Proceedings of the American Academy of Arts and Sciences.

Vou. 59, No. 5.— NOVEMBER, 1923.


By P. W. Brineman.

( Continued from page 3 of cover.)


Duranp, Ex1as J.— The Genera Midotis, lonomidotis and Cordierites. pp.1-18. 2 pls. September, 1923. $.80.

BaxTER, GREGORY PAUL AND Scott, ARTHUR FERDINAND.— A Revision of the Atomic Weight of Boron. The Analysis of Boron Trichloride and Boron Tribromide. pp. 19-48. September, 1923. $.80.

Lipka, Josern.— Trajectory Surfaces and a Generalization of the Principal Directions in any Space. pp. 49-77. September, 1923. $1.00,

Pierce, GeorGe W.— Piezoelectric Crystal Resonators and Crystal Oscillators Applied to the Precision Calibration of Wavemeters. pp. 79-106. October, 1923. $1.00.

BripcmMan, P. W.— The Compressibility and Pressure Coefficient of Resistance of Rhodium and Iridium. pp. 107-115. November, 1923. $.50.

es : Men: Ne Po Nps CNIS ikea ae Oy Si, itn tie Ri tes Sy

bate Popenihsce SS es

i Ph

Proceedings of the American Academy of Arts and Sciences.

VoL. 59, No. 5.—NOvEMBER, 1923.


By P. W. BripGMman.


By P. W. BripGMAN.

Received Oct. 6, 1923. Presented Oct. 10, 1923.


THE data of this paper are supplementary to determinations of compressibility and pressure coefficient of resistance which I have previously given for a number of substances.! It is my intention to keep this work as up to date as possible by making new determinations either for new materials or else for materials of greater purity than have hitherto been available. I have not hitherto made any measure- ments on either rhodium or iridium, and I cannot find any records of others, so that these results are quite new.

The two metals were obtained from Johnson and Mathey, London. They were provided in the form of round wires approximately 10 cm. long and 0.08 cm. in diameter. They were shaped by swaging at a red heat, it not being possible to draw them, and therefore they were not perfectly uniform in diameter. The extreme variation from the average diameter of the rhodium was 6.5% and of the iridium 3.2%. Departures from geometrical perfection are entirely without effect on the measurements of pressure coefficient of resistance or of compressi- bility, but if the departures are too large, the accuracy of the specific resistance, which was determined incidentally, may be somewhat affected. For these two wires, however, the departure from uni- formity was so small that no error in the specific resistance is to be feared.

I have no chemical analysis of the materials, but some idea of the probable purity may be obtained from the temperature coefficient of resistance at atmospheric pressure. That of the rhodium is high, and this material is probably as pure as any on which previous measure- ments (specific resistance and temperature coefficient of resistance) have been published. The temperature coefficient of the iridium, on the other hand, is low, and its purity is probably not as high as might possibly be obtained.

The methods of measuring both pressure coefficient of resistance and compressibility were the same in all respects as those used on other materials, and which have been previously described. The


same apparatus was used, and the usual pressure range of 12000 kg/em?. Resistance was measured by the potentiometer method adapted to specimens of small resistance (the resistance of these speci- mens was of the order of 0.01 ohm). The current leads and the potential leads were small copper wires soft soldered to the specimen. For the compressibility measurements the specimens were mounted as tension specimens in the lever apparatus for long specimens. Be- fore measurement, the wires were annealed to nearly a white heat in a bunsen flame. The results follow.


Compressibility. The compressibility was measured at 30° and 75°. The points obtained were entirely regular, and all were used in the computation without discarding any. The results are expressed in the formulas:

> AV : F , At 30°, | an —10-7 [3.72 2.67 K 10 p] p, _, AV At 75 TV, = —10-’ [3.81 2.67 K 10 p] p.

At 30° the average deviation of the 14 observed points from a smooth curve was 0.28% of the maximum observed effect, and at 75° 0.19%. These deviations are on the measured difference of compressibility between rhodium and iron; the corresponding accuracy of the actual compressibility (assuming no error in the value used for iron) is about three times greater.

The compressibility is very nearly the same as that of molybdenum It is to be noticed that the departure from linearity, shown by the second degree term in the formula above, is greater than it is for iron. This is not usual, and is the first example I have found for a metal, it being the usual rule that the departure from linearity is greater for those metals with the greater compressibility.

Resistance. The electrical resistance was measured as a function of pressure at 30° and 65°. An attempt at 95° did not give good results. The dimensions of the specimen were not as advantageous to accurate measurement of the resistance as of the compressibility, and the results obtained were not as good. The results are expressed in the following formulas, which give the fractional change of resistance at atmospheric pressure and the temperature in question as a function of pressure:


- AR At 30°, = —10-* [1.738 9.7 K 10> p] p Ro,30° AR At 65°, —— = —10-6 [1.776 10.1 X 10-5 p] p. Ro,65°

At 30° the average departure of the observed points from a smooth curve, (no discards) was 0.60%, and at 65°, 0.95% of the maximum effect.

At atmospheric pressure the specific resistance was calculated from the actual resistance and the dimensions and was found to be 4.90 X 10-§ at 0°. This is somewhat higher than 4.70 X 10-°, the value found by Broniewski and Hackspill.?

Between and 100° the resistance changes linearly with tempera- ture at atmospheric pressure, and the mean coefficient over this range, in terms of the resistance at 0°, is 0.00399. From the values of Bron- iewski and Hackspill for the specific resistance I calculate for their temperature coefficient 0.00404. The agreement is within the limit of error indicated by their significant figures.

The pressure coefficient of resistance is of the order of 10% lower than for the other two metals of this class previously measured, platinum and palladium, and the temperature coefficient is slightly higher than for the best platinum, indicating high purity.


Compressibility. The compressibility was measured at 30° and 75°. The results are reproduced in the formulas:

7 Vo

At 75° ll = —10-7 [2.81 2.2 X 10-° p] p 4 e 3 Vo a ee .

At 30° = —10-’ [2.68 1.3 K 10-5 p] p,

The average deviation of the observed points from a smooth line (no discards) was 0.094% of the maximum effect at 30°, and 0.16% at 75°. This deviation is on the difference of compressibility between iridium and iron; the error on the absolute compressibility is about one half of this.

The departure from linearity is here less than it is for iron, which is as to be expected. The variation with temperature of the second degree term is unusually high, as is also the temperature coefficient of the initial compressibility. It will be found on making the computa-


tion that the total change of volume between atmospheric pressure and 12000 kg. is only 1% greater at 75° than at 30°. Possibly the formula gives a temperature variation of both first and second degree terms larger than it should be, one compensating the other.

The absolute compressibility of iridium is smaller than any metal hitherto measured; the lowest value hitherto found was for swaged tungsten, whose absolute compressibility at 30° was 2.93 X 10-7.

Resistance. The electrical resistance was measured as a function of pressure to 12000 kg/cm? at 30°, 65°, and 95°. The results are repro- duced by the formulas:

_, AR . , At 30°, gg 10-* [1.353 4.0 K 10-* pl p, _ AR At 65°, ——— = —10-® [1.280 3.9 X 10-* p] p, Ro,65° _, AR At 95°, ee 10-6 [1.340 3.9 X 10-* p] p. 0;95°

At 30° the deviation from a smooth curve of the observed points (no discards) was 0.8% of the maximum effect, at 65° it was 1.1%, and at 95° 2.1%. The formulas show that the initial coefficient has a mini- mum value at 65°. This is unusual, but other examples have been found. The temperature variation seems to be outside the limits allowed by the probable error, and is probably real for this material. Whether it would be found for absolutely pure iridium is another question.

The pressure coefficient of resistance is materially dower than that for rhodium, as would be expected from its smaller compressibility and higher melting point, and is close to that previously found for tungsten.

The specific resistance at atmospheric pressure, found from the measured resistance and the dimensions, was 6.61 * 10-® ohms per em. cube, considerably higher than 6.10 X 10-®, the value of Bro- niewski and Hackspill.?

The temperature coefficient of resistance at atmospheric pressure between and 100° I found to be 0.00322. The relation between temperature and resistance is not quite linear, but the average coeffi- cient between and 50° is 0.8% less than between and 100°. This is the normal direction for departure from linearity, and is the reverse of the direction shown by platinum. I calculate from the values of Broniewski and Hackspill for the specific resistance that the tempera- ture coefficient of their iridium between and 100° was 0.00361.


This is too much above my value to be accounted for by errors of measurements, and doubtless indicates a perceptible amount of impurity in my iridium. The effect of this impurity on the com- pressibility is probably very small, on the pressure coefficient of re- sistance somewhat larger, and on the temperature coefficient of resistance largest of all, bringing the value from 0.0032 to somewhere around 0.0040. It is to be anticipated, judging from experience with other metals, that the compressibility of pure iridium will be slightly less than the value given above, and the pressure coefficient of resist- ance somewhat greater numerically.


A comparative study of the properties of the three series of chemi- cally related elements Fe, Co, Ni; Ru, Rh, Pd; and Os, Ir, Pt is of interest. In Table I are collected various data for these elements, namely atomic weight, melting point, atomic volume, compressibility at 30°, and pressure coefficient of electrical resistance at 30°. The two latter data are missing for Os and Ru. The table discloses irregu- larities in the progression of the properties; what the exact significance of these irregularities is I shall not attempt to discuss. In the first place the order of atomic weights in the first series is not the same as that of the atomic numbers, as is well known, there being an inversion between Co and Ni. This inversion is not found in the other two series, but it will be noticed that in the second series the difference of atomic weights between Ru and Rh is much less than between Rh and Pd, whereas in the third series the atomic weights are almost exactly evenly spaced. There is here evidently some sort of progressive change as the outer electron structure of the atom becomes more com- plicated. In each series the melting point progresses regularly, being least for the member with greatest atomic number. The atomic volumes, on the other hand, do not progress regularly, but there is a reversal in passing from the light to the heavy series; in the first series the lightest member has the greatest volume, and in the third the heaviest, the order being irregular in the second series. The com- pressibility also shows no regular order, but does follow the order of atomic volumes as far as the data are known, the element with the greatest atomic volume in any series also having the greatest com- pressibility, which is what would be expected. This would lead us to expect that the compressibility of Os would be even smaller than that of Ir, and the smallest known for a metal. It is even not unlikely that



Compressibility at 30°

Pressure coefficient of resistance

87 X 1077

5.39X 107

Fe Co Ni Atomic weight 55.84 58 .97 58 .68 Melting point 1530° C 1480 1452 Atomic volume 1 6.7


Compressibility at 30°

Pressure coefficient of resistance at 30°

2.68 10~"


at 30° —2.42X10- |-9.3 x1077 |—1.90x107 Ru Rh Pd Atomic weight 101.7 102.9 106.7 Melting point 2450? 1950 1550 Atomic volume 9.0 9.2 Compressibility at 30° 3.72107 5. 19 10-7 2.28 Pressure coefficient of resistance at 30° —1.74x10°* |—1.96x 107 Os Ir Pt Atomic weight 190.9 193.1 195.2 Melting point 2700 2350 1755 Atomic volume 8.5 9.2

3.60 ~7 eo “—


it may be less than for diamond.

The pressure coefficient of resistance

in each of the series decreases numerically from the third to the second member of the series. In the Fe, Co, Ni series the coefficient increases again on passing from Co to Fe; the data have not yet been obtained

to show whether the behavior in the other two series will be the same.



Measurements by previous methods of compressibility and pressure coefficient of electrical resistance have been extended to rhodium and iridium. The compressibility of rhodium is about that of molyb- denum, and that of iridium is somewhat less than that of tungsten, and is the lowest yet measured for a metal. The pressure coefficient of resistance of rhodium is of the order of that of platinum and palladium, and that of iridium is close to that of tungsten. Measurements are also given of the specific resistance at and the mean temperature coefficient of resistance between and 100° at atmospheric pressure. In the discussion attention is called to certain irregularities in the progression of properties in the three chemically related series Fe, Co, Ni; Ru, Rh, Pd; Os, Ir, Pt.

I am indebted to my assistant Mr. I. M. Kerney for making many of the readings.

THE JEFFERSON PuysicaL LABORATORY, Harvard University, Cambridge, Mass.


1P. W. Bridgman, Proc. Amer. Acad. 52, No. 9, 1917; 56, No. 3, 1921; 58, No. 4, 1923; 68, No. 5, 1923. 2 Broniewski et Hackspill, C.R. 163, 814-816, 1911.