Time-Resolved Carrier Dynamics near the InsulatorMetal Transition

Chapter 11

Time-Resolved Carrier Dynamics near the Insulator—Metal Transition 1

2

3

4

3

M . J. Feldstein , C. D. Keating , W. Zheng , Y. H. Liau , A. G. MacDiarmid , Michael J. Natan , and N.F. Scherer Downloaded by COLUMBIA UNIV on September 13, 2012 | http://pubs.acs.org Publication Date: December 18, 1997 | doi: 10.1021/bk-1997-0679.ch011

2

4

1

Center for Biomolecular Science and Engineering, Naval Research Laboratory, Washington, DC 20375 Department of Chemistry, The Pennsylvania State University, University Park, PA 16802-6300 Department of Chemistry and Laboratory for Research on the Structure of Matter, University of Pennsylvania, Philadelphia, PA 19104-6323 Department of Chemistry, University of Chicago, Chicago, IL 60637 2

3

4

Carrier dynamics in polyaniline and colloidal Au films have been examined using a combined approach of time resolved laser spectroscopy and atomic force microscopy (AFM). These systems exhibit insulator-metal transitions (IMT) in conjunction with synthetic modification of their structure. The relationship between structure and reactivity, in terms of hot carrier lifetimes and transport, has been identified by correlating changes in the dynamics with the directly measured morphology. Physical insight into the processes affecting carrier lifetimes and localization and the nature of the IMT has been derived from an analysis of the experimental data. The results and conclusions presented herein have implications for directing research and development in device applications based on thin film technologies. I. Introduction Metallic systems play a prominent role in material science and physical-chemistry research. The principal attribute that defines a material as metallic may be subject to the theoretical perspective^ 1) but, charge conduction is clearly a primary characteristic. The specific nature and detailed physics of a material's carrier dynamics and dielectric response has significant implications for the resultant properties of the system. In real systems, structural aspects greatly influence these dynamics as well as their opto-electronic

© 1997 American Chemical Society

In Nanostructured Materials; Shalaev, V., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1997.

141

Downloaded by COLUMBIA UNIV on September 13, 2012 | http://pubs.acs.org Publication Date: December 18, 1997 | doi: 10.1021/bk-1997-0679.ch011

142

NANOSTRUCTURED M A T E R I A L S

properties, and, consequently, the overall material properties. For example, the optical excitation of collective modes, such as surface plasmon polaritons (SPP), depends strongly on the structural and dielectric properties at the material's interface.(2) Knowledge of the structure-reactivity relationship of materials and interfaces impacts our understanding of surface phenomenon and dynamics in thin films. This knowledge becomes crucial for the design and development of electronic and opto electronic components which are shrinking to the nanometer and single electron scale. (3) The traditional, well developed theories have primarily focused on bulk metallic systems and have relied on indirect dynamical measurements.(4) Few have tried to understand, using direct measurements, how nanometer length scale structure effects real time carrier dynamics. Theoretical models which do take into account structural effects can typically only address the relationship between structure and function for idealized or completely random geometries.(5)(6) There are theoretical models capable of calculating the dielectric response and electromagnetic interactions for arbitrary shapes.(7) However, they have not yet been widely applied and, moreover, it may not be trivial to apply these models to amorphous materials, such as conducting polymers. Direct measurements of both dynamics and structure can facilitate obtaining deeper understanding carrier dynamics and electro-optic properties in thin film metallic systems, such experiments are reported in this paper. Further, the study of insulator-metal transitions (IMTs) can provide insight into the dynamical properties of complex, disordered systems.(8)(9) The subtleties of phenomena like carrier localization and transport in systems such as conducting polymers and colloidal metal films can be probed and elucidated. This chapter reports the results of optically excited carrier dynamics near the IMT in a series of thin films composed of colloidal gold particles 12 nm in diameter and a series of thin protonated polyaniline (PANI) films of varying conductivity. Carrier lifetimes and dynamics have been probed directly by time resolved laser spectroscopy measurements. Additionally, the films' mesoscopic structure has been determined with atomic force microscopy (AFM). Correlations are made between film structure, carrier dynamics and conductivity. The issues of carrier localization, transport, and scattering are addressed. Π. Experimental Time resolved carrier dynamics were studied by way of transient transmission measure ments, performed using a standard pump-probe configuration. The laser system is based on a home-built, cavity-dumped Ti: Sapphire laser(lO) which produces sub-20 fs pulses centered at 800 nm. Pulse energy's are in excess of 50 nJ/ pulse (peak powers >10 W) and a repetition rate of 500 kHz was used. The cavity-dumped pulse is dispersion compensated using a prism-pair sequence and then split 90:10% into a two beams; a strong pump and weak probe. The pump pulse travels along a fixed optical delay line while the probe passes through a variable delay line. The two beams are focused, at near normal incidence, to a common spot on the sample, where each pump pulse has an energy of approximately 15 nJ yielding a fluence of -0.15 mJ/cm , which is in the low fluence 6

2


11.

FELDSTEIN E T A L .

Time-Resolved Carrier Dynamics

143

regime for metallic systems. (11) The samples were continuously rotated at -20 Hz about an axis normal to the surface to provide "fresh" sample and avoid excessively heating a single location.


III. Results and Discussion Films composed of multi-layers of 12 nm A u colloids tethered to the substrate were prepared according to previously published procedures.(12) Figure 1 shows the topography of two such films; a comparison of the two layer and the four layer films shows colloidal aggregation and growth of domain sizes. The A u colloidal films, which vary in thickness, from approximately 10 to 60 nm, and proportionally in domain (or aggregate) size, exhibit structurally dependent hot-electron lifetimes.(13) Specifically, the hot-electron lifetimes vary from 1-3 ps in a predictable manner with the film's growth. These responses were measured by femtosecond optical spectroscopy detecting transient photo-bleach signal arisingfromhot-electron relaxation via electron-phonon coupling and are shown in Figure 2. In a simplified model, the carrier lifetime, τ, due to electronphonon scattering is simply related by the electron heat capacity, C , as: τ = C/G, where G is the electron-phonon coupling constant. The structural dependence of τ has been accounted for by considering the effect of the domain size on the coupling constant G.(14) Two competing phenomena which determine G have been identified. The first of these, as illustrated in Figure 3, is electron oscillation frequency-phonon resonance detuning (EOPRD) which increases as the domain size, d, decreases due to an increased electron oscillation frequency. The effective oscillation frequency arises from elastic scattering of electrons from the domain boundary. The detuning is significant since without spectral overlap with the phonon spectrum the electron-phonon coupling is reduced. The second phenomenon effecting G is inelastic surface scattering (ISS) which increases as the domain size, d, decreases due to the greater number of carrier-surface collisions. Even though ISS is less effective for energy transfer than electron scatter from bulk phonons, ISS yields enhanced electron-phonon coupling since it is an additional source of coupling. The relative contribution of EOPRD and ISS can be determined by first assuming the following functional form: G / G = ISS/EOPRD , where Q$ is the electron-phonon coupling constant in the bulk limit. ISS can be calculated according to: (15) ISS = 1 + 3(1-Ρ)/8β + 7 α/5, where β is a film thickness parameter defined as d/Λ and d is taken as the average film height measured with an A F M , Λ is the effective electron mean free path, Ρ is a specular reflection parameter of the electron from the film surfaces (and is taken to be zero), α is the grain diameter parameter defined as (R/1-R)(A/D), where D is the grain size and is fixed at 12 nm for the colloids presently under study, and R is the reflectivity of electrons at grain boundaries and is taken to be -0.2.(16) The values of ISS for each film, shown graphically in Figure 4, have the expected trend of increasing with decreasing film dimension as collisions with the boundary become more frequent. The values of EOPRD have been solved for and are shown in Figure 4. In these results, the trend anticipated in the work of Tomchuk(14) is found: EOPRD increases with e

B



144


Figure 1 A F M images of A u colloid films.

Delay Time (ρs) Figure 2 Pump-probe lifetime results for A u colloid films. increases with number of layers (2-5) for samples A - E .

Conductivity


11.

FELDSTEIN E T A L .

145


α EOPRD Η

domain liameter, d

H

η

Phonon \ Spectrum \

ISS


\ œDebye

J

v /2d Electron Oscillation Frequency Fenûi

Figure 3 Electron-phonon coupling mechanims.

10

Figure 4

20

30

40

Average Film Height (nm)

Factors affecting electron-phonon coupling vs. Film height.



146


decreasing dimensions as the electron oscillation frequency goes farther out of resonance with the upper hrnit of the phonon spectrum. The calculated EOPRD for the thinnest film begins to approach yet never reaches the 2 orders of magnitude predicted by Tomchuk for nanometer scale metal islands. This is not surprising since the reflectivity factor, R, used to calculate ISS may not appropriately take into account surface defects and traps or the elliptical boundary conditions relevant to the colloids. These factors could yield a reduced value for R and increased coupling leading to a larger theoretically estimated value for EOPRD. Nevertheless, the fact that both trends agree with theoretical predictions makes it possible to conclude that the two phenomenon are in competition and that a balance between them yields the effective electron-phonon coupling determined experimentally herein. It is evident from the present results that the electrons exhibit inter-colloid mobility even though the colloids are not in direct metallic contact. Carrier scattering time and EOPRD, as shown in Figure 4, correlate with film growth by colloidal aggregation. A comparison of EOPRD and the electron oscillation frequency, where the domain size used to determine the electron oscillationfrequencyis the average film thickness (related to the aggregate size), not the single particle radius, also shows a positive correlation.(17) Thus, if the EOPRD changes with film growth it can only be the case that the electrons are not confined to the volume of a single colloid but are mobile within the aggregated domains. Electron mobility in this system is also consistent with the fact that electrons exhibit enhanced tunneling probability (vs. vacuum) through functionalized alkane chains.(18)(19)(20) However, the electron dynamics in colloidal A u films are different than in bulk metals where electrons are fully mobile. Specifically, a comparison (not shown) of EOPRD with the calculated electron oscillation frequency shows a non-zero intercept which suggests that even for films with the greatest aggregation {i.e., smallest frequency difference) the electrons lack compete mobility and the colloidal size still maintains an influence on the e-ph coupling.(13) The coupling between aggregated particles has been shown to be of significant importance for the optical and electronic properties of the colloidal films. This coupling takes the form of both the well known dipolar coupling to produce the collective plasmon optical resonance at -800 nm and the newly reported inter-colloid coupling that increases the hot-electron decay rate due to inelastic electron-phonon collisions. The mesoscopic structure of the films has been measured with scanning probe microscopy techniques and a simple model is reported which accounts for the dependence of the hot-electron lifetime on the films' structure. Finally, the results consistent with the DC electrical conductivity mechanism of percolation based on activated hopping.(13) Thin (-40 nm) polyaniline emeraldine base films, doped with camphorsulfonic acid and spin cast from chloroform solutions, were prepared in a range of conductivities by "secondary doping" with m-cresol vapor. Figure 5 shows the A F M topography for low (σ = 0.03 S/cm) and high (σ = 0.9 S/cm) conductivity spin cast films. Secondary doping functions through enhanced solvation of the charged {i.e., doped) polymer chains. This process promotes rearrangement into more favorable molecular confirmations and tends to uncoil the polymer chains.(21) The result of secondary doping is better intra-chain



11.

FELDSTEIN E T A L .


Figure 5 A F M images of low and high conductivity poly aniline films.


147


148


homogeneity and inter-chain contact and interaction thereby yielding higher conductivity films. The more highly conductive material might be expected to have a more "smooth" and uniform morphology, as observed in Figure 5, since the degree of crystallinity (or order) increases. By contrast, the disordered low conductivity films would have a more heterogeneous morphology. Two time scales for optically excited carrier relaxation are observed in optical pump-probe experiments and are shown in Figure 6. However, only the relatively slowly relaxing carriers which have exponenetial decay constants in the picosecond range change as the measured conductivity in poly aniline changes: the results for samples Β (σ=0.033 S/cm) and Ε (σ= 0.940 S/cm) in Figure 6 indicate that the trend in the 'slow' carrier scattering time correlates with the conductivity. The results of Figures 5 and 6 indicate that the lifetime of weakly localized carriers increases with decreasing disorder. These conclusions regarding the primary importance of the slow carriers are rather interesting since it has been shown herein and previously(9) that only a small fraction of the total carrier population is quasi-localized and thereby relax with the longer scattering times. Presumably, these carriers are quasi-ID localized, as identified by Kohlman et al, where the localization is the result of inhomogeneous disorder, such as crystalline defects or impurities.(9) Since the scattering time increases with decreased disorder it seems clear that the localization length also increases and that the quasi-ID carriers contribution to the film's conductivity is based on the percolation mechanism. Figure 7 depicts the idea that a localization length, L , is associated with the carriers in materials governed by the percolation mechanism. When this length exceeds the inter-metallic spacing for a statis-tically significant percent of the film (50% for a 2-D film) an IMT occurs and conductivity rises exponentially. An IMT may occur when either L or the density of metallic centers is increased. Associated with increased DC conductivity, o , is an increased carrier scattering time, τ, which according to the Drude model is: σ ω 2 τ / 4 π , where ω is the plasma frequency. The correlation of increased conductivity with increased order is evident in the A F M images of Figure 5. Further, the height correlation function analysis (not shown) of these images shows quantitatively that the low conductivity film is dominated by short range order and large roughness while the high conductivity film exhibits long range order and less roughness indicating a film that is more ordered. The results of time domain measurements of the carrier dynamics in PANI films near the IMT suggest that scattering times and conductivity are not necessarily directly related as predicted by the Drude model. That is, measuring one property is not sufficient to uniquely determine the other quantity. Rather, the direct measurement of both, as well as the plasma frequency, would provide a better characterization of the dynamical time scales and theoretical predictions of conductivity. Still, predictions from the Drude model do agree within an order of magnitude of the measured values. loc

l o c

DC

πο=

ρ

ρ

IV. Conclusion The implications of this work may be valuable to direct research and development in device application. For example, from a synthetic perspective, this work confirms the understanding of the need for increased film order to achieve maximal conductivity. Both


FELDSTEIN E T A L .

1 0


El 9 H 8 H

j

7 6 "


5 "

Figure 6 Pump-probe measurement of carrier relaxation in doped polyaniline films.

Figure 7 Cartoon of ordered and disordered regions of polyaniline film.


150



of the studies of these thin film systems that undergo an IMT have implications for the design and development of devices. For example, the development of opto-electronic devices necessitates a working understanding of the relationship between structural features and performance characteristics. The results reported herein suggest a guide for the design of opto-electronic devices with synthetically tuneable hot-carrier lifetimes. Specifically, since both of these materials exhibit large nonlinear optical responses, the combination of the two could yield the development of devices where the optical response as well as the carrier dynamics are adjustable parameters. Designed control over the system properties (i.e., carrier lifetimes, etc.) could potentially yield better optimized device performance.

Literature Cited 1. C.M. Hurd, Electrons in Metals, R.E. Kreiger Publishing Co., Florida, 1981, chapter 2. 2. H. Raether, Surface Plasmons, Springer-Verlag: Berlin(1988). 3. J.I. Pascual, J. Méndez, J. Gómez-Herrero, A.M. Baró, N. Garcia, U. Landman, W. Luedtke, E.M Bognachek, H.-P. Cheng, Science, 267, 1753 (195). nd

4. Sir N.F. Mott, Conduction in Non-Crystalline Matterials, 2 Ed., Clarendon Press: Oxford (1993). 5. M. Moskovits, Rev.Mod.Phys., 53(3), 783 (1985). 6. M.I. Stockman, T.F. George, V.M. Shaleav, Phys.Rev.B, 44, 115 (1991). 7. W.H. Yang, G.C. Schatz, R.P Van Duyne, J.Chem.Phys., 103, 869 (1995). 8. Sir N.F. Mott, Metal-Insulator Transitions, Taylor & Francis: London (1990). 9. R.S. Kohlman, J. Joo, Y.G. Min, A G . MacDiarmid, A.J. Epstein, Phys.Rev.Lett, 77(13), 2766 (1996). 10. D.C. Arnett, P. Vöhringer, N.F. Scherer, J.Am.Chem.Soc., 117(49), 12262 (1995). 11. H.E. Elsayed-Ali, T. Juhasz, G.O. Smith, W.E. Bron, Phys.Rev.B., 43, 4488, (1991). 12. R.G. Freeman, K.C. Grabar, K.J. Allison, R.M. Bright, J.A. Davis, A.P. Guthrie, M.B. Hommer, M.A. Jackson, P.C. Smith, D.W. Walter, M.J. Natan, Science, 267, 1629 (1995). 13. M.J. Feldstein, C. D. Keating, Y.H. Liau, M.J. Natan, N.F. Scherer, J.Am.Chem.Soc., in press 1997. 14. S.A. Gorban, S.A. Nepijko, P.M. Tomchuck, Int.J.Electronics, 70(3), 485 (1991).


11.

FELDSTEIN E T AL.


151

15. T.Q. Qui, C.L. Tien, Int.J.Heat.MassTransfer, 167, 25 (1988). 16. J.W.C. de Vries, Thin.Sol.Films., 1988 167, 25. 17. The frequency difference is calculated as: v / - ω where y is the electron (Fermi) velocity, is the average film height, as determined by AFM, and ω , the Debye frequency, is 2.2 x 10 s . F

D

D

13

-1

14


18. Finklea, H. O.; Hanshew, D. D.; J.Am.Chem.Soc.; 1992, 114, 3173.

10

8

19. Smalley, J. F.; Feldberg, S. W.; Chidsey, C. E. D.; Linford, M. R.; Newton, M. D.; Liu, Y-P; J.Phys.Chem, 1995, 99, 13141. 20. The through-bond electron tunneling rate constant decay parameter has been found to be on the order of 1 per methyl group. Given the limiting rate constant, k = 6 x sec , the number of methyl groups in the crosslinker, the number of carriers per colloid as 5 x 10 , and assuming a first order rate equation, one would expect an electron transfer rate on the order of 100 femtoseconds. If the number of crosslinkers between adjacent colloids is taken into account an even high rate would be predicted. Thus, the rate of electron mobility is significantly faster than the e-ph coupling times reported herein and it is reasonable to predict that electron dynamics within a given colloid are sensitive to the presence of adjacent colloids. 18,19

n-0

-1

4

21. A.J. Epstein, A.G. MacDiarmid, Synth.Met., 69, 179 (1995).


Time-Resolved Carrier Dynamics near the InsulatorMetal Transition

Recommend Documents