6640
J. Phys. Chem. 1992, 96,6640-6650
tution. As before, the calculated shifts compare well with the experimental values. In conclusion, the main results of the present theoretical analysis of the vibrational IR spectra of the isolated H3PX and HzPXH (X = 0, S) species may be outlined as follows: (a) Ab initio calculations at the electron correlation MP2/6-31G** and MP2/6-3 1 1G** levels predict accurately (within harmonic a p proximation) the vibrational IR spectra of the molecules in question and also the shifts of the IR wavenumbers of these species upon isotopic substitution. (b) Although the SCF/6-31G** vibrational IR spectra of the molecules under study compare slightly worse with the experimental data, this approximation also might be helpful for prediction of the vibrational spectra of the molecules with phosphorus or/and sulfur, particularly for much larger species. (c) Although the SCF/6-31GS* level is accurate enougb for qualitatively correct prediction of the IR spectra of the phosphines in question, the calculations at electron correlation level are necessary for rigorous prediction of the dipole moments or/and relative stabilities of different tautomeric forms of these molecules.
Acknowledgment. We are very grateful to authors of ref 6 for sending us the experimental data prior to publication and especially to Dr. Zofia Mielke (University of Wroclaw, Poland) for her helpful correspondence. This study was supported by a contract (DAAL03-89-0038) from the Army High Performance Computing Research Center. J.S.K. is also grateful to the Komitet Badan Naukowych (Poland) for support his travel expenses (grant KBN-400789101). We thank the Mississippi Center for Supercomputing and the Minnesota Supercomputer Center for an allotment of computer time for the calculations presented here.
References and Notes (1) Withnall, R.; Andrews, L. J . Phys. Chem. 1987, 91. 784. (2) Andrews, L.; Withnall, R. J . Am. Chem. Soc. 1988.110, 5605. (3) Withnall, R.; Andrews, L. J . Phys. Chem. 1989,92, 4610. (4) Mielke, Z.; McCluskey, M.; Andrews, L. Chem. Phys. Leu. 1990,165, 146. ( 5 ) Mielke, Z.; Brabson, G. D.; Andrews, L. J. Phys. Chem. 1991,95,75. ( 6 ) Mielke, Z. Private information (unpublished results of experimental
studies in the laboratory of L. Andrews, Chemistry Department, University of Virginia, Charlottesville, Virginia); Mielke, Z.; Brabson, G. D.; Andrews, L., manuscript in preparation. (7) Person, W. B.; Kwiatkowski, J. S.; Bartlett, R. J. J. Mol.Srrucr. 1987, 157, 237. (8) Kwiatkowski, J. S.; Lcszczynski, J. J . Mol. Srrucr. (Theochem) 1992, 257, 85. (9) Kwiatkowski, J. S.; Leszczynski, J. Mol.Phys. 1992, 73. (IO) Hehre, W. J.; Radom, L.; Schleyer, P. v. R.; Pople, J. A. Ab Znirio Molecular Orbital Theory; Wiley: New York, 1986. (11) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639. Akrishnan, R.; Binkley, J. S.;Seeger, R.; Pople, J. A. J . Chem.-Phys. 1980, 72, 650. (12) Frisch, M. J.; Head-Gordon, M.; Truclts, G. W.; Foresman,J. B.; Schlegel, H. B.; Raghavachari, K.; Robb, M.; Binkley, J. S.;Gonzalez, C.; Defrces, D. J.; Fox, D. J.; Whiteside, R. A.; Sffiger, R.; Melius, C. F.; Baker, J.; Martin, R. L.; Kahn, L. R.; Stewart, J. J. P.; Topiol, S.; Pople, J. A. GAUSSIAN 90, Revision H, Gaussian, Inc., Pittsburgh, PA, 1990. ( 1 3) Schachschneider, J. H. Technical Report, Shell Development Company, Emeryville, CA, 1969. (14) KuBulat, K. Ph.D. Dissertation, University of Florida, 1989. Unpublished results (private information). (15) Schmidt, M. W.; Yabushita, S.; Gordon, M. S. J . Phys. Chem. 1984, 88, 382. Schmidt, M. W.; Gordon, M. S . Can. J . Chem. 1985, 63, 1609. Schmidt, M. W.; Gordon, M. S. J. Am. Chem. Soc. 1985, 107, 1922. (16) Streitwieser, Jr., A.; Raja, A.; McDowell, R. S.; Glaser, R. J . Compur. Chem. 1987,8,788. (17) Schneider, W.; Thiel, W.; Komornicki, A. J. Phys. Chem. 1988,92, 5611.
Ligand Macrocycle Effects on the Photophysical Properties of Rhodium( II I) Complexes. A Detailed Investigation of cis- and trans-Dicyano( l14,8,11-tetraazacyclotetradecane)rhodium(III)and Related Species Lawrence J. McClure' and Peter C. Ford* Department of Chemistry, University of California, Santa Barbara, Santa Barbara, California 93106 (Received: January 29, 1992; In Final Form: April 17, 1992)
Described are photophysical Propertiesfor the dicyano tetravunacrocycle Rh(II1) complexes tram- and cis-Rh([14]aneN4)(CN),+ and tram-Rh([ 15]aneN4)(CN),+ (I, IV, and 11, respectively, [ 14]aneN4= 1,4,8,1l-tetraazacyclotetradecane,[1S]aneN4 = 1,4,8,12-tetraazacyclopentadecane)plus related compounds. The electronic absorption spectra of the complexes are characterized by broad bands with A- in the near-ultraviolet region (e < 300 M-' cm-') assigned as spin-allowed ligand field (LF) transitions. Emission spectra are characterized by broad, unstructured, visible range bands assigned as spin-forbidden transitions from the lowest energy 'LF states. A Franck-Condon analysis of the emission spectrum based on a two acceptor mode model was performed, and the results were consistent with average bond displacements in the excited state of ca.0.14 A for a FUI-N bond and 0.02 A for the N-H bonds. Both I and I1 luminesce strongly in ambient temperature aqueous solutions, but the measured lifetime (T,,,) of the former proved to be an order of magnitude longer. The r, values were measured over the range 77-300 K, and lifetimes of both I and I1 exhibited temperature-independent (at lower T) and temperatureactivated regions. Notably, the activation energy for nonradiative decay (k,) of the 3LF state of I is significantly larger than that observed for 11. At 77 K,values of T, for the various complexes were found to correlate with the estimated voo of the emission bands, and the isotope effect on the nonradiative rate resulting from D/H exchanging the amine protons was shown to increase with the number of N-H bonds in the complex. Molecular mechanics calculations were also camed out to probe the potential effects of ligand field excitation on the total strain energy of these tetraazamacrocycle complexes resulting from the connectivity between the amine nitrogens. These photophysical properties are discussed in terms of competitive weak and strong coupling mechanisms for nonradiative deactivation.
Introduction In 1983 Kane-Maguire et al., reported that the luminescence from the ligand field (LF) excited state (ES) of the rhodium(II1) complex ion tran~-Rh([l4]aneN~)(CN)~+ (I, [ 14]aneN4 = 1,4,8,1I-tetraazacyclotetradeoine, called ucy~lm",a Figure 0022-3654/92/2096-6640503.00/0
1) is unusually bright for a Rh(II1) amine and displays a lifetime (8.1 PS in ambient aqueous solution, 292 ps in 77 K dimethylfOmamide &sS) several orders of magnitude longer than cornmOdY Observed for Rh(III) amine complexes? In addition I does not show the ligand photolability generally observed for such 0 1992 American Chemical Society
Ligand Macrocycle Effects on Rh(II1) Complexes
n [14]aneN,
n [l5]aneN4
trans-Rh((1 4]aneN4)(CN),+ (I)
Figure 1. Macrocyclic amine ligands used plus 'stick" and space filling strain energy minimized structures of rruns-Rh( [ 14]aneN4)(CN)*+(I) from MM2 calculations using Tektronix CAChe software (see Experimental Section).
complexes. This behavior was attributed to the presence of the strong field CN- ligands along the z-axis of this (essentially) D4* complex, the result being that the lowest lying LF ES has 3A2g symmetry with changes in electron density focused in the xy-plane of the macrocycle. Consequently, nonradiative and photochemical deactivation would be inhibited by the relatively rigid tetraamine macrocyclic ligand occupying the equatorial coordination sites, the unusually long lifetime of I being attributed to the slower nonradiative relaxation of the macrocyclic complex relative to the mono- and bidentate ammine complexes. Studies in this laboratory have long been concerned with the photochemical and photophysical properties of d6transition metal complexe~.~ Many of these have focused on the rhodium(II1) amines, principally the penta- and tetraamine derivatives of Rh(NH3)5L"+and Rh(NH,),b'"+, as prototypes for the behavior of ligand field excited states.4-1° The goal of the present studies is to extend these investigations of characteristics and mechanism@)of the nonradiative decay processes in 3LFstates of Rh(II1) amines by the synthesis of complexes different in macrocycle size and coordination, by investigation of the temperature and solvent dependence of luminescence lifetime, and by calculation of ES geometry and macrocyclic ring strain. These data will be interpreted in terms of the current theories regarding mechanisms for Es decay. Experimental Section Materials. Unless otherwise stated, all chemicals were ACS reagent grade. When used as a solvent, water was first deionized and then distilled from KMn04. Methanol was distilled from calcium hydride under nitrogen. Gases were prepurified and were passed through BASF Deox columns to remove oxygen. Rhodium trichloride was provided on loan by Johnson Matthey, Inc. The ligands [ 14]aneN4, [ 15]aneN4 (1,4,8,12-tetraazacyclopentadecane), and [9]aneN3 ( 1,4,7-triazacyclononane)were purchased from Aldrich Chemicals, Inc., and [ 141pyaneN, (3,7,11,17-tetraazabicyclo[11.3.l]hetadeca-l(17),13,15-triene) was synthesized
The Journal of Physical Chemistry, Vol. 96, No.16, 1992 6641 according to literature methods." Illustrations of these ligands are given in Figure 1. Syntheses. cis-Dichloro( [ 14]aneN4)rhodium(III) chloride, trans-dichloro( [ 14]aneN4)rhodium(III) chloride,I2 bis( [9]aneN3)rhodium(III) chloride,13 and cis- and trans-dicyano([ 14]aneN4)rhodium(III)perchlorate14were prepared according to published procedures. The salt [Rh(Me4en)4](CF3C02)3was provided by Professor A. Ludi of the University of Berne, Switzerland. tranS-Dichlor~([l!$"~)rbodium(III) chloride, This complex was prepared in a manner analogous to the published method for trans-(Rh( [ 14]aneN4)C12)C1.12 Rhodium trichloride (RhC13* 3H20, 0.5 g, 1.9 mmol) and [ 15]aneN4 (0.5 g, 2.3 mmol) were dissolved in methanol (25 mL) and heated at reflux for 2 h during which time the color of the solution turned from deep red to orange. The reaction mixture was filtered hot and HCl was added dropwise to precipitate any unreacted [ 15]aneN4 and/or cis[Rh([15]aneN4)C12]C1chloride. A small amount of light yellow precipitate was removed by filtering with a medium porosity sintered glass frit. The filtrate was evaporated to dryness, and the resulting solid was recrystallized from 3 M HCl and then from 6 M HCl. Yield: 585 mg (73%). Anal. Calcd for [Rh(CllH26N4C12]Cl: C, 29.9; H, 6.40; N, 12.68. Found: C, 29.9; H, 6.42; N, 12.46. b.arrs-Dichloro([141pyaneN4)hOdium(III)Chloride. Rhodium trichloride (200 mg, 0.765 mmol) and [ 14]pyaneN4 (200 mg, 0.849 mmol) were dissolved in methanol (10 mL) and heated at reflux for 2.5 h. A red precipitate (24 mg) formed as the solution turned a deep yellow. Addition of concentrated HCl (1 mL) produced no added precipitate. The filtrate was evaporated to dryness. The resulting solid was dissolved in a minimum of 0.3 M HCl and passed down a Dowex 400W-X2 cation exchange column, but no separation was observed. The product was dissolved in a minimum of hot water and then precipitated as the perchlorate salt by the addition of a small amount of concentrated aqueous NaC104. Yield: 150 mg (39%). Anal. Calcd for [Rh(N4C13H23)C12]C104*0.5H20: C, 30.1 1; H, 4.66; N, 10.80. Found: C,30.19; H, 4.62; N, 10.74. fac-Trichloro(1 , 3 , 9 - t r i a z a c y c l ) M u m (III). Rhodium trichloride (200 mg, 0.765 mmol) was dissolved in methanol (15 mL) and warmed to ca. 40 OC. Triazacyclononane (200 mg, 1.58 mmol) was added, and a yellow precipitate formed immediately. The precipitate was washed with boiling water, boiling methanol, ethanol, and acetone. Yield: 130 mg (51%). Anal. Calcd for [Rh(N3C6H&13]: c, 21.28; H, 4.46; N, 12.41. Found: c, 21.16; H, 4.44; N, 12.34. b.arrs-Dicyano([ls)ewN4)hOdium(III) Perchlorate. (Caution: All perchlorate salts are potentially explosive and should be handled with care.) trans-Dichloro([ 151aneN4)rhodium(III) chloride (200 mg, 0.471 mmol) and NaCN (2.3 g, 46.9 mmol) were dissolved in water (15 mL) and heated at reflux for a p proximately 24 h. Concentrated aqueous NaC104 was added to the solution, and a white precipitate formed. The precipitate was collected on a medium porosity glass frit, washed with ethanol and acetone, and then airdried. A second fraction, collected upon cooling the reaction mixture, was added to the first fraction. The combined fractions were recrystallized by dissolving in a minimum of boiling water, concentrating by heating on a hot plate, and then cooling in an ice bath. Yield: 106 mg (48%). Anal. Calcd for [Rh(CIIH2~N4)(CN)2]ClO4.0.5H20: C, 32.68; H, 5.66; N, 17.58. Found: C, 32.46; H, 5.49; N, 17.27. trarrs-Dicyano([141pyaneN4)rbodium(III)Perchlorate. transDichloro( [ 14]pyaneN4)rhodium(III)perchlorate (1 15 mg, 0.22 mmol) and NaCN (1.33 g, 33.2 mmol) were dissolved in water (10 mL), and the solution was heated at reflux for 12 h. During this time, the yellow solution became colorless. Concentrated aqueous NaC104 was added to the solution, and the resulting white precipitate was collected by suction filtration on a paper filter. The product was washed with ethanol and acetone. A small second crop was collected after cooling in an ice bath. The two crops were combined and reprecipitated by dissolution in hot water and addition of sodium perchlorate. Yield: 50 mg (45%). Calcd for
6642 The Journal of Physical Chemistry, Vol. 96, No. 16, 1992
McClure and Ford
for approximately 2 h. The D 2 0 was removed by evaporation [Rh(N,CI3H23)(CN)2]C104.H20:C, 35.48; H, 4.96; N, 16.55. under reduced pressure. Extent of deuteration was monitored by Found: C, 35.8; H, 4.85; N, 16.48. the appearance of N-D stretching bands in the 2400-cm-’ region ci~-Diammine([14JiI”~)rhodium(III)Perchlorate. cis-Diof the IR spectrum concomitant with the decrease in the intensity chloro([ 14]aneN4)rhodium(III) chloride (750 mg, 1.82 mmol) of the N-H bands in the 3200-cm-’ region. The process of disand trifluoromethanesulfonic (“triflic”) acid ( 10 g, 66.7 mmol) solving, heating, and drying was repeated until deuteration, as were placed in a scnrpulously clean and dry round-bottomed flask measured by the IR spectrum, was complete. Typically, this fitted with a condenser and heated at 110 OC under a slow stream procedure was performed 4 times. of nitrogen for 3 days. Upon addition of the cis-dichloro([ 141Instnunentation. Electronic spectra were recorded on either aneN,)rhodium(III) chloride to the triflic acid, the effluent gas a Hewlett-Packard HP 8452 Model spectrophotometer or a Cary was passed through a solution of 1 M AgN03 and a white pre118C spectrophotometer. In general, spectra were run in a 1-cm cipitate, AgCI, was formed. After a few moments, precipitation quartz cell in aqueous acidic solutions (pH 4, HCIO,). All molar of AgCl was complete. The weighed product, 1 equiv of AgCl, extinction coefficient measurements were repeated at least twice. was the result of reaction of C1- counterion with the triflic acid Infrared spectra were recorded using a Bio-Rad FTS60 FTIR to yield HCl. The effluent gas was then passed through an oil spectrophotometer. IR samples were prepared in a 1% KBr disk, bubbler to maintain a slight overpressure of nitrogen. After 3 pressed in a stainless steel die and recorded at 2 cm-I resolution days, the triflic acid was evaporated under a stream of nitrogen, over 64 scans using a nitrogen background. and liquid ammonia (50 mL) was condensed into the flask. After the ammonia had evaporated (a. 3 h), the off-white product was Emission and excitation spectra were recorded utilizing a SPEX dissolved in water, filtered, and 2 mL of HCl was added. Addition Fluorolog 2 spectrophotometer equipped with a Hamamatsu R928A water-cooled PMT configured for photon counting and of ethanol did not yield a precipitate, so the solution was evapointerfaced with a SPEX Datamate I1 data station. Spectra were rated to dryness under reduced pressure. A white product was corrected for PMT response. Excitation spectra were corrected collected in a slurry of ethanol (100%) on a medium porosity glass for lamp intensity variation by the ratio method with a Rhodamine frit and washed with acetone and ether and air-dried (1.2 8). 6G reference. Spectra of solution samples were recorded in 10 Approximately 500 mg was dissolved in a minimum hot water mm X 10 mm Suprasil fluorescence cells, the solvents being either and 2 mL of concentrated NaC10, was added, yielding a snow water (pH 4) or methanol/water (4/ 1 (v/v)). Spectra of solid white powder. This reprecipitation process was repeated twice. samples were recorded as mixtures in 3% KBr pellets prepared Anal. Calcd for [Rh(N4C10H24)(NH3)2[(C104)2Cl: C, 21.01; by pressing in a hydraulic press. Spectra at 77 K were obtained H, 5.29; N, 14.70. Found: C, 21.08; H, 5.30; N, 14.46. by cooling the samples in a quartz Dewar flask filled with liquid trans-Dia”iae([15~eN4)rhodium(III)Chloride. trans-Dinitrogen. An 0-52 Coming glass filter was used on the emission chloro( [ 15]aneN4)rhodium(III) chloride (220 mg, 0.47 mmol) to remove light source scatter. If the emission intensity was was heated at 110 OC in triflic acid (10 g) for 3 days under a sufficiently weak, a 7-54 Corning filter was used in conjunction steady stream of nitrogen and argon. After 3 days, the triflic acid with wide 5-mm slits on the excitation beam. was removed by evaporation; liquid ammonia (50mL) was then Luminescence spectra and lifetimes were also recorded using added and allowed to evaporate over a 3-h period. The product an AVCO ERL C-950 pulsed nitrogen laser as the 337-nm exwas dissolved in water, and the resulting solution filtered. Apcitation source. Emission was monitored at a right angle to the proximately 3 mL of concentrated HCl was added, and the solution excitation beam. The emitted light was passed through a Fastwas evaporated to dryness under reduced pressure. The red solid ieEbert 0.8-m scanning grating monochromator and then an 0-52 was washed with ethanol (loo%), giving a yellowish white solid. filter to remove laser scatter and was detected by an RCA 8852 The attempt at recrystallization from HCl (3 M) gave a yellow PMT operating at 800-1200 V. The PMT output was monitored solid (unreacted starting material). Ether added to the filtrate with a Princeton Applied Research 4400 boxcar averager with yielded a white solid. Yield: 80 mg (33%). Anal. Calcd for a 4422 grated integrator. Data were collected by scanning over [Rh(N4CI1H26)(NH3)2](ClO4)3*0.5H20: C, 20.09; H,4.65; N, the wavelength of interest at a set time window (generally 100 12.88. Found: C, 19.5; H, 4.50; N, 13.09. ns) after the pulse. For lifetime measurements, data were collected fhms -Dia”ine([ 14]aneN4)rbodium(III)Perchlorate. transat a single wavelength and stored as PMT output vs time signal Dichloro( [ 14]aneN4)rhodium(III)chloride (650 mg, 1.59 mmol) in waveform mode (scanning time gate). A Pacific Photoinand triflic acid (10 g) were heated at 110 “C under a slow steady struments Model 50 PMT provided the trigger pulse for the stream of nitrogen in a scrupulously clean, air-tight, round-botboxcar. tomed flask. The mixture was allowed to evaporate under nitrogen. Lifetime measurements were recorded utilizing a Quanta Ray Liquid ammonia (50 mL) was condensed into the flask and alDCR-1A Q-switched Nd:YAG pulsed laser operating at 10 Hz lowed to evaporate under a slow stream of nitrogen. Another as the excitation source. The system was equipped with an HG-1 50-mL aliquot of ammonia was then added and allowed to harmonic generator and a PHS- 1 harmonic separator to isolate evaporate. The solid product was dissolved in a minimum amount the desired frequency. The method of detection has been described of water and 4 mL of HCl was added. The solution was evapopreviously. l5 rated under reduced pressure and an off-white solid product collected in a slurry of ethanol. The product was washed with Variable-temperature studies were performed using an Air acetone and ether and air-dried (yield, 400 mg). Half of this Products Heli-Trans Model LT-3-110 liquid transfer refrigeration product (200 mg) was redissolved in triflic acid and heated at 110 system linked to a 25-L liquid nitrogen Dewar and a Scientific OC for a further 6 days under a steady stream of nitrogen. A white Instruments Model 5500-1-25 microprocessor-based temperature controller. The temperature controller’s accuracy is rated at f0.5 solid flecked with black and rosy impurities formed. This was K between 35 and 450 K with a resolution of 0.2 K over the same dissolved in liquid ammonia (50 mL) which was then allowed to evaporate. The compound was then dissolved in a minimum of temperature range. The temperature control unit was interfaced with the Heli-Trans unit by a silicon diode thermometer (Model 1 M HCl, filtered, and evaporated under reduced pressure. The S-14566). The laser beam was passed at 45O through round white solid precipitate was washed with acetone and ether and air-dried. (72 mg) The precipitate was dissolved in a minimum Quartz windows spaced at 2 mm. Two cannula were affixed to of hot water and concentrated aqueous NaC10, was added. The the sample cell and led to two Schlenk tubes on the outside of solution was oooled overnight in a refrigerator. The resulting white the cold tip. Samples were syringed into the sample cell of the precipitate was washed with acetone and ether. Yield: 42 mg cold tip from a reservoir of sample in a Schlenk tube that had (8%). Anal. Calcd for [Rh(N4CloH26)(NH~)2](C104)3~H20 C, bem pressurized slightly. 19.2; H, 4.84; N, 13.02. Found: C, 20.09; H, 4.92; N, 12.93. In a typical experiment, the cell was flushed first with water and then 3 times with the solvent or solvent mixture of the exDeuteration of the Amine Hydrogem. Deuteration of the amine periment. Equilibrium at each T was considered achieved by hydrogens was accomplished by dissolving the appropriate complex waiting 15 min after the time the temperature controller reached in a small amount of D 2 0 and heating at 30 OC under nitrogen
Ligand Macrocycle Effects on Rh(II1) Complexes
The Journal of Physical Chemistry, Vol. 96, No. 16, 1992 6643
a steady state at the set temperature. The sample was allowed 5 min between each lifetime measurements at a given T to dissipate any heat due to local heating of the sample by the laser pulses. Lifetimes were measured by scanning both from high to low T and then low to high. Each experimtnt was repeated at least twice. Temperaturedependentmeasurements in the 293-330 K range were also performed with samples in sealed l-cm2 square cells and a water-cooled, constant-temperature heat bath system. Luminescence quantum yields were measured on the Spex Fluorolog 2 by a modified Parker-Rees technique.I6 A 7-54 Corning filter and 1.25 mm slit were used on the excitation beam to ensure spectral purity. An 0-52 filter was used on the emission to reduce the scatter of the excitation source. Anthracene was used as the standard (@ = 0.42).17 The samples were optically dilute (A C 0.1) and solvent blanks were run in identical cuvettes. The spectra were corrected for PMT sensitivity. The working equation is @x = @r(rlx/~y)(Ex/Ey)(Dx/D,)(Ar/Ax), where @ is the absolute quantum yield, q is the solvent refractive index, D is the integrated area under the corrected emission spectrum, and A is the absorbance at the excitation wavelength. The subscripts x and r refer to the unknown and reference, respectively. Integrated intensity was calculated by summing the intensity quanta over the wavelength range according to D = C(dZ/dX) dX. Calculation Methods: Emission Spectrum Fitting. The Franck-Condon band shape analysis of the emission spectra was based on a two acceptor mode in the nonradiative rate expresion.'"-^' Spectra were recorded for frozen solutions at 77 K in MeOH:H20 (4/ 1 (v/v)). After correction for PMT sensitivity and background noise, the intensity measurements as a function of emission energy were transferred to a VAX/VMS 11/750 computer. For the curvefitting routines, between 75 and 85 points in increments of 3 nm were used to define the spectrum. An abscissa linear in energy was obtained by multiplying the intensity values by X2/c.22The data were normalized to I,, = 1 and fitted to the theoretical band shape equation. A program utilizing the Marquadt method, a variation on steeptdescent algorithm, was used to fit the emission spectrum to the band shape equation. It minimized the dimensionless quantity x2, which compares the experimental and calculated values to the standard error of measurement, as a function of fitted parameters. The routines which performed a single iteration of the Marquadt method were obtained from the l i t e r a t ~ r e ?and ~ routines were developed to define the minimization function, call input data, and read the number and initial guess of each variable parameter. Derivatives of the fitting function with respect to each parameter were first obtained by hand calculation and checked using Mathematica software on a Macintosh IIcx computer. Minimization was tracked by flagging the x2 value before and after an iteration. Minimization was considered achieved when the difference between p r e and postiteration values was less than 0.1% for five consecutive iterations. The standard deviation of the intensity incorporated into the x2 value was assigned the intensity reading at 4a from the maximum intensity of the emission curve (based on normal approximation of the emission curve). A second program was written that calculated and plotted both the best fit and experimental spectra. These spectra were then transferred to a Mac IIcx computer and plotted using Cricket Graph software. The line shape fitting equation for emission spectra is21-22 I ( v ) = CumC,l(vm - UmY, - U I Y I / U ~ ) ~ ( S ~x~ ~ / U , ! ) (S"'/ufl) expK-4 In (2)((v - ym + umym + W)/AVA)~I (1) where um is the zero-zero energy of the electronic transition, v, and S, are the energy and distortion factor for the high frequency acceptor, q and SIare the energy and distortion factor of the low energy acceptor, and Aus is the solvent broadening term of the individual vibronic transitions. The terms v, and y are the number of quanta of the high and low energy acceptor modes contributing to the intensity expression. As with all curve-fitting procedures which subject a limited data set to a many parameter fit, care must be taken in the interpretation of the best fit values. Different initial guess values for the parameters were used to ensure that
the program minimized to global rather than local minima. Values to which the parameters converge were restricted to physically possible values. For the Rh(II1) tetraamines, a variety of experimental and theoretical models exist that provide initial guesses of the curvefit parameters. The zero-zero energy of the transition was eatimated according to the 1% rule and the assumption of Gaussian emission band shape (uoo umX + 1 . 2 9 A ~ ~ /The ~ ) . energies ~ of the high and low accepting modes were estimated from IR spectra (for the N-H mode) and typical M-L symmetric stretches (for the Rh-N mode). Estimation of the distortion parameters was accomplished by providing "reasonable numbers" for the following log
(VOO/nNHSmvm)
(In
=
(UNH/~ND)/2~~00)[~ND'NH/uNH - uNDl
+
(2)
where uoo is the 0-0energy, YNH and vND are the energies of the accepting mode and the protio and deuterio analogs. Of these parameters, the thermal parameter is least amenable to prediction, and this was predicted by taking an average value from the current literature.22 For tram-Rh( [ 14]aneN4)(CN),]', the initial values used were uoo = 24000 cm-', v, = 3200 cm-I, u1 = 300 cm-I, S, = 1.5, SI = 0.15, and Aus = 1500 cm-I. Molecular Mechanics Calculations. MM2 calculations were performed using a Tektronix Computer Aided Chemistry (CAChe) software package interfaced to a Mac IIcx computer. The initial guess structure was entered manually by use of the Molecular Editor application of the CAChe system. The structure corresponding to the minimum strain energy was then calculated using either a Conjugate Gradient, Steepest Descent or Blockdiagonal Newton Ralphson minimization technique to minimize the strain energy. Contributing forces in expression for the strain energy, Utot,included bond stretch, bond angle, dihedral angle, improper torsion, van der Waals forces, electrostatic interactions, and hydrogen bonds. Up to 300 iterations of the minimization algorithm were performed during any one experiment with the relaxation factor set at the default value of 1.00, but if minimization was not complete by the 300th update, more iterations were allowed. The convergence criteria for the minimization algorithm was 0.0010 kcal/mol. The parameter list supplied with the software package was developed by Alliinger for use on organic systems2' and updated for use with main group and transition metals. In general, only changes in unstrained bond length, ro, force constant for bond stretch, kf,and force constant for angle bending were necessary to improve correspondence between the calculated and known X-ray structures. The coordinates of the atoms corresponding to the minimum strain energy were written to a molecule file in the Molecular Editor application from which structural data were gathered. The CAChe "bat guess" algorithm to approximate the M-L parameters did not give suitable minimized structures for Rh(II1) amines. For this reason, the parameter file was updated to include specific parameters for Rh(II1) amines and altered to include the effects of coordination on the force constants of the ligand bonds. The ideal unstrained bond length, ro, for amine complexes is generally taken as the length of the M-N bond in the M(NH3)6nC complex, but no structure for a hexaammine Rh(II1) salt is registered in the Cambridge Data SeMce. The Ru-N bond length in [ R u ( N H ~ ) ~ ] C is Ireported ~ to be 2.12 A?' and since a small contraction of the Rh(II1) radius relative to Ru(II1) would be expected, the value of 2.07 A used by H a n ~ o c kfor ~ ~the Rh-N bond was considered reasonable. Values for the force constants were obtained by a 47-body normal coordinate analysis IR analysis by Borch et al.30and Williamson et al.31 The altered parameters include the Rh-N stretch, C-N stretch, and N-H stretch (to account for changes in bond strength upon coordination) and the bond angle bends of C-N-Rh, H-N-Rh, and N-Rh-N. No alteration of the parameters for torsion angle or improper torsion angle was made, since these values are uncertain and their effect upon the minimized geometry is small. Results and Discussion Assignment of the cis and trans isomers was based on the
6644 The Journal of Physical Chemistry, Vol. 96, No. 16, 1992
McClure and Ford
large ligand field splittings are expected only when the macrocycle coordinates to the metal with low steric strain.34 ref PBOtoreactklty. The photochemical consequences of 3 13-nm irradiation in pH 4 aqueous solutions were briefly explored for 2 several of the tetraazamacrocycle Rh(II1) complex ions. Each C of the dicyano complexes investigated, cis- and tram-Rh( [ 141ancN,)(CN),+, trans-Rh( [ 15]aneN4)(CN)2+,and trans-RhC ([ 14]pyaneN4)(CN)2+proved to be photoinert in the context that 2 long-term irradiation led to no changes in the absorption spectra, c and upper limits for quantum yields of disappearance could be 4 estimated to be 1104 mol/einstein. In contrast, the dichloro C analogs proved to be photoactive with absorption spectra changes c 4 consistent with the aquation of a single C1 in each case. Each C of the three trans dichloro complexes gave aCl 0.01, while c cis-Rh([14]aneN4)C12+gave @cI 0.31 (experimental uncerc tainties flOW). The aCl values for the cis- and tranr-Rh([l4]'Resolution of peak maxima of the absorption bands,,,,,,A,, is f2 nm. aneN4)C12+ions agreed with those reported earlier.35 Aquation Experimental uncertainties in the determination of the molar extinction cocffiof NH3 (eA 0.2) was observed for cis-Rh( [ 14]aneN4)(NH3)2+. cients are fS%. 'This work. dRoman numeral used in text to designate the Initial excitation into the 'LF band of a rhodium(II1) amine Rh(II1) complex cation. complex generally is followed by efficient intersystem crossing to the lowest lying triplet state(s) from which the observable photochemical and photoluminescence events O C C U ~ . ~ To - ~ a~ ~ ~ first approximation, the coordination symmetry of the rhodium 1 tetraazamacrocycle complexes tranr-RhN4X4is Ddh. Thus, the two lowest 3LF states are the and 'E with the one electron 0.8 configurations (dxzdyz)3(d,)2(dt) and (dxzdy,)4(d,)1 (d+4', respectively, and split by 35/4Dt, where Dt is Dq(xy) - Dq(z), 0.6 the field strength difference between the equatorial and axial ligands. The ES order depends on the nature of the axial ligands, 0.4 e.g., when X = C1- or NH3, Dt > 0 and the lowest energy ES is the 'E,but when X = CN-, Dt < 0 and the lowest energy ES is the 3A2state. In the former case, axial l a b h t i o n of the complex might be expected, indeed this is the photochemistry observed. In the latter case, equatorial labilization would be predicted by the ES symmetry; however, the nitrogen coordination sites of the 200 250 300 350 400 450 500 macrocycle ligands are mechanically constrained from being reWavelength (nm) leased into solution. Thus, ligand photosubstitutionswould not Figure 2. Electronic spectra in acidic (pH 4 HCI04) aqueous solution: be expected, a prediction consistent with the experimental ob(A) [tr~m-Rh[lS]aneN,C1~](CIO~) (3.3 X lo-' M); (B) [tram-Rhservations. [1S]aneN4(NHp)z](C104)3(6.3 X lo-' M); (C) [tram-Rh[15]aneN4Although the cis-RhN4X2isomers have Cb coordination symM). (CN),](ClO4) (1.0 X metry, two of the axes are identical to each other but different from the third. If the ligand fields on each axis are averaged, assumed stereoretentive substitution reactions of the known cisthe holohedralized symmetry becomes D41,3' thus simplifying and tranr-[Rh([14]aneN4)C12]+starting material whose identities discussion of electronic state symmetries and providing more were established by the characteristic differences in the IR and convenient comparisons with trans analogs. In this context, a electronic spectra. Perchlorate salts of the trans-dicyano complexes strong field X leads to a lowest energy ES having 'E symmetry, of [ 14]aneN4(I),[1S]aneN4(11), and [14]pyaneN4(111) displayed while a X with a weaker ligand field than the N4 macrocycle would a single band in the v(CN) region at 2124,2118, and 2130 cm-I, give a 3A2lowest energy ES. The latter ES would be expected respectively, while the corresponding cis isomer of the dicyano to show equatorial labilization of a monodentate ligand X as was [14]aneN4 compelx (IV) displayed two v(CN) bands', at 2125 observed for both X = Cl- and NH3. It is more difficult to predict and 2134 cm-I. Compounds with no CN groups were identified the reactivity of the former case, although photolabilization of by diagnostic differences in the 790-910 cm-' region of the IR a macrocycle nitrogen along the unique ligand axis would be spectrum and well-characterized differences in the electronic suggested. In this context, the MM2 calculations (see below) show spectra.), that distortion of one of these Rh-N bonds would be more facile Electronic Absorption Spectroscopy. Spectra of the various than would be the case for an equatorial Rh-N bond in the complexes described here are summarized in Table I. Consistent with the spectra of other Rh(II1) amine c o m p l e ~ e s , ~ - ' the ~ ~ ~ ~ ~ analogous ' ~ ~ ' ~ trans isomer; however, the complete absence of speztral changes upon the photolysis of either I or IV indicates that neither broad, w b k (t < 300 M-' cm-') bands in the region 270-405 nm isomer is photoreactive toward net ligand labilization under the (Figure 2) are assigned as Laporte 'forbidden", spin-allowed ligand photolysis conditions. field transitions. More intense bands (a 2 104 M-' cm-I) at shorter Emisdoa Spectril Roperties. The emission spectrum of each wavelengths are probably ligand to metal charge transfer (LMCT) Rh(II1) macrocyclic amine complex in 77 K MeOH/H20 (4/1 in character. The [14]pyaneN4 complexes all possess a strong (v/v)) glasses exhibited a single broad band in the visible or absorption band at ca. 262 nm (a 2 4OOO M-' cm-')characteristic near-IR region (Figure 3). The peak maxima and band widths T * transition. of a ligand localized r The energies of the LF bands of the t r a n r - [ ~ ( [ l S ] ~ e N 4 ) X 2 ] ~ are summarized in Table 11. Spectra were corrected for both background scatter and PMT sensitivity. Also listed are estimates species follow the expected order for various X, i.e., CN- > NH3 of zero-"energies vm for the emitting ES. Marked dependences > CI-. The relative field strengths of the macrocyclic ligands can on ligand X, tetraamine macrocycle size and coordination geombe constructed by comparison of the transdichloro isomers. For etry are evident in the emission spectra in 77 K MeOH/H20 the tranr-RhN4CI2+complexes, values of the lowest energy spin glasses. For the trans-dicyanocomplexes, the emission energies allowed transition range give the order: [ 14]pyaneN4> [ 14]aneN4 fdlow the order [14]aneN4> [15]aneN4> [14]pyaneN4,but little > (en), > (NH,), > [lS]aneN,. Notably, [15]aneN4 appears difference is Seen between the tr~nr-[Rh([14]aneN~)(NH~)~]~+ to have the weakest field of these equatorial ligands despite being (V) and trans-[Rh([ 15]aneN4)(NH3)2]3+ (VI) salts. a secondary amine, a feature which supports the contention that
TABLE I: Eketroaic Spectra of Rh(UI) Tetruzrmrcrocyclic
Comdexea in A a u " Acidic Solution (DH4 HCIOJ commund Am,, (nm)' t (M-' cm-')b rram-[Rh([14]aneN4)C12](C104)~ 404,310 65,75 242,204 3300, 37000 421,303 79,216 tram-[Rh([ 1 S]aneN,)Cl,](CIO,), 248. 208 4100,31000 402. 320 116,525 rranr-[Rh( [ 14]pyaneN4)CIz](CIO,), 260 4500 248, 361 cis- [Rh([ 14JancN4)Clz](ClOd, 350,297 384,360 130,296 ~ac-[Rh([9]aneN,)Cl,](CI04), 270,2200 rram-[Rh([14]aneN4)(CN)2]CI04 (I)d 270,218 rram-[Rh([lS]aneN4)(CN)~]CI04 (11) 290 (broad) 286 317,4200 rranc-[Rh((14]pyaneN4)(CN)2]CI04(111) 294,264 cis-[Rh([ 14]an~N,)(CN)~]C10~ (IV) 284 440 96, 105 rras-[Rh([14]aneN4)(NH,)2](C10,)3 (V) 288,255 114,110 rram-[Rh([1S]aneN,)(NH3)2](C104)J (VI) 310,262 390,290 ~is-[Rh([14]aneN~)(NH,)~](ClO~), (VII) 308,260 ~~
-
-
'
-
-
The Journal of Physical Chemistry, Vol. 96, No. 16, 1992 6645
Ligand Macrocycle Effects on Rh(II1) Complexes
TABLE II: Emission B d Chracteristics of the M c y m rad Meinmine Tetrnaumrcrocycle Rh(III) Complexes in 77 K MeOH/H20 GI.sses
compound ,Y (10’ cm-I)“ A Y , -, ~(10, cm-lIb um 7-d __ (lo3 cm-lIc 21.2 4.4 26.9 286 tram-[Rh( [ 141aneN4)(CN),] (CIO,)’ 18.9 3.9 23.9 56.0 trans-IRh(I151aneN,)(CN),1(C101) tram-iRh( i 14jpyaneN4)(CNi,](~10,) 16.4 3.4 20.7 15.4 21.3 17.7 cis-[Rh( [ 14]aneN4)(CN),](C104) 16.9 3.4 22.7 41.9 [Rh([91aneN3)21(Clo,)3 17.6 4.0 17.4 3.3 22.0 35.6 tram-[Rh([ 14]aneN4)(NH3)2](C104)3 17.2 3.5 21.7 39.7 tram- [Rh([ 15]aneN4)(NHJ2] (C104), cis- [Rh([ 14]aneN4)(NH3),] (C104), 16.2 3.4 20.6 8.0 16.5 3.5 21.0 12.1 [Rh(en(CH,),),I(CF3C02)3 “Energy of emission maxima. Experimental uncertainties are S*lO%. bFull width at half height. estimated according to 1% rule and assumed Gaussian emission band shape (ref 6). dEmission lifetime in CIS. ‘Emission quantum yield at 77 K. ’First reported in ref 2. -I
1 0
3
a
10-
0.8
n.fi
-
erp’t . ewefit
l1
TABLE I11 Best Fit Parameters for Calculated Emission Spectre” complex voo vm VI Sm SI Avs rrrmr-IRhIll4lane-N.IICN~,lICIO.I 22.9 3.22 r r u r - i ~ h i i i ~ j a n e - ~ ; j i ~ ~ j ~20.1 j i ~ ~ 3.45 ~;j 18.9 3.22 Rh([9laneN~)~l(C10,)~ ~is-[Rh([l4]ane-N,)(CN)~](CIO,)18.4 3.52 cis-[Rh([14]ane-N4)(NH3),](CI04), 17.4 3.47
0.656 0.125 4.52 2.17 0.5150 0.105 7.10 2.72 0.505 0.125 6.25 2.85 0.451 0.112 6.79 1.87 0.429 0.110 6.69 2.03
“All energies are 10’ cm-I. The parameters vm, ,,Y and V I are the zero zero energy and the energiea of the high and low energy acceptor modes,respectively. The dimensionlessquantities S, and SI are distortion terms, and Au, is the solvent broadening term for the vibronic transition.
-
dicating that the calculations were settling into global minima rather than local minima. For ~is-[Rh([l4]aneN,)(NH~)~]~+ and Rh([9]aneN3) I1 > IV, the same as found in the E, = (AE- E,J2/4E, calculation described above; however, the magnitudes of these values arc smaller, indeed much smaller for I1 and IV. Given that the orbital parentage of the ligand field ES involves the population of a u-orbital antibonding with respect to the Rh-N bonds, the assumption of equal ES and GS k(Rh-N) values would give an upper limit for the E, to the surface crossing. Smaller values of k(Rh-N) give lower estimates for this E,; indeed use of k(Rh-N) = 0.4 mdyn/A gave the values -0.6, 0.37, and 0.24 pm-l for I, 11, and IV, respectively, values now much more closely approaching the experimental E, values for k, determined from the T dependence of 7,. The quantitative failure of this model is unsurprising, since the GS potential is 50 steep in the curve crossing region that any variation in the ES surface will have major consequences with regard to the curve crossing energy. These will be quite sensitive to the actual Rh-N bond length change in the ES,the size of the force constant used, probable anharmonicities in the Rh-N bond, the possible roles of other vibrations (e.g. the N-Rh-N bend), and mixing between states near their intersection. Thus we conclude that the MM2 modeled electronic states compare favorably with the trend of the experimentallydetermined Ea’s and are consistent with the conclusion that nonradiative deactivation in the temperature-dependent region is dominated by a strong coupling mechanism.
100
3 5 w
80
60
40
20 1 6
1 8
2 0
2 2
2 4
2 6
Rh.N (Angstroms)
F i 6. MM2 calculated molecular strain energies as a function of the Rh-N bond lengths assuming k(Rh-N) is zero and that the four RhN(macrocyc1e) bonds are stretched symmetrically: (a) [tranr-Rh[ 141aneN,(CN),]+; (b) [trans-Rh[ 1 5]aneN4(CN),]+; (c) [cis-Rh[ 141aneN4(CN) ,]+.
and -70 kcal/mol for four. The analogous plots for the cis isomer VI1 show it to be more flexible; stretching all four Rh-N(mac) bonds from 2.15 (the SE minimum) to 2.7 A led to but a 30 kcal/mol increase in strain energy. The situation is different for the [15]aneN4 complexes VI and VIII. For the trans complex VI, the calculations indicated substantial relief of strain as the Rh-N(mac) bonds are stretched from 2.05 to 2.3 A. Thus, a LF ES distorted along these bonds from an estimated GS Rh-N of 2.1 A may have a more favorable size match between macrocycle and metal for a [ 15]aneN4species than for the comparable LFES of the [14]aneN4 analog. The total SE calculated for the cis isomer VI11 was substantially greater than for the trans, even at the calculated Rh-N(mac) minimum (2.3 A), and this may provide a possible explanation for the failure to observe formation of cis isomer in any of the syntheses. A second type of MM2 calculation was used in an attempt to model ground and LF excited state potential surfaces for transand cis-Rh([ 14]aneN4)(CN)2+and trans-Rh( [ 15]aneN4)(CN)2+
Spmmory
It is clear from the above that none of the models discussed give an unequivocal, quantitative explanation of the nonradiative deactivationdynamics observed for the different Rh(II1) complexes examined here. Nonetheless, the remarkably long ligand field excited state lifetimes of the strongly luminescent macrocycle amine complex tranr-Rh([14]aneN4)(CN)2+can qualitatively be 300
trorls-Rhl[14laneN4I(CNl2+
200
100
G.S. (k=1.5) 0
1.6
1.8
2.0
2.2
2.4
2.6
2.8
1.6
1.8
2.0
2.2
2.4
2.6
2.8
Rh-N (A) Figure 7. MM2 calculated molecular strain energies as a function of the Rh-N bond lengths assuming k(Rh-N) = 1.5 mdyn/A for the ground state and the ligand field excited state. Curves on the left represent calculations for I, those on the right represent calculations for 11.
6650 The Journal of Physical Chemistry, Vol. 96, No. 16, 199'2
attributed to two factors. First the ligand field strength of the [14]aneN4macrocycle leads to a higher uoo for the emitting state. As a consequence noruadiative deactivation via both weak coupling and strong coupling mechanisms is decreased relative to analogous Rh(II1) amines including trans-Rh([ 15]aneN4)(CN),+ion. Weak coupling is concluded to be the predominate deactivation pathway in the lower T regime (
