( 2 ) Arthur, P., Vanderkam, R. K., Ibid., p. 765 (1961). ( 3 ) Barker, G. C., Anal. Chim. Acta 18, 118 (1958). ( 4 ) TBockris, J. O’M., Azzam, A4. M., ‘1ature 165, 403 (1950). ( 5 ) Brdicka, R., Collection Czech. Chem. Commun. 8, 13 (1936). ( 6 ) Cooke, W. D., Kelley, M. T., Fisher, D. J., ANAL. CHEM.33, 1209 (1961). ( 7 ) DeFord, D. D., George A. Philbrick Researches, Inc., Applicatons Bull. 2-1 19-57. (8) Devay, J., Acta Chim. Hung. 35, 255 (1963). (9) Fisher, D. J., Stelzner, R. W., Oak
Ridge Xational Laboratory, Oak Ridge,
Tenn., private communication, 1963. (10) Ilkovic, D., Collection Czech. Chem. Commun. 4, 480 (1932). ( 1 1 ) Ilkovic, D., Zbid., 8, 13 (1936). (12) Kaspar, C., Trans. Electrochem. SOC. 77,357 (1940). (13) Kelley, M. T., Fisher, D. J., Cooke, W. D., Jones, H. C., “Advances in Polarography,” I. S. Longmuir, ed., Vol. I, p. 158, Pergamon Press, 1960. (14) Kelley, hl. T., Fisher, D. J., Jones, H.C., ANAL.CHEM.32, 1262 (1960). (15) Kelley, M. T., Jones, H. C., Fisher, D. J., Zbid., 31, 1475 (1959). (16) Kolthoff, I. M., Marshall, J. C., Gupta, S. L., J . Electroanal. Chem. 3, 209 (1962).
(17) National Bureau of Standards, Circular 514, August 10, 1951. (18) Schaap, W. B., McKinney, P. S., ANAL.CHEM.36, 29 (1964). (19) Taylor, J. K., Smith, S. W., J . Res. iVatL. Bur. Stds. 56, 143 (1956). (20) Windsor, P., M.S. thesis, Indiana University, Bloomington, Ind., 1957.
RECEIVEDfor review January 8,. 1964. Accepted A ril 8, 1964. Division of Analytical Ehemistry, 145th Meeting, ACS, New York, S . Y., September 1963. Work supported by a grant from the E.S. Atomic Energy Commission, Contract S o . AT(11-1)-256. Taken in part from the Ph.D. thesis of Peter S. McKinney.
Polarographic Behavior of Dysprosium(ll1) in Aqueous Solut ons ROBERT F. LARGE’ and ANDRE Y TIMNICK Kedzie Chemical Laboratory, Michigan State University, East Lansing, Mich.
b Two-step polarograms are recorded with Dy(lll) in aqueous 0.1M LiCI, 0.01% gelatin. On varying the pH from 2.6 to 4, Eliz for the first step changes from - 1.85 to - 1.735 volts vs. S.C.E. At constant pH, id for the first step varies linearly over the concentration range 0.2 to 5.0 mM. El,z for the first step is shifted to more negative potentials and the second step disappears when DzO is used as the solvent. A suspension of a lanthanide hydrous oxide yields a one-step wave which coincides with the second step of the normally recorded polarogram. These observations lead to the conclusion that two-step polarograms recorded with solutions containing lanthanide ions which do not yield dipositive ions in aqueous solutions, result from the reduction of hydrogen ions produced by the hydrolysisof the hydrated lanthanide ions in the immediate vicinity of the electrode and the subsequent reduction of additional hydrogen ions from water associated with the adsorbed lanthanide hydrous oxide.
S
the appearance of the initial report pertaining to the polarographic behavior of lanthanum-and the tripositive lanthanide ions by h’oddack and Druckl ( l a ) , numerous conflicting results have been published. Expected stepwise reduction has been observed for europium( H I ) , ytterbium(II1) (g), and samarium(II1) (17, 19), and direct reduction of samarium(II1) (4, 19) to the metal has also been reported. Stepwiae reductions (12, I d ) , direct IKCE
Present address, Eastman Kodak Co., Rochester, N. Y.
1258
ANALYTICAL CHEMISTRY
reduction to the metals ( 2 , 6 ,16,16,19), reduction of hydrogen ions from hydrated lanthanide species (IO), and hydride formation (18) have been proposed as the processes reflecting the polarographic behavior of the lanthanide ions in aqueous solution. Kolthoff and Lingane (7) and, more recently, Hall and Flanigan (3) have summarized the status of lanthanide ion polarography. This investigation is an attempt to obtain further information which will lead to an elucidation of the electrode processes associated with polarographic waves recorded with aqueous solution containing lanthanide ions. The dysprosium(II1) ion was selected as representative of the group and most of the experimental work was conducted with solutions of this ion. EXPERIMENTAL
Apparatus. The current-potential curves were recorded with a Sargent Model XXI polarograph, with which a Sargent S-30260 potentiometer was employed for potential measurements. The current-time curves were displayed on a Tetronix Model 504 oscilloscope, and photographed with a Polaroid Model 110.1 camera. A modified H-cell was employed. The sample compartment was separated from the central compartment by a fine porosity glass disk. The central compartment was filled with supporting electrolyte and was separated from the third compartment by a medium porosity sintered glass disk. The third compartment contained saturated KCI to allow connection with the saturated calomel reference electrode. This arrangement provided adequate electrical connection while minimizing contamina-
tion of the .sample solution by foreign ions or by agar from the side arm of the reference electrode. The central and reference compartments were filled with fresh solutions prior to each run. The capillary employed in this work had a drop time of 4.05 seconds and an m value of 1.12 mg. per second a t an applied potential of - 1.40 volts us. S.C.E and an heff value of 69.6 cm., as measured in 1 m J I solution of dysprosium(II1) a t a pH of 3.0 in 0 , l M LiCl plus 0.01% gelatin. Reagents. The lanthanide sesquioxides (99.9%) employed in this work were obtained from the Michigan Chemical Corp. and were ignited to constant weight before use. The deuterium oxide (99.5%) was obtained from Xichem, Inc. -111 water employed in this work was redistilled from alkaline permanganate solution. The tetramethylammonium iodide employed was purified by repeated recrystallization from an ethanol-water mixture. All other chemicals were reagent grade and were used without additional purificntion. Experimental Procedures. Dissolved oxygen was removed from all solutions in the electrolysis cell by bubbling a stream of prepurified nitrogen through the solution for 15 minutes. I n recording the current-potential curves, extra care was taken to ensure that the capillary tip was as clean as possible. Before each use, the capillary, with mercury flowing through it, was immersed in concentrated HSOa for a few minutes, rinsed thoroughly with distilled water, and then irnniersed in distilled water for approximately 15 minutes. Just prior to insertion into the cell, the capillary was rinsed thoroughly with a stream of redistilled
I
spa.
It
W
a a 3 0
c -1.15
-1.45 -1.75 Edo V 8 . SC.E.,VOLTS
Figure 1 . A.
0. I?.
-2.05
Polarograms
1 m M Dy(lll) in 0.1 M LiCI, 0.01 %gelatin, pH 3 HCIOa,0.1 M LiCI, 0.01 ?& gelatin, pH 3 0.1M L X I , 0.01 % gelatin
water and the sides were wiped dry with absorbent filter paper If the capillary remained in the cell between recording of additional data on 1;he same solution, care was taken to ensure that all hydrogen bubbles wliich might have accumulated during the previous electrolysis were remoxJed. Since the i-E curves obtained in this invest,igation were noi, symmetrical, an arbitrary procedure had to be developed through which a practical half-wave pot,ential, Ellz, and the diffusion current, i d , values could be reproducibly evaluated from these curves. Curve d of Figure 1 represents the mid-points of the oscillations of a recorded polarogram. The first step is due to the reduction of hydrogen ions present in the acid solution. The second and third steps, hereafter referred to as the 6rst and second lanthanide waves, were observed only when lanthanide ions were present in solution. To evaluate EIl2 for the first lanthanide wave, one line was drawn through the upper hydrogen wave plateau and another through the upper first lanthanide wave plateau. A line was then drawn through the ascending portion of the first lanthanide step. El,zwas located a t the mid-point of the segment of this line terminating at the points of intersection with the plateau lines. Evaluation of Ellz for the second lanthanide wave was impossible, since no upper plateau developed. A horizontal line was drawn through t h r point' of highest current attaincrl in ('-.e second lanthanide wave. .4 l i n r .:;as then drawn through the ascending portion of the second lanthanide wave. The potential a t the pcint of intersection of this line with the upper horizontal line, designated as E,, is the identifying Imtential evaluated for the second lanthanide wave instead of &.
The diffusion current id, for the first lanthanide wave, was measured from the upper hydrogen wave plateau to the mid-point of the firqt lanthanide plateau. Since a true diffusion current could not be evaluated for the second lanthanide wave, a current designated as i, was measured. Its magnitude was measured from the mid-point of the upper plateau of the first lanthanide wave to the maximum current level attained for the second lanthanide wave. A considerable portion of this current is due to discharge of the supporting electrolyte, and where possible the current value for the supporting electrolyte a t this point was subtracted from the measured i, value. Because of these complications, the i, values are not of high accuracy. The graphical evaluation of the wave parameters for the two steps is demonstrated in Figure 1. All potential values were corrected for iR drop through the cell and measuring circuit in the customary manner. T o obtain instantaneous current-time (i-t) curves, a 100-ohm + 0.5y0 resistor was inserted in series with the polarographic cell and the potential drop across this resistor followed with an oscilloscope. The output of the Sargent Model XXI polarograph was employed a$ a potential source, with the recording portion of the instrument disconnected. The i-t curves as displayed on the oscilloscope were photographed with the Polaroid camera during one complete sweep. The usual operating conditions for the Tetronix Model 504 oscilloscope were: a sweep time of 1 cm. per second, recurrent triggering, and a sensitivity between 0.2 and 2 mv. per cm. DISCUSSION OF RESULTS
Shown in Figure 1 are representations of the i-E curves obtained for 1 m M dysprosium(II1) a t p H 3 (curve A ) , perchloric acid a t p H 3 (curve B ) , and that of the supporting electroll te, 0.1M
LiCl plus O.Olyogelatin (curve C). Of special interest is the fact that the wave resulting from the reduction of hydrogen ion precedes the waves resulting from the presence of the lanthanide ion. Therefore, in considering the processes involved in the waves due to the lanthanide ion, the fact that another electrochemical reaction is actively taking place prior and during t,hese processes must be kept in mind. Also of interest is the development of the waves resulting from the lanthanide ion in the presence of varying concentrations of free hydrogen ion. As the p H of the solution is increased, the first wave resulting from the lanthanide ion becomes less well defined and difficulties in measurement of the wave parameters result. Therefore, to assure adequate development of the waves, some excess free hydrogen ion must always be present in the solution. However, if the hydrogen ion concentration is too high, the current resulting from the discharge of hydrogen ion is too large to allow accurate measurement of the current due to the presence of the lanthanide ion. Thus for general purposes, the most practical range of hydrogen ion concentrations is that equivalent to the p H range 2.5 to 4.5. For simplicity in the following discussion, the lanthanide ion-induced waves are separated from the hydrogen wave and identified in order as the first wave and the second wave, and discussed separately as far as is feasible. Characteristics of First Wave. T o test the waves obtained for dysprosium(II1) in 0.1Jf LiCl plus O.O1yc gelatin for reversibility, t,he waves were analyzed by plotting E,, against log i/i, - i. The plots obtained show curvature, are not symmetrical about the zero log intercept or true R l i z ,and give nonintegral values of n. From these facts, it can be assumed that the process involved is not a reversible one, and in such case, predicting an n value from the slope of the plots is impossible. Further evidence of the irreversibility of the process giving rise to this wave is the high value, +5 mv. per degree, of the temperature coefficient of E,,* obtained over the temperature range 20" to 3 O O C . T o study the effect of the concentration of free hydrogen ion on the wave parameters and id, solutions of dysprosium(II1) in 0.1Jf LiCl plus 0.01% gelatin were prepared varying only in p H and were polarographed under similar conditions (Table I ) . The wave shows a marked pH effect with becoming more negative with a decrease in pH. The slope of an E1,z us. p H plot for 0.8 m41 dysproPium (111) in 0.1M LiCl plus 0.01% gelatin is 82 mv. per pII unit and is a represent'ative value. This shift in El,, VOL. 36, NO. 7, JUNE 1964
1259
I
Table I. Variation of Current-Potential Characteristics of First W a v e with Change in pH
0 8 mJP dysprosium(II1)in 0 . l X LiCl plus 0 01% gelatin E , volts Zd, PH zs P C. E. 2 59 -1 849 3 42 3 50 2 78 -1 828 3 60 2 98 -1 806 3 78 3 18 -1 795 3 92 3 53 -1 768 -1 735 4 02 4 02 *1
Table 11. Variation of Current-Potential Characteristics of First W a v e with Change in Dysprosium(ll1) Concentration at Constant pH
0 1M LiCl Rith 0.01% gelatin at a pH of 3.00
Concn , mM
0 0 0 1 2 5
2 5 8 0 0 0
E1 2, volts us S.C E -1 801 -1 802 -1 803 -1 799 -1 813 -1 815
td,
pa 1 00 2 48
3 91 5 03 10 79 27 33
with change in p H is in the same direction and of the same order of magnitude as the shift in with change in concent,ration observed for a weak acid in unbuffered solutions when t,he process giving rise to a polarographic wave is the reduction of hydrogen ion produced by the dissociation of the weak acid (5)* The lanthanide ions hydrolyze a t relatively low pH and form the gelatinous hydrous oxides a t a pH between 6 and 7 . The conditions a t the electrode surface and in the solution in the immediate vicinity of the electrode are drastically affected by the diffusion-controlled reduction of hydrogen ion. With such depletion of the hydrogen ion concentration, the solution conditions in the immediate vicinity of the electrode should approximate conditions favorable to hydrolysis of the lanthanide species diffusing into this volume of solution. However, the actual conditions are determined by the flux of hydrogen ions from the bulk of the solution, and are thus dependent on the hydrogen ion concentration in the bulk of the solution., Changes in the effective free hydrogen ion concentration thus established in the immediate vicinity of the electrode would affect the protolytic dissociation (hydrolysis) of the hydrated lanthanide species. An increase in the flux of hydrogen ions to the electrode represqes dissociation of the hydrated lanthanide ion there. Thus, the variation of El 2 with changes in the hydrogen ion concen1260
ANALYTICAL CHEMISTRY
tration suggests that the actual reduction process associated with the lanthanide ion is the reduction of hydrogen ions produced by protolytic dissociation of the hydrated lanthanide ion a t the electrode surface. The decrease in id with decrease in pH as shown in Table I is also in keeping with this protolytic behavior. h s the flux of hydrogen ions to the electrode is increased, the protolytic dissociation of the lanthanide ion is repressed, which results in a lower id. This pattern of behavior is indicative of a current controlled by an antecedent chemical reaction, and is representative of the general classification of polarographic kinetic currents. -kt this point,, a comparison of the observed current values to those calculated by the IlkoviE equation is of interest, and to this end an estimation of the diffusion coefficient of the dysprosium(II1) ion is required. The diffusion coefficient a t infinite dilution of dysprosium(II1) was calculated by the Sernst expression to be 5.85 x sq. em. per second, employing the value of the equivalent conductivity a t infinite dilution for DyC13 of 142.0 mhos per em., as given by Dye and Ypedding (1) and the usually accepted value of 76.3 for the equivalent conductivity a t infinite dilution of the chloride ion. Employing the above calculated value of D , experimentally determined values of m and t of 1.12 mg. per second and 4.05 seconds. respectively, and a value of 3 for n, the id as calculated by the IlkoviE equation is 6.00 pa. for 1 m M dysprosium(II1) in 0 . 1 X LiCl plus 0.01% gelatin a t a p H of 3.0. Th. experimentally determined i d values are approximately five sixths of this value, which leads to the assumption that if n is 3, the observed current values are less than the calculated i d values because of the kinetic character of the processes giving rise to the wave. The relationship between the wave parameters i d and El:*and the concentration of lanthanide ion a t constant pH is shown in Table I1 for dysprosium (111) concentrations between 0.2 and 5.0 m J i in 0.1M LiCl with 0.01% gelatin a t a pH of 3.00. The observed id is proportional to the concentration over the range t,eeted a t a constant pH value. From Table I1 it can be seen that there is no variation in E1,2with change in concentration of the lanthanide ion at constant pH other than that to be expected in measurements over such a concentration range. Thus, the concentration of the lanthanide ion has little or no effect on E,,* as compared to that of pH, and control of the flux of hydrogen ions from the bulk of the solution to the electrode by keeping the pH a t a constant value as the concen-
tration varies serves as a "buffer" to level any effect on the dissociation process by the concentration of the lanthanide ion. Sovak (13) has shown that the El for the reduction of the deuterium ion from DC1 in D20 is more negative than that for the reduction of an equal concentration of hydrogen ion from HC'I in H 2 0 , and has also noted that the extent of the shift of E12 to more negative potentials is determined by the relative amount of H 2 0 present in the DzO. The values of the potmtial shift for particular solvent compositions were: 87 m y . more negative in 99.67, DzO, 80 mv. in 99.17, D 2 0 , 63 rnv. in 94.6% D 2 0 , and 31 niv. in 76.57, DyO. Furthermore, he established that such shifts in Eipzto more negative potentials were not observed for procwes not involving the reduction of hydrogen (deuterium) ion. Thus, if the reduction process associated with the lanthanide ion wave actually involves the reducion of hydrogen ions, a shift of El to more negative potentials should be observed if the process is carried out in
D20. Solutions of dysprosium(II1) were prepared containing the same concentrations of dysprosium(III), HC10,. supporting electrolyte, and gelatin as a reference solution in H20, and were electrolyzed under identical conditions. The i-E curves obtained for these solutions in D 2 0 are compared to the reference solutions in H20 in Figure 2. -1marked shift along the potential axis is evident in the waves. Two attempts were made and the value of the shift obtained in the first attempt was 60 mv. and in the second was 140 mv. However, in the first attempt, the exact dilution of the D 2 0 with H20 was not known, but the sample was considerably more dilute than in the second attempt. In the second attempt, the solution preparation and handling techniques were improved and the electrolysis was carried out immediately after preparation; thus the D 2 0 content of the sample was much higher than the first. A value of the potential shift in El,z greater than the 80 mv. obtained by Novak is not unexpected if the reduction process is preceded by a chemical reaction, as the D 2 0 would affect the rate of any chemical reaction, which in turn Jvould give rise to an additional shift to more negative potentials. I t was also found that' changing the solvent composition by the addition of a slight amount of water shifts the wave to more positive values than are observed with high D20 content. The solutions in DzO, as initially prepared, were estimated to be ai)proximately 98yc D 2 0a t best, and were of course continually being contaminated with H?O while in the cell. On
E&
V S.
Figure 2.
S.C.E., VOLTS
Polcirograrns
A.
1 m M Dy(lll) in 0.1M LiCI, 0.01%
B.
PH 3 1 m M Dy(lll) in 0.1M LiCI, 0.01% gelatin, p H 3, D 2 0 solvent
gelatin,
this basis, determinations of precise values of the EIiz shifts involved are impossible. .Isshown in Figure 2, the most striking observation in the comparison of the i-E curves is the disappearance of the second wave. However, the shift to more negative potentials puts this process into a potential region where the discharge of the iiupporting electrolyte gives a considerable current, and it is probable that the second process blends in with the discharge of the supporting electro1:qte and thus no second wave is observed. Solutions were prepared employing various salts as supporting electrolyte at the same concentrat:on and with the same concentrations of dysprosium(II1) and gelatin. Two wavN2s related to the lanthanide ion were 'obtained in all cases, and the choice of supporting electrolyte had no par5cular effect on the waves. The effect of gelatin on the wave is primarily due to the effect of gelatin
1
&,
1
7.0
cm.t
on the hydrogen wave, not on the wave induced by the lanthanide ion. I n the absence of any gelatin and at. hydrogen ion concentrations sufficient to give a well-defined hydrogen wave, t,he hydrogen wave is drawn out because of a prolonged maximum, and merges with the wave produced by the lanthanide ion, so that there is no visible first wave directly related to t'he presence of the lanthanide ion. However, the addition of a slight amount of gelatin allows t,he development of the lanthanide ion first wave. Triton X-100 gives very similar results. .It lower hydrogen ion concentrations (pH 3.5), the lanthanide ion wave is visible before the addition of gelatin. In either instance, the lanthanide ion wave exhibits a slight maximum until the concentration of maximum suppressor approaches 0.01 %. The observations by other workers (17) concerning the disappearance of polarographic waves for lanthanide ions in the presence of citrate or tartrate ions are consistent with the electrode process presented in this work, since complexation of the lanthanide ion would reduce the extent of hydrolysis. The improvement of yields in electrolytic separations of the lanthanide ions in the presence of citrate or tartrate ions can also be explained by consideration of this reduction in the extent of hydrolysis. Plots of i d against for the wave in question are shown in Figures 3 and 4. For 1 m M dysprosium(II1) in 0.lM LiCl with O . O l ~ ogelatin and with HClOd concentrations sufficient to yield a p H of approximately 3, a straight' line is obtained which does not pass through the origin and denotes behavior intermediate between dif-
I
I
8.0
Figure 4. Variatioii in i,, vs. heif relationship of first step with change in pH. 1 rnM Dy(lll) in 0.1M LiCI, 0.01 % gelatin A.
pH 3.20
8. p H 2.95 C.
p H 2.78
;
'
3'
h&,
I
5'
cm.t
'
Figure 3. Variation in step with changes in h,ff
7I
'
id of
9
I
first
A.
2 m M Dy(lll), pH 3.55 B. 1 m M Dy(lll), pH 3 . 2 0 C. 1 m M Dy(lll), pH 2.78 Supporting electrolyte. 0.1 M gelatin
LiCI,
0.01 7 0
fusion and kinetic control. With 2 m M dysprosium(II1) in the same medium but with a lower acid concentration, the line begins to approximate that to be expected for diffusion control. Of particular interest is the variation in the i d against he,, relationship observed as the p H varies with a 1 m M dysprosium(II1) solution. When the p H is decreased from 3.20 to 2.78, the plot of i d against (heff)l/Zobtained for the lower p H value shows IPSP diffusion control than that for the higher value, as illustrated in Figure 4. This behavior is in keeping with the variations of i d and with p H previously
Figure 5. i - t curves recorded with 1 rnM Dy(lll) in 0.1 M LiCI, 0.01 gelatin, pH 3
%
First step A. Plateau of hydrogen wave, E d r - 1.70 volts B. Ascending portion, E d c - 1.79 volts c. Ascending portion, E & - 1.84 volts D . Plateau of wave, E d e - 1.89 volts Second step E . Ascending portion, E d $ - 1.98 volts F. Ascending portion, E d < - 2.01 volts G. Fall in current prior to flnol rise, E d r - 2.05 volts All potentials VI. S. C. E. VOL. 36, N O . 7, JUNE 1 9 6 4
1261
mentioned and is again indicative of some kinetic control. A comparison of the plot of id against (hefJ1/*obtained for a solution of dysprosium(II1) in D20 to that obt,ained for a solution in H10 of similar composition is also of interest. As would be expected, the plots for the D20 solution show a greater kinetic control of the current, and the current values are also somewhat less. I s predicted by the IlkoviE equation, the current variation during the course of a single mercury drop is approximately proportional to the onesixth power of time. This relationship has been shown to hold over the entire course of a wave, from the beginning of t,his step until the limiting current is reached, for a reversible polarographic process. However, for an irreversible process, the current-time relationship varies during the course of the wave (8),the current being proportional to the two-thirds power of time a t very small values of the current as compared to the limiting current,, and with a gradual decrease over the course of the wave until a current dependence on the one-sixth power of time is attained in the limited current portion of the wave. h further effect, conclusively demonstrated by Kuta and Smoler (@, is that the i-t curve for the first drop after application of a potential is very different from that observed with any subsequent drops because of an inherited concentration depletion of the solution into which the second and further drops grow. Thus for any drop other t,han the first, t,he current dependence on t'ime will never reach the theoretical value, and the exponent of time varies over t'he course of the drop. An additional factor of importance is that the addit,ion of gelatin slight'ly alters the i-t relationships during the course of the drop, and tends to lower the value of the limiting current. However, with all deviations from the theoretical behavior combined, the i-t curves experimentally observed for well-defined polarographic processes are still smooth curves which are suggestive of the behavior predicted by theory. To test the experimental technique employed in this work, i-t curves were obtained for millimolar solutions of cadmium(II.), lead(II), and hydrogen ion, The i-t curves observed during the course of the polarographic reduction of these ions were all smooth curves closely resemblirig the expected behavior. Shown in Figure 5 are t'he i-t curves obtained a t various potentials with a solution of dysprosium(II1) in 0.1J.I LiCl plus 0.01% gelatin a t a pH of approximately 3. Curve A of Figure Figure 5 shows t,he i-t curve obta,ined 1262
ANALYTICAL CHEMISTRY
r
I
3
I
5
G O N G . , KIM
Figure 6. Variation in i, for second wave with change in concentration at constant pH Dy(lll) in 0.1 M LiCI, 0.01
70gelatin, pH 3.00
a t a potential on the plateau of the hydrogen wave, which is regular. Curves B and C show the i-t curves obtained a t potentials on the rising portion of the first wave related to lanthanide ion, and curve D shows the i-t curve obtained a t a potential on the plateau of the first \Tave related to the lanthanide ion, which is again regular. The irregularities seen on curves B and C are reproducible, and are not seen with a solution equivalent to that being observed but without the lanthanide ion. An exact interpretation of the phenomena responsible for the irregularities is not possible a t this time. However, the nature of the irregularity suggests that it' is the result of one or more of the following phenomena: an abnormal increase in the size of the drop as a result of a change in surface tension, an abnormal change in the charging rate of the drop as reflected by changes in the double layer capacitance, or an abnormal mode of supply of depolarizer to the electrode surface. Any or all of these phenomena might be possible when the tripositive lanthanide ion hydrolyzes and produces a gelatinous solid in the immediate vicinity of the electrode. Further study concerning the conditions a t the electrode surface, such as double layer capacitance measurements, might yield information to allow an exact interpretation of the processes occurring a t the electrode. Characteristics of Second Wave. I n the measu ments related to the second several problems arise which are not encountered with the first wave. The second wave shows no plateau, just a drop in current prior to the final current rise due to discharge of the supporting electrolyte. However, in any attempt to interpret the wave, some reference point for current and potential measurements is necessary, and to this end the mid-point' of the oscillation giving the highest cur-
&
rent value prior to the fall in current, was chosen. The actual significance of the current a t this particular point is in doubt, but the fact that it is the highest current value prior to the final current rise and can be reproducibly evaluated suggests its use aq a reference point,. The current between this reference point and the point of measurement of id for the first i-iave was defined as i,. S o significance will he at,tached to the magnitude of this current. The supporting electrolytes employed in this work begin to be discharged and thus yield an appreciable current during the potential region \There t,he second wave occurs. Thus, the measured i, values contain current other than that wholly arising from a process which involves the lanthanide ion. Correct'ion of the measured current values by subtraction of the residual current is possible, but this leads to large uncertainties in the result,ing values and sometimes bends to complicate the problem rather than simplify it. hnother problem arising in these measurements is t'hat in the pot,ential region under investigation, hydrogen bubbles can be seen t,o accumulate on the tip of the capillary. The result of this accumulation of gas is a disturbance of t.he drop growth, a shielding of a portion of the mercury drop, and undoubtedly some stirring of the solution. 111 of these factors combined yield an irreproducibility in the wave of such magnitude that accurate measurements of the wave parameters are impossible. Thus, in the discussion to follow, emphasis is placed on trends suggested by a set of values rather than on the magnitude of the values. The variation of i, with variation of concentration as noted for solutions of dysprosium(II1) in 0.1M LiCl with 0.01% gelatin a t a constant p H of 3.00 is shown in Figure 6. As shown in the figure, the current is not linearly proportional to the concentration of dysprosium(II1). The current values plot'ted in Figure 6 are not corrected for any residual current contribution to the total current. The actual contribution by residual current to i, is greater for the higher concentration values, so the plot should actually show a greater curvature. When a current does not vary linearly with concentration over the concentration region in question, a current resulting from an adsorption process may be suspected. In the process giving rise to the first wave lanthanide hydrous oxide is produced in the immediate vicinity of the electrode and could be adsorbed. The current resulting from the subsequent reduction process would then be controlled by the degree of coverage of the surface of the
4 Figure 7. Effect of addition of N a O H on i for 1 m M Dy(ll1) in 0.1 M LiCI, 0.01 % gelatin PH
A. B.
3.70
D. 7.20
5.65 6.60
E.
C. All curves start a t
E curve
7.30
- 1.30 volts vs.
S.C.E.
Table 111. Variation of CurrentPotential Characteristics of Second W a v e with Changes in pH
Ede VS. S.C.E., VOLTS
0.8 m M dysprosium(II1) in 0.1M LiCl with 0.01 % gelatin
mercury drop, and thus would not vary linearly with concentralion. The variations in 2', and E , with changes in pH are s h o , m in Table I11 for 0.8 mJ1 dysprosium(II1) in 0.1M LiCl with 0.01% gelatin. E , shows a shift to more negative potentials with a decrease in pH. This shift is expected with this process following another process involving the same species which shows a marked potential shift. The variation in i, with change in pH as shown in Table 111 is probably somewhat exaggerated. The current values are those obtained after subtract,ion of the contribution to i, by discharge of the wpporting electrolyte, and the i, values for lower pH have a much larger contribution from the discharge of the #supporting electrolyte than do those a t higher pH, because of the shift of the wave to more negative potent:.als with a decrease in pH. However, some decrease in the current for the second lanthanide ion process with a decrease in pH would be expect'sd if the process requires the presence of the lanthanide hydrous oxide, since the effective hydrogen ion Concentration in the immediate vicinity of the electrode as determined by the pH in the bulk of the solution would affect the amount of lanthanide hydrous oxide which could persist. Figure 7 shows the i-E curves observed with the successive addition of NaOH to a 1 mM solution of dysprosium(II1) in 0.1M LiCl with 0.01% gelatin. The height of the first wave related to the lanthanide ion is immediately diminished as the pH is increased to 6.60, shows successive decreases in height as more NaOH is added, and is almost indistinguishable when the pH reaches '7.20. However, the second wave related to the lanthanide ion does not show such a marked decrease in height, and in fact decreases very little, even though a precipitate is plainly visible when the pH reaches 7. h susDension of previously precipitated lanthanide hydrous oxide in 0.1M LiC1 also shows
i-E curves of the same nature as curve E of Figure 7 . Thus the second wave appears to be definitely related to the presence of the lanthanide hydrous oxide. '4 greater concentration of hydrogen gas is visible around the electrode in this potential region than a t potentials on the plateau of the first lanthanide wave. Thus it appears that the reduction process subsequent to the adsorption of the lanthanide hydrous oxide also involves the reduction of hydrogen ion, with the most likely source of additional hydrogen ion being from water associated with the hydrous oxide. This second wave related to the lanthanide ion was also observed with solutions employing lithium perchlorate, lithium sulfate, or tetramethylammonium iodide as the supporting electrolyte. The addition of gelatin or triton X100 has no noticeable effect on the characteristics of the wave. The second wave was not observed when DzOwas employed as a solvent. The i-t curves obtained at various potentials during the development of the second lanthanide wave for a 1 mM solution of dysprosium( 11I) at p H 3 in 0.liM LiCl with 0.01% gplatin are shown in Figure 5. Curves E and F show the i-t curves obtained on the rising portion of the second lanthanide wave, and curve G shows the i-t curve obtained in the region where the current decreases prior to the final current rise. The gross irregularities shown in these figures were reproducible and were not observed with a solution of the same composition but without the lanthanide ion. Again, it is not possible to interpret i-t curves of this type, but there can be no doubt that some process is violently disturbing the normal electrode solution interface. With the other evidence at hand, it is probable that the process involves the adsorption of the lanthanide hydrous oxide formed by hydrolysis, and subsequent
E,, volts PH
vs. S. C. E.
i,, ra.
2.59 2.78 2.98 3.18 3.53 4.02
-2.025 -2.013 -2.009 -1.996 -1.980 -1.967
0.25 2.48 3.61 3.87 4.60 5.19
reactions which produce an extreme turbulence a t the electrode. Electrocapillary curves obtained in the presence of the lanthanide ion also show an irregularity in the potential region where the maximum value of the current resulting from th second process related to the lanthanide ion is found. This phenomenon may again be interpreted as indicative of an adsorption process. I
Comparisons with Other Lanthanide Ions. Shown in Figure 8 is a comparison of the i-E curves obtained for 1 m J I l a n t h a n u m ( I I I ) , 1 mill gadolinium(III), 1 mM luteciu,n(III), and 1 m M dysprosiuni(III), all in 0.1M LiCl with O O l ~ ogelatin and a t
J .60
-1.75 I
-2.05 I
-1.90 8
Ede vs. S . C . E . , VOLTS
Figure 8. Polarograms for 1 mM lanthanide ion in 0.1M LiCI, 0.01% gelatin A. 8.
Lu(lll1, pH 3.45 Dy(lll), pH 3.00
C. D.
Gd(lll), pH 2.90 Lallll), pH 2.80
VOL. 36, NO. 7 , JUNE 1964
1263
Table IV. Comparison of Half-Wave Potentials of First W a v e for Selected Lanthanide Ions
1 mM in lanthanide ion in 0.1M LiCl with 0.017, gelatin at pH shown Eiint
Ion Lanthanum(II1) Gadolinium(II1) Dysprosium( 111) Lutecium( 111)
pH 2 80
3 80 2 90 3.85 2.78 3.20 2.85 3.45
volts us. S. C.sE. -1 907 -1 849 - 1,820 -1.767 - 1.807 -1.783 -1.789 -1.751
the p H values shown. All of these ions yield i-E curves of a nature similar t o that observed for dysprosium(II1). The first wave related to the lanthanide ion in the i-E curve for lanthanum(II1) is difficult to see, a t higher pH values, and in fact might be missed if other i-E curves such as those shown in Figure 8 were not available for ready reference. Since both waves for lanthanum(II1) occur a t more negative potentials than those for dysprosium (111), the second wave might not be noted at lower p H values, since the shift in the naves with p H puts this wave into a potential region where it would tend to blend in with the discharge of the supporting electrolyte.
Table IV compares the Eli2 values obtained for the first wave related to the lanthanide ions, lanthanum(III), gadolinium( 111), lutecium(111),and dysprosium(II1). There is a transition to more positive potentials going from lanthanum( I1I) to gadolinium( I1I) to lutecium(III), which must reflect the differences in basicity of the three ions. The relative order established by the El,z values of the ions shown in Table IV and Figure 8 parallels the order of basicity established by Moeller and Kremers (11). SUMMARY
The polarographic waves observed for aqueous solutions of the dysprosium(II1) ion (and thus by inference all the tripositive lanthanide ions exhibiting no stable dipositive state in aqueous solution) are the result of the reduction of hydrogen ions produced by the protolytic dissociation of the hydrated lanthanide ion in the immediate vicinity of the electrode surface, followed by adsorption of the lanthanide hydrous oxide thus produced, and the accompanying reduction of additional hydrogen ions from water associated with the lanthanide hydrous oxide. LITERATURE CITED
(1) Dye, J. L., Spedding, F. H., J . Am. Chem. Sac. 76,879 (1954).
(2) Estee, C. R., Glockler, G., Ibid., 70,
1344 (1948). 13) Hall. L. C.. Flaniaan. D. A., ANAL. CHEM.’35, 2108 (1963). ’ (4) Iwase, A4., S i p p o n Kagaku Zasshi 78, 1656 (1957); C. A . 52, 16090 (1958). ( 5 ) Ibid., 81, 1266 (1960); C. A . 55,21913 (1961). (6) Kolthoff, F. M., Lingane, J. J., “Polarography,” Vol. I, 2nd ed., p. 243, Interscience, Kew York, 1952. (7) Ibid., I7oL 11, Chap. XXII. (8) Kuta, J., Srnoler, I., “Progress in Polarography,” Vo1. I, p. 43, Interscience, New York, 1962. (9) Laitinen, H. A., Taebel, W. A,, IND. ENG.CHEM.,ANAL.ED. 13, 825 (1941). (10) Misumi, S.,Ide, Y., Bull. Chem. SOC. J a p a n 33, 836 (1960). (11) hloeller, T., Krerners, H. E., Chem. Revs. 37, 97 (1945). (12) Noddack, W.,Bruckl, A., dngew. Chem. 50, 362 (1937). (13) Novak, J., Collection Czech. Chem. Commun. 9, 207 (1937). (14) Puruhhottam, A , , Raghava Rao, Bh. S.V., ilnal. Chim. .4cta 12, 589 (1955). (15) Rabideau, S . W., Glockler, G., J . ilm. Chem. Soc. 70, 1342 (1948). (16) Swensen, A. W., Glockler, G., Ibid., 71, 1641 (1949). (17) Timnick, A , , Glockler, G., Ibid., 70, 1347 (1948). (18) Treindl, L., Collection Czech. Chem. Commun. 24, 3389 (1959). (19) Yakubson, S. I., Kastromina, N. A., Z h . :Yeorgan. Khim. 2 , 349 (1957); C. A . 51, 16149 (1957). RECEIVEDfor review December 26, 1963. Accepted March 30, 1964. The authors are grateful for a Socony blobil fellowship awarded to Robert F. Large.
X-Ray Fluorescence Analysis of Roman Coins GILES F. CARTER 7 422 Bucknell Road, Wilrnington, Del. 7 9803
b Previous work has indicated that x-ray fluorescence analysis of museum objects is not trustworthy because of a surface enrichment effect. The present work shows that only a few elements are enriched or depleted at the surfaces of Roman copper and silver coins. When coins are electropolished in certain solutions, accurate analyses may b e obtained by x-ray fluorescence. Analyses of four silver and four copper Roman coins are presented.
T
HE RAPIDITY, accuracy, and nondestructiveness of x-ray fluorescence analysis have been demonstrated for numerous alloys ( I S ) . The surface of an alloy must be abraded or polished if an accurate analysis is to be made by x-ray fluorescence. I n cases where the alloy consists of two or more phases, one must take care that the surface is
1264
ANALYTICAL CHEMISTRY
prepared in such a way that all phases are exposed in their true proportions. For instance Kilday and Michaelis (22) found that a steel containing lead should be polished with diamond dust to prevent smearing of the lead phase on the surface. It is necessary to remove the surface oxide from alloys if one is to analyze the alloys accurately by x-ray fluorescence. There are some specimens, such as museum objects and particularly ancient coins, where normal cleaning and surface abrasion are unacceptable. The most extensive x-ray fluorescence analyses of coins and museum objects have been carried out by Hall (5-9). Hall stated repeatedly that one must view x-ray fluorescence analyses of coins with care because of a surface enrichment effect. The objective of the present work i.; to demonstrate that x-ray fluorescence analysis can be used
to analyze ancient coins with reasonable accuracy provided the surfaces are properly electropolished. EXPERIMENTAL
A General Electric X-Ray Diffraction-5 unit was used for all analyses. The sample holder used for determining aluminum, silicon, and phosphorus was covered with pure lead foil, and the sample area exposed was 1.2 cm. by 1.9 cm. For the determination of all other elements an aluminum holder was used, having an opening near the x-ray tube window of 0.64 cm. in diameter. When the tungsten spectrum interfered with the determination of an element, for example nickel or gold, a chromium target s-ray tube (AEG-50s Machlett) was used. Otherwise a tungsten target tube was used except for elements having a n atomic number less than vanadium. These elements were usually determined ac-
