J . Phys. Chem. 1984, 88, 918-923
918
photochemical system consisting of Ru(bpy)32+,MV2+, EDTA, and Pt-based catalysts a t p H 5,36the turnover number for the formation of H2is of the order of 100. Two factors may account for this enhanced efficiency compared to that for the radiolytic system described in this paper: (1) in the photolytic system, the catalyst is not exposed to an ionizing radiation field nor to background H2; (2) in the photolytic system, EDTA is present. While it is known37that ionizing radiation has some effect on the properties of colloids, there seems to be no significant difference in the H2-generating ability of metal sols toward radiolytically or photolytically generated radical^.^ Induction periods for the generation of H2 in the popular photochemical system have been o b s e r ~ e d ~although * , ~ ~ it is not clear what is the relationship between H 2 generation and hydrogenation as a function of exposure time. The presence of EDTA appears to be very important to the stability of the photochemical system, in addition to its role (37) Johnston, F. J.; Ross, A. B.; Helman, W. P. Radiat. Phys. Chem. 169. (38) Moradpour, A.; Amouyal, E.; Keller, P.; Kagan, H. N o w . J . Chim.
1982, 19,
1978, 2, 547.
(39) Kiwi, J.; GrBtzel, M. J. Am. Chem. SOC.1979, 101, 7214.
as a sacrificial electron donor for the reduction of R ~ ( b p y ) , ~ + generated in the excited-state electron-transfer quenching reaction. We have found@that, in the radiation chemical system identical with the one described in this paper except with EDTA having replaced 2-propanol as the H atom and O H radical scavenger, the Pt-catalyzed generation of H 2 at pH 4.6is close to quantitative with the hydrogenation of MV+. minimized. It appears that EDTA acts as a "poison" toward the hydrogenation process thereby enhancing the H 2 yield. These results emphasize that there is a strong dependence of the nature of the catalyst and its specific behavior on the solution medium.
Acknowledgment. We thank Dr. D. Meisel for making ref 24 available to them in preprint form. Registry No. MV2+,4685-14-7; MV'., 25239-55-8; (CH3)*COH, 5131-95-3; Ru(bpy)J2+,15158-62-0; H2, 1333-74-0;Pt, 7440-06-4; 2propanol, 67-63-0; hydrogenated methyl viologen, 27236-70-0. ~
(40) Mulazzani, Q.G.; Venturi, M.; Hoffman, M. 2."Abstracts of Fourth International Conference on Photochemical Conversion and Storage of Solar Energy"; Jerusalem, Israel, Aug 8-13, 1982; p 181.
X-ray Diffraction and Nuclear Magnetic Resonance Studies of the Relationship between Electronic Structure and the Tetragonal Distortion in ZrH, Joseph S. Cantrell,* Chemistry Department, Miami University, Oxford, Ohio 45056
R. C. Bowman, Jr., and D. B. Sullenger Monsanto Research Corporation-Mound,t In Final Form: July 1 , 1983)
Miamisburg, Ohio 45342 (Received: April 19, 1983;
Low-temperature (LT) X-ray diffraction (XRD) studies of 13 ZrH, samples in the range 1.50 5 x 5 2.00 were made by using a Guinier-Simon (GS) focusing camera and other methods. The ZrH, phase diagram was made more complete for the fcc (&phase) and fct (€-phase) regions from ambient to low temperature (i.e,, to about 110 K). No new low-temperature phases were obbserved by XRD. Careful determinations of the composition and temperature dependences of the following parameters were made: c / a ratio, V (unit-cell volume), and Al/(ZAT). When the XRD results are combined with the solid-state proton NMR parameters that reflect the density of electron states at the Fermi level, mutual support is obtained for the Jahn-Teller splitting of the Zr d bands to produce the tetragonal distortion of the ideal fcc lattice when the stoichiometry exceeds x = 1.66. The lattice constants from XRD confirm that the maximum in the values of the proton NMR parameters that are observed when 1.80 5 x I 1.85 is due primarily to the Fermi level falling at the maximum of the lower split d band and is not merely a lattice-temperature effect.
Introduction Phase relationships in the transition metal-hydrogen systems have been extensively studied.'S2 Although many factors can influence phase transformation behavior, the cubic (fcc) to tetragonal (fct) lattice distortions for the nominal dihydrides of the 4B metals (i.e., Ti, Zr, and Hf) appear to be directly associated with a distinctive feature of their electronic ~tructures.~'Namely, the Fermi energy levels, EF,for the cubic dihydrides fall on sharp peaks in the theoretical electron density of states N ( E ) curve^.^-^ A tetragonal distortion that splits the degenerate Fermi level states for the fcc structure can stabilize the distorted hydride phase by means of a solid-state Jahn-Teller effect (e.g., discussions by S~itendick,.~). In the past few years there have been several experimental studies that support the contention that the Jahn-Teller mechanism produces the fcc to fct transition in TiH, and ZrH,. A recent
'MRC-Mound
is operated for the US. Department of Energy under
Contract No. DE-AC04-76-DP00053.
0022-3654/84/2088-0918$01.50/0
summary of those experimental results and the relevant theoretical treatments was given by Bowman et a1.* Through analyses of the composition (Le., x value) and temperature dependences of the proton spin-lattice relaxation times (T,) and proton Knight shifts (uK) from N M R experiments on both cubic 6-phase and tetragonal t-phase ZrH, samples over the range 1.50 5 x 5 2.00, (1) R.L. Beck and W. M. Mueller in "Metal Hydrides", W. M. Mueller, J. P. Blackledge, and G. G. Libowitz, Eds., Academic Press, New York, 1968, p 241. (2) G. Alefeld and J. VBlkl, Eds., "Hydrogen in Metals II", SpringerVerlag, West Berlin, 1978. (3) A. C. Switendick in "Transition Metal Hydrides", R. Bau, Ed., American Chemical Society, Washington, DC, 1978, p 264, Adu. Chem. Ser. No. 167. (4) A. C. Switendick, 2.Phys. Chem. (Frankfurt am Main),117, 89 (1979). (5) M. Gupta and J. P. Burger, Phys. Reo. E, 24, 7099 (1981). (6) M. Gupta, Solid State Commun., 29, 47 (1979). (7) A. Fujimori and N. Tsuda, Solid State Commun., 41, 491 (1982). (8) R. C. Bowman, Jr., E. L. Venturini, B. D. Craft, A. Attalla, and D. B. Sullenger, Phys. Reo. E , 27, 1474 (1983).
0 1984 American Chemical Society
The Journal of Physical Chemistry, Vol. 88, No. 5, 1984 919
Electronic Structure in ZrH, A. FCC structure
I
,
B. Tetragonal Structure ( $ < 1.0) X = 1.80 X = l 9.0.
-v
x=200
n
EF @ X = 1 8 0
Figure 1. Schematic densities of states of d bands for (A) fcc and (B) fct ZrH,. Solid vertical lines are locations of E, from proton (TleT)-1'2 parameters while dashed vertical line in (A) is predicted (ref 5) EF position for fcc ZrH2. A E ( c / a ) is energy difference between N ( E )
maxima. it was proposed* that the sharp peak in N ( E ) a t the Fermi level for 6-ZrH, (theoretical model) is split into a resolved doublet for c-ZrH, as shown in Figure 1. Furthermore, the proton N M R parameters indicated that EF shifts with x to produce a maximum in N(EF) when x N 1.83. An important implicit assumption of that analysis was the absence of both low-temperature phase transitions and large lattice parameter variations. Although there have been several previous systematic room-temperature X-ray diffraction (XRD) investigationsg-" of ZrH, phases, including measurement of their lattice parameters, very few crystallographic studies of ZrH, have been a t low temperatures (Le., between 80 and 300 K). In the present paper we present results from a careful X R D examination of the ZrH, system throughout the composition range 1.50 5 x I1.997 a t ambient and lower temperatures (down to 108 K) to detect lattice changes.
Experimental Details The polycrystalline ZrH, samples that were examined by X R D were portions of the high-purity materials previously used in the recent proton N M R experiment^.'.'^ Zone-refined Zr metal foils and purified hydrogen gas had been used to prepare 13 ZrH, compositions in the range 1.50 Ix I2.00 with maximum uncertainties in x of fO.0 1. Further details on synthesis procedures as well as initial values of the ZrH, phase compositions and room-temperature lattice parameters have been published elsehere.^.^^ The ZrH, foils and powders had been stored under vacuum or purified argon prior to air exposure during the loading of the fused-silica X-ray capillaries. However, several subsequent (9) W. L. Korst, USAEC Report NAA-SR-6880, unpublished. (10) R. L. Beck, Trans. Am. SOC.Met., 55, 542 (1962). (11) K. G. Barraclough and C. J. Beevers, J . Nucl. Mater., 34, 125 (1970). (12) H. L. Yakel, Jr., Acto Crysfallogr., 11, 46 (1958). (13) I. A. Naskidashvili, Sou. Phys.-Solid Sfate (Engl. Trans/.),18, 874 (1976); I. N. Bydlinskaya, I . A. Naskidashvili, V. A. Melik-Shakhnozarov, and V. I. Savin, ibid., 22, 517 (1980); L. S. Topchyan, I. A. Naskidashvili, R. A. Andrievskii, and V. I. Savin, ibid., 15, 1461 (1973). (14) R. C. Bowman, Jr., E. L. Venturini, and W. K . Rhim, Phys. Reu. B, 26, 2652 (1982).
X-ray powder patterns for ZrH, have given no indication of detrimental effects from ambient air exposures for periods up to several months. The N M R experimental procedure has been described in detail elsewhere.' X-ray diffraction studies were made by using a Guinier-Simon (GS) focusing powder camera with a quartz monochromator and an Enraf Nonius cryogenic system for work between ambient temperature and -165 O C (108 K) and is described by Simon.I5 Cooling is by boiling liquid nitrogen and passing a stream of cold gas over the sample. Temperature was controlled and measured near the sample to within f l K. The sample location was found to be 3 K warmer than the position monitored but did not fluctuate; the temperature was measured to within fO.l K and was continuously recorded during all low-temperature (LT) runs. Intermediate temperatures were obtained by using an electrical resistance heater in the low-temperature Dewar tube. The heater was operated by using a Sorenson power supply which was under microprocessor (North Star) control. This experimental assembly allowed all intermediate temperatures to be achieved in the range of ambient to -165 "C (108 K). The temperatures during the X R D measurements were reproducible and accurate to within f l K between 108 and 120 K and f 2 K between 120 K and ambient where a greater tendency to oscillate a t the higher temperatures resulted in the larger error. The L T cryostat was also used with a Debye-Scherrer (DS) camera on a Rigaku rotating anode X-ray generator. This experimental arrangement shortened exposure times from 8-1 0 to 1/2-1 h and resulted in better temperature control. The DS camera recorded high-angle data and thus more accurate lattice parameters were obtained. Ambient X R D data were also recorded on a Norelco diffractometer using thin powder specimens and a focusing geometry which resulted in t w o 4 values with precision of f0.01'. All X R D data were recorded with copper Kcu radiation (A = 1.54178 A for &, and 1.54051 A for a ' ) . The GS camera was designed for multiple exposures by moving the film cassette between exposures. A series of exposures, including an N B S (National Bureau of Standards) Si standard, were recorded on the same film which permitted more accurate determinations of the shifts in lines for a change in composition (x) or a change in temperature for a single x value (film shrinkage and camera alignment, etc., were much more accurately controlled by the use of multiple exposures). Full matrix least-squares procedures were used in all computer analyses of X R D data. Errors were computed from the standard deviations from the least-squares procedures. Multiple runs were used and averaged wherever possible. Severe X R D peak overlap and poorly resolved X R D peaks were omitted whenever feasible and computations were made with and without the suspect values. The unit-cell parameters given in Table I (a, c, c / a , and V') are reported with the quoted standard deviations and represent values that have been duplicated; many were repeated 3 times.
Results and Discussions N M R Results. The schematic N ( E ) curves for 6- and c-phase ZrH, presented in Figure 1 have been based mainly upon the temperature and stoichiometry dependences of proton rKand TI data.' Figure 2 summarizes the ( TI,T)-'/*and uK parameters from these measurements where T I ,is the conduction electron contribution to the proton spin-lattice relaxation time. Within the free-electron approximation16 for hyperfine interactions in transition metals, TI, for protons in metal hydrides can have two contributions' as follows: R ( T ) = (TleT)-' = 4rh?,2kB([Hhl(S) Ns(EF)12$- [Hhf(d) Nd(EF)12ql (1) Here, h is Planck's constant; y , is the gyromagnetic ratio for the resonant nuclei; kB is Boltzmann's constant; N,(E,) and Nd(EF) are the s- and d-band density of states a t the Fermi level, reA. Simon, .! Appl. Crysfallogr.,3, 11, 18 (1970); 4, 138 (1971). A. Narath in 'Hyperfine Interactions", A. J. Freeman and R. B. Frankel, Eds., Academic Press, New York, 1967, p 287. (15) (16)
920
The Journal of Physical Chemistry, Vol. 88, No. 5, 1984
Cantrell et al.
TABLE I: Summary of XRD Parameters for ZrH, Samples Xa
T, K
1.997 1.997 1.997 1.997 1.950 1.901 1.901 1.852 1.852 1.83Se 1.801 1.801 1.801 1.801 1.801 1.776 1.776 1.750 1.750 1.701 1.701
108 123 178 293 293 118 293 108 293 293 108 145 180 220 293 118 293 118 293 108 293
1.701 1.701 1.651 1.651 1.603 1.603 1.552 1.501
108 293 108 293 108 293 293 293
a,
nm
0.498 08 (3f 0.498 11 (5) 0.498 10 (6) 0.498 14 (3) 0.497 93 (3) 0.496 75 (4) 0.496 97 (3) 0.495 80 (3) 0.496 44 ( 3 ) 0.495 97 (3) 0.494 05 (4) 0.494 16 (5) 0.494 25 (6) 0.494 40 (6) 0.494 5 5 (3) 0.492 10 (4) 0.493 00 (3) 0.490 87 (4) 0.491 67 (3) 0.490 35 (5) 0.49053 (4)
c, nm
0.443 36 (3) 0.443 34 (5) 0.443 38 (7) 0.444 31 (3) 0.445 11 (3) 0.444 15 (4) 0.44541 (3) 0.445 48 (4) 0.446 88 (3) 0.448 25 (4) 0.447 91 (4) 0.448 30 (5) 0.448 71 (6) 0.449 20 (6) 0.450 13 (3) 0.44962 (4) 0.452 04 (3) 0.452 20 (4) 0.454 18 (3) 0.454 92 (5) 0.457 OS (3)
c /a
1O6[(Aa/ a)(l/An]
V, nm3
€-Phase (fct) 0.8901 (3) 0.8900 (5) 0.8901 (6) 0.8919 (3) 0.8939 (3) 0.8941 (4) 0.8963 (3) 0.8985 (4) 0.9002 (3) 0.9038 (4) 0.9066 (4) 0.9072 (5) 0.9078 (6) 0.9086 (6) 0.9102 (3) 0.9137 (4) 0.9169 (3) 0.9212 (5) 0.9237 (3) 0.9274 (5) 0.9321 (4) 6-Phase (fcc)
0.109 99 (8) 0.110 00 (9) 0.110 00 (9) 0.110 25 (8) 0.11036 (8) 0.10960 (8) 0.11001 (8) 0.10951 (9) 0.110 1 3 (8) 0.110 26 (9) 0.10933 (9) 0.10947 (9) 0.10961 (10) 0.109 80 (10) 0.11009 (8) 0.108 88 (9) 0.10987 (8) 0.108 96 (9) 0.10979 (8) 0.109 46 (9) 0.109 89 (9)
0.6 (4)c
10.7 (4)c
2.3 (4)
14.8 (4)
0.4 (4)
11.6 (4)
5.1 (4)
24.7 (4)
9.6 (5)
28.0 (5)
8.5 (5)
23.0 (5)
8.8 (5)
27.2 (5)
7c
65
115 K
s
1
GS, DS, diff GS GS. DS, diff GS, DS, diff GS GS GS GS GS, DS, diff
DS GS, DS, diff DS GS, DS, diff GS GS, DS, diff GS GS, DS, diff GS GS, DS, diff GS GS, DS, diff GS, DS, diff GS, DS, diff
Calculated for the largest .60
(TI eT)’ -40
306 K
ic
s 55 v)
- 50 Y-
(17) R.C.Bowman, Jr., and W. K. Rhim, Phys. Reu. B, 24,2232 (1981). (18) R. G6ring, R. Lukas, and K. Bohmhammel, J . Phys. C,14, 5675 (1981).
GS GS GS GS, DS, diff GS, DS, diff
1
60
% where the major influence to the R(T ) temperature dependence is the relative position of E F to a peak. However, if N(EF) varies strongly with an external parameter (e+, an increasing tetragonal distortion from the Jahn-Teller effect), R(T ) will be proportional to the N(EF) changes and eq 2 should make only secondary contributions. The reduction in Nd(EF) is presumably responsible1’J8 for the temperature behavior of the proton uK and (T,,T)3-1/2 parameters in the TiH, when x 2 1.8. The most prominent features of the proton (T,,T)-’/*parameters in Figure 2 are (1) a peak centered a t x N 1.83 and (2) a negative temperature dependence only when 1.80 5 x I1.85. A peak in N(E) can produce the negative temperature dependence in (T,,T)-1/2through eq 2 since d2N(E)/dE2< 0 at E F under this condition. Thus, the ( T1,T)-’/2peak in t-ZrH, should directly correspond to the Fermi level moving past the lower N ( E ) peak
I
I
1”9”6“ K“
methodb
DS
0.477 35 (5) 0.108 77 (9) 0.47841 (4) 0.10950 (9) 11.1 (5) 0.477 51 ( 5 ) 0.108 88 (9) 0.47860 (4) 0.10963 (9) 11.4 (5) 0.477 29 (5) 0.10873 (9) 0.478 07 (4) 0.109 26 (9) 8.2 (5) 0.477 40 (5) 0.108 80 (9) 0.478 57 (4) 0.10961 (9) = H/Zr. GS = Guinier-Simon focusing camera. DS = Debye-Scherrer camera. diff = diffractonieter. AT. Standard deviation of least significant digit. e x = fO.01. spectively; H d s ) is the Fermi-contact hyperfine field for unpaired s electrons at EF; H h d d ) is the core-polarization hyperfine field of spin-paired s orbitals below E F from the unpaired d electrons at EF; and q is a reduction factor for d-orbital degeneracy a t E F as described by Narath.16 Although &As) is positive, Hhf(d) is usually negativeI6 for transition metals. The observeds negative proton uK values for t-ZrH, as well as the APW band theory ca~culations3for fcc ZrH2 indicate Nd(EF) >> N,(EF) in this phase; hence, the second term will dominate eq 1. If EFlies in a region of N(E) that has sharp structure (Le., peaks, etc.), the thermal broadening of the electron distribution with increasing temperature can produce a temperature dependence in R(T ) through the relation
lo6[(AC/
c)(l/Ar)l
-30
h
% .20 ox
E 45
40
35
Figure 2. Composition dependence of proton (T1,n3-’/* and U K parameters for various temperatures. Also given in ref 8 of this paper. when the stoichiometry increases toward x = 2.0 as illustrated in the lower portion of Figure 1. The negative temperature dependences of proton uK for the ZrHI,80and ZrH,,% samples while uK values for ZrHl,a and ZrH,,, are independent of temperature
Electronic Structure in ZrH,
The Journal of Physical Chemistry, Vol. 88, No. 5, 1984 92 1 0.940
0.500 293 K
0 293 K ( R . T . )
r~
0.490
0 108 K (L.T.) 108 K
-
0.930
V KORST (1962) (R.T.)
0.460
0.470
0.460
0.450
r
A BARRACLOUGH & BEEVERS (1970)
0 293 K (R.T.)
108 K (L.T.)
0.920
-
0.910
-
cl a
-
0.440
-
0.430
1
0.900
1
I
1
I
1 '
I
I
1
1
1
1.65
1.70
1.75
1.80
1.85
1.90
0.890'
I
3
Figure 3. Unit-cell axes, c and a, vs. x (H/Zr) for room temperature (293 K) and the lowest temperature studied (108 or 118 K for samples, as indicated in Table I; treated as equivalent). Error bars not indicated
fall within the symbol. are also consistents with EFfalling near the sharp peak only when = 1.85. During the analysis of the proton N M R parameter it was implicitly assumeds that the tetragonal t-ZrH, phase did not undergo any other structural transition in the temperature range 100-300 K. Furthermore, there should be no anomalous temperature dependence to the lattice parameters when 1.80 I x I 1.85 when compared to the other ZrH, compositions. X-ray Diffraction Results. Table I summarizes the XRD results and gives the temperatures studied, the methods used, and the errors (standard deviations) in the least significant digit. The root mean square is used to obtain error values for c/a, V, and Al/lAT. We have carefully examined the 13 ZrH, preparations, 1.501 5 x 5 1.997, for a wide range of temperatures from ambient to 108 K (-165 "C) to check for any new phases or unusual thermal changes in the lattice parameters. N o new phases were found by X R D although hydrogen ordering phase transitions had been implied from mechanical properties experiment^.'^ Other lowtemperature studied4 have not provided support for these proposed hydrogen site ordering transitions. In addition no unusual thermal effects in the lattice parameters were determined. The LT experiments at 108 and 118 K were treated as equivalent for Figures 2-6. The crystallographic parameters given in Table I (a, c, c / a , V, and A l l l A T ) show only a smooth change with either composition variation or temperature variation, as shown by Figures 3-6. The c / a ratio vs. x (Figure 4) has an inflection point near x = 1.83 but the curve is smoothly continuous. The unit-cell volume vs. composition (Figure 5) does not shown any discontinuities for either ambient or L T (108 K) data as the 6 to c phase boundary is crossed. These results are consistent with predictions made for Jahn-Teller distortions in the solid state as given by Bowman et aLS It is not possible from these data to conclude that the 6 to e phase change is second order, but a second-order phase transition model could be consistent with these X R D results. Minor XRD peaks of the y-phase are observed for compositions below x = 1.55 but these data were not included in the results reported here. Comparison between Lattice Properties and Electronic Structure. In the composition range 1.50 5 x 5 1.55 there are two phases,' the y-phase (tetragonal, fct, c > a ) and the &phase (fcc) with the &phase being the predominant phase and the yphase diffraction lines being very weak and probably representing
, I
1.95
2.00
HlZr (atomic ratio) Figure 4. c / a vs. x (H/Zr) for room temperature (293 K) and the lowest temperature studied (108 or 118 K for samples, as indicated in Table I; treated as equivalent). Error bars for these data are indicated but errors are not given for other work. The errors are based upon root mean square standard deviations.
x
0 293 K (R.T.)
o.iiia
108 K (L.T.)
o.iioa
V
(nrn)l
0.109c
0.108( 1 0
1.50
1.60
1.70
1.80
1.90
2.00
HlZr (atomic ratio) Figure 5. Unit-cell volume, V, vs. x (H/Zr) for room temperature (293 K) and the lowest temperature studied (108 or 118 K for samples, as indicated in Table I; treated as equivalent). Error bars are indicated and are based upon root mean square values. only a few percent of the sample. The width of the two-phase region between the fcc &phase and the fct ( c < a ) ephase has not been firmly established although it may be only a few tenths atomic percent in width. Becklo estimates the width of the two-phase region to be 0.1 atomic %. Pressure-concentrationtemperature (PCT)s t ~ d i e sof~ the ~ . ~Zr-0-H ~ system have shown (19) R. K.
Edwards, P. Levesque, and D. Cubicciotti, J . Am. Chem. SOC.,
77, 1307 (1955).
922
The Journal of Physical Chemistry, Vol. 88, No. 5, I984
.
28
1
0 C.
4 33
L
AXIS
a . AXE j . PHASE
1
HlZr (atomic ratio)
Figure 6. A l / l A T for a and c axes vs. x (H/Zr) for room temperature (293 K) and the lowest temperature studied (108 or 118 K for samples, as indicated in Table I; treated as equivalent). Error bars are shown and
are based upon root mean square values. that oxygen concentrations over 3.5 atomic % have significantly affected the boundaries of the 6 e two-phase region. Edwards and Levesque,' showed that oxygen shifts the 6 to c boundary to lower hydrogen content by introduction of a broad two-phase region. However, both Korst9 and Libowitz2, predict boundaries for the two-phase region of approximately 1.64-1.73 d= 0.01 for x in high-purity ZrH,. In the present study the x = 1.701 sample consisted of approximately equal amounts of 6- and €-phaseswhile the x = 1.650 was single phase (6) and the x = 1.750 sample was single phase ( e ) ; hence, the width of the two-phase region must be bracketed by these two compositions. The unit-cell volume decreases continuously as the stoichiometry is reduced from x = 2 to 1.50 with no distinct breaks in the curve a t the phase boundaries for both ambient and L T (108 K) studies. This smooth change in unit-cell volume is consistent with the type of behavior predicted23for the electronic band Jahn-Teller effect. Both the a and c axes do not change very much with increasing x beyond 1.90. This is consistent with the model of splitting the peak in density of states (Jahn-Teller effect) to form a broad "valley" between the two peaks, as shown in Figure 1. The valley corresponds to 1.90 5 x I2.00 and thus the distortion in the crystal lattice occurs primarily in the composition range of 1.65 < x < 1.90. The inflection point on the c / a plot (Figure 4) occurs a t approximately x = 1.83 and the steepest part of this c / a ratio also occurs in the region of the (TleT)-'I2peak shown by Figure 2. The peak in ( T1,T)-'/* vs. x, the inflection point in c / a vs. x, the region of the greatest stoichiometry dependence in the c axis and a axis, and the rapid decrease in the A l / l A T values as x increases above x N 1.80 can be directly related to the Jahn-Teller effect, Le., to the splitting of the degeneracy of the density of electronic states to give a tetragonal lattice with two broad peaks for N ( E ) in the d-band region near E F as represented in Figure 1. If the strength of the Jahn-Teller interaction in c-ZrH, is
+
(20) K. P. Singh and J . G. Parr, Trans. Faraday SOC.,59, 2248 (1963). (21) R. K. Edwards and P. Levesque, J . Am. Chem. Soc., 77, 1312 (1955). (22) G. G. Libowitz, J . Nucl. Mater., 5, 228 (1962). (23) J. Labbe and J. J. Friedel, J . Phys. Radium, 27, 708 (1966).
Cantrell et al. sufficient to produce the resolved N ( E ) peaks schematically shown in Figure lB, the general stoichiometry and temperature behavior of the proton NMR data in Figure 2 and the present lattice parameters can be consistently explained. When x 2 1.7, the Fermi level in the fcc &phase falls far enough into the sharp N ( E ) peak arising from the degenerate d-band states5 that the system is stabilized (Le., total energy is lowered3j4) by a spontaneous tetragonal distortion. Although the energy difference A E ( c / a ) between the N ( E ) maxima in t-ZrH, is certainly dependent upon the c / a ratio, it has not yet been quantitatively determined by either experimental or theoretical methods. However, the photoemission spectroscopy results of Weaver et aI.24imply U ( c / a ) N 1 eV in e-ZrH,. In the Fermi level region the primary effect of varying hydrogen content is e ~ p e c t e d ~to- ~be shifts in E F without major alterations of the N ( E ) peaks. This hypothesis led to assignments of the E F positions in Figure 1B from the proton NMR data.* However, large changes in the c / a ratios can also increase A E ( c / a ) even should the Fermi level remain invariant for different stoichiometries. Since the proton uK and ( TleT)-1/2 parameters are proportional to N ( E F )a t any particular composition, they cannot readily distinguish between these alternative mechanisms. The negative temperature dependences of the proton gKand (TleT)-'12 when 1.80 Ix I1.85 has been attributed8 to thermal broadening of the electron distribution function at a sharp peak in N(E) that was essentially independent of the c / a ratio. The measured c / a parameters in Figure 4 support this assumption. For example, since the (TIeT)-lI2 peak for 293 K occurs a t x = 1.83 where c / a is about 0.904, the same c / a ratio at 108 K would occur near x = 1.8 1. If N(EF)depends primarily upon the absolute c / a ratio, Figure 4 indicates the (Tl,T)-'12 peak a t 108 K would be observed at x = 1.81. Yet, Figure 2 clearly shows that the ( TleT)3-1/2 maximum remains in the immediate vicinity of ZrH,,,, between 300 and 115 K. Similar arguments also hold at the other t-ZrH, compositions. Hence, variations in the c / a ratio cannot be the major factor responsible for the ( TleT)-'12peak near x = 1.83. However, a secondary relationship apparently exists between the lower N(E) peak and the temperature dependence of the €-phase lattice parameters. For example, the values of Ac/(cAT) and A u / ( a A v decrease dramatically for stoichiometries above x N 1.80. Since EF lies above the peak for x 2 1.85, thermal effects in the electron distribution should have only minor effects on lattice parameters through any additional effects of the Jahn-Teller mechanism. However, any factor that increases electron population a t EF (Le., the narrowing of the Fermi distribution function with decreasing temperatures) should enhance the Jahn-Teller energy level splittings whenever x lies below the peak since more lower energy states result from an increase in tetragonal distortion. In particular, the c axis should decrease more rapidly as the temperature is decreased. This view is supported by the data in Figures 3 and 6 where the largest temperature dependences occur for 1.70 Ix 5 1.80. The smaller volumes at lower temperatures (see Figure 5 ) cause greater orbital overlap which should also enhance the Jahn-Teller tetragonal distortion as the temperature decreases. This latter contribution may account for the relatively large Ac/(cAT) values as x approaches 2.0. Barraclough and be ever^,^ consider the effect of filling the tetrahedral intersticies of the &phase (defect fluorite structure) with hydrogen atoms beyond the composition ZrH1,66 to cause a martensitic transformation to the t-phase which then exists up to the stoichiometric composition ZrH,. Labbe and Friede123 proposed that the martensitic transition is driven by the electronic band Jahn-Teller effect. Bowman et aL8 have shown, as illustrated by Figures 1 and 2 in this paper, that the proton NMR pardmeters are consistent with the Jahn-Teller model and the present XRD data are also consistent with this model. According to Anderson and Blout,26 Landau's general theory of phase transitions states (24) J. H. Weaver, D. J. Peterman, D. T. Petersone, and A. Franciosi, Phys. Reo. E, 23, 1962 (1981). (25) K . G. Barraclough and C. J. Beevers, J Less-Common Met., 35, 177 (1974). (26) P. W . Anderson and E. I. Blout, Phys. Reo. Lett., 14, 217 (1965).
J. Phys. Chem. 1984, 88, 923-933
923
Conclusion that second-order transitions usually involve some change in internal symmetry other than mere strain. The usual martensitic This work supports the Jahn-Teller model in the following ways: transitions are first order and probably involve only strain and (1) the volume change is smooth and continuous across the 6 to thus do not have other driving factors such as loss of center of e phase boundary and no breaks or discontinuities occur in Figure symmetry. Beck,1oWhitwham?’ and Moore and Young28question 5 of unit-cell volume vs. x; (2) the absence of any low-temperature the postulated martensitic mechanism for the 6-c transition in phase changes or transitions indicates that the ( TleT)-’I2 vs. x ZrH, because of the observed constancy of the unit-cell dimensions plot is due to a peak in N ( E ) , density of states, and not due to in the two-phase region as shown by Moore and Young28and by any further modification of the crystal lattice; (3) there are no Sidhu et al.29 Typically, martensitic transitions show unit-cell breaks or unusual temperature effects that could have caused the dimensions that are composition dependent.30 However, the ZrH, N M R results reported by Bowman et a1.;8 (4)there is an inflection system shows such a narrow two-phase region that it is very in the c / a vs. x plot near x = 1.83 as noted in Figure 4;( 5 ) the difficult to determine unit-cell dimensions as a function of comregion near x = 1.83 has the greatest stoichiometric dependence position in the two-phase region. As mentioned previously at the in the c axis and the a axis; and ( 6 ) the Al/IAT values rapidly beginning of this section, the reported wide two-phase region may decrease as x increases above x = 1.83. Therefore, it is the well be due to oxygen contamination and thus cannot be relied conclusion of this study that the Jahn-Teller effect is a suitable upon to answer this question. It appears that this system undergoes model to explain the lowering of the fcc lattice symmetry to fct a pseudomartensitic transition that is first order with a very narrow as hydrogen is added to the host lattice. two-phase region. In addition, this system is probably quite Acknowledgment. We thank E. L. Venturini for the synthesis sensitive’ to oxygen impurities which can alter its phase diagram and has led to the rather large reported d i f f e r e n c e ~ ~ - ~in~the . ~ ~ * * ~of high-purity ZrH, materials. We also thank A. Attalla and B. D. Craft for sample handling and assistance with N M R meaextent of the mixed (6 + e) region. Additional measurements on surements. For cryogenic X-ray equipment we thank E. Jendrek, carefully characterized samples are required to resolve the specific C. Hudgens, D. Etter, and C. Wiedenheft. In addition, we thank roles of oxygen (and, perhaps, other impurities) on the 6 to e phase L. Smith for some X-ray measurements and E. Jendrek for the transition. computer programs used to analyze the X R D data. J.S.C. was partially supported by the U S . Department of Energy through (27) D. Whitwham, Mem. Sci. Rev. Metall., 57, 1 (1960). the Mound Research Participation Program. This work was also (28) K. E. Moore and W. A. Young, J . Nucl. Murer., 27, 316 (1968). supported by the Chemical Sciences Division, Office of Energy (29) S. S. Sidhu, N. S. Satya Murthy, F. P. Campos, and D. D. Zauberis in ‘Nonstoichiometric Compounds”, R. Ward, Ed., American Chemical SoResearch, U S . Department of Energy, and by a Faculty Research ciety, Washington, DC, 1963, p 87, A h . Chem. Ser. No. 39. Committee Grant from Miami University. (30) G. M. Wolten in ‘Encyclopedia of X-rays and Gamma Rays”, G. L. Registry No. ZrH,, 1110.5-16-1. Clark, Ed., Reinhold, New York, 1963, p 580.
Organic Photovoltaic Cells. Correlations between Cell Performance and Molecular Structure D. L. Morel,* E. L. Stogryn, A. K. Ghosh, T. Feng, P. E. Purwin, R. F. Shaw, C. Fishman, EXXON Research and Engineering Company, Linden, New Jersey 07036
G . R. Bird, and A. P. Piechowski Department of Chemistry, Rutgers, The State University of New Jersey, New Brunswick, New Jersey 08903 (Received: December 6, 1982; In Final Form: June 15, 1983)
A systematic study of the influence of chemical structure on the performance of merocyanine dyes as solar energy converters has been made. Key correlations between current-generating capability and chemical properties such as chain length, electronegativity, and acid strength have been established. While these correlations are helpful in leading to systematic improvements in performance, certain chemical group substitutions give rise to disproportionate increases signaling the simultaneous advancement of several underlying mechanisms. While the highest sunlight efficiencies achieved to date have been with binuclear structures, certain trinuclear structures have been developed which exhibit broader spectral width and higher quantum efficiency than their binuclear counterparts. The inherently higher solar efficiency potential of devices made from these structures has not yet been realized, however, because of instability problems. Squarylium dyes also exhibit performance indicative of potential high efficiency. Broad spectral widths occur for certain structures due to aggregate formation, High quantum efficiency and open circuit voltage, however, seem to correlate with narrow solid-state absorption bands rather than with broadened aggregate-induced absorption.
Introduction Over the years, organic materials in thin film form have been used extensively in the fields of photography and electrophotography. Each of these processes involves the collection and use of light energy. The production and movement of charge as an *Present address: A R C 0 Solar Industries, P.O. Box 4400, Woodland Hills, CA 91365.
0022-3654/84/2088-0923$01.50/0
integral part of these processes suggests a close similarity to the photovoltaic process. The possibility of using organic materials for the production of low-cost, efficient solar cells is an intriguing one. Color film technology offers an appropriate model for a manufacturing scheme. Color film making requires tight manufacturing tolerances, and yet color film can be mass produced at relatively modest costs. The preparation of a photovoltaic cell is less elaborate than for a color film (3-4 layers rather than 9-10),
0 1984 American Chemical Society
