Research Paper 2453 rOOmOI of R~eo"h of The NO tiO ~;;:if ~;=asoni;l~~i~;~~~::

Russell W . Mebs, John H. Darr, and John D. Grimsley

A study was made of the applicability of a number of metals and a lloys for therm all y stable ultrasonic delay lin es. A preliminary iavestigation was made of different types of pressure holders and ad­ hesi ves for use in crystal t ransducer attachments and of the iafluence of specimen length on attenuation for valious m etals and alloy. The effect of cold-work, annealing, and specimen cross section on attenuation was a lso determined for a representati ve isoelastic alloy. Measurements of temperature variation of signal attenuation, distort ion, a nd delay t im e on a number of assembled delay lines indicated t hat an isoelastic a lloy employing over cured epoxy-resin crystal attachments gave best over-all t ransmission characteristics. No correlation was obtainable between strength and SOLl nd-transmission characteristics with varioLls cemented joints.

1. Introduction variations in pulse attenuation and distortion with acceleration or change of temperature. The ultrasonic delay line has come into wide use Only a general survey will be given of the basic in recent years with the development of radar, com­ theory of ultrasonic transmission as related to delay ~uters , and other electronic devices [1 , 2, 3, 4V It lines of 50 tLsec or less. More comprehensive 's a means for delaying a signal for a predetermined treatments have been given elsewhere [2, 4, and 5]. ~hort period to be accurately reproduced for use at _tn appropriate later instant. The delay line consists 2.1 Attenuation 'of (a) a transducer for converting the electronic signal to a sound signal, and (b) a suitable medium The signals being considered arc 10-Ylc waves of 10f appropriate dimensions in contact with the trans­ short pulse length. The attenuation or diminution p.ucer that will transmit the sound signal to a second of the signal in traveling through the delay line is determined by (a) the conver ion in the transducers ~rans du cer at the opposite end of the medium, or permit its refl ection back to the original transducer ; of the electric ignal to a sound ignal, and vice it t is then converted to a usable electric signal. versa, (b) the effi ciency of transfer of the sound The most comm.on materials now used for delay signal betwecn transducer and delay line, and (c) lines are quartz [1] and mercury [5, 6]. Water, the loss in signal strength in passing throu".h the glass, and magnes ium alloys have also been used in line. Such attenuation i measured in decibel: (db) ~xperimental lines. Where acceleration or changes which can be calculated according to formula 1 ~n temperature of the line may be involved, many bf these materials are eliminated from usc. db = 20 loglo ~;, (1) The M etallurgy Division of the National Bureau pf Standards was asked by the Ordnance Electronics where Ao and AI are the output and input signal iDivision to assist in the search for a more suitable strengths, respectively. The factors involved in :material for delay lines by studies of metals that the loss in energy due to conversion between the might be useful for the purpose. Two specific electric and sound signals in the transducer will not eqUirement s were that it should satisfactorily be treated in this discussion. ransmit lO-Mc ultrasonic pulses at temperatures The efficiency of transmission of the sound wave ~'anging from - 50 0 to + 200 0 0, and possess a from the crystal into the delay line, or vice versa, egligible temperature coefficient of sound velo city is determined by the intimacy' of contact between ver this temperature range. the crystal and line and by the degree of matching of the acoustic impedances of the crystal and delay line materials. For purposes to be discussed later, f. Elements of High-frequency Sound-Pulse the side of the crystal opposite the delay line may Transmission in Delay Lines be held under pressure by an appropriate material ; the acoustic impedance of such material will like­ The factors to be considered in the selection of wise affect the transmission efficiency of the sound flllaterials for delay lines arc (a) the length of delay wave between crystal and delay line [2 , 4]. required, (b) the carrier frequency, pulse length. The attenuation losses within a metal delay line and pulse shape to be transmitted, (c) the allowable have been attributed to the elastic hysteresis, the variations in delay time with acceleration or change Rayleigh sound scattering, and the sound-diffusion of temperature to which the line may be submitted, process. The loss due to elastic hysteresis is attrib­ (d) the permitted maximum amounts of pulse uted to an absorption from the sound wave of the attenuation and distortion, and (e) the allowed energy required for cyclic stressing of the material 1 Figures in brackets refer to the literature references at the end of this paper. [7] . This loss is a significant part of the total loss

209 only at relatively low frequencies in the kilocycle there was no transformation to the compressional . range, being generally proportional to the frequency, mode; such transformations are obtained only upon whereas the other losses increase much more rapidly. reflection at certain oblique angles. Sound scattering and diffusion occur only in the In the present investigation, emphasis was placed megacycle range; they are approximately propor­ on determining which materials would give the least · tional to the fourth power of the frequency when the change in delay time per unit length with change in grain size of the metal used is somewhat smaller temperature between - 50 0 and + 200 0 C. The than the sound wavelength fLnd approximately delay time, L D , of a delay line can be expressed as proportional to the square of the frequency when the grain size is somewhat greater than the sound wave­ l LD=--J (4) length [8, 9]. Such scattering and diffusion are due Vshear to the differences of the elastic coefficients in adj acent grains of the metal in the direction of propagation where is the physical length of the line. From of the sound wave. eq (3 ) In addition to the principal signal transmitted through the line, there may be echoes caused by (5) I reflection of a portion of the signal from line ends and crystal surfaces. These are detected at a later time than the principal signal, and for the proper Recognizing that l, p, and G are all temperature­ operation of the delay line in some applications, dependent, and assuming isotropic expansion of a their amplitude should not exceed a small fraction of polycrystalline material, it can be shown the principal signal.

2.2. Distortion The pulses transmitted through the delay line The temperature coefficient of expansion (l/Z) may not only be attenuated but may also become (ol/oT) for most metals is relatively small, and ' distorted. for the alloys of iron and nickel accounts for less A primary cause of pulse distortion is the super­ than 0.25 percent of the change in delay time over a position upon a weak pulse transmitted directly 250 0 0 range. Thus a satisfactory thermally stable through the line of additional signals due to portions material of this type 'would be one with nearly zero or of the original signal that have been previously small negative temperature coefficient of the modulus scattered and then reflected or refracted t·o the (l/G) (oG/aT) over the required temperature range. receiving crystal over a path longer than the direct Little study has previously been made of the influ­ pulse. Other causes may be attributed to poor ence of temperature change on pulse attenuation contact between crystal and delay line, to improper and distortion. For temperatures ranging up to cutt,jng of the crystals, or to improperly prepared 200 0 0 , little or no change occurs in the microstruc­ delay lines. ture of many metals. Oonversely, cemented crystal attachments may be quite susceptible to temperature 2.3. Efhct of Temperature on Delay Time, Attenua­ change. Therefore, one might expect temperature tion, and Distortion variation in transducer loss due to change in the nature of the crystal contact with the delay line. In The velocit,y of sound within a rlelay line, and discussing the effect of temperature on attenuation I hence the delay time, are determined by the specific and distortion, a differentiation will be made in this physical properties of the material used. The veloci­ paper between that obtained within the delay line ties for longitudinal and shear sound waves differ and and that obtained at the crystal connections. are expressed as follows: 3. Test Materials, Specimens, Procedures/ (2) and Apparatus

3.1. Materials (3) The 14 metals and alloys investigated and their compositions and conditions are listed in table 1. where E is Young's modulus, A is the bulk modulus of The magnesium alloys were studied as metals elasticity, G, the shear modulus, and p the density of known to have good sonic properties and were also the material. The longitudinal velocity therefore is employed in the buffer lines to be described later. greater than tbe shear velocity. The high-purity and commercial nickel, Invar, 32- In the present series of tests, AT and AO cut quartz percent-nickel iron and 18:8 chromium-nickel steel crystals were used as transducers, and hence only were tested to ascertain whether any relationship shear waves were produced by the sending crystal. existed between the nickel content and degree of Furthermore, the directly transmitted portion of the signal attenuation and distortion obtained. The signal was normal to the receiving crystal so that aluminum single crystal was prepared in the Bureau's

210 hemical composition and conditicn of the mat erials used

Chemica l composition, percentage by weight

Designation ------~----~------~----.------,------.-- Condition Cr 1'11 0 T i Al Si C :l1n Others ------1------11------.------.------.---,-- Mg 95,64 2.7 o.a 0.3 rCu . 05 }Extruded, Zn 1.0 Mg 91.9·1 6,5 . 3 .2 rCu .05 }Extrllded. Zn 1. 0 0,01 , 03 {Cu . 003 . 10 S , 003 }Annealed . . 14 ,24 Cli ,0-1 Cold·drawn. _.- - _.------Reduced 45 percent in area by cold-drawing . . 14 . 75 . RI Reduced 51 percent in area by cold-drawing. 18:8 Cr- ' i steel '_._._._ 72.86 b __ •• ••• g,OO _. _ . . . _._ 18,00 . 14 Cold-drawn . Carbon too l steel ' .... _ 98,4 b._._ ••• _ .•...... __ .20 1. 20 . 20 Annealed, Alnminum. ____ ._._._._ . ___ . ______. _. __ .. _._ 90. 99 Single crystal grown by Bridgman method, Isoelast ic A d ._. _. _____ 49,2 _._. _____ 41, 5_. __ .. _._ 9,2 . 31 Reduced 30.5 percent in area by COld-swaging; annealed 1 br, at 500° C.

Isoelas tic n , . -_ .. _---- 52.8_._ ... _.. 36,7...... _.. 6,8 . 56 0,22 . 62 . 096 . 45 W , 01 Cold-drawn . , 53.4 _._ . ____ _ 37,0 _. __ .. ___ 7,0 , 18 f,e ,27 Isoelas tic C ------. 06 . 46 . 20 . 43 W . 03 }Hot-roll ed. Isoelas tic D d 47,6 _._. __ . __ 42,79 .. _ .. _ .. 5,97 I. 87 ,01 ,53 M g 18 IIot-rolled . 51.4. . ___ . ___ Isoelastic E o.:::::::::: 35,6 ------ll.l ,11 , 17 ------Reduced approximately 50 percent in area by cold· drawing

,'r ype analysis, b By difIcrcnce. 0 Determined by the Analytical Chemistry Section, National B ureau or Standards. d Determined by manuracturer.

M etallurgy Division . Of the isoelastic alloys, A was investigation [11] . It was found necessary Lo check an experimental alloy [10]; B, C, and D were com­ all crystals carefully as to uniformity of thickness mercial alloys; and E was prepared and cold -rolled (a O.OOOl-in. variation V,Tas permissible) and to in the M etallurgy Divi ion. indicate Lhe direction of the X -axis. Crystals were obtained with bo th faces silver-plaLed. 3.2. Specimens The compression fi tting used for holding a quar tz Most tests were made on specimen approximately crystal against a line during a test consisted of a 6 in. long. However, o ther specimens ranging in ~ amina te d plastic preSS Ul'e holder con taining a back­ physical length from X in. to over 6 in. were tes Led ; ll1g block and a socket-joint screw-tigh tening ar­ these permitted the measuremel1t of the attenuation rangement (flg. 1). The backing block was made of per unit length of a number of materials. lead or steel, with the cry tal facing s urf~ce ? ein~ At the beginning of the investigation the specim ens carefully polIshed. In some tests a 0.0005-m . tmfoil used were 0.35 in. in diameter over the cen tral por­ layer was placed between the crystal and specimen . tion, and Yz in. in diameter over a Yz-in. length at All facing surfaces were refinished between tesLs. each end. Pressure holder compression fi ttings were A similar set of compression fittings made of alumi­ I. used with this type specimen for holding quartz num were found to be unsatisfactory. They added ery tals diTec tly against the line during tcs t or for e 'cessive stray capacitance to the circuits. I mounting cemented crystals before test. Later tests The following adhesives were used in cemen ting indicated that specimen cross section was of little crystals to the delay lines. consequence, providing it exceeded the transducer area. H ence specimens of various sizes were sub­ sequ ently ,used, with a Yz-in. diameter the most D esignat ion T ype of adhesi ve , common size. A special jig was used for specimen preparation. AA Eleva ted-tem pera t Uf e thermosetting It consisted of a steel cylinder with its two fla t ends epoxy cement . made effec tively parallel by surface grinding. A BB Room- and elevated-tern perat u re t he /"- square hole was carefully machined along its cylinder mosetting epoxy cement. CC Silver paste. axis so as to be eff ectively normal to the end surfaces. DD Silicone. Setscrew held the specimen in place during surface E E Ba kelite. grinding and polishing (wet laps). The finished FF Beeswax. surfaces were fla t and p arallel wi thin a few ten thousandLh s of an in ch . 3.3. Transducer Assembly All the adhesives except the beeswax were applied separately to line and crystal and allowed to dry at One-quar ter-ineh squarc AC and AT cut crystals room temperature or in an oven before assembly. and ,)i-in . rOllnd AT cut crystals were used in this All connections were allowed to se t under pressllre,

211. B E F A

D

FIGUR}] 1. Bakelite compression fittings. A, Specimen ; n, Bakelite pressure holder; C, crystal; D, b ackl ll g plate a od electrica l lead : E, socket plate; x' , pressure scrcw with ball hcad.

~ --.--- r --~

FlGURB 2. Buffet· line assembly. A, Specimen; B, clamp holder; 0 , magnesium-all oy buffel' with cement at­ tached crystal; D, shielded brass spring connector with clectricalleads.

and additional curing was sometimes given the con­ modification of the method devised by McSkimin [13] . nection after the pressure ·was removed. Adhesive The two buffers consis ted of FS- 1 magnesium rods AA [12] was set at 170 0 0 or higher before the re­ 1 Y2 in. long. The Y2-in. length at each end was Y2 in. moval of pressure. Adhesive BB showed a tendency in diameter, and the central portion was 0.3 5 in. in to absorb moisture, which destroyed its sound­ diameter. Quartz crystals were cemented to one end transmission qualities. Occasional baking out or of each buffer . By means of the special clamp shown storage of the assembly in a desiccator removed this in figure 2, the unmounted ends of the two buffers trouble. were held under pressure against a specimen of any Attaching crystals to lines for each test by pressure desired length . It was necessary to lap the ends of holders or by cementing was time-consuming. Also the buffers adjacent to the test line between tests, after several tests with pressure holders a crystal and to regrind these ends whenever erratic test usually fractured, and cemen ted crystals could not results indicated the need for this operation. be salvaged for reuse. Furthermore, transmission efficiency varies with different crystals, and succes­ 3 .4. Electronic Apparatus sive tests with compression fittings, using a single pair of crystals, was found to produce sufficient Figure 3 is a block diagram of the apparatus used scatter of results to make difficult the measurement for generating square-wave pulses and for measuring of attenuation per unit length on low-loss materials. input and output signals for the delay line. A pulse Hence the buffer technique was developed and used generator capable of producing pulse lengths up to in many tests made at room temperature. This is a 10 J.Lsec and pulse repetition rates up to 5,000 cis

212 DELAY uffi.ciently slow to enable acc urate mea urement GENERATOR I of the temperature variaLion of input and output PULSE MARKER I signal strengths and of delay Lime. A negligible GENERATOR GE NERATOR temperature gradient was obLained along Lhe speci­ t CATHODE men during these measurements. AMPLIFIER DELAY LINE I f OllOWER ~ OSCI LLOS CO PE ~ SIG NAL 4. Effect of Type of Transducer Assembly on l GENERATOR I the Transmission of Ultrasound in Mag­ nesium Alloy FS- l FIGVHE 3. Diagram of test equipment. Table 2 lists the attenuation value obtained with a 6-in.-long FS- 1 magnesium-alloy specimen, using triggered the output of a signal generator adjusted pressure holders or cemented transducers. A torque to produce a 10-Mc constant-amplitude wave. The wrench was used with the pressure holders and the re ultant pulsed lif (high-frequency) waves were then torque adjusted to obtain minimum attenuation; amplified and fed into the input crystal transducer. fairly reproducible results were obtained by this The hf voltage generated in the output crystal method. It is apparent that the use of the cathode transducers was fed through the cathode follower follower greatly decreased the attenuation attained. into the cathode-ray oscilloscope where the amplified signal was placed on its vertical plates. The effec­ TABLE 2. Altemwtion at room temperature fo r various tra ns­ tive output impedence of the delay line is better ducer connections on 6-inch-long FS- l magnesium-alloy lines matched by the input impedence of the cathode fol­ lower than by that of the oscilloscope. This permits A ttelluation a more efficient transfer of output enrrgy to the oscil­ M ethod of holeling crystal against Pulse specim ens length Principal Fd~~ ~C:;Y - loscope. In some early tests, the cathode follower SIgnal signal was not used; a much higher effective signal attenua­ 1------tion thereby was obtained (table 2). Ilsec lib db Steel backing plate ______1.0 32.0 42.8 The oscilloscope sweep applied to the horizontal Do ______1. 0 • 53. 1 • 61. 3 plates is triggered through the delay generator by the Steel backing pla te with 0.0005-in. layer of tin fo il ______1.0 pulse generator; the delay generator can be adjusted Lead backing plate ______1.0 • 70. 4 • 57. 8 • 74.0 to retard the sweep by as much as 1,000 ,usec. The L ead backing p la te with 0.0005-in . layer of tin fo iL ______1.0 • 62. 8 • 85. 8 marker generator, which triggers the pulse generator, Adhesive AA______1. 0 30.5 36.7 Do______1. 0 • 47. 2 a 52.3 also produces small vertical markers every micro­ Do ______. ______2.8 28.4 second and larger markers every 10 ,usec along the Adhesive AA with lea d backing plates a nd lO-in .-lb torque o n pressure holdeL_ 1. 0 34.1 48.8 horizontal axis of the oscilloscope screen. An Adhesive BB ______1.0 a 58.9 Beeswax FF ______.______1.0 • 45. 8 alternate circuit permits placing the input signal • 59. 8 a 67. on the vertical plates. By accounting for the circuit a'rhese lines were tested with ou tpu t connected directly to oscilloscope, re­ amplification and attenuation of the various stages, m aining lines were testeel with Oll tput con nected throug h cathode follower to the ratio of output to input signal can be measured oscilloscope. by the 0 cilloscope. A Polaroid-Land camera was Examination of results indicates that lines ce­ used for photographing these signal shapes. mented with adhesivcs AA or BB gave the lowest attenuation of the principal pulse. The usc of 3.5. Measurements at Elevated and Low Tempera­ beeswax FF as a crystal adhesive resulted in a tures fairly large attenuation. Although a fairly high attenuation was obtained with ·thc lead-backed A nichrome-wound alundum-tube furnace was crystals when 0.0005-in. tinfoil wa inserted between used. This employed automatic control, providing crystal and line, successive echoes on this line were a temperature that varied less than 5° C over the greatly diminished in comparison with the lines specimen length as measured by thermocouples without the foil. The echoes obtained with the line attached to the center and ends of the specimen. cemented with adhesive AA were somewhat larger Some few tests were made 'with a shorter 5-in. than those obtained with any combination of holder furnace in which 6-in. test lines had their transducers and backing plates. Furthermore, when a crystal extending beyond the ends of the furnace. Thus cemented with adhesive AA was placed in a pressure the temperature variation of transmission character­ holder and held with lead backing plates, the istics of the test material, exclusive of the transducers, attenuation of the principal pulse was increased only could be obtained. All tests were made on specimens a few decibels, whereas the first echo attenuation with cemented quartz crystals. increased by about 12 db. In order to obtain subzero temperatures, the test Of particular interest in these tests is the effect specimen was placed inside and midway as to the of transmission on wave shape. There are three length of a thick-walled Hi -in. copper tube; dry ice general methods of evaluation of this effect, namely, was placed around this tube. (a) the visible change, as noted on the oscilloscope The rate of cooling after the application of dry and which can be photographed; (b) the range of ice and heating llpon sublimation of the ice were frequencies over which the amplitude of the principal

213

~------~ 60.-----.------,------r-----,------,----~ , SENT 50 IS/S'STAIN LESS STEEL_ o~ I I ~ HEAT TREATED CARBON I 40 ~-~ -:. TOOL STEEL...... j

30 MAGN ES IUM ALLOY F S -1 ...... 0 0-0 70 20 ~------~3~2~~NIC~K~EL------~

.-'/ ALLOY> IF> 60 -- ....J COLO DRAWN ...,...... O 0 ...... w N ICKEL--a-~~ RECEIVED i3 50 ~ ~ ~ ~ ~-:._' N_V _A R~~~------la £ 40 0Li~ o ¢ ~ => 30 801------., z W ..Is'Ra l­ I­

.. ..~

: ~ . iIlor' , : •• ' . \ --.'". ~ I't .' -

FIGURE 6 NIicrographs of structure of various materials. ~<500

A. Cold·rolled 36% nickel, 12% ch ro mium, 52% iran alloy E, cross section, of nitric (co nc.) and 4 parts of glacial acetic acid. D. Cold·drawn isoelastic alloy etch cd in Precision No.2 reagent. 'I' he rather complex formula and method of B, longitudinal section , etched in1 part of ferric ch loride, 10 parts of hydrochloric preparing this reagent have becn described in [14]. B. Cold·rofIed 36% nickel, acid and 20 parts of water. E. Hot·rolled isoelastic all oy C, cross section, etched 12% chrominm, 52% iwn all oy E , longitudinal section, etched in Precision No.2 in Precision No.2 reagent. F. Hot·rolled isoelastic all oy C, longitudinal section, re~gent. C. Anncaled hi gh·purity ni ckel. longitudinal section, etched i11 6 parts etched in Precision No.2 reagent.

215 lower than Invar. High-purity nickel was the 70 ,----,-----,-----,-----r-----,-----,----, poores t transmitter of sound of the materials shown, although it did not show a correspondingly high value of buffer loss. 60 A short length of alloy E was found to give ex­ tremely high values of attenuation, as did the single crystal of aluminum. Attenuation per unit length could not be evaluated for these two materials. A study of figure 5 and table 3 reveals that no general relationship exists between nickel content and attenuation per unit lenr th for the respective m etals z and alloys. Factors such as other constituents, ."'.... degree of cold work, and grain structure must also ... be considered as important. The isoelastic alloy E 30 (fig. 6 A and B) and the high-purity nickel (fig. 6, C) both possessed large grains, which ma.y account, in 20 part, for the high attenuation per unit length ob­ 0 2 3 4 6 7 tained. The very low attenuation per unit length LI NE LENGTH , INCHES of the quenched and tempered carbon tool steel can be attributed to its structure; when annealed, the FIGURE 7. I nfluence of annealing treatment and additional tool steel was found to give a somewhat higher value. cold-swaging on the attenuation-length characteristics of iso­ Of the commercial isoelast.ic alloys tested, alloy B elastic alloy A. was cold worked a greater amount, with the accom­ Curve I, as received (cold-swaged to 30.5% reduction of area and stress-relief panying refinement of its structure (fig . 6, D), than annealed, 0.250-in. dia meter); curve 2, fully annealed (0.250-in. diameter); curve 3, additionally cold-swaged to a total of 70 % red uction o[ area (0.165-in. diameter); was alloy C (fig . 6, E and F ). This, accordingly, curve 4, additionall y cold-swaged to a total o[ 70% reduction o[ area (0.165-in. diameter) plus stress-annealed at 400° C [o r 1 hour; curve 5, as received, ane! ma­ might be presumed to be a prime factor for the chined to 0.165-in. diameter. low attenuation for alloy B. However, a few tests on another heat of hot-worked alloy C with a struc­ ture similar to that of figures 6, E and F , indicated It should be noted in this group of curves, how­ this material to have as low attenuation per unit ever, that all except the curve for the annealed alloy length as alloy B. possess nearly the same slope for the portion lying between zero and 1 in. in length, which is the extent 6. Effect of Cold Deformation, Annealing of some of the upper curves_ Thus the principal Treatment, and Sound-Path Cross Section difference in attenuation values for these curves on the Attenuation of Ultrasound can be attributed to a difference in buffer loss. The principal effect of the additional cold-swaging, A rather extensive study was made of the ul tra­ therefore, exclusive of the eff ect of reduction of sonic properties of isoelastic alloy A during this cross section, was its apparent effect on the trans­ investigation. This material was received in X-in.­ mission characteristics of the buffer-specimen inter­ diameter size, having been cold-reduced 30.5 percent face, i. e., the additional cold-working evidently and stress-relief annealed at 500 0 C during manufac­ tended to impede the transmission of sound be­ ture. This material, as received, possessed a rather tween such surfaces. The increase in buffer loss by large attenuation per unit length and a low buffer full annealing indicates that optimum sound-energy loss, as shown in curve 1, figure 7 (also shown in transfer occurred at the interfaces when this material fig . 5). Full annealing of this material tended to had been only moderately cold-worked and stress­ raise the attenuation length curve, as show in curve relief annealed (compare curves 1 and 2). 2. When the as-received material was cold-swaged In the above tests, specimen diameters varied to 0.165-in. diameter, that is, to a total cold reduc­ from 0.25 to 0.165 in., thereby affecting somewhat tion of 70 percent, an appreciable rise occurred in the sound-transmission effici ency, inasmuch as the the attenuation values (curve 3). Stress-relief an­ effective sound-path cross section at the input crystal nealing of the swaged material caused a slight lower­ could be as great as X in. sq. Some tests were there­ ing of attenuation (curve 4). fore made to determine the effect of increasing speci­ In order to determine whether the increase of men cross section beyond that of the crystal on the attenuation was due to the actual cold-working of sound-transmission effi ciency_ One-inch-long speci­ the material or whether it was due to the reduction mens of magnesium alloy J- l were m achined to effective cross section of the sound path, som e of the diameters ranging from }~ to X in. and tested between as-received material was machined to 0.165-in. buffer lines. A maximum of only 3-db variation of diameter. Curve 5 for the machined material lies attenuation was obtained in these tests, with no between curves 1 and 4, and thus indicates that the regular variation with diameter. Therefore, the rise in attenus.tion of the swaged material was effect of specimen cross section exceeding the crystal partly due to the effect of cold-working and partly area evidently does not have a significan t effect on to reduction of cross section. the effi ciency of sound transmission.

216 7 . Effect of Temperature on the Attenuation fmnace on a similar specimen, with cooling air blast of Ultrasonic Pulses in Various Materials directed at the transducers, indicated a lowering of attenuation by as much as 8 db at the higher temper­ Tests were conducted to deLermine the effect of atures. Such evidence suggests that the attenuation Lemperatme on boLh the delay-line material and the per unit length of the test alloy aCLually diminishes transducer connections attached to the delay line. at higher temperatmes. Figures 8 and 9 show the variation of attenuation The remaining curves in these figures were ob­ with temperatme of various assembled delay lines. tained with the entire delay line situated in the The pulse lengths are indicated in the legend of each longer furnace. oW-hen adhesive AA was used with figme. Wi th the excep Lion of cmve 3 of figme 9, the recommended 200 0 C curing temperature, a the optimum pulse length .vas employed. sharp rise of attenuation (emve 2) occurred at Figure 8 provides a comparison of various com­ elevated test temperatures in contrast to the less binations of cements and cming conditions on 6-in.­ abrupt rise obtained after overcuring at 275 0 C long alloy B delay lines. In a preliminary test, (curve 3). employing the shorter 5-in. furnace, with transducer Curves 4 and 5, obtained with adhesive BB, cured connections extending outside the furnace, the at the optimum temperature 2 of 280 0 C for this attenuation was found to show negligible variation, cement, show the variations of attenuation as ob­ excepL for a small decrease at 120 0 C (curve 1). Thus tained with the principal signal and the first re­ a temperature change of the delay-line material, in flection. Although a small temperature variation of the range shown, caused little variation of attenua­ attenuation is secured, the strong reflected signal and tion--the transducer connection probably is the chief the large pulse length required for maximum signal element affected. Later test with the shorter are both deterring factors for satisfactory ultrasonic pulse transmission in such a delav line. Somewhat 7or-----,-----,------r-----,------,----, shorter optimum pulse lengths \vere obtained with adhesive AA (curves 2 and 3). C urve 6, obtained with silver-paste adhesive CC, 60 indicates remarkably small variation of attenuation with temperature and a low level of attenuation, althou&h .requiring a long pul e length for optimum 50 transmlSSlon. C urve 7 and 8, obtained with silicone adhesive DD, illustrate the flattening of the attenuation- 70 40 r------~ ' The optimum temperature for any cement is that minimum cnring tempera, tUI'C resulLing in the least temperature variation of attenuation. v ~. 70 v z o o ;: 60 4 ~ 50 z ,...

50 50

40

TEMPERATURE. 'c I FIGU RE 8. Influence of temperature on attenuation of 10 mega­ cycle ultrasonic pulses in isoelastic alloy B , crystals cemented 30~0--~'~401~== ~~~eo ~~h---~--~~~280 as indicated. M easurements made for initial received signal TEMPERATURE · C with entire line at indicated temperature, except curve 1. FIGU RE 9. Influence of tem perature on attenuation of 10 mega­ Curve I, 6-iu.line, adhesive AA, cured 200° C, tested in 5-in. furnace, optimum cycle ultrasonic pulses of designated length in various delay pulse length exceeds 10 microseconds; curve 2, 6-in. liue, adhesive AA, cured 200° C, optimum pulse length 2.5 to 6.8 microseconds; curve 3, 6-in.line, adhesive AA, lines, crystals cemented as indicated. M easurements made for cured 275° C after test 2, optimum pulse length 2.75 to 5.4 m icroseconds; curve 4, initial received signal with entire line at indicated tem perature. 5.3-in. line, adhesive BB, cured 280° C, optimum pulse length 5.4 microseconds' curve 5, first reflection of test 4; curve 6, 5.3-in. line, adhesive CC, cured 550 0 C; C.urve 1, 6-in. l!S-l magn~sium-alloy line, adhesi ve AA, cured 200 0 C, optimum pulse lcngth exceeds 10 microseco nds; curve 7, 5.3-in. line, adhesive DD, optimum pulse length 1 mICrosecond; curve 2, 6-i n. FS- l lllagnesiuID ­ cured.250~ 0, optimum pulse length exceeds,lO microseconds; curve 8, 5.3-in. line, alloy line, adhesive BB, cured 200 0 0 , optimum pulse length exceed s 10 micro­ adheSive DD, cw-ed 350° C after test 7, optimum pulse length exceeds 10 micro­ seconds; curve 3, same as test 2 with pulse length 3.7 microseco nds' curve 4 6-in. seconds; eurve 9, 5.3-ill. line, adhesive EE, optimum pulse length exceeds 10 i soe l astic-~ll oy C linc, adhesive BB, cured 200° 0, pulse varies 0.8 to 1.4 rnicro­ mieroseconds. seco nds.

275121-53--- 2 217 temperature curve by elevating the curing tempera­ better signal transmISSIOn qualities. Thus the de­ ture of this adhesive from 250 0 to 350 0 C. However, velopment of good transducer connections becomes curve 8 li es somewhat above curve 6 and was ob­ purely empirical, inasmuch as the sound and mechani­ tained with a larger optimum pulse length than was cal properties of these elements cannot be correlated. curve 4. The apparent plastic condition of silicone cemcnts, even at elevated temperatures, obviously 8. Effect of Temperature on the Delay Time detracts from their usefulness for transducer connec­ of Ultrasonic Pulses in Various Materials tions in delay lines. Curve 9, for Bakelite cement EE, indicates a fairly high attenuation with a low Figure 10 shows the influence of temperature on temperature variation. A fairly long pulse length delay time, or time of transmission, of 10-Mc. ultra­ is required. sonic pulses in assembled delay lines of various ma­ Figure 9 shows the variation of attenuation with terials. Some of these tests were extended to sub­ temperature for additional materials. Curves 1, 2, zero temperatures. The FS- 1 magnesium-alloy line and 3 were obtained with FS- 1 magnesium alloy. showed the greatest change in delay time with tem­ Adhesive AA cured at 200 0 C (curve 1) provides a perature. Of the isoelastic alloys tested, alloy C fairly shor t optimum pulse length, although produc­ alone shows no variation in delay time up to the ina a large variation of attenuation with temperature highest test temperature, 194 0 C. The delay time for 0 0 ab"'ove 120 C. Adhesive BB, cured at 200 C, gave alloy B does not change up to 180 0 C, but rises there­ a smaller variation of attenuation, although it had a after. The delay time for alloy D rises gradually with I somewhat higher room-temperature value. The rise temperature up to 130 0 C, above which it is nearly in attenuation produced by shortening the pulse constant. length from its optimum value of 10 ,usec (curve 2) to The sensitivity of measurement of delay time in 3.7 ,usec is shown in curve 3. Data for these latter these tests was approximately 0.1 ,usee. For a 50- two curves were obtained during a single tempera­ ,usee delay line, the limiting variation of the temper­ ture-attenuation test. For this reason, curve 3 be­ ature coefficient of delay, measurable over a 200 0 C tween the points at the two high est temperatures is range is 10 parts per million. Thus for alloy 0, this , drawn approximately parallel to curve 2. coefficient eviden tly is 8 parts per million or smaller. Curve 4 is for isoelastic alloy C with adhesive BB. By comparison, the temperature coefficient of delay A small rise of attenuation occurs at about 150 0 C. for the FS- 1 magnesium-alloy line averages about Additional tests were made on other delay Jines 380 parts per million over the test temperature range. with similarly cemen ted crystals for temperatures The small change in delay time above 180 0 C ob­ ranging to as low as - 60 0 0. Negligible variations served with isoelastic alloy B may not be due to an in attenuation were noted below room temperature; actual change in properties of the alloy. A rather these curves are not plotted. poor received-pulse shape was obtained with this It is evident from the above results th at perhaps material; it is possible that a change in pulse shape the principal variable in the efficacy of ultrasonic at the upper temperature levels due to some change transmission in the isoelastic alloy delay lines is the transducer assembly. It was therefore considered of some interest to determine whether any relationship 59 , existed between the mechanical strength and sound­ transmission efficiency of the cements for the various ISO£LA$TIC - 8 ~ curing temperatures employed. 58 fr-.6-...6-~ A ~ ,Q ~~-----J Some mechanical tests were made in which carbon­ steel tensile specimens were cut in two, the cut sur­ faces being carefully polished and rejoined with the 57 55 various adhesives by the same methods employed in attaching crystals to delay lines. These specimens 54 were then tested in tension. Maximum tensile strengths were obtained with joints of adhesives AA and BB, when cured at 130 0 53 and 150 0 C, respectively. It will be recalled that the optimum curing temperatures for these cements for best signal transmission were 275 0 and 280 0 0, 52 ~ respectively. At these latter temperatures, the mechanical strengths of the joined bars were appreci­ ably lower than maximum. It was also observed 51 that sil ver-paste adhesive OC, which gave low signal attenuation, possessed quite low mechanical strength at all curing temperatures. 50

Eviden tly high strength and optimum signal trans­ -80 -40 40 80 120 160 200 mission are not synonymous. The higher curing TEMPERATU RE, 0 c temperature, though probably causing a more brittle FIGURE 10. I nfluence of temperature on delay time for delay and weaker joint, may provide in such brittleness lines of Val'ious materials.

218 in Lhe propertic of Lhe acl hesive might effeoL a shift Thi laLLer attachmcnt m cLhod al 0 produccd the in position of the received-pul e peak and thus effec­ smallc L refl ected signal . tively change the m ea ul'ed valu es of delay time. 2. The attenuation per unit length of a maLerial is best measured by the buffer LesL meLhod. IL varies 9. Delay-Line Characteristics from a very small valu e for t he FS- l magnesium a ~l o. y to very. hi~h values f,or an~eal e cl b} gl, -pmi.LY The res ults or Lhe above tests indicate limitat ions n ickel and a smgle crystal oJ alummum. fhe burIer as to the optimum delay-line characteri tic ob­ loss ranges from small values ror the FS- 1 magne ium tainable. Wher eas a magne ium delay lin e may alloy to somewhat higher values f01' Lhe isoelastic provide satisfactory attenuation and pulse-repl'oduc­ alloys, Invar, and 18 : 8 Cr-N i steel. tion characteristics, its delay time varies appr eciably 3. Although very coarse-grained maLerials pre­ with temperature. Constancy of delay time with sumably give a high attenuation loss, such losses do low attenuation is most neady fulfilled by isoclastic not show a fixed relationship with grain size for the alloys Band C; alloy C exhibits no m easurable finer structure commercial isoclastic alloys. temperature variation of delay time. \iVhen crystals 4. Although pure nickel possesse a high attenua­ are attached to an isoelastic alloy C delay line with tion per unit length, there is no regular variation of adhesive BB and cured at 200 0 C, a reasonably high iLttenuation with nickel content. Invar, a 36- attenuation is secured (fig. 9, cLll' ve 5) that is n early percent-nickel iron alloy was found to have an independent of temperature. In oreler to obtain a attenuation pel' unit length of the same magnitude temperature invariant attenuation at a lower loss as magnesium FS- 1 alloy and quench ed and tem­ level , as typifi ed b~T curves 4, 6, and of fi.gur e for pet'ed 1.2-percent-cal'bon Loo l steel. isoelastic alloy B , it is necessary to use delay lines 5. In Lests on the eff ect of cold-work and annealing th~ t produce gr eater igna,l distortion .and thus r e­ 011 one of the isoelastic alloys, it was determined qLllre a gr eaLer pulse length for optImum trans­ that the attenuation per unit length was no t affected ; mission. These tcsts wou ld indicate that for t he e however, the buffer loss wa Lhe lowest at an inter­ alloys a low attenuation is not compatible with a med.iate stage of cold-work. short opLimum pulse length. The increase in buffel' loss of Lhe everely swaged The number of variables t hat affect sound Lrans­ isoelastic alloy was due only parLly to the change in mission make it difficult to pro\Tide a simple explana­ stru cture of t he material , being al so du e partly to tion of the occurrence of the above r elationships. the r eduction in cross section of the material. vVhere Some of these variables are (a) the damping of the the cross-sectional area of the specimen exceeded crystals by the cements and delay line, (b ) the effect that of the crystal, the eff ect of change of section of a poorly clamped resonating crystal on signal co uld not be detected. distortion, (c) the relative acoustic impedance of 6. Of the various metals and all oy tested, two crystals and line materials, and (d) the influence of commercial isoelastic alloys, of differen t manufac­ the cement-layer characteristics on the sound trans­ t ure, wer e found to possess the least variation of mission. Possibly other variables affect . these r e­ delay time with temperature over the range -500 lationship , necessitating an empirical study of each to + 200° C. Both alloys contained approximately crystal, cement, and material combination in order 36 percent of nickel, 7 to 8 percent of chromium, plus to determine their suitability for use in delay lines. other minor constituents. The variation of over-all These tests indicate that the transducer assembly attenuation with temperature for these alloys was is the greatest variable in obtaining satisfactory found to be due primarily to the transducer assembly delay-line transmission over a wide temperature employed ; a mall deer'ease of internal attenuation range. with rise in temperature was indicated on one of A primary r equirement is a thermally stable these materials. A negligible variation of attenua­ cement layer of suitable transmission characteristic. tion ~vi th temperature up to 200 0 C, at a low attenua­ Even with an ideal cement, transmission is limited tion level, was secmed by cementing the crystals to by the matching of the acoustic impedance of the the more severely cold-worked of these two alloys crystal and delay line. with a ilver-paste cement or one of the proprietary epoxy-type cemen ts that had been overcured after 10. Conclusions attachment. For the isoelastic alloys low attenuation was found From measurements made of the transmission of to be incompatible with short optimum pulse length 10-Mc ultrasonic pulses in variou metals and alloys, for any of the transducer connections tested. The the following conclusions may be drawn: limitations of obtaining good tran mi ion appar- 1. The most efficient of the methods employed for entlv are associated with the characteristics of the conver ion of electrical energy into sound energy cements and the relatively poor impedance match of within a delay line, and vice versa, is to attach quartz crystal s with isoelastic alloys. quartz crystals to the delay line by an epoxy-type Somewh at higher curing temperatures were n eces­ cement. The best reproduction of the principal sary for producing cementecl joints having best tran - . transmitLed signal wave shape, however, was ob- mission qualities, than fo r procuring maximum tained by 1J e of laminatcd plastic pressure holders mechanical strength. in which the quartz crystal wer e backed up with 7. Although the buffer technique presents a rapid lead blocks and separated from the line by tinfoil. means for determining which metals possess a low 219 attenuation per unit length or low transducer .losses [3] A. G. Emslie and R. L. McConnell, Radar system engi­ neering, 1, chap. 16 (McGraw-Hill Book Co., Inc., New at room temperature, the more t edious procedure of York, N . Y., 1947) . empirical determination of crystal, cement, and [4] H. B. Hunt ington, A. G. Engle, and V. W. Hughes, Ultra­ metal delay line combinations tested over the tem­ sonic delay lines, J. Franklin Inst . 245, 101 (1948). perature range to which the delay line will be sub­ [5] A. E. Benfield, A. G. Emsli e, and H . B. Huntington, On jected, is necessary for satisfactory development of the theory and performance of liquid delay lines, MIT Radiation Laboratory Report 792 (September 21,1945). such lines. [6] 1. L. Auerbach, J. P . Eckert, R. F . Shaw, and C. B. Sheppard, Mercury delay line memory using a pulse rate of several megacycles, Proc. IRE 37,855 (1949) . Grateful acknowledgment is extended to W. M. A. [7] R. L. Wegel and H. Walther, Internal dissipation in Andersen of the Andersen Laboratories, Inc.; to :M. solids for small cyclic strains, Physics 6, 141 (1935). E. Fine and W. C. Ellis of the Bell Telephone Labora­ [8] W. P . Mason and H. J . McSkimin, Att enuat ion and scat­ t ering of high frequency sound waves in metals and tories; to W. A. Mudge and E. M. Wise of the Inter­ glasses, J . Acous. Soc. Am. 19, 464- 473 (1947) . national Nickel Co.; and to C. H. Fetter of the [9] W. P. Mason and H . J . McSkimin, Energy losses of sound American Time Products, Inc., for technical advice waves in m etals due to s catt ering and diffusion, J. Appl. and for furnishing some of the isoelastic alloys in­ Phys. 19, 940- 46 (1948). [10] M. E. Fine and W. C. Ellis, Thermal variation of Young's vestigated. The authors are also indebted to W. E. modulus in some Fe-Ni-Mo alloys, Trans. AIME 191. Eberly, Carpenter Steel Co. for cold-drawing the 761 (1941) . 32-percent-nickel iron alloy and for furnishing the [U] W. P . Mason, Piezo-electric crystals and their applica­ Invar alloy. Many thanks are due the following tion to ultrasonics, chap. XV (D. Van Jostrand Co., Inc., New York, N. Y., 1950). members of the staff of the National Bureau of [12] A new bonding resin, detailed results of laboratory tests Standards: L. T. Sogn, for providing much tech­ on araldite, Sheet Metal Ind. 26, 1966-1988 (1949). nical information on quartz crystals; T. E. Orem, [13] H. J . McSkimin, Ultrasonic measurement technique, J . for the aluminum single crystal ; and R. H . Harwell, Acous. Soc. Am. 22, 413- 18 (1950) . for preparing isoelastic alloy E. The authors also [14] Charles T. Flachbarth and Chester S. Pondo, Three new etchants for nonferrous m e tal ~, Metal Pro gr. 56, 688- acknowledge the extensive assistance and advice 691 (1949). provided this work by E. Voznak of the Ordnance Electronics Division, the sponsoring agency.

11. References [1] David L. Arenberg, Ultrasonic solid delay lines, J. Acous. Soc. Am. 20, 1 (1948). [2] Crystal Research Laboratories Interim Engineering Re­ ports, Contract W28- 009 and IV28- 110 (July 1946 to December 1948). WASHINGTON, June 8, 1953.

220