l \ [ E R C U R I C OXIDE BY GUY B . TAYLOR L W D GEORGE A . HULETT
The atomic weight of mercury as a t present accepted, is based on the analysis of the chloride and bromide, salts of quite similar character. Confirmation from other classes of compounds is desirable From our work on the dissociation of rIicrcziric oxidcl we have become convinced that this compound can be prepared in a high state of purity by the direct union of mercury and oxygen. Mercury can probably be prepared purer than any other single substance’ and electrolytic oxygen leaves little to be desired. I n the previous work the oxide was made by heating the two elements together in sealed tubes of about 2 0 0 cc capacity. From so small a volume of oxygen but little of the oxide could be produced a t each heating, and part of i t adhered so tightly to the walls of the tube, that scraping was necessary for its renioval, causing traces of silica t o occur in the product. It mas also difficult t o open the tubes with their high vacua without introducing bits of broken glass. Mfth the apparatus sketched in the accompanying figure (Fig. I ) we were able t o heat the two elements together and supply the oxygen under pressure as fast as it was used up. Tube A, capacity about 2 j 0 cc was half filled with mercury and placed in an electric furnace. The two traps T and ‘f’ contained mercury. n‘hen oxygen was compressed in the cylinder C, by pouring mercury into the separatory funnel, the gas bubbled through the mercury in T’ and into the reaction tube A. At the same time the mercury in trap T rose in the left arm, preventing pressure being put upon the other parts of the apparatus The oxygen cylinder C could be refilled a t any time by closing the cock of the funnel and drawing out the mercury from the bottom. The gas entered from the generator by bubbling through trap T and trap T’ ~~
~
’ Jour
P11y-s Chem , 17,j 6 j (1913) Hulett Phps Re\ , 33, 310 11911)
756
Guy B . Taylor and George A. Hulett
prevented the release of the pressure in the reaction tube A. At the beginning the air was pumped out through tube B and then this tube sealed off. The oxygen was generated electrolytically in a type of generator (E, E) that has given good results in this laboratory. It consisted of two cylindrical vessels, capacity about two liters each, connected a t the bottom by a short piece of wide glass tube with ground joints. The apparatus could not be made in one piece, but the joints were well ground and held tight without lubrication. Tubes for leading off hydrogen
U
9 C
J Fig.
I
and oxygen were connected a t the top of each vessel by means of ground connections made gas tight with a trace of marine glue. Platinum wires sealed into tubes just above the ground joints carried the platinum gauze electrodes. For electrolyte we used a 30 peicent solution of Kahlbaum’s best potassium hydroxide, kept permanently free from carbonate by an excess of barium hydroxide. The oxygen was passed over an electrically glowed platinum spiral, G. Moisture was removed by freezing out or passing over. phosphoric anhydride as indicated in sketch. The apparatus serves equally well for hydrogen. All parts of the apparatus were glass and only one stop-
N e r c u r i c Oxide
7 57
cock with which the oxygen came in contact was necessary. This was well ground and required only a trace of lubricant. At first we used syrupy phosphoric acid, but since this required frequent renewing, and i t was necessary to operate the apparatus continuously for many days, we later substituted a good quality grease. Chance of contamination of the oxide in the reaction tube from this source is too remote for consideration. Four times distilled mercury was heated in the reaction tube A a t 420 * I O ' C under an oxygen pressure of 2-3 atmospheres continuously for from five to seven days. Only t h a t portion of the oxide floating on the surface of the metal or falling loosely from the tube was taken while that adhering to the tube was rejected. The yield was IO-Ij grams HgO per run. The reaction is so extremely slow and the yield so. small that our supply of oxide has necessarily been limited. To emphasize this difficulty i t may be remarked in passing t h a t the reaction tube exploded on three separate occasions after the run had been going several days. The oxide was finally freed from all traces of uncombined metal by heating to 400' in a rapid stream of oxygen under a pressure of one atmosphere. The gas was taken from a cylinder furnished as pure oxygen by the S. S.White Dental Mfg. Co. and dried by sulphuric acid, soda lime, and calcium chloride. At 400' C the rate of decomposition of the oxide is insignificant and the vapor pressure of mercury something over two atmospheres, so we concluded that any mercury mechanically enclosed by the crystals would be expelled. The oxide was then bottled and kept in a desiccator until used.
Analysis of t h e Oxide The principle of the analysis consisted in reduction of the oxide by metallic iron and weighing the mercury as metal. The ground oxide was weighed in a hard glass tube about 2 0 0 mm long and 7.5 mm internal diameter, closed a t one end, This was thoroughly mixed in the tube with about twice its volume of finely divided metallic iron. This mixture was
Gzz~’B. Taylor a?zd George
758
A. Hztleti
covered with a layer of iron powder and the whole held in place by an asbestos plug. The tube was then constricted in two places, the air exhausted to a pressure of 0.1mm or less with a mercury pump, and sealed off a t the constriction nearer the pump. The decomposition was effected by putting the inverted tube in an electric heater with its empty end protruding below and remaining a t room temperature to condense the evolved mercury. The temperature was raised slowly and held a t 3jo-37j O for 2 or 3 hours, or until practically all the mercuric oxide had been reduced. Then the temperature was raised to about 600’ C and held there 2 3 t o 36 hours longer to insure corrplete decomposition. During the last half hour, the protruding end of the tube where the mercury was condensed mas cooled to o o with ice Whether there is enough iron in actual intimate contact with the oxide to effect its complete reduction is immaterial, since the iron oxide formed rapidly catalyzes the dissociation into mercury and oxygen, but the total iron present must be sufficient to take up all the oxygen liberated in order to prevent reoxidation. A t the end of the experiment the tube was cut open and the mercury transferred as a single globule to a porcelain crucible and weighed. The metal obtained in this way required no washing or drying, an operation involving some difficulties as has been pointed out by Perdue and Hulett’ and Easley.? The finely divided iron was obtained by reduction of Baker’s “analyzed” ferric oxide in pure, dry hydrogen. By the above method the following results were obtained. The weighings were made by the method of substitution, using a Ruprecht balance capable of giving an agreement of 0 . 0 2 mg on consecutive weights. This balance was kept in a room where temperature and humidity were under control. Ample time was always allowed for temperature differences to adjust themselves after each opening of the balance case. Carefully calibrated weights were used and vacuum correc-
~~~
1 2
_ _ ~ ~ Jour Phyq C h e m , 15, 147, I j j ~ 1 9 1 1 ) Jour Am Chem S O C ,32, 1 1 1 7 (1910)
JIerciiric- Oxide tions applied, 0.03 mg per gram HgO and Hg. ~
Sample
~~~
~
~~-
~
\vt
\Tt HgO
759 0.0jj
mg per gram
~-
Hg
Percint H g
92.604 92,601
9 2 . E06 92.605 92.606 92.603 92,609 92.602 92.612
Then for 0
=
General mean 92 6053 16 Hg = 200.37 * 0.o.j.
*
o oooS
The constancy of the results obtained is all that could be desired; the probable error of a single determination of the percent Hg in HgO is * 0 . 0 0 2 Fercent and of the average percentage * 0.0008 percent. X o w mercury oxide was weighed as the starting point and the oxygen determined by difference, not a standard method for an atomic-weight determination : nevertheless, the results are so consistent that the calculated atomic weight of mercury 2 ~ 0 . 3 7 (0= 16) shows a probable error of only * 0 . 0 2 5 . whichshould give us the first decimal place and a good idea of the second, provided there are no constant errors. The manipulation and method were tested by weighing pure mercury into a tube, adding the iron, evacuating, and proceeding as in the analysis. This would determine whether any mercury was relained by the iron or lost by faulty condensation. S o loss could be detected. The HgC) was prepared from direct union of oxygen and mercury, using only the purest materials with no contact with other substances. The oxide could include some mercury, but subsequent heating in a current of oxygen to 400' seemed t o entirely eliminate this source of error, and as the calculated atomic weight is smaller than the latest determination, this point seems to be
Gzq'
760
B. TajNlor and George il. Hulett
above criticism ; that the oxide was not completely decomposed is improbable for the heating in contact with the catalyzer, ferric oxide, a t 600' was always 24 hours and several of the determinations 36 hours, with the same results. The accepted atomic weight of mercury is based upon the excellent work of Easley' and of Easley and Brann.' These investigators worked with the chloride and bromide. From their published results we have calculated the following : Hg : HgCl? ( 6 determinations) 7 3 . 8 8 3 3 * o . o o o 7 g = 200.621
HgClz :
HgBrz :
2
2
-1gC1 ( 6 determinations) 9 4 . 7 1 7 7
AgBr
(I I
*
0.0023
200.617 determinations) 9 j . 969 7 * 0.0013 200.622
* o.coj8
*
=
0.0067 =
*
0.00j2
As far as the experimental results are concerned, this work should locate the atomic weight of mercury a t 200.62, with the second decimal place quite certain, provided there are no constant errors. The mercury was weighed direct and the chlorine determined as silver chloride, the results agreeing so closely that one is inclined to consider the methods without sensible error and the percentage of mercury in the mercuric chloride used as 73.8833 percent. If this salt was pure we are justified in the above value for the atomic weight and the bromide results give added confidence. Our result, 200.37, shows a marked deviation from this value but our method was so simple and direct that we place a great deal of confidence in 92.605 percent being the percentage of mercury in our oxide. In preparing the oxide everything was excluded but oxygen and mercury with no contact with other substances. Since there appears to be too little mercury, inclusion of the metal is not to be considered, nor oxygen as an impurity in solution in HgO. But if some higher oxide is capable of existence and is soluble in HgO, it might be expected to be present, for in the system oxygen and mercury in equilibrium with both elements in Jour. Am. Chem. Sac., 31, -Ibid., 34, 137 ( 1 9 1 2 ) .
1
1207
(1909); 32, 1 1 1 7 (1910).
M e r u t Tic Oxide
761
excess, every possible combination of Hg and 0 must have been present while the HgO was forming. Bredig and IVeinmayr' have noticed a periodic catalysis of the decomposition of hydrogen peroxide in solution by metallic mercury. Antropoff? has shown that this is due to the alternate formation and shearing away of a film of a mercury compound on the surface of the metal, which he has isolated and identified as HgOs. A very slight amount ( 0 13 percent) of such a peroxide in our preparation would account for our low value, and would be exceedingly difficult to detect if present as solid solution. The results obtained from this analysis of our mercury oxide seem to us to call for a further investigation of the atomic weight of mercury. Laboratory of Piigsicai C h e i u / s t r j , Princeton, October, 1913 1
Zeit phys Chem , 42, 601 (1903) Ibid , 6 2 , j 1 3 (1908): Jour prak'. Chem, 77, 2 7 3 (1908)
