J. Phys. Chem. 1984, 88, 4497-4502 rotational cooling of the nozzle expansion and are isotropically distributed, since there is no preferred orientation. The orbital angular momentum is given by L = pr X vrel = p b X vrel, where p is the reduced mass and b is the impact parameter. L must be perpendicular to vrel, and thus the distribution in L has only cylindrical symmetry about vrel. Our data indicate that the cross section for the reaction drops rapidly as the energy is decreased toward the threshold. We hope to measure this explicitly in the near future. At low energies the reaction takes place at small b (small L), and the distribution in d will be roughly spherically symmetric and so will the product distribution. As the energy is raised, the reaction takes place at larger b so that the distribution in J is constrained to be more nearly perpendicular to vreI. If the complex dissociates so that a large fraction of d goes into the final orbital angular momentum L', then L' will be roughly perpendicular to vRI. The final relative velocity vie{ must be perpendicular to L'. A careful analysis1° of the probabilities shows that, in the limit where L' = L, the C M cross section will be proportional to (sin e')-', where 8' is the angle between vreI and vre,', that is, Pc will be largest near vIel, as we observe at higher energies. It is no great surprise that the reaction occurs by way of a long-lived complex because the product ions are attracted by a strong Coulombic force. Using a model of spherical ions with radii of 2 A each, with the charge localized at the centers, we find that the Coulomb well is roughly 3.5 eV deep. Because of the long range of the Coulomb potential, there can be no centrifugal barrier in the exit channel. It is not likely that there is a chemical (electronic) barrier or surface crossing either. It is not clear yet what the surface looks like on the entrance side of the well. We find that the reaction is rougly 3-eV endothermic from reactants to products which would indicate that the reaction is exothermic (10) W. B. Miller, S. A. Safron, and D. R. Herschbach, Discuss. Faraday SOC.,44, 108 (1967).
4497
from reactants to the bottom of the well. This is substantiated by Arnett's measurements in solution.6 When the product ions are solvated, the dissociation energy of the complex is much less than 3.5 eV, and the overall reaction is exothermic for tertiary and secondary carbocations and endothermic for primary ions. The measure of product angular and energy distributions gives information on the product side of the potential-energy surface. There are several beam experiments which probe the reactant side. One of these is the measurement of the cross section vs. translational and vibrational energy of the reactants. We have done similar experiments on another system and hope to start them on the present system soon. We have additional, preliminary evidence that an ionic-ovalent surface crossing occurs in the reactant side of the surface. We have found several organic bases which undergo charge exchange with SbF, SbFS + B
-
SbF5-
+ B+
(7)
In a few cases the charge exchange and the halide abstraction reaction both occur. This strongly suggests that the reaction starts with an electron jump to form an ion pair (RX+)(SbF,-) held in a Coulomb well. This may dissociate back to reactants or to the ion pair, or it may exchange a halogen to give R+ and SbF5X-. In the case of benzyl and benzoyl chloride the R+ ion is very stable, and (1) and (3) are the only reactions seen. The reactions of SbF5 and organic halides form one of the few examples of a gas-phase chemiionization reaction that involves a chemical exchange, rather than just charge transfer. The reactions exhibit complicated, yet tractable dynamics.
Acknowledgment. Research support from the National Science Foundation under Grants CHE-7826137 and CHE-8201164 is gratefully acknowledged. Registry No. SbF5, 7783-70-2; C6HSCOC1,98-88-4; Sb2F,,,, 8904329-8; C6H5CHSC1, 100-44-7.
Photoionization of Isolated Nickel Atom Clusters E. A. Rohlfing, D. M. Cox, and A. Kaldor* Corporate Research-Science Laboratories, Exxon Research and Engineering Company, Annandale, New Jersey 08801 (Received: March 23, 1983)
We have used a combination of photoionization and time-of-flight mass spectrometry to obtain a coarse-grained measure of the ionization potential of nickel clusters (Nix) for n = 2-23 as a function on ionizing laser intensity and frequency. The IPSof Nix do not monotonically approach the work function as x increases but exhibit an oscillation as a function of cluster size. Changes in the intensity dependences of the Nix+ signals are interpreted with the aid of two proposed mechanisms.
Introduction Considerable research, both theoretical and experimental, has been carried out to characterize reactions with, by, and on supported transition-metals clusters.' However, the intrinsic properties of isolated naked metal atoms clusters (M,) in the size range from two to several hundred atoms have been elusive to obtain. This size range is of considerable scientific importance because the electronic, structural, magnetic, and chemical properties are expected to change from predominately molecular to bulk in character as the clusters increase in size. Most experimental work on bare metal atom clusters to date has, of necessity, been focused upon low boiling point (e2000 "C) metals (alkalies,2 Pb,'Sb? ( 1 ) T. N. Rhcdin and G. Ertl, Ed., "The Nature of the Surface Chemical Bond", North Holland, Amsterdam, 1979.
0022-3654/84/2088-4497$01.50/0
14) since they can be easily vaporized in a hot oven and formed into a molecular beam for subsequent study. More recently metal cluster beams for somewhat higher boiling point (e3000 "C) metals have been produced by using an improved design5 for a (2) ):( E. Schumacher, W. H. Gerber, H. P. Harri, M. Hoffman, and E. Scholl, Metal Bonding and Interactions in High Temperature Systems", J. L. Gole and W. C. Stwalley, Ed., American Chemical Society, Washington, DC, 1982, ACS Symp. Ser. No 179, p 83. (b) A. Hermann, E. Schumacher, and L. Woste, Helu. Chim. Acta, 61, 453 (1978). (c) M. M. Kappes, R. W. Kung, and E. Schumacher, Chem. Phys. Lett., 91,413 (1982), and references therein. (3) K. Sattler, J. Muhlbach, and E. Rechnagel, Phys. Reu. Left.,45, 821 (1980). (4) A. Hoareau, B. Cabaud, and P. Melinon, Surf.Sci., 106, 195 (1981). ( 5 ) S. J. Riley, E. K. Parks, C. R. Mao, L. G. Dobo, and S.Wexler, J . Phys. Chem., 86, 3911 (1982).
0 1984 American Chemical Society
4498
The Journal of Physical Chemistry, Vol. 88, No. 20, 1984 Vaporlzatlon Laser Beam
I
Ion
I
Vertical Dellectlon
(6) (a) T. G. Dietz, M. A. Duncan, D. E. Powers, and R. E. Smalley, J . Chem. Phys., 74, 6511 (1981). (b) D. W. Powers, S. G. Hansen, M. E. Guesic, A. C. Pulu, J. B. Hopkins, T. G. Dietz, M. A. Duncan, P. R. R. Langridge-Smith, and R. E. Smalley, J . Phys. Chem., 86, 2556 (1982). (c) D. L. Michalopoulos, M. E. Guesic, S. G. Hansen, D. E. Powers, and R. E. Smalley, Ibid.,86, 2556 (1982). (7) (a) V. E. Bondybey, J . Phys. Chem., 86, 3396 (1982). (b) J. L. Gole, J. H. English, and V. E. Bondybey, Ibid.,86,2560 (1982). (c) V. E. Bondybey and J. H.English, J . Chem. Phys., 76, 2165 (1982).
Rohlfing et al. to approximately a 1-mm-diameter spot onto a rod inside a face plate attached to the front of the pulsed valve. The laser is timed to fire at the peak of the helium intensity in the nozzle throat and creates a plasma which is quenched and cooled by the helium gas. The metal target rod of 1/8 in. in diameter is continuously rotated and translated in order to continually expose a fresh area to the vaporizing laser. If this is not done, sufficient material is vaporized within a few laser pulses to dig a pit and the metal cluster signals significantly drop. Cluster formation occurs as cooled metal atoms aggregate in the high-pressure region while He provides the third-body collisions necessary to remove excess energy. The metal atom clusters are vibrationally and rotationally cooled in the adiabatic expansion process. The extent of cluster formation can be controlled by varying the length of the nozzle channel after the target rod and before the expansion into the vacuum. In our Nix experiments, clusters ranging in size from 2 to 30 atoms were produced from a combination of the faceplate channel (2 mm diameter by 6 mm long) and an extender channel (1.5 mm diameter by 6.25 mm long). Because of the long channel used the Nix clusters emerge from the nozzle at the same time, travelling near the H e beam velocity and all arrive at the detection region within a few microseconds of each other. The vaporization laser has a temporal pulse width of -25 ns, and typical laser energies incident on the target rod were 15-20 mJ/pulse of 0.532-pm light plus approximately 5 mJ of residual 1.06 pm from the pump laser. The metal cluster/He beam expands into a molecular beam apparatus that has been described previ0us1y.l~ The beam is skimmed twice by two 5-mm-diameter conical skimmers located 20 and 70 cm from the nozzle. The photoionization region, defined by the point where the ionizing laser crosses the molecular beam axis, is 85 cm from the nozzle. Two laser systems were used to produce photoions in these experiments. The first laser, a Lambda-Physik excimer laser (Model EMGlOl), was operated on the ArF line at 193 nm (6.42 eV) and the KrF line at 248 nm (4.99 eV). The excimer laser beam was apertured to produce a 1.5 mm high by 10 mm wide rectangular spot on the molecular beam. A Quanta-Ray PDL-WEX system was used to obtain tunable UV light; this beam was focused with a 29-cm focal length cyiindrical lens which produced a roughly rectangular spot at the molecular beam. The length of this spot was always -3 mm, and the height could be varied from less than 10 bm to about 300 pm, depending on the distance from lens to molecular beam. For intensity dependencies, the ionizing laser beams were attenuated with one or more neutral density filters and the laser pulse energy was measured with an energy meter. Photoions produced are mass analyzed with a time-of-flight mass spectrometer (TOFMS) which has been described in detail elsewhere.* Typically, the repeller (lower plate) is held at +3200 V and the extractor (upper plate) is held at +2600 V. Since all of the metal clusters are travelling near or at the He beam velocity (1.8 X lo5 cm/s), the higher mass cluster ions have appreciable kinetic energy in the beam direction. Two parallel plates above the extractor deflect the ions and compensate for this component of translational energy. This device is effectively an energy selector and therefore the optimum (maximum ion signal) deflection voltage is mass dependent and scales roughly as TOFMS signals are amplified and then digitized by a transient digitizer (LeCroy Model 2256AS) and all data acquisition and processing are handled by a CAMAC based MIK 11/23 processor (Standard Engineering). For typical mass spectra, 100 pulses are averaged to improve signal to noise. Results and Discussion A typical TOF mass spectrum for Nix clusters is presented in Figure 2. The experimental conditions are given in the figure caption. In addition to the Nix+ peaks, Ni,O+ and Ni,02+ are the two satellite peaks associated with each Nix+peak. The exact origin of the oxide peaks is unknown. Although we have used ultrahigh-purity helium and continually purge the nozzle during (8) T. G. Dietz, M. A. Duncan, R. E. Smalley, D. M. Cox, J. A. Horsley, and A. Kaldor, J . Chem. Phys., 77, 4417 (1982).
The Journal of Physical Chemistry, Vol. 88, No. 20, 1984 4499
Photoionization of Isolated Nix
15000
-a g
-6
106
"f
10000
I
5000
0
5
10
20
15
25
30
NIX Cluster Sire (Atoms)
Figure 2. The photoionization time-of-flight mass spectrum for Nix clusters is shown for two vertical deflection plate voltages using the ArF laser line (6.4eV). The upper curve was taken with 500 V and the lower one with 800 V on the vertical deflection plates. From studies of ion signals as a function of deflection plate volate we find 500 V maximizes ion signals for masses near 530 amu (-Ni9) and 800 V maximizes the 1300 (-Niz2) ion signal. The two scans in this figure have been normalized to their respective ArF laser intensities, 0.6 MW/cmZ for the upper curve and 0.8 MW/cm2for the lower curve. In addition, the lower curve has been multiplied by a factor of 5 . 14
I
I
I
1
12 l3
1
i
:
I
- 301 108
' '
'
"""
(b) N14
t
1
'
" I " '
I,/@
'
'
"""
1
I
I
102
10'2
io.' ArF Laser Energy (mJ)
10
IO'
lti'
io
ArF Laser Energy (mJ)
Figure 4. The Nix+ dependence on laser ArF energy is shown for (a) Ni3+,(b) Ni4+,(c) Ni6+,and (d) Ni12+.The straight lines shown are the integer slopes (1 or 2) which best represent the data; they are not fitted lines.
h 20
10
15
Cluster Size (Atoms)
Figure 3. A composite Nix mass spectrum is shown in which the ion
signals for the individual Nix+clusters have been adjusted to account for the deflection plate voltage. The ArF ionizing laser intensity was about 40 kW/cm2. a run, we have not been able to completely eliminate the oxide peaks. We speculate that the oxide peaks may be due to residual oxygen impurity in the helium source gas, or a surface oxide on the Ni target rod, or chemical "gettering" of residual oxygen in the vacuum system by the highly reactive bare Nix clusters. Whatever the explanation, oxide peaks have also been observed ~ F e z clusters. for C U , ~and The distortion of mass spectra due to different deflection plate voltages is dramatically evident in Figure 2 where low mass clusters are completely swept out by increasing the deflection plate voltage from 500 to 800 V. A composite Nix mass spectrum obtained for a fixed A r F laser intensity is shown in Figure 3. Each Nix+ ion signal amplitude plotted in this figure is the value measured at the deflection plate voltage which maximizes that particular Nix+ ion signal. Thus the ion signals are free from instrumental distortion and their magnitudes are determined solely by physical factors. These include the concentration of neutral clusters in the beam, the order of the photoionization process (Le., first order, single photon, or higher order multiphoton), the photoionization (9) E.A. Rohlfing, D. M. Cox, and A. Kaldor, Chem. Phys. Leu., 99, 161 (1983).
cross section at 6.42 eV, and fragmentation processes which can be induced by the ionizing laser. The details of the effects of ionization order and fragmentation will be discussed in detail later. For the present discussion, we note that only Ni3 and Nix clusters for x 2 10 are both single photon ionized and free from fragmentation effects at the laser intensity at which the distribution shown in Figure 3 was taken. Therefore the relative magnitudes shown there are determined only by the photoionization cross section4 and the concentrations of the neutral Nix clusters. Assuming roughly the same cross section for all clusters, we conclude that the observed decrease in ion signals for Ni,+ and Nixt for x 2 10 means that the concentration of nickel clusters produced by the laser vaporization/expansion source decreases monotonically with increasing cluster size. This trend is entirely consistent with a single atom addition mechanism for the buildup of larger clusters from the atomic vapor. We are able to bracket the I P for Nix clusters as a function of cluster size by examining the dependence of the Nix+ ion signals on ionizing laser intensity for different ionizing laser frequencies. The order of the ionization process (linear, quadratic, etc.) is then used to determine a minimum/maximym value of the IP for Nix ( x = 2-23). For example, if the photon energy of the ionizing laser is 5.58 eV (Quanta Ray-PDL-WEX laser system), all Nix+ ( x = 1-23) ion signals are found tQexhibit a quadratic dependence on laser intensity. We interpret these results to mean that all Nix (x = 1-23) clusters have IPS greater than 5.58 eV but less than 11.16 eV. If the I P is above 11.16 eV, at least a three photon process would be required to produce ions and if the IP is less than 5.58 eV we would produce ions via a one-photon ionization process. For a photon energy of 6.42 eV from the ArF ionizing laser the intensity dependence of the ion signals is more complicated. Starting with the nickel atom and working up in cluster size, we observe that both Ni and rJi2 show a quadratic dependence on the ArF laser intensity, indicating that their IPSlies above 6.42 eV but below 12.84 eV. However, from the 5.58-eV data we known that the upper limit is 1 1.16 eV and thus we can bracket
4500 The Journal of Physical Chemistry, Vol. 88, No. 20, 1984 the Ni and Niz IPS between 6.42 and 11.16 eV. Representative examples of power law plots for cluster ion signals in the range of three to twelve atoms are shown in Figure 4. These log-log plots show the Ni3+, Ni4+, Ni5+, and Nilz+ as functions of ArF laser energy. Since the temporal and spatial characteristics of the ionizing laser are constant, the laser energy differs from the laser intensity by only a scale factor. In Figure 4a, Ni3+exhibits a first-order dependence on laser energy (unity slope), except at the highest energies where an increase in order is seen. For Ni4+, Figure 4b, a first-order dependence is also observed at low laser intensities, however, the order increases as the ionizing laser intensity increases, becoming very nearly second order. Very similar behavior is seen for Ni5+, Ni7+, Ni8+, and Ni9+, except that the first-order depencence extends to successively higher laser energy as the cluster size increases. The exception to this is Ni6+, Figure 4c, which shows an ionization order of nearly two a t all but the highest energies where a slight decrease in slope occurs. The power law plot for Nilz+, Figure 4d, is representative of all Nix+ signals for x = 10-23. It shows a first-order slope except at high energies where a decrease in order is seen. Taking into account the 5.58-eV data, we can again use our "bracketing" arguments to conclude that the IPS of Ni3 and Nilo to Niz3 are all less than 6.42 eV but above 5.58 eV. Also the I P of Ni6 must be greater than 6.42 eV and less than 11.16 eV. For Ni4 thru Ni9, the interpretation is not so straightforward due to the intensity-dependent ionization orders observed in the ArF data. We argue that, for these clusters, the first-order processes observed at low ionizing laser intensities is the result of single-photon ionization of the indigenous Nix clusters in the beam and that the IPSof Ni4, Ni5, and Ni7-Ni9 all lie below 6.42 eV but above 5.58 eV. At higher laser intensities, it is obvious that other, higher order channels are contributing to the observed Nix+ signals. It is implicit in our interpretation that the metal clusters are assumed to be predominately in their ground electronic state prior to ionization. We fell this is a reasonable assumption, since experimental spectroscopic data taken for a variety of metal cluster systems have shown that metal dimers (CuZ,Crz, Moz) produced by the same laser vaporization/expansion technique are indeed in their ground electronic state and, in addition, are vibrationally and rotationally cooled by the supersonic expansiom6 As an aid to our possible explanations for the features observed in the case of ArF laser ionization of Nix clusters, it is useful to examine mass spectra which have been normalized to the ionizing laser energy. These provide a more general and a linear way of looking a t the trends seen in the power law plots (Figure 4). Normalized spectra are shown in Figure 5 for three different laser intensities which span the range used in our experiments. Figure 5a is a spectrum taken in the low intensity regime in which the Ni4+-Ni9+ (except Ni6+) show first-order behavior. In Figure 5b, the ionizing laser intensity lies in the range where the Ni4+-Ni9+ signals start to show higher order effects and this is evidenced as an increase in the normalized ion signals for these clusters. Note that for the single-photon ionized Ni3, the energy normalized Ni3+ signal is remaining constant. Finally, in Figure 5c, the laser intensity is high and the higher order effects for Ni4+-Ni8+ are pronounced, effectively shifting the peak of cluster ion distribution toward lower mass clusters. Also note that, at this high intensity, the normalized Ni3+ signal shows an increase while there is a decrease in the normalized cluster ion signal for Nix+, x 1 10. We propose two possible explanations for this mixed one- and two-photon behavior seen in the Ni4+-Nilo+signals. First, Nix+ may arise from parent cluster ionization Nix
hv
Nix+
+ e-
(1)
or also from photofragmentation of larger clusters. Niy
hv
Nix+
+ NiYx + e-
(2)
Such photofragmentation processes may be laser intensity dependent and the minimum energy for process (2) will be higher by at least the cluster binding energy than that needed for process (1). In addition a large photofragmentation channel would tend
Rohlfing et al.
1000 I
5
O
0
1
i
i
I
1000
500
0
0
5
io
15
20
Nix Cluster Size (Atoms)
Figure 5. Nix time-of-flightmass spectra for three different ArF laser intensities. Each spectra has been normalized to its respective laser
energy which are, with corresponding intensities in parentheses as follows: (a) 0.029 mJ (19 kW/cm2, (b) 0.235 mJ (160 kW/cm2), and (c) 1.01 mJ (670 kW/cm2). to mask the parent cluster ion production. Such an effect could account for the increased intensity dependence observed on the smaller clusters (Ni4+, Ni5+, Ni7-10+)at the intermediate laser intensities. For Ni6+ (Figure 4), we observe only a quadratic intensity dependence. If the Ni6+ IP is indeed greater than 6.42 eV, fragment ions if formed should exhibit at least an I3 intensity dependence. Alternatively, the Ni6+ intensity dependence could be explained as entirely a 1 plus 1 photon detection of fragment Ni6 species, if one assumes the IP of Ni6 is less than 6.42 eV and that Ni6 is not present initially in the molecular beam. Although this possibility cannot be completely ruled out, the presence of every other cluster size giving rise to parent ion signals (at least at low fluence) makes the absence of one size cluster seem highly improbable. In addition, photofragmentation would imply that parent clusters which fragment should show a reduced dependence on laser intensity for intensities where small clusters show enhanced intensity dependence. The observation of reduced signal at the higher laser intensities on Nilz+ and larger clusters would be consistent with this interpretation. Alternatively, this high-intensity behavior could be rationalized as a saturation effect in the sense that for sufficiently high photon flux nearly all Nilz+and larger clusters within the irradiation zone are ionized. A second mechanism to explain the mixed one- and two-photon behavior of the ion signal is based upon consideration of their electronic structure. Consider Ni, as an example; recent ab initio calculation^^^ indicate overlap between the s,p- and d-like ionization manifolds of Ni4. However, the highest occupied molecular
The Journal of Physical Chemistry, Vol. 88, No. 20, 1984 4501
Photoionization of Isolated Nix TABLE I: Electron Work Function for Nickel
a
$3 eV
ref
method
5.15 5.22 (100) 5.04 (110) 5.35 (111) 5.40 (111) 5.22-5.21
17 18 18 18 19
photoelectric photoelectric photoelectric photoelectric photoelectric
20
RPD
i
Experimental value.
orbital (HOMO) of s,p character is calculated to be -0.7 eV lower in ionization energy than the d-like HOMO. Thus for an ionizing photon energy which lies between these two ionization energies, single-photon ionization from the s,p "band" can account for the observed first-order behavior at low intensities. At higher ionizing laser intensities, two-photon ionization can access the d manifold and this process may dominate s "band" ionization because of the much higher density of states in the d manifold. It should be clear by now that our simple power dependence data cannot provide enough information to allow us to choose between either of the two mechanisms detailed above. For the fragmentation question, pump and probe experiments in which the spatial qualities of the two lasers are used to probe only photofragments ejected from the molecular beam may give insight into the fragmentation patterns and bond energies of Nix clusters. The band structure of Nix clusters can be probed by experiments in which the ionizing photon energy can be scanned in the threshold regions and, more directly, by doing photoelectron spectroscopy on these clusters. The summary of our ionization potential measurements for Nix is presented in Figure 6. The vertical bars on the figures delimit the value of the I P for the individual clusters. From early mass spectrometer work a value of 6.6 eV has been reported for the appearance potential of nickel dimer" which is consistent with our experimental measurements. Table I summarizes the values of the work function obtained from the literature. The crosshatched region on Figure 6 delimits the range of values reported for the nickel work function. Even for the coarse measurements of the I P S shown in Figure 6, we observe a gross oscillation of the I P as the cluster size increases, not simply a monotonic approach toward the work function. This behavior is found to be even more dramatic for Fe, cluster^.^ Theoretically calculated values for Nix IPSfor x = 1-612-13 and for NiI3,N19,and Nzo14also are shown in Figure 6. Quantitatively the calculated values fall 1.5-2.0 eV below the measured values. Qualitatively the general trends in IPS predicted by theory are reasonably consistent with the trends observed experimentally. For Ni3 and Ni,, the IPS for different geometrical configurations12J6have been calculated. In Ni3 the linear structure has been found theoretically to be the most stable structure. The calculated IP is found to decrease monotonically (in magnitude) as the angle varies from 180' (linear) to 60' (D3h).On the other hand, a nickel containing species isolated in solid argon has been reported2' to be bent Ni, with an apex angle of between 90° and 100'. It may well be that the cluster structure depends critically (10) J. 0. Noell, M. D. Newton, P. J. Hay, R. L. Martin, and F. W. Bobrowicz, J . Chem. Phys., 73, 2360 (1980), and references therein. (11) A. Kant, J. Chem. Phys., 41, 1872 (1964). (12) H. Basch, M. D. Newton, and J. W. Moskowitz, J . Chem. Phys., 73, 4492 (1980). (13) M. D. Newton, Chem. Phys. Lett., 90, 291 (1982). (14) (a) C. F. Melius, T. H. Upton, and W. A. Goddard 111, Solid State Commun., 28, 501 (1978). (b) T. H. Upton, W. A. Goddard 111, and C. F. Melius, J . Vac. Sci. Technol., 16, 531 (1979). (1 5) "CRC Handbook of Chemistry and Physics", R. C. Weast Ed., 62nd ed, CRC Press, Boca Raton, FL, 1981-2. (16) M. D. Newton, private communication. (17) D. E. Eastman, Phys. Reu. B, 2, l(1970). (18) B. G . Baker, E. B. Johnson, and G. I. C. Maire, Surf. Sci., 24, 572 (1971). (19) S. R. Kelemen and T. E. Fischer, Surf. Sci., 102, 45 (1981). (20) A. A. Holscher, Surf. Sci., 4, 89 (1966). (21) M. Moskovitz and D. P. DiLelle, J. Chem. Phys., 72, 2267 (1980).
Mi, Cluster Size (Atoms)
Figure 6. The maximum and minimum values of the ionization potentials for nickel clusters are plotted as a function of cluster size. The cross hatched area presents the range of values of the work function for nickel (see Table I). The solid circle represents the value of the IP of atomic N i (7.63 eV).lS The squares and circles are theoretical predictions from ref 12, 13, and 14, respectively. The two theoretical values for Ni3 and Ni4 represent IP calculations for two different molecular structures. For Ni3 the IPSfor the linear (solid square) and equilateral triangular (open square) structures were calculated whereas the IPSfor the square planar (solid square) and tetrahedral (open square) structures were calculated for Ni,. The argon fluoride excimer laser photon energy is 6.4 eV whereas the highest energy photon we could generate from a Quanta Ray PDL-WEX laser was 5.6 eV.
on its local environment, e.g., isolated in a molecular beam or trapped in a matrix. For Ni,, the square-planar structure is calculated to be more stable than a tetrahedral structure and gives rise to a higher value of the IP. It is enticing to speculate that with improved theoretical calculations and experimental measurements it may become possible to infer the structures of these small clusters from consideration of fundamental electronic properties such as IPS. We note that the higher mass clusters (e.g., Ni23) have approached to within -0.4 to 1.2 V of the work function of bulk polycrystalline nickel (4 = 5.2 eV). It has been proposedz2in the literature that, if one considers these clusters as small metal spheres, an additional energy of e 2 / 2 R is required to remove an electron from a curved surface of radius R , compared to that needed to remove an electron from an infinite flat surface. For the larger Nix clusters (x = 20-23) this additional energy is calculated to be -2 eV.23 The IP of Ni23 would be expected to be -7.2 eV by using this classical picture, a value 0.8 to 1.6 eV higher than observed experimentally. These results for Nix clusters are in contrast to those found for Na, clusters where the classical droplet model was found to give values much closer to the experimental values for clusters of only 8 to 12 atoms in size. Thus for the larger nickel clusters we find the IP is closer to the bulk work function than to the IP estimated for a small metal droplet. This raises the question as to whether one should consider these larger clusters as small bits of macroscopic nickel or as unique chemical entities whose particular structural and electronic properties still depend critically upon the number of atoms in the cluster. Finally, in addition to bracketing the IPS for bare Nix clusters, we have bracketed the IPS for Ni,O and Ni,02 clusters for x = 2 to 13. These results will be discussed elsewhere, but we note that the IP for certain oxides (Ni30, Ni40, Ni70, and the corresponding dioxides) are higher than those for the corresponding Nix cluster. Summary and Conclusions The first measurements of the ionization potentials for isolated nickel atom clusters ranging in size from 2 to 23 atoms are reported. A variation in the ionization potential with cluster size (22) A. Hermann, E. Schumacher, and L. Woste, J . Chem. Phys., 68,2327 (1978). (23) R is taken as (23)1/3D0/2where D is the nickel lattice spacing of 2.492
A.
4502
J . Phys. Chem. 1984, 88, 4502-4505
is observed, not simply a monotonic decrease of the IP toward the value of the work function as the cluster size increases. A comparison of these results with theory indicates that while the quantitative theoretical predictions are poor, qualitatively, the trends predicted theoretically are in reasonable agreement with the measured experimental trends. Even though the I P S of Nix clusters have been measured only within a fairly coarse grid of photon energies, we have been able to gain insights into possible alterations of electronic and structural characteristics of metal atom clusters and to demonstrate the limitations of the present theoretical calculations and experimental measurements. Further experimental work using tunable lasers will allow us to more
precisely define the metal cluster IPS.Such information represents a first step toward the characterization of the electronic and structural features of small metal clusters.
Acknowledgment. We are particularly indebted to R. Smalley and D. J. Trevor for valuable discussions and R. Smalley and J. Hopkins for valuable technical assistance with the details of the pulsed solenoid valve as well as for providing some data acquisition software. We gratefully acknowledge the valuable technical assistance provided by Ken Reichmann during the course of these experiments. Registry No. Nickel, 7440-02-0.
Mass Spectral and Electric Deflection Study of Acetic Acid Clusters$ R. Sievert? I. Cadei,* J. Van Doren,I and A. W. Castleman, Jr.* Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802 (Received: January 31, 1983)
Acetic acid clusters, (CH3COOH),, up to n = 10, were produced in a supersonic beam expansion and analyzed in a molecular beam quadrupole mass spectrometer. A general mechanism for their mass spectral fragmentation was deduced. Polarity of the first four clusters was determined in a molecular beam electric deflection experiment, providing evidence for their vapor-phase structure.
Introduction Owing to the unusually large amount of dimer in the vapor phase,' studies of the vapor-phase polymers of acetic acid are of interest for testing theories of homogeneous nucleation: elucidating the nature of hydrogen bonding? and investigating the mechanism of cluster g r ~ w t h . Althoqgh ~ there has been extensive investigation of the acetic acid vapor p h a ~ eand ~ , ~the properties of the dimer are very well-known,' there is still some controversy about the amount of higher-order clusters in the gas phase's8 and their structures and proper tie^.^ The present study was undertaken to provide additional data on the polymers of acetic acid vapor, with attention to the problem of fragmentation mechanisms which inherently can influence the interpretation of the mass spectra of clusters. Evidence for the polarity of clusters containing the first four monomer subunits was obtained from the mass spectrometric experiments coupled with a quadrupole electric deflection technique. The first known evidence for the focusing of high-order polymers was obtained and are reported herein. These findings have provided the basis for making certain deductions concerning the structure of these clusters. Experimental Section The apparatus consists of four major parts: the stagnation chamber, the differential pumping chamber with nozzle exhaust, the focusing region, and the detection chamber (Figure 1). Clusters are produced by expanding the vapor of the stagnation chamber at about 1 atm through a sonic nozzle to form a free jet. The stagnation chamber consists of a heatable glass tube (0.d. = 16 mm) that is drawn down to form a nozzle with a throat diameter of 100 or 150 pm." The liquid acid (Baker, 99.9% pure) Preliminary phases of this research were undertaken at the University of Colorado. NATO Fellow, 1981-2. Present address: Institut fur Physikal Chemie, Tammann Str. 6, D-3400 G6ttingen, West Germany. 3 Visiting Fellow, University of Colorado, 1981-2 and Visiting Associate Professor, Pennsylvania State University, 1982. Present address: Institut za fiziku, P.O. Box 57, Studentski trg 12/Y, 11000 Beograd, Yugoslavia. Present address: Department of Chemistry, University of Colorado, Boulder. CO 80309.
is heated in a glass flask connected to the tube. Its temperature is held constant as monitored by a thermocouple, and the partial pressure of the acid is determined by interpolating from known vapor pressure curves. A regulated flow of carrier gas is bubbled through the liquid. The total pressure of the stagnation chamber is measured with a Bourdon type gauge. The whole glass tube is heated resistively; its temperature, as well as that of the nozzle, is measured by a thermocouple mounted at each location. The nozzle temperature has to be set higher than that of the bulb temperature in order to prevent condensation. The nozzle exhaust is a stainless steel chamber (id. = 30 mm) pumped by a freon baffled oil diffusion pump. It is separated from the differential pumping chamber by a skimmer having a I-mm aperture. The pressure during operation with the gas load is approximately 5 X 104-10-3 torr in the nozzle exhaust and IO" in the differential pumping region. The movable nozzle is aligned with the skimmer and the axis of the apparatus by optimizing the mass spectrometric signal of the cluster beam. The molecular beam enters the focusing region through a hole (i.d. = 5 mm). The region is a stainless steel cyclinder pumped by a liquid nitrogen baffled oil diffusion pump. The typical pressure under a gas load is lo-' torr. This chamber houses the electric deflection field. It is comprised of a set of stainless steel quadrupole rods (length 57.15 cm, diameter 0.476 cm), which are mounted in a support of machinable glass and stainless steel that (1) Chao, J., Zwolinski, B. J . Phys. Chem. Ref. Data 1978, F l , 363. (2) Heist, H. R.; Colling, K. M.; DuPuis, C. S. J . Chem. Phys. 1976, 65, 5147. (3) Joesten, M. D.; Schaad, L. J. "Hydrogen Bonding", Marcel Dekker: .~ New York, 1974. (4) Castleman, A. W., Jr.; Kay, B. D.; Sievert, R.; Stephan, K.; Van Doren, J.: Miirk, T. D. Presented at the Seventh International Symposium on Gas . . Kinetics, Gottingen, 1982. ( 5 ) Karle, J.; Brockway, L. 0. J. Am. Chem. SOC.1944,66, 574. (6) Clague, A. D. H.; Bernstein, J. H. Spectrochim. Acta, Part A 1969, 25, 593. (7) Frurip, D. J.; Curtiss, L. A.; Blander, M. J. Am. Chem. SOC.1980, 102, 2610. (8) Ritter, H. L.; Simons, J. H. J. Am. Chem. SOC.1945, 67, 757. (9) Johnson, E. W.; Nahs, L. K. J . Am. Chem. SOC.1950, 72, 547. (10) Kay, B. D. Ph.D. Thesis, University of Colorado, Boulder, 1981. (11) Kay, B. D.; Lindeman, T. G.; Castleman, A. W., Jr. Reu. Sci. Instrum. 1982, 53, 473.
0022-365418412088-4502$01.50/0 0 1984 American Chemical Society
