Devices for Gradient Elution in Chromatography M.BOCK and NAN-SING LING
ROBERT
D e p a r t m e n t o f Biochemistry, University
o f Wisconsin, Madison, Wis.
Improvement of column chronlatographic separations appeared possible if devices existed for delivering an eluting agent whose composition varied in a controlled arid predictable manner. The described devices permit delivery of eluting agent to a chromatographic column in such a manner. When this method was applied to several test mixtures improved separation was obtained. The devices should be of aid in ion exchange or column Chromatographic separations f o r a wide variety of mixtures.
Q
G-ISTITATIVE elution of substaricee from a column or bed of precipitated or adsorbent is often facilitated if the composition of the eluent can be changed in a smooth uniform manner toward a stronger eluent. Alm, Williams, arid Tiselius ( I ) have considered the advantages of such a system over a stepwise change of eluting strength. Busch, Hurlbert, and Potter (2,4 ) have found a constanbvolume mixer (Figure 1) useful in separating organic acids, and for resolving complex mixtures of nucleo-
1
I
1
I
of tailing aiid collected in small volumes of e1uat.e even though the column may be operated at flow rates which preclude equilibrium conditions. Ijakshmanan and Lieberman ( 6 )have described a system which gives a more useful concentration gradient than that provided by the simple constant volume mixer. Parr (6) has recently reported :t simple system which gives useful concent,ration gradients. In thiR paper a system is described in which gravitational leveling of t.he liquids in two or more connected vessels of differing shape makes it possible to deliver a solution a t a concentration that can be made any desired function of the volume delivered. Desreux ( 3 ) has used a relat,ed principle to blend organic solvents for fractional solution of polymers. In Figures 2 through 5, typical systems are sketched together with the resulting concentration us. volume relations. The roncentration axis is the relative concentration, st.arting a t the concentration of solution in vessel 1 and ending with the concentration in vessel 2. The volume axis is the fraction of volume of the total system which has been delivered, except' in Figure 1, where it is the volume delivered divided by the volume of the mixing flask (flask 1) alone. A complete assembly consist,s of the containers. apertures for filling (and application of air pressure, if desired), and narrow diameter out1et.s leading to a small mixing chamber, and then through narrow diameter tubes to the chromatographic column, OI'ERATIhG PRlhCIPLE OF COMPOSITIOK CONTROLLER
If n vessels of identical height but varied in cross section are filled with solutions of concentrations C,,, and connected to a common discharge tube, and if liquid is then slowly withdrawn, the composition of this liquid is determined by the relative volumes contributed by each of the n vessels. If an amount of liquid is withdrawn, so that the meniscus in the assembly of vessels is lowered by a small distance Ah, the volume of liquid withdrawn
0
O . ? I ~
1.0
1. 5
2.0
/VI
Figure 1.
Constant Volume Mixer
Delivers concentration equa1,to C1
- (C? - CI)
e-'/Yi
In addition to the chromatographic uses of a solution of gradually changing concentration, for many operations thisprocess is laboriously and often inefficiently approximated by manually changing a wash liquid stepwise toward a greater or lesser concentration. The sulfuric acid regeneration (7) of calcium-loaded ion exchange resin is carried out in such a stepwise manner. The purification of proteins, which have been precipitated on a large surface of inert filler, may be rendered chromatographic in nature if the solvent, washed over the filler, is gradually changed to take into solution one by one the various components of the protein mixture. During the elution of substances from ion exchange resins it has been observed that if the concentration, C , of the eluting agent either rises linearly with the volume, u, of eluent passed or rises a t an increasing rate-Le., C 0:v2-substances covering a wide range of exchange coefficient can be separated with a minimum
0
0. 2
0. 6
0.4
0.
a
1.0
"/v
Figure 2.
Cylinder Containing Conical Flask (r = kh)
+
-
-
For curve a, C = CI (C? C I ) ( U / V ) Z .When inner flask has = kh, concentration varies linearly (curve b ) ; C CI (CZ CI) (U/V)
72
1543
+
ANALYTICAL CHEMISTRY
1544 from any vessel n is A,.hAh, where A,.* is the cross-sectional area of vessel n a t height h. The composition (volume fraction scale) of the solution withdrawn is
The theory of gradient elution chromatography has not advanced to a stage where it is possible to predict the concentration us. volume relationships which are most useful for a specific t,ask. At present, a system must be found empirically to give satisfactory results. However, if theory gives this information or if certain concentration-volume relationships are to be investigated, an apparatus can readily be devised to make the concentration any desired function of volume. If the concentration is to vary as some arbitrary function of volume delivered
c = c1 +f(u)
0.
(8)
then an inner vessel may be constructed whose radius squared varies in this same manner.
kr2 = f(v) Since
c
=
(9)
ci + k T 2
(10)
then
c
ED
=
c,
(11)
+f(v)
The authors have constructed such inner vessels by building a template which fitted the curve kr2 = f(u). An Erlenmeyer flask is mounted in a glass lathe, heated, and then blown and shaped until its radius fits the template.
el
g
1 C1
CONCEWI'TrRATION
Figure 3.
c2
Rectangular Tank with Baffle
C given as any desired function of volume
If the densities of the n solutions are similar, the Ah, terms cancel out. For example, in a system of two vessels (Figure 2) consisting of a cylinder and a radially symmetric vessel which occupies some of the space within the cylinder, the denominator of Equation 1 is a constant, 7rri (the cross-sectional area of the cylinder). The concentration of any element of volume withdrawn is CI
0
0. 2
0. 6
0.4
0.6
1.0
'/v Figure 4.
Thus, the concentration of the solution withdrawn varies with the square of the radius of the inner vessel. If the radius of this vessel is a linear function of the distance, h, from the top of the vessel-Le., if the inner vessel is a cone-the concentration varies with the square of h and of the volume delivered (Figure 2, a ) .
reahav
C = C1 Therefore,
C = C1
+ kri
+ k'V2
(3) (4) (5)
If an inner flask shaped like a bell jar ( r 2 a h ) is used in place of a cone, the concentration of the solution delivered varies linearly with the volume delivered (Figure 2, b). Ti
ah av
=
c 1
Therefore,
c
+ k"v
Mixer Composed of Plane-Walled Tanks
Tanks 2 and 3 both contain eluent z at concentration Cw but, tank 3 also contains eluent 21 at concentration C:. Left-hand famdy of curves refers to eluent z.and right-hand family to eluent 21. b equals b* --- b equals 2b* b equals 4b*
------
A device which permits a general solution of C = f(v) can also be constructed in a rectangular tank. The desired concentration-volume relationship is plotted on a graph in which the concentration axis is equal in length to the base of the rectangular mixing tank employed, and the volume axis is equal to the height of the tank. A metal bafRe is now shaped, so that its edge fits the C us. u graph. This plate is sealed into the rectangular tank (Figure 3). Upon operation of this tank in the manner described for mixers 2, 3, and 4,the concentration of the eluent delivered varies with the volume delivered in the manner desired. This may be extended to multiple component systems of the type described in Figure 4. In some applications it is simpler to employ separate plane walled tanks of the type illustrated in Figure 4 than to make a divided tank. In Figure 4, the depend-
1545
V O L U M E 26, NO. 10, O C T O B E R 1 9 5 4 enee of the concentration us. volume relationship on the ratio of the base lengths of these tanks is illustrated. Operating prccedure is the same &E for previously described devices. One tank is shown further divided to give a type of mixer which is useful when the concentrations of two or more eluting agents are to he varied independently. Chamber 3 is equipped with a standpipe which allows its solution level to be made equal to that in chambers 1 and 2. This mixer may also he constructed in a rectangular tank. Several baffles may he placed in the same tank and the chambers formed may he either used or isolated depending on the situation. Such a system can he used a8 either % one- or two-eluent mixer, and any of the concentration gradients illustrated may be selected by simply turning a stopcock. ,
ties d, and d,, the meniscus of solution 1 will he dJd, times as high nhave the point of mixing a8 will the meniscus of solution 2. The change in meniscus height Ah,, of solution 1 during a withdrawal of % small volume of solution will he dr/dl times as large as Ah%. The operation of the described devices will he effected because the menisci will be a t levels which depend on their density. However, if the menisci m e a t the same height as the paint of mixing, no error will he introduced by density differences. Thus, if i t is desired to produce concentration 08. volume behavior, which closely iollows theory for widely varying densities, the appmatu8 i8 80 constructed that the point of miring moves a t the Same rate as the menisci. The composition of the solutions delivered will he precisdy known in terms of volumes of solutions 1and 2. Volume changes on mixing are not corrected hut can he readily detorrnined if the operator desires to express his data in any particular concentration scale. Although it is not difficult to construct a device nhich moves the point of mixing-for instance, a float-it is generally not necessary unless large density differences are involved, and a precise description of the eluting agent is required. Highly reproducible and aoourate results o m usuall? he ohtained if the density effects are minimiaed by mounting the point of mixing a t the average height of the menisci during the over-all experiment.
F i g u r e % Mixing S ~ - s t e m U s i n g P r i n o i p l e Illustrated i n Figure 2 Figure 6 shoms a gradient producing device of the type used hy I'arr (6). This system consists of two vertical-walled containers with cross-sectiond &rea, A I , and Ax, and solutions of initial concentration C, and C?. Container 1 must he stirred continuously. This system delivers a solution of oonoentration
C
=
C, - (Ca - CJ(1
--v)AdAi
(12)
when the densities of the tvo solutions are similar
Figur e 6 . Mixer System Giving C = Cz Cx)(l - . ) M A , a, dr = ZA1: b, A . = AI: of arose
C.
-
(Cs -
2Aa = A,. where Ai and A , ere area=
MATERIALS O F CONSTRUCTION
Vessels. FIGURE 2, CYLINDER. If the system is run under slight pressure, an anaerobic fermentation jar, such as J3475, Scientific Glass Apparatus Co., can be employed. If gravity feed or microoumr, . . feed is employed, an open cylinder may be used. FIGURE 2, INNERVESSEL. An Erlenmeyer h s k of approximately the correct volume is first sealed to a tube, and the flask is worked in the glass blower's h t b e (using template when needed) until the proper shape is reached. FIGURES 3 AND 4. Metal sheets, bent or cut as needed, ?re soldered to farm the tanks or,a baffle is cementod into a jar similar to J3525 or 03770. Siientific Glass Apparatus Co. MICROPUMP, Technicon Chromatography Carp. EFFECT OF DENSITY
UIFFERENCES
..-... are in communication with each other through tubes, joined at. the 'point of mixing," have densi\X,hon
t w n snlntin". I___
\VI ... iich
The systoms described in this article have been constructed and tested in these hhoratories. The performance has been tested by measuring the concentration of aliquots withdrawn a t various stages during operation, The mixing efficiency has heen examined by using a 10% glycerol solution containing a blue dye. The homogeneity of effluent color was observed. Motor aperated mixing is not needed for the misers shown in Figures 2 through 5. If very slow rates of flow were used, it was not convenient to employ capillaries small enough to cause mixing by turbulent flow. I n these cases the solution can be passed through a short bed of silica. sand or fine glass heads. The laminar flow patterns are then sufficiently disturbed to give well-mixed solutions. These systems are easily eanstycted and very simple to
ANALYTICAL CHEMISTRY
1546 operate, and give concentration os. volume relationships which are both reproducible and predictable. Analytical chromatography of nucleotides, amino acids, and organic acids in these laboratories has been simplified and improved by use of the described gradients. ACKNOWLEDGMENT
The authors are indebted to L. J. Gosting for helpful discussions of the theory, and to F. N. Hepburn and J. C. Alexander for testing some of these systems on amino acid mixtures and for pointing out some of the required concentration relationships.
LITERA’CURE CITED
(1) LUm, R. S., Williams. R. J. P., and Tiselius, A, Acta Chem. Scand., 6, 8 2 6 (1952). (2) Husch, H., Hurlbert, R. E . , and Potter, V. R., J . Rid. Chem., 196, 717 (1952). (3) Desreux, V., Rec. trav. chiin., 68, 789 (1949). (4) Hurlbert, R. B., and Potter, T’. It.. personal communication.
(5) Lakshmanan, T. K., and Lieberman. S., Arch. Biochem. arid Biophys., 45, 235 (1933). (6) Parr, C. W., Proe. Biochem. Soc., 324th meeting, XXVII. (7) Rohm and Haas Co., Form 20R,dmberlite IR-120. K E C E I Y Efor D review hpril 9, 1054. -4ccepted .July 26, 1954.
Determination of Tin, Iron, and Molybdenum in Titanium Using Paper Chromatography IRVING KOLIER
and
Picatinny Arsenal, Dover,
CHARLES RIBAUDO
N. 1,
Although paper chromatography has found wide usage in the separation of the elements, no application to the separation of elements from titanium metal and alloys has been reported. As part of a program to develop methods of analysis for titanium metals and alloys, an investigation of the separation of trace metals from titanium by paper chromatographic techniques was undertaken in this laboratory. Tin, iron, and molybdenum were successfully separated from a titanium metal sample and quantitatively determined by polarographic and colorimetric methods. If such separations can be performed with a wide variety of metals, the procedures should widen the scope of the techniques available to the analytical chemist.
T
ITASIUM metal, in commercial form, may contain one or more of the following elements, usually in trace amounts: iron, chromium, maganese, alumillurn, molybdenum, magnesium, vanadium, tungsten, silicon, nickel, cobalt,, copper, zirconium, tiu, niobium, tantalum, beryllium, thorium, boron, sodium, calcium, lead, and silver. I n the determination of any of these elements a preliminary separation from titanium would bc most advantageous, and in many cases, necessary. A separation of the titanium is often carried out by chemical means Tl-ith very litt81euse of the physical means of separation. Although the chemical separation of titanium is satisfact,ory in some instances, the constant danger of coprecipitation of the trace elements, if t,he separation is one of precipitation, is aln-ays present. The use of paper chromatography as a means of separating metals from other metals has been a successful technique (6, 7 ) , although no work has been done toward its application t.o the separation of trace metals from titanium. Previous work involving the migration of tin, iron, molybdenum, and titanium using various solvent systems (6, 7 ) indicated that a separation \vas possible when these rnptals were present in more or less equal amounts; however, very little work had been done in the separation of trace metals from large amounts of another metal. An investigation was therefore initiated to explore the possibilities of finding a suitable solvent system, one in which titanium, the major constituent, would have little or no movement while the metals sought (tin, iron, and molyl~denum)would have sufficient movement for separation. Such a solvent system was found to be n-butyl alcohol saturated with 3 S hydrochloric acid. Titanium metal was dissolved in hydrochloric acid, and the dissolved metal oxidized with hydrogen peroxide and diluted to a specific volume. An aliquot was placed on a paper strip and
chromatographed. The separated mctals were determined by polarographic and colorimetric methods. APP.4RATUS
For the chromatographic separation of tin, iron, and molybdenum from titanium the following xere used: a Fisher chromatographic assembly (Catalog No. 5-724) containing a stainless steel support (painted with acidproof paint to prevent its corrosion by hydrochloric acid vapors), glass tray, Petri dish, four glass rods to hold t,he paper strips and a borosilicate glass jar, 2 feet high and 1 foot in diametrr: a roll (600 feet) of Fisher chromatographic filter paper Whatnian No. 1, 1.5 inches wide (Catalog S o . 5-716); a sprayer; and a micropipet calibrated to deliver 0.1 ml. The three elements were determined by use of a Sargent polarograph, Model SSI, ivith polarographic cell and a Beckman spectrophotometer, Model DU. CHROlM.4TOGRAPHIC SEPAKATIOS OF TIN, IRON, A S D MOLYBDENUM FROM TITANIUM
Reagents. Hydrogen peroxide, 30%; hydrochloric acid. n-BLtyl alcohol-hydrochloric acid solution. Add 3N hydrochloric acid to R volume of n-butyl alcohol in a separatory funnel. Shake and continue the addition of arid until two layers appear. Discard the lower layer. Alizarin-ammonia spray solution. Expose a 0.5% solution of alizarin in ethyl alcohol to ammonia vapors for 1 hour. The color of the solution changes from amber to violet. Procedure. Transfer quantitatively to a 50-ml. beaker an accurately weighed 0.50-gram portion of the titanium sample. Add approximately 7 ml. of coiiceiitrated hydrochloric acid, rover the beaker with a watrh glass and place on a steam bath. Add a few more milliliters of concentrated hydrochloric. acid if all thP t,itanium sample is not dissolved. When the entire sample has been dissolved, remove the beaker from the steam bath, cool to room temperature and add dropwise, with swirling, 30% hydrogen pewside. Continue the addition of hydrogen peroxide until the color of the solution changes from violet to yellow. Add 5 drops of hydrogen peroxide in excess and place o n a steam bath. Khen the dark red color obtained upon the addition of excess peroxide changes to yellow, remove from the steam bath, and cool to room temperature. Transfer to a 10-ml. volumetric flask and dilute to the inark with water. From a roll of I’/z inch Whatman S o . 1 chromatographic filter paper, cut 6 strips to i~ lrngth of 22 inches. Draw a pencil line horizontally across earh of the strips a t a distance of 5 inches from one end. Spread a 0.1-ml. portion of the dissolve‘d sample across the pencil line on each of five strips by means of a micropipet. Two of these strips are used in the determination of the elements sought, one strip is used for the determination of the hand location of the element hy spraying the strip with the Alizarin-ammonia spray solution, thr fourth and fifth strips are used for the colorimetric. determination of iron and molybdenum, if no separation is obtainrd, and the sixth strip is used for running a blank on the reagent and paper. -illow the spotted strips to dry in air for 1 hour. APPAR.ITL-S FOR CHROK~TOGRIPHING. Place the end of the
