1104
ANALYTICAL CHEMISTRY
photometric end point is reached and the reproducibility indicates that the method could be used for the spectrophotometric titration of thallium(1) by the technique adopted by Sweetser and Bricker ( 1 5 ) for the titration of magnesium, calcium, etc. COULOMETRIC OXIDATION OF THALLIUM
where D is the diffusion coefficient in square centimeters per second, d is the thickness of the diffusion layer, A is the electrode area, and V is the volume in milliliters. As the plot of log current against time did not give a straight line, the reaction is doubtless complex. MacNevin and Baker (9) report similar findings for the primary coulometric oxidation of arsenic( 111).
Between 0.3 and 1.0 meq. of thallium may be estimated by primary coulometric analysis a t a platinum anode in an acid solution with an accuracy within 0.3%. The procedure and apparatus are the same as those employed by RlacSevin and Baker ( 9 ) , except that the electrolyte ( l M sulfuric acid) is scavenged a t an anode potential of f1.38 volts us. the S.C.E. for 15 minutes. The anode potential is reduced to +1.34 volts and the thallium(1) solution is added. The oxidation is considered to be over when the current falls to 0.2 ma. The primary coulometric oxidation of thallium has more theoretical than practical interest. The anodic oxidation of thallium(1) did not obey the equation proposed by Lingane (8) for single reactions occurring pr-ith 100% efficiency: i t = &-kt
LITERATURE CITED
Besson, J., Compt. rend. 224, 1226 (1947). Flsschka, H., Mikrochemie 4 0 , 42-5 (1952). Gallo, G., Cenni, G., Gazz. chim. ital. 39, 285-96 (1908). Gutbier, h.,Dieterle, W., 2. Elektrochem. 29, 457-67 (1923). Haar, K. ter, Bazen, J., A n a l . Chim. Acta 10, 23-8 (1954). Heiberg, XI. E., 2 . arwrg. Chem. 35, 347 (1903). Lamphere, R. W., ASAL. CHEY.23, 258-60 (1951). Lingane, J. J., J . Am. Chem. S O C .67, 1916-22 (1945). JIacSevin. W. AI., Baker, B. B., ANAL.CHEM.24,986-9 (1952). JIartens, R. I., Githens, R. E.. Ibid., 2 4 , 991-3 (1952). Sorwitz. G., A n a l . Chim. Acta 5 , 518-20 (1951). Peitier, S., Duval, C., Ibid., 2 , 210-17 (1948). Pribil, R., Chem. Listy 4 5 , 85-7 (1951). Pribil, R., Zabronsky, Z., Collection Czechoslou. Chem. Comrnuns. 16, 237-9 (1951).
(1)
Sweetser, P. E., Bricker, C. E., AXAL.CHEM.26, 195-8 (1954). Taentnershver, AI., Trebackaiewicz, T., 2. physik. Chem. A165
where io is the current a t zero time, i, is the current a t time t, and 1, is given approximately by
367-71 (1933). R E C E I V Efor D review November 2 6 , 1955. Accepted April 7, 1956. Work supported b y the Defence Research Board of Canada Grant 7510-19. Project D 44-75-10-19.
Analysis of Cyclopentadiene Dimer Concentrates Review of Existing Methods and Description of Mass Spectrometer Method E. B. CLAIBORNE, H. M. DAVIS, and C. A. RIVET,
JR.'
Fsso Standard Oil Co., Baton Rouge, La.
Existing methods for cyclopentadiene analysis hale been retiewed. A new anal3 tical method utilizing the mass spectrometer is described in w-hich cyclopentadiene dimer concentrates are depolymerized in a specially constructed craclting chamber attached to the inlet system of the mass spectrometer. The method is accurate, reproducible, rapid, and ideal for plant and product quality control in a cyclopentadiene recovery process.
I
N A process for the recovery and purification of cyclopenta-
diene, complex mixtures of the dimers, codimers, and higher polymers of cyclopentadiene and its homologs and acyclic pentudienes, such as piperylene and isoprene, are encountered. Beraiise cyclopentadiene itself is the component of most commerci 11 importance, it was necessary to obtain an accurate, reproduchle method for determining cyclopentadiene in the presence of the less vuluable components. This paper contains a brief discussion of several analytical methods which had been used and 3 mole detailed description of a mass spectrometer method whicah ha6 been developed and is currently in use. A true analysis of such dime1 Concentrates would show the total quantity of the various compounds that are actually piesent However, the basic information desired is the quantity of the individual monomers, particularly cyclopentadiene, which could he recoverable by some depolymerization pi ocess The degree of depolymerization of a dimer concentrate is dependent upon temperature, residence time, and pressure (vapor phase). It is also a function of the types of dimers present in the concentrate itself-for example, cyclopentadiene dimer is more easily cracked 1
Present address, Creole Petroleum Corp.. rlmuay Bay. Venezuela.
to monomer than is the codimer of cyclopentadiene and isoprene. Also, because it is difficult to crack cyclodiene dimers without some degradation, any method of analysis which is based on depolymerization can represent only the quantity of monomer3 uvail:thle under the specific cracking conditions employed. Theoretically, a t least, the cracking conditions that produce the yrentest amount of depolymerization with the least' amount of degradation should permit the most accurate determination of :ivuilalile cyclopent,adiene. Incomplete or inconsistent depolymerization has been primarily responsible for the difficultied encountered with most of the analytical methods that have been tried. PREPARATION OF AYALYTICAL ST.ASD.IRDS
Pure dimers of cylcopentadiene, methylcyclopentadiene, and diniethylcyclopentadiene were prepared in the laboratory by conventional methods (2, 6). The final products were vacuum distilled to remove traces of low boiling and high boiling impurities. l l a s s spectrometer analyses of heart cuts from these distillations showed less than lyOof the adjacent homologs arid, for calibrat,ion purposes, the dimers were considered t,o be essentially 100% pure. The crystalline dicyclopentadiene which was obtained compared favorably wit,h the same material isolated by Edson, Powell, and Fisher ( 2 ) . Qualitative mass spectrometer analysis of the cracked products from plant dimer concentrates indicated the presence of significant concentrations of thermally stable codimers of cyclic pentadienes and acyclic pentadienes; therefore, it was essential to obtain these codimers for calibration purposes. Codimer concentrates of approximately 90% purity were isolated from the dimer concentrate by vacuum distillation and were also synthesized from pure isoprene, cyclopentadiene, and methylcyclopentadiene. Dipen-
V O L U M E 28, NO. 7, J U L Y 1 9 5 6 tcnc, mother known impurity, was obtained from I-dr, rwpcc:tiveIy. Xo attempt wa.3 made to apply tlirir metho11 t o the an;ilysiP of high purity ~-oncei!trates. The exhaustive tube cracking mrthod consist; of cracking ilie dimer concentr:itt~in the vapor phase in a hot tithr a t 700" 1;. to a still pot, n-here they are fractionated 11s containing \\-hat is assumed to he looc; cyclopentndicrie and methylcyrlopentadiene, respectivrly. The material is exhaustively cracnked by cwntinuouply recj-cling the still bottoms through the cracking tubes. This method is l~nsicallyan assay type procedure a n d \vas used before t h r ulti,aviolet and maps spectrometer methods \yere applied. The ultraviolet method n-hich was used in this 1:il)oratoi.y is that of Po~velland Edson (3)and is based on tlie determination of only cyclopentadiene and methyl(~yc~Iopenta(1i~~~ie. An isoocztane soliition of the dicyclopentadiene concwitrate ic cwcked a t ;355" to 3G0C. in a hot wire cracker, followed ljj- n!wsiirement of the ultraviolet absorption of a highly diluted portion of the cracked products a t 240 and 258 nip. The concrntration of tlie two components is determinrd f r o m the solution of tn-o siinultaneous equations. The ultra-giolet method as described by Powell and Etlson is natui~:~lly limited to samples in which no cyclodienes other than ryclopentadiene and methylcyclopentadiene are present. .4n o include the C7 homolog in a three-component' , but no routine use was ever made of this pi,ocedure. iiromatics and other materials that are not changed when passed over the cracker do not interfere, as their absorption ip vorrected for by use of a blank. Any codimers of cyclopentaclime with ultraviolet absorbing materials such as isoprene m-hich c.r;iixk a t 355" C. may cause high results. On the other hand, the pi'esence of an appreciable quantity of these eodimers may yicxld low results due to their failure to crack completely and mxke the cyclopentadiene available for measurement. The dimei,s and codimers themselves do not appreciably absorb in the ultraviolet, region. Powell and Edson (3)claim an accuracy within about =k0.8Cc on each component for the ultraviolet method. .4nalysis of two known mixtures of the pure dimers prepared in this laboratory ~ the actual perehorved a maximum deviation of 4 ~ 1 . 9 7from centage. The data are shown in Table I. .4 precision study made in this laboratory with five different analyses has shown a maximum deviation from the mean of +2.3c: for either component and for the total cyclodiene. The standard deviation (g) was found to be 1.20 for cyclopentadiene, 1.05 for methylcyclopentadiene, and 1.14 for the total (Table 11). For samples containing a higher proportion of methylcyclopentadiene, the deviation from the average will probably be greater
Analysis of Know-n .\fixtures by ultraviolet .4bsorption SIethod '&-eight,
Component Cyclopentadiene dimer llethylcyclopentadiene dimer Tutal
types:
Cyclopentadirnr dimer ~Ieth3-lcyclopentadiene dimer Total
'lable 11.
Operator 1
('.?lopentadiene, \Vt. C' 76.7 73 6
..-
13.3 100.0
84 3 13.7
85 3
i o d
1" ? 100 0
\Iethylcyclopentadiene. \t-t. vo
Total Purity, 1Vt. .; '
12 0 15.4 14 1
88.7 89.0 89.7 851.4 88.9 91.7 91.6. 80 2 91 8
0
73 9
14.3
-,.3.L' -
- .I-
76.7 I
.Irerage c'td. drv.
86.7
15.2 100.0
~
13.9 13.4 13.7 13 ;i 14 4 14 6 13. D 1.5 8 1.5 8 15 4 14.2 14 8
13.5
73.3 78.0 i8 1 74 8 77.2 i5.4 i f ;3 74.9
J
.inalysis
84.8
Cltra\iolet inalysis of Dicyclopentadiene Concentrate
75
2 3 4
Synthesis
1.20
1.05
:,;
90 7 !)0. G
90.9 89.8 90.4 1 14 .-
for the methylryclopc~itadie~ic fig111~1, since the a h r p t i o r i of the methylcyclopcntadiene is mrasiired on the sick of tlici absorption band. The presence of the higher homolog (dirncthylcyclopentadirnc) will also interfere Ivith the analysis. Because of the difficulty encountered in complete cracking of all dimers containing cyclopentadiene and the insufficient rrproc1iicil)ility of the iilt,raviolPt method, the mass spectrometer method was cleveloped. XIASS SPECTROMETER METHOD
I n the original form of this method, the dimers were cixc*kccl in the presence of a diluent in a hot \Tire cracker a t 300" t o 400" C. (same a s the ultraviolet method) arid the cracked products were charged to the mass spectrometer. The diluent, usually isooctane, acted as an internal standard. Some inherent disadvantages of this method are reduced sensitivity because of large escesa of inert diluent and nec~=ssityfor maintaining the cracked pi,oducts a t dry ice temperature t o prevent redimerization. The next modification consisted of heating the entirc inlet system of the mass spectrometer to -400' C. The dimer samples were charged to the instrument x i t h an ordinary capillary dipper and were cracked directly in the inlet, system. There were several uncontrollable "cold spots'' in this system, which probably resulted in incomplete or inronsistent cracking and cont1ens;ition of heavy components. Thrre were also uncontrollable ',hot spots," xhich resulted in frequent overheating and collapsing of tlie glass inlet system. In general, reproducibility by this method was not very good. Dimer samples are no\\- cracked in an external low prcssurr borosilicate glass cracking chamber which is connected directly to the inlet system of the mass spectrometer. A diagram of the cracker is shown in Figure 1. I n operation the cracker is evacuated to normal inlct system mm.). .4 magnetically operated gallium valve pressure ( located at the exit of the cracker is closed, and t'he sample is introduced through a heated mercury orifice. After 2 to 5 minutes at 350' to 450" C., t,he gallium valve is opened and the craeked products are expanded into the main inlet system of the mass spectrometer. The inlet system is constructed so the cracker can be evacuated and charged with a second sample xhile t h e first sample is being analyzed.
1106
ANALYTICAL CHEMISTRY
Both calibration compounds and unknown samples are charged to the cracker on a constant volume basis; therefore, the initial mass spectra which are obtained are expressed in terms of peak height (divisions) per unit volume of material charged. Calibration data have been converted to a basis of peak height per unit weight by dividing the mass spectra of the pure compounds by their respective densities and are thereby made independent of the type (monomer or dimer) of calibration compound used. The cracking procedure now used gives complete control of cracking temperature and residence time and has materially improved reproducibility. Complete depolymerization of all of the dimers and codimers present is not accomplished even a t the low pressure and high temperature used in the cracking step. This was evident from the presence of prominent parent dimer peaks in the mass spectrum of the cracker effluent. The peaks ( m / e )used to determine the individual components are as follows: cyclopentadiene, 65 and 66; isoprene, 68; methylcyclopentadiene, 79 and 80; dimethylcyclopentadiene, 94; cyclopentadiene-acyclic codimer, 134; dipentene, 136; and methylcyclopentadiene-acyclic codimer, 148. The mass spectrum of the unknown sample is converted to a basis of peak height per unit weight by dividing the absolute peak heights by the density of the sample. A sample calculation is presented in Table 111. The heaviest component present is calculated first. For example, the peak height-unit weight of sample a t 149 mass divided by the peak height-unit weight of pure codimer a t 148 mass represents weight per cent of 148 mass codimer present in the cracked products from the unknown sample. The other components, in order of decreasing molecular weight, are calculated in a like manner after the individual peaks have been corrected for contributions from heavier components. The calculated values for cyclopentadiene a t 65 and 66 peaks normally check within lye, and an average of the two figures is used. The agreement between the check values for cyclopentadiene and methylcyclopentadiene is taken as evidence of proper calibrations for the heavier components. The concentration of uncracked cyclic-acyclic codimers indicated by the height of the 134 and 148 peaks is very sensitive
Table 111. Component Cyclopentadiene Isoprene Methylcyclopentadiene Dimethylcgclopentadiene 134 mass codimer Dipentene 148 mass codimer 5
Sample Calculations
Peak Height Divisions ConOriginal verteda 861.00 1920.00
Mass Peak 65 66 68 79 80
Calculated Weight, % NormalNatural iaed 74.9 5.9
68.8 427.0 246.0
94 _.
17.3
5.0 1.6 1.1 0.5
0.62 0.6 134 0.27 0.3 136 0.94 0.9 148 0.14 0.1 100.4 100.0 Original spectrum divided by density of sample, 0.965 gram per cc.
Table IV. Component, Wt. % Cyclopentadiene Methylcyclopentadiene Dimethyloyclopentadiene Isoprene 134 massoodimer Dipentene (136 mass) 148mass codimer
5.2 1.7 1.1 0.5
- -
Effect of Cracking Severity on Component Balance Cracking Conditions 350' C. 400' C. 45OO.C. 450' C. 450' C. 400' C. Std. 2 min. 2 min. 2 min. 5 min. 15 min. 15 min. Dev." 73.7 74.2 7 4 . 5 76.1 76.3 75.7 0.20 15.0
15.1
15.5
15.7
15.9
15.90.10
0.9 2.9 5.4
3.5 4.4
0.9
0.9 4.8 2.7
0.8 5.8 0.5
0.8 6.1 0.1
0 . 8 0.01 5.70.20 0 . 7 0.20
0.8 1.3 100.0 450" C., 5 minutes.
0.8 0.9 1.1 0.7 1 0 0 . 0 100.0
0.9 0.8 1 . 0 0.30 0.2 0.0 0 . 2 0.05 100.0 100.0 100.0
OT ; = - -
INLET SYSTEM
OF
(MOLTEN GALLIUM)
)I!
A
.
~-
II
Figure 1. Low pressure cracker for mass spectrometer analysis of cyclopentadiene dimer concentrates Entire cracking chamber, from mercury orifice
to gallium valve, is wrapped with heater wire
and insulated
to changes in cracking severity. The data in Table IV show that as the cracking severity is increased from 2-minute residence time a t 350' C. to 15-minute at 450' C., the amount of uncracked 134 mass codimer decreases from 5.4 weight 76 to 0.1 weight % and that this decrease is accompanied by proport,ionate increases in the concentrations of cyclopentadiene and acyclic pentadienes (calculated as isoprene). I t is concluded from this that the codimers which are being used for calibration purposes are representative of the thermally stable codimers which actually exist in the dimer concentrates and that any cyclopentadiene combined in such codimers is theoretically recoverable. I t should be remembered, however, t.hat the cracking conditions used in this test approach the ideal and that the actual amount of cyclopentadiene monomer which can be recovered in a commercial process might be considerably lower because of degradation. The analyses in Table IV have been normalized, as it, is believed that essentially all components (99% or better) have been accounted for. The unnormalized totals range from 95 to 105% and on the average are close t,o loo'%. Five minutes' residence time a t 450' C. has been found to represent the practical limit of cracking severity for this method. However, from an analytical standpoint, the figures obtained a t low cracking severity are just as good as the figures at high severity. Reproducibility and component balance in both cases are very good. -4lthough this work represents only one application of combined low pressure cracking and mass spectral analysis, modification of the equipment and technique to reach higher temperatures should permit the study of other depolymerization or pyrolysis reactions. LITERATURE CITED
(1) Bergman, F., Jophe, H., ANAL.CHEM.20, 146 (1948). (2) Edson, K.C . , Powell, J. S., Fisher, E. L., Ind. Eng. Chem. 40, 1526 (1948). (3) Powell, J. S., Edson, K. C., ANAL.CHEM.20, 510 (1948). (4) Powell, J. S., Edson, K. C., Fisher, E. L., Ibid., 20, 213 (1948). (5) Uhrig, K., Lynch, E., Becker, H. C.. IND.ENG. CHEM.,ANAL ED. 18, 551 (1948). (6) Wilson, P . J., Wells, J. H., Chem. Reus. 34, 1-50 (1944). R ~ C E I V Efor D review January 3, 1956. Accepted April 21. 1956. Division of Analytical Chemistry, 128th Meeting, ACS, Minneapolis, Minn., September 1955.
