3100
Anal. Chem. 1986, 58,3100-3103
(7) Turk, G.C.;Watters, R. L. Anal. Chem. 1985,5 7 , 1979-1983, (8) Hart, L. P.; Smlth, E. W.; Omenetto, N. Spectrochim. Acta, Pari 6 1985, 408, 1637-1649. (9) Breckenridge, W. H.;Bllckensderfer, R. P.; Fitzpatrick, J.; Oba, D. J . Chem. Phys. 1979, 7 0 , 4751-4760. (IO) Natl. Bur. Stand. Circ. ( U S . ) 1952, N o . 467. (11) Green, R . E.; Havrilla, G. J.; Trask, T. 0. Appl, Spectrosc. 1980, 3 4 , 561-569. (12) Havrilla, G. J.; Green, R. 6. Anal. Chem. 1980,5 2 , 2376-2383.
(13) Huber. K. P.; Herzberg, G. Molecular Spectra and Molecular Structure I V . Constants Of Diatomic Molecules; Van Nostrand Reinhold Co.: New York, 1979. (14) Andersen, T.; Sorensen, G. J . Quant. Spectrosc. Radiat. Transfer 1973, 73, 369-376.
RECEIVED for review May 12, 1986. Accepted July 25, 1986.
X-ray Photoelectron Spectroscopy of Potential Technetium-Based Organ Imaging Agents Michael Thompson*
Department of Chemistry, University of 'Toronto, 80 S t . George Street, Toronto, Ontario M5S I A I , Canada Adrian D. Nunn and Elizabeth N. Treher
The Squibb Institute for Medical Research, P.O. Box 191, New Brunswick, New Jersey 08903
Technethrm-99 3dbindlng energies were measured for a set of 12 compounds whkh included model specles and several potential radiopharmaceutlcais. The range of compounds from the free metal to the technetlum(VI1) valence state gave a span of 4.9 eV. Anlonlc and coordinated halogen was distlnguished in the chlorine 2p and bromine 3d spectra of a number of phosphine complexes. Similar spectra of two dloxbne complexeg indicated signtfkantly more electron density on the halogen than is the case for a coordinated species. The result is Indicative of weakening of the metal-chlorine bond. Analogous spectra were obtained when bromine was substituted for chlorlne. The boron 1s Mnding energy for boronic acid present In the dioxlme cornpiexes was In agreement wlth the proposed oxygen population on the boron atom.
Technetium-99m radiopharmaceuticals are used extensively in nuclear medicine as in vivo diagnostic agents in part because of the favorable decay characteristics of this radionuclide ( I ) . Selective imaging of the organs is heavily dependent on the pharmacokinetic properties of complexes as influenced by the nature of the particular ligand employed. The latter is varied comprehensively to obtain structure-in vivo distribution correlations, which in turn allow radiopharmaceutical chemists to target labeled material to specific tissue. Of significant interest in this area has been imaging of the heart and the more difficult problem of imaging the brain, which involves passage of the technetium complex across the intact bloodbrain barrier (2, 3). A potentially important factor in the in vivo behavior of metal complexes is the overall charge on the complex and charge distribution on the central metal and ligand atoms. X-ray photoelectron spectroscopy (XPS) can yield relative information on the charge distribution and metal valence state through an examination of binding energy shifts compared to values from model compounds. The potential identification of metal oxidation states by XPS has attracted continued interest with mixed results. For example, qualitative correlation of metal binding energies with variable valence state in arrays of palladium (41, vanadium
Table I. Technetium 3dSIzBinding Energies and Metal Environment in Chelates and Model Compounds" compd
I
I1 I11 IV V VI VI1 VI11 IX X XI XI1
formula
Tc oxidn state n
NH~TcO~ [ ( ~ - C ~ H ~ ) , N ] T C O C ~ 5~ 4 [(CHJ~NITCC~G 4 (NH4)2TcBr6 [T~(dmpe)~Cl~]Cl 3 3 [Tc(d~pe),Cl~lCl 3 [T~(dmpe)~Br~]Br 3 [Tc(diars),Brz]Br TcC1[(dmg),bub] 3 3 TcBr[(dmg),bub] TcCl[( ~ d o ) ~ m b ] 3 0 Tc
Tc 3d5jz f0.2 eV
258.8 257.5 256.9 256.1 254.8 255.0 254.6 254.9 254.9 254.9 255.0 253.9
q:
1.6 1.3 0.9 0.7 0.3 0.3 0.2 0.3 0.3 0.3 0.3 0
"dmpe, 1,2-bis(dimethylphosphine)ethane;dppe, 1,2-bis(dipheny1phosphine)ethane; diars, o-phenylenebis(dimethy1arsine); dmg, dimethylglyoxime; cdo, cyclohexyldioxime; bub, n-butylboronic acid; mb, methylboronic acid. *Values for chelates V-XI obtained from plot of calculated qp vs. binding energy shift with respect to Tc metal from this work and ref 14.
( 5 ) ,silver (6),antimony (7), rhenium (8), molybdenum (9), tin (lo),ruthenium ( l l ) and , osmium (12) complexes has been attempted. In a semiquantitative approach Larsson et al. (13) obtained a good correlation of estimate of partial charge on the metal atom for 2p3jzbinding energies of copper in a series of analogous compounds. Charge calibration was achieved by theoretical calculations for a standard compound. The studies have shown that there is often a rough correlation of XPS chemical shift with metal oxidation, but ligand polarizability and T- and metal-metal bonding invariably provide influences that obscure such relationships. Furthermore, it is difficult to distinguish different oxidation states in the heavier metals where the range of binding energies is small or where low oxidation states as a group are involved. A dearth of binding energy data is evident in the case of compounds of technetium. Gerasimov et al. (14) measured Tc 3d,,z values for a set of simple inorganic compounds, and a good correlation of binding energy with estimated Pauling partial charge was obtained. In addition, the internal conversion electron technique has been used to examine the va-
62 1986 American Chemical Society 0003-2700/86/0358-3100$01.50/0
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
3101
Tc 3dy
Figure 1. General structure of technetium-dioxime complexes (IX-XI).
lence condition of technetium-99m in a small number of similar compounds ( 1 5 1 6 ) . The present paper describes the first XPS examination of 99Tc chelates, including several compounds that show considerable promise as organ imaging agents. EXPERIMENTAL SECTION Materials. All technetium compounds were made using the long-lived 99'c isotope. NH4Tc04was obtained from Oak Ridge National Laboratory. After recrystallizationit was used to prepare other compounds. The other reagents used as starting materials were obtained commercially as reagent grade chemicals and were used without further purification. The Tc compounds examined in this work are given in Table I. Compounds I-VI11 were prepared by using previously published methods (17). The three remaining new complexes (E-XI) were synthesized by combining stoichiometric amounts of NH4Tc04,dioxime, and boronic acid in an ethanol/l M HCl or HBr (4:l) solution. Stannous chloride in 6 M HCl (HBr) was added to reduce Tc(VII), yielding a dark red-orange solution. Removal of about half the solvent by rotary evaporation followed by addition of 3 M HCl (HBr) gave a red-orangeprecipitate. After washing with water, the solid was dissolved in warm ethanol and reprecipitated by adding 3 M HC1 (HBr). The solid was recrystallized from hot ethanol. Characterization by infrared and UV-vis spectroscopyand elemental analysis was performed before X P S examination. The general structure of the dioxime complexes provided by X-ray crystallography is shown in Figure 1. Finally, for comparison purposes a bis boron-capped iron dimethyl glyoxime complex was prepared by using an existing procedure (18). X-ray Photoelectron Spectroscopy. Measurements were made on a Hewlett-Packard 5950A spectrometer using a Mg K a source. Samples were prepared for analysis by adhering powders to tape on a copper platen. A gold mask and an electron flood gun at a setting of 10 eV were used to control sample charging. Narrow-scan spectra were obtained for the measurement of electron kinetic energies and for approximate checks of atomic ratios. Binding energies for all compounds were referenced to the hydrocarbon signal for C 1s of 284.3 eV for purposes of relative comparison. This is the same standard used by Gerasimov et al. (14).
RESULTS AND DISCUSSION Generally in the body the cells separating the extracellular space of an organ and the vascular space are not tightly joined, and so solutes can pass from the blood to the organ through channels between the cells. One exception is the brain, where the cells are tightly joined so that all solutes must pass through a cell in order to permeate from one space to another. This barrier to movement is called the blood-brain barrier (19)and serves to restrict access to the brain. The uptake of a solute into a cell is determined by its ability to cross the cell membrane, which can be by active or passive means. Charged technetium compounds such as those shown in Table I are known not to cross the intact blood-brain barrier to any appreciable extent, whereas neutral technetium species can (20). In this respect it is an accepted goal in brain imaging to design a neutral and lipophilic agent. An important aspect of the passive diffusion of a complex across lipid bilayer membranes is the unfavorable Born energy barrier, which is
Tc-Cl
1
A
'3-
V
a
I
1
W
i W
78
58
58
BINDING
ENERGY
(eV)
Figure 2. X-ray photoelectron spectra: (a) Tc 3d for TCCI[(~O)~BCH,], (b) CI 2p for [Tc(dppe),CI,]CI, and (c) Br 3d for [Tc(dmpe),Br,]Br.
associated with the presence of a formal charge or dipole in the molecule. Despite considerable effort in the characterization of technetium complexes by several methods, the one technique capable of yielding direct information on charge distribution, XPS, has been virtually ignored. Accordingly, we begin with the measurements of the Tc 3d5,2 binding energy, which is in the vicinity of 256 eV (the 3dIjzsignal is shifted 3.7 eV to higher energy-a typical spectrum is shown in Figure 2a). The values obtained for [(n-C4Hg),N]TcOC14, 257.5 eV, and NH4Tc04,258.8 eV, are in excellent agreement with those given by Gerasimov and co-workers (14),257.4 eV and 258.9 eV, respectively. The remaining Tc 3d5j2levels for the compounds studied in this work together with suggested formal oxidation states of the metal are given in Table I. As expected a rather narrow range of 4.9 eV is found for the extremes of the free metal and Tc(VI1). Also, shown in Table I are the Pauling partial charges (qp) for the metal in the simple inorganic compounds. The latter are calculated from the partial ionicities of the bonds formed by the atom and the formal residual charge on the atom. Correlation of XPS chemical shifts for different metal oxidation states and for the effect of overall electronegativity of the ligand with the charge parameter has been attempted several times, but has met with limited success. Compounds V-VI11 are known to contain Tc(III), and so the similar Tc 3d5!2 binding energies c o n f i i that chelates IX-XI have a technetium oxidation state of three, with the charge on the metal being about 0.3 on the Pauling scale. As has been noted previously (8), higher oxidation numbers do reflect greater charge on the metal, but there is no direct correspondence between the two parameters.
3102
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
Table 11. Ligand and Anion Binding Energies for Technetium Compoundsn (k0.2 eV) binding energy, eV N 1s 0 Is 401.4 401.7 402.4
B ls
Fe 2p3,,
As 3d
531.1 532.9
130.1 197.9 121 130.4 67.9 66.2 68.1 66.5 68.2
(NH4)2TcBr6 [Tc(dmpe),Br,] Br [Tc(diars),Br,] Br TcCl[(dmg),bub] TcBr[ (dmg),bub] TcCl[ (cdo),mb] Fe(dmg)3(BOC,H9)2
402.1 (1) (2) (1) (2)
130.3
197.1 67.7 197.1
42.1 400.1 400.0 400.1 400.1
532.4 531.9 531.8 532.2
190.0 190.2 190.4 191.1
709.6
Values are k0.2 eV, and intensity ratio is in parentheses.
This prompted us to examine the charge condition on ligand atoms, particularly the halogen species. Binding energies measured for ligand and anion atoms are given in Table 11. Both C1 3p spectra of [Tc(dmpe),Cl2]C1 (Figure 2b) and [T~(dppe)~Cl,]Cl exhibit two signals with an average split of 2.1 eV in the intensity ratio of 2:1, which we ascribe to covalent and anionic chlorine, respectively. It is interesting to note that the mean value for covalent C1 3pSl2in the phosphine complexes, [ (n-C4H9),N]TcOCl4and [ (CH3)4N]2TcC1,,198.0 eV in a range of 0.5 eV, is remarkably close to the value of 197.9 f 0.5 eV for a large array of rhenium complexes, when binding energies are corrected to the same calibration level (8). The measurement of 197.1 eV for C1 3p3j2 in both TcCl[ (dmg),bub] and TcC1[(cdo),),mb] indicates significantly more electron density is present on coordinated chlorine in these complexes than for the other compounds mentioned above. Such an effect is undoubtedly associated with an increase of the technetium-to-chlorine bond length. In this respect Clark et al. (21) observed an average of 1.0 eV lower binding energy (C1 3pSl2for cis compared to trans chlorine for a set of square planar phosphine complexes of Pt(I1) and Pd(I1). This shift was correlated with an increase in bond length from 2.29 to 2.39 8,and with results from NQR spectroscopy (22). The reason for the weakening of the Tc-to-C1 bond in the dioxime complexes studied here may lie in the nature of the coordinate structure around the metal. Chlorine-to-metal bond lengths of 2.32-2.36 8, have been observed for six-coordinate technetium complexes of variable oxidation number, whereas a value of 2.44 8, was measured for an eight-coordinate species (23). Accordingly, it is possible that the weakening of the bond in the dioxime complexes may be associated with the coordination number of seven found in these compounds. Finally, we have found that the chlorine bond weakening is reflected in the chemistry of the complexes with respect to facilitated nucleophilic displacement. For example, the chlorine atom can be exchanged relatively easily by the addition of Br- to a solution of the technetium C1dioxime complex. The bromine 3d spectra of [T~(drnpe)~Br,]Br and [Tc(diars),Br2]Br (Figure 2c) again show the characteristic split in the intensity ratio of 2:l for coordinated and anionic halogen, respectively. As expected the discrepancy in 3d5,12values is an average of 1.8 eV. Compared to the chlorine case discussed above, the measurement for Br 3d5,, in TcBr[(dmg),bub] (67.7 eV) is closer to the average for covalent bromine (68.0 eV). Boron Is binding energies are found in the approximate range of 185-195 eV ( 2 4 , 2 5 ) . In the present work the mea-
surements for the dioxime complexes are compared with that of Fe(dmg),(BOC4H9),as a standard. As expected TcC1[(cdo),mb] exhibits a slightly higher B 1s energy than TcCl[ (dmg),bub] due to the different electronegativities of the two alkyl groups. Also, the values for the dioxime complexes are about 1.0 eV lower than for the iron complex, indicating higher electron density on boron in these compounds. This result is consistent with the placement of four electronegative oxygens on the boron atom in the iron complex ( l e ) ,compared to three for the T d i o x i m e system. Reduction of the oxygen-on-boron population by one appears to result in a lowering of B 1s energy by about 1.0 eV. The measurements for N Is and P 2p3,2 are as expected with the quaternary nitrogen species being about 2.0 eV higher than the dioxime nitrogens, which yield highly consistent binding energies (and a very useful internal standard for the calibration procedure). Finally, with respect to 0 Is, the value for [(nC4H9),N]TcOC14is significantly higher (532.9 eV) than for the remaining compounds due to the influence of the -TcCl, moiety. The nonequivalent oxygen atoms in the Fe complex (ratio 3:l) have not been resolved. In conclusion, XPS has demonstrated that the Pauling partial charges on the technetium(II1) in the neutral dioxime complexes is -0.3. The remaining charge is distributed throughout the molecule and manifests itself in part as a weakening of the halogen-technetium bond strength. The Tc-dioxime complexes described above appear to combine low metal valence state with the absence of formal charge or significant presence of a dipole condition. The 99"Tc complex of one of the compounds (XI) is currently in clinical trials as a myocardial imaging agent and has allowed the visualization of exercise-induced ischemia. Other members of the series have shown brain uptake and retention in animals commensurate with their physiochemical properties. Registry No. I, 13598-66-8; 11, 92622-25-8; 111, 104716-19-0; IV, 75493-13-9; V, 80537-73-1; VI, 91467-32-2; VII, 80529-07-3; VIII, 80529-08-4;M, 104716-20-3;X, 104716-21-4;XI, 104716-22-5; XII, 7440-26-8; Fe(dmg),(BOC4H9)2, 39060-39-4; Br,, 7726-95-6; P, 7723-14-0; N2, 7727-37-9; 0 2 , 7782-44-7; B, 7440-42-8; Clz, 7782-50-5; As, 7440-38-2; Fe, 7439-89-6. LITERATURE CITED (1) Eckelman, W. C.; Levenson. S. M. I n t . J. Appl. Radiat. Isot. 1977, 28, 67-82. (2) Ell, P.J.; Holman, B. L. Computed Emission Tomography; Oxford University Press: Oxford, 1982. (3) Radionuclide Imaglng of the Brain; Hoiman, E. L., Ed.; Churchill-Livingstone: New York. 1985. (4) Kumar, G.; Blackburn, J. R.;Albridge. R G.; Moddeman. W. E ; Jones, M. M. Inorg. Chem. 1972, 1 4 , 73 (5) Larsson, R.;Fokesson, E.; Schon. G. Chem. Scr 1973, 3, 88
Anal. Chem. 1986, 58,3103-3108 (6) Murtha, D. P.; Walton, R. A. Inorg. Chem. 1973, 12, 368. (7) Birchall, T.; Connor, J. A,; Hillier, I. H. J . Chem. SOC.,Dalton Trans. 1975, 2393. ( 8 ) Chatt, J.; Elson, C. M.; Hooper, N. E.; Leigh, G. J. J . Chem. Soc., Dalton Trans. 1975, 2393. (9) Chatt, J.; Eison, C. M.; Leigh, G. J.; Connor, J. A. J . Chem. SOC., Dalton Trans. 1976, 1351. (IO) Umapathy, P.; Badrinarayanan, S.:Sinha, A. P. 8. J . Electron Spectrosc. Reiat. Phenom. 1983, 28, 261. (11) Demanet, C.M. South Afr. J . Chem. 1982. 35, 45. (12) Lin'Ko. I . V.; Zaitzev. 8. E.: Molodkin, A. K.: Ivanova, T. M.; Lin'Ko, R. V. Zb. Neorg. Khim. 1983, 28, 1520. (13) Folkesson. B.; Sundberg, P.: Johansson, L.; Larsson, R. J . Electron Spectrosc. Reiat. Phenom. 1983, 32, 245. (14) Gerasimov, V. N.; Kryuchkov, S. V.; Kuzina, A. F.; Kulakov, V. M.; Pirozhkov, S. V.; Spitsyn, V. I. Doki. Akad. Nauk. SSSR, Engi. Transi. 1982, 266, 148. (15) Gerasimov, V. N.; Zelenkov, G. A.; Kuiakov, V. M.; Pcheiin, V. A,; Sokolovskaya, M. V.; Soldatov, A. A,; Chistyakov, L. V. Zh. Eksp. Teor. Fir. 1984, 86, 1169. (16) Fiser, M.; Brabec, V.; Dragoun, 0.: Kovalik, A,; Frana, J.; Rysavy, M. Int. J . Appl. Radiat. b o t . 1985, 36, 219.
3103
(17) Deutsch, E.; Nicollni, M.; Wagner, H. N., Jr. Technetium in Chemistry and Nuclear Medicne; Cortina International: Verona, Italy (Distributed by Raven Press, New York), 1983. (18) Jackels, S.B.; Rose, N. J. Inorg. Chem. 1973, 12, 1232. (19) Bradbury, M. The Concept of a Blood-Brain Barrier; Wiley: New York, 1979. (20) Lister-James, J. In Radionuclide Imaging of the Brain; Hoiman, B. L., Ed.: Churchill-Llvingstone: New York, 1985; p 75. (21) Clark, D. T.; Adams, D. B.; Briggs, D. J . Chem. SOC. D 1971, 12, 602. (22) Fryer, C. W.; Smith, J. A. S.J . Chem. SOC.A 1970, 1030. (23) Bandoli, G.; Mazzi, U.; Roncari, E.; Deutsch, E. Coord. Chem. Rev. 1962, 4 4 , 191. (24) Handbook of Photoelectron Spectroscopy;Perkin-Elmer: Eden Prairie, MN, 1979. (25) Bremser. W.; Linneman, F. Chem. Ztg. 1972, 96,36.
RECEIVED for review May 2,1986. Accepted August 6, 1986. The support of the Natural Sciences and Engineering Research Council of Canada to M.T. is gratefully acknowledged.
Theoretical Basis for Line Number to Line Intensity Logarithmic Relationship Alexander Scheeline
School of Chemical Sciences, University of Illinois, 1209 West California Avenue, 79 R A L Box 48, Urbana, Illinois 61801
A linear relationship between the logarlthm of the number of lines having a given spectral Inlenslty and the logarithm of the relative lntenslty of those llnes wlth respect to the weakest observable llnes In the spectrum Is crltlcally evaluated. Use of exact quantum mechanical formulas tor electrlc dlpole transltlons In hydrogen allows slmulatlon of spectral behavlor that can be generallzed to elements whose spectra can be described In the Russell-Saunders limit. The linear relationship Is shown to be of narrower appllcablllty than orlglnally concelved. Some apparent anomalles In llterature data are explalned. Results are dlscussed In terms of Interferences expected from line overlaps In emlsslon spectrochemical analysis.
In a recent article ( I ) , experimental evidence for a general relationship between the number of observable lines in an emission spectrum and the relative intensities of those lines was discussed. An empirical relationship proposed by Learner (2) was critically evaluated, based on the spectrum of neutral arsenic obtained under conditions similar to those used by Learner in generating his hypothesis (3). The relationship employed was loglo Nk = logl, (No)- km (1) where N k is the number of lines with emission intensity in octave k , N o is the number of lines emitting a t the limit of detection (signal approximately 3 times the observation system limiting noise), k is the octave number (recall an octave is a factor of 2 in emission intensity), and m is the slope of the supposedly linear relationship between octave number and number of lines. Learner found that m was very close to 0.1505, which is log,, 21/2,for several elements excited in hollow cathode lamps. Analysis of arsenic data ( I , 3) also revealed that, for data collected on the McMath solar telescope in-
terferometer, the sameslope line could be fit to observed data. What remained unclear was whether the slope of 0.1505 was a true physical constant, an artifact of measurement, or a coincidental value that applied only to the species observed. In the absence of instrumentation having sufficiently well characterized spectral response to independently generate intensity data, a theoretical approach to the problem was pursued. The only neutral element for which spectral properties can be accurately calculated in closed form (ignoring relativistic corrections) is hydrogen. The appropriate formulas can be found in general works on atomic spectra ( 4 ) ,specific references on quantum mechanics (5), and the journal literature (6). Consideration was restricted to dipole-allowed transitions. Oscillator strengths for hydrogen have been tabulated for all absorption transitions between n (principal quantum number) levels up to n = 50 (6),and tabulations for restricted sets of transitions including levels up to n = 500 are available (7). There is thus a sufficient data base to allow simulation of the intensity distribution in the hydrogen spectrum. Comparison of the theoretical distribution to that predicted by Learner allows for critical appraisal of the validity of eq 1. Clearly, the hydrogen spectrum differs from that of all other elements in that there are no electron-electron interactions. However, judicious use of various theorems concerning transition probabilities allows extrapolation to more complicated atoms. There appears to be some parallel between the current discussion and the statistical theory of spectra (8). However, such theory has been directed mainly a t the distribution of energy levels rather than the intensity of transitions between those levels. Indeed, other than some general sum rules for oscillator strengths, the entire issue of distribution of transition moments among various levels appears to be largely empirically based. In the following sections, the necessary formulas for the intensity distribution in the hydrogen spectrum are presented
0003-2700/86/035S-3103$01.50/0 0 1986 American Chemical Society
