Journal of Research of the National Bureau of Standards Vo!' 47, No. , December 1951 Research Paper RP2270 Thickness of Glass Electrodes 1. 1. Diamond and Donald Hubbard
A nondestru ctiv e magnetic m ethod has been developed for t he meas ure ment of t he t hickness of glass eleelrode bulbs. It involves the use of t~ e Brenner :\1agne-Gage as t he measunng lI1 strument a nd ca rbon y lll"on powder as the baclollg material. Each of t he fo ur g;lasses inves.tigated was found to have a characteri stic " d ~ part ure t hickn ess, " rangin g fr om ()4 to 130 mI crons. A glass electrode whose mlllllnUm thIckness was equal to or less than this departure thickness was found to give the full 59 millivolts p er pH unit electrode response. Bulbs of in creasin gl y greater thickness showed in creasing departures from t his theoretical response. The deparllll"e thicknesses of the glasses investigated were fo und to be a fun ction of their h yg ro sc opic i t i e~. 1. Introduction ments on paint, porcelain enamel, plastic, and other film s or coatings that arc or can be backed by iron The effect of the thickness of the membrane of a 01' steel plates. glass electrode on its electrode fLm etion is of interest . The aciaptation made. by us is lhe use of carbonyl in connection with a continuing sLudy of the glass Iron powder as the backmg material and the use of a electrode [10 , 12, 13, 16j.l Kratz [18] made the sLate corresponding direct calibration prO CedlU" Cl. ment that the theoretical variation of potential with pH can be obtained only with thin-walled membranes. 2. Experimental Work Kahler and DeEds [17] found tha t there was 1"0 ughly a linear relationship between the "average electrical 2.1. Method of Measuring Thickness thickness" of an elec trode and Lhe difference, in m illi volts per pH unit, beLween its electrode response and The operation of the Magne-G age is based on the that of a hydrogen el ectrode. This difference ap dec rease in magnetic aLL racLion resulting from the proached zero as the membrane thickness approached interposition of a nonmagn etic material between a 7:ero . . Dole [7], however, quotes von Stciger [20] to magnet and a magn etic material. The apparatus, the effect that glass electrode potential is completely shown in figure 1, consisLs of a lever a.rm Lhat is independent of the thickness of the glass. The pres connected Lo a coiled spring, which is in turn con ent study has been made in order Lo clarify the nected to a dial and 1000b. Suspended from the ond matter. of the lever arm is one of se veral interchangeable MacInnes and Dole [19] determined the thicknes permanent bar magnets of different weighLs and pole of their thin membranes by weighing a measUTed area strengths. Two of the magncLs available have been of tmiform thiclmess, this UJliformiLy being deter found to cover the range of thicknesses of in tere t in mined by Lh e usc of interference colors. Knowing this sLudy. the glass density, they calculated their films to be In use, Lhe steel-backed specimen is placed in posi about 1M thiclc Von Steiger [20] measlll"ed the thick tion under the magnet, and the knob is roLated ness of his Cremer-Haber electrodes [5 , 9] by bringing counterclockwise unLil the presser arm, which. is slip a glowing glass rod up to a selected di stance from the mOLm ted on the assembly, brings the rounded Lip of glass surface. By observing the reHection against a the magn eL in to con tact with the surface of the speci black background, he observed interferen ce fringes men. The counterclockwise rotation is continued and then calculated thicknesses. Y oshimma [22], in until the built-in stop is reached, or until a trial or his study of the effect of thickness on asymmetry previolls experience indicates that the tension on the potential, assumed the thicknesses of his electrodes spring has been released suffi ciently so that the were in the order of their electrical resistances and magnet will remain in contact with Lhe specimen had no direct knowledge of the thiclO1 esses involved. when the knob is rotated clockwise, taking the presser Kahler ancl DeEds [17] measured the thiclO1esses of arm off the lever arm and magnet. The 1010b is then their electrodes by using a microscope with a cali rotated clockwise, increasing the tension on the brated fine fo cus adjustmen t to focus on the t·wo spring until the force is sufficient to pull the magnet surfaces of the membrane. The difference in setting, away from the specimen surface. The dial reading corrected for the refractive index of the glass, gave at this point is a Itmction of the thickness of the the thiclmess. nonmagnetic material. When the instrument is For the purpose of this stuely, a nondestructive properly calibrated with material of known thick magnetic me~hod for the measurement of the glas~ ness, the dial reading can be converted to a thickness membrane tluckness has been developed. It involves reading. the use of the Magne-Gage 2 [1], developed by A Cremer-Haber type glass elecLrode is typically a Brenner [2 , 3], for the measurement of electroplated thin glass bulb about 10 to 20 mm in diameter blown coatings on metals, and is also used for measure- on one end of a glass tube a few millimeters in diam eter. Since it is impossible to use a steel plate as a I Figures in braekels indicate the literature references at the end of this paper. 2 J\I[ anufacturecl by the American I nstrument Co., Sil ver Spring, Md. backing material, for a thiclmess measurement on an 443 160
150
140
130
120
110
!Q 100 o a: o i . 90 en en ~ 80 o "% .... 70
60
50
FIG UR E 1. Aminco- Brenner JYfagne-Gage. An instrument for the measurement of fi lm thi ckness. o~ __~ __~~ __~ __~ __~ __~ __~ __~ __~~ 50 60 70 80 90 100 110 120 130 140 150 1",0 ?bject of this shape, carbonyl iron powder is used SCALE READING- mstead. The powder is placed in the clean, dry elec FIe l' HE 2. Calibration C1l1've showing scale reading as a f1lnction trode by a standard procedure, and its open end is of thickness f or magnet 2 of the 1IIagne- Gage. st opp~r e d. .After adjusting the magnet to the ap Carbonyl iron powder L (20 p. average parLi cle size), Vibration-packed to propnate heIght b)T means of the rack and pinion the equilibrium, used as backing material. bulb is placed in contact with the plastic g~Lard around the magnet, and the measurement is made. fLlms or coatings, the Magne-Gage is calibrated Contact with the guard serves to steady the specimen against thickness standards supplied by the manu during measurement, and, if maintained during mul facturer. These standards are squares of steel elec tiple measurements made to obtain a precise value troplated with known thicknesses of a nonmagnetic for the dial reading, will insure that the readings are metal. The resulting calibration curves are useless being made on the same spot. This is important, as for interpreting dial readings obtained with powder all such glass bulbs are quite nonuniform in thickness. backed samples. We use instead thickness standards One additional precaution must be observed in han made by cementing aluminium or platinum foils of dling th e instrument when making measuremen ts on known and uniform thickness 3 over the open ends powder-b.acked, thin membranes. It is easy for the of glass tubes of about 9-mm diameter. These simu magnet tIP to dent soft or thin metal or to puncture lated electrodes are filled with carbonyl iron and a thin glass membrane unless con tact is macl e gently measured in the same manner as the electrodes. The and no pressure is exerted on the specimen by further calibration curves resulting for magnets 2 and 3 are counterclockwise rotation of the instrum ent knob. shown in figures 2 and 3. But in the usual case, the spring tension must be de In his work on the thickness of electroplated met creased to insure that the magnet will remain in con als, .Brenner [2 , 3) found that the geometry of the tact with the specimen when the presser arm is lifted. specunell may have an appreciable effect on the scale This is accomplished before contact is made and reading obtained. In order to prove that no serious without removing the specimen from contact with errors are introduced by the difference in shape be the guard, by rotating the knob counterclockwise tween the Cremer-Haber el~ctrod e and the straight while using the thumb to prevent the presser al'I~ tu be standards, the followmg experiment was per from also rotating. formed. A thiele-walled bulb 15.5 mm in ouLside
When used for the measurement of steel-backed 3 Foil-tbicknes8 measurements made by A. O. Strang. 444 T ARLE 1. NJagne-Gage scale reading as ajlected by vibration tim.e and )JOI·tial size
1 ron powder g- rade Vibration t.i m e L I (51'Tfl) I ( 8~E ) I (201') ---'---1 U 81' s traigh t-Lllbe thickness standard Ascragc sca!e read in g (m ag- net :1)
min 0.5 ______I 1 ______I. 5 :1() 2 1. 5 !5 10. 5 1 ..\ ______2 ______10 28 14 10 2.,j ______20. 5 3 ______10 27 20 n. 5 II> 4 19. Z 5 ______------__- 26. \ 5 13. 5 0 lU. 5 13. 5 cr 6 ______26 19.5 () 8 ______i 280 25. 5 . 10 ______26 ..II> '"z 280 Typical Cremer-Flabc!' type glass eleelrod e () A \' era~e sca le reading (m agnet 2) "x .... 240 1 ______----- 127 120.3 2______138.5 125.5 120.2 220 125 L :::::::: i:iii --- 125 6______135 200 8______134.5 10______13,1. 5
180 was packed by holding the filled electrode in the 160 !l0rm~l bulb-·down position and l'lmning a mechan lOul Vibrator up and down the Ollt ide of the glass 140 tube. In these experiments, a Sargent Ideal electric marker, set for a fairly high vibration amplitude 120 was used. Preliminary experiments indicated that vibration time required to reach equilibrium and 10 20 30 40 50 60 70 80 90 Magne-Gage scale reading at equilibrium both de SCALE READING pended on the energy of vibration. No quan titative data were obtained on this poin t, but it must be FICURE 3 . . Cct/1;bralion CUTve showing scale Teading as a Junction oj thickness JOT magnet S oj the ll1fagne-Gage. emphasized that the method and energy of vibration selected must be the same during measurement as Carbony l iron powdcr L )20 I' average particle size). v ibration-packed to eQUl IIIJrHlIII, used as backlllg m aterial. during standardization. From the data summarized in table] several dial!leter was blown at the end of a piece of 6-mm facts arc evident. For each of the fOllr ~ r ades of t ~bmg . A second similar electrode was prepared carbonyl iron powder tried , vibration will bl'esult in WIth a 1g.O -mm bulb. The ends of the bulbs were packing to an equilibrium condition. The time ground off to leave holes about 2 to 3 mm in diameter . needed to reaeh equilibrium decr eases with increas Pieces of aluminum foil 118 IJ. in thickness were then ing particle size. Figure 4, from the data for the cemented over these holes. These simulated elec straight-tube standard in table 1, illustrates this trodes were filled with powder the foil thicknesses t.en~len cy . In addition., the scale readings at equi measured with the Magne-Gage and the readincrs l~bnum ?ec)'ease n~meI'l c any with increasing particle compared with that obtained fro~ the 118- IJ. straight SIze, as Illustrated 111 figure 5. As can be seen from tube standard. Scale readings of 141 141 and 142 the ~~li?ration curves shown in figures 2 and 3, the were obtained, the higher readincr for: the'15.5-mm sensItIVIty of measurement, as microns per scale bulb. This is of the order of the ~xperimental errors division , decreases with increasing scale reading. involved. Partially to keep this sensitivity as high as possible From th? very n~tur e of a powder, it seemed likely for any given thickness, but mainly to take advantacre that the ehal readmg obtamed on the Magne-Gage of the saving in vibration time involved, we sele c t~d would depend on the degree of compacting of the 20-IJ. iron powder (grade L) as standard. M easure powder, as \~e ll as on its particle size. Experiments ments were made after vibrating for l.5 minutes. were accordlllgly made on foul' of the five sizes The data, table 2, indicate the precision possible. av ai.lable. 4 Th e }~ were performed on the 118-1l ~1ea s urement s were made on two electrodes, using straI~ht -tub e thICkness standard and essentially magnet 21 filling 'wi th iron powder L , and vibrating duplIcated on a tYPIcal glass electrode. The powder for 1.5 mmutes. After every reacling, the electrode was held bulb up, still stoppered, vibrated for about 4 AvaiJ able from Autam DiVision, General Anilinr and Film C0rp. 30 seconds until the powder separated from the 445 ------. ---
T A B L E 2. M 1dti ple measw·ement.< of the thickness of two electrode tip and then held bulb down and repacked glass elecM'ode bulbs for 1.5 minutes before the next reading was taken. With ordinary care, scale readings can easily be Scale read ings reproduced to within one unit . From an examina tion of the calibration curves shown in figures 2 and 3, E lectrode 1 Elec trod e 2 it can be seen that one scale unit amounts to about 12l. 2 131. 0 2 to 3 percen t of the thickness m easured, for thick 120. 1 131. 0 nesses down to about 30 IJ- . For thinner membranes, 120. 1 130. i 120. I 130. 9 this percentage increases somewhat, but measure 120. I 130.2 120.0 131. 0 menL to 1 IJ- is easily obtained. J 2O. 2 131. 0 120. 1 130.9 121. 0 11 1. 0 2.2, Measurement of Electrode Response 120.5 1:11. 1 120. 2 ~ ------121. 0 ------E ach elec trode was prepared for voltage response measuremen t by soaking in distilled water over X=Av erage ...... 120. 4 n o. 9 I----!I----- night or longer. It was th en filled with mercury S=Standard dev ia tion, -J T- D 2/ N - L ...... 0. 43 0.26 [21], into which the measurement lead was dipped. Sx = Standard erro r of mean , SI-JN...... 12 . 08 Limits for true average (statistical proba bil· A Beckman model G pH meter was used to meaSUl'e Ity=O.99) ...... ±. 12 ±.09 the differences in poten tial developed in a series of Britton universal buffer solu tions [4] between the z 7 ,--,---,- -,--,-- ,---,--,--,---,---, experimental el ectrode and a well-conditioned com ~ mercial glass electrode (Beckman 1190) serving as a ~ reference electrode. The difference b etween the .~ 6 en measured poten tials for the buff er solutions of pH 4 :J and 8, divided by four, gave the departure of the ~ 5 w electrod e from the theoretical, in millivolts per pH :I: o uni t. T his departure subtracted from th e theoretical ~ 4 59 mv per pH unit, gave the electrode response of a: f2 th e experimental elec trode. W 3 ~ ~ z 3. Results Q 2 !;( 3 .1. Thickness Profile of a Glass Electrode a: CD > IL-__~ __~ __~ ___L-_~ __~ __~ ____L- __~ __~ It is obvious from the natur e of the glass blowing 2 4 6 8 10 12 14 16 18 20 22 process that a bulb blown at the end of a glass tube AVERAGE PARTICLE SIZE, MICRONS will be nonuniform in thickn ess. M an:v measure- FIG URE 4. Duration of vibration Tequil'ed to reach eq1!ilibrium state of packing, as a f unction of average particle size of the cMbonyl i ron powders.
2 8r---.----r---.----.---.----r---.----,---,---~
26
24 ~ :Ja: CD 22 :J 5 0 W 20 ~ z "0 18 ~ UJ II: w 16 -J ~ 0 12 10 2 4 6 8 10 12 14 16 18 20 22 AVERAGE PART ICLE SIZE, MICRONS FIGURE 6. Thickness profi le of a commercial glass electrode Fr GURE 5. Scale Teading after packin g to equilibrium as a Azimuthal equidistant projection , wit h bulb ti p ta ken as central point. T hick - function of average particle size of the carbonyl iron powders. nesses in microns. 446 - mont, of the Chickn'" of gl'" ,ket,-o", bulb" using 100 1 the powder-backed magnetic method, have shown 448