioactive background of less than 0.1 cps.
According to the invention, innovative use of MP-de,~iL~l"etry includes coinri~lPnt detection and pulse shape analysis to achieve an excellent signal to background ratio ~S/B) for low activity sources, and optimal use of photons with different energies. The following working definitions are used: low energy photons: 10 keV < E < 100 keV, and high energy photons: E > 100 keV. Fullll~llllore, photons emitted in the atomic shell rearran~,~lllellL~
are called X-rays and photons emitted by the nuclear transitions and/or positron annihilzlfions are called gamma-rays. The low energy photons (LEP) of importance in this invention are either X-rays or gamma-rays, but high energy photons (HEP) are always gamma-rays. To conclude this brief introduction we disclose a list of radioisotopes that are particularly useful in MP-densitometry, because they emit more than one photon. These fall into two main f~miliPs:
Family 1: Positrons emitted by the source ~nnihil~t~ with a nearby electron into two gamma rays, each with an energy of 511 keV. These gamma rays are emitted back-to-back because of momentum consel~tion and their coincident detection is strongly angle-correlated. In positron-nuclear gamma (PG) emitters an emitted positron is followed by a 3~6 PCTrUS97/02224 nuclear gamma tr~n.~ifion. The positron ~nnihil~tes, as described above and ~ere is also an additional nuclear gamma ray. In nuclear cascade gamma (NC) emittt?r.c, a m~t~t~ble nucleus c.~c~c~es from a highly excited m~f~t~hle state, and produces several gamma rays.
Tne radioisotopes from the PG and NC groups that are illl~OlL~llt for MP-densitomehy are S listed in Table 1.
Family 2: Eleckon cap~re (EC) isotopes. In this case all atomic X-ray and a nuclear gamma ray are emitted (see Table 2).
Tables 1 and 2 show the lifetimes of the radioisotopes in years and days. The energies of the photons are listed in Handbook of Chemistry and Physics, Edition CRC, New York, relevant portion of which are hereby incorporated by l~rc;L~llce.

Table 1. Isotope Family 1: Long life l.osiLIc,~ (PG) and nnrlP~r f~ (NC) ~kl Na22 ( 2,58 y);
Al26 (> 105 y);
CoS6 ( 77.3 d); CoS8 ( 71.0 d); Cos~ ( 5.27 y);
Zn65 (245.0 d) As74 (18.0 d);
Rb84 ( 33.0 d);
y 88 (108.0 d);
Tc95m ( 60d);
Rh99 (16d); Rhl02 (206 d);
Il26 (13.2d);
EUls~ ( 13y);
Hgl94 (130d);
Bi206 (15.3 d) Subtotal: 16 isotopes~0 Table 2. Isotope Family 2: Long Life 13:C isotopes.
Be 7 ( 53.6 d);
Ar3' ( 34.3 d);
Ti44 ( 1,000 y);
V49 ( 330.0 d);
CrS' ( 27.8 d);
Mns4 ( 291.0 d);
Fess ( 2.7 y);
CoS6 ( 77.3 d); CoS7 (270.0 d); CoS8 ( 71.0 d);
Zn65 (245.0 d);
As73 ( 76.0 d); As74 ( 18.0 d);
-CA 0224~813 1998-08-06 W O 97/29356 PCT~US97/02224 Se75- (120.0 d);
Rb83 (83.0 d); Rb84 ( 33.0 d);
Sr82 (25.2 d); Sr85 ( 64.0 d);
y 88 (108.0 d);
- 5 Zr88 ( 85.0 d);
Tc95m (60 d);
Rh99 ( ~6 d); Rhl~l (5 y); Rhl02 (206 d);
Pdl03 ( 17 d);
gl05 (40 d); Aglo8m ( >5 y);
Cdl09 (470 d);
Inll3 (118 d);
Snll3 (118 d);
Sbll9 (158 d);
Tell8 ( 60 d); Tell9 ( 45 d); Tel2l { 17 d);
I 125 ( 60 d); I 126 ( 13.2 d);
Bal3l ( 11.6 d);
Cel39 (140 d);
Pml43 (265 d); Pml44 (440 d); Pml45 ( 18 y); Pml~ (710 d);
Pml58m ( 40. 6 d);
Sml45 (340 d);
Eul48 ( 54 d); Eul49 (120 d); Eul50 ( 5 y); EUls2 ( 13 y);
Gdl46 (48 d); Gdlsl (120 d); ~dl53 (200 d);
Tbl~ (73 d);
Tml68 (85 d);
Ybl69 (32 d);
Lul73 ( 1.3 y); LUl74m ( 165 d);
Hfl75 (70 d);
Wl8l (130 d) Rel83 (71 d); Rel84 ( 50 d);
Osl85 (94 d);
Irl89 ( 11 d); Irl~ ( 11 d); Irl92 ( 74 d);
Aul95 (200 d);
Hgl94 (130 d);
Tl202 ( 12 y); Tl204 ( 3 9 y);
Bi206 (15.3 y); Bi207 (30 y);
Pu237 (-45.6 d);
Cm24l (35 d).
Subtotal: 72 isotopes It is beneficial to be able to use a plurality of different MP-densitometer compatible sources. In some applications not only density but also composition need to be ~l~t~rminpd~
e.g., multiphase mea~ulelllellL~ in petroleum industry. This can be accomplished using numerous potential sources with emission energies properly selected to enable differentiation of the materials and to facilitate elimin~tion of various backgrounds. Furthermore, sources with lifetime longer than a month are pl~,rt;llc;d.
-g WO 97/29356 PCT~US97/~2224 The use of low activity sources is highly advantageous. The power and practicability of the disclosed procedures are conn~cttqd with the use of new, very low background and highly selLsiLive ~letectinn techniques applicable to some radioisotopes. Specifically, the use of Multiple Photon Detectors (MPD) and a~l~liaLe radioisotopes from the f~milies mentioned above are of crucial i.npolLa~ce.
Very low radioactive background detectors (VLBD) are opL"~i~ed for multiphoton radioisotopes, inrl~l-ling positron-gamlna (PG) ellliLL~s, nuclear cascade (NC) ~ , and electron capture ~EC) c~ iLLel~. Instruments for (luauLiryi.lg PG C~lliLk;l'7 were disclosed in United States Patent No. 5,083,026 entitled Met*od, Apparatus and Applicatzons of the Quantitation of Mu~tiple Gamma-Photon Producing Isotopes wit~ Increased Sensitivi~y.
United States Patent No. 5,532,122 entitled Quantitation of Gamma and X-ray Emitting Isotopes describes detection of EC radioisotopes. Such instruments are distinct from the use of ml~ltiphoton isotopes for densitometry.
In C~Li".i~ g the pelÇo~ allce of gamlna ~el.~ ..",eters the main challenge is to t1imini~h the radioactive ba~,kgl~ulld and thus to achieve with a low activity source an excellent signal to background ratio, i.e., S/B > 100, and high accuracy in density measurements. Basic elements that may be upLilll.~ed are:
* the source;
* the detector; and * the geom~trir~l configuration of the system.
Other elements for o~li"~ nn are:
* the coincidence trigger and on-line shape analysis;
* the L~lllp~ldLulc; st~hili7~tion/compçn~tion scheme;
* the calibration procedure; and 2~ * the data acquisition and sLaLi~,Lical procedures.
In the following we show that t_e innovative use of modified geometry, mllltiphoton coincidence and sophisticated signal analysis permits development of MP-densitometers with p~,lr(,l "~ nres considerably better than prior-art gamma densitometers. System optimi~tion depends on the sizes and densities of measured samples. These differences imply the use of photons with dirr~l~lll energies, because the range of photons is strongly energy dependent.
In the case of PG ~llliLLt~l~, there are additional advantages due to the presence of back-to back ~nnihil~ti~n photons, i.e., the geometry is better defined for this case. Also, in PG
~llliLLel~" additional third high energy photons permit re~nn(l~nt measurement, and allow CA 0224~813 1998-08-06 W 097t29356 PCTrUS97/02224 estim~tion of the average atomic number of the object under study. In the case of EC
CllliLLcl~ there are typically only two coincident photons, and one of t_ose is always a rather low energy X-ray (emitted due to atomic shell lcall~ngc.l,ellL). Thus, use of EC e~liLLcl~.
leads to an asymmetric geometry, where two very differene detectors are used. The NC
tllliLLcl~ pl(.l)liaLc for MP-densitometry emit spatially non-correlated photons. When they emit two photons of very dirrel~cll~ energies, the instrumentation is similar to the case of EC
. . s. The use of NC ellliLLcl~. is important because some of them produce two very high energy photons, like 60Co with energies of 1,173 and 1,332 keV, respectively. In this case the geometrical configuration may be dirrclcllt from the case of PG ell,iLLcl~. and both detectors can be located at the same side of the sample. FulLhc~ ore, one can use PG
CllliLLcl~7 where high energy photons, e.g., ~nnihil~ti~n photons are used in a coincidence scheme to reject background. Because very fast scintillators can be used to detect photons, important advantages of m~xim~l count rate may be realized.
Each of these situations requires the use of photon detectors with ~lirrclcllL properties, leading to very ~irrel~ t~chnir~l implelllcllL~Lions. Thus for each of the cases the specific tradeoffs between the detection efficiency (DE), energy resolution (ER) and temporal response of the detectors should be selected. Furthermore, the photon energies considerably infllTP-nre the choice and dimension of ~letectors. In the case of high energy photon (HEP) e~ tcl~7, only heavy inorganic scintillators are practical and economical, whereas many dirrelclll types of L;~P detectors are available including scintillators, semicnn~l~cting and gas detectors.
Scintillators are the most popular detectors for qn~ntit~tion of hard X-rays and gamma-rays. Good energy resolution is a major advantage in the design of low r~ ctive ba~;hgl~ d detectors. Among the scintillators NaI(Tl) and CsI(Tl) produce the best light yield, while photomultiplier tubes (PMTs) permit counting of single photoelectrons. The Nal(Tl)/PMT combination is very popular, but other scintillator-PMT colllbillaLions are also of importance, including CsI(Tl), CsF, BF2, CaF2(Eu), YAP(Ce), BGO, and GSO(Ce).Here, we will dirÇelellliate between two classes of detectors appropliate for detection of soft X-rays ( NaI(Tl), CaF2(Eu)) and high energy photons (NaI(Tl), CsI(Tl), BGO, BaF2, YAP(Ce), CeF3, and GSO(Ce)), lci,~e~;lively.
Cost considerations should be emph~i7~-1 There are three elements which enhance practicability and ~limini~h the cost of application of gamma densitometers:
* use of low level radioactive sources which do not require licensing;

W O 97/29356 PCT~US97/02224 * small size, portability and user fri~mllinPc~; and * low cost of the instrument.
Use of low level radioactive isotopes decreases the cost in three ways. First, the use of stronger sources requires licensed, specially trained operators and additional, often S considerable, costs of transportation, reporting and waste disposal. Second, the stronger source requires larger shields and decreases the portability and user fri~n~llinPss of the system. Finally, cost of isotopes increases and some of them are not available at larger amounts.
The MP-dwlsiLollleL~,~ system should be portable, robust and user-friendly. Many of the operations are in remote locations, i.e., system should be self-cont~in~-l Thus, low cost ~ o~ 3ters which use scintillators were implemented. Furthermore, the read out electronics has been simplified and some functions which previously were imp!em~ntlo~l in haldwale are ~iullellLly accomplished in software. Particular combinations of l~l-lwal~ and software are disclosed.
The applications of MP-den~it~m~-try, especially in petroleum, chemical and nuclear in-hlstri~os require:
* good density resolution, say beLw~el~ 0.3 % and 3%;
* reasonably high dynamic range, say from 0.05 g/cc to 1.2 g/cc for gasloil/water ~Lul~s, and from 1 g/cc to 10 g/cc for solid samples;
* better than 0.~% reproducibility; and * self-calibration and self-diagnostics.
The invention is further described below. In particular, means for achieving long term stability and repeatability of L,elrolllla,lce of MP-densitometers and advantages of different geometries and proper h~nt~l;ng of temperature drffls and pileup artifacts are discussed.
DETAILED DESCRIPTION
In describing ~ler~ d embo-lim~nt~ of the present invention described here and illustrated in the drawings, specific terminology is employed for the sake of clarity.
However, the invention is not intended to be limited to the specific terrninology so selected, and it is to be nn-l~r~tood that each specific element includes all technical e~Iuivalents which operate in a similar marmer to accomplish a similar purpose.
The general principle underlying the present hlvellLiol1 co~ lises a method according to which sources e~uill;.~g con~;ullenLly a plurality of energetic photons (E > 10 keV) are WO 97/29356 PCT~US97/02224 detected in setups with an u~Li~ cd configuration. The a~dldlus implem.?nting the optimal detection srhPm~s are disclosed.
Standard g~mn~ n~if~ tçr and MP~ c~ ; Existing gamma de~iLollleters use the simplest possible configuration: the sample is placed between the radiation source and the detector (see Figure l) In this configuration only a few elements can be optimi7e~l;
energy/activity of the source and the detector itself. In the following, we disclose a MP-den~itom~ter that elimin~t~s the limitations of prior-art gamma de~ o~ tçrs.
In many applications, current gamma densitometers are severely limited by the high level of required activity. For example, the Gamma-Trol d~ n~ tçr produced by Tracerco Inc., Fngl~n-l uses about 30 mCi source. The den~h-)m~t~rs produced by other m~mlf~rtllrers have sources with similar activities. The thousand-fold background illl~Lovclllent achieved in MP-densitometers permits the use of a few microCurie sources, that are license exempt.
Ful~llel~llore, the prior-art gamma del~iilo,lleters often lead to artifacts in measurement. The ext~?rn~l ~elll~cld~ulc change leads to count rate drifts which are mi~hllt;lyl~t~ d as change of density of a measured sample. This problem is enh~n~erl because gamma densitometers applied in petroleum production are often used in a desert cllvholllllclll, where diurnal temperature changes can be as large as 30-50~C.
Reproducibility and reliability are iUlpOll~lllL features for gamma-densitometers. In many detectors, typical sources of unc~ hlLy are self-absorption and errors due to variations in the source and detector positioning. Frequently used geometry is a ffat scintillator coupled to a single PMT. Absorption artifacts are typically diffîcult to account for in this geometry.
These artifacts can be considerably ~1imini~htofl when using two essentially identical flat detectors, each with independent readout ele~ ,nics. We call such a configuration a "sandwich " detector, because both the source and the measured object are placed between the two detectors. When source self-absorption is negligible and the pl~emlont of the sample is correct, both detectors give essent;~lly the sarne count rate. A sandwich configuration facilitates the use of highly efficient dirrc~cllLiaLion/compensation ~
The background due to Cosmic Rays (CR) is an important component of the total bacl~l~u..d in a single detector configuration. When two detectors are used, the energy ~ 30 deposited in one detector is often very dirrtlcllL from the energy deposited in the second detector, which permits rejection of background due to CR plilllaly particles. Also, as secondary particles shower initi~tf~d by CR in atmosphere have a characl~ lic density which considerably limits the sensiliviLy because of the back~,ound due to secondaries.

WO 97/293S6 PCT~US97/Q2224 Furthermore, use of coincidence scheme is very efficient in elimin~ting radioactive ba~L~luulld in scintillators, PMTs and shields.
Another important advantage of the sandwich geometry is its versatility. A ~l~;fellcd --~ implementation of a fltqtector according to the invention uses two ess.onti~lly identical crystals.
This is a good choice when ~nnihil~tion photons are ~l~otec~(l, and for some nuclear cascade isotopes, e. g., 60Co. However, many other important sources emit photons of quite dirrt l~nL
energies; often one photon is a soft X-ray (E < 50 keV) while the second photon is a nuclear gamma-ray, say E > 500 keV. This configuration is described in following section.
In such case a sandwich detector COI~ of two dirrcl~cllL scintillators with ~lirrcl~elll crystal thi~ s~es is optimal.
Use of this geometry leads to double the number of elements, including PMTs which are a ei~iric~.,l source of background. Furthermore, passive shielding is considerably more costly and the overall shape of the detector assembly is somewhat a~ dld and tliminieh~s portability. Finally, the sandwich geolucLl y and coin~irlPnre Aimini~h the detection 1~ efficiency. Thus, the use of sandwich geometry is an innovative and rather counter-hl~uiLive solution.
Conclusion 1: The prior-art gamma deneit-)m~fers are limited by the use of single photon emitters. Use of multiphoton e lliLLcl~ permits the sandwich detector configuration and tholle~n-lfold lower background. This leads to improved signal/back~lvulld ratio and permits high sellsiLiviLy with sources of low activity. Such Mp-~1en~itompter~e can be operated with a very low activity source, typically below 10 microCi, with better than 1% accuracy.
Diverse Mp~ r cvl.rl~."alions: MP-densitometers can operate in very-dir~rellL configurations as "symmetric" and "a~ylllmcLlic" sandwiches, respectively. The use of sandwich geometry- is counter-hlluiLivc but enables the tradeoff between a lot of 2~ uiLelllents because it:
* permits u~ l;On of S/B ratio;
* permits operation in both non-coincident (OR) and coincident (AND) modes;
* permits to increase ~e sensitivity- with respect to the sample density;
* minimi7t?S ~ ion artifacts; and * ~li."~ s many short term instabilities.
The simplest "symmetric sal~lwich" geometry is the one in which the MPD source is placed inside of the object whose ~fopG, ~ies are measured. This is the ~lcre~l~d implçm~nt~tion when ~ropcl ~ies of gas/fluid mixLulcs in flow are measured. The " symmetric CA 0224~813 1998-08-06 WO 97/29356 PCTr~S97/02224 sandwich" geometry is shown in Figure 2. More specifically, for mea~.u,~ L of gas/fluid ixLul~ densit.,v, a "spool" can be placed between two fr~gm~nt~ of a normal pipeline. The "spool" has diameter equal to the main pipe but is made of material with low aLL~,lualion coefficient, e.g., beryllium, plastic enforced with kevlar/carbon fiber, ~ , VAnz~
or ~ t;l~lll. The "spool" is made according to stringent mechanical tolerances, i.e., bo~ the geometric dimensions and wall thickness are made with better than 0.1% precision. This facilitates the calibration of the MP-densitometer. This "spool" contains in the center a small, hydrodyn~mir~lly shaped container in which the source is placed. This geometry is especi~lly efficient and easy to calibrate in t_e case of PG ~ L~l~, wherein the two ~nnihil~tion photons are emitted back-to-back. Thus, the count rates in both detectors should be very close to each other. This facilitates self-diagnostics and self-calibration.
The "symmetric sandwich" geometry was op~illl~ed for a particular situation of liquid flow. It allows measurement of laminar flow of gas/fluid mixtures, wherein the fluid moves close to the vertical pipeline walls and tne flow of gas is mostly in the center of the vertical pipe. Note, that in this situation the velocity of gas and fluid are quite different leading to additional "dynamic pressure" colnpl~ssion of gas and modifying the efficient density.
However, it can also be implemented in some other applications, such as in the measurement of ~ .,.lies of gr~mll~tt-d material, e.g., grain, seeds or cement.
For horizontal positioning of pipes, the l'~.yllJ~ ;Llic sandwich" geometry also permits d~l "~ g if flow is turbulent (no dirrelGllce between top and bottom detectors) or laminar (a few percent dirr~lGnce between top and bottom detectors). Furthermore, one can establish whc~Lll~l in the mllltirh~e flow instabilities tlr)min~t~, i. e., if diî~lcllL phases of petroleum/water/gas mixture are separated. Another important feature of this configuration is that the total count is ."i-.i",i~l and pileup artifacts ~limini~h~-l A somewhat more complicated "modified symmetric sandwich" geometry is shown in Figure 3. The advantage of this geometry is that it is applicable also in cases when the use of a "spool" is not practical. However, it retains the main advantages of a symmetric sandwich geometry, namely insensitivity to flow pelLulbation, improved pileup rejection and self-caTibration capability. For example, because of overlapping absorption cones the effects ''30 of flow inhomogeneity can be tlimini~h.ocl The "COlll~!ell.~ sandwich" geometry is shown in Figure 4. Herein the source is placed between two objects; a measured sample whose pLo~ ies are unknown, and a "standard sample" with similar and perfectly known properties. A pair of m~tch~-l detectors is ~ ,ftlled, with almost identical detection efficiency, energy resolution and LG~ t;.dLu,G
depenrl~n~e. The concept of "relative" measurements is implement~l, i.e., di~rGlculce in count rate is proportional to di~rclGnce of densities between the "unknown" and "standard".
~ This method is very efficient, when mea,lllclllents of relatively small objects are required, S especially when better than 0.1% precision in density/physical dilllcl~,ions is needed. This configuration is "robust" and permits high reliability self-calibration. It is especially important, when the artifacts due to pileup and due to multiple ~,ca~cliug are important.
Another innovative application is wherein the "relative density/-lim~ncions" of an object are to be measured. By using the known ~,L~dard, the majority of calibration artifacts, e.g., Lclll~tldLulG dependence can be eli~
We disclose special MP-dcl~,ilolllchl configurations advantageous for large objects, say tubes with diameter larger than 6 inches. In such cases the '~a~,ylllulcLlic sandwich"
configuration shown is Figure 5 is the most useful. Herein, the PG source, e.g., 2~Na, is placed close to a relatively small, say 2" in ~ met~r fast scintillator. A very large detector, say 6" in diameter, is placed on the other side of the pipeline. Optionally, a big detector Col~ i"g of a plurality of smaller detectors operated in parallel may be used.
Note, that the large detector does not need to be placed exactly opposite the source at the other side of the object. This leads to a configuration called "shifted asymm~tric sandwich"
(see Figure 6). When the second detector is placed under an angle, higher absorption length is achieved, permitt;ng better relation between density precision and sf~tictir~l uncertainty in the count rate. The change of angle leads to some increase of absorption in pipe walls and increases considerably the absorption in petroleum. Thus, the efficient signal decreases and ~r~liaLG calibrations should be used. Furthermore, only part of the flow is sampled in density çstim,.tion. Actually, in this case absorption in petroleum is higher than absorption in the walls of pipe, and thedecrease in count rate should be c-.lll~n~ d by the increase of mea~,u.elllGllL time. Once more, this configuration cannot be used in prior-art gamma dçll~itomPtçrs which are essentially limited by their signal/bacl~,l~u,ld ratio. It is, however, use~ul in MP-densitometers wherein the sGllsiLiviLy is limited by st,.ti.~tics rather than a signal/background ratio.
The "modified asymmetric sandwich" shown in Figure 7 is especially useful, when objects of very large diameter are studied using PG isotopes, including objects so large, that even ~nnihil,.ti~n photons are subst~nti~lly absorbed. For example, for 10 inch pipelines with 0.375" steel walls about 95% of ~nnihil,.tion photons are absorbed. In this case, the two CA 0224~813 1998-08-06 WO 97/29356 PCTrUS97102224 detectors have very different functions. The detector close to the source detects nn~c~ttered photons and serves as trigger. It can be relatively small, say 2 inches in ~ mlot~r.
However, to (limini~h the pileup artifacts it should be made of a fast scintillator. The second ~ ~ detector measures the number and spectrum of scattered photons. It should be relatively S large with size comparable to the thickn~ of the object to be studied. For example, in one of the implement~ti~ns~ the large detector consists of two slabs of relatively slow CsI
scintillator, each about 2" thick and 16" long. Each CsI scintillator is coupled to a separate PMT and chain of electronics. Using this configuration, the signal/background ratio better than 100 was achieved using only 1 microCi source for 10"-diameter pipes filled with water.
An important, "triangular" configuration is disclosed in Figure 8. This configuration is not applicable to PG isotopes but is a~lo~ lG for NC isotopes, wllGlGhl two high energy and directionally uncorrelated photons are emitted, e.g., such is a case of 60Co. This geometry is ~ pliate for density measurement in the largest pipes, wherein the highest possible energy of photons is an advantage because it limits the alLGl~ualion and permits reasonable st:~ti~tirs. Note, however, that only a fraction of the object volume is efficiently measured, which in the case of inhomogeneous objects including turbulent flow of liquids in pipes may lead to some artifacts. Data analysis minimi7es these artifacts.
For large pipelines, a special "spool" can be produced from ~ll.. i.. ~ or ~ -il,.. rather than st~inles~ steel. For both materials, the stopping power is smaller for a thickn~cc with the same mechanical properties. Thus, a few times lower attenuation in metal is achieved, and a lower activity source can be used. Furthermore, instead of a "spool" with essentially circular cross-section, the oval shape can be used. With a~plupli~te dimensions, the change on the flow speed will be minim~l but the attenuation will be a few-fold ~imini~ht?fi Finally, the "spool" can consist of two parallel channels, i.e., the liquid flow can be bypassed. Let us consider the 24" pipe with 1 " walls. ~ lming an 8" bypass the walls needs to be only 0.25" thick. Thus, the ~LlelluaLion ~iminich~s about fourfold. Taking in account the coincident method of readout used in the MP-(len~itf~metry; the activity of a source may be about 16-fold lower.
Conclusion 2: For dirr~ lL MP-densitometer configurations optimized for PG and NC
radioisotopes, respectively, the performance advantages demon~trat~l in tests fully justify the use of these configurations.
MP-~lPn~itom~t~r conri~u-~lions o~ l for EC sources: We disclosed above the MP-densitometers optimized for measuring densities of large ( > 2" thickn~cs) objects.

W O 97/293S6 PCT~S97/~2224 However, t~ere are important applications for MP-densitometry for smaller objects, such as - measurement of thickness of thin foils, flats and plates of plastic, paper and metal. In this case, the ~tteml~tion of gamma rays is small and sources with lower energy should be used.
The MP-den~itometPrs work best, when the sample absorbs from about 50 to 80% of emitted radiation. For the X-ray, however, the ~ttenll~tion coefficient is very dependent on the atomic number of the absorbing media because the photoelectric effect d~ s. Thus, the applicable energy range is from about 20 keV for paper/plastic/~1l.l..il...l.- foils to about 100 keV for steel or cooper foils.
In our US Patent No. 5,532,122 and U.S.S.N. 08/669,970, the total counting rate is very small; typically less than 100 cps where SIB = 10, is typically judged adequate.FulLl~ ore, in the biomedical applic~tions of MPD, the ~tteml~tion by a s~nple is totally negligible and count rates in both detectors are practically identical. Also, the typical t~nre from one detector to another is very small and heavy shielding is often practical.
In contrast, with MP--lP-n.~h- m~try using EC sources:
1) the count rate is actually ~uite high, typically 100,000 cpm but excellent S/B > 1,000 is required;
2) the " asymmetric sandwich" or "triangular" configurations are advantageous and are used instead of "~yll~ Llic sandwich" configuration;
3) portability is an important issue, and the passive shielding should be considerably smaller than in stationary biomedical applications; and - 4) m~tr.flin~ of detectors properties is extremely ilnL~ol~L in MP--1t n.citom-otry using FC
sources.
Typically a measurement uncertainty of a few percent is required in prior MPD
applications. It is mainly due to either radioactive background or st~ti~t~ uncertainty; often 2~ only a few to a few hul~ ds of photons are collectP(I In MP-~ o~.. P,ters, e.g., used to ~asule the thicL-n.es~ of thin layers, the measurement precision of 0.3 % is typically required and 100,000's of photons are collectecl The uncertainty is limited mainly by the ~y~.".~lir effects and alLird ;L~. Thus, some further contrasts are:
a) in prior MPD systems the count rates in both detectors are iclenti-.~l, whereas in MP-densitometer the dirr~ ce of count rate btlw~en two detectors is the si~nal;
b) influence of L~ ldLul~ drifts is negligible in MPD but is crucial in MP-delLsilollleters;
c) the pileup and dead-time artifacts are very low in MPD but their elimin~tion is over important in MP-den~it--m~try; and CA 0224~813 1998-08-06 d) in MPD applications, the throughput is largely controlled by the user, e.g. measurement can almost always be repeated and time of measurement o~L~ l, wherein in MP-densitometry any measurements inLt;llu~Lions (dead-time) are unacceptable.
Thus, even if some technical solutions enabling the low r~-lio~c tive backgroundincorporated into MPD and MP-densitometers are similar, the two devices are very dirrt;,cL"
which forced a plurality of irmovative modifications.
For small objects, exrçllent st~ti~tirs can be obtained with a small activity source.
g that 50% of photons are absorbed in the object, and that the detection efficiency is about 50%, a 1 microCi source leads to a count rate of about 8,000 cps (50,000 cpm).
This permits 1% st~ti~tir~l precision in density measurement for data ~cq~ iti~n time of a few seconds. However, at this count rate, pileup artifacts are not negligible and dead-time corrections important, making on-line pulse shape analysis difficult.
We ~ ignt-(1 and implemented the MP-den~itom~ter based on use of low energy EC
emitters. In the following, we call this device the X-ray densitometer. An a~ ial~ EC
source is 1~5I, which emits photons with 27, 31 and 35 keV. For this source we achieved the ra-lin~ctive background of less than 1 cpd. Thus bach~,loul~d irnprovement of about 50,000 when com~al~,d with prior-art devices has been achieved. MP-densitometers optimized for the thin films application are of three different types, namely:
* "c~ ensdLed flat symmetric sandwich" (see Figure 9);
* "asymmetric sandwich with separator" (see Figure 10); and * "modified triangular" (see Figures 11 and 12).
Figure 9 provides a s~h~ tic view of a "compensated flat ~ylmllcLlic sandwich"
configuration of a MP-densitometer. A single EC source is placed between two samples, one of unknown density/dimensions and another being a standard sample. Two m~tch~-l detectors 2~ are used each with ~ mrtrr much larger than the thirknPss of the objects. This is a geometry which facilitates the relative measurement and compensation of the artifacts providing the properties of two detectors are well matched. We expect that about 1.0%
precision in thirkn~s~ can be obtained for films with thickness of about 1 mm, and about 0.2 % for plates with thickn.ss.~ of about 1 cm. Herein, the ~C isotope with very dirr.,~llL
energies of X-ray and gamma ray are used. The comparison of coincident rates on each side for dirrelelll energies permits artifacts elimin~tion. The use of a compensation scheme makes the measurement very reliable, e.g. if the foil from the same batch can be used as standard so that errors due to the variability in chemical composition of the sample can be elimin~ttoc7 W 097/293~6 PCTAUS97/02224 A schematic view of an "asymmetric sandwich with separator" cor~lguration of MP-densitometer is shown in Figure 10. A single EC source is placed on one side of the object whose density and/or ~lim~ncions are to be 5t~ ierl Small X-ray detector is placed close to a source and larger gamma ray detector on the ot_er side of the sample. The two detectors work in coincident mode. The advantage of this cor~lguration is its versatility. Using it, objects of diverse shape, not only flats can be ev~ln~ted Furthermore, the instrument can be relatively easily rescaled for samples with dirre.~;llL dimensions. However, the ~tt~ml~ti~n o~ellies of material from which the sample is fabricated has to be known rather well and a reliable calibration procedure is required. Actually, we disclose the use of an EC-clencitomFter in this corLfiguration, where the dual or plural energy isotopes are used. By CC~ dlillg the relative count rate at dirr~ L energy, the dLLellu~Lion ~Lc,~e.Lies of material can be calclllzlt-~d However, the second detector should have the best energy resolution possible. Typically it will be a Ge detector or NaI(Tl) scintillator. Note the placement of the "separator" around the EC source. The use of separator permits suhst~nti~l fliminichm,ont lS of cross-talk between the two deLecLul~, leading to low r~iio~ctive background. The use of a se~dldlul ~iiminiches the requirements on speed of pulse-shape analysis and permits to overcome the pileup problem. This cor~lguration may employ digital signal processor (DSP) cards.
The disclosed "asymmPtric sandwich" cor~lgurations is enh~n.~e~l by use of sepalatol~,.
In this case, one of the photons is an X-ray with relatively low energy and second photon is a gamma ray, whose energy can be selected to be quite high. Obviously, the soft X-ray cannot peneL~ a large object but it can be used as a trigger to fliminich the background.
This configuration leads to the lowest background, because a X-ray detector can be very small and thin. Using this configuration, a background lower than 0.1 cpm was achieved.
Furthermore, such detectors are much more portable because the X-ray detector can be mi~ 1 From a large class of EC radioisotopes (see Table 2), the energy of a gamma ray can be selected to o~Li~ the resolution of a MP-cle-rlcitcm~ter for a diverse object.
Figure 11 provides a s~ e~ ic view of a "modified triangular" configuration of a MP-densitometer. A single EC source is placed on one side of the object whose density/~1ime.~cions are to be studied. Two gamma ray detectors are placed on the other side of the object. The size of detectors and the ~lict~nre between them is much larger than the ~hirl~n~-ss of the object. The t~,vo detectors work in coincident mode and are s~ald~ed by a passive shield. The main advantage of this geometry is that both detectors are placed on the CA 0224~813 1998-08-06 W O 97t29356 PCTrUSg7/02224 same side of a studied object. Actually, in many applications, the density of fast moving - flats/plates is studied. The m~r,h~nir,~l limitations of the transport system means that the volume under the flats is at premium. In this case the triangular geometry is advantageous because the EC source can be of very small dimensions, down to mm3. A~ dLivt:ly~ the source can be a thin ( < 1 mm in diameter), capillary or plastic rod. Another advantage of this geometry is that two detectors placed in parallel have much lower background tnat two detectors placed face-to-face. A factor of ten improvement in S/B is possible. Furthermore, in this geometry a sophi~tir~te~l graded passive shield can be placed between the detectors.
It can be further enh~nrecl by the use of active anticoincidence detector as an anti-CR shield (see Figure 12). The disadvantage of triangular geometry is the lower geometric ~lptpction efficiency than other configurations.
When compared with previously disclosed configurations adequate for MP-gamma ~1P~ PtPrs~ a new element is the use of a "s~alalol", a thin sheet of high atomic number metal with a hole in w_ich the source is placed. The use of the se~aldlor, permits further background rejection. Note, that for high energy photons, the separator has to be so thick that it considerably limits the geometrical detection efficiency. For soft ~-rays, the mi11imeter thick sepalaL(Jr made of heavy metals (L~ e~ gold, lead) absorbs almost all of the photons. It limits a crosstalk without ~ignifi~nt change of g~omel.ical rlPt~octi~n efficiency.
In X-ray densitometer it is ~pLupLiale to physically isolate the scinti11~tors as much as possible to reduce in(l11cec~ X-ray crosstalk between them. If the thirkn~ss of the samples is relatively small (a few millimeters) this can be achieved by incorporating a few rnm thick sheet of lead or copper into the sample holder. Isolation of the detectors very effectively reduces the background in the single-photon l25I region of interest (ROI). NaI(TI) or CsI(Tl) based detectors show considerable st;lL~iliviLy to the geometry and diameter of the opening in the s~alaluL. Detector isolation has less effect on the background in CaF2(Eu)-based systems because these detectors do not produce secondary X-rays in the 1~5I ROI.Conclusion 3: Generally, the opLil~ Lion of MP-den~itom~ters for EC sources is adifficult, m1~1~ip~ eter process. It is highly sensitive to tne properties of samples to be measured and the required measurement se~ iviLy. It is made even more complicated, because the sources of background are variable in respect to both time and location variable.
Thus, the implement~tion of optimized MPD-densitometers using EC sources is not easy to model theoretically or numerically.

W 097/293S6 PCTrUS97/02224 U~cc.Lail.ly soulces in ~i....rq flPn~it~ ; The most important source of uncertainties is st~ticti~l limitations. In any measurement based on assembly of large number, N, of discrete measurements the etz~tietir~l uncertainty is sqrt(N). Thus, in both gamma d~,iLc,llleter and MP-de~ ll.Pter, at least 1,000,000 pulses are acquired to have the st~tietir~ lcelL~illLy of about 0.1 % in a count rate. The uncel~i.lLy of density measurements is proportional to the uncertainty in count rate, but the coefficient is strongly variable with application. For example, in petroleum applications it depends on the pipeline fii~mloter, thickn~ee of the walls, ~ t~nre between the detectors, and fluid density. In the case of MP-densitometry and using 2~Na source, the coefficient is 3, 2.5 and 2 for pipes with rli:lm~ters of 3", 6" and 9", respectively. Thus, for 3" pipelines the st~tietir-~l ullcelL~ Ly in number of counts of 0.1% leads to density uncertainty of about 0.3%. St~tietir~l u~ L~illLy can be Aiminieh-od by i~ asillg either the source activity or the measurement time. In practice, the existing gamma densitometers are operated in a mode in which other sources of uncertainty are much larger than the st~tictir~l uncertainty. In contrast, the MP-densitometers have been optimized so that the st~ti.etic~l unc~ .Lies have the same contribution to the overall error in density measurement as all other sources of uncert~inti~-~e~
The main limitation of prior-art gamma densiLo..leters can be traced back to:
* high r~Aio~rtive background;
* short time instabilities, mainly due to pileup and dead-time artifacts; and * temperature dependence of the count rate.
In prior-art detectors these sources of uncertainties lead to a few percent error in the density measurement. Note the complicated nature of the sources of uncertainty. For example, to overcome the high level radioactive back~,loulld, the prior-art gamma densitometers use very strong sources (30 mCi or 1.1 x 109 decays per second). Even when ~e~ )g that only 1% of photons are dete~t~A~ this leads to a 1.1 x 107 counts per second detection rate. At this rate, a pileup in a NaI(Tl) scintillator is above 50%, i.e., the pileup deforms the shape and heights of every second photon. Thus, even rudimentary, pulse shape analysis is impossible. Even when using the faster BGO, about 10% of pulses pile up. In contrast, when using MP-densitometry, only 10 microCi source is used, leading to 1.1 x 1O6 decays per second. Taking in account the ~u~llua~ion and the requirement of coincidence, counts rates are about 1O4cps. At this level, pileup is still important in NaI(Tl) but totally n~gligihle when using BGO, BaF2 ,CsF or other fast detectors. Actually, even when using NaI(Tl), electronic means exist to minimi7~ the influence of the pileup. The spurious signals CA 0224~813 1998-08-06 W O 97/29356 PCTrUS97/02224 due to pulses overlap in the crystal itself can be partially accounted for by analyzing the pulse shapes for the selected sample of pulses; this method permits rejection of about 90% of pileup. Even with 10 mCi source in an a~lol,liate configuration, the pileup and dead-time ~ ~ corrections are known to within 0.1 percent. However, this requires the availability of sophistie~tr(l on-line pulse shape analysis. In MP-densiLollleters we have been able to elimin~tr sources of uncertainty. The methods and technical details of pileup/dead-time corrections in MP-de~itullleter are disclosed in the following.
An important uncertainty in del~iLollleLLy is a few percent overnight drift due to temperature sel~ilivily of system elements, mainly scintill~tors. In many applications the densitometers operate outside, e. g., in the petroleum industry in desert conditions the diurnal temperature dirr~ lce can be as large as 50 oC. In recent tests of the MP-dellsilollleter in the CONOCO Test Facility at Lafayette, LA, a shadow of pipeline led to about 2 oC change in the t~ )eldlul~, of the detector. Typically, the light yields of scintillators change about 3% for every 10 oC change of the t~ JeldLulc of scintill~tor, which in the above example 1~ would lead to about 0.6% uncertainty in density if measurements are pelr,lllled with a standard gamma densil{llleter. Furthermore, the change of amplification in a PMT in~ ced by ~elllL)eldLuLe dependent drifts of high voltage and change of amplification in amplifiers have to be accounted for. In MP-den~it >mPtrr, these uncertainties have been elimin~trd by methods disclosed in the following.
In pekoleum applications, there are also sources of uncertainties due to absorption in pipe walls. Note that these limit:lt;ons carmot be evaluated during m~mlf~ctllring, but must be calibrated in field at the petroleum inct~ tion. For example, this is a case of limitations due to ~ ion in steel of pipes. Herein, the merh~nir~l tolerances on the dimensions and wall thicknrs~ are about 5 %, wherein the density measul ~lllent with precision better than I %
is required. In case of MP-densitometry, the PG source, e.g., ~2Na, emits not only two ~nnihil~tion photons, but also 1275 keV photon. Colll~ali~on of the attenuation of these photons with and without fluid, permits to factor out the influence of pipe on the count rates.
Furthermore, we observed that the mech~nic~l precision of the placement of the pipe within the densitometer may lead to a systematic error of about 2-3 % in the case of prior-art garnma den~itometers. These kinds of errors are easy to detect and elimin~te when using the MP-de. ~
Conclusion 4: To overcome radioactive background problem, prior art gamma densitometers had used radioactive sources about 1000 times higher than used in MP-W 097/29356 ~CT~US97/02224 de~sitometry. Thus, pileup and dead-time co~ ,Lions are the main source of unceLL~ y in prior-art gamma del~ik3~eters, but can be minimi7~cl in MP-densitometers. When the background is pushed to below 10 cpm, the domin~ting sources of uncertainty are pileup corrections which limit the precision of NaI(Tl) based MP-densitometer. At that level, Le.. pel~uie corrections become d~ ;-lg and il~ dliV~; methods of their eliminzltion used in MP-~ n...~t~rs are disclosed.
Sources of ba~. vu. d in Multi Pho~on D~ . (MPD) ~y~ There are many sources of background and noise in nuclear radiation detectQrs. We believe that multiphoton based devices are the first very low r~io~ctive background detectors (VIBDs) which can be reasonably priced (< $30,000) and small enough (10 x 10 x S0 cm) for general use in field applications.
We documented the following sources of background in scintillator-based gamma-densitometers and in MP--lencitom~oters:
A1 radioactive cont~min~tion of detectors;
A2 radioactive co. .~ ion of PMTs;
A3 radioactive co..~ .*lit-n of shields;
- A4 high energy g~mm~ from the ellvh~ ellL, Bl neutron infl~cerl ~mm~ from the detectors, PMTs and shields;
B2 cosmic ray infhlce~l g~mm~ from the detectors, PMTs and shields;
B3 direct hits due to cosmic rays;

Cl dark ~;u~ L~ of PMTs;
C2 cosmic ray in~ ed dark ~;Ullt;llL~ of PMTs;
Dl electronic pickup; and D2 vibrational pickup.

Sources of background A1-A4 and Bl-B3 are dependent on detector size; typically they grow linearly with the mass of the detector. Sources Cl-C2 are dependent on the size of PMTs, and grow proportionally to their surface. The background sources D1-D2 tend to be independent of the size or type of the detector. Furthermore, they are much more Ull~)Ol L~ L
for photons with energy lower than 50 keV.

CA 0224~813 1998-08-06 W 097/29356 PCT~US97/02224 The first group of backgrounds (A1-A4) leads to backgrounds of the order of a few cps.
Also, the second group of backgrounds (B1-B3) is an important source of background in prior-art gamma densitometers. Actually, the background due to CR in~ ce~l dark ~;Ul~ S
in PMTs (C2) are proportional to background due to a direct hit by CR (B3), which is much easier to count. Thus, in MP-de~ eters, the background due to direct hits by CRs is periodically measured and used to evaluate the other sources of ba~h~ ulld (B1, B2 and C3).
If the sum of these background is above a preset threshold (periods of high CR activity) they can be accounted for in the density mea~u~ lellL. The background sources from the third group (Cl-C2) are the most ~lifficnlt to reject by haldwdl~ means.
MP-densitometers use a synergistic combination of coincidence, haldv~ale means and sophictir~t~tl pulse shape analysis. Coincidence permits about a 100-fold decrease in background, and additionally our pulse shape analysis software permits rejection of about 90 % of the background. This requires use of on-line pulse shape analysis, including a~pLoplidle ~orLwdl~. Software for background rejection limits the m~xim~l count rate, and is very difficult to implement when the actual detector count-rate is much larger than 10 k~z.
This limit~ti~n leads to highly innovative ~lldlegy of interleaving mea~ulelllents and calibration runs.
The electronic and vibrational pickup noise is very much system and site dependent. In petroleum industry applications it tends to be large, because the detectors are placed close to pipeline, whereas the data acquisition system, i.e., the COlll~)Ult~l, iS often placed in a trailer. For example, during experimental tests of MP-densitometer in CONOCO Test Facility, Lafayette, Louisiana, this ~1ict~nre was about 90 feet. Even when using well-grounded coax cables, the observed electromagnetic noise without coincidence is about 100 cpm. In prior-art garnma densitometers, this source of noise is considerably lower than the r~-lio~ five background. However, in MP-densitometers wherein the radioactive background was ~iiminich~od a thousandfold, the electrom~nPtir illl~lr~ nce is an important source of ullce~illLy. We disclose the use of:
* coincidence counting wherein the signal from two detectors must arrive within 100 nsec;
* pulse shape analysis to evaluate the inflll~n~e of electronic h~ rel~llces;
* use of specially shielded BNC cables, or preferably fiber-optics, to ~imini.ch this source of background; and * use of custom-designed low noise electronics.

W O 97/29356 PCT~US97/02224 All sources of background can be cl~ if~-d into two categories. First, there are sources which are essentially constant, e.g., A1-A3 and C1. These sources of background can be accounted for by a~r~lidl~ calibration, and their influence can be ;.~ cl (but not elimin~t~cl). However, there are many sources of background which are highly variable, S both geographically and as a function of time. For example, ellviiul~llental r~ active background is often ~lomin~t~t1 by radon and is strongly weather dependent. Also, the background due to CR is not only geographic position dependent, but is also motl~ ted by solar flares activity leading to changes of geom~gn~tic cutoffs. Thus, the radioactive bacl~,l~und in prior-art gamma densitometers cannot be calibrated, leading to spurious resi-1us~1~ in density mea~.ul~lllents. On the other hand, the background in MP-deLeeLc.l . is .u~ ssed about a thousandfold and the background variations are negligible.
(~onclusion 5: Very low back~luulld levels can be achieved at the earth's surface only through the synergistic use of a plurality of methods, of which coincidence and sophi~tir~tPA
pulse shape analysis are the most important. MP-del~ o...eters are superior to prior-art gamma densitometers because all time/geographic location artifacts due to bach~ ulld are elimin~tt~d Tmrl~ ;nn~ of Mp~ In MP-densitometers according to the invention, the coincidence beLw~ell two photons emitted by an applu~liate source is used to ~limini~h the r~io~tive background. For example, when using a suitable long life PG
source ~22Na), scalL~lillg of the ~nnihil~ti~n photons leads to change of both total llulllb~l of ~let~cte~l photons and to ~.ub~ Lial change of the ~ec~l.ull. AlgoliLllllls for estim~tin~ t_e density of t_e studied object depend on this change of the spectrum. Thus, the use of scintillators with the highest energy resolution, namely NaI(Tl) is disclosed. However, for larger objects, this leads to either reduction of detection efficiency when 3 " thick crystals are used or increase of background when 4" or 5" -dialllelel crystals are used. Por example, in de~si~ollletry of petroleum products in flow, the optimal NaI(Tl) crystal ~i~m~ter is a function of pipeline (1i~mPt~r. For pipelines with a rli~m~tt~r up to 6", the 3" crystals are optimal. For 10" and 16" pipeline ~ mPterS~ the optimal crystal diameter is 4" and S", le~e-;LiYely.
The rise-time of scintillation light in NaI(TI) is about 200 ns, and typically about 500 ns pulse shaping is used. For sources in excess of 5 microCi this leads to tne pileup artifacts.
We observed the uncertainty of about 0.5% in count rate for 3" pipelines filled with petroleum. This leads to 1.5% precision in density measurement. However, the use of the CA 0224~813 1998-08-06 geometry shown in Figure 3 with each source about 5 microCi ~imini.cht~:s by a factor of two pileup and permits a better than 1% precision in density mea~,ulc~ lll. Furthermore, the uncertainty in count rate has a ~1i".il.ixl,il~g infl~1enre on precision of density measurements for larger pipes, e.g., for 6" pipe the density measurements uncertainty of about 0.5% is possible with two 5 microCi sources.
MP-densitometers may be improved by the use of scintillators different from NaI(Tl).
For example, CsI(Tl) has only lllal~,hlally lower energy resolution but about twice higher stopping power. Alas, it is a slower scintillator than NaI(Tl) and tnus leads to higher pileup alLi~dcL~,; CsI(Tl) should not be used when the source activity is above 1 microCi. Use of BGO is suggested, even if this crystal has about four times lower energy resolution. The advantages of BGO crystals are about three times higher stopping power and faster scintillation. Thus, for the same stopping power the pileup is about a factor ten smaller and also background is sulllewlldl lower. Finally, when a very high precision of density measurement is required, say 0.1%, the higher activity sources may be used to provide the 1~ a~ idle st~ti~til~s. In this case even in the symmetric sandwich geometry, the errors are mainly due to pileup. To ~limini~h pileup, the fastest detector with a~rvpliate stopping power should be used. Thus, we disclose the use of BaF2 GSO(Ce), YAP(Ce), CsF, and CeF3, which are ten times faster than BGO and a hundred times faster than NaI(Tl). Note, that if the pileup is not critical, the use of BaF2, GSO(Ce), YAP~Ce?, and CsF is highly counter-intuitive because it features lower stopping power, worse energy resolution and requires very costly quartz PMTs.
In MP-densitometers using EC sources, the X-ray detector may be built of CaF2(Eu) rather than NaI(TI). Using CaF2(Eu) some sources of background (especi~lly the dark ~;ull~nl~,) can be rejected better than when using NaI(Tl). Alas, CaF2(Eu) is also a much slower scintillator, and typically the shaping time of a few microseconds is used. Thus, it should be used only when the radioactive source is below 0.5 microCi, i. e., only in the case of rather small objects.
Finally, we disclose the use of yttrium~ .. i .. ,-phosphate scintill~tor (YAP(Ce)) instead of NaI(Tl) as a X-ray detector in an a~,yl~ ic configuration of the MP-densiLollleter.
YAP(Ce) is not-hygroscopic and is more robust mech~nic~lly than NaI(Tl). Second, it is a very fast scintillator and pileup is negligible even for 10 microCi sources.
Conclusion 6: We disclose the use of a plurality of heavy inorganic scintillators in construction of MP-densitometers. Their advantages and limitations in diverse applications o 97n9356 PCTrUS97/022Z4 are rli~cn~se~l. More specifically, the utility of fast scintill~tors (YAP(Ce), BG0, BaF2, CsF, and CeF 3)iS disclosed.
Te~ .aLu~ stabil;, ~ h./co~ .t..~ n sl-h~mP~ ;n MP~ n~i~n..~ rcnrzlt~
measurement of density requires a high stability detector system. The observed drifts are predu..,i.~,."~ly due to changes in ambient temperature, which cause drifts in the yield of the scintillators as well as changes in gains of the PMT's and amplifiers. In prior-art gamma densitometers, the temperature drifts are the domin~ting sources of mea~u ell~ent errors. For example, many oil wells and pipelines are located in deserts, where the diurnal L~nlLIclaLure variations can be as large as 50~C. Similarly, the temperature gradients are very large and site dependent in the ml-t~ rgic hl~lu~Lly where, the gamma densitometers are often placed very close to moving slabs of very hot metal.
Typically, the light yield from NaI(Tl) scintillator is changed by up to 3 % when temperature varies by 10~C. The typical dependence of a coinrirl~nt count rate in NaI(Tl) based MP-del~iLun~c~l on L~ eldLulc is shown in Figure 13. For majority of scintillators including NaI(Tl), the light yield is a monotonic function of temperature and decreases for lower temperature. However, the temperature dependence of scintillator may be more complicated in particular cases. For example, in BaF2 there are two components, one very fast e..,i~ E an UV light and another much slower ~,..illi..g in visible. The first component is almost temperature independent, wherein the second component is ~Lloll~,ly temperature dependent and becomes negligible at above 65 ~C . Preferably, with BaF2 in MPD-densi~ etry, the crystal temperature is kept above 40 oC to ~limini~h artifacts.The main method of correcting for lel~)eld~Ul~ artifacts is to stabilize the temperature of the crystals. Each detector may be independently stabilized in temperature. The sçh~m:-tir drawings of the telll~el~LIllc stabilization system are provided in Figure 14. For example, in the case of petroleum industry, each detector is placed in a massive, explosion proof box typically made of ~1-..,1;ll--lll The a~l.,plidl~ Dewar with internal dimensions of about 3.5"
is mounted inside this box. PMTs are very bad heat conductors due to a high vacuum inside and only a very thin wall of glass. Thus the Dewar covers essçnti~lly only a scintill~tor and PMT assembly and is additionally inc~ tl~-l around the PMT neck with porous plastic material. The fast reacting ohmic heaters, e.g., high power transistors are placed on the good conductor metal plates, typically ~lntninl-nn or copper attached to the crystal. The temperature is homogenized inside the Dewar by the use of mini~hlri7P~ fans. Thetemperature stabilization is obtained by the direct feed back using a fast electronically CA 0224~813 1998-08-06 W 0 97/29356 ~CTrUS97/02224 readable thermometer. We also disclose a somewhat more sophi~tir~tPd temperaturestabilization system in which all elements of stabilization system (heaters, fans, thermostats) are COI11~ULG1 modifiable. A fee-lha(~k loop is established between the output of scintillators (position of photoabsorption peak) and the parameters of the temperature stabilization system.
Practically, it is preferable to use only heating rather than both heating and cooling. In this case, the LGnlpGldLul-, is established higher than the expected m~ximllm temperature of the ~llV ir OlllllGllL .
Such a temperature stabilized system permits operation at an ess~nti~lly stable temperature regime, where the le~elaLulG drifts and fillchl~tions are lower than loC. However, even this L~ peldlu~G stabilization is not sufficient for 0.1% count rate stability. Thus, the additional temperature compensation system is used in MP~ trr (see Figure 15).
Herein, three ~Gmpe~dLulG sensors are placed in each detector; two on the crystal and one on the integral high voltage supply (HVS)/amplifier module. The thermn...~Le,~ output is arnplified by a DC-DC amplifier to a signal of about 5V. The information from these sensors is sent several times a second to a digital oscilloscope card and is on-line analyzed by the co",~uleL. We disclose the use of multiplexing system, wherein the outputs of the three thermometers are sequentially inputted on DSO. To ~limini~h the electronicil~t~,.r~lGnce, the analog signal is sent by well-shielded BNC cables.
Optionally, the voltages from electronic thermo,lleLt;l~ are electronically modified in a voltage-to-frequency C(311VG1~ , and are sent by the cables that are used for tr~n~mi~sion of pulses from detectors to data ~cq~ ition CO1n~ULGL~ In yet other implern~nt~tion, the electronic thermometer has a digital RS320 interface and is directly coupled to the input ports of the data acquisition CO111~ULG1. After reception the data are software filtered to remove spikes due to electrom~gn~tir hlL~lrGll,nce and averaged. In all these implement~ticn~ the reliable LGlllpGldLurG mea~ulGnl,GllL~ with precision of about 0.2% were achieved. Note, that the tG"lp~lature of crystals is often a few degree Celsius different from temperature of PMT
base. This is accounted in software in which three LGlll~.;ldLulG compensation curves are cs~lc -l~te-l independently, for scintill~ti~ln crystals, high voltage supply/dividers and signal amplifiers, respectively.
Both heating and cooling by means of the Peltier effect may be used so that the crystals LG1~C1~LU1~G is stabilized below the ambient L~",~e,~LulG. This permits use of the operational temperature, in which the temperature coefficient of the scintillator is minims.l.

CA 0224~813 1998-08-06 W 097/29356 PCTrUS97/02224 The temperature dependent drifts of the high voltage supplies, preamplifiers and ADC's are imporLant, and it is preferable to use stable electronics and high voltage power supply for the PMTs. The high voltage power supplies for the PMTs have been stabilized by introducing a high gain negative fee-lh~r1~ and using as a reference Max 580 chips with low temperature coefficient (1-2 ppm/~C). We also use 1% metal film resistors with a low temperature coefficient. The voltage divider for the PMT is based on the same type of resistors. The temperature dependence of the preamplifiers has been minimi7~1 by using highly stable operational amplifiers and di~.Llil)uLillg the gain so that each amplifier cascade has a gain lower than five. The result of the work of temperature colllpcnsdlion scheme is shown in Figure 16, where the row data, and corrected data of overnight measurement are plotted.
The ~empcldLulc se~ ivily and long term stability of DSO's are especially important in coincident systems. The relative drifts of two ch~nn~ls as a function of ambient lclll~cldlulc change is often larger than the intrinsic resolution of DSO. Thus, we disclose innovati~e 1~ controls for our MP-de~v.i~oll,cL~l~.. First, at the known repetition frequency (typically every few minntçs), the two detectors are co....l...l~d into two channels of DSO. This permits us to distinguish if long term drifts are due to scintill~tcrs/front electronics or data acquisition, especially ADC's. Second, after each measurement period, the calibration procedure is inhi~t~-l wherein the pulse from the well-stabilized pulse ~,ell~ldLol is applied in parallel to both channels of a data acquisition system and counted in both OR and AND modes.Actually, to check for possible nonlin~rifies of data acquisition, the pulses of diverse amplitude are counted and their shapes registered.
Conclusion 7: Advanced temperature stabilization/colll~el~dlion systems can be used in MP-densitometry. The tempeLa~S of the crystals are stabilized within a loC by active temperature control system. Additional and independent electronic systems measure the actual temperature of the scintill~t~7rs and HVS/amplifiers with precision better than 0.2 oC.
Software permittin~ on-line c~ el~.alion of the count rate is disclosed.
Two modes of operation of MP~ n~;lu~ .; TheapplicationsofMP-densitometers can be divided into two classes. In the first class, density measurement with precision of a few percent is required and long term stability of the count rate should be about 0.3 to 0.5 % .
The above described t~çhniqllçs of telllL,tl~ul~ stabilization and compensation are fully sufficient. Furthermore, a periodical calibration of detectors is recommended.

CA 0224~813 1998-08-06 WO 97/29356 PCTrUS97/02224 In the second class of applications, e. g., in two phase measurements of oil/water mixtures, better than 1% density precision is required. Thus, the count rate uncertainty should be 0.1 - ().3 % which is really rlimrlllt to achieve. To enable this count-stability a few innovative techniques are used in MP-dencitomet~rs to further ~liminich the rem~ining S Le~ eldLulc inrlnre-l drifts in the PMTs and scintillators. These drifts can be colll~ cal~d by adjusting the gains of the amplifiers, so that the acquired spectra will be unchanged.
Electronics permits control of these gains. When coupled with Lclll~e.alulc readout devices placed in the detector system this allows adjusting the gains to compensate for telll~cl~ture in~ e-l drifts. The lclll~cldLulc-to-gain calibration for each detector can be efficiently obtained by a series of measurements in non-coincident (OR) mode and use both OR and AND mode acquisition. The entire calibration procedure is ~lltom~t~d and can be periodically performed to colll~clls~te for long-term drifts.
An optional element of the long term stabilization of the detectors is the use of calibrated low energy photon sources concicting of at least two low energy lines. These calibration lS energies are sufficiently dirrcl~llL so that these two lines can be distinguished by a detector.
Preferably, this is a few hundred nCi source, that is mech~n;r~lly shielded and the obscurator is colll~uLel activated.
Therefore, the MP-dencitom~-ters use two modes of data acquisition: measurement mode and calibration mode. In measurement mode the main effort is to ~liminich the pileup and dead-time effects. Thus, the fast electronics and system of fast counters are used. In the calibration mode, the main effort is to evaluate the influence of dirr~lc.l~ background sources (natural radioactivity, CRs, electronic hlL~lrt;lcl~ce). Also, the efficiency of temperature stabilization/coll~ellsa~ion system is checked. In calibration mode, the pulses from the c~,ullLel~ go to a fast digital storage oscilloscope (DSO) working in the sampling mode.
These data are dumped into colll~uLel memory and processed in parallel with data~q~ ition. The software performs the following steps:
* measuring the pulse shapes;
* calc~ ting an energy spectrum;
* estim~ting the pileup and dead-time corrections; and * measuring the external electrom~gn~oti~ hlL~lrcl.,llces.
From such short calibration runs, the look-up calibration tables are upgraded which are used to adjust the raw data. In calibration mode the data are obtained for both detectors in CA 0224~813 1998-08-06 - W O 97/29356 PCT~US97/02224 OR as well as in AND modes. Thus any differences between the detectors showing amalfunction can be analyzed.
The MPD normally operates in OR and the AND mode, with considerable background reduction through pulse shape analysis. In a 3" ~ m~tt~r and 3" thick NaI(Tl) crystal, the typical background without shape analysis is about 100 counts per second. Using the pulse shape analysis, in MPD the background in the OR mode is 50 counts per minute. In AND
mode, the MPD instruments permit about thousandfold lower background. During the data acquisition both non-coincident and coincident events are identified and counted. Upon completion of the acquisition the MPD outputs both OR and AND counting data. The AND
data serve to establish the material density and the OR data are also analyzed (either on-line or periodically) to permit the reliable calibration of the system, e.g., to monitor the temperature drifts and pile up problems.
The software r~,L~ s the pileup corrections to the counting rate. These corrections are very different at OR and AND mode of operation. It is hllpolL~,.L that the pileup corrections are ~stim~t~C~ at the level of raw data, i. e., before background rejection. Self-diagnostics and self-calibration programs permit concordance of the OR and AND counting rate data with accuracy of about 0.2%.
In an ~uLo~ Lic device calibration/ROI setting algol illllll, the calibration procedure depends on the MP--1~neitnm~t~r configuration, and in the following the case of "modified symmetric sandwich" using 22Na source, is described as an example. The user can request the software to perform this procedure at any time but typically the calibration runs are performed each hour. No calibrated source is n~ e~ l, because the reasonably high activity (1- 5 microCi) 22Na source is already present. The program acquires 100,000 events with all pulse-height and pulse-shape rejections enabled. Once the acquisition is fi~iehP-1 the spectrum is analyzed to determine the ROI for a photoelectric absorbtion peak of ~nnihil~tion photons. The count rate (cpm) within this ROI is determin~od. As the actual activity (dpm) in the source is known, the detection efficiency may be established from the ratio of cpm/dpm. To estimate the absolute activity of the calibration source, the program starts a second round of data acquisition with pulse-shape rejection disabled (as the calibration source activity is high at the ~11 keV line, there is no need to reject the background, and no real events are discarded). Spectra for both detectors (A and B~ are built until 200,000 events are acquired. Subseql~e-ntly the spectrum of each detector is analyzed to estimate the count rates in the l-photon and 2-photon peaks and efficiencies of the detectors are c~ te-l The W O 97/29356 PCTrUS97/02224 data are checked for c~ r-y by cc~ d~ g the çstim~tes from detectors A and detector - B. Also, the pileup pulses are ~l~tectecl and t_eir frequency estim~t~l With the pulse shape analysis available, the pileup pulses are easily ~etPct~d due to a characteristic, two peak structure. With st~ti~tic~s of about 200,000 pulses analyzed the sub-set of about 2,000 pileup pulses are typically analyzed, for which the ~ t~nre between two peaks is measured. This permits st~ti~tir~l analysis of the pileup artifacts.
After the autocalibration/ROI setting is complete, the system is ready to count samples of the same geometry as the last calibration sample used. Acq~ itiQn is pel~~ ed either for a preset time or until a preset llulllbel of counts (determining the st~t;~tir~l uncelLdi,l~y of counting) have been acquired in the ROI. The program then e~ s the actual decay rate in the source using the previously computed DE. The sample counting data are then stored to disk in an ASCII data file which can be lL~ll~d to any database or spr~ h~oet program for analysis.
If possible, the calibration is pe,ro,ll,ed with pipelines empty or filled with water so as to permit cc",l~dlison with previous calibration runs. However, the calibration can be performed with any fluid in the pipeline because it is mainly based on pulse shape analysis of non-attenuated :mnihil~ti~n photons in OR mode.
Conclusion 8: In a calibration mode using pulse shape analysis based on a digital storage oscilloscope, interleaving the measurement and calibration runs perrnits improved background n.. ~ i"~ It also permits better control of artifacts due to pileup and ~ eldLulc;
in~lllcefl drifts. The calibration methods are highly cO~ U~ lL~l~ive.
EXAMPLES
Specific embodi-ll~l,L~ of the invention will now be described in more detail, by way of example only.
C~ Ja~dlive tests of MP- and ~nli~rd d~ncitom~tf~rs: An MP - d~llsiL~llleter formeasurement of gas/oil/water mixture density in the density range 0 - 1.05 g/cc was tested both in the laboratory and in a CONOCO Test Facility at Lafayette, LA.
The two detectors with 3 " NaI (Tl) crystals were coupled to 3 " PMT and bases (HVPS/amplifier/shaper) and assembled inside explosion proof al- l . l l i l ll l l l - boxes . The signals were processed by amplifier/shaper electronics and then Lldll~r~lled to a PC 486/66 equipped with proprietary data acquisition cards. The colll~uL~;l was in~t~11e~ in the control room about 90 feet away from the detectors. The coincident trigger selecte~l the coincident signal CA 0224~813 1998-08-06 W O 97/293S6 PCTrUS97/02224 from two detectors within the energy interval 100 - 600 keV, which correspond to both scattered and not scattered ~nnihil~ti-~n photons.
The MP-densitometer is working in three modes: in two calibration modes and in the data ~rq~licition mode. In the first calibration mode the signal from a low activity (0.3 microCi 22Na) source is processed by a proprietary coincident card and is subsequently analyzed with the DSO. The coincident trigger is tuned for a selecte~l region of interest (ROI) in energy.
The ROI depends on a geometrical configuration of the system and may be dirr~,,GllL for two detectors in an asymmetric configuration.
For density measurements the MP-densitometer is calibrated using the 2 microCi source in two density points: with an empty and water filled pipe. In this mode the smoothed count rate and Le~ Glatures (see below) are stored and lenl~eldture co~ el~.dLion look-up tables are C~ t~
In data accLuisition mode (with 2 microCi source) the signal from a coincident trigger is ~rcllmlll~te~1 for a given number of times (100 times per minute). The use of a st~ticti(~l rejection procedure is disclosed, wL~ ill the count rates that are more than two sqrt(N) off from the average value N are rejected and replaced by a weighted average over the closest four points. Then all the counts are summ~-l over and this number is accepted as a real number of counts during a given time interval (1 min). The same procedure is applied to the signals from two temperature sensors installed on the two detectors. 3~very minute the signals from the Lelll~e~dlul~ sensors, coincident (AND) count-rate and non-coincident (OR) count rates are written in the file. The density of the l~ Lule inside the pipeline is c~ t~-1 and stored using the coincident count rate corrected for the current detector temperature, and the reference points obtained during the calibration run. The time dependence of the l~lll~eldLul~s of both detectors, coincident count rate and density are presented on the monitor in the real time. The digital output density signal is converted into analog DC voltage in the range 1 - 5 V.
Selected results of the tests are L~l~SellL~d in the Figures 17-19. In Figure 17 the results of measurer~ents of oil/water mixture density for dirr~lellL fractions of oil in the salt water are shown. The density was measured by MP-densitometer in~f~lw at the vertical 3 " pipe two feet below Gamma Trol densitometer using 28 mCi source of l37Cs. The density was averaged every 60 seconds by both densitometers. The density of oil and water were measured separately in before mixing by Solartron densitometer with precision better than 0.1%. The densities of oil/water l~ LuleS calculated on the basis of these me~ u7_ments a~

CA 0224~813 1998-08-06 W 097/29356 PCT~US97/02224 the-readings from CONOCO Test Facility flowmeters are shown in Figure 17 as the data by Solartron. The precision achieved for 60 second measurement time by the prototype of MP-densitometer was comparable to the one achieved by Gamma Trol densitometer, which used 15,000 more active source. On the other hand the system~tif~l uncertainties of Gamma Trol densitometer in the region of low water fractions are substantially higher than these of MP-~lencit- m~ot~r. The same data after averaging during 4 minute intervals are shown in Figure 18. In this case the uncertainties of MP-densitometer with 2 microCurie source are close the uncertainties of (:~mm~ Trol d~ t~r.
The results of measurement of gas/water mixture density are shown in Figure 19. In this case the precision of measurement is not ~let~rmin~d by st~tictir~l uncertainties, but by systematical unc~l iahl~ies of a few percent due to turbulent flow of gas/fluid mixture in the pipe. Both densitometers gave close results for density measurements and for both del~il~,lllc:lel~ even for one minute measul~lllent time the precision of density measurements was d~ .ed by the ~y~ l uncertainty.
Conclusion 9: The use of MP-densitometers permits comparable or better density measurements than prior-art gamma densitom~t~rs even if 15,000 lower activity of source is used.
MP-d~n~if~m~1t~r ol?l;...;,..~-l for EC sources: There are important applications of MP-densitometry for smaller objects (< 3")~ ~uch as the measurement of tluckness of tbin foils and slabs of metal, plastic and paper. ~- this case, the attenuation of gamma rays is small and sources with lower energy should be used. The MP-densitometers works best of all, when the sample absorbs from about 50 to 80% of emitted radiation. For the X-ray, however, the attenuation coefficient is very dependent on the atomic number of the absorbing media because the photoelectric effect dul.lill~l~s. Thus, the practicable energy range is from ~5 about 20 keV for paper/plastic/~lllmimlm foils to about 100 keV for steel or cooper foils.
The a~,plupliate configurations are the "compensated flat ~ .,llellic sandwich " (see Figure 9), "asymmetric sandwich with separator" (see Figure 10), and "modified triangular"
(see Figure il). The relative advantages of this configurations have been discussed above.
In the three detector configuration (see Figure 12), the third, anticoincidence detector is used to elimin~t~ the cross-talk between the two signal detectors.
For small objects excellent statistics can be obtained with a small activity source.
~.csllming, that 50% of photons are absorbed in the object, and that the detection efficiency is about 50%, the 1 microCi source leads to a count rate of about 8,000 cps ~50,000 cpm).

CA 0224~813 1998-08-06 W 097/29356 PCT~US97/02224 This permits the 1% st~ti~tic~l precision in density measurement for a data ~q~ ition time of a few seconds. Alas, at this count rate pileup artifacts are important. Thus, the software based pulse shape analysis using DSOs and subsequent software analysis is not ~lecrl~te and the background due to CR inrlnre-1 cross-taL~ between detectors will dominate. To implement the cross-talk rejection in "cu~ ensdl~d flat symmetric sandwich " and "asymm~tric sandwich with s~alaLol " we disclose the use of a fast digital storage processor (DSP) based pulse shape analysis.
Generally, NaI(Tl) scintillator crystals are believed to be optimal for delec~ioll of X-rays.
In coincident mode, the background rejection is inversely proportional to the square of the energy resolution, which for NaI(TI) is about 50% better than for other scintill~t rs.
Fu~ ore, among scintillators with reasonable energy resolution (NaI(TI), CsI(TI), YAP(Ce) and CaF2(Eu)), the YAP(Ce) and NaI(TI) scintillators are the fastest. Once more, according to prior-art practice, background rejection is proportional to the square of the timing resolution. The theoretical estim~t~s suggest that the background in an MPD based 1~ on NaI(Tl) scintill~tQr should be about four times lower than when using other scintill~tors.
In practice, an MPD system can achieve a background of about 0.5 counts per week (0.5 cpw).
However, we also observed some real inconveniences of NaI(Tl) based devices:
* NaI(Tl) is mech~ni~lly fragile, e.g., it often cracks when submitted to ~ eldLule gradients and/or during transportation;
* NaI(Tl) must be hermetically sealed, and the thin Al or Be foils covering the front surface of the scintillator is easy to rip off when samples are placed in its vicinity; and * when samples of larger ~ are used, the background in MPD systems based on NaI~:ll) deteriorates rather fast due to crosstaLk between the crystals.
For example, there are some changes in NaI(Tl) crystal propclLies over a 2-year period in about 40 NaI(Tl) crystals which we have tested. Some (about 10%) crack from thermal stresses while others (about 15 %) are turning yellow due to their hygroscopicity . In contrast, no ~ignificzint variability in crystal plvpel~ies was observed for twenty CaF2(Eu) crystals over one year interval.
A series of mea~ule~ L~ in which for soft-X rays NaI(TI) was replaced with CaF2(Eu) scintillator confirmed our theoretical prediction that a simple replacement leads to a factor of few i~ cases in background. However, following a step-by-step process of irmovative changes in readout electronics and after developing an additional 10,000 lines of software, CA 0224~813 1998-08-06 W 097129356 PCTrUS97/02224 we have achieved considerable improvements. ~ ellLly, CaF2(Eu) and NaI(Tl) based MPD
~ systems have comparable background.
CaF2(Eu) has a ~ulyli~hlgly low background for 125I detection because in AND mode the background is do,.~ d by soft X-ray emitted and absorbed in one crystal coincident with some source of energy ~l~t~-ctP.-1 in the second crystal. When NaI(Tl) is used, any absorbtion of an external photon with E > 35 keV in the crystal leads to remission of 26 keV or 32 keV photons from lca,ldllgt;lll~l~L of atomic shells. Thus, the characteristic iodine X-rays are emitted when the NaI(Tl) crystal is used. These cannot be ~ tin~ h~l from the 25 and 31 keV photons emitted by l2sI. FO1LU1~L~1Y~ CaF2(Eu) consists of only low atomic l~u~bel e~ t~. Thus, the characteristic X-rays have less than 15 keV energy and can be dirrt;,~ ted from radio-iodine X-rays. Thus, as a rule of thumb, NaI(Tl) or CsI should be used for EC radioisotopes with atomic number of either less than 40 or larger than 70. For EC isotopes with atomic number between 40 and 70, CaF2(Eu) is a ylc;r~;llc;d scintill~tnr.
In MP-dellsiLuu,eter using EC sources the crystals themselves are preferably very thin and the surface of the PMT cathodes is about fifteen times larger than the surface of the crystals.
A high energy CR striking one of the PMT anodes gives rise to an avalanche of electrons which are subsequently amplified. Such pulses lead to an ayyalellL energy deposition much lower than the energy of the CR, i.e., there is a considerable overlap between the energy ~ye~;Llulll of CR infl~lced dark current pulses in the PMTs and the energy from radio-iodine.
In MP-densitometers optimized for EC sources and using scintillator-PMT combinations this source of background is quite illl~?olL~u~L. FolLullatt:ly, the energy deposited in the scintillator typically leads to pulses longer than the CR in~ çe~l pulses in the PMT, which has a char~cteri~tir time constant of less than a nanosecond. The dirr~ ce in pulse rise times between CR inrlllce(l PMT pulses and those created in CaF2(Eu) is large; 0.1 ns and a few microseconds respectively. In an embodirnent of a CaF2(Eu) based MPD system, about 95 %
of ~R in-lllre-l PMT artifacts are rejected on-line.
For low-background counting the scintillators used in X-ray detectors should be shielded from radioactivitv in the photomultipliers to which they are coupled, such as from beta particles and low energy photons. For example, a 5-mm thick ~uartz window is placed between the PMT and the scintillator. Quartz is ~l~;rell~d both for its excellent optical pLopelLies and its high purity; quartz matches the optical density of CaF2(Eu) very well and is acceptable for NaI(Tl). We did not observe any increase in background due to radioactive co"~i...,i.~tion in quartz. Further improvement can be achieved by using materials with W O 97/29356 PCT~US97/02224 higher stopping power than quartz. We disclose the use of high-purity GeO2 and g.,, ~
based glasses for this purpose. They have very low intrinsic radioactive background and their higher atomic number and density is an advantage over quartz windows. Such windows with a few milli-~ . thirkn~se will efrlciently stop low energy photons as well as beta particles without degrading the optical qualities of the scintill~tor/PMT ~.y~.Lellls. Actually, the optical plup~lLies of gelica and ~ glass match NaI(Tl) better than quartz does.
Even higher density glasses based on lead may be used as well as high density l~ alc;llL
crystals such as PbF2 and bi~ lULll ge. ..,~..il~ (BGO). In the case of BGO, undoped crystals should be used so as not to gen~,ldle artifacts due to scintill~tir)n within the BGO. The optical density of these m~tPri~le is higher than for CaF2(Eu) or NaI(Tl). The use of a thin layer of special optical greases, e.g., powdered PbF~ in silicon grease can match the optical properties of the scintill~tor and window and the window and PMT.
Both background and ~l~tPction efficiency in the modified aSymm~tric sandwich geometry depends on the ~limPneione of the small crystal. For many applications the source can be small, often about 1 mm in ~I;AII~ . An optimal cor~lguration is one with two crystals of both dirrel~ m~tPri~l and dirr~lGllL ~ mPt~r~. The detector close to the source should be small and thin, typically 0.75" or 1" in rli~m~ter. However, the detector on the other side of t'ne ~ed~.uled object should be much larger, typically 3" or 4" in rli~mPtpr. In a ~ r~ d design of MP-densitometer using EC isotopes, the two detectors are dirrtlG..I;~l~cl not only by their dimensions but also use dirr.,.el,L scintill~tors; depending of the type of source used it should be either NaI(TI), YAP~Ce), CsF or BGO.
A pl~r~ d implemPnt~tion of the MP-densitometer o~Li~ d for det~ction of EC' isotopes, e.g., l25I, is an "asymmetric sandwich", c~ g of two modules, the smaller of them preferably placed in a low radioactive background shield con~icting of heavy metals, 2~ typically a sandwich co~ g of a few millimPters of Pb /Sn/Cu. The shield is essentially cylinf7ri~ 1, with the thin window close to where the scint~ tQr is placed. Also, there is a thinner shield be~wt:ell the PMT and the PMT base cont~ining high voltage supply, high voltage divider, and preamplifier. Additionally, the somewhat thicker shield is placed after the said PMT base.
Each of the detector modules consists of the following elements:
* inorganic scintillator with thir.knP~ ed for a given ellliLL~l, high purity optical window, placed between scintillator and PMT;
* selected, low radioactive background PMT;

CA 0224~813 1998-08-06 W O 97/29356 PCT~US9710222 * a graded passive shield between the PMT and PMT base assembly; and - * PMT base assembly con~i~ting of HV supply, high voltage divider and preamplifier.
I25I is an applo~liate source for a plurality of MP-del~iLollletry applications, preferably using flat detectors (1 mm thick NaI(Tl) or 1.5 mm thick CaF2(Eu)). Thinner scintillators decrease the DE of the system while for larger crystals the signal to background ratio effectively ~limini~h-os. The scintillators are coupled through quartz windows 3-5 rnrn thick to a high resolution PMTs which are selected for low background. The PMT signals are amplified and shaped using ~lv~lictary electronics built into the PMT bases. To reduce the flux of background photons from the bases to the scintillators the bases are isolated from the PMTs with 5 mm of lead and 1 mm of copper plating with holes for the PMT pins.
At least one of the detectors is placed in graded lead+tin+copper shields (typically, 0.25"
lead, 2 mm tin, 1 mm copper). The detectors are placed face-to-face a few millimeters from one another and a crosstaLk eli.,~ tor is placed between one of the detectors and the sample.
Typically, this is a 1 mm thick copper sheet mounted in a lead frame. Openings are left in the copper sheet for the source. A plastic, e.g. delrin, guide ensures that foil is centered in the detector system.
An illlp~lldll~ element of the ~lcrcll~d implement~tion is the use of selected PMTs made of glass with low cont~min~tion by 40K. More specifically, we disclose use of the Electron Tubes Inc. 2" or 3" PMTs preferably selected to present less than 0.1 cps background. An element of the design is decoupling the PMT base from the PMT by means of a graded shield con~i~ting of three layers of metal with very dirrelc;llL atomic numbers. Typically, such a graded shield consists of about 0.5" of Pb, 0.15" of Sn and about 0.1" of Cu. The use of PMT base fabricated from selectt-(l materials with low r~-1in~rtive background is pl~r~lled, e.g., using pure ~1 for the supporting frame, resistors and capacitors selected for low radioactive background, and in-free solder, e.g., made of pure Sn or Sn/Pb alloy. All passive and active elements of PMT base are selected to have a very low l~lllp~lalule drift, and active colllpellsalion techniques to elimin~te temperature dependent gain drift are disclosed.
Data acquisition in MP-densitometry is based on amplifying and shaping the signals from the PMTs coupled to each detector and building a combined energy spectrum for subsequent analysis. The counts in an a~lu~liaL~ energy region of interest (ROI) for the desired isotope are then illlegldled to determine the count rate. An important part of the pl~r~ d implementation is the use of both OR and AND modes for data acquisition and analysis. The CA 0224~813 1998-08-06 WO 97/29356 PCT~US97/02224 use of a mnlfich~nnt?l DSO for on line background rejection is disclosed. The use of tri~n~ll~r shaping and software rejection of fast pulses due to signals in~ cerl by Cosmic Rays in PMT is disclosed. The particular optimi7~ti~n~ when using CaF2(Eu) scintillators is to have a pulse rise time of < 0.75 microseconds and a fall time of about 3 microseconds.
~elf-diagnostic and self-calibration for reliably nlalchillg the count rates in OR and AND
modes are illl~30~ l, including on-line baseline restoration and pileup rejection techni~lles.
In coincident mode the use of DSO to match the shape and temporal coincidence of pulses from two detector modules are preferred. There is a tradeoff between the need to estim~te the pulse coincidence to within better than 100 nsec, ~limini~h pileup and yet permit rejection of long duration pulse pe.. il l illg rejection of dark ~;UllGn~ from PMT. On-line software-based pulse fitting procedures overcome these conflicting requirements.
Conversion of the count rate (cpm) into the actual ~tt~ml~tion in the sample requires knowledge of the detection efficiency (DE) of the counter. For IlZ5 the DE can be ~leterminPd from the spectrum itself using the Eldridge formula. Preferably, one ~letermin~s the DE for each detector sG~alalGly, which allows ill~pl'~VUlg the calibration and accuracy, and testing system integrity, and enhancing DE evaluation and for diagnostic purposes.
- An important co~ullent of the non-radioactive background in MP-densitometer using EC
sources is due to dark pulses in the photom~ irliers. The shape of these pulses is different from those produced by scin~ ti~-n in the detectors, making pulse-shape based disL. i"~ ion possible. In calibration mode, we acquire pulse shape(s) for each event using a PC-based dual input plug-in DSO card and perform fast pulse shape analysis. This allows rejection of PMT dark pulses as well as other electrom~gn~fic and vibrational artifacts. After pulse-shape-based rejection, the background in the system is almost flat for energies in the 15-100 keV range and is l~l"~.k;l~ly stable, ind~elldelll of the activities in the vicinity of 2~ the detector.
The data acquisition haldWa~l~; iS mounted inside the de~ic~te~l PC controlling the MP-densitometer. The data ~cqlli~ition electronics consists of a triggering circuit, amplification/attenuation modules for each detector, 3 digital timer/counters, and a dual channel 20 MHz digital storage oscilloscope (DSO) which is used both as a 2-input pulse shapelheight analyzer. Furthermore, additional DSO or two ADC are used to monitor detector temperature.
The triggering circuit produces a rectangular trigger pulse whenever a pulse exceetling a preset threshold amplitude is registered in either detector. If pulses are registered CA 0224~813 1998-08-06 W O 97/29356 PCTrUS97/02224 ~lml-lt~n~ously in both detectors, a higher amplitude trigger pulse is produced. It is thus possible to count separately coincident and non-coincident events. The trigger pulse is sent to the external kigger input of the DSO or DSP. The triggering circuit is proprietary and disclosed in the following. The amplification/attenuation modules adjust the amplitudes of the pulses so that the region of interest is within the 0-1 Volt window of the DSO or DSP
and that particles of the same energy produce pulses of the same amplitude in both channels.
The first of the 3 timer/counters is used as a precise acquisition time timer (c~ullLulg the 2.5 kHz reference pulses). The second timer/counter counts all trigger pulses produced by the triggering circuit, while the third counts only triggers associated with coincident events.
The data acquired from these counters are used to directly evaluate losses due to acquisition system dead time and thus enable the system to correctly count high activity sources.
The DSO (for example a CSLite m~mlf~tllred by Gage Inc.) is capable of ~imlllt~nPous sampling of two input channels with 8-bit accuracy and up to 20 MHz sampling rate and has an additional extt~:rn~l trigger input. The data are stored in onboard memory and can be Lldl~ir~,~ed to the host PC RAM by ~L~ al.l memory-to-memory Ll~l~r~;r via the DSO's 8-bit access to the PC bus. The dead time is strictly non-extendable and by means of the counters described above we always correct for dead time losses. The DSO is rearmed and initi~i7ed after each acquired and processed event. The pulse traces are ~ ~r~ ,d from the DSO to the host PC memory and are analyzed for amplitude and shape by software.
Initially the DSO is set up to continuously chart the input voltages in the two channels and wait for a triggering pulse in the trigger input. When a trigger pulse is registered, the DSO
is allowed to capture a predetermined number of post-trigger points and is then stopped. The relevant portion of the traces (typically 20 pre-trigger and 108 post-trigger points at 20 MHz sampling) is L.~rt;-l~,d to the host PC~ memory for analysis. The transfer procedure takes less than 200 microseconds per trace for a 486-DX66 co~ u~l.
The analysis begins with con-yulaLion of the baseline and the pulse amplitudes in each detector. This shows whether the event occurred in detector A, detector B or both. The pulse amplitudes are adjusted for the current baselines, and if the latter are unacceptably distorted the event is rejected. Then, a number of pulse shape parameters are evaluated and compared with the ranges of acceptable values established by the software at a system setup.
These include pulse widths at 1/4, 1/2 and 3/4 of peak pulse height. Fast integer-based algoli~"lls for these computations run very efficiently on Intel processors. ~Iowever, when software implementation is too slow and leads to pileup/dead-time artifacts, one may use a WO 971293~6 PCT~US97/02224 two processor system, in which one of the processors is r1P~liC~tP~1 to data acquisition using DSO card, and the second is analyzing the data. Also, the use of fast bus with 32 bits for "flash" read-out of data from the DSO is desirable. In other implem~nt~fions, the on-line - - pulse shape analysis is pclrolllled by the DSP card which permits hal.lwalc realization of the above said algc,liLl~lls. In all these cases, the coincidence of events and additional rise-time alignment check is performed. After pulse-shape, pulse-height and coincidence/anticoinr~ nre analysis a not reiected event is added to the ~lJloplidlc spectrum (~e~;LLU111 of a detector A, sl,ec;L ull- of a detector B, or 2D spectrum of coincident events).
After an event has been processed, the DSO is reset for :lcqllieition of the next event.
Any events Oc~;uLlil~g during the processing time are lost ~dead time). When the spectral data are processed, they are ~-lj--ste~ for this dead time. The acquisition can be preset to collect data either for a given interval of time or until given st~tictiCs is acquired. The ~rqllicition can also be termin~t~1 by the user at any time.
The software program contains a simple data file browser which allows viewing and analysis of data from single and series of measurements. The user can select a data file to be viewed through a system of menus, see the Petim~t~o~l density and measurementuncertainty, plot the count rates vs. time from be~h l iug of measurement series and print out the data with stsltieti(~ uncertainties. Both the tCll~Cld~Ul~ c~ c..saL~d and raw data can be ~rc~cced. A more intensive analysis and merging of data can be pe.ro.ll.ed using a commercial spre~lch~et program.
The MP-d~,.c;loi-~ter sorL~.c may be coded in Borland Pascal and Assembly language (to speed up the pulse proceeeing) and can operate under DOS using a Windows-like GUI
shell, or under MS-Windows using the DelphiTM software development system (Borland Il~Lc~Lional), which uses the ext.on~eA Borland Pascal l~n~l~ge. The software sets the 2~ ar.ll,ixiL~on/rejection p~r~m~ter~ for each newly assembled detector system. This program determines the optimal trigger levels and pulse shape rejection parameters for the system and creates internal data files to store these parameters. These parameters generally do not have to be redt~ ed during the lifetime of the system unless a major component (e.g., a PMT/base or DSO card) is replaced.
The embodiments illustrated and ~lic~, ,c~ed in this specification are int~n~ fl only to teach those skilled in the art the best way known to the illY~llLu-~ to make and use the invention.
Nothing in this specification should be considered as limiting the scope of the present invention. Modifications and variations of the above-described embo~lim~nfc of the invention W097/29356 PCT~US97102224 are possible without departing from the invention, as appreciated by those sldlled in the art ~ in light of the above te~f hing.~. It is therefore to be understood that, within the scope of the clairns and their equivalents, the invention may be practiced otherwise than as specifically ~ ~ described.

Claims (38)

What is claimed is:

1. A method for determining the density of a sample comprising:
(a) placing a radioactive source comprising a radioisotope capable of concurrently emitting at least two photons so that emitted photons pass through the sample, (b) detecting at least two photons concurrently emitted by the radioisotope, as signals in at least two detectors, and (c) measuring the attenuation in the sample of at least one of the photons concurrently emitted by the radioisotope.

2. The method of claim 1 wherein the radioisotope is selected from the group consisting of long life positron-gamma radioemitters listed in Table 1, long life nuclear cascade radioemitters listed in Table 1, and long life electron capture radioemitters listed in Table 2.

3. The method of claim 1 wherein the total source activity is below 10 microCi and further comprising rejecting background in the photon detectors by (i) selecting low natural radioactive background elements (ii) analyzing pulse height and pulse shape; (iii) using coincidence signature; and combinations.

4. The method of claim 1, further comprising the signals from at least one of the detectors to have width of less than 2.0 microseconds, and using a fast coincidence circuit to diminish pile-up artifacts.

5. The method of claim 1, further comprising shaping the signal from at least one of the detectors to a width less than 0.5 microseconds and using a very fast coincidence circuit to diminish the pile-up artifacts.

6. The method of claim 1, further comprising analyzing the signals from the detectors on-line or off-line by DSO to establish the fraction of pile-ups, and using appropriate software correction to estimate the true count rate.

7. The method of claim 1, wherein at least one detector comprises a scintillator, an amplifier, and a photomultiplier tube, and further comprising methods to eliminate the temperature dependent drifts of count rate, selected from the use of elements with low temperature coefficients, temperature stabilization, temperature compensation, and combinations thereof.

8. The method of claim 7, further comprising placing the scintillator in a Dewar and stabilizing its temperature electronically with a precision of a few degrees Celsius

9. The method of claim 7, wherein temperature is stabilized by both the appropriate heater (preferably ohmic heater) and cooler (preferably Peltier element) and means to homogenize temperature by forced air flow.

10. The method of claim 7, wherein temperatures of both scintillator and amplifier are sensed with precision of at least 0.2 °C by temperature to voltage converter (electronic thermometer) and wherein the equivalent voltage is then measured by a voltage sensor in a central processing unit.

11. The method of claim 7, further comprising measuring the temperature of the scintillator and the amplifier in a voltage sensor in a central processing unit by (i) sending an analog voltage signal equivalent to a temperature to an ADC or DSO voltage sensor by shielded coaxial cable, (ii) transforming the analog voltage signal equivalent to a temperature in an optocoupler, sending by fiber optics, decoding by light to voltage converter, and then measuring in the voltage sensor, or (iii) transforming the analog voltage signal equivalent to a temperature in a voltage to frequency converter, sending by shielded cable, and measuring by a frequenciometer in the central processing unit.

12. The method of claim 11, wherein the temperature measuring step is performed at least 10 times per second.

13. The method of claim 7, further comprising a statistical rejection procedure whereby a preset number of temperature measurements (N) is acquired, and temperature values that deviate from running average of temperature by more than 95% (more than two sigma) are rejected and replaced by weighted average over the closest four measurements of temperature.

14. The method according to claim 7, wherein temperature measurements are used to compensate for count rate drifts due to temperature changes and to calculate the true count rate.

15. The method according to claim 14, wherein the temperature compensation is performed by means of a calibration curve which is a product of three polynomial curves taking into account temperature sensitivity of the scintillator, photomultiplier tube, and amplifier, respectively.

16. A method according to claim 7 wherein the temperature dependence of DSO is periodically measured with a system containing the means for signal multiplexing and the means to generate the pulses of well known and temperature independent shape.

17. A method according to claim 7 wherein a temperature gain calibration for each detector is obtained by a series of measurements in both OR and AND mode acquisition.

18. The method according to claim 1, further comprising steps to eliminate long term drifts of count rate, selected from (i) the use of external sources of X-ray and gamma-ray photons with known energies with at least two different lines, and mechanical means for removing or shielding the sources, (ii) the use of a separate measurement mode and calibration mode, and combinations thereof.

19. The method of claim 1, wherein at least three counters are used in measurement mode and at least two channels of DSO are used in calibration mode.

20. The method of claim 1, wherein calibration mode comprises the steps of measuring pulse shapes, calculating the energy spectrum, estimating pile-up and dead-time corrections, and measuring external electromagnetic interference.

21. The method of claim 1, further comprising comparing the data in the OR and AND
data acquisition mode, including the acquisition of at least 100,000 events with both pulse-height and pulse shape rejection enabled followed by establishing detection efficiencies by comparing the count rate rates in the detectors A and B without the use of rejection.

22. The method of claim 1, wherein the sample is a fluid comprising a liquid, gas, or mixture, and the influence of container walls is accounted for by calibrations performed when the container is empty or filled with liquid of known density.

23. The method of claim 22, wherein the sample is either water or hydrocarbons with well known density, and the calibration source has activity lower than 0.5 microCi.

24. The method of claim 22, wherein the sample is either water or hydrocarbons with well known density, and the calibration source has activity lower than 10 microCi.

25. The method of claim 22, wherein a statistical rejection procedure is implemented, wherein a preset number of coincidence counts (N) is acquired, and count rates that deviate more than two sqrt(N) from the average value N are rejected and replaced by weighted average over closest four measurements of count rate.

26. An apparatus for measuring the density of a sample, comprising:
(a) a holder for placing a radioactive source near a sample so that photons from the source pass through the sample, (b) at least two detectors capable of detecting at least two photons concurrently emitted by a radioisotope and generating a corresponding signal, at least one of the detectors being placed across the sample from the source and capable of measuring the attenuation and preferably the scatter of at least one of the photons concurrently emitted by the source, and (c) a data processor for converting the signals to density measurements.

27. The apparatus of claim 26, wherein the source comprises a radioisotope of Table 1, and the detectors are operated in coincidence mode in a symmetric sandwich configuration as in drawing 2; in a modified symmetric sandwich configuration as in drawing 3; in a compensated sandwich configuration as in drawing 4; in a triangular configuration as in drawing 8; or for thin foil or plate samples, in a compensated flat symmetric sandwich configuration as in drawing 9, in an asymmetric sandwich with separator configuration as in drawing 10, or in a modified triangular configuration as in drawing 11.

28. The apparatus of claim 26, wherein the detectors are operated in coincidence mode in an asymmetric sandwich configuration as in drawing 5; in a shifted asymmetric sandwich configuration as in drawing 6; or in a modified asymmetric sandwich configuration as in drawing 7.

29. The apparatus of claim 26, wherein the holder is a spool made of strong material with low atomic number selected from beryllium, plastic reinforced with appropriate fiber, aluminum, vanadium, titanium, and combinations.

30. The apparatus of claim 29, wherein the spool has essentially elliptic cross-section, with total cross-section surface close to the cross-sectional surface of two pipes it is joining.

31. The apparatus of claim 26, wherein the holder is placed inside a bypass made of strong material with low atomic number selected from beryllium, plastic reinforced with appropriate fiber, aluminum, vanadium, titanium and combinations.

32. The apparatus of claim 26, wherein the bypass has essentially elliptical cross-section, with total cross-section surface much smaller than the cross-section of the main pipe.

33. The apparatus of claim 26, further comprising a third, anticoincidence detector to diminish background.

34. The apparatus of claim 26, wherein at least one of the detectors comprises ascintillator composed of NaI(Tl) with a thickness of at least 2" coupled to a photomultiplier tube.

35. The apparatus of claim 26, wherein there are two detectors of substantially the same size.

36. The apparatus of claim 26, wherein at least one of the detectors has a scintillator coupled to a photomultiplier tube, the scintillator selected from BGO, BaF2, GSO(Ce), CsF, abd CeF3.

37. The apparatus of claim 26, wherein a radioisotope from Family 2 is used, at least one the detectors is a NaI(Tl), CaF2(Eu) or YAP scintillator with thickness of less than 0.5"
coupled to a photomultiplier tube, and background is diminished by measures selected from (i) using appropriate thick optical windows made of quartz or other ultrapure, optically transparent materials, (ii) placing an essentially cylindrical shield including at least one heavy metal component around at least one of the detectors, (iii) placing a separator around the source and close to one of the detectors, (iv) selecting photomultiplier tubes to have radioactive background of less than 0.1 cps, and combinations thereof.

38. The apparatus of claim 37, wherein the shield comprises a few millimeters ofPb/Sn/Cu, the Cu being placed closest to the scintillator and further comprising a shield between the photomultiplier tube and its base.