1425
Molecular Structure and Magnetic Properties of p-Dihydroxo-bis[ 2,6-pyridinedicarboxylatoaquoiron(111)] and ~-Dihydroxo-bis[4-hydroxo-2,6-pyridinedicarboxylatoaquoiron( 111)] Tetrahydrate John A. Thich, Chia Chih Ou, Dana Powers, Byron Vasiliou, Donald Mastropaolo, Joseph A. Potenza.*132 and Harvey J. Schugar*' Contribution from the School of Chemistry, Rutgers University, New Brunswick, New Jersey 08903. Received January 22, 1975
Abstract: The crystal and molecular structures of the title complexes have been determined from single-crystal three-dimensional x-ray data collected by counter methods. [Dipic(H2O)FeOH]* (Dipic or Dipicolinate = 2,6-pyridinedicarboxylate) crystallizes as green prisms in space group P i ( C i ' , No. 2): with Z = 1; a = 8.844 (4), b = 11.001 ( 5 ) , c = 7.303 (4) A; LY = 94.06 (4), p = 111.56 ( l o ) , y = 131.54 (4)'; dc&d = 1.927, dot& = 1.93 (1) g/cm3. Least-squares refinement of 1436 reflections having F 2 > 30 gave a conventional R factor of 0.052. The structure consists of centrosymmetric dimeric units in which crystallographically equivalent Fe(II1) ions are bridged by two hydroxyl groups. The planar [ Fe2(OH)2I4+ unit has Fe-OH-Fe bridging angles of 103.6 (2)O and a Fen-Fe separation of 3.089 (2) A. Nonequivalent Fe-OH distances of 1.938 (5) and 1.993 (5) A reflect the trans influences of the ring N atom and of the coordinated H20 molecule, respectively. The distorted octahedral coordination geometry of the iron ions is completed by a water molecule and two carboxylate atoms from the tridentate Dipic ligand. [Chel(H2O)FeOH]z.4H20 (Chel or chelidamate = 4-hydroxo-2,6-pyridinedicarboxylate) also crystallizes as green prisms in space group Pi(Cj',No. 2): with Z = 1; a = 7.972 (4), b = 11.106 (4), c = 7.051 ( 3 ) A; 01 = 67.71 (3), p = 112.73 (3), y = 95.81 (4)O; dcalcd = 1.923, dobs,j = 1.92 ( 1 ) g/cm3. Least-squares refinement of 1543 reflections having F2 > 3a gave a conventional R factor of 0.044. The structural parameters of [Chel(H2O)FeOH]r4H20 are nearly identical with those of [Dipic(H2O)FeOH]z; the only significant difference between the valence polyhedra is the slight shortening of the Fe-Fe separation from 3.089 (2) 8, in the Dipic structure to 3.078 (2) A in the Chel structure. In both structures, dimers are linked by hydrogen bonds, although the hydrogen bonding network is more extensive in [Chel(HzO)FeOH]2.4H20. The magnetic susceptibilities of [Dipic(H2O)FeOH]z and [Chel(H2O)FeOH]2.4H20 have been measured over the 85-300 K range. In the context of the H = -2Js1132 spin-spin coupling model, the susceptibility data can be fit by respective J values of -1 1.4 (4) and -7.3 (4) cm-' (31= S 2 = 72, g = 2.00, TIP = 0). The observed extent of antiferromagnetism is discussed in terms of structural and electronic effects. A review of the magnetic properties attributed to the parent Fe(II1) aquo dimer [(H20)4FeOH]z4+is based on those observations.
T h e phenomenon of reduced paramagnetism resulting from the hydrolysis and polymerization of aqueous Fe(II1) has interested research workers for more t h a n 60 yeam3q4 Despite t h e importance of basic polynuclear Fe(II1) aquo species to t h e inorganic, geochemical, and bioinorganic chemistry of iron, their structural, magnetic, a n d electronic properties a r e still neither adequately characterized nor clearly understood. T h e magnetic properties of even t h e simplest polymerization product, t h e so-called aquo dimer of probable structure [(H20)4Fe(OH)2Fe(H2O)4l4+,remain subject to dispute. T h e unavailability of crystalline salts of the aquo dimer has prevented its definitive study by t h e usual combination of x-ray crystallographic and magnetochemical techniques. A number of workers have attempted to determine t h e magnetic properties of this dimer as either a solution species o r a s a n adsorbed species on a sulfonate-type ion exchange Both techniques a r e plagued by the presence of substantial amounts of S = 5/2 Fe(II1) monomers and t h e possible irreversible formation of high molecular weight Fe(II1) polymers. ( I t would appear, however, t h a t a t 25OC the magnetic properties of the polymers and aquo dimer m a y not be grossly d i ~ s i m i l a r . ~While ) all workers have agreed that the dimer exhibits antiferromagnetic spin-spin coupling, estimates of t h e magnetic moment ( p ) per Fe(II1) a t 25OC have ranged from_ Oto m3.7 p ~ In. terms of a model based on t h e usual - 2 J S l - s ~ exchange Hamiltonian, t h e upper figure of -3.7 p~ corresponds to J N -40 cm-I. It is not yet certain whether the -40 cm-' value is in error or is related to the marked variation in J values observed for a series of [ C u 2 ( 0 H ) 2 I 2 + complexes having different C u OH-Cu bridging angles.8 Substantially smaller spin-spin
coupling (10-17 cm-I) has recently been observed for crystalline Fe(II1) chelates containing [Fe2(OH)2I4+ or [Fe2(alkoxo)2I4+ bridging units;9-" however, structural information derived from x-ray diffraction d a t a is available only for two [Fe2(alkoxo)214+ type dimers." W e have chosen to characterize t h e [Fe2(OH)2I4+ unit k d e t a i l in order to determine its magnetic properties and variation, if any, with the geometric and chemical features of t h e bridging a n d nonbridging ligands. W e report here t h e crystal structures and magnetic properties of [ Dipic( H 2 0 ) F e O H I 2 (Dipic = 2,6-pyridinedicarboxylate) and [ C h e l ( H 2 0 ) F e O H 1 ~ 4 H 2 0 (Chel =4-hydroxo-2,6-pyridinedicarboxy late).
Experimental Section (1) Preparation and Characterization of Complexes. The ligand 2,6-pyridinedicarboxylic acid (Aldrich Chemical Co.) was used as received. [Dipic(H2O)FeOH]2 was obtained as green prisms by maintaining equimolar mixtures of 2,6-pyridinedicarboxylicacid, FeC13-6H20, and urea at 9OoC for 24 hr. In a typical experiment, a 1-1. aqueous solution 0.05 M in each reactant yielded 7.5 g (60%) of product which was collected by filtration, washed with H2O (low solubility at 25OC), and air-dried. All solutions of reactants used to prepare the complexes were prefiltered through 0.22 pm pore size millipore filters. A solution of the product in cold aqueous HC104 gave a negative test for CI-. Anal. Calcd for FeC7NH603: Fe, 21.82; C , 32.84; N, 5.47; H, 2.36. Found: Fe, 20.8 (iodometry); C, 32.71; N , 5.48; H, 2.01. Chelidamic acid (Aldrich Chemical Co.) was received as an orange-brown powder and was purified by a variation of a published procedure.I2 Sufficient concentrated NH3 was added to dissolve a slurry of 50 g of the crude material in 350 ml of hot H2O (final pH = 3.5). After the dark brown solution was stirred with 15 g of activated charcoal for 15 min and filtered, the product was precipitat-
Potenza, S c h u g a r , et al.
1 [Dipic(HrO)FeOH]2
and [Chel(H2O)FeOH]2-4H20
1426 Table 1. Crystal Data
___ [Chel(H,O)FeOH J ,.4H,O
[Dipic(H,O)FeOH J Body-centered Primitive
Primitive -____
-__
Space group
2
Z
Ii
1
11.085 (7) 11.001 (5) 7.303 (4) 94.09 (4) 89.87 (10) 96.46 (4)
Pi
1
8.844 (4) 11.001 (5) 7.303 (4) 94.06 (4) 111.56 (10) 131.54 (4) 1.93 (1) 1.927 441.3 17.7 0.71069
882.6 -
Pi
1.972 (4) 11.106 (4) 7.051 (3) 67.71 (3) 112.73 (3) 95.81 (4) 1.92 (1) 1.923 532.0 15.0 0.71069
Transformation matrix: 0
0
1
ed by lowering the pH to approximately 2.5 with concentrated HCI. A repetition of this procedure afforded a light yellow crystalline product which decomposed with gas evolution at 26OOC (uncorrected, lit. 248-25OoC).l2 The preparation of [Chel(H20)FeOH]2-4H20 was similar to that of [Dipic(H2O)FeOH]z; a 1-1. aqueous solution that contained 0.05 M each of FeCI3-6H20, urea, and chelidamic acid yielded 10.1 g (65%) of a green crystalline product which also gave a negative test for CI-. Anal. Calcd for FeC7NH1609: Fe, 18.13; C , 27.30; N , 4.65; H , 3.27. Found: Fe, 18.0 (iodometry); C, 27.72; N, 4.65; H, 3.61. Identical procedures, except for longer reaction periods (several days), were used to prepare the Cr(II1) and AI(II1) analogues of [Chel(HzO)FeOH]2.4H20. On the basis of similar crystal morphologies and x-ray powder diffraction patterns, these three complexes appear to be isostructural. However, these observations are insufficient to ascertain whether the Cr(llI) and AI(II1) analogues crystallize in space group P i (as does the Fe(l1l) complex) or in space group P 1 , The urea hydrolysis procedure was also used to prepare crystalline [Dipic(H2O)Cr(OH)]2 and [Dipic(H2O)AI(OH)]2. These complexes were isostructural (powder diffraction and crystal morphology), but different from the Fe(II1) analogue. However, a crystallographic study of [Dipic(HzO)Cr(OH)]2 showed that its molecular structure is comparable to that reported below for [Dipic( H20)Fe(OH)]2.13 (2) Spectroscopic and Magnetic Measurements. Variable temperature magnetic susceptibility studies were performed in the laboratory of Professor H. B. Gray with a PAR FM-I vibrating sample magnetometer equipped with an Andonian Dewar; the apparatus and techniques used to collect the data have been described and 152 X e1~ewhere.l~ Diamagnetic corrections of 114 X cgs per Fe(1II) were used for [Dipic(H2O)FeOH]2 and [Chel(H2O)FeOH]2.4H20, respectively; these were calculated from Pascal’s constant^.'^ Infrared spectra were measured with a Perkin-Elmer Model 225 spectrometer. Reactions run at ]Loth scale in D 2 0 were used to prepare the deuterated complexes. Microscopic examination of both Fe(II1) complexes indicated that they were obtained as pure single phases. X-Ray powder diffraction patterns of the bulk phases (used for the magnetic and spectral measurements) corresponded to the patterns calculated from the single-crystal data. (3) X-Ray Data and Structure Solution: [Dipic(H2O)FeOH]2. A crystal approximately 0.28 X 0.10 X 0.10 mm was mounted along its a axis in a sealed capillary. Preliminary Weissenberg and precession photographs revealed a body-centered triclinic cell ([I or I ? ) with the systematic absence h k I = 2n 1 . Subsequent axis transformation and solution of the structure revealed P i as the corrcct space group. Unit cell parameters were obtained by a least-squares fit of 33 moderately intense reflections using graphite monochromated Mo
+ +
Journal of the American Chemical Society
+
/
98.6
K a radiation ( A 0.71069 A) and an Enraf-Nonius CAD-3 automated diffractometer. Values of the centered and primitive cell parameters are given in Table I along with the transformation matrix relating bcdpcentered and primitive reflection indices. The density of 1.927 g/cm3 calculated for Z = 1 (primitive cell) agreed well with the value of 1.93 (1) g/cm3 measured by floatation of several crystals in a solution of CBr4 in CC14. Diffractometer data based on the centered cell were collected at 22 =k 1°C. Graphite-monochromated Mo Ka radiation was detected with a scintillation counter and a pulse height analyzer set to admit approximately 90% of the K a peak. A 6-26 scan was used to collect a unique data set to a maximum of 20 = 60°. Reflections with 26 < 4O were shielded by the beam stop and were not recorded. The scan range S was a function of 6 chosen according to S = (1.50 0.1 tan 6)’. Each reflection was scanned before being recorded, and zirconium foil attenuators were automatically inserted if the intensity of the diffracted beam exceeded 6000 counts/sec. A circular aperture 0.5 mm in diameter was placed 4.1 cm from the crystal. Background measurements were made at the beginning and end of each scan with the counter stationary; the total time for background counts equalled the scan time. The scan rate was 1 /6O per second, and each reflection was scanned repeatedly to a maximum of six scans or until 6000 total counts were obtained. Intensities were placed on a common scale by dividing by the number of scans. The intensity of a standard reflection, measured at 50 reflection intervals, was consistent to 1 2 % and showed no significant trend. A total of 2603 independent reflections was collected; these were reduced using the transformation matrix and then corrected for Lorentz and polarization effects. The linear absorption coefficient, p = 17.7 cm-’ for Mo K a radiation, is so small that absorption corrections were not considered necessary for a crystal of the size used to collect data. Standard deviations were assigned to F2 values according to
+
1
u ( F 2 ) = - (N,
+
(0.02Nn)2)1/2 LP where Nt is the total count (scan plus background), N, is the net count, and 0.02 is an estimate of instrumental instability. The overall scale factor was initially estimated using Wilson’s method and was then refined. The structure was solved by the heavy atom method using 1436 reflections with F2 2 3u(F2) and refined using full-matrix leastsquares techniques.I6 We initially assumed that the dimer was centrosymmetric. The space group P i has two general positions, and the presence of one dimer per unit cell requires one monomer per asymmetric unit. Approximate coordinates for the iron atom and two oxygen atoms were obtained from a normal sharpened Patterson map. A difference map based on the iron and oxygen phases revealed the coordinates of the remaining non-hydrogen atoms. With all non-hydrogen scattering matter present, the initial agreement factor R F = zllFd - IFdl/zlFd was 0.21 1. Isotropic refinement was initiated using atomic scattering factors from Cromer and Waber for Fe, 0, N , C, and H.” All atoms were treated as neutral species. Both real and imaginary parts of the anomalous dispersion corrections were applied to Fe.’* Initial refinement was based on F2, and weights were set according to w = I/u2. Three refinement cycles of all atomic positional and thermal parameters reduced R F to 0.104. Two additional cycles of refinement with anisotropic thermal parameters for all non-hydrogenic atoms reduced R Fto 0.076. Further refinement was based on F . A weighting scheme, chosen by an analysis of variance,I9 led to the following assignments for dF0): u(F,) = 1.03 - 0.014)Fd;2.90 < I F 4
< 15.3
o(F,) = 0.054Fd - O.OI2;IFd> 15.3
Two cycles of refinement reduced R Fto 0.064 and R,F = [Zw(FO - F,)2/ZwF,2]1/2 to 0.078. The pyridine ring hydrogen positions were calculated, and hydrogen atoms were included at these positions for further refinement with isotropic temperature factors equal to the overall temperature factor obtained from the Wilson plot. Hydrogen atom positional and thermal parameters were not refined. Two additional cycles of refinement reduced R F to 0.059 and R,F to 0.075. A difference Fourier map at this point revealed
/ March 17, 1976
1427 Table 11. Fractional Atomic Coordinatesa and Thermal Parametersb for [Dipic(H,O)FeOH] and [Chel(H,O)FeOH] ;4H,O
Atom
-2373.8 (10) -1188 ( 5 ) 1621 (5) -1647 ( 5 ) 951 (5) -5713 ( 5 ) -903 ( 5 ) 877 (5) 1999 (6) 4096 (7) 4997 (8) 3811 (7) 1753 (6) 714 (6) 254 (7) 5004 6640 4500 -3600 -1000 -800
1347.9 (7) 3805 (3) 6693 (4) -64 (4) 12 (4) -559 (4) 2007 (4) 3133 (4) 4825 (5) 6015 (6) 5420 (6) 3646 (6) 2563 (5) 5144 (5) 671 (5) 7379 6317 3177 1000 1200 2600
P , , or B P22 [Dipic(H,O)FeOH] , 589.5 (11) 121.3 (8) 48.5 (6) 1326 (6) 205 (5) 79 (3) 3858 (6) 235 ( 5 ) 71 (3) 1145 (6) 215 (6) 67 (3) 121 (4) 3453 (6) 283 (5) -1933 (5) 173 ( 5 ) 98 (4) -1210 (6) 21 1 ( 5 ) 96 (4) 3649 (6) 157 (6) 67 (4) 4647 (7) 141 (7) 60 (5) 6765 (8) 165 (8) 89 (5) 7779 (8) 180 (8) 116 (6) 6660 (8) 170 (8) 104 (6) 4557 (7) 159 (7), 82 (5) 3180 (7) 168 (7) 62 (4) 2967 (8) 184 (7) 74 (5) 7622 1.69 9447 1.69 7497 1.69 3600 1.69 -2000 1.69 -1800 1.69
-1363.2 (9) -3810 (4) -6702 (4) 86 (4) 208 (4) 586 (4) -2007 ( 5 ) -6394 (4) 3766 (5) 308 ( 5 ) -3121 ( 5 ) -4801 (6) -5994 (6) -5371 (6) -3577 (6) -2507 (6) -5 184 (6) -578 (6) -7345 -3044 1110 -1200 -2800 -7400 3800 4800 600 1400
1038.5 (7) 145 (3) 279 (3) 2647 (3) 4679 (3) -191 (3) 1545 (3) 4979 (3) 2478 (4) 2769 (4) 2378 (4) 2055 (4) 2868 ( 5 ) 4084 ( 5 ) 4414 (4) 3523 (4) 721 (4) 3654 ( 5 ) 2568 5318 -365 1900 1000 4900 1900 2400 3500 2700
[Chel(H,O)FeOH] ;4H,O -2042.2 (12) 55.2 (9) -2489 (5j 90 ( 5 ) -2867 ( 5 ) -2128 ( 5 ) -2237 (6) -1229 ( 5 ) -5358 (5) -2346 (6) 2479 (6) 2871 (6) -2299 (6) -2414 (7) -2412 (7) -2297 (7) -2174 (8) -2200 (7) -2600 (7) -2166 (7) - 2490 -1980 -2315 -6000 -6000 -2200 1800 2600 3400 2600
Y
X
z
033
P,7
79.6 (12) 112 (7) 128 (7) 128 (8) 165 (8) 74 (7) 137 (8) 73 (7) 115 (10) 99 (10) 124 (11) 114 (10) 95 (9) 108 (10) 130 (11)
53.8 ( 5 ) 105 (3) 101 (3) 91 (3) 161 (3) 89 (3) 113 (3) 79 (3) 65 (4) 79 ( 5 ) 96 (5) 94 ( 5 ) 90 (4) 83 (4) 91 (4)
129 (2) 190 (10) 216 (10) 213 (10) 283 (12) 132 (9) 148 (10) 261 (12) 276 (14) 307 (14) 113 (11) 96 (12) 125 (13) 121 (13) 143 (14) 100 (12) 104 (13) 140 (14) 125 128 146 142 142 243 267 267 297 297
-1.8 (6) -13 (3) -36 (3) -9 (3) -21 (3) 18 (3) -18 (3) 1 (3) 9 (4) -14 (4) -12 (4) -7 (4) -9 (4) -4 (4) -5 (4) -8 (4) -8 (4) -8 (4) -5 1 11 -9 -9 3 5 5 -13 -13
0,-
-
P, 3 3.4 (6) 18 (4) 22 (4) 12 (4) 49 (4) 6 (4) 41 (4) 15 (4) 27 (5) 14 (5) 38 (6) 37 (6) 32 ( 5 ) 25 (5) 32 ( 5 )
-28 -44 -65 -48 -59 -37 -41 -56 -63
(1) (3) (4) (3) (4) (3) (3) (4) (4) -50 (4) -31 (4) -22 (4) -26 ( 5 ) -28 (4) -22 ( 5 ) -21 (4) -17 (4) -28 (5) - 36 -36 -30 -38 -38 -52 -59 -59
-50 -50 Atomic coordinates are X lo4. b Anistropic thermal parameters are X l o 4 ; the form of the anisotropic thermal ellipsoid is exp[-(p, ,hZ+ P Z 2 k 2+ P J 2 + 2P,,hk + 2P,,hl+ 2P,&)l. the coordinates of the three remaining hydrogen atoms. Two addipendent diffraction maxima. With all non-hydrogen scattering tional cycles of refinement resulted in final values of R F = 0.052 matter present, R Fwas 0.18. Anisotropic refinement using weights and R,F = 0.067. For the last refinement cycle, all positional and derived from counting statistics reduced R F to 0.064. An analysis thermal parameter changes were within their estimated standard of variance then led to the following assignments for a(F,): deviation. A final difference map showed a general background of a(F,) = 1.22 - 0.0301Fd; 2.0 S IFd 6 20.0 0.5 e/A3 and no peaks larger than 0.9 e/A3: all Deaks above background were residuals of known atoms.20 o(F,) = 0.008lFd 0.456;IFd > 20.0 (4) X-Ray and Structure Solution: [Chel(H2O)FeOH]2*4HzO. Data collection and reduction were equivalent to that for [DipHydrogen atoms were located as above. These were added as fixed ic(HzO)FeOH]z with the following exceptions:(l) the crystal size atom contributors to the structure factor and assigned anisotropic was approximately 0.25 X 0.25 X 0.20 mm; (2) the scan range S temperature factors equal to those of the carbon atoms to which was selected as S = (1.20 0.87 tan 0)' based on analysis of repthey are bonded. Five cycles of anisotropic refinement (excluding resentative peak profiles; and (3) cell constants were determined H ) reduced R F and R,F to their final values of 0.044 and 0.049, from a least-squares refinement of 16 moderately intense reflecrespectively. All positional and thermal parameter changes were tions centered on the diffractometer. The calculated density of within their estimated standard deviation for the final refinement 1.923 g/cm3 agreed well with the value of 1.92 ( I ) g/cm3 meacycle. A final difference map showed a general background of 0.4 sured by flotation of several crystals in a solution of CC14 and e/A3 and no peaks larger than 0.6 e/A3; all peaks above backCHBr2CHBr2. Pertinent crystal data for [Chel(HzO)FeOH]z. ground were residuals of known atoms. 4 H 2 0 are shown in Table I . Final atomic parameters for both structures, together with estiThe structure was solved by the heavy atom method using 1543 mated standard deviations from the least-squares refinement, are reflections with FZ 2 3u(F2) from a total of 3329 measured indegiven in Table 11. A view of both dimers, showing the atom num-
+
+
Potenza, Schugar, et al.
/
fDipic(HzO)FeOH]z and [ C h e l ( H ~ O ) F e O H ] ~ . 4 H ~ O
1428
Figure 1. Molecular structure of [Dipic(H2O)FeOH]2showing the atom numbering scheme and molecular structure of [Chel(HzO)FeOH]2.4HzO. The numbering scheme is similar to that for [Dipic(H2O)FeOH]2; the ring hydroxyl atom is O ( 7 ) . Lattice oxygen atoms have been omitted for clarity.
Figure 2. (Top) Stereoscopic packing diagram for [Dipic(H2O)FeOH]2. The origin of the unit cell has been translated ’h along a in order to clarify both the centrosymmetric nature of this dimeric complex and its similarity to the one shown below. (Bottom) Stereoscopic packing diagram for [Chel(H20) FeOH] 2-4H20.
bering scheme, is given in Figure 1. Lists of observed and calculated structure factors are available.20
Description of the Structures Both structures consist of centrosymmetric [ L ( H 2 0 ) F e O H ] 2 dimeric units in which crystallographically equivalent Fe(I1I) ions a r e bridged by two hydroxyl groups. T h e Fe(l1l) ions have a distorted octahedral coordination geometry. Chel and Dipic occupy three coordination sites; the remaining three a r e occupied by the two bridging hydroxyl groups and a terminal water molecule. Although the pyridine rings a r e situated on opposite sides of t h e distorted edge-shared octahedral units in both complexes, there a p pears to be no compelling intramolecular steric grounds for exclusive formation of the “trans” isomer. As a referee as noted, this result m a y be caused by crystal packing forces. A comparison of bond distances and angles for both structures (Table 111) shows that they a r e strikingly similar. T h e largest distortion from octahedral symmetry is evidenced by the angles O( 1)-Fe-0(3) (149.7 (150.6)O) and is d u e to the limited bite of t h e tridentate ligand. This distortion is also demonstrated by deviations from t h e leastsquares planes given in Table IV. Atoms N, 0 ( 6 ) , 0(6’), a n d O(5’) (plane l ) , which d o not involve t h e carboxylate oxygen atoms, show a m a x i m u m deviation from planarity of
Journal of the American Chemical Society
/
98.6
/
0.1 1 A. The Fe(II1) ions lie on plane 1 within experimental error. I n contrast, atoms 0 ( 1 ) , 0 ( 3 ) , 0(5’), and O(6) (plane 3 ) show much larger deviations from planarity. T h e Fe(II1) ions a r e displaced more than 0.15 8, from these planes. T h e Fe-Fe separations in [ C h e l ( H 2 0 ) F e O H ] 2 - 4 H 2 0 a n d [Dipic(H2O)FeOH]2 of 3.078 and 3.089 A, respectively, a r e comparable, although they a r e nonequivalent (3.80) within experimental error. Both separations a r e large enough to effectively preclude significant direct overlap of metal ion orbitals based on t h e ionic radius (0.65-0.79 A) of high-spin six-coordinate Fe(III).21 T h e F e - O H - F e bridging angles (103.6 (103.2)O) in t h e strictly planar [ Fe2(OH)2I4+ units a r e equivalent in both structures, as a r e t h e corresponding F e - O H distances. Thus, the only signific a n t difference between the [Fe*(OH)2l4+ units in both structures is t h e Fe-Fe distance, and this difference is small. Bond distances and angles of the Dipic and Chel ligands a r e similar and agree with those reported by other workers for free and complexed Dipic.22 For this reason, they will not be considered in detail here. W e have been unable to loc a t e prior crystallographic studies of either free or complexed Chel. Since the ring hydroxyl hydrogen a t o m was located on a difference map, and the C ( 4 ) - 0 ( 7 ) bond length
March 17, 1976
1429 Table 111. Bond Distances (A) and Angles (deg) in [Dipic(H,O)FeOHI and [Chel(H,O)FeOHI ;4H,0Q Atoms Fe. . .Fe' Fe-O(1) Fe-0(3) Fe-0(5) O( 1)-C(6) 0(2)-C(6) O(31-W) 0(4)-C(7) 0(7)-C(3) N-C(l) N-C(5) 0(5)-H(0(5)) 0(6)-H(0(6)-1) 0(6)-H(0(6)-2) 0(7)-H(0(7)) O( 8)-H(0(8 )-1) 0(8)-H(0(8)-2) O(9)- H(O(9 k 11 0(9)-H(0(9)-2)
Distance 3.089 (2) 2.078 (4) 2.053 (5) 1.938 (5) 1.261 (5) 1.252 (6) 1.285 (6) 1.225 (8)
Atoms (3.078 (2)) (2.064 (4)) (2.021 (4)) (1.938 (4)) (1.285 (7)) (1.231 (6)) (1.272 (7)) (1.235 (6)) (1.330 (7)) (1.330 (6)) (1.333 (6)) (1.083) (0.902) (0.9 3 8) (0.839) (0.943) (0.805) (0.988) (0.976)
-
1.344 ( 7 ) 1.326 (8) 1.033 0.956 0.782
-
Fe-0(5') Fe-0(6) Fe-N
Distance ~1.993 (5) (1.989 (4)) 2.021 (5) (2.044 (4)) 2.070 (6) (2.057 (5))
C(l)-C(2) C(l)-C(6) C(2)-C(3) C(3) C(4 ) C(4)-C(5) C(5)-C(7)
1.389 (6) 1.501 (8) 1.393 (10) 1.416 (9) 1.379 (6) 1.5 17 (7)
(1.377 (8)) (1.527 (7)) (1.412 (7)) (1.417 (7)) (1.380 (8)) (1.522 (7))
C(2)-H(C(2)) C(3)-H (C(31) C(4)-H(C(4))
1.091 1.094 1.091
(1.083)
-
-
(1.086)
Hydrogen Bonding Contact Distances [Chel(H,O)FeOH] ;4H,O O(1). . . O(8) O(2) ' . O(5) O(2). . O(6) O(4) . . O(7) O(4). . O(9) O(6). . . O(9) O(8) . . . O(9)
[Dipic(H,O)FeOH] O(6). . . O(4) O(6). . . O(2)
2.914 (6) 2.743 (6) 2.691 (6) 2.723 (6) 2.745 (6) 2.610 (6) 2.935 ( 7 )
2.630 (7) 2.721 (7)
~
Angle
Atoms Fe'-Fe-O(1) Fe'-Fe-0(3) Fe'-Fe-0(6) Fe-O( 1)-C(6) Fe-O(3)-C(7) Fe-O(5)-Fe' Fe-N-C(l) Fe -N-C( 3) Fe-N-C(5) 0(1)-Fe-0(3) 0(1)-Fe-0(5) 0(1)-Fe-0(5') 0(1)-Fe-0(6) O( 3)- Fe-0 (5) 0(3)-Fe-0(5') 0(3)-Fe-0(6) 0(5)-Fe-0(5') 0(5)-Fe-0(6) 0(5')-Fe-0(6) O(l)-Fe-N O(3)-Fe-N 0(5)-Fe-N O(5')-Fe-N O(6)-Fe-N
102.9 (2) 102.3 (2) 134.0 (1) 118.9 (4) 119.7 (4) 103.6 (2) 118.2 (4) 176.6 (3) 119.1 (4) 149.7 (1) 108.5 (2) 92.1 (1) 86.5 (2) 101.7 (2) 97.9 (2) 87.8 (2) 76.4 (2) 95.3 (2) 170.7 (1) 75.5 (2) 75.5 (2) 168.0 (2) 92.3 (2) 96.3 ( 2 )
Atoms (103.4 (1)) (100.5 (1)) (131.8 (1)) (119.4 (3)) (120.6 (3)) (103.2 (2)) (119.6 (2)) (175.7 (2)) (118.9 (4)) (150.6 (2)) (110.5 (2)) (90.9 (1)) (86.9 (2)) (98.8 (2)) (97.7 (1)) (89.8 (1)) (76.8 (2)) (93.1 (2)) (168.2 (2)) (75.7 (2)) (75.8 (2)) (168.3 (2)) (93.5 (2)) (97.1 (2))
O(l)-C(6)-0(2) 0(3)-C(7)-0(4) O(l)-C(6)-C(l) 0(2)-C(6)-C( 1) 0(3)-C(7)-C(5) 0(4)-C(7)-C(5) 0(7)-C(3)-C(2) 0(7)-C(3)-C(4) N-C(l)-C(Z) N-C(1)-C(6) N-C(5)-C(4) N-C(5)-C(7) C(3)-0(7)-H(0(7)) C( 1)-N-C(5) C(l)-C(2)-C(3) C(l)-C(2)-H(C(2)) C(2)-C( 1)-C(6) H(C(2))-C(2)-C(3) C(Z)-C(3)-C(4) C(2)-C(3)-H(C(3)) H(C(3))- C(3)-C(4 1 C(3)-C(4)-C(5) C(3)-C(4)-H(C(4 1) H(C(4))-C(4)-C(5 C(4)-C(5)-C(7)
Angle 125.9 7 5 ) 126.0 (4) 115.7 (5) 118.4 (3) 114.2 (5) 119.8 (4)
-
119.6 (3) 111.4 (3) 121.6 (5) 111.1 (3) 122.7 (3) 118.5 (5) 120.8 (6) 129.0 (5) 120.7 (4) 120.6 (4) 120.6 (5) 118.8 (7) 116.9 (6) 119.8 (4) 123.2 (5) 127.2 (6)
(124.9 (5)) (124.8 (5)) (114.1 (4)) (121.1 (5)) (114.0 (4)) (121.2 (5)) (123.6 (4)) (116.6 (4)) (122.4 (5)) (110.9 (4)) (121.4 (4)) (110.3 (5)) (128.8 (4)) (121.4 (5)) (117.2 (4)) (120.3 (5)) (126.7 (4)) (122.4 (5)) (119.8 (5)) (117.8 (5)) (120.8 (5)) (121.3 (5)) (128.3 (4))
Qvaluesfor [Chel$H,O)FeOH];4H,O are given in parentheses. is typical of t h a t for a carbon-oxygen single bond,23 Chel is clearly bonded in t h e enolI2 form. T h e pyridine ring atoms a r e coplanar within experimental error (plane 4 ) . However, rather large deviations from planarity a r e found for t h e entire ligand in both structures (plane 5); these a r e attributed primarily to the nonbonding carboxylate oxygen a t o m s O(2) and 0 ( 4 ) , both of which a r e approximately 0.15 %, removed from the plane of t h e remaining ligand atoms. T h e iron atoms and t h e bridging hydroxyl oxygen atoms show large deviations from this plane in t h e opposite direction; this gives a further indication of the distorted geometry in both complexes. In the Dipic structure, individual dimeric units a r e linked by a hydrogen bonding network involving the coordinated
water hydrogen atoms and t h e carboxylate oxygen atoms of neighboring molecules. In t h e Chel structure, a n extensive hydrogen bonding network, involving t h e lattice and coordinated water molecules, t h e Chel hydroxyl group, and t h e carboxylate oxygen atoms O(2) a n d 0(4), links dimers together. Intermolecular hydrogen bonding distances a r e listed in Table 111, while Figure 2 illustrates the packing for both structures. No evidence was found in either structure for intramolecular hydrogen bonding. Although Dipic and Chel dimers show different intermolecular bonding, this does not appear to affect the structural parameters of the [Fez(OH)2I4+ units significantly; however, hydrogen bonding m a y in part account for t h e rather large deviations of atoms O(2) a n d O(4) from the planes of the dicarboxylate
Potenza, S c h u g a r , et al.
/
/Dipic(HzOJFeOH]z a n d [Chel(H2O)FeOH]2.4H20
1430 Table IV. Deviations from Least-Squares Planes ( I ) Plane defined by N. 0(6), 0 ( 5 ) , O(5')
0.058 -0.058 -0.073
N
O(6) O(5)
Distance from Plane (0.111) O(5') (-0.099) Fe (0.092)
0.073 0.002
(2) Plane defined by 0(1), 0(3), N , O ( 5 ) Distance from Plane O(1) -0.147 (-0,116) O(5) 0.116 O(3) -0.155 (-0.126) Fe -0.05 1 N 0.186 (0.177)
(-0.103) (-0.003)
(0.064) (-0.05 6)
(3) Plane defined by 0(1), 0(3),0(6), O ( 5 ' ) Distance from Plane O(1) 0.344 (-0.358) O(5') -0.313 O(3) 0.327 (-0.336) Fe -0.185 O(6) -0.357 (0.390) (4) Plane defined by N, C(1), C(2), C(3), C(4), C(5) Distance from Plane C(4) 0.001 C(1) -0.010 (-0.002) C(2) -0.004 (0.000) C(5) -0.015 0.020 (0.0002) N C(3) 0.008 ( 5 ) Plane defined by N, C(l)-C(7), 0(1)-0(4) Distance from Plane N 0.089 O(1) 0.023 (0.037) C(l) 0.036 (0.030) O(2) -0.135 O(3) 0.074 (0.024) C(2) 0.022 C(3) 0.041 (0.027) O(4) -0.181 O(7) C(4) 0.059 (0.035) C(5) 0.061 (0.028) Fe 0.247 C(6) -0.021 (-0.012) O(5) 0.779 C(7) -0.010 (-0.013)
(0.303) (0.158)
(0.004) (-0.006) (0.003)
(0.041) (-0.110) (0.035) (-0.1 23) (-0.0 12) (0.199) (0.685)
a Values
b Unit
for [ Chel(H,O)FeOH] ,.4H,O are given in parentheses. weights were employed in the calculation of all planes.
ligands. T o a good approximation, both structures m a y b e regarded a s consisting of virtually discrete and magnetically dilute dimers. Infrared Spectra. Selected O H ( D ) modes of t h e title complexes and of related dimers having C r 2 ( 0 H ) 2 4 + and A12(OH)24+ units a r e presented in T a b l e V . Powder diffraction studies indicate that [Chel(H2O)FeOH]2.4H20 is isostructural with its Cr(II1) a n d AI(II1) analogues. Similarly, [Dipic(HzO)CrOH]2 and [Dipic(H2O)AIOH]2 a r e isostructural, but exhibit a powder pattern different from that of the Fe(II1) analogue. Crystallographic studies of [Dipic(HzO)CrOH]2 show t h a t its molecular structure, but not its crystal structure, is comparable to t h a t found for t h e Fe(II1) ana10gue.l~ T h u s a combination of isotopic and metal ion substitutions m a y be used to identify t h e O H ( D ) modes in the above series of dihydroxo bridged dimers. Except for a few instances in which a mode of interest was obscured by other vibrations, deuteration resulted in a n isotopic red shift of -v"?. F e ( O H ) S 0 4 and F e ( O H ) C r 0 4 a r e isostructural lattice polymers containing chains of hydroxo bridged Fe(II1) ions, e.g., ( F e O H F e O H F e O H ) , . T h e identification of t h e stretching and deformation modes of the bridging OH a t -3500 and -950 cm-], respectively, was facilitated by t h e absence of ligandllattice water.24 Comparable absorptions a r e exhibited by the dihydroxo bridged dimers (Table V ) . T h e series of Dipic complexes exhibits a broad strong absorption a t 3410-3560 cm-l assigned to t h e stretching mode of the bridging O H ; t h e corresponding deformation mode appears a t 900-980 c m - ' . Except for varying degrees of band splitting, these absorptions a r e similar for the Chel complexes. T h e OH stretching modes for [Chel(HzO)F e O H I z . 4 H 2 0 and its Cr(II1) analogue have weak shoulders a t higher energy. However, only a single deformation J o u r n a l of the American Chemical Society
1 98:6 /
mode of t h e bridging OH (900 and 950 cm-I, respectively) was observed for these two complexes. T h e AI(II1) analogue exhibits about equally intense OH stretching modes a t 3540 and 3430 cm-' along with two corresponding O H deformation modes a t 1020 and 950 c m - ] . W e note that t h e Al(II1) analogue was obtained a s a single phase whose powder diffraction pattern essentially was identical with that of [Chel(H2O)FeOH].4H20. Such a comparison will not differentiate between the possible space groups of P Ior Pi for the AI(II1) complex. However, factor and site group considerations indicate that two OH stretching and two O H deformation modes a r e allowed in both space groups. All of t h e complexes exhibit a broad band system centered a t 3000-3 100 cm-' which is assigned to t h e deformation modes of ligand and, for t h e Chel complexes, lattice H 2 0 . T h e corresponding O H 2 deformation mode, expected to appear a t -1600 cm-', is obscured by strong carboxylate absorption in that spectral region. However, a weak absorption a t 1205-1225 cm-] exhibited by all of the deuterated dimers m a y reasonably be assigned to the isotopically shifted deformation mode of the lattice/ligand H20 molecules. Unique to the Chel complexes a r e absorptions a t 36103630 and 1220-1245 cm-' which a r e assigned to the stretching and deformation modes, respectively, of the ring OH group. Thus, the ir spectra reflect the fact that the Chel ligand is present in the enol, rather than keto, form. Magnetic Susceptibility. Magnetic susceptibility d a t a in t h e temperature range 85-300 K, corrected for diamagnetism, a r e presented as leff per Fe(II1) in Table VI. T h e usual spin-spin interaction model based on the exchange Hamiltonian H = -2531.s2 with S1 = S2 = s/2, g = 2.00, and T I P = 0 leads to the relationship: pefj) = 12.006
+ 30z'O + 1 4 ~ +' 5z24 ~ + z28 + 9z'O + 7218 + 5224 + 3228 + 230
55 11
where z = exp[-J/kT].I5 A J value for each temperature (Table VI) was obtained graphically from the experimental Fer?vs. T data and plots of leffZ vs. k T / J calculated using t h e above relationship. Average J values for [Dipic( H * O ) F e O H ] 2 and [ C h e l ( H 2 0 ) F e O H ] ~ . 4 H z O a r e -1 1.4 (4) and -7.3 ( 5 ) cm-I. respectively. A more detailed analysis of the susceptibility d a t a by least-squares fitting techniques supports the assumptions of g = 2.00 and T I P = 0 and yields the same J values. Thus, while both complexes exhibit only modest coupling, t h e coupling constant for the Dipic complex is clearly larger in magnitude t h a n that of t h e Chel complex. Magnetic moments per Fe(II1) calculated using these average J values in the above equation a r e also shown in Table VI. These moments a r e typically within 0.1 W B of those determined experimentally. Although the agreement between observed and calculated moments m a y be improved by using a n additional parameter J' corresponding to a higher order exchange term in the Hamiltonian v(,?1.s2)2], the significance of such a treatment is not clear, and higher order terms were neglected here.25,26
Discussion [Dipic(H*O)FeOH]2 and [Chel(H2O)FeOH]*.4H20 a r e t h e first two complexes for which the F ~ z ( O H ) ~unit ~ + has been characterized by x-ray crystallography. A related complex formulated as [(Pic)zFeOH]z where Pic = pyridine-2-carboxylate has been previously characterized both a s a solution2' and as a crystalline ~ p e c i e s .In ~ contrast to t h e Fe(II1)-Pic system, detailed studies of t h e solution equilibria of the Fe(II1)-Dipic system gave no indication of dimer formation a t 2 5 ° C . 2 7 Although we have observed that [Dipic(HzO)FeOH]2 precipitates from aqueous solutions of NazDipic and Fe(1II) a t 25OC, no further attempts
M a r c h 17, 1976
1431 Table V. Selected OH(D) Ir Modes (cm-I) of M,(OH),'+ Complexes with Dipic and Chel
Mode0 Ring OH stretch Ring OD stretch Ring OH def Ring OD def Bridge OH stretch
M = Fe(II1)
-
[Dipic(H,O)MOH] M = Cr(II1) -
-
-
,
~
M = AI(II1) -
3410 s, b
3470 s, b
3560 s, b
Bridge OD stretch
2530 s
2580 s
2640 s
Bridge OH def
-900 mb
-940 m, shb
Bridge OD def
-680b
-
Ligand-lattice OH, stretch Ligand-lattice OD, stretch Ligand-lattice OD, def
-3000 vb -2250 vb -1215 w, sh
720 m, sh -3070 vb -2350 vb 1225 w
[Chel(H,O)MOH] ,.4H,O __._ M = Fe(II1) M = Cr(II1) M = Al(1II)
-
980 s -C
-3100 vb -2350 vb 1236 w
3630 m, sp 2680 m, sp 1220 m 980 m 3460 w, sh 3400 s 2615 w, sp 2510 s -900 mb
3610 m, sp 2680 m, sp 1230 m 990 m 3520 w, sh 3430 s 2620 w, sp 2550 s 950 mb
695 m
728 m
-3100 vb -2200 vb 1208 w
-3060 vb -2240 vb 1205 w
3620 m, sp 2690 m, sp 1245 m 1000 m 3540 s, sp 3430 s 2620 s, sp 2530 s, sp 1020 m -950 m b -770 m b -690 mb 3070 vb -2200 vb 1210 w
-
QAbbreviationsused are: strong (s), medium (m), weak (w), broad (b), sharp (sp), shoulder (sh), very (v). b Overlaps other modes. Approximate band positions were estimated from spectral changes caused by isotopic substitution. CObscured by other strong absorptions in the 750-800 cm-' region. Table VI. Magnetic Susceptibility Results for [ Dipic(H,O)FeOH] and [Chel(H,O)FeOH] ;4H,O
[Dipic(H,O)FeOH] ,
Table VII. Exchange Interactions in Dihydroxo- and
Dialkoxo-Bridged Fe(II1) Dimers Complex
[Chel(H,O)FeOH] ,4H,O
-J (cm-')a
[Dipic(H,O)FeOH 1 [Chel(H,O)FeOH] ;:4H,O [(Pic),FeOHl , b [(Acac)FeOCH,],C [ (DPM),FeOC,H, j ,d [ (DPM),FeO-n-C,H,j , [ (DPM),FeOCH,j , [ (DPM),FeO-i-C,H,] , [ Fe(SALPA)(SALPA-H) ] ,e [Fe(SALPA)Cl],C,H,CH, ~
299.6 285.9 273.2 260.5 246.5 232.4 217.7 202.2 186.0 168.6 150.0 129.5 106.0 96.0 85.0
4.86 4.82 4.81 4.78 4.73 4.68 4.62 4.54 4.40 4.26 4.08 3.81 3.44 3.23 3.03
12.1 12.0 11.6 11.4 11.3 11.1 10.9 10.8 11.0 11.0 11.0 11.3 11.6 12.0 11.9
4.93 4.88 4.83 4.78 4.71 4.64 4.56 4.41 4.35 4.21 4.03 3.79 3.47 3.30 3.10
5.24 5.21 5.21 5.17 5.13 5.15 5.10 5.06 4.97 4.88 4.73 4.56 4.20 3.94 3.71
7.8 7.8 7.4 7.4 7.4 7.9 6.8 6.6 6.8 6.7 6.9 6.8 7.2 7.7 7.9
5.28 5.25 5.22 5.18 5.15 5.10 5.04 4.97 4.89 4.79 4.65 4.47 4.18 4.03 3.84
11.4 7.3 8 11 11 11 8.5 10 17 17
Ref This work This work 9 10 10 10 10 10 lla llb
acalculated from the -USIS, model for g = 2.00 and TIP = 0. pyridine-2-carboxylate. CAcac = anion of acetylacetone. dDPM = anion of dipivalomethane. CSALPA = dianion of Schiff s base from salicylaldehyde and 3-aminopropanol-1. The ferric ions in this dimeric complex are bridged by two propoxide groups. b Pic =
species yielded J values of ca. - 17 c m - ' . If the differences in the nature of the nonbridging ligands and in the coordiaCalculated in p g at each temperature from the corrected molar nation number of the Fe(1II) ions m a y be ignored, then the susceptibility per Fe(II1) using the formula ueff2= 7.998 Xcorr(T). systematic variation of magnetic behavior with bridging bcalculated in cm-l at each temperature by the procedure described angle observed for a series of C U ~ ( O H ) species* ~~+ is not m the text. CCalculated in p g at each temperature using the exhibited by the two Fe2(OH)14+ and two Fer(propoxequation described in the text. ide)z4+ species described above. However, the number and nature of t h e exchange paths in t h e Cu(I1) and Fe(II1) diwere m a d e t o determine if either of t h e title complexes also mers a r e different.ls W e a r e a t this time unable to account exists as solution species. for the observed variations in weak antiferromagnetic couT h e extent of antiferromagnetism exhibited by the title pling (Table VII) that have been reported for various complexes is comparable to t h a t reported for other Fe(II1) Fe2(OH)24+ and Fe2(aIkoxo)r4+ species. T h e effects of complexes formulated as either F e ~ ( 0 H ) 2 ~or + Fe2(alknonbridging ligands, Fe(II1) coordination number, and Feo x 0 ) 2 ~ +species (Table VII). O f special interest a r e t h e re0 - F e bridging angles d o appear to be magnetically invisible cent crystallographic and magnetochemical studies of two dimeric Fe(1II)-Schiff base complexes which have a F e 2 ( ~ - ~ in the more strongly coupled Fe204* units ( J = -95 cm-1).28 T h e relative importance of such factors in the ropoxide)14+ unit. I n comparison to the title complexes, the Fe2(0H)24+ and Fe2(alkox0)2~+units remains to be estabF e 2 ( ~ r o p o x i d e ) 2 ~unit + containing six-coordinate Fe(II1)' l a lished. was found to have a longer Fe-Fe separation (3.217 (7) A), Consistent with theoretical expectations, magnetic coucomparable F e - 0 distances ((1.988 (20), 1.972 (21); 1.970 pling via the direct overlap of the Fe(ll1) d orbitals does not ( 19), 1.9 19 (1 9) A), and larger bridging angles (1 10.6 (9) appear to be significant. As noted above, one of t h e Fezand 108.2 (9)O). Magnetic susceptibility data obtained a t (pr0poxide)2~+dimers has both a greater Fe-Fe separation 298, 195, and 77 K have been reported, but were not a n a and larger antiferromagnetism than were found for the title lyzed in detail. O u r analysis of these d a t a yielded a J value complexes. Still larger antiferromagnetism is exhibited by of -17 cm-' for g = 2.00 and TIP = 0. T h e Fez(propoxF e r 0 4 + dimers ( J N -95 c m - ' ) 2 s and F e 3 0 7 + trimers ( J ide)24+ unit containing five-coordinate Fe( II1)l Ib was -30 cm-')19 where the Fe-Fe separations a r e 3.6 and found to be structurally similar to the title complexes in re3.32 A, respectively. Similar results have been observed for gard to its Fe-Fe separation of 3.089 (6) 8, and F e - 0 Cr2(OH)z4+ species30 and for various dimeric Cu(1I) carbridging angle of 104.1 ( 6 ) ' . Magnetochemical studies of b o x y l a t e ~ . ~It' appears likely that "direct" spin-spin interthis latter dimer a s both a toluene solvate and an unsolvated Potenza, S c h u g a r , et al.
/
[Dipic(HrO)FeOH]l a n d [Chel(H>O)FeOH]2-4H20
1432 actions m a y be ignored for polynuclear Werner-type complexes of t h e first transition series metal ions. P a r t of our incentive for a crystallographic study of both title complexes was to see if a structural basis for t h e observed difference in their magnetic behavior could be found. T h e electronic effects of substituting OH for H in the 4pyridine position include a n increased basicity of t h e ring nitrogen of Chel relative to t h a t of D i p i ~ . O~n~this , ~ basis, ~ we had expected t h a t t h e F e - O H bond trans to t h e F e - N bond in the Chel complex would be lengthened relative to the corresponding F e - O H bond in t h e Dipic complex. Since t h e paths of exchange coupling should depend on the F e - 0 bond distances in the bridges,26 we had correctly predicted in advance the observed order of exchange coupling (IJ(Dipic)l > IJ(Che1)l). However, the F e - O H and F e - N bond distances and the F e - O H bridging angles in both complexes were found to be identical within experimental error. Therefore, there is no structural support for the above chemical intuition. T h e substitution of a n electron withdrawing group such as halogen, sulfone, etc., in t h e 4-position would allow a greater variation in t h e range of ligand basicities. Since crystallographic and preliminary magnetic studies ( J N -95 cm-I) show the Fe(II1) complex of the 4-chloro derivative to be a F e 2 0 4 + system,34 such a comparison m a y not be feasible for Fe(II1). A more fruitful study m a y be possible for Cr(II1) which has, relative to Fe(III), a larger affinity for N ligands. As noted above, both [ C h e l ( H 2 O ) C r O H ] 2 * 4 H 2 0 and [Dipic(HzO)CrOH]z have the C r ~ ( 0 H ) 2 ~unit. + Finally, we consider t h e saga of t h e aquo dimer [(H20)4FeOH]14+ for which a J value of c a . -40 cm-l has been estimated.' W e see no compelling reasons why t h e aquo dimer should be either structurally or magnetically anomalous to the title complexes. Unfortunately, t h e unavailability of crystalline salts of the a q u o dimer rules out definitive magnetic or structural studies. T h e mineral fibroferrite (Fe203.2SO3.11 H 2 0 ) does bear a compositional similarity to A1203-2SO3.11 H2O which contains discrete [(H20)4A10H]24+ ions (AI-OH-AI angle, 100.4'; AI-AI distance 2.86 A).35 However, the d spacings reported for t h e AI phase a r e different from those reported for fibroferrite36 (and verified independently in this laboratory). T h e large and rapid water solubility of the AI phase in contrast to the poor solubility of the Fe phase furthermore suggests t h a t fibroferrite consists of polymeric rather than dimeric units. T h e title complexes constitute t h e best available models for the magnetic properties of t h e parent Fe(II1) aquo dimer. If the geometries of t h e Fe2(OH)24+ units in these species a r e similar, their magnetic properties will very likely be comparable. Their geometrical equivalence m a y be approached in an indirect way by comparing the features of t h e A12(OH)24+ units in the Chel and Dipic complexes with those reported for the [(H20)4A1(OH)2A1(H20)4]4+ species in t h e A1203.2S03.11H20 phase.35 A final comment concerns the magnetic properties a t 77, 196, and 297 K which have been attributed to the Fe(II1) aquo dimer as an adsorbed species on a sulfonate-type ion exchange resin.7 The bulk susceptibility d a t a (corrected for high-spin Fe(II1) species) were accounted for by a J value for the dimer of -40 to -42 cm-I. W e have recently characterized a basic Fe(II1) sulfate of composition Ks[(H20)3(S04)6Fe30].6H2O which, in analogy to the wellknown trimeric Fe( 111) carboxylates, contains discrete [ ( H20)3(S04)6Fe30]5- units.37 Antiferromagnetic coupling between the crystallographically equivalent Fe( 111) ions is described well by the triangular cluster model for J = -26.0 cm-I, g = 2.00, and T I P = 0. Similar species may also have formed on the sulfonate resin. Analysis of the tri-
Journal of the American Chemical Society
/
98.6
/
mer's magnetism by a model appropriate for a dimer yields, assuming g = 2.00 and T I P = 0, J values which systematically decrease from a value of -33.3 cm-l a t 299.6 K to -24.5 cm-' a t 96 K. T h e J value of c a . -40 cm-' that has been attributed to t h e Fe(II1) a q u o dimer is inconsistent with those found for related Fe(II1) dimers (Table VII) and cannot readily be rationalized by considering contributions from more highly condensed and antiferromagnetic species such as the trimeric Fe3O7+ unit.
Acknowledgments. This research was supported by the Rutgers Computing Center (H.J.S., J.A.P.), the Rutgers Research Council (H.J.S.), a n d a Rutgers Biological Sciences Support G r a n t ( H . J . S . ) . Magnetic measurements were m a d e in t h e laboratory of Professor H a r r y B. G r a y a t t h e California Institute of Technology; H.J.S. thanks Professor G r a y for his hospitality. Supplementary Material Available: structure factor tables for [Chel(HzO)FeOH]z.4H20 and [Dipic(H20)FeOH]~(16 pages). Ordering information is given on any current masterhead page.
References and Notes (1)Author to whom correspondence should be addressed. (2)A. P. Sloan Fellow, 1971-1973. (3)L. N. Mulay and P. W. Seiwood, J. Am. Chem. SOC.,77,2693 (1955). (4)T. G. Spiro and P. Saltman, Struct. Bonding(Berlin), 6, 116 (1969). (5)H. J. Schugar, C. Walling, R. B. Jones, and H. B. Gray, J. Am. Chem. Soc., 69, 3712 (1967). (6)J. Mathe and E. Bakk-Mathe, Rev. Roum. Chim., 11, 255 (1966). (7)T. Nortia and E. Kontas, Suom. Kemlstil. E, 44, 406 (1971). (8) K. T. McGregor, N. T. Watkins, D. L. Lewis, R. F. Drake, D. J. Hodgson, and W. E. Hatfield, Inorg. Nucl. Cbem. Lett.,9, 423 (1973). (9)H. J. Schuaar, G. R. Rossman. and H. B. Gray. J. Am. Chem. Soc., 91, 4564 (1965). IO) C. S. Wu, G. R. Rossman, H. B. Gray, G. S. Wammond, and H. J. Schugar, Inorg. Chem., 11, 990 (1972). 11) (a) J. A. Bertrand and P. G. Elier, lnorg. Chem., 13, 927 (1974);published susceptibility data at 298, 195,and 77 K correspond to J = -17 cm-' for g = 2.00 and TIP = 0. (b) J. A. Bertrand, J. L. Breece, and P. G. Eiler, ibid., 13, 125 (1974). 12) S.P. Gag, 0.Fernando, and H. Fraser, lnorg. Chem., 1, 887 (1962). 13) C . Ou. W. Borowski, J. A. Potenza, and H. J. Schugar, to be submitted for publication. [Dipic(HpO)CrOH]z crystallizes in the monoclinic space grou CZ/,,, with 2 = 2,a = 12.366 (5)A, b = 10.847(7)A, c = 7.162 (4) = 117.51 (3)', and d 0 w = 1.95 (1)g/cm3. RF currently is
0
4.9%. 14) H. J. Schugar, G. R. Rossman, C. G. Barraclough, and H. B. Gray, J. Am. Chem. SOC.,94, 2683 (1972). (15) See A. Earnshaw. "Magnetochemistry", Academic Press, New York, N.Y., 1969,p 77. (16) in addition to local programs for the IBM 360/67computer, local modi-
fications of the following programs were employed: Zalkin's FORDAP Fourier program: Johnson's ORTEP 11 thermal ellipsoid plotting program: Busing, Martin, and Levy's ORFFE error function and ORFLS least-squares programs. (17)D. T. Cromer and J. T. Waber. Acta CrystaIlogr., 18, 104 (1965). (18)International Tables for X-Ray Crystallography, Vol. 111, Kynoch Press, Birmingham, England, 1962.p 201-213. (19)The analysis of variance was performed using the program NANOVA obtained from Professor I. Bernal; see J. S.Ricci, Jr., C. A. Eggers. and I. Bernal, Inorg. Chim. Acta, 6,97 (1972). (20) See paragraph at end of paper regarding supplementary material. (21)R. D. Shannon and C. T. Prewitt, Acta Crystallogr.. Sect. B, 25, 925
(1969). (22)F. Takusagawa, K. Hirotsu, and A. Shimada. Bull. Cbem. SOC.Jpn., 46, 2020 (1973), and references therein. (23)B. R. Penfold, Acta Crystallogr., 6,591 (1953). (24)D. Powers, G. R. Rossman. H. J. Schugar, and H. B. Gray, J. Solidstate Cbem., 13, l(1975). (25)J. S. Griffith, Struct. Bonding(Berlin), 10, 87 (1972). (26)A. P. Ginsberg. lnorg. Chim. Acta Rev., 5, 45 (1971),and references therin.
(27)von G. Anderegg. Helv. Chim. Acta, 43, 1530 (1960). (28)K. S.Murray, Coord. Chem. Rev., 12, l(1974). (29)E. M. Holt, S.L. Holt, W. F. Tucker, R. 0. Aspiund, and K. J. Watson, J. Am. Chem. Soc., 96,2621 (1974). (30)J. T. Veal, W. E. Hatfield, and D. J. Hodgson, Acta Crystallogr., Sect. E, 29, 12 (1973). (31)J. A . Moreiand and R. J. Doedens, J. Chem. SOC.,Chem. Commun., 28 (1974). (32)von E. Blasius and B. Brozio. Ber. Bunsenges. Pbys. Chem., 68, 52 (1964).
March 17, 1976
1433 (33) A. Albert in “Physical Methods in Heterocyclic Chemistry”, Vol. 111, A. R. Katritzky, Ed., Academic Press, New York, N.Y., 1971, Chapter 9. (34) C. Ou, D. Powers, J. A. Potenza, and H. J. Schugar. unpublished results. (35) G. Johansson, Acta Chem. Scand., 16, 403 (1962).
(36) F. Cesbron, Bull. SOC.Fr. Mineral. Cristallogr., 87, 125 (1964); Struct. Rep.. 20, 343 (1956). (37) J. A. Thich, B. Vasiliou, D. Mastropaolo, D. Powers, J. A. Potenza, and H. J. Schugar, to be submitted for publication.
Acid-Base Properties and Gas-Phase Ion Chemistry of (CH3)3B M. K. Murphyla and J. L. Beauchamp*lb Contribution No. 51 I7 from the Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, California 91 125. Received May 28, 1975
Abstract: The ion-molecule reactions of trimethylborane, (CH3)3B, both alone and in mixtures with other molecules, have been investigated by ion cyclotron resonance spectroscopy. Reaction pathways, product distributions, and reaction rate constants have been determined for a variety of reactions involving both positive and negative ions. Where possible thermochemical data have been obtained, including the gas-phase Bronsted acidity, PA[(CH3)2B=CH2-] = 365 & 5 kcal/mol. Attempts to determine the Bronsted base strength were frustrated by failure to observe protonated (CH3)3B. Exothermic proton transfer reactions lead exclusively to the formation of (CH3)2B+ and CHI. The thermoneutral CH3- transfer reaction (CH3)2I0B+ (CH3)3I1B* (CH3)31°B (CH3)2I1B+ is observed to be rapid with kf = k , = 2.6 X cm3 molecule-’ sec-I. Transfer of D- and F- from appropriate reagent anions to (CH3)3B produces the four-coordinate anions (CH3)jBDand (CH3)3BF-, which are discussed in light of the electron pair acceptor capabilities of the vacant B 2p valence orbital in the neutral. Photoionization efficiency curves for ions generated in (CH3)3B between 9.7 and 10.8 eV photon energies have been obtained, yielding the adiabatic IP[(CH3)3B] = 10.01 & 0.02 eV, the fragmentation threshold, AP[(CH3)2B+] = 10.35 f 0.05 eV, and the B-CH3 bond dissociation energy in the parent radical ion, D[(CH3)2B+-CH3] = 7.8 f 1 kcal/mol.
+
+
Ion cyclotron resonance ( I C R ) mass spectrometry provides a tool uniquely suited to the detailed study of ionmolecule chemistry in t h e gas phase, in t h e absence of complicating solvation phenomena. Recently reported from this laboratory have been t h e reactions of Li+, NO+, and carbonium ions exemplary of their behavior a s Lewis acid^.^-^ Studies of relative Lewis acidities of carbonium ions toward t h e reference bases H-, F-, and Br- have been used t o measure relative carbonium ion stabilities in the gas p h a ~ e . ~A - natural ~ extension of this work is t h e determination of t h e relative Lewis acidities of various neutral acceptors toward these s a m e Lewis bases. Recently reported gas phase studies along these lines have discussed t h e relative binding energies of F- and CI- t o hydrogen halides ( H X where X = F, CI, Br),* compounds possessing t h e hydroxyl functional group (including water, several alcohols, a n d carboxylic acids)9 and a variety of inorganic Lewis acids.I0 As part of a n extensive investigation of the group 3 Lewis acids M X 3 (where M = B, AI, -; X = H, alkyl, halogen) in this laboratory, this report describes I C R and photoionization mass spectrometry ( P I M S ) studies of the thermochemical properties and reactions of positive and negative ions derived from trimethylborane, (CH3)sB. This molecule, unlike m a n y group 3 Lewis acids, exists a s a trigonal planar monomer in the gas phase a t room temperature.” As such, a n investigation of its ion chemistry m a y provide insights into t h e effects of valence shell electron deficiency, resulting from the presence of a vacant B 2 p orbital in the neutral, on ionic reactivity and stability. Investigation of Lewis acidity of boron compounds has been extensive, but has dealt largely with neutral adducts formed between acids BX3 and various n-donor bases (e.g., amines, phosphine^).'^^'^ Previous gas phase studies of t h e ion chemistry of boron containing compounds have considered boron hydrides,14 tri-n-butylborane,I5 boron halides,1° and borazine.I6 Unlike (CH3)3B, interaction of boron atoms via multicenter bonding present in even t h e simplest
boron hydrides complicates efforts to elucidate the effects of a single vacant B 2p 0rbita1.I~S u c h is also the case in t h e monomeric boron trihalides, where t h e B 2p orbital is extensively involved in dative T bonding to t h e halide substituents.l* Previous investigations of ions derived from (CH3)3B utilizing conventional electron impact mass spectrometry deal only with positive ion^.'^-^^ These studies of (CH3)3B entail ionization (IP) and appearance potential (AP) measurements, correlation of fragmentation patterns for structural determination, and t h e effects of isotopic substitution (I.*H, ‘ B , 1 2 , 1 3 C )on fragmentation patterns.20-22 Thermochemical d a t a on ions containing boron a r e limited, with considerable disagreement a m o n g reported values (Table 1).
Experimental Section The instrumentation and techniques associated with ICR spectrometry are described in detail e l ~ e w h e r e .In ~ ~studies . ~ ~ reported here, both a modified Varian V-5900 spectrometer and a larger, high field instrument, constructed in the laboratory are emp l ~ y e dGas . ~ ~mixtures utilized are prepared directly in the ICR cell by admission of the appropriate sample components through separate variable leak valves in a parallel inlet manifold. Absolute gas pressures are determined using a Schulz-Phelps ionization gauge, adjacent to the ICR cell, calibrated searately for each component against an MKS Baratron Model 90Hl-E capacitance man ~ m e t e rA . ~linear calibration of Baratron pressure vs. ionization gauge current affords pressure determinations over a range of lo-’ to Torr. The overall accuracy in pressure measurement for these studies is estimated to be f 2 0 % , and represents the major source of error in reported reaction rate constants. Photoionization measurements of (CH3)3B utilize the CaltechJ P L facility, which has been previously described.25 Pertinent operating conditions include: source temperature, ambient (22OC); ion source sample pressure, 1.6 X 1 0-4 Torr; resolution, 1 A; repeller field, 0.1 V/cm; ion energy for mass analysis, 20 eV. The hydrogen many-line spectrum is utilized as the photon source for the wavelength range studied (1290-1 140 A). Photon intensities are
Murphy, Beauchamp
/
Acid-Base Properties of (CH3)3B
