Energy of the Quasi-free Electron in H2 , D2 and O2 : Probing Intermolecular Potentials within the Local Wigner-Seitz Model C. M. Evans,1, a) Kamil Krynski,1 Zachary Streeter,2 and G. L. Findley2, b) 1) Department of Chemistry and Biochemistry, Queens College – CUNY, Flushing, NY 11367 2) School of Sciences, University of Louisiana at Monroe, Monroe, LA 71209 (Dated: 2 November 2015) We present for the first time the quasi-free electron energy V0 (ρ) for H2 , D2 and O2 from gas to liquid densities, on noncritical isotherms and on a near critical isotherm in each fluid. These data illustrate the ability of field enhanced photoemission (FEP) to determine V0 (ρ) accurately in strongly absorbing fluids (e.g., O2 ) and fluids with extremely low critical temperatures (e.g., H2 and D2 ). We also show that the isotropic local Wigner-Seitz model for V0 (ρ) – when coupled with thermodynamic data for the fluid – can yield optimized parameters for intermolecular potentials, as well as zero kinetic energy electron scattering lengths. a) cherice.evans@qc.cuny.edu b) findley@ulm.edu 1 I. INTRODUCTION The quasi-free electron energy V0 (ρ) (where ρ is the fluid number density) is important in understanding electron transport through a fluid,1–4 as well as for modeling electron attachment reactions in fluids.2,5 The recent development of field enhanced photoemission6–8 as a technique to directly and accurately measure V0 (ρ) has opened the possibility of investigating the density and temperature dependence of the quasi-free electron energy in a variety of problematic fluids (e.g., fluids with extremely low critical temperatures,7 strongly absorbing fluids8 (i.e., fluids that absorb light in the ultraviolet or vacuum ultraviolet regions) and polar fluids9 ). We have shown in previous studies7,10 that the isotropic local Wigner-Seitz model for V0 (ρ) can give insight into the determination of the intermolecular potential for a fluid. In this paper, we present the quasi-free electron energy V0 (ρ) in H2 , D2 and O2 . We use these data, coupled with thermodynamic data for the fluids, to optimize intermolecular potential parameters. We also show that nonlinear analyses of V0 (ρ) within the isotropic local Wigner-Seitz model using these intermolecular potentials yield accurate zero kinetic energy electron scattering lengths for each fluid. II. ISOTROPIC LOCAL WIGNER-SEITZ MODEL Within the local Wigner-Seitz model, V0 (ρ) is given by10,11 3 V0 (ρ) = P− (ρ) + Ek (ρ) + kB T , 2 (1) where P− (ρ) is the ensemble average electron/perturber polarization energy, Ek (ρ) is the zero point kinetic energy of the quasi-free electron, and (3/2)kB T (kB ≡ Boltzmann’s constant) is the thermal energy of the quasi-free electron. The density dependent average polarization energy P− (ρ) depends upon the position ri of each of the N fluid constituents relative to the quasi-free electron at the moment of electron/constituent interaction. If the interaction potential is isotropic and is written as a sum of pairwise potentials, then the total electron/fluid interaction potential is4,12–19 N ∑ 1 w− (r1 , . . . , rN ) = − α e2 ri−4 f (ri ) , 2 i 2 (2) where α is the polarizability of the fluid, e is the charge on the electron, and f (r) is a screening function that incorporates the repulsive interactions between the induced dipoles in the fluid. The screening function f (r) is4 ∫ ∞ f (r) = 1 − απρ 1 ds 2 g(s) s 0 ∫ r+s dt f (t) θ(r, s, t) , (3) |r−s| with θ(r, s, t) = 3 2 (s + t2 − r2 )(s2 − t2 + r2 ) + (r2 + t2 − s2 ) , 2 2s (4) and with g(s) being the fluid radial distribution function. (In eqs. (3) – (4), the integration variables s and t represent the distance between the constituent of interest and all other constituents of the fluid, respectively.) Under the assumption of a canonical distribution, the probability of sampling a particular polarization energy W is given by12–21 ∫ P (W ) = ∫ ··· δ(W − w− (r1 , . . . , rN )) × e−U (r1 ,...,rN )/(kB T ) /∫ i ∫ ··· ∏ e −U (r1 ,...,rN )/(kB T ) ∏ dri (5) dri , i where U (r1 , . . . , rN ) is the multi-dimensional potential energy prior to the electron/fluid interaction. A moment analysis of the Fourier transform of eq. (5) yields the average polarization energy P− (ρ) through the first moment, or12–21 ∫ m1 = −4πρ ∞ g(r)w− (r)r2 dr ≡ P− (ρ) . (6) 0 The zero-point kinetic energy Ek (ρ) is obtained from solving the Schrödinger equation10,11,22 ∇2 ψ + 2me [Ek − Vloc (r)]ψ = 0 ~2 (7) for the quasi-free electron in a dense fluid. In eq. (7), Vloc (r) is a short-ranged potential that accounts for local dynamic polarization of a fluid constituent by the quasi-free electron, me is the mass of the electron and ~ is the reduced Planck constant. For isotropic fluids, the local potential Vloc (r) satisfies an average translational symmetry10,11,22 Vloc (r) = Vloc (r + 2rb ) , 3 (8) where rb is the interaction range. At any density, the minimum distance between a quasifree electron with low kinetic energy and a single fluid constituent is given by the absolute value of the scattering length A. Under the assumption that interactions in the first solvent shell dominate the dynamics of the problem, the maximum distance rℓ for a short-ranged interaction is one-half of the spacing between two fluid constituents in this shell. This distance, otherwise known as the local Wigner-Seitz radius, is10,11 √ rℓ ≡ 3 3 , 4πgmax ρ (9) where gmax is the maximum of the fluid radial distribution function g(r). Thus, the interaction range for the local short-ranged potential is rb = rℓ − |A|. By applying these boundary conditions to the asymptotic solutions of eq. (7), the zero-point kinetic energy of the quasi-free electron is10,11,22 Ek (ρ) = ~2 η02 . 2me (rℓ − |A|)2 (10) In eq. (10), η0 is a phase shift induced by the short-ranged potential, and the density dependence arises from rℓ . The phase shift η0 represents the only empirical parameter in the local Wigner-Seitz model, if the zero kinetic energy electron scattering length A has been determined independently from other experimental techniques (see, for example, cross section measurements23–25 or measurements of the shift of high-n dopant Rydberg state energies26,27 for the fluids in this study) and if the polarization potential P− (ρ) is known. (The theoretical treatment of the short range potential leading to η0 is a question of active interest that will be the subject of a future publication by us.) III. COMPUTATIONAL METHODOLOGY Since the calculation of P− (ρ) involves g(r), while rℓ is ascertained from the maximum of g(r), accurately determining the radial distribution function for each fluid over a large density and temperature range is extremely important. We have shown that a direct integration of the Ornstein-Zernike relation with a Percus-Yevick closure reduced numerical instabilities in the determination of g(r) at high densities and on near critical isotherms.10,26 4 This direct integration is accomplished through the homogenous Percus-Yevick model28 g(r) = r−1 e−U (r)/kB T Y (r) , where ∫ r Y (r) = dt 0 (11) dY (t) dt (12) with ∫ ∞ [ ] dY (r) = 1 + 2πρ dt e−U (r)/kB T − 1 Y (t) dr 0 [ ] r − t −U (|r−t|)/kB T −U (r+t)/kB T × e Y (r + t) − e Y (|r − t|) − 2t . |r − t| (13) Stable numerical integration of eqs. (11) – (13) was achieved using an optimization routine developed by Ng et al.,29 which employs a weighted linear combination of m ≥ 3 previous iterations to generate the input for the (m + 1)th iteration. Since one of the goals of this study was to test the use of the isotropic local WignerSeitz model for V0 (ρ) as a means of selecting an adequate intermolecular potential, we chose to calculate radial distribution functions via eqs. (11) – (13) using the two most common intermolecular potentials, namely the Lennard-Jones 6-12 potential (LJ6-12) and the exponential-6 (Exp-6) potential. The functional forms of these potentials are given in Table I. Since recent studies30–35 have shown that different cut-off radii for the OrnsteinZernike approach to determining radial distribution functions can yield different potential parameters, we chose to validate the radial distribution functions obtained from the chosen intermolecular potentials by comparison with thermodynamic data via36 2 p(ρ, T ) = kB T ρ − πρ2 3 ∫ rc dr r3 g(r) 0 dU (r) , dr (14) where rc is a cut-off radius. In this study, we found that rc ≈ 8σ ensured numerical stability in the near critical point calculations. Moreover, increasing rc beyond 8σ did not lead to appreciable change in the intermolecular potential parameters optimized to thermodynamic data at noncritical temperatures (both above and below the critical temperature). 5 IV. EXPERIMENTAL PROCEDURES The direct determination of the quasi-free electron energy V0 (ρ) using field enhanced photoemission (FEP)6–8 requires the acquisition of photoemission spectra under the influence of a minimum of four different electric fields at each fluid density and for each temperature. The difference ∆i in the near threshold photoemission from a metal electrode (which was Pt for the experiments reported here) in the presence of a fluid and under the influence of two different electric fields is6–8 (√ √ ) ∆i = [hν − ϕ(ρ)] ΛH − ΛL B(T, ρ) 1 + (ΛH − ΛL ) 2 (15) where e2 Ej εr (T, ρ) Λj = (16) with j = L, H for the low and high applied electric field Ej , respectively. In eqs. (15) and (16), B(T, ρ) is an empirical constant that depends on the fraction of electrons that can absorb a photon and escape, hν is the photon energy, ϕ(ρ) is the metal work function in the presence of the fluid, and εr ≡ ε(T, ρ)/ε0 is the temperature and density dependent fluid relative permittivity, with ε0 being the permittivity of the vacuum. Additional rearrangement of eq. (15) gives the primary FEP equation:6–8 hν = b∆iE + a(ρ) , (17) with a slope of b = 1/B(T, ρ) and an intercept of a(ρ) = ϕ(ρ) − √ ) 1 (√ ΛH + ΛL , 2 (18) where ∆iE is defined as ∆iE ≡ √ ∆i √ . ΛH − ΛL (19) Thus, the FEP signal is a plot of photon energy as a function of ∆iE via eq. (17). Fitting the linear region of this signal by least squares analysis yields the work function ϕ(ρ) from the intercept a(ρ) in the limit of zero electric field. The density dependent quasi-free electron 6 energy V0 (ρ) is then directly determined from V0 (ρ) = ϕ(ρ) − ϕ0 , (20) where ϕ0 is the reference work function for the specific metal electrode. Because ϕ(ρ) is temperature dependent, and in order to minimize errors introduced by fluid/electrode interactions, a series of photoemission spectra are obtained at very low fluid density and at various voltages for each isotherm to ascertain the reference work function ϕ0 . All photoemission data for these studies were measured with monochromatic vacuum ultraviolet synchrotron radiation using the University of Wisconsin Synchrotron Radiation Center stainless steel Seya-Namioka beamline equipped with a high energy (5–35 eV) grating.26,37 The radiation, which had a resolution of ∼ 10 meV in the energy region of interest, entered a copper sample cell fitted with a 3 mm thick MgF2 window capable of withstanding up to 100 bar of pressure. The photoemitter was a 10 nm thick strip of Pt sputter deposited along the window diameter. A stainless steel electrode was attached to the window perpendicular to this strip of Pt, with the spacing between the two electrodes being 0.1 cm. The electric field was applied to the stainless steel electrode, while the photocurrent was detected at the Pt electrode using a Keithley 6514 electrometer. (To increase the signal-to-noise ratio for these low signal measurements, the high voltage power supply and the electrometer drew power through an AC line conditioner to remove electrical line noise.) The flux from the Seya beamline was monitored for consistency using a nickel mesh intersecting the beam prior to the sample cell. However, the photoemission spectra were normalized to monochromator flux using a GaAsP diode adjusted for the beam current in the Aladdin storage ring, since the lowest photoemission threshold of the nickel mesh occurred in the spectral region of interest. The cell temperature was adjusted, and maintained to within ±0.05 K, using an Advanced Research Systems DE-204SB 4K closed cycle helium cryostat system. Oxygen (Matheson Trigas, 99.998%), hydrogen (Matheson Trigas, 99.9999%) and deuterium (AirGas, 99.999%, 99.8% isotopic enrichment) were used without further purification. Prior to the introduction of any of these gases to the sample cell, the gas handling system and sample cell were baked to a base pressure of at least low 10−8 Torr. The number densities of O2 ,38,39 H2 40 and D2 41 were calculated using the NIST Standard Reference Database 7 23 Version 9.1.42 Significant loss of photoemission current was encountered in the measurement of photoemission spectra of Pt in all of these fluids at very high densities (i.e., ρ > 1.8 × 1022 cm−3 or ρr ≡ ρ/ρc > 1.7) and, therefore, only a few data points were obtained in this regime. Since the minimum field strength used in these studies was 1.0 kV/cm (corresponding to a kinetic energy of 100 eV for a free electron in a vacuum) while the maximum field strength was in excess of 40 kV/cm, electron localization and electron attachment reactions to fluid impurities should be minimized, even when one accounts for the significant reduction in the kinetic energy of the electron due to electron/fluid interactions. Therefore, we attribute the almost complete loss of photoemission signal at very high densities to fluid stabilized dissociative electron attachment.43,44 Each FEP spectrum depends on the relative permittivity of the fluid and, therefore, this latter parameter was obtained by a least squares analysis of the total charge on the electrodes, in the absence of photons, as a function of voltage across the electrodes at all fluid densities, temperatures and applied electric fields. The Clausius-Mossotti functions45,46 for the experimentally determined relative permittivities for all fluids in this study are shown in Fig. 1 plotted versus the reduced fluid number density ρr . These data indicate that the relative permittivity exhibits a slight critical point effect, which was not observed in our earlier measurements of He7 and N2 ,8 and which – to the best of our knowledge – has not been reported previously. Modeling of these results, along with relative permittivity measurements from simple polar fluids, is currently in progress.47 V. RESULTS AND DISCUSSION Because of the large number of photoemission spectra required for this study, we have chosen for brevity to show in Fig. 2 representative spectra of Pt in H2 , D2 and O2 obtained for various electric fields near the critical point of each fluid. A distinct enhancement of the photocurrent as a function of electric field is observed in all systems. Similar series of photoemission spectra were obtained at various fluid number densities on noncritical isotherms and on an isotherm near the critical temperature. The near critical isotherm was selected to be +0.5 K above the critical temperature so as to maximize temperature and pressure stability near the critical point and to minimize the possibility of forming two phases 8 (i.e., gas and liquid) in the sample cell. Electric fields were chosen to maximize the field effect and to minimize the probability for electron localization or electron attachment reactions, while not creating dielectric breakdown in the cell. Because ϕ(ρ) is temperature dependent, a set of four low density field dependent photoemission spectra were measured at every temperature, thereby providing a ‘zero’ density signal that corrects for these temperature variations. Obtaining FEP signals from the photoemission data and then fitting these signals by linear least squares analyses of eq. (17) yields intercepts a(ρ) for each density at each √ √ temperature. These intercepts are shown in Fig. 3 plotted as a function of ΛH + ΛL for select densities in H2 , D2 and O2 at noncritical temperatures. Fig. 4 shows similar data for the near critical isotherms in H2 , D2 and O2 . Extrapolating a(ρ) to zero field yields the work function ϕ(ρ) (i.e., eq. (18)), which can then be used to obtain the quasi-free electron energy V0 (ρ) from eq. (20). (We should note here that the reference work function was set to ϕ0 ≈ ϕ(3.0 × 1019 cm−3 ) for each fluid at each temperature in this study.) Fig. 5 presents V0 (ρ) plotted as a function of the reduced fluid number density ρr ≡ ρ/ρc (with the fluid critical density42 ρc (in units of 1021 cm−3 ) being 9.3397, 10.435 and 8.209 for H2 , D2 and O2 , respectively). To apply the isotropic local Wigner-Seitz model to the data shown in Fig. 5, we began by determining optimized Lennard-Jones 6-12 and exponential-6 potential parameters for H2 , D2 and O2 using eq. (14) with the thermodynamic data of each fluid provided by NIST.42 The optimal parameters were defined as those that minimized the average percent deviation in the pressure as calculated from eq. (14) in comparison to the experimental thermodynamic conditions for the fluid when Tr (≡ T /Tc ) > 1, at a near critical temperature (i.e., Tr ≈ 1), and when Tr < 1, with the critical temperature42 Tc being 33.145 K, 38.34 K and 154.58 K for H2 , D2 and O2 , respectively. (We should note here that we were unable to obtain V0 (ρ) data at temperatures Tr < 1 for H2 and D2 . However, to ensure consistency over the broadest density range (i.e., from gas phase up to the density of the triple point liquid), we optimized potential parameters using additional thermodynamic data42 at Tr < 1 for all systems in this study.) Fig. 6 shows the optimization results for a Lennard-Jones 6-12 potential for O2 when Tr > 1, when Tr ≈ 1 and when Tr < 1. The solid point in the individual panels of Fig. 6 represents the area of overlap for each temperature region. Using this technique allows one to obtain the parameters that best fit the entire fluid phase of O2 , 9 as well as the error in these parameters. (Contour plots similar to that shown in Fig. 6 were generated for each potential (i.e., LJ6-12 and Exp-6) for each system (i.e., H2 , D2 and O2 ), but are not shown for brevity.) Table II gives the intermolecular potential parameters for the exponential-6 and Lennard-Jones 6-12 potentials for H2 , D2 and O2 as determined from these contour plots in comparison to parameters obtained from other computational methods. The radial distribution functions from the optimized potential parameters were used subsequently to calculate the average electron/fluid polarization energy P− (ρ), from eq. (6), and the local Wigner-Seitz radius rℓ for use in the determination of the kinetic energy Ek (ρ) of the quasi-free electron (i.e., eq. (10)). (The polarizability for each fluid was obtained from S. V. Khristenko et al..48 ) Once an optimized intermolecular potential was found, the only adjustable parameters in the isotropic local Wigner-Seitz model are in the calculation of the kinetic energy Ek (ρ) of the quasi-free electron, namely the zero kinetic energy electron scattering length A and the phase shift η0 . Thus, optimization of the isotropic local WignerSeitz model involves the adjustment of one or both of these parameters. However, for the fluids in this study, the zero kinetic energy electron scattering length A had been previously determined.27 Nevertheless, we decided to ascertain if one could obtain the zero kinetic energy electron scattering length directly from the quasi-free electron energy data through the local Wigner-Seitz model. Thus, employing a technique similar to the one used in the optimization of the intermolecular potential parameters, we measured the average percent deviation between the calculated V0 (ρ) in comparison to experimental data as a function of both A and η0 . As an example, Fig. 7 shows the contour plots for the optimization of V0 (ρ) for D2 at noncritical temperatures and near the critical temperature using the exponential-6 potential for all calculations. The solid point in the individual panels of Fig. 7 represents the area of overlap between the noncritical and critical isotherm data for those contours that deviate from experiment by less than the experimental error. (Similar contour plots were generated for the other fluids but are not shown here.) The results of these optimizations are presented in Table III along with those obtained from the calculation of V0 (ρ) (i.e., η0 ) with fixed A. Employing the fully optimized results, one notices that the scattering length obtained from the best intermolecular potential (as selected using thermodynamic data and the quasi-free electron energy data) is within 2% of that determined from dopant field ionization studies.27 10 (Using the zero kinetic energy electron scattering lengths obtained here to determine the total scattering cross section, via the relationship σT = 4πA2 , gives σT = 3.88 Å2 for H2 and σT = 0.149 Å2 for O2 . The zero kinetic energy total scattering cross section obtained for hydrogen is comparable to those obtained from other experimental techniques, which range from 2.5 Å2 to 5.75 Å2 .27,49–51 The difficulties in directly measuring the zero kinetic energy total electron cross section of O2 have been discussed by various groups.52–56 However, the value presented here is in line with the estimate that σT ≤ 0.4 Å2 .52–55 ) Therefore, choosing the intermolecular potential that best fits the thermodynamic data allows one to obtain the zero kinetic energy electron scattering length from the experimentally measured quasi-free electron energy in the fluid. The calculated quasi-free electron energy V0 (ρ) using the isotropic local Wigner-Seitz model are shown as lines in Fig. 5. (We should note here that the lines were calculated to the triple point liquid density (i.e., ρr ≈ 2.5), although we were unable to obtain experimental data in this region because, as mentioned previously, the fluids in this study have significant dissociative electron attachment channels.43,44 ) VI. CONCLUSIONS We have measured (cf. Fig 5) the quasi-free electron energy in H2 , D2 and O2 from gas to liquid densities at various temperatures, including an isotherm near the critical temperature. Moreover, we have shown that these data can be interpreted within the local Wigner-Seitz model for V0 (ρ). To the best of our knowledge, these results represent the first investigation of electron energy in these fluids over such a wide density range. Thus, we have further illustrated8 that field enhanced photoemission can be used to obtain the quasi-free electron energy in molecular fluids. We have also shown that combining the thermodynamic data of the fluid with the isotropic local Wigner-Seitz model for the quasi-free electron energy allows one to select optimized intermolecular potentials and to determine the zero kinetic energy electron scattering length of the fluid. Since the investigation of the zero kinetic energy electron scattering length in many complex molecular fluids is difficult, the ability to extract A accurately from the quasi-free electron energy will allow for future study of more complex molecular fluids (such as polar fluids). We are currently in the process of extending the local Wigner-Seitz model to anisotropic fluids.9 11 ACKNOWLEDGMENTS We thank Dr. Yevgeniy Lushtak (SAES, Inc), Ms. Ollianna Burke (NAVY Officer Training Command) and Mr. Holden Smith (Louisiana State University) for their assistance in obtaining the experimental data presented here. All measurements were performed at the University of Wisconsin Synchrotron Radiation Center, a facility primarily funded by the University of Wisconsin – Madison with supplimental support from facility Users and the University of Wisconsin – Milwaukee. This work was supported by a grant from the National Science Foundation (NSF CHE-0956719). REFERENCES 1 R. A. Holroyd and W. F. Schmidt, Ann. Rev. Phys. Chem. 40, 439 (1989), and references therein. 2 R. A. Holroyd, “Electron kinetics in nonpolar liquids – energy and pressure effects,” in The Liquid State and Its Electrical Properties, Vol. NATO ASI Series Volume 193 (Springer, New York, 1988) pp. 221–233, and references therein. 3 D. Chandler and K. Leung, Ann. Rev. Phys. Chem. 45, 557 (1994), and references therein. 4 J. Lekner, Phys. Rev. 158, 130 (1967). 5 R. A. Holroyd, J. R. Miller, A. R. Cook, and M. Nishikawa, J. Phys. Chem. B 118, 2164 (2014). 6 Y. Lushtak, C. M. Evans, and G. L. Findley, Chem. Phys. Lett. 515, 190 (2011). 7 Y. Lushtak, S. B. Dannenberg, C. M. Evans, and G. L. Findley, Chem. Phys. Lett. 538, 46 (2012). 8 Y. Lushtak, C. M. Evans, and G. L. Findley, Chem. Phys. Lett. 546, 18 (2012). 9 C. M. Evans, K. Yalikun, K. K. Krynski, and G. L. Findley, J. Chem. Phys. (2016), Energy of the excess electron in CO and HD: The anisotropic local Wigner-Seitz model, in preparation. 10 X. Shi, L. Li, G. L. Findley, and C. M. Evans, Chem. Phys. Lett. 481, 183 (2009). 11 C. M. Evans and G. L. Findley, Phys. Rev. A 72, 022717 (2005). 12 A. K. Al-Omari, R. Reininger, and D. L. Huber, J. Chem. Phys. 109, 7663 (1998). 13 K. N. Altmann and R. Reininger, J. Chem. Phys. 107, 1759 (1997). 12 14 A. K. Al-Omari, K. N. Altmann, and R. Reininger, J. Chem. Phys. 105, 1305 (1996). 15 A. K. Al-Omari and R. Reininger, J. Electron Spectrosc. Relat. Phenom. 79, 463 (1996). 16 A. K. Al-Omari and R. Reininger, J. Chem. Phys. 103, 4484 (1995). 17 A. K. Al-Omari and R. Reininger, J. Chem. Phys. 103, 506 (1995). 18 A. K. Al-Omari, R. Reininger, and D. L. Huber, Phys. Rev. Lett. 74, 820 (1995). 19 J. Meyer and D. L. Huber, Phys. Rev. A 47, R3491 (1993). 20 I. Messing, B. Raz, and J. Jortner, J. Chem. Phys. 66, 4577 (1977). 21 I. Messing, B. Raz, and J. Jortner, J. Chem. Phys. 66, 2239 (1977). 22 B. E. Springett, J. Jortner, and M. H. Cohen, J. Chem. Phys. 48, 2720 (1968). 23 Y. Itikawa, A. Ichimura, K. Onda, K. Sakimoto, K. Takayanagi, Y. Hatano, M. Hayashi, H. Nishimura, and S. Tsurubuchi, J. Phys. Chem. Ref. Data 18, 23 (1989). 24 J. S. Yoon, M. Y. Song, J. M. Han, S. H. Hwang, W. S. Chang, B. J. Lee, and Y. Itikawa, J. Phys. Chem. Ref. Data 37, 913 (2008). 25 Y. Itikawa, J. Phys. Chem. Ref. Data 38, 1 (1990). 26 X. Shi, Energy of the quasi-free electron in atomic and molecular fluids, Ph.D. thesis, The Graduate Center – CUNY (2010), and references therein. 27 C. M. Evans, H. T. Smith, O. Burke, Y. Lushtak, and G. L. Findley, J. Phys. B: At. Mol. Opt. Phys. 47, 035204 (2014). 28 A. A. Broyles, J. Chem. Phys. 35, 493 (1961). 29 K. C. Ng, J. Chem. Phys. 61, 2680 (1974). 30 F. Mandal, J. Chem. Phys. 59, 3907 (1973). 31 S. Fishman and M. E. Fisher, Physica 108A, 1 (1981). 32 G. A. Martynov, J. Chem. Phys. 129, 244509 (2008). 33 J. M. Brader, Int. J. Thermophysics 27, 394 (2006). 34 P. T. Cummings and P. A. Monson, J. Chem. Phys. 82, 4303 (1985). 35 J. J. Brey and A. Santos, J. Chem. Phys. 79, 4652 (1983). 36 D. McQuarrie, Statistical Mechanics (University Science Books, South Orange, New Jersey, 2000). 37 Y. Lushtak, Energy of the quasi-free electron in repulsive atomic and molecular fluids, Ph.D. thesis, The Graduate Center – CUNY (2012), and references therein. 38 R. Schmidt and W. Wagner, Fluid Phase Equilibria 19, 175 (1985). 39 R. B. Stewart, R. T. Jacobsen, and W. Wagner, J. Phys. Chem. Ref. Data 20, 917 (1991). 13 40 J. W. Leachman, R. T. Jacobsen, S. G. Penoncello, and E. W. Lemmon, J. Phys. Chem. Ref. Data 38, 721 (2009). 41 I. A. Richardson, J. W. Leachman, and E. W. Lemmon, J. Phys. Chem. Ref. Data 43, 013103 (2014). 42 E. W. Lemmon, M. L. Huber, and M. O. McLinden, NIST Standard Reference Database 23: Reference Fluid Thermodynamic and Transport Properties – REFPROP, Version 9.0 (National Institute of Standards and Technology, Gaithersburg, MD, 2010). 43 E. Krishnakumar, S. Denifl, I. Čadež, S. Markelj, and N. J. Mason, Phys. Rev. Lett. 106, 243201 (2011), and references therein. 44 V. Laporta, R. Celiberto, and J. Tennyson, Phys. Rev. A 91, 012701 (2015), and references therein. 45 P. V. Rysselberghe, J. Phys. Chem. 36, 1152 (1932). 46 J. W. Schmidt and M. R. Moldover, Int. J. Thermophys. 24, 375 (2003). 47 C. M. Evans, K. K. Krynski, K. Yalikun, and G. L. Findley, J. Chem. Phys. (2016), Charge screening and dielectric permittivity in dense diatomic fluids, in preparation. 48 S. V. Khristenko, A. I. Maslov, and V. P. Shevelko, Molecules and Their Spectroscopic Properties (Springer-Verlag, New York, 1998). 49 U. Asaf, W. S. Felps, K. Rupnik, S. P. McGlynn, and G. Ascarelli, J. Chem. Phys. 91, 5170 (1989). 50 J. Ferch, W. Raith, and K. Schroeder, J. Phys. B: At. Mol. Phys. 13, 1481 (1980). 51 E. S. Chang, J. Phys. B: At. Mol. Phys. 14, 893 (1981). 52 A. Zecca, G. Karwasz, and R. S. Brusa, Riv. Nuovo Cimento 19, 1 (1996), and references therein. 53 I. Shimamura, Sci. Pap. Inst. Phys. Chem. Res. (Rikagaku Kenkyusho) 82, 1 (1989). 54 S. A. Lawton and A. V. Phelps, J. Chem. Phys. 69, 1055 (1978). 55 I. D. Reid and R. W. Crompton, Aust. J. Phys. 33, 215 (1980). 56 J. Randell, S. L. Lunt, G. Mrotzek, J. P. Ziesel, and D. Field, J. Phys. B: At. Mol. Opt. Phys. 27, 2369 (1994). 14 TABLE I. Intermolecular potentials used in this study. For these functions, σ and rm are collision radii (rm = 21/6 σ for the Lennard-Jones 6-12 potential), ϵ is the potential well-depth, and γ gauges the steepness of the intermolecular potential. Name Functional form [ ( rm )6 ] 6 γ(1−r/rm ) ϵ U (r) = 1−6/γ γ e − r [( ) ] ( )6 12 U (r) = 4 ϵ σr − σr Exponential-6 (Exp-6) Lennard Jones 6-12 (LJ6-12) 15 TABLE II. Intermolecular potential parameters for an exponential-6 (Exp-6) and a Lennard Jones 6-12 (LJ6-12) potential for H2 , D2 and O2 obtained from this study, as well as selected parameters from previous investigations. Fluid H2 Potential Exp-6 ϵ/kB (K) 25.6 ± 0.1 37.3 36.4 rm (Å) 3.291 ± 0.003 3.337 3.43 γ 12.0 ± 0.1 14.0 11.1 Ref this work b c H2 LJ6-12 28.3 ± 0.1 32.6 37.0 36.7 37.00 36.0 3.243 ± 0.006 3.32 3.29 3.32 3.287 3.29 — — — — — — this work a a a b f D2 Exp-6 28.4 ± 0.1 3.392 ± 0.005 12.0 ± 0.1 this work D2 LJ6-12 31.6 ± 0.1 34.4 37.0 35.2 3.358 ± 0.004 3.34 3.29 3.31 7 — — — — this work a a a O2 Exp-6 112.8 ± 0.1 125.0 125.0 99.495 3.625 ± 0.006 3.86 3.84 3.781 13.5 ± 0.1 13.2 13.0 14.45 This study d e e O2 LJ6-12 126.0 ± 0.1 106.7 116 3.587 ± 0.004 3.892 3.92 — — — this work e f a I. H. Hillier, M. S. Islam and J. Walkley, J. Chem. Phys. 43, 3705 (1965). b J. Van de Ree, J. Los ad A. E. DeVries, Physica 34, 66 (1967). c W. J. Nellis, M. Ross, A. C. Mitchell, M. van Thiel, D. A. Young, F. H. Ree and R. J. Trainor, Phys. Rev. A 27, 608 (1983). d Q. F. Chen, L. C. Cai, Y. Zhang and Y. J. Gu, J. Chem. Phys. 128, 104512 (2008). e A. Belonoshko and S. K. Saxena, Geochimica et Cosmochimica Acta 55, 3191 (1991). f M. D. McKinley and T. M. Reed III, J. Chem. Phys. 42, 3891 (1965). 16 TABLE III. The zero kinetic energy electron scattering length A, the phase shift η0 , and the average percent deviation %χnc and %χcr between the calculated quasi-free electron energy V0 (ρ) and the experimentally measured values for noncritical temperatures and for a near critical point temperature, respectively. The radial distribution functions used in these calculations are obtained using either the Lennard-Jones 6-12 potential (LJ6-12) or the exponential-6 potential (Exp-6) parameters. The zero kinetic energy electron scattering length was obtained either from Evans et al 27 using dopant field ionization (DFI) or was determined here by optimizing the fit to the quasi-free electron energy (this work). Fluid H2 Potential Exp-6 A (Å) 0.541 (DFI) 0.556 ± 0.001 (this work) η0 0.66 0.651 ± 0.002 %χnc 2.1 1.8 %χcr 2.5 1.5 H2 LJ6-12 0.541 (DFI) 0.56 ± 0.01 (this work) 0.71 0.69 ± 0.01 4.5 3.1 10 7.5 D2 Exp-6 0.575 (DFI) 0.589 ± 0.002 (this work) 0.68 0.670 ± 0.001 2.4 1.2 2.1 1.4 D2 LJ6-12 0.575 (DFI) 0.58 ± 0.01 (this work) 0.73 0.70 ± 0.01 5.1 2.1 11 6.8 O2 Exp-6 −0.091 (DFI) −0.109 ± 0.004 (this work) 0.60 0.592 ± 0.002 3.0 1.4 2.9 1.8 O2 LJ6-12 −0.091 (DFI) −0.112 ± 0.01 (this work) 0.65 0.67 ± 0.01 7.2 3.5 18 10 17 8 6 e )2 + r ( / )1 r ( × (a) 4 e 01 2 2 0 0.0 0.5 1.0 1.5 2.0 2.5 1.5 2.0 2.5 1.5 2.0 2.5 rr 8 6 e )2 + r ( / )1 r ( × (b) 4 e 01 2 2 0 0.0 0.5 1.0 rr 8 (c) e )2 + r ( / )1 r ( × 6 4 e 01 2 2 0 0.0 0.5 1.0 rr FIG. 1. The Clausius-Mossotti function [(εr − 1)/(εr + 2)], where εr ≡ ε(T, ρ)/ε0 with ε0 being the vacuum permittivity, plotted as a function of reduced fluid number density ρr ≡ ρ/ρc for (a) H2 , (b) D2 and (c) O2 . The temperatures for H2 were (•) 39.00 ± 0.05 K and (◦) 33.55 ± 0.05 K, which is near the critical temperature of Tc = 33.15 K. The temperatures for D2 were () 48.09 ± 0.05 K, (N) 43.03 ± 0.05 K and () 39.02 ± 0.05 K, which is near the critical temperature of Tc = 38.34 K. The temperatures for O2 were (H) 158.57 ± 0.05 K, () various noncritical temperatures below the critical temperature of Tc = 154.58 K, and the near critical isotherm of (△) 155.27 ± 0.05 K. The solid lines represents nonlinear least squares fits of the noncritical temperature data to [(εr − 1)/(εr + 2)] = c1 ρr + c2 ρ2r + c3 ρ3r . For H2 : ρc = 9.22 × 1021 cm−3 , c1 = 1.2 ± 0.1 × 10−2 , c2 = 7.3±0.2×10−3 and c3 = −1.7±0.1×10−3 . For D2 : ρc = 1.32×1022 cm−3 , c1 = 9.8±0.2×10−3 , c2 = 8.9±0.3×10−3 and c3 = −2.0±0.1×10−3 . For O2 : ρc = 8.20×1021 cm−3 , c1 = 4.5±0.2×10−2 , c2 = −9.1 ± 0.3 × 10−3 and c3 = −1.6 ± 0.3 × 10−3 . 18 8 8 8 (a) (b) 6 6 2 0 6 )stinu .bra( i )stinu .bra( i )stinu .bra( i 4 6 (c) 4 2 0 7 8 9 Energy (eV) 10 6 4 2 0 7 8 9 Energy (eV) 10 6 7 8 9 10 Energy (eV) FIG. 2. Photocurrent plotted as a function of photon energy for (a) H2 , (b) D2 and (c) O2 . These data were obtained near the critical point of each fluid with the temperatures and densities being (a) Tr (≡ T /Tc ) = 1.011 and ρr (≡ ρ/ρc ) = 0.962, (b) Tr = 1.018 and ρr = 1.01 and (c) Tr = 1.004 and ρr = 1.02. (The critical temperature42 Tc is 33.145 K, 38.34 K and 154.58 K; while the critical density42 ρc (in units of 1021 cm−3 ) is 9.3397, 10.435 and 8.209 for H2 , D2 and O2 , respectively.) The applied fields (from bottom to top) are (a) 10.0 kV/cm, 14.0 kV/cm, 18.0 kV/cm and 22.0 kV/cm; (b) 24.5 kV/cm, 31.0 kV/cm, 37.0 kV/cm and 44.0 kV/cm; and (c) 7.00 kV/cm, 8.50 kV/cm, 12.0 kV/cm and 14.0 kV/cm. 19 9.0 (a) 8.8 )Ve( 8.6 ( ra ) 8.4 8.2 8.0 7.8 0.00 0.05 0.10 LH LL 1/2 + 0.15 1/2 0.20 (eV) 9.2 (b) 9.0 8.8 )Ve( 8.6 ) ( ra 8.4 8.2 8.0 7.8 0.00 0.05 LH 1/2 + 0.10 LL 1/2 0.15 (eV) 8.4 (c) 8.2 )Ve( 8.0 ( ra ) 7.8 7.6 7.4 7.2 0.00 0.02 LH 0.04 1/2 + 0.06 LL 1/2 0.08 0.10 (eV) FIG. 3. The FEP intercept a(ρ) (cf. eq. (18)) of representative field enhanced photoemission plots for (a) H2 at Tr = 1.182, (b) D2 at Tr = 1.253 and (c) O2 at Tr = 1.03. In all graphs (•) represents the density of 3.00 × 10−19 cm−3 or a reduced density of ρr ≈ 3.0 × 10−3 . The reduced densities ρr are (a) (◦) 0.225, () 0.455, () 0.696, (N) 0.942, (△) 1.18, and (H) 1.36; (b) (◦) 0.144, () 0.422, () 0.719 and (N) 1.01; (c) (◦) 0.242, () 0.479, () 0.966, (N) 1.25 and (△) 1.41. See text for discussion. 20 9.0 (a) 8.8 8.6 )Ve( 8.4 ( ra ) 8.2 8.0 7.8 7.6 0.00 0.05 0.10 LH LL 1/2 + 0.15 1/2 0.20 (eV) 9.4 (b) 9.2 9.0 ) 8.4 8.6 ( ra )Ve( 8.8 8.2 8.0 7.8 7.6 0.00 0.05 0.10 LH LL 1/2 + 0.15 1/2 0.20 (eV) 8.5 (c) 8.0 )Ve( ( ra ) 7.5 7.0 6.5 0.00 0.05 LH 1/2 + 0.10 LL 1/2 0.15 (eV) FIG. 4. The FEP intercept a(ρ) of representative field enhanced photoemission plots for (a) H2 at Tr = 1.011, (b) D2 at Tr = 1.018 and (c) O2 at Tr = 1.004. In all graphs (•) represents the density of 3.00 × 10−19 cm−3 or a reduced density of ρr ≈ 3.0 × 10−3 . The densities ρ are (a) (◦) 0.311, () 0.455, () 0.696, (N) 0.964, and (△) 1.80; (b) (◦) 0.173, () 0.400, () 0.614, (N) 0.918, (△) 1.39, (H) 1.76, and (▽) 1.87; (c) (◦) 0.295, () 0.731, () 0.931, (N) 1.34, (△) 1.47, and (H) 1.65. See text for discussion. 21 1.2 (a) 1.0 0.8 )Ve( 0.6 0 V 0.4 0.2 0.0 -0.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 2.0 2.5 3.0 rr 1.2 (b) 1.0 0.8 )Ve( 0.6 0 V 0.4 0.2 0.0 -0.2 0.0 0.5 1.0 1.5 rr (c) FIG. 5. The quasi-free electron energy V0 (ρ) in (a) H2 , (b) D2 and (c) O2 – determined from eq. (20) using data in Figs. 3 and 4 (as well as data at other densities and temperatures that are not shown for brevity) – plotted as a function of reduced number density ρr . In all cases, (•) represents data obtained at noncritical temperatures above the critical temperature (i.e., Tr > 1), (, N, H) represents data obtained at noncritical temperatures where Tr < 1, and (◦) represent data obtained near the critical temperature (i.e. Tr ≈ 1.02). The lines are the optimal isotropic local Wigner-Seitz calculation. See text for discussion. 22 (a) (b) 0.7 5 0. 1 0.1 1 0.2 140 140 100 80 90 40 140 140 120 100 90 70 128 90 70 50 30 3.20 20 30 40 3.15 60 70 60 0. 11 80 0. 0. 127 180 90 41 .0 0 1 0.0 5 .5 5 .9 0 1 31 .4 0 0.6 0.5 0.4 0.5 .6 5.5 5 5.4 5. 8 0. .8 0 1 .0 0 0 126 Bk / e 0.8 125 200 200 50 3.10 124 120 )K( 0.7 5.5 3.15 3.25 6.0 0.9 1 .1 220 .0 160 0.8 1 0.11 .2 0 8.0 0.9 0.4 .0 0.5 5 3.10 123 3.30 1 1 3.20 5.3 0 0.0 Bk / e 5.5 5.4 0.3 Bk / e 3.15 9.0 3.25 )K( )K( 3.20 0.4 3.25 (c) 3.30 7 0. 3.30 123 3.10 124 125 s (Å) 126 s (Å) 127 128 123 124 100 0 10 90 125 126 127 128 s (Å) FIG. 6. Contour plots showing the percent average deviation of the pressure calculated from eq. (14) in comparison to experimental thermodynamic pressures for O2 at low densities up to the triple point liquid density at temperatures (a) above the critical temperature, (b) near the critical temperature, and (c) below the critical temperature as a function of the Lennard-Jones potential parameters ϵ/kB and σ. The solid point represents the area of overlap among the three panels. See text for discussion. 23 (a) (b) 0.69 0.69 8.5 9.5 5.5 6.5 0.68 4 7.5 4.5 5.5 8.5 9.5 4.5 5 6.5 9.5 8 h 4.5 3.5 3 5 8 7.5 4.5 3.5 6 6.5 8 7 5 6.5 9 8.5 6.5 5.5 0.57 0.58 5 0.59 7.5 8.5 4.5 0.66 0.56 4.5 0.67 5.5 7 6.5 9 6 0 h 0 0.67 8.5 7 7 5.5 13 11 10 4 0.68 5 12 10.5 6 7.5 8 6 9 11 10 9 7 13 12 10.5 0.66 0.60 10 0.56 A| (Å) 9 6 5.5 7 5 4 8 0.57 0.58 0.59 0.60 A| (Å) | | FIG. 7. Contour plots showing the percent average deviation of the quasi-free electron energy V0 (ρ) calculated from eq. (1) in comparison to V0 (ρ) in D2 obtained from field enhanced photoemission (cf. Fig. 5) at (a) a noncritical temperature (i.e., Tr = 1.253 and (b) near the critical temperature (i.e., Tr = 1.018) as a function of the zero kinetic energy electron scattering length A and the phase shift η0 . The solid point represents the area of overlap among the two panels. See text for discussion. 24