Phase Relations in Heat Transfer Salt Systems J . ALEX;ISI)ER, JR.. ~ S DS . (;. HINDI\ Hotcdr? Prowss Corporation
of
P ~ rnsylrcrriiu, i \IiLrcub Hook, Pu.
A phase diagram is presented w1iic.h gi\es in detail the initial freezing points of the system sodium and pots-iuni nitrites and nitrates, a portion of which represents the. heat transfer material used in Houdry fixed-bed catal? til. craching units. Other diagrams are giben which s h o ~the effects on the commercialIy used portion of this s.rstein of certain chemical reactions M hivh tend to take place : t i the
THE
success of the Houdry fixed-bed catalytic cracking process in the petroleum indust,ry has been due not only to the catalytic principles involved but also t o the solution of a number of engineering design problems. One of these design features ib the liquid heat-transfer system Ivhich is used to maintain esscntially con-taut temperature conditions in the cracking reactors. A total of 30,000,000 pounds of heat-transfer liquid is n o x employed for temperature control in these fixed-bed catalytic cracking units. Thii liquid is circulated through the reactors and HUSiliary heat exchange equipment. In the course of its circulation it maintains proper cracking temperatures in the reactors xhirh art' on-stream, removes heat from reactors which are bcing rt'generated, and dissipates the excess heat by preheating the oil i t w i , heating the regeneration air, arid pt,oducing the rteam necrssary For the uperation of the plant. The liquid used for this purposc is a eutectic niisture of molterr potassium nitrate and sodium nitrite. This mixture was c h o s t ~ because of it3 high heat capacit.?. (approsinlately 0.35 l3.t.u. per pound per ' F.', its low melting point (285" F.), its negligible corrosion on ordinary carbon steel a t temperatures up to approxiniately 1000" F., and its low cost (6-8 cents per pound). The chemical propwties of this mixture and the design of the equipment in which it is used arc sucli that it is estrcnie1)- stable a t its normal operating temperature in fixed-bed units-that is, bt tween 800-900" F. For example. in one typical unit, after 3.5 years of operation the freezing point of thr mixture had incrt only from 283' to 324"F. The salt systems have been designed to prevent, vithin practical limits, contact between air or other gases and the molten salt. However, under operating conditions certain chemical reart ions normally take place slowly which, if allowed to continue, ultimately affect the utility of these systems by raising the freezing points beyond practical limits, or by forming insoluble materials Fhich settle out in the systems. The survey described in thih paper was made primarily as an aid to the detection, evaluation, and control of these reactions. It was limited b>-practical considerations to a determination of the initial freezing points, which are defined here as t>hetemperatures a t which the first solid phase appears. This was done because it is the absence or presrncr of solid matter in these mixtures which determines their usefulness as heat transfer media. All data are given in 'F. becausc of the fact that the primary application of such data is in connection with commercially operated plants in. this country.
mixture ages. These reactions include the ovidation of nitrite ion to nitrate ion, and the formation of hydroxide and carbonate ions as a result of internal oxitlationreduction followed b y the absorption of water \apor and carhon dioxide. Processes for the commercial control of these reactions are discussed. The evperimental method6 used in the study are also described.
time. Freeziug point data for all four binary systrms containing a common ion, indicated numerically on the edges of Figure 1, are given in the International Critical Tables (Z), and numerous other references to the use of individual mixtures are contained in the literature ( 1 ) . FTnwcver, it b%asnot until 1938-1939 that full commercial use
v n'.KN02 024
PHYSICAL 4 - D CHEMICAL BEHAVIOR O F SYSTEV
The uae of mixtures of the nitrates and nitrites of sodium and potassium for heat-transfer purp0se.r has been known for yomc
Figure 1. Freezing Points of Alkali Nitrate-Nitrite Mixtures with Compositions Given in Weight Per Cent
1044
August 1947 "50
INDUSTRIAL AND ENGINEERING CHEMISTRY I
1
!
1045
I
I
1
J
KOL
L b Q W r A T E IUM
.5 APPIQI
0
'
' I I ,
i* 1 b R 3 0 h A i A S l a (0,
Figure 3. Effect of Carbonate Ion on Freezing Points of Alkali Nitrate-Nitrite hlixtures Letters refer to compositions in Figure 1.
was made of any of the minimum freezing point mixtures of t,hr four ions. By 1939-1940, a t least ten Houdry units were in operation, in which a commercial heat transfer salt containing 40% sodium nitrite, 7% sodium nitrate, and 5370 potassium nitrite manufactured by E. I. du Pont de Xemours & Company, Inc., was used ( I ) . It later became apparent that an even more profitable mixture of these compounds could be employed. This latter mixt,ure was established on the basis of the following data. An examination of the basic salt system covering all mixture# of the four ions, SaT, K + , NO,-, and SO,-, indicates that i t can he represented conveniently on a diamond-shaped graph, the four apexes of which represent, respectively, 10070sodium nit'rate. sodium nitrite, potassium nitrate, and potassium nitrite. This system is shown in Figure 1, which is plotted in terms of \\-eight percentages in order to facilitate its use in plant practice. Tht, configuration of the isotherms on this graph s h o w that the quaternary eutectic crosses the mid-line connecting 100% soduim nitrite and 10070 potassium nitrate a t a point, A , representing tht, composition 45% sodium nitrite and 55% potassium nitrate. It is this eutectic mixture which was chosen for use in all Houdry 5alt systems since 1941, because of its simple composition coupld with its unusually low melting point. In use, three chemical reactions take place in either of thest. commercial mixtures which tend to change their composition and physical properties. OXIDATIOSOF KITRITE. The nitrite ion is oxidized to nitrartj ion, usually through contact with air, according to the simplts equation:
2N02-
+ 02
+ 2S03.-
This reaction results in the formation of mixtures which are represented by points located on the line A E (Figure 1). Line F-3E represents all salts having a constant Na+-K+ mole ratio of 1.20. As the composition shifts in the direction of increasing sodium nitrate concentration, a correspondingly increasing freezing point occurs until a maximum of approximately 430' F. is reached a t the point where all nitrite has been converted to nitrate. The maximum allowable freezing point in a given plant is dependent on the pressure and degree of superheat of the steam available for use in the storage-tank heating coils. In most plants it has proved desirable to hold the fretzing point below 350" F. FORXITION OF HYDROXIDE l u x . Hydroxide ion is formed as a result of the internal oxidation-reduction of small amounts of nitrite ion with the simultaneous &sorption of water vapor.
The influence of quantities ~ r fhydroxide ion up to 5 molr 70 of the total anion content is to lower the freezing point by amounts up to 20" F. This is shown graphically in Figure 2, in which the effect of added hydroxide on four representative mixes of increasing nitrate ion content (indicated on Figure 1 as points A , B , C, and D ) is described. The limit of this freezing point lowering ie greater the higher the freezing point of the basic salt mixture. I s the hydroxide ion concentration continues to increase, the freezing point goes through a minimum and rises again to valuee higher than that of the original salt, with the hydroxide separating out BP a glassy precipitate at the temperature indicat,ed.
IOE
50 q31.
4DC
D ION
Figure 4. Comparison of Influences of Hydroxide and Carbonate Ions on Freezing Points of -4lkali NitrateWitrite \Iixtiires
The curves in Figure 2 are slightly high in the S-1570 hydroxide region because of the presence of about 170 carbonate in the alkali used for making up the experimental mixes. (The extent of this error can be estimated from Figures 7-10, in which the dotted lines represent the minimum amount of carbonate present for B given alkali concentration.) The weight % scales in Figures 2 and 3 are average scales. The exact conversion factors for mole % to weight % vary slightly with changing composition of the basic salt mixture. I n plant practice the concentration of hydroxide has never yet exceeded that corresponding to the minimum melting point. However, the presence of alkali undoubtedly increases the rate of
1046
INDUSTRIAL AND ENGINEERING CHEMISTRY
e Figure 5. Tetrahedron Representing Alkali NitrateYitrite 3lixtures Containing Hydroxide and Carbonate Ions Letters refer to compositions in Figure 1.
Na+-K' ratio
=
1.20
ahsorption of acid gases such as carbon dioxide as mentioned helox. FoRxiTIoN OF CARBOS.~TE 10s. Carbon dioxide, if present in the atmosphere above the molten salt, is absorbed readi1)- by the frce alkali, forming c.arbonates which are insoluble in the salt system. This ease of ahsorption of carbon diosidr was brought out in the course of the experimental work whPn it became apparent that reproducible freezing points could not I)(, obtained on salt mixtures because of the presence of carbonate, even when they were mixed and melted in a beaker, covered \vith a watch glass, into which a stream of nitrogen was passing. It was only.when this design was further modified and the molten salts were prepared in a glass-stoppercd flask, equipped ivith nitrogen inlet and outlet, that complete freedom from thr influence of extraneous carbon dioxide was reached. Figure 3, giving the effect of carbonate ion on misturcs -1, H , C, and D , indicates that amounts over 1 mole yo of the total anion content precipitate out below 600" F. as a finely divided white solid. Only 0.3 mole yo of carbonate ion is soluble at 400" F. Within the experimental error no eutectic is formed b+ tween the nitrate-nitrite mixture and the carbonates. Instead, the freezing point of the mixture remains constant until sufficient carbonate is present to separate out of the system a t higher temperatures. It is probable, however, that a eutectic mixture is actually formed. This can be deduced from the behavior of the binary system-,sodium nitrate-sodium carbonate, and potassium nitrate-potassium csrbonate, as given in the International Critical Tables ( 2 ) . Since there are sections of the salt systems which operate at temperatures of 550-600' F., it is advisable that carbonate concentrations be maintained below 0.5 mole yo so as avoid carbonate precipitation in quiet spots of the heat exchange equipment. ULTIhlilTE EFFECT O F HYDROXIDE .4ND CARBOK.ITE I O N S O S FREEZING POISTB. Figure 4 indicates roughly the limits of the influence of the two mentioned ions on heat transfer systems. This graph was constructed by extrapolating the experimental data toward the freezing point figures for 100% sodium hydroxidcpotassium hydroxide and 100% sodium carbonate-potassium carbonate ( E a + - K + mole ratio 1.20) obtained from the International Critical Tables.
Vol. 39, No. 8
Thew curves show that the effect of alkali o n the frcwing points of commercial heat transfer systems is never of great ronsequence, 5ince the masimum freezing point obtainable with any of the mistures tested is less than 450" F. The frrczing point of mixtures containing carbonate swiiis t o teadily, reaching a masimum a t the freezing point of the mixed carbonates. This family of curves, when plotted on semilogarithmic paper, given almost a straight, line relation. It is knonn that decoinpoeition of carbonate takes place at temperatures 1 w e r than this masimum. IIo\T-evcr, during our invcstigation the temperature of initial dtwmposition n-as never erceetleci, except for short periods of time. COMBIXED EFFECTO F REACTIOSS. A s all of the rc described trnd to take. p1ac.e simultaneously, the resulting f point systcsm is comples; it consists of six ions, Sa', ICS02~ , , KO:$-, O W , and CO:,--. Since the cation r:ttio remains cona': stant throughout tiirse changes, thr system can he ~,cprcw~nted a rrgular tetrahedron, one edge of which indicstcs the nitrat(,nitrite ratio espressed in terms of the freezing point of thc hydroside- and carbonate-free salt, 17-hile the opposite edgc connects OH- and 100 mole yo COa--. points representing 100 mole Figure 5 is a diagrammatic sketch of this tetrahedron resting on its cdgc 9 E . This edge corresponds to the line similarly represented in Figure 1. As all the information of practical interest is included in the section of this tetrahedron ADPR, this section ha.s been distorted and replotted on three-dimensional rectangular coordinatcs in Figure 6, in order to make it more legible. It is given in only enough detail to shon- the trends involved. Figure 7 represents four planes of this figwe, which give the same data in a form in n-hich they may be read quantitatively. I n thews graphs the weight yo scales are exact.
1.2
I 1
v
"
MOL.
' X
r
1
'
HYDROXIDE ION
'
'
'
I
'
15
Figure 6. Freezing Points of .illtali Nitrate-Xitrite 3Iixtures (:ontaining Hydroxide and Carbonate Ions for Section .4DPR in Figure S
Conclusions which inay be drawn from thew graphs are: (a)Very generally, the effects of added hydroxide and carbonate are independent of each other. ( b ) More specifically, l o a concentrations of carbonate have the effect of raising the freezing point of mixtures containing over 10% hydroside ion. High concentrations of carbonate cause precipitates to form well above the freezing points of mixtures containing only hydroside. EIowever, minimum freezing points are still present at successively higher hydroxide concentrations. ( c ) Low conccntrations of hydroxide loFer the initial freezing points of mixtures containing
August 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
I 1R I F (
I
' W TWITI
2
INITIL
1047
FR. PT. OF
3
Y
32PF.
._
PT. E, FIG. I ) '
5
1
I
6
UT.% MsOH
MOL.% OH.
0
1
2
3
Y
5
5
UT. % naOH
Figure 7.
UT.$ RaOH
Freezing Points of Alkali Nitrate-Nitrite Mixtures Containing Hydroxide and Carbonate Ions
small ainouiitb of carbonate but do riot materially affect the solubility curves of carbonate. High concentrations of hydroxide lower the solubility of the carbonate and tend to eliminate the constant temperature portions of the carbonate solubility curves. MIXTITIESCOXTAIXISGOVER 20% EXCESSKK02. As an interesting sideline to this freezing point investigation, our experiinental work has shown t,hat the majority of that portion of the basic salt system containing more than 207, excess potasaiuni nitrite (region of Figure 1 below the dotted line GI,) has the peculiar property of solidifying slowly as a transparent glass without ever undergoing a sharp transition from liquid to solid state. This phenomenon is present to varying extents in different portions of this area. For example, 011 the edges connecting sodium nitrite and potassium nitrite, and potassium nitrite and potassium nitrate, respectively, freezing points are obtainable, but odd viscosity effects are experienced in thrl tmiprrature region
above the freezing points. The center of the disturbance seems to be within the dotted elipse, and in this section it is virtually impossible to obtain tt freezing point by the method used i n this paper. PLANT CONTROL OF S A L T COMPOSITION
The commercial control of the reactions described in the previous section and the commercial reconditioning of salts which have deteriorated beyond the desirable point proved to be coniparatively simple. A brief summary of methods which are commercially usable for these purposes is as follows: METHODSFOR FREEZISG POINTREDUCTIOS. ( a ) The reduction of nitrate to nitritr ion ma? be accomplished by passing ti portion of the salt through an absorption tower in which it is brought into contact with hydrogen. At plant operating tempera-
1048
Vol. 39, No. 8
INDUSTRIAL AND ENGINEERING CHEMISTRY
tures, the hydrogen reacts with the nitrate according to the general equation:
NO*-
+ Hz
+
SOZ-
+ H20
Hydroxide ion may be formed as a by-product of this reaction to the extent of about 5% of the N02- produced.
0
EXPERIMENTAL WORK
The reagents used in all experimental work Lvere General Chemical Company reagent-grade materials. They were either substantially anhydrous or were corrected for their moisture contents. They were not' further purified, but contained no conflicting impurities, except for the hydroside mistures which con1,aincd about 1% carbonate. Temperatures to 580' F. were measured on 3-:nch-immcrsion 20-580" F. A.S.T.LI. thermometers; above 580" F., 20-760" F. thermometers n-ere used. These thermometers iyere calibrated relative to one another. Their self-correspondence was within
2' F. C 3 R I STIPPERS
T E S T TUBE-3222 HH.4. D.
CLASS J A P
INSULA710N
j
t
-
3
34 -v
Figure 8. .ipparatur? Used for Determination of Freezing Points of Alkali Kitrate-Nitrite RIixtures
( b ) Kitrite ion concentration can be increased by discarding a portion oi the mixture and replacing it Tvith a salt misture of 50,370 potassium nitrite-i9.570 sodium nitrite (indicated at point F on Figure 1). When this occurs the compoiition of the system moves back along the line A E (Figure 1) to the eutectic .4, and if enough of the nitrite-rich misture is added, the spnteni may even be described by points on the dotted line F A . Sormally a market is available for the salt discarded under these circumstances. ELI.\IIS.*.TIOS O F CARBOXATE A S D I I Y D R O X I D E . ( a ) Carbonate and hpdroxide have most commonly been eliminated by conversion to nitrate. This is accomplished by adding concentrated nitric acid T O the salt system, usuallv in a liquid-filled absorption tov-er. .Additional nitrate thus accumulated should then be reduced t o nitrite by one of the means previously mentioned. This rcdu~,rionto nitrite is not necessary when carbonate and hydroside ;$re converted to a mixture of nitrate and nitrite by reaction with gaseous nitrogen dioside. This latter method has not been applied commercially until recent,ly because nitrogen dioxide has not been available in large quantities. ( b ) Carbonate concentrations map be reduced to approximately 0.3 mole Yo COS-- by taking advantage of the low solubility of carbonate in heat transfer salt. This is accomplished by cooling a side st,ream of salt from the main salt system in a conventional exchanger, allowing the carbonate t o settle in a properly designed tank following the cooler and then withdrawing this precipitate. ( c ) Carbonate may also be eliminated from the salt system by precipitation with calcium nitrate and subsequent filtering In a separate unit. I n this method it is necessary to take into solution the total batch of salt being treated by the addition of approximately 10% water. Calcium nitrate is added t o the mixture, and the precipitated calcium carbonate is filtered out in a gravel filter. The r e t salt is returned to the system by injection in ?mall quantities into the main salt storage tank.
DETERlIINATIOX O F FREEZING POIST O F ~ I T R I T E - ~ I ' I ' l 7 I T E MIXES. The aceurat,ely weighed salts were transferred to a 500-ml. iodine flask fitted with a two-inlet ground-glass stopper which was used to maintain an atmosphere of dry nitrogen above the sample. An electric hot plate was then used to raise the sample temperature to 730" * 25' F. and to maintain it there for 5-10 minutes in order to remove the moisture present. The sample was then quickly transferred to the test tube used in the freezing point determination (Figure 8 ) , the tube \vas stoppered, and the sample was stirred a t a rate of two strokes per second. Temperature readings were taken every 30 seconds. The temperature dropped 12-14' per minute a t 400" F. and 6-8" p t r minute a t 300" F. At the freezing point a fine precipitate appeared in the solution, and within a minute or two the entire m a s clouded up. With the eutectic mixture (285' F.) the teinperLiture leveled off a t t,he freezing point and remained quite constant for 4-5 minutes; with more highly oxidized mixtures the change in slope of the cooling curve mas less abrupt. In all cases the mass became solid in a brief time. The freezing pnint was then determined from the cooling curve. It checked quite c l o d ? with the appearance of the first crystals in the solution. T o check, the test tube was warmed to 100" F. above the freezing point and then allowed to cool as before. Checks wete usually within 2-3' F. Theaccuracy of the freezing points given in the following table is probably better than *5' F. EFFECTOF ALKALIAND/OR CARBOXATE.For the survev of the effects of the addition of hydroside and carbonate, four basic nitrate-nitrite mixtures xere prepared which mere chosen to simulate heat transfer salts a t various stages of oxidation. The compositions of these mixtures and their freezing points are shown i n Figure 1 a t points A , B, C, and D, and are given in the following table: Composition. Freezing Point. F.
JIixture Mixture Mixture Mxture
A B C
D
285 323 367 405
KYOl 55.0 53.2 51.9 50.9
Yo by W t . Nay02
SZhNO1
45.0 29.7 17.9 8.3
0.0 17.1 30 2 40.8
The mole ratio of sodium to potasium i j essentia!ly constant in these mixtures and equals 1.20. .Iccurately weighoti amount3 of hydroxide and/or carbunate were added to each of the four basic mixes, hydroxide being adcleil as a sodium hydroxide-potassium hydroxide mixture and carbonate as a sodium carbonate-potassium carbonate mixture (Na'-K+ mole ratio 1.20). In preparing mixes containing both ions, the carbonate iinpurities in the alkali were taken into consideration. The resulting mixes were melted and transferred to t,he freezing point apparatus as described. A slightly different technique of measurement was required, however, because of the fact that no appreciable change in the slope of the cooling curve was noticed except with high hydroside concentrations. The point at which carbonate or small amounts of hydroxide produced a solid phase had to be noted visually with the temperatures
August 1947
I N D U S T R I A L AND E N G I N E E R I N G C H E M I S T R Y
dropping at 10-12" per minute. This visual observation was further hampered by the etching of the glass surfaces by the hot alkali. As a result, the data on these mixtures are accurate only to * l o o F.
1049
the Sun Oil Company, who contributed to the original standardization of the specific freezing point technique' described. LITERATURE C l T E D
(1) Kirst, W. E., Nagle, W. M., and Castner, J. B., Trans. A m . ACKNOWLEDGXZENT
The authors wish to acknowledge the help of Charles S. Pennington, who obtained the majority of the recent experimental data. They also wish to acknoxledge the work of I. W. Mills of
Inst. Chem. Engrs., 36, 371-94 (1940). (2) Washburn, E. IT,, et al., International Critical Tables, Vol. I11 (1928). PRESEVTED before the Diiision of Petroleum Cl~emlstry a t t h e 110th Meeting of the AMERICAY CHEMICAL SOCIETY, Chicago, I11
Solubility of 'Nitrogen and Methane in Sulfur Dioxide hl R . DEAS AND W . S. WALLS Phillips Petroleum Company, Bartlesville, O k l n . T h e solubilities of nitrogen and methane i n d f ~ dir oxide haie been determined experimentall> o\er a pressure range up to 515 pounds per square inch absolute and at temperatures of 83" and -23.7" F. The compositions of the equilibrium \apor phases at the higher pressures were also determined, and the iaporization equilibrium K constants were computed. It was found that the solubilities increased with temperature and were proportional to the partial pressure of the gas in accordance with Henry's law. The Kuenen absorption coefficients were also computed. The solubility of nitrogen was found to be considerabl) less than the value previously reported in the literature.
T
HE Kuenen absorption coefficients for nitrogen and oxygen in sulfur dioxide \\-ere reported by Dornte ( 1 ) . The data indicated the solubilities to be relatively large, and it was desircd t o make a comparison with the solubility of methane. Since no solubility values for methane or nitrogen in sulfur dioxide a t high pressure were found in the literature, the measurements reported here were made over a range of pressures u p to 515 pounds per square inch absolute. Several experiments, in which the gas partial pressure was 1 atmosphere, were carried out in order to obtain solubilities for comparison n-ith those computed from thc high pressure data.
HIGH PRESSURE APPARATPS AXD PRO(:EUURE
The high-pressure apparatus (Figure 1) consisted of twin steel cella, each of approsini:ttcly 240-cc. capacit!., mountrd on trunnions in a constant temperature bath. The citeel tubing fastcmed into the bottom of each cell lead to the pressure gagc :ind to a calibrated piston-type niercury pump. The pump \\-as constructed in such a manricr that the volume of mercury punipcd could be continuously measured to 0.1 cc. The cffectivc volumes )f the cells were variccl by introducing or \vithdran.ing inc'rcury with the pump. One cell was employed to contain the gas and sulfur dioxide, where they were brought into equilibrium by vigorous shaking of the cell. The other cell !vas cmplo>-ed in taking a sample of the vapor phase. The pressure was nieasured with a Rourdon tub