- Email: [email protected]
Nanocrystal Self Assemblies: Fabrication and collective properties
Studies in Surface Science and Catalysis 132 Y. Iwasawa, N. Oyama and H. Kunieda (Editors) c 2001 Elsevier Science B.V. All rights reserved.
237
Nanocrystal Self Assemblies: Fabrication and collective properties M.P.Pileni
Laboratoire S.R.S.I, URA CNRS 1662, Universite P. at M. Curie (Paris VI), BP 52, 4 place Jussleu, 75252 Paris cedex 05, France. I. Fabrication Nanocrystal are made by using reverse micelles (1-3). At the end of the synthesis, the nanocrystals are coated with dodecanethiol . To get 2D and 3D superlattices, the deposition procedure and the nanocrystal concentration play a key roles. A drop of the solution is deposited on HOPG substrate with a filter paper underneath. The solution migrate from the substrate to the filter paper and after few seconds the solvent Is totally evaporated. At rather low concentration (particle volume fraction, (t)=0.01%), monolayer made of nanocrystals are observed on a very large domain (50^m) (4). The nanocrystals are organized in compact hexagonal network. The average distance between particles is 1.8nm. Immediately after the solution containing the nanocrystals is deposited on the substrate, the solvent begins to evaporate and droplets form. The nanocrystals themselves are fully solvated (or "dressed") by the heptane, which prevents their assembly into dense structures (5). As the droplets grow and begin to merge, some of the particles (which are still mobile because of the thin solvent layer present on the HOPG surface) are expelled away from the merge center. These dressed particles form compact monolayer islands, whose density increases after all of the solvent evaporates and interdigitation of the alkyl chains on the nanocrystal occurs. Other particles are caught in the center of the droplet merge point. The pressure exerted on these particles by the droplet menisci is large, and while a monolayer initially form, continued
238 droplet coalescence engenders the formation of a 3D structure. The three dimensional structure of dressed particles dries out as the solvent evaporates, and thus interdigitation of the particles' alkyl chain coating occurs in 3D instead of 2D. The sizes of the 3D aggregates are similar, which implies that the merge regions between growing solvent droplets are also similar in size. Enhancement of the TEM pattern shows that the nanocrystals self assemble In four-fold symmetry (6,7). This can be attributed to the [001] plane of an FCC lattice can easily be seen. The center-to-center distance between
two
nanocrystals along the [010] plane is ca. 11 nm. The average particle diameter is 5.8 nm, and the shortest center-to-center distance is 7.8 nm, leaving a 2 nm edge-to-edge separation that is consistent with alkyl chain interdigitation. By deposition of a drop of solution containing nanocrystal on HOPG substrate with an anticapillary tweezer formation of rings instead of monolayer self organized in compact hexagonal network (8). Similar behavior is observed with spherical silver, cobalt, ferrite and with flat tringular CdS nanocrystals. When nanocrystals dispersed in hexane are deposited on a TEM grid under a "quasi" saturated atmosphere, they are then randomly dispersed without any ring formation. Such a change In the nanocrystal organization from rings to a random dispersion is due to the evaporation rate. As matter of fact, the evaporation time (3 minutes) Increases compared to that under air
(30
seconds). Hence formation of rings made of nanocrystals are related to the evaporation rate. This is strongly confirmed by the calculated values of temperature gradient (AT) and Marangoni number (Mg). The estimated values of temperature gradient and Marangoni number are 29 and 10^ under air and 4.8 and 1.8.10"^ under saturated hexane respectively. Hence, the decrease in the evaporation rate induces a decrease in the AT and M^. This induces a decrease in the instabilities. By reducing the evaporation time, the system equilibrates faster than the heat loss by the evaporation process. Under such conditions, instabilities disappear and nanocrystals are randomly distributed
239
on the carbon film. This means that formation of rings is related to the instabilities, Induced by a fast evaporation process. The physical properties of the nanomaterials used (semiconductors, metals, oxides) and their shape (spheres or triangle) are not related to ring formation.
II.
COLLECTIVE
OPTICAL
AND
ELECTRONIC
TRANSPORT
PROPERTIES Both the experimental
and simulated
absorption
spectra
of silver
nanocrystals show a decrease in the plasmon resonance band intensity and increase in bandwidth with decreasing particle size. When silver nanocrystals are organized into a 2D lattice, the plasmon resonance peak Is shifted to energies lower than what is obtained for dilute solutions of isolated particles. UV/vis polarization spectroscopy can reveal information about interparticle electromagnetic interactions.
In s-polarizatlon, the electric field vector is
oriented parallel to the plane of the substrate at all incidence angles 9. Plasmon resonance modes with components polarized perpendicular to the plane of the substrate are not seen when the incident light Is s-polarized. On the other hand, p-polarized light, whose electric field is parallel to the plane of incidence can probe plasmon resonance excitations whose components are either parallel or perpendicular to the substrate. In s-polarlzation, the absorption spectra are virtually independent of incidence angle and show a plasmon resonance band centered at 2.9 eV, which is similar to that seen in isolated silver nanocrystals. In p-polarization, a second band appears at higher energy as the incidence angle is increased. At large angle (60°), the two peaks are well-defined: the first is close in energy (2.8 eV) to the absorption maximum for isolated particles (2.9 eV), but the second is centered at ca. 3.8 eV. The high energy band at 3.8 eV is attributed to the self-organization of the
240 silver nanocrystals into a hexagonal network (10). The position of the peak can be explained in terms of local field effects. When a single silver nanoparticle is deposited on a gold 111 substrate, the scanning tunneling spectroscopy
measurement indicates a double tunnel
junction. Upon increasing the applied bias voltage V, the capacitor elements (defined by the tip-particle interface and particle-substrate charged up,
interface) are
and the detected current I is inititally close to zero. Above a
certain threshold voltage, electrons can tunnel through the interfaces and the current increases with the applied voltage. A plot of dl/dV versus V clearly shows that the derivative reaches zero at zero V. The non-linear profile of the l(V) curve and zero dl/dV at zero bias voltage are characteristic of the wellknown Coulomb blockade effect. For silver nanoparticles self-organized in a 2D superlattice on an Au(111) substrate, the current is an order of magnitude lower than that observed for Isolated particles. The Coulomb gap Is small (ca. 0.45 V, compared to the 2V seen in the isolated particles), and the overall l(V) curve is more linear. This indicates an increase in the ohmic contribution to the current.
In other words, the tunneling contribution to the total current
decreases, and more conductive pathways between particles are established. The derivative curve indicates a metallic conduction behavior with dl/dV ^^ 0 at zero bias voltage.
From the l(V) and dl/dV curves, it can be concluded that
when the particles are arranged in a 2D lattice, the tunneling current exhibits both metallic and Coulomb contributions. This indicates that lateral tunneling between adjacent particles is very important and contributes to the total electron transport process. When silver nanocrystals are assembled in a 3D FCC structure, the l(V) curve shows a linear ohmic behavior. The dl/dV curve is essentially flat, indicating metallic behavior without Coulomb staircases. The ohmic behavior cannot be attributed to the coalescence of the particles. Thus we conclude that the FCC structure of the superlattice induces an increase in the tunneling rate via a decrease in resistance between the particles.
The
241
electron tunneling between adjacent particles becomes a major contribution to conduction, and the Coulomb blockade effect in the l(V) curves is inhibited (11).
The mechanism may involve an enhanced dipole-dipole interaction
along the vertical (z) axis. When subjected to a voltage bias, the Fermi level of the individual nanocrystals is also perturbed.
The details remain to be
uncovered, but it is clear that a supercrystal of coated metal nanoparticles can behave as a metal.
III. COLLECTIVE MAGNETIC PROPERTIES A comparison of the magnetic properties of isolated magnetic nanocrystals and 2D hexagonal assemblies reveals cooperative effects in the latter system (12,13). The magnetization curves for cobalt nanocrystals deposited on an HOPG substrate, for magnetic fields applied parallel and perpendicular to the substrate are compared to that observed when nanocrystals are isolated in a matrix. When the magnetic field is parallel to the substrate, the Mr/Ms ratio is 0.60, and the hysteresis loop is squarer than that obtained for particles In solution. When the field is perpendicular to the substrate, the loop is less square, and the Mr/Ms ratio decreases to 0.40. These results show that for a given saturation magnetization, the remanence magnetization markedly varies with the orientation of the magnetic field. The observed changes cannot be attributed to coalescence of the nanocrystals, since TEM images taken over large areas of the sample show no evidence of this. Among the possible explanations for the change in magnetic properties (isolated particles versus 2D assemblies) is magnetic dipolar coupling between particles. This lead to enhanced magnetization. The scenario for dipole coupling enhancements is an enhancement due to the long-range order of the 2D lattice and collective "flips" of the magnetic dipoles. The dependence of the magnetization cun/es on field orientation is calculated. The simulated cun/es resemble the experimental
242
curves showing variations when the magnetic field is applied parallel and perpendicular
to the
substrate
(14).
From these
theory-experiment
comparisons, it seems reasonable to conclude that the collective magnetic properties observed when the cobalt particles are arranged in a 2D lattice are due to an increase in magnetic dipole-dipole interactions. REFERENCES 1. M.P. Pileni, ed. Reactivity in Reverse Micelles. Amsterdam, Elsevier, 1989. 2. M.P. Pileni, J Phys Chem 97 (1993) 6961. 3.. M.P. Pileni,. Langmuir 13 (1997) 3266. 4. L Motte, F Billoudet, MP Pileni. J Phys Chem 99 (1995) 16425. 5. L Motte, E Lacaze, M Maillard, MP Pileni. Langmuir 16, (2000) 3803. 6.L Motte, F Billoudet, E Lacaze, MP Pileni. Adv Mater. (1996) 8:1018- , 1996. 7. L Motte, F Billoudet, E Lacaze, J Douin, MP Pileni. J.Phys.Chem, B 101 (1997),138. 8. M Maillard, L Motte, T Ngo, MP Pileni J.Phys.Chem. (2000) in press 9.A Taleb, C Petit, MP Pileni. J.Phys.Chem, B 102 (1998) 2214 10. A Taleb. V Russier, A Courty, MP Pileni. Phys Rev B 59 (1999) 13350. 11. A.Taleb, F.Silly, O.Gusev, F.Charra, M.P.Pileni, Adv. Mat.12, (2000), 119. 12. C. Petit, A Taleb, . M.P. Pileni,. Adv.Mater 10 (1998) 259.13. C Petit, A Taleb, MP Pileni. J. Phys; Chem. B 103 (1999) 1805. 14. V. Russier, C. Petit, J. Legrand,. M.P. Pileni Phys. Rev. B 62 (2000) 3910.
237
Nanocrystal Self Assemblies: Fabrication and collective properties M.P.Pileni
Laboratoire S.R.S.I, URA CNRS 1662, Universite P. at M. Curie (Paris VI), BP 52, 4 place Jussleu, 75252 Paris cedex 05, France. I. Fabrication Nanocrystal are made by using reverse micelles (1-3). At the end of the synthesis, the nanocrystals are coated with dodecanethiol . To get 2D and 3D superlattices, the deposition procedure and the nanocrystal concentration play a key roles. A drop of the solution is deposited on HOPG substrate with a filter paper underneath. The solution migrate from the substrate to the filter paper and after few seconds the solvent Is totally evaporated. At rather low concentration (particle volume fraction, (t)=0.01%), monolayer made of nanocrystals are observed on a very large domain (50^m) (4). The nanocrystals are organized in compact hexagonal network. The average distance between particles is 1.8nm. Immediately after the solution containing the nanocrystals is deposited on the substrate, the solvent begins to evaporate and droplets form. The nanocrystals themselves are fully solvated (or "dressed") by the heptane, which prevents their assembly into dense structures (5). As the droplets grow and begin to merge, some of the particles (which are still mobile because of the thin solvent layer present on the HOPG surface) are expelled away from the merge center. These dressed particles form compact monolayer islands, whose density increases after all of the solvent evaporates and interdigitation of the alkyl chains on the nanocrystal occurs. Other particles are caught in the center of the droplet merge point. The pressure exerted on these particles by the droplet menisci is large, and while a monolayer initially form, continued
238 droplet coalescence engenders the formation of a 3D structure. The three dimensional structure of dressed particles dries out as the solvent evaporates, and thus interdigitation of the particles' alkyl chain coating occurs in 3D instead of 2D. The sizes of the 3D aggregates are similar, which implies that the merge regions between growing solvent droplets are also similar in size. Enhancement of the TEM pattern shows that the nanocrystals self assemble In four-fold symmetry (6,7). This can be attributed to the [001] plane of an FCC lattice can easily be seen. The center-to-center distance between
two
nanocrystals along the [010] plane is ca. 11 nm. The average particle diameter is 5.8 nm, and the shortest center-to-center distance is 7.8 nm, leaving a 2 nm edge-to-edge separation that is consistent with alkyl chain interdigitation. By deposition of a drop of solution containing nanocrystal on HOPG substrate with an anticapillary tweezer formation of rings instead of monolayer self organized in compact hexagonal network (8). Similar behavior is observed with spherical silver, cobalt, ferrite and with flat tringular CdS nanocrystals. When nanocrystals dispersed in hexane are deposited on a TEM grid under a "quasi" saturated atmosphere, they are then randomly dispersed without any ring formation. Such a change In the nanocrystal organization from rings to a random dispersion is due to the evaporation rate. As matter of fact, the evaporation time (3 minutes) Increases compared to that under air
(30
seconds). Hence formation of rings made of nanocrystals are related to the evaporation rate. This is strongly confirmed by the calculated values of temperature gradient (AT) and Marangoni number (Mg). The estimated values of temperature gradient and Marangoni number are 29 and 10^ under air and 4.8 and 1.8.10"^ under saturated hexane respectively. Hence, the decrease in the evaporation rate induces a decrease in the AT and M^. This induces a decrease in the instabilities. By reducing the evaporation time, the system equilibrates faster than the heat loss by the evaporation process. Under such conditions, instabilities disappear and nanocrystals are randomly distributed
239
on the carbon film. This means that formation of rings is related to the instabilities, Induced by a fast evaporation process. The physical properties of the nanomaterials used (semiconductors, metals, oxides) and their shape (spheres or triangle) are not related to ring formation.
II.
COLLECTIVE
OPTICAL
AND
ELECTRONIC
TRANSPORT
PROPERTIES Both the experimental
and simulated
absorption
spectra
of silver
nanocrystals show a decrease in the plasmon resonance band intensity and increase in bandwidth with decreasing particle size. When silver nanocrystals are organized into a 2D lattice, the plasmon resonance peak Is shifted to energies lower than what is obtained for dilute solutions of isolated particles. UV/vis polarization spectroscopy can reveal information about interparticle electromagnetic interactions.
In s-polarizatlon, the electric field vector is
oriented parallel to the plane of the substrate at all incidence angles 9. Plasmon resonance modes with components polarized perpendicular to the plane of the substrate are not seen when the incident light Is s-polarized. On the other hand, p-polarized light, whose electric field is parallel to the plane of incidence can probe plasmon resonance excitations whose components are either parallel or perpendicular to the substrate. In s-polarlzation, the absorption spectra are virtually independent of incidence angle and show a plasmon resonance band centered at 2.9 eV, which is similar to that seen in isolated silver nanocrystals. In p-polarization, a second band appears at higher energy as the incidence angle is increased. At large angle (60°), the two peaks are well-defined: the first is close in energy (2.8 eV) to the absorption maximum for isolated particles (2.9 eV), but the second is centered at ca. 3.8 eV. The high energy band at 3.8 eV is attributed to the self-organization of the
240 silver nanocrystals into a hexagonal network (10). The position of the peak can be explained in terms of local field effects. When a single silver nanoparticle is deposited on a gold 111 substrate, the scanning tunneling spectroscopy
measurement indicates a double tunnel
junction. Upon increasing the applied bias voltage V, the capacitor elements (defined by the tip-particle interface and particle-substrate charged up,
interface) are
and the detected current I is inititally close to zero. Above a
certain threshold voltage, electrons can tunnel through the interfaces and the current increases with the applied voltage. A plot of dl/dV versus V clearly shows that the derivative reaches zero at zero V. The non-linear profile of the l(V) curve and zero dl/dV at zero bias voltage are characteristic of the wellknown Coulomb blockade effect. For silver nanoparticles self-organized in a 2D superlattice on an Au(111) substrate, the current is an order of magnitude lower than that observed for Isolated particles. The Coulomb gap Is small (ca. 0.45 V, compared to the 2V seen in the isolated particles), and the overall l(V) curve is more linear. This indicates an increase in the ohmic contribution to the current.
In other words, the tunneling contribution to the total current
decreases, and more conductive pathways between particles are established. The derivative curve indicates a metallic conduction behavior with dl/dV ^^ 0 at zero bias voltage.
From the l(V) and dl/dV curves, it can be concluded that
when the particles are arranged in a 2D lattice, the tunneling current exhibits both metallic and Coulomb contributions. This indicates that lateral tunneling between adjacent particles is very important and contributes to the total electron transport process. When silver nanocrystals are assembled in a 3D FCC structure, the l(V) curve shows a linear ohmic behavior. The dl/dV curve is essentially flat, indicating metallic behavior without Coulomb staircases. The ohmic behavior cannot be attributed to the coalescence of the particles. Thus we conclude that the FCC structure of the superlattice induces an increase in the tunneling rate via a decrease in resistance between the particles.
The
241
electron tunneling between adjacent particles becomes a major contribution to conduction, and the Coulomb blockade effect in the l(V) curves is inhibited (11).
The mechanism may involve an enhanced dipole-dipole interaction
along the vertical (z) axis. When subjected to a voltage bias, the Fermi level of the individual nanocrystals is also perturbed.
The details remain to be
uncovered, but it is clear that a supercrystal of coated metal nanoparticles can behave as a metal.
III. COLLECTIVE MAGNETIC PROPERTIES A comparison of the magnetic properties of isolated magnetic nanocrystals and 2D hexagonal assemblies reveals cooperative effects in the latter system (12,13). The magnetization curves for cobalt nanocrystals deposited on an HOPG substrate, for magnetic fields applied parallel and perpendicular to the substrate are compared to that observed when nanocrystals are isolated in a matrix. When the magnetic field is parallel to the substrate, the Mr/Ms ratio is 0.60, and the hysteresis loop is squarer than that obtained for particles In solution. When the field is perpendicular to the substrate, the loop is less square, and the Mr/Ms ratio decreases to 0.40. These results show that for a given saturation magnetization, the remanence magnetization markedly varies with the orientation of the magnetic field. The observed changes cannot be attributed to coalescence of the nanocrystals, since TEM images taken over large areas of the sample show no evidence of this. Among the possible explanations for the change in magnetic properties (isolated particles versus 2D assemblies) is magnetic dipolar coupling between particles. This lead to enhanced magnetization. The scenario for dipole coupling enhancements is an enhancement due to the long-range order of the 2D lattice and collective "flips" of the magnetic dipoles. The dependence of the magnetization cun/es on field orientation is calculated. The simulated cun/es resemble the experimental
242
curves showing variations when the magnetic field is applied parallel and perpendicular
to the
substrate
(14).
From these
theory-experiment
comparisons, it seems reasonable to conclude that the collective magnetic properties observed when the cobalt particles are arranged in a 2D lattice are due to an increase in magnetic dipole-dipole interactions. REFERENCES 1. M.P. Pileni, ed. Reactivity in Reverse Micelles. Amsterdam, Elsevier, 1989. 2. M.P. Pileni, J Phys Chem 97 (1993) 6961. 3.. M.P. Pileni,. Langmuir 13 (1997) 3266. 4. L Motte, F Billoudet, MP Pileni. J Phys Chem 99 (1995) 16425. 5. L Motte, E Lacaze, M Maillard, MP Pileni. Langmuir 16, (2000) 3803. 6.L Motte, F Billoudet, E Lacaze, MP Pileni. Adv Mater. (1996) 8:1018- , 1996. 7. L Motte, F Billoudet, E Lacaze, J Douin, MP Pileni. J.Phys.Chem, B 101 (1997),138. 8. M Maillard, L Motte, T Ngo, MP Pileni J.Phys.Chem. (2000) in press 9.A Taleb, C Petit, MP Pileni. J.Phys.Chem, B 102 (1998) 2214 10. A Taleb. V Russier, A Courty, MP Pileni. Phys Rev B 59 (1999) 13350. 11. A.Taleb, F.Silly, O.Gusev, F.Charra, M.P.Pileni, Adv. Mat.12, (2000), 119. 12. C. Petit, A Taleb, . M.P. Pileni,. Adv.Mater 10 (1998) 259.13. C Petit, A Taleb, MP Pileni. J. Phys; Chem. B 103 (1999) 1805. 14. V. Russier, C. Petit, J. Legrand,. M.P. Pileni Phys. Rev. B 62 (2000) 3910.