UNIVER S I TY iOF MI C HI GAN Department of Chemistry FAST GEIGER-MUELLER COUNTERS A Literature Review to May 1947 W. Wayne Meinke AEC NUCLEAR CHEMISTRY Project No. 7 Contract No. AT(11-1)-70 U. S. ATOMIC ENERGY COMMISSION

ACKNOWLEDGMENTS The author wishes to express his sincere appreciation to Dr. Luke E. Steiner, under whose direction this work was carried out, for his continued advice and encouragement; to the other members of the Chemistry Department for their helpful suggestions; and to Miss Marilynn H. Hayward for her help in the preparation of the graphs and manus cript.

TABLE OF CONTENTS Page INTRODUCTION 1 REVIEW OF THE LITERATURE 4 The Wire 5 The Cathode 7 The Filling 11 Contamination 16 Diffusion 18 Adsorption 18 Temperature Effects 19 Decomposition 20 Spurious Counting 20 Geometry of the Tube 21 BIBLIOGRAPHY 22 ii

FOREWORD The following report is an excerpt of a thesis submitted to the Faculty of Oberlin College in May 1947 in partial fulfillment of the requirements for Honors at Graduation in the Department of Chemistry. Much of the original work in the thesis is now outdated but the literature survey presented here remains a pertinent review of information published prior to 1947 on the characteristics of Geiger-Mueller counters. iii

INTRODUCTION A counter consists basically of two electrodes separated by a gas and operated at a difference of potential, An electric field, existing as a result of the potential difference, attracts or repels charged particles passing between the electrodes and charged particles formed within the field. The charges collect on the- electrodes and are discharged through an external circuit. The process of collection, discharge, and recovery of the original voltage by the wire constitutes a voltage pulse. These pulses can be counted electronically by the external circuit. The type of counter most frequently -used consists of a cylindrical metal cathode and an axial wire anode. The main function of the cylinder is to distribute the potential and to form a volume in which the electric field is defined by the geometry of the electrodes, We shall consider only counters operating in a voltage range within which the size of the pulse is independent of the event initiating the pulse, Such a constant size pulse can be formed regardless of whether an alpha particle, an electron, or a gamma ray enters the counting tube. The range of voltage producing pulses of constant size is called the Geiger region, and counters operating in this region are called Geiger counters. If we plot the counting rate of a Geiger counter as a function of the voltage we obtain a characteristic curve similar to Figure l. The literature disagrees on nomenclature,- hence the following terminology will be used in this paper (Korff21): Starting potential -- A -- The voltage which must be applied to a counter to cause it to count with the particular recording circuit which may be attached, ---

Threshold voltage for Geiger counting action -- B -- The lowest voltage at which all pulses produced in the counter by any ionizing event are of the same size, regardless of the size of the primary ionizing event. Plateau -- B - C -- The more or less horizontal portion of the curve of counting rate as a function of voltage* (On this section of the characteristic curve the pulses are all of the same size. ) Geiger-Mueller (G-M) counter -- A type of counter with a large sensitive area designed by Geiger and Mueller to operate in the Geiger region. Quenchin -- The process of terminating the discharge in a counter. Slow counter -- A Counter containing no self-quenching gas. Recovery time -- The time interval, after a count is recorded, before the pulses produced by the next ionizing event in the counter are of full size. Slow counters record only a few hundred particles a minute because of the time required. for quenching and recovery. External circuits like t'h Neher-Harper circuit were devised to accelerate the quenching. However in 1937 Trost discovered that the addition to counting gases of certain vapors like alcohol would speed up the dispersion of the charge. These self-quenching tubes with alcohol added were defined as fast counting tubes., A satisfactory counter should have a negligible number of spurinOuts or false counts under operating conditions and should deliver pulses of.warly the same size. The counter should also have a wide plateau (200 volts or more) to insure a counting rate independent of fluctuations in voltage. -2 -

Figure 1 Characteristic curve of a Geiger counter Figure 2 (Trost45 ) Dependence of the starting potential upon the cathode diameter at different wire diameters Wire diameters: 0E2 manE 0.1 mm. Filling: 90 mm. argon 10 mmA ethyl alcohol -2a

70 -E 60o E L 50 a~~.3m ClE 0. I mm 0.2mm 0.3mm c ~ 40 -C 0 Q) 0 c 30 -U 20 -1000 1200 1400 Voltage on Tube Startinq Potential (volts) Figure 1 Figure 2

In this laboratory a counter was required for internal counting of the tritium in radioactive alcohol The following properties were considered necessary: it must be a fast counter; it must have a plateau of 200 volts or more; it must operate in the Geiger region; and it must be constructed of a circular metal: cathode and a central wire anodte to allow internal countirng -3 -

REVIEW OF THE LITERATURE Several general sources of information on counter tubes give a good foundation for the understanding of the more specialized problems presented later. Bothe condenses much of the history of Geiger counters and lists may of their uses in an article commemorating the 60th birthday of Hans Geiger. May8, supplemented by Craggs,8 gives a fairly concise summary of developments in the field of counter tubes up to 1943. Trost 5 in his article "Uber Zahlrohre Mit Dampfzusatz" presents a good basis for fast counter preparation and theory. Montgomery and Montgomery give a good summary of general counter theory and experimentation. And finally Korff in his book "Electron and Nuclear Counters" gives a fine coordinating basis for the study of any type of counter. No attempt will be made here to enumerate or evaluate any of the more recent theories on the action of the self-quenching counter. Articles t42 22 by Montgomery and Montgomeryo Stever, and Korff and Present indicate that the action of the counting tube is no longer a mystery. Trost first published his discovery of the self-quenching tube in 1957. Hence we did not consider it necessary to thoroughly check literature prior to 1937. However this report does contain as complete a coverage of suitable articles on G-M counter tubes from 1937 to February 10~ 1947 as the available library facilities would permit. In Trost's45 experiments the characteristics of counter tubes were determined for variations in electrode size and material as well as gas fillings. His findings, expanded and augmented by the literature, follow. -4 -

THE WIRE Wire material. The wire in Trost's tubes was made of white steel although many workers prefer tungsten because of its lower reactivity, 30 Medicus showed that the mEaterial of the wire does not matter much since counters using brass, copper, silver, iron, steel, zinc, or aluminum wires can be used with apparently little difference in their effect upon the count12 ing. Davis and Curtiss2 sed untreated steel and tungsten wire with no difference in performance traceable to material or construction. Milatz and Kate52 used platinum wire in their counters. Locher26 found that the wire must be very free from conducting points, including dust which may be precipitated on it while the. tube is in use. These particles lead to spurious counts and destroy the precision of the counting. Most workers use tungsten wire because it can be easily sealed to glass and can be heated while the tube is being evacuated. The tungsten can be cleaned in a bath of molten sodium nitrite (340~C.) and then washed in distilled water. (Collie and Roaf5) Any sharp points of metal, dust particles, or grease remaining, burn off when the metal is glowed for several seconds. The process involves attaching the ends of the wire to a Variac and slowly raising the voltage. (Korff21) Glowing also forces adsorbed gases out of the wire. Thus if the tube is kept evacuated during the process, practically all foreign gases will be eliminated. Adsorbed gases on the wire appear to be particularly troublesome; several writers mention them as one of the major reasons why many G-M tubes will not operate quantitatively. Milatz and. Kate3 indicate that for best performance the wire (platinum in their work) must be well stretched.. They placed a metal spring between the leads and the wire, thus kIeeping the wire co-axial -5 -

with the cathode The geometry of the tube will be discussed more fully later. Wire sizeo Trost presented data demonstrating the change of starting potential with size of wire (Figure 2). Moon5 agrees with Trost and states that in general the working voltage increases with the thickness of the wire but that the diameter of the wire is not very critical. 14 However, Haines14 showed that the situation is reversed at low pressures (around 1 cm total gas pressure in the tube), the starting potentials diminishing with an increase in wire size, but since 10 cm. is the normal gas pressure in counting tubes,, his results will not affect ordinary work. Table I lists the diameters of wires used. by various workers in their counter tubes. Table I Wires Used by Various Workers Wire Wire Diam. Worker Date Material (in mm. ) Lawrence25 1941 Tungsten 0.02 Curran and Petrzilka1O 1939 Tungsten 0.05 Shonka40 199 Tungsten 0.075 Hoagi7 1938 Tungsten 0.075 34 Montgomery and Montgomery 1941 'Tungsten 0 075 Rochester and Janossy8 1943 Tungsten 0,075 Strong45 1943 Tungsten 0.1-0.2 Strong4 1943 Copper 0.1-0.2 Collie and Roaf5 1940 Tungsten 0.2 Milatz and Kate52 1940 Platinum 0.2 Halpern and Simpson15 1937 Tungsten 0.25 McLellan 1945 Tungsten 0.25

THE CATHODE Trost and others agree that neither the material nor the treatment of the anode affects the plateau of a counter. The cathode, however, has a much more criticalinfluence o tube characteristics., Cathode material Trost found that the size of the plateau and its slope depend essentially on the material and treatment of the cathode, Figure 3 shows characteristic curves for various counters with untreated cathodes. The preparation of these tubes included polishing with emery and washing with alcohol and water before assembly. A period of perhaps two hours elapsed between the polising and the sealing of the tube onto the vacuum line. It can be seen that with an alcohol-argon mixture, usable plateaus (slope under 10% per '100 "v.olts) were obtained with brass, gold., and chromium,, Iron, silver and. copper were essentially poorer, aluminum yet poorer, and 1astly nickel and cadmium completely unusable. Trost attributed the diff erence in characteristic curves to the various kinds of surfaces the metals present. To make the cathode surfaces uniform he coated them with a mixture of Zaponlack (a kind of cellulose nitrate varnish) made by diluting a few drops of the varnish with 50 ml. of amyl alcohol. Figure 4 shows the characteristic curves of some of these treated, cylinders. It can be seen from Trost s work that many metals can become suitable Cathode -material if given the proper trfeatment. Trost's work did establish the fact that among the common metals brass at least could be -used in satisfactory counter tubes without additional treatment. Several other workers however have: disregarded his results. and recommrend methods involving the careful cleaning of the counters with acid, frequent ri.naings with water,: and baking in the presence of gases rich in oxygen, -7 -

Fig u 3 (Trost45) Characteristic curves for different metal cathodes (Tube- dimsnions 30; 0,2 indicate a cathode diameter of 30 mra and a wire diameter of Filling,: 90 0mi. argon 10 T.im ethyl alcohol (Trost-45) Characteristic curves for ash1Elacked metalt cathodes Tube dimen'io-ns: Cathode diameter - indicated in mm, Wir diameter - constant Filling: 90 amm. arxon 10 - ethyl al.o.hol -7a

Cd Ni Al / Cu * I I / 1/ *''r 01. I, -;.... I I A F ig b i i 19;a20; 0.2 - - CII~~ j/~~~ 1~30; 0.2. j I40; 2 1100 1200 1 o I facqe on Tube 3oo 500 Figure 3 400.43~~~~~~~~~~- ~~ A32 -Figur C 0r N3

Cathode treatment. For making fast counters, Strong43 recommends a procedure in which a copper cathode is cleaned thoroughly with 6N and then 0.1N nitric acid after which it is rinsed and dried. The tube is next heated with dry air inside, evacuated, NO2 admitted, and heated again until a coating of Cu20 forms. A mixture of argon and xylene is used as a filling. Hoagl7 eliminates the N02 heating process but requires that a first oxide coating be washed off with nitric acid before the final sensitive coating 4o be made. Shonka, using the methods of Hoag, obtained tubes with plateaus 27 of over 1000 volts. Locher on the basis of experience with the production of more than 1500 tubes advocates the use of specially treated cathode surfaces. Collie and Roaf5 recommend treated copper cathodes for counting 42 tubes. Stever states that a treatment of the cathode surface improves fast counter action. The N02 treatment, by producing a velvety black or dark brown coating of oxide on the surface of the copper cylinder, results in a low photoelectric emission and a high work function, and hence in 6 better tube characteristics. Copp and Greenberg indicate that the best results in their experiments were obtained with copper tubes which had received the treatment with N02 (very similar to that described in Strong). 29 McLellan 9 uses a copper foil cathode treated by the N02 method. Hans Weltin48 on the other hand eliminates the cleaning process of the copper cathode by coating the cathode with Aquadag, applied with a small brush. Curtiss1l prepared a sensitive counter tube by coating the metal (copper, aluminum, or brass) with a lacquer made from bakelite and amyl alcohol. This lacquer was left in a soft state and apparently helped in the quenching process. -8 -

Untreated dathodes. Other workers however, while admitting that these elaborate procedures result in reliable counters, maintain that from their own experience most of the chemical treatment is unnecessary. Davis 12 and Curtiss emphasize that in their work on counter tubes no special treatment of the barrel or central wire was required to make a successful counter. All their counters were made up with materials as taken from the stockroom. Copper, aluminum, and steel were used for the barrels with no noticeable difference in performance traceable to the materials or construction. Curran and Petrzilka confirmed Trost's conclusion that polished brass which had been allowed to stand in air for some time before assembly was a good material for counter walls. They also found. that if copper was heated gently in air so that a very thin layer of oxide is present on the surface, it was quite as good as brass. They avoided the use of Zaponlack in improving the working of tubes because of its solvent effect upon picein. However, they did find a method of oxidizing aluminum sufficiently so that it would work consistently and well in counters for periods of months or.e-ven years." Figure. 5 shows characteristic curves for tubes prepared by their methods. 38 After preparation of nearly 100 counters Rochester and Janossy8 state that careful treatment of the counter sheath is not necessary for an efficient counter. They used a copper-in-glass type of counter with the cathode sheath 0.1 mm. thick of light copper, cleaned by rubbing over with a rag soaked in benzene. The counters, varying in di'ameter from 3.0 to 3.5 cm. and in length from 20 to 80 cm., were filled immediately as they came from the glass blower with a mixture of argon and alcohol and then sealed -9 -

Figure 5 (Curran and Petrzilka10) Characteristic curves for different metal cathodes Thick walled Thin walled --- Copper 0 Aluminum ) Brass Figure 6 (Trost45) Characteristic curves for shellacked gold cathodes of varying diameters. Wire diameter constant. Filling: 90 mm. argon 10 mm. ethyl alcohol -9a

iI -;,,,-/ --- - - 6.-,6 500. 7 5m 1500 I I~~~~~~~~~~~~~~~~~~~~~ 400.40m0 I I I I CD JII~~~~~E ~ '300 II "- I I I..4 30mm C I I d c~~~ L. II II o c~. L.) _ ~,/2~~~~~~~~~~~~~~~00 CL 20mm II P CI II00 o 2 U,/ 5.I-" -- -- O - - -O —J)- - ~ I~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - 50' I O0 12 1100 2 100 100 400 1600 I 0OO II~~~~~~~~~~ ~~~~Voltage on Tube [4/~~~~ ~Figure 6, I I II~~~~~~~~~~,l I II I 1000 1100 1200 1300 1400 1500 Voiltage on Tube Figure 5

off, the whole operation for each counter taking only one quarter of an hour. The counters had the following properties: efficiency of 99.5%, starting potential of 1000 volts, and length of plateau of 300 volts. Counters prepared in this way were in use for two years without appreciable changes in their properties, Lawrence25 in preparing counters for the determination of H3 gas used a tathode of brass (5 cm. in diameter) that had been carefully cleaned with nitr-ic acid and rinsed copiously with distilled water. Brown and Curtiss3 describe a method for making thin-walled aluminum counters from commercially available tooth-paste tubes thus eliminating the expense of boring metal tubes to obtain a suitable thin cathode. 35 Moon states that for counting beta rays the cathode may be made of thin metal sheet, thin glass, or if very thin walls are desired, of foil wrapped around a perforated metal cylinder. He also reports that the outer electrode has relatively little influence on the plateau. In fact it need not even be a perfect cylinder, although cylindrical symmetry is necessary for a uniform field at the wire. Thus minor imperfections or scratches on the cathode will not give rise to spurious counts as they would on the wire. 26 Locher in corroborating this fact states that the smoothness of the cylindrical cathode is of little importance, Cathode size. Variation of the diameter of the cathode will affect the characdteristic curves as shown in Figure 6 (Trost). Increasing the diameter raises the starting potential and, by increasing the sensitive volume of the counter, raises the rate of counting. However, Lawrence25 states that some counter diameters are limited by the fact that in counters of large diameter, H disintegrations near the wall are not counted efficiently.

Variation of the length of the counter itself is useful at times. 43 Strong states that the length should always be at least 5 times the diameter of the tube. Again with increase in length, the sensitive volume is increased and consequently the counting rate increased. In the final analysis counter length and diameter are determined by the nature of the problem to be studied (Korff21). THE FILLING A third major variable in counting tubes is the filling. Many gases and gas mixtures have been suggested and tried by various workers but we will consider here only the "fast" fillings. Montgomery and Montgomery34 found that the gas filling determines: 1) the starting potential of the counter,. 2) the efficiency of the counter, and 3) the magnitude of time lags between the passage of the ionizing ray and an appreciable change in potential of the counter wire. Previous to 1937, counters had such appreciable time lags (as in 3 above) that several hundred counts a minute was the limit of recording. Since the counters were used mostly as instruments in laboratories, very little was known of their actual mechanism of reaction. Furthermore, their properties were not reproducible, Trost however found experimentally that some counter tubes, though counting very pporly when first assembled and sealed off, would count well after a period of several days. His construction methods included polishing the brass tubes with emery, washing them with alcohol, and drying them before assembly. He reasoned that the added activity after several days must be due to gases adsorbed on the walls of the cathode. Further experiments in which known amounts of alcohol were added to the original filling of argon or air produced a -11 -

satisfactory counter that had the characteristic of being able to quench itself with no external help (as from a Neher-Harper quenching circuit). This was the first fast counter tube. The principal gas. The gas filling of a fast counter tube consists of the principal counting gas and a small volume of quenching gas. Curran and PetrzilkalO determined characteristic curves for various counting gases with all other variables constant (including a constant volume addition of alcohol to serve as a quencher). From these curves (Figure 7) it can be seen that argon gives the best counting curve with neon and helium not far behind. We can see that the inert gases behave similarly in the counting tube. Hydrogen, oxygen, and nitrogen on the other hand are poor counting gases. Hence one would expect that air would also be a poor filler for fast counters, a fact readily shown when a small amount of air is allowed to leak into a tube which is counting. The similarity of the inert gases is particularly beneficial since argon is not suitable for all types of work. Argon becomes radioactive under neutron bombardment but a helium filling may be used for neutron counters. Helium does not become radioactive and, in spite of a gradual rise of the plateau in its characteristc curve, is a good 19 counting gas. In 1941 Kapur, Sarna, and Charanjit presented the results of some of their work done with helium filled counters and more 29 recently McLellan described the construction of a helium filled tube with a plateau of from 200 to 300 volts. For the majority of general counting measurements however, it will be found that argon is the most suitable filling. -12 -

Figure 7 (Curran and PetrzilkalO) Characteristic curves for different counting gases Cathode: Thin aluminum tube Total pressure: Constant Alcohol addition: Constant Principal counting gas: 1. Argon 2. Helium 3. Neon 4,, Air 5. EHydrogen 6. Oxygen 7. Nitrogen -12a

800 600 C 0 200 1100 1200 1300 1400 1500 1600 Volftage on Tube Figure 7 12b

The quenching gas. The ideal quenching gas would have a high vapor pressure and would not be readily adsorbed on the walls of the counting tube. However very few gases will satisfy both of these requirements at room temperature and consequently any gas used will of necessity be a compromise. The work of Trost in 1935 indicated that benzene is useless as a quencher and that no definite plateau is obtained with chloroform. He found that alcohol worked well but had a low vapor pressure and was easily adsorbed. In 1937 he reported that in principle, most vapors act the same as alcohol in trying to quench the tube charge. However in practice only a few vapors are useable, partly because the resulting plateau is poor and partly because too few impulses are counted. He tested many gases for ideal quenching properties and finally tried, besides ethyl alcohol, methylal and the two esters, methyl formate and methyl acetate. He also tried methyl chloride, ethyl chloride, acetone, and acetaldehyde, all of which had poorer counting regions than alcohol. He found that tubes filled with the esters would become almost unusable after a month, the slope of the curve increasing with timeo After a month not even thorough washing out and refilling of the tube would restore its original characteristics. However he did find that if a tube was filled with methyl acetate for a day, pumped out, and then refilled with ethyl alcohol, a much longer plateau (500 volts) would be obtained than normally would be expected with an alcohol mixture alone. (A brass tube was used in this experiment.) Trost concluded that among the gases he had tried, alcohol was the most suitable for quenching. His work also contained illustrations of the characteristic curves of halogen compounds of hydrocarbons, showing that chemically related substances have a corresponding regularity in their counting curves. -13 -

Following Trost's lead most workers use a misture of argon and 12 alcohol as a filling. However Davis and Curtiss report that they have found amyl acetate a more reliable filling than alcohol, possibly because its heavier molecule has a higher quenching action. Amyl acetate does seem 11 to have definite quenching properties since Curtiss, while using it as a solvent for a Bakelite lacquer which he applied to the wall of the tube, found that if the lacquer remained in a moist sticky state, the tube would show good counting properties with a filling of alcohol and argon, but that if the lacquer were allowed to dry, the tube would be ruined for satisfactory service. Korff, Spatz, and Hilberry23 found that methane and boron trifluoride are suitable quenchers, having no appreciable temperature coefficients between 55.~C. and minus 220C (Figure 1)o. It should be noted however that the methane quenched. tube requires a much higher starting potential then the alcohol quenched tube. Kapur and co-workers19 report that methyl and. ethyl alcohols are the best quenchers for helium filled tubes. McLellan29 confirms this and states that Oo5 cm. of alcohol produces good quenching with 7 cmo total pressure of helium and alcohol mixture. 34 Montgomery and Montgomery suggest that if an alcohol filling is used with a treated cathode the alcohol may react with copper oxide present-. 20 Organometallic fillings. Keston presented an interesting discussion on self-quenching G-M counters containing organometallic compounds. He found that good tubes could be obtained with PbMe4 pressures from 0o8 to 2.5 cm.,, the tubes at 18 to?0O cm. pressure giving plateaus of 500 volts starting at 1500 volts. Unlike the argon-alcohol counters that show a very steep rise from threshold to plateau, these counters require a much

larger change in voltage between threshold and plateau. He found that less satisfactory counters could be made with dimethyl mercury while iron carbonyl gives a self-quenching counter that is decidedly unstable, decomposing overnight. Filling pressure. Trost showed the dependability of the plateau upon both the total pressure of the filling and the partial pressure of the quenching gas. (Figure 8). The absolute maximum of the group of curves lies around 100 mm. total pressure with 10 mm. of alcohol added. Further investigation showed that the absolute maximum comes at higher additions with tubes of smaller diameters. With rich mixtures however, on the one hand come unwanted addition and temperature effects, and on the other hand a sharp increase in voltage. Thus Trost does not advise the use of a mixture containing more than 10% alcohol. For tubes with very much smaller diameters (under 20 mm.) a somewhat higher argon pressure seems advisable. Rochester and Janossy38 report that the efficiency of a fast G-M counter is not changed when the pressure of the argon is increased from 11 cm. to 74.5 cm. 47 Paul Weisz7 in experiments using an argon-ethyl ether filling found that apparently the absolute pressure or density of the quenching gas alone determines the quenching characteristics of the tube, regardless 59 of what the total pressure may be. Rochester and McCusker corroborate these conclusions. They find that the alcohol counter can be used only in a narrow range of alcohol pressures which is determined by the vapor pressure of the alcohol, The useful pressure range is not much affected by the argon which merely acts as.a filler to increase the efficienty of the tube, Figures 9 and 10 show the effect of alcohol and argon pressures on plateau length and starting potential. -15 -

Figure 8 (Trost45) Plateau length as a function of total gas pressure at different partial alcohol pressures Tube dimens ions: Cathode diameter - 50 mm* Wire diameter - 0.2 mm. Cathode s)rface: Shellacked -15a

'~1~ ~~~1'1 ~~~~~~ ~~~~~ ~~~~,.tr ~~~~~'~~,~~~.~.~..~. i~~~~~~t 1I i tf t I + —. - -~.~~~.... '~~~. —.~~~ ~1......- * 0 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~J t I t tf:::::::: ''' t' t::: -::: —: - t:::X:::::1:::::::::: Il l: I t:::::l: ~::;s:: 1.I;;;.I;;;; I"'. i..L....../~.! f.r... 1 - - 1:~",~~,~~~ --:~~~ "" ".~, --,- ~~~~I~~~, I:1 /....11..... 1 I:XX < H H W X34 <1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...'1~~ g...... |.... t..... g... _ '.:!' * 'I ' ' ' I ' ' ' '' ' I' ' ''.I.'..................... '`'~~~~ ~"' "'r~~~~~ '~'~ "~~ '~'~~~~~~~~~~~ ~ ~~~~ ~~~I.~.~.,ir r II....... ~'<~(~,~~,~1 ~ ~<ri~~~~~~~ ~~~.~...<..S..T -e..........1.. ~~~~~~~~~~ ~~~~~~~. ~.....~~~..... ~~~................ ~~~~~~~~~~~~~~~~~~~~~~~~~........... I:100-i 1401 1- 1 —1-X - — I ---l — T 1...1.....1 1 1 1 1 1 11................:1 5b......; t ~ ~1 ~~..................... r~ ~~~ ~~~ ~..... ""l ~ ~s'~'( ~ ~ ~I~~~1 ~~~~ ~~.........,....( "' "" "" "'~ ""~'-.. ~.~~ ~.~~.~~~...~~~~~~~~~~~~ r............ "" "" "" '~" ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~,~,~,~,...~...,,.,,.......... '~~~~ ~~~' '~'~ ~~" ~~~ ~". '~'~ ~'.~).~..~~~~~~ ~1, ~ ~~~~~~1~~1.our....... ~~- ~~~~ ~............... ~'~' '.~, ~~,. ~.,.,,~ (I,~ ~,II IIII II.,, 00I............. ~~"' ~~~" "~' "~' "'..'.' ~." "~' ".,.I~I~~~~~~~~..~1 1 t I1I( I,.................. "" "'' "' "".....~~~.~~, ~~~~~~~~~ "...... 1,................. "~~'1 —' ~~~~~~~~~~~-~~~~-~1~~.. )I(; t 111( (;( i s......... ' '' '"'..........:..................... I I~, 1,, ~~~~1~- ~~l~~~~r~~~~~~~~~~~~~~~~~~~~~~~~~~~~.~....................... ( t 1 '( i ' ' '~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..............

Figure 9 (Rochester and. McCusker39) Effect of alcohol and argon pressure on plateau length Figure 10 (Rochester and McCusker39) Effect of alcohol and argon pressure on starting potential

..... '....' '-.....;'!!!! -!;!!::!.-.. _ -.;;.! —. '-'. _ ____...'....T::.-...... ""~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~....................,. m -.X.o _. + ' 1',,. _....;-|1......................................... L..::: t: _........................... ' ".-. —.!! _!:::: '... _..... _..',......... "'.............. "'..::.'. - ' —:'::::*:!!it,,,,,,,,,,......... -........................... _:..::::...:.-..... ' t I..,' 1" '.... I.....,.... '..::::: '! 1 I ~. I ~ '.." '::.... '::.::::: 1-. t..,. - 1 0 4 ---.., j_ F - T............................... '.... ~ ~ ~ ~ ~ ~ ~ ~ ~_ ~...........:.::.........., -.......;.! ' I I ' il,: ' I, I.I i'..... i. ' ".1 - " "... '.... 1. ":.I I '... '..............'-, -: '. -:: '.......... ': ':::::::*::;;.... I...... ---........-.....:1., o= -- - 1 --- -1- ---- - - - - --- -1- - - --- t- > - - --— 1 ---- --. I,1.. —. ~.1~- ~~ ~ ~: -, ~._t.-:.......,~.;~~~.... _........... ~ ~..:. --- —- - --...... -t —: ---i-.. --- —-. —...-+ --- —— t --- — ----... t ---$ ---l-....... 1.... -i.... ':':'1:: —1'1:: -'" L ' ' r | l: t '1 '' -1' ' - I I I I I I ' ' I' 1- '1 7-~" ~~~~~~~~~~~~~~~~~~~~~~~~~......... ~~~~~~~~~~~~~~~~~~~, i..............:-:::::::::l::::::....... I.. ' '..:..: I:..:;...::.1::..:.::1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:m..:......" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~!'~ -.:!~~~~ ~ ~~~~~~~~~~~~~~~ i i. ~, $-.1l.;...;-.-l —.1: -...,, '.....!T....:1-:i;1: ~ ~ ~ ~ ~ ~ ~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~........!"-: -..t:~....iiii...: ~.... ~.1 1::}....... -~.1.......- I........ ~. -.. —..:.. 1 I i ': i~:!-:: -':: i:,':- 1- I *:i ~~~~~~~~~~~~~~~~~~~~......, ~.~............'............... —.-l. --- -I-.-. —... -. — --- - --- - -': '::. I.I.:;.:i:::'-:.i:-:-t:-!!:: i'.!:::.:,:.i:.: "::,:'"~......:~ '~ ~ l ~;,'- ~ ':............. ' ~ ' ~'...... ~... -::::....~....:t:'''.....t':! ' ';..... —.-..... ' -:...' '~.......,......"'i'':'::'i:..:::": -':1:' '!..' i ~" ':'! '.::'.'.~':':...:.:...:. 7:I:::i7 iit:'.:-:. ':. i-T,; -:.... _L__~ ~......... '........:..... ':: I'-: "'"" i:::::~':':::::-,-:-:............-.... 77-, ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ '':I......t.... ~i",.! -: '/.;.-. —.-' _ ':i'...'.':':'::!.....:,~: _: ~~ ~ ~~~~~~~ --- —!:::!....i:i': i:~::,, -' -,:-c- -—:~.r-':?~ — ~' —k........ f-' ---'t-:'- '' —'- '' —':~~:i.:ii~'!~t:~ir~............:':-'-':....:"~'....; ~J.::::'~'.~ '..:....~::: 't:'..... t....:.:~:::,-4 ---.:::.~9.,, ~:., __~-!........-.:L..... ~ '.... i...t ~ ~ i,;,,,..::::..,,~.~ ~.- ~.... r: ~,.... '............ ' % '~....... ~-..-.::......... ~............... -,............! ' I I I i.; i I,,,. ' I. '. ''-;'!.... ~ ' ~..... 1....~......!!!::.... ~.......i; —, -4... ~..i... '... ---. ~~ "i'''~':~i..: I 1:::.::::: [. '-':..!!! --. I.' ' I ' -. — i: 7.:::'::.:..,,,, ' ' ', ~ ~' ' ':I:::-::'~~ '_i':!';:::;;;::::?....! —!.... - - ~ ---~ ' " '-~...i...... — 'F!:,:~~:~~~~::~~::::::::::::::::::::::::I ~:~.. i - ~,: --- --- i ' i:~.,": —'"-+ —~-~ ----. -.':: --- — '.-_,'. ' ' ' i:' t.:~!O i'. < ~~~~~~~~~~~~~~~~~~;:::::::::::::::,....:":::::::.'.:l ~;:: i " i ' ~i~ '~, ' ---' —~-' ' '.~~1~~~''~:',::::::::::::::'. "i ' I, ~: '. I 1 " ~:; I '.: ':" ":' ':::::::::;...:;~..... ~ ~~:... { 1,...,... ''

Technique of filling~~ In filling tubes, Collie and Roaf5 recommend the use of a cold trap (salt-ice mixture a little below 0~C. ) through which the alcohol vapor is admitted. This trap eliminates any water as well as benzene which might be present, both of these substances affecting adversely the characteristics of the tube. Montgomery and Montgomery34 also advise the use of a cold trap to remove harmful substances from the alcohol vapor, CONTAMINATION Various contaminants may still be present in the filling however. 41 Spatz treated the subject of argon impurity and showed (Table II) the dependence of plateau slope and starting potential on argon purity and alcohol content, Table II Influence of Argon Purity on Tube Characteristics Starting pot. Plateau slope %1ale. vapor A purity volts % / volt 5 99.8 840 0.01 10 99.8 960 0.02 10 98.0 1o00 0.15 10 90.0 1125 0.25 20 99.8 1080 0. 05 20 90.0 1220 0.35 From this table we see that impurities such as air or oxygen increase both the plateau slope and the starting potential of argon-alcohol mixtures Since air is soluble to 0.7% for a 20% alcohol mixture, it causes an increase of plateau slope of 0.05% per volt. This increase can be eliminated by degassing the alcohol. -16 -

Curran and Petrzilka10 show how the characteristic curves vary with different partial pressure (O to 5 cm.) of air added to a counting mixture. Spatz also found an increase of plateau slope with an increase in alcohol content of the tube. 31 Medicus reported that the presence of traces of water vapor in the counter encouraged spurious discharges, not appearing in a carefully outgassed tube. On the other hand many authors have found that counters with very carefully outgassed electrodes and purified gases fail to count at all. To produce workable counters a trace of some impurity such as oxygen is allowed to be present, evidently forming a surface layer on the cathode. This surface layer is apparently essential to the working of the counter. Many authors seem to be able to reach the right degree of impurity by using commerical gases without extra purification (May)28. At one time mercury, a contaminant present in most every high vacuum system unless trapped out, had been thought to cause trouble in the counting mixture. Medicus31 however claimed that the presence of mercury reduced the number of spurious counts. More recently Korff and Present made a study of the effects of mercury vapor on the counting>i mixture. They found that because of the extremely low vapor pressure of mercury at room temperature, the mercury contaminates a counter to about one part in 105 without creating any noticeable effect. Higher mercury concentrations will change the characteristics but fortunately these concentrations can not be effected at room temperature. -17 -

DIFFUSION Diffusion seems to cause considerable trouble during the process of filling the counter tubes. Spatz41 states that his counters required 24 hours for complete mixing and diffusion of the gas and vapor. He uses a mixture of 5% alcohol and 95% argon. Korff21 indicates that at least an hour should be allowed for.diffusion equilibrium to be established after 46 the mixture is made. Weisz in a review states that proper mixing of gases is necessary. Davis and Curtiss12 indicate that the difficulties experienced with their counters arise almost exclusively from failure to secure a mixture of gas and vapor of the proper proportions, brought about because of the slow diffusion rate of the gases. However other authors whose results indicate that their counters are accurate and reproducible have apparently not allowed much time for 58 diffusion. Rochester and Janossy filled and sealed their tubes within one quarter of an hour. No mention is made in the papers of Trost45, 10 29 Curran and Rtrzilka, or McLellan29 of any definite time allotted for diffusion. ADSORPTION Adsorption seems to be another one of the big problems of counter construction. Many authors have had trouble with alcohol adsorption on the waxes, resins, and plastics which they use to seal their tubes. Curran and PetrzilkalO found that ebonite adsorbed alcohol vapor and shortened the life of their tubes. Trost44 indicates that adsorption is one of the main disadvantages of alcohol filling, for within a month, alcohol vapor in his tubes appeared to have been depositied on the insulators and metal parts in the form -18 -

of a hard rubbery substance. However these tubes could be regenerated by simple warming. Others found that picein readily absorbs the alcohol as do hard rubber insulators and stoppers (Curran and Petrzilka1O)o The adsorption is dependent upon the temperature and thus gives the tube an undesirable temperature correction. Strongly alkaline glass tends to absorb vapors and should be avoided. Adsorbed gases in the tube can be eliminated from the wire by glowing and from the cathode and tube walls by baking. Thus tubes completely sealed in a pyrex envelope and thoroughly outgassed would have a minimum adsorptive tendency. ITEMPERATURE EFFECTS Several authors report temperature effects on various types of fast counting tubes. Trost observed a rise of counting voltage of 2 volts per degree temperature lowering. He explained this rise by the tendency of the vapor to precipitate out on the wall of the tube even above its saturation temperature. Cowie7 noted an increase of counting rate with rise in temperature and attributed the change to thermionic emission from the walls of the counter. He also found that a slow counter at 100~Co started to count 100 volts lower than at room temperature. Korff, Spatz., and Hilberry23 made a study of this temperature coefficient and explained it in terms of the Montgomery and Montgomery theory of counter action. As the temperature is lowered, too few quenching molecules remain to efficiently quench the discharge. Thus Korff and co-workers tried filling tubes with gases such as methane and boron trifluoride that are not saturated vapors at ordinarily encountered temperatures. Figure 11 indicates the difference in characteristic curves of the methane and alcohol quenched tubes at 55, 26, 0, and minus 22~C. We see then, that -19 -

Figure 11 (Korff, Spatz, and Hilberry23) Characteristic curves for counters with argon-alcohol and pethane fillings at various temperatures. Lower voltage scale (higher voltages) applies to methane counter only. Argon-alcohol counters show temperature effects below but not above room temperature; methane counter has negligible temperature coefficient in this range. Temperature s -220 Co ------. 0 C, ------ 26~ Co --- --- 55~ C ----— o.1'9a —

q6T TT emnTI aqnjl uo au6. lOA oo00 0061 00 o1 0OOl 00021 0001 00n1 ooZI 0001 o o 0/~I i ##0080 +HD~ ~e~ c~-V 0001 009 // ooe 3 -- 0001 0091 0081

using methane as a quencher has eliminated the bad temperature effect of alcohol quenched tubes. Following this explanation, McLellan29 prepared tubes with 5 mm. pressure of alcohol vapor added. At this pressure, the vapor condenses only below minus 150 C. These tubes proved workable down to at least 6~ C., the starting potential of the tubes remaining constant for varying temperatures o DECOMPOSITION While investigating factors that influence the plateau characteristics of self-quenching G-M counters, Spatz41 found that the alcohol in a fast counter decomposes constantly with use. Korff and Present22 state that about 1010 alcohol molecules are decomposed at the cathode for every discharge. Since there are about 1020 alcohol molecules in the counter volume, the counter will go bad after about 10 counts. The primary decomposition products are usually free radicals which then combine to form organic compounds, some with and some without quenching properties. With continuous use the larger vapor molecules break up into non-quenching gases 41 such as oxygen and hydrogen, or hydrocarbon polymers. Spatz lists the decomposition products of alcohol vapor in a discharge (condensed from data by Cummings and Bleakney9)b Yaddanapalli49 presents the same type of data for methane. SPURIOUS COUNTING Several authors have investigated spurious counting and inconsistencies of tubes. Nunn May lists certain precautions to be followed in the elimination of these spurious counts. Hull18 mentions a warming up -20 -

correction which seems to be an inherent property of the tube itself. 34 Montgomery and Montgomery discuss causes of spurious counting and list several tests for spurious counts. Greisen and Nereson13 discussed the efficiencies of alcohol filled counters used in coincidence work and cosmic 38 ray counting. Rochester and Janossy compare efficiencies for different types of fillings of copper and brass in glass counters. Their data are taken from six observers with different methods of preparing the tubes. GEOMETRY OF THE TUBE The geometry of the tube affects the electric field around the wire considerably. May28 states that any slight asymmetry of the electrodes leads to variation in the electric field in different parts of the counter. Thus the critical value at which counting begins is not attained at the 35 same voltage over the entire tube. Moon reports that the cathode does not even have to be a perfect cylinder but that cylindrical symmetry is necessary for a uniform field at the wire. Christophl4 investigated the effect of deliberately placing the wire at a variable distance from the axis of the outer cylinder. He found that even a small displacement had a marked effect on the plateau of the characteristic cirve. Thus the good counting tube requires a symmetrical electric field. As previously stated however, any considerations of size are entirely dependent upon the nature of the problem to be studied.

BIBLIOGRAPHY 1. Black, J. F., and J. Chiem. Phys., 11, 395, (1943) Taylor, H. S.,7 2. Bothe, W., Naturwissensehaften, 30, 593, (1942) 3. Brown, Burrell W. and J. Research Natl. Bur. Stds., Curtiss, L. F., 35, 147, (1945) 4. Christoph, Ann. Phys, Lpz., 266, 145, (1936) 5. Collie, C. H. and Proc. Phy. Soc. London, 52, 186, (1940) Roaf, D., 6. Copp, D. H. and Rev. Sci. Inst. 14, 205, (1943) Greenberg, David M., 7. Cowie, D. B., Phys. Rev., 50, 385, (1936) 8. Craggs, J. D. Repts. Progress Physics, 9, 137, (1942 - 4) - (Publ. by Phys. Soc. of London) 9. Cummings, C. S. and Phys. Rev., 58, 787, (1940) Bleakney, W., 10. Curran, S. C. and Proc. Camb. Phil. Soc., 35, 309, (1939) Petrzilka, V., 11. Curtiss, Leon F., J. Research Natl. Bur. Stds., 25, 369, (194o) 12. Davis, Francis J. and J. Research Natl. Bur. Stds., 29, Curtiss, Leon F., 405, (1942) 13. Greisen, H. and Phys. Rev., 62, 326, (1942) Nereson, H., 14. Haines, C. L. Rev. Sci. Inst., 7, 411, (1936) 15. Halpern, J. and Rev. Sci. Inst., 8, 172, (1937) Simpson,. 0. C.., 16. Heldman, Julius D., Techniques of Glass Manipulation in Scientific Research, Prentiss Hall, 1946 -22 -

17. Hoag, J. B., Electron and Nuclear Physics, p. 431, D. Van Nostrand, 1938 18. Hull, D. E., Rev. Sci. Inst., 11, 404, (1940) 19. Kapur, P. L., Current Sci., 10, 521, (1941) Sarna, H. R. and zCharanjit 20. Keston, Albert S., Rev. Sci, Inst., 14, 293, (1943) 21E Korff, Serge A., Electron and Nuclear Counters, D. Van Nostrand, (1946) 22. Korff, S. A. and Phys. Rev., 65, 274, (1944) Present, R. D., 23. Korff, S A,, Rev. Sci. Inst., 13, 127, (1942) Spatz, W. and Hilberry, N., 24. Lange, Norbert A., Handbook of Chemistry, p. 226, Handbook Publishers, 1946 25,. Lawrence, E. O. National Defense Research Committee of OSRD, Serial %66, August 26, 1941 26. Locher, G. La, Phys. Rev., 47, 326, (1935) 27. Locher, G. L., Phys. Rev., 55, 675, (1941) 28. May, A. Nunn, Reports on Progress in Physics (Publ. by Phys. Soc. of London) 5, 39 (1 938 - 39) 29. Mellan, A. G., New Zealand J. Sci. Tech., 27b, 263, (1945) 30. Medicus, G., Zeit. f. Phys., 74, 350, (1932) 31. Medicus, G., Zeit. f. Phys., 103_, 76, (1936) 32. Milatz, J. M. W. and Physica, 7, 779, (1940) Kate, H. Ten, 33. Miller, Warren Wy Science, 105, 123, (1947) 34. Montgomery, C. G. and J. Franklin Inst., 231, 447, 509, (1941) Montgomery, D. D., -23.

35. Moon, P. B., J. Sci. Inst., 14, 189, (1937) 36. O'Neal, R. D., Phys. Rev., 60, 359, (1941) 37. O0Neal, R. D., and Phys. Rev., 58, 574, (1940) Goldhaber, M., 38. Rochester, G. D. and Phys. Rev., 63, 42, (1943) Janossy, L., 39. Rochester, G. D. and Nature, 156, 366, (1945) McCusker, C. B. A., 40. Shonka, Francis R., Phys. Rev., 55, 24, (1939) 41. Spatz, W. D. B,. Phys. Rev., 64, 236, (1943) 42. Stever, H. G., Phys. Rev., 61, 38, (1942) 43. Strong, John, Proce:dures in Experimental Physics, pp. 259-305, Prentiss Hall, 1943 44. Trost, A., Phys. Z., 36, 801, (1935) 45. Trost, A., Zeits. f. Physik, 105, 399, (1937) 46. Weisz, Paul, Phys. Rev., 61, 392, (1942) 47. Weisz, Paul, Phys. Rev., 62, 477, (1942) 48. Weltin, Hans, Rev. Sci. Inst., 14, 278, (1943) 49. Yaddanapalli, L.M., J. Chem. Phys., 10, 249, (1942) 50. Yarwood., J., High Vacuum Techniques, John Wiley and Sons, 1946 -24 -