Samorí , Paolo : Self-assembly of conjugated (macro)molecules: nanostructures for molecular electronics

68

Kapitel 6. Results and discussions

In this chapter are presented the main experimental results with relative discussions.

6.1 Self-assembly of thiols on metallic substrates

6.1.1 Introduction

The electronic properties of single molecules or of a small ensemble of molecules can be probed following two different strategies as introduced in paragraph 4.3. The first of them is using the STM set-up to measure the conductivity of a highly ordered adsorbate. In this context, self-assembly of thiol functionalized molecules on metallic substrates [Ulm91] has been studied intensively as a model for controlled solid-liquid interface reactions. Gold has been widely used as a substrate because it is a rather inert metal for many gases and liquids [Ham95] and in addition alkanethiols pack on it in a simple stable commensurate radicradic3R30° overlayer with a lattice parameter of a=5.00 Å [Del96]. Such self-assembled monolayers (SAM)s have been characterized both in Ultra-High Vacuum (UHV) and in ambient conditions by Fourier Transform Infra-Red spectroscopy [Por87, Nuz90], Second Harmonic Generation [Buc91], X-ray Photoelectron Spectroscopy [Bai89a,Nuz87, Bai89b], ellipsometry [Por87, Chi90], Grazing Incidence X-ray Diffraction [Cam93a], Transmission Electron Diffraction [Str88], Near Edge X-ray Absorption Fine Structure Spectroscopy [Häh93], He scattering [Cam93b, Chi89], Atomic Force Microscopy [Alv92] and Scanning Tunneling Microscopy (STM) [Wid91,Kim92,Del96,Poi97]. The mechanism of alkanethiol self-assembly has been recently studied by Poirer and Pylant by means of UHV-STM measurements [Poi96]. On the other hand, a very little work has been performed on SAMs of thiols on Ag(111) so far. By FTIR spectroscopy and Raman spectroscopy significantly smaller tilt angles than on Au(111) were reported. In UHV, dosing of dimethyl disulfide on annealed Ag(111) results in a LEED pattern, which can be indexed as two domains of a (v7Xv7)R10.9° coincidence structure. It is explained by the cleavage of the S-S bond to form a methanethiolate film. The same structure was found previously for adsorption of H2S and sulfur on Ag(111). It was assigned to two dimensional crystals of gamma-Ag2S. For octadecanethiolate, however, Eisenberger and coworkers proved a larger adsorbate lattice using LEHD and GIXD. Recently Heinz and Rabe provided insight into alkanethiols SAMs


69

on Ag(111) by studying the packing and conductivity of the adlayer at increasing lengths of the aliphatic chain from CH3SH to C11H23SH [Hei95]. In the first part of this chapter, varying the metallic substrate and the chemisorbed type of thiol-functionalized molecule, the structure and the electrical conductivity of the formed SAM are investigated using STM.

6.1.2 Sublimed Au and Ag substrates

Key issues for a reproducible SAM formation are the flatness and the crystallinity of the metallic substrate. Therefore metallic supports have been developed using different procedures.

Figure 6.1: STM images of Ag(111). Flat crystallites with sharp angles of 120°.
Tip bias (Ut)= 1.5 V; Average tunneling current (It) =0.03 nA.

First, Ag and Au surfaces were sublimed on freshly cleaved muscovite mica discs in high vacuum (HV) varying systematically the parameters of the deposition, i.e. the temperature of the substrate during sublimation (Tsub) and the nominal thickness of the metallic adlayer (Z), in order to find conditions leading to a pseudo epitaxial film. In the case of silver, the best surfaces have been obtained at Tsub=275 °C and Z=50 nm. The Ag films, on a micrometer scale, exhibit atomically flat terraces with hexagonal domains (Fig. 6.1) with 120° angles that indicate macroscopic evidence for epitaxial growth.

Unfortunately, the surface of Ag in contact with air gets oxidized very rapidly. Indeed films that have been exposed to air for 30 minutes show a crystalline lattice with a spacing of 5 Å (Fig. 6.2) that can be assigned to Ag2O. This oxidation is confirmed by the presence of clusters on micrometer scale images. Because of this reason, a series of precautions have


70

been taken, including venting the UHV chamber with argon, and bathing Ag films in organic solutions under Ar reflux. In addition, the STM set-up was put in a box (box B in Fig. 5.4) which could be filled with an inert gas (Ar, He, N2).

Figure 6.2: STM constant current image of uncoated Ag surface after 30 min.exposure to air.

Also gold surfaces have been prepared by sublimation in HV. Film sublimed at Tsub=350 °C and Z= 50 nm have exhibited a polycrystalline morphology with crystallites extending on a scale of several tens of nanometers (Fig. 6.3).

Subliming Au films at Tsub=400 °C and Z= 200 nm lead to films that exhibit a morphology made of flat crystallites extending on an area of some µm2 (Fig. 6.4). The surface flatness was characterized by its root mean square roughness:

Rrms= (6.1)


71

Figure 6.3: STM Constant current image of uncoated Au surface (Tsub=350 °C, Z= 50nm), exhibiting a polycrystalline structure. The gray scale height is (h)=20 nm.

Figure 6.4: STM constant current image of uncoated Au surface (Tsub=400 °C, Z= 200nm),
exhibiting epitaxial terraces. The gray scale height is (h)=1.78 nm.

where N×N is the number of pixels (512*512), hmn is the height value of the pixel mn and is the mean height of the pixel calculated from the N×N values. It was also characterized by its average roughness: RA= (6.2)

On an area of 1µm2 these parameters amount to Rrms=19 Å and RA= 15 Å, respectively.

With the aim of increasing the flatness and epitaxial character of the Au surface, different types of thermal annealing post treatments have been carried out. All of them were very fast processes where the temperature could hardly be controlled.


    72

  1. Flame-annealing (Fig. 6.5);
  2. Annealing with a halogen lamp from the front side (Fig. 6.6);
  3. Annealing with a halogen lamp from the rear side (Fig. 6.7).

Figure 6.5: STM constant current images of Au flame-annealed sample recorded in different areas of the surface. A) Ut = 0.1 V; It =0.3 nA; scan rate = 7.4 Hz; scan length (L) = 1.5 µm, Rrms=4.94 Å resolution 250X250 px; B) 0.13 V; 0.2 nA; 10.8 Hz; L = 1 µm, Rrms=6.14 Å; 512X512 px.

Figure 6.6: STM constant current image of Au shock-annealed sample (from the front) recorded in different areas of the surface. A) Ut = 0.12 V ; It = 0.2 nA; scan rate = 11.1 Hz; L = 1 µm, Rrms=5.52 Å; 250X250 px; B) 0.13 V ; 0.2 nA ; 10.3 Hz; L = 1 µm, Rrms=4.56 Å; 512X512 px.


73

Figure 6.7: STM constant current image of Au shock-annealed sample (from the back).
Ut = 0.13 V ; It = 0.348 nA ; scan rate = 9.8Hz; L = 0.5 µm, Rrms=1.88 Å; 512X512 px.

An increase of the epitaxial character with all the thermal annealings can be recognized from the STM images by the existence of several crystallites with sharp boundaries separated by angles of 60°, forming triangular domains. Between these 3 types of samples the flame annealed ones (Fig. 6.5) seem to exhibit an higher epitaxiality, although also samples II and III are extremely flat with existence of big crystallites.

6.1.3 Template Stripped Gold substrates

For enhancing further the quality of the gold substrate, a method to produce ultra large, atomically flat gold films glued onto Si wafers was reported by Hegner et al. [Heg93, Wag95]. This method, known as Template Stripped Gold (TSG), is based on a) deposition of Au onto mica sheets, b) supporting the free Au surface by gluing it to a Si wafer (using epoxy or ceramic glues), and c) stripping the Au film from the mica. The limitation of this procedure, for the case of epoxy glues, is the instability of the multilayer against commonly used organic solvents. Ceramic glues, that do not suffer the same type of limitation, oblige to a rather complicated procedure for the TSG preparation [Wag95]. Similarly Stamou and coworkers have recently prepared TSG by evaporating a Au film on a smooth silicon wafer and supporting the Au free surface by gluing it to a glass slide [Sta97]. A novel simple method based on a different way of supporting the Au film have been developed. These surfaces have been produced by first sublimating a thin epitaxial gold film on mica, then electroplating a thick nickel layer (~200 µm) and finally stripping the metal from the mica that in the following case is done mechanically, although it can be performed also chemically


74

(see Fig. 5.1) . [Sam99b]

Figure 6.8: Unfiltered topographical STM image of a coated Ni supported-
TSG surface (Ut=20 mV, average It= 800 pA). Height range= 11.4 Å; Rrms=2.3 Å.

Fig. 6.8 shows an STM image of the Ni supported TSG surface coated with an undecanthiol monolayer recorded on the micrometer scale in the constant current mode. The image reveals three typical topographical features of a Au surface where alkanethiols had chemisorbed, as recently reviewed by Delamarche et al. [Del96]. First, the gold surface exhibits atomically flat terraces extending over up to several hundreds of nanometers. Second, triangular areas with 60° angles, typical of epitaxial Au(111) grown in UHV, can be recognized (indicated with white arrow in Fig 6.8). Third, randomly distributed depressions, either one or two gold steps deep (2.4 Å and 4.8 Å), due to the etching of the Au surface during the alkanethiols chemisorption, are also present.

The values for Rrms=(2.2±0.2) Å and RA=(1.7±0.2) Å, measured on an area of 1µm2, are remarkably about one order of magnitude lower than those obtained on the free Au surface evaporated on mica in high-vacuum (HV) at 400°C substrate temperature (Fig. 6.4), and half or one third of the ones annealed (Fig. 6.5 - Fig. 6.7). It is also smaller than values found for TSG films prepared by evaporating Au both onto mica [Wag95] an onto a Si wafer [Sta97], most likely because the supporting procedure with glue introduces in the system more mechanical stress than the electroplating. Obviously, the atomical flatness of our TSG substrates extends at least over the micrometer scale. Ni supported TSGs produced at Tsub= 35°C during the Au evaporation exhibit a Rrms which is approximately only twice as large as


75

the one obtained at Tsub= 300°C, even though they do not exhibit atomically flat terraces extended over several hundred nanometers with typical 60° angles. The reproducibility and easiness of the mechanical pealing for all samples have been proven by the absence of an insulating tunneling barrier, which would be expected for mica leftovers [Wag95]. In addition, the intertness of our multilayer makes it possible that pealing is carried out also chemically with THF. This allows to avoid any mechanical stress that could be induced in the film during the pealing process carried out with tweezers.

It is worth to note that this alternative route to produce TSGs leads to a Au surface with an increased flatness, which is ascribed to the minimized mechanical stress introduced in the system during each step of the substrate preparation. Moreover, since the procedure described here does not require any gluing between the Au film and the support, the application of this TSG does not suffer from any limitation due the physical and chemical stability of epoxy glues. This is important both for the choice of the solvent to be used for molecular assembly on the substrate and for any chemical in-situ modification of the SAM to be carried out. This method described for TSG preparation is very simple and easily transferable to large scale production. The thick film of Ni grown on the upper Au surface can be replaced with any other thick (and preferably inexpensive) metallic layer, which is stiff, chemically inert and stable in time. The metal supported TSG may become the golden support of choice for SAMs formation and scanning probe microscopy imaging both in biology and in material science. [Sam99b]

6.1.3.1 SAMs on Template Stripped Gold substrates

Figure 6.9 shows STM images of 1-undecanthiol chemisorbed on TSG recorded in constant height mode on the nanometer scale (unfiltered and filtered). A hexagonal pattern of undecanthiol molecules was visualized at a high gap impedance, resulting from rather high voltages (~800 mV) and low tunneling currents (30-60 pA). The minimal tunneling resistance to maintain molecularly resolved imaging on undecanthiol and nonanthiol films were 27-10 G\|[OHgr ]\| and 1 G\|[OHgr ]\|, respectively. This strong dependence on the alkyl chain length is an indication for the non-destructive imaging of the SAM [Hei95]. The average spacing was (5.2±0.3) Å, consistent with the radicradic3R30° adsorbate layer on Au(111) surfaces (alpha-phase).


76

Figure 6.9: STM constant current image of undecanthiol on Ni supported-TSG surface. Ut = 800 mV; I = 40 pA. a) Unfiltered image with the 2D-Fourier Transform showing the periodicity of the hexagonal lattice (alpha-phase); b) band pass filtered image. Although a high frequency noise blurs the raw image (as evident also from the FFT), the lattice of the adlayer can be seen in both images.

6.1.4 Conductivity of SAMs of Alkenes and Alkanes

Alkanethiols are linear, flexible saturated hydrocarbons, which form insulating materials. When they are crystallized in a SAM their structure is tightly packed and stable. On the other hand alkenethiols are non-saturated oligomeric model systems for poly(acetylene), which is a stiff polymer, forming a semiconductive material. The aim of this work is to design a prototype system of a molecular wire by preparing SAMs of unsaturated alkenethiols on Ag(111) and Au (111) and to study the average conductivity of their self-assembled monolayers with STM. Alkanethiols on Ag(111) form a radic7*radic7 R10.9° adlayer with the alkyl chains oriented nearly perpendicular to the basal plane of the substrate [Hei95]. In fact the tilt angle with respect to the surface normal is 12° while on Au(111) it is 27°, thus the molecules are more tightly packed on Ag than on Au. This suggests Ag(111) as an ideal substrate for SAM preparation of alkyl chains assembled perpendicular to the substrate.

The experiments can be divided into four separate sets I to IV:

I) Three different types of alkenethiols varying the number of carbon atoms and consequently of double bonds in the main chain have been investigated:


77

C9A

C11A

C12A

Ag(111) surfaces have been immersed for 6 hours in 5 mM mixtures of unsaturated and saturated alkanethiols solutions in benzene.

Unsaturated

0 %

5 %

20 %

80 %

100 %

Saturated

100 %

95 %

80 %

20 %

0 %

With the naked eye the films with a bigger amount of unsaturated alkenethiols appear less shiny. The light reflectivity of the sample increases with the percentage of saturated alkanethiols. The poor shine of the film by naked eye indicates that the surface of the film possesses a roughness on the order of micrometers. This feature makes STM and AFM studies difficult. In fact the maximum excursion of a piezo in the Z axis is usually just a few micrometers (less than 5). Such a “macroscopic“ rearrangement of the surface can be explained only with a drastic process. An oxidation of the Ag or of the double bound is not enough to give rise to this rough morphology. It seems more likely to explain the phenomenon with a rearranging of the Ag (111) surface, as suggested by Seidel [Sei93]. Indeed it is well known that strong chemisorption forces can override the substrate-substrate bonding, which in turn can cause a reconstruction of the surface followed by a decrease of the surface atom coordination number [Ulm91].


78

II) Three different types of alkenethiols with 9 carbon atoms in the main chain have been used:

C9A

C9B

C9C

The samples produced by bathing the Ag for 1 hour either in a 1 mM or in 5 mM solution of C9A reveal a similar morphology on the micrometer scale, suggesting that the large roughness seen in the previous experiments was probably due mainly to an excessive immersion time, and to a much lesser degree to the concentration of the solution.

Based on previous experiments, it was decided to decrease the concentrations to 1 mM and to reduce the time of immersion of the Ag film in the solution to 1 hour. Comparing these experiments with the previous ones, the films appear more shiny. Indeed the surface roughness is smaller allowing these samples to be studied with STM.

Nevertheless, the film roughness existing on a micrometer scale does not permit to achieve molecular resolution imaging on either fully and partially unsaturated alkenethiol films.

III) Thiol end functionalized - C11 pentaene, both in a solution: a) with > 90% all-trans and; b) with >85% all-trans has been diluted with benzene and the following solutions have been used for chemisorption onto the metallic surfaces:

C11A


79

Table 6.1: Sample of unsaturated C11A mixed with fully saturated C11H23SH

Unsaturated

Saturated

Stock solution

Total concentration of alkanethiols

Type of Au

substrate

Sample

Name

100 %

0 %

1

0.66 mM

TSG

C11A1

95 %

5 %

1

2 mM

Sublimed

C11A2

80 %

20 %

2

2 mM

TSG

C11A3

20 %

80 %

2

2 mM

Sublimed

C11A4

0 %

100 %

 

2 mM

TSG

C11A5

Note: stock solution 1 is > 90% all-trans conformation and solution 2 > 85% all trans.

TSG surfaces coated with saturated alkanethiols (C11A5) have been imaged at ambient conditions in the constant current mode and described in paragraph 6.1.3.1. They exhibit crystallites extending over several hundreds of nanometers with steps at 60° angles typical for Au (111) surfaces as in Fig. 6.8. Also for the case of alkenethiols (C11A1) the morphology is made of small holes on a sub-micrometer scale. This feature, which is due to the etching of the Au surface (Fig. 6.10, 6.11) confirms the occurrence of the reaction of thiol functionalized molecules with gold. Noteworthy, this feature has been visualized on all the samples.

Figure 6.10: STM constant current image of TSG coated surface (Ut = 800 mV; I =30 pA) (sample C11A1). Typical holes proving the reaction of the thiols with gold. Height range= 7.2 Å; Rrms=1.4 Å.

An STM investigation of sample C11A1 on the molecular scale did not reveal any periodic lattice, although some periodical features with spacings in the range of 5 Å have been observed. After being stored for 4 days adsorbed on Au in the dark, sample C11A1 exhibits a roughness increased of almost one order of magnitude on the several nm scale (Fig. 6.12 to be compared to Fig. 6.10). This suggests a notable instability of the organic interface.


80

Figure 6.11: STM constant current image of TSG coated surface (Ut = 800 mV; I = 20 pA) (sample C11A1). Typical holes proving the reaction of the thiols with gold. Height range= 7.2 Å; Rrms=1.5 Å.

Figure 6.12: STM constant current image of C11A1 on TSG (Ut = 800 mV; I = 50 pA). Typical holes proving the reaction of the thiols with gold. Height range = 49.6 Å; Rrms=12.0 Å .

On the other hand, sample C11A3 exhibits on a 200 nm scan length micrograph a defined structure with different alignments on two different crystallites of Au (Fig. 6.13).


81

Figure 6.13: STM constant current image of C11A3 on TSG (Ut = 500 mV; I = 50 pA). Different Au crystallites with different alignments of adsorbed molecules indicated with black arrows.

On a sub-nanometer scale a periodic pattern has been recorded in Fig. 6.14.

Figure 6.14: STM image of sample C11A3 on Ni supported-TSG surface. Ut = 500 mV; I = 50 pA. a) Unfiltered image; b) band pass filtered image. Although a high frequency noise blurs the raw image, the lattice of the adlayer can be recognized in both images.

It appears clear that the chemical stability of the system, both in solution and as a thin film on Au, is decreasing with an increasing quantity of conjugated molecules. Since a well defined hexagonal lattice has been visualized completely only on fully saturated alkanethiols (sample C11E in Fig. 6.9), it is hazardous to conclude something about the different electrical resistivity measured on the different samples, i.e. to estimate the effective resistance of the different molecules investigated. Nevertheless we can note that contrasts due to single ends of alkanethiol molecules have been imaged using the following tunneling parameters:


82

Sample C11A1: 400 mV and 3.5 nA (corresponding to R= 114 M\|[OHgr ]\|);

Sample C11A3: 800 mV and 200 pA ; 500 mV and 50 pA (corresponding to R= 4-10 G\|[OHgr ]\|);

Sample C11A5: 700 mV and 65 pA ; 800 mV and 40 pA (corresponding to R= 10.7 - 20 G\|[OHgr ]\|).

This indicates that the electrical resistivity increases with the increasing quantity of fully saturated alkanethiols in the layer.

Sample C11A2 and C11A4, due to the lower quality of the Au surface, did not exhibit any extremely flat area and, because of this reason molecular resolution images have not been achieved.

IV) Thiol end-functionalized C11 pentaene have been self-assembled on Au sublimed films:

C11A

Deposition of the organic solutions onto Au discs have been performed both in Strasbourg with ultrafresh polyenes (for samples C11A10-C11A17, see table 6.2) and in Berlin with some of them stored frozen for 9 days (samples C11A18- C11A24).

In table 6.2 the 2mM solutions in benzene, are listed which have been used for the chemisorption.

As previously observed:

The gold etching confirms that alkanethiol molecules have chemisorbed on the Au surface;

The stability of the system, both in solution and in the thin film on Au, is decreasing with the increasing quantity of conjugated molecules. Moreover the roughness of the surface increases tremendously with the enhancing storage time of the organic films in a dark air environment;


83

An estimation of the average electrical resistivity of the layer calculated for images with molecular resolution is listed in table 6.3.

Table 6.2: Sample of C11A mixed with saturated ones

Sample

Unsaturated

Saturated

Bath time

C11A10

100 %

0 %

18 hours

C11A11

95 %

5 %

18 hours

C11A12

80 %

20 %

18 hours

C11A13

20 %

80 %

18 hours

C11A14

5 %

95 %

18 hours

C11A15

100 %

50 %

0 %

50 %

for 1 hour

for 17 hours

C11A16

100 %

50 %

0 %

50 %

for 6 hour

for 12 hours

C11A17

100 %

0 %

0 %

100 %

for 2 hour

for 16 hours

C11A18

100 %

0 %

24 hours

C11A19

95 %

5 %

24 hours

C11A20

80 %

20 %

24 hours

C11A21

20 %

80 %

24 hours

C11A22

5 %

95 %

24 hours

C11A23

0 %

100 %

24 hours

C11A24

100 %

0 %

0 %

100 %

for 4 hour

for 20 hours


84

Table 6.3: Sample of C11A mixed with the fully saturated analogue: electric properties

Unsaturated

Saturated

Ut

It

Mean Resistance

100%

0%

400 mV

3.5 nA

114 M\|[OHgr ]\|

80%

20 %

800 mV

200 pA

4-10 G\|[OHgr ]\|

0%

100%

700 mV

65 pA

10.7 - 20 G\|[OHgr ]\|

These results clearly indicate an increasing resistance with the increasing quantity of fully saturated alkanethiols.

In contrast with the previous experiments a periodic lattice has been visualized also on 100% unsaturated alkenethiols (image not shown), although the quality of the imaging was poorly reproducible in terms of lattice spacing. In the present case samples C11A15, C11A16, C11A17, C11A24, which were prepared according to different procedures, did not exhibit a different morphology and stability, also if compared to the related pure unsaturated samples C11A10 and C11A18.

Thus, it is concluded that the stability of the synthesized polyenes is rather poor on both Au and Ag substrates. This did not allow to achieve reproducibly molecular resolution STM imaging and quantitative evaluation of the resistance of the adsorbed monolayer, even though the results are qualitatively in line with an increased conductivity with the unsaturation of the alkyl chains.

6.2 Role of the substrate in physisorption

The requirement for a well controlled and reproducible physisorption of organic layers on solid substrates is an extreme flatness of the support, a well defined chemical composition and cleanliness of the surface to be coated.

As introduced in chapter 4, molecules tend to adsorb on the surfaces in different ways, with one parameter being the conductivity of the support.

A typical conductive substrate used for STM investigations is highly oriented pyrolitic


85

graphite (HOPG) [Bin87] (Fig. 6.15) which is a layered substrate that can be freshly prepared by cleaving its surface with an adhesive tape. An atomically flat surface on a micrometer scale is in this way made to appear. This flat interface is neutral and inert to organic solvents. Therefore it should be ideal for the self-assembly of either neutral or ionic adsorbates, since electrostatic interactions can be neglected.

Figure 6.15: HOPG crystallographic structure.

Figure 6.16: Layer structure of muscovite mica structure: A) Side view; B) Top view: the cleaved plane is the basal plane, composed of hexagonal array of oxygen ions with regular vacancies which are randomly filled with potassium. Unit cell: a= 5.2Å , b=9.0 Å.

Other typically flat layered conductive substrates used for STM studies are dichalcogenides [Wil69]. They posses various chemical compositions and structures, which makes it possible to select the ideal support for a given study. Unfortunately they are only rarely commercially available, and therefore they are not so wide spread as HOPG. Moreover they are chemically less stable.


86

On the other hand, the most used insulating support for SFM investigations is ruby mica (muscovite) which is a composite of sheets, belonging to the phyllosilicates. In this crystal a layer of octahedrally coordinated aluminum cations is sandwiched between two identical layers of linked (Si,Al)O4 tetrahedra (Fig. 3a). Two of these tetrahedrally coordinated sheets are linked by an interlayer of potassium cations. The lamellar cleavage takes place between these tetrahedral layers exposing a basal plane of oxygens (Fig. 3b) with a structural imbalance of charge and a partial potassium coverage. The charge imbalance is due to the isomorphous substitution of the cations coordinated in the tetrahedral sheet or in the octahedral layer that is located underneath. In the case that Al(III) replaces Si(IV), the overall charge gets negative by one elementary charge. This charge imbalance is neutralized in the solid state by interlayer potassium cations [Dee65]. The maximum theoretical lattice imbalance is taken as the number of surface sites per square meter NS = 2 × 1018 sites/m2 of surface, or one charge site per 46.8 Å2 [Nis94,Nis95]; this leads to a surface charge density sigma = 0.34 C/m2. According to this, a very high limiting “local“ concentration (estimated to be up to 33.5 M) of cations on the mica surface is reached [Sam96]. When mica is cleaved the plane indicated by the arrow in Fig. 3a is split, and potassium cations left on the two split surfaces only partially screen their negative charges.

6.3 Phenyleneethynylene trimers

6.3.1 Introduction

A fine tuning of the performance of molecular based electronic devices depends on the spatial arrangement of the molecules. Therefore a crucial issue is to drive the molecular self-assembly onto flat solid substrates towards highly ordered, reproducible and thermodynamically stable supramolecular structures [Leh93]. In this context oligomers are widely investigated as model compounds of their related macromolecules for their electronic properties [Mül98a]. Among conjugated species, para-phenyleneethynylene derivatives have received a reduced attention [Tou96, Gie96, Bun00]; besides their optoelectronic properties [Tad96, Wed96], they possess a remarkable stiffness and linearity along the conjugated backbone [Mor94, Wau96] which are features that can play a pivotal role in the 2D and 3D self-assembly into well-defined nanostructures. The capability of Scanning Probe


87

Microscopies to achieve true atomic resolution imaging renders them the only techniques that up to now can give evidence of structural defects at the atomic level [Bin83b, Ohn93].

6.3.2 STM on physisorbed monolayers

A monodisperse molecular system physisorbing at the interface between its almost saturated solution and an Highly Oriented Pyrolitic Graphite (HOPG) substrate can assemble in an epitaxial monolayer following the three-fold symmetry of the support [Rab91a]. The monocrystalline structure of alpha-phenylethynyl-omega-phenyl-ter[1,4-(2,5-dihexylphenylene) ethynylene)] (2) (Fig 6.17) is displayed in Fig. 6.18.

Both the aliphatic side chains and the conjugated skeletons lie flat on the (0001) plane of graphite. Since the contrast in STM imaging is mainly ruled by the energy difference between the electronic states of the substrate and the ones of the adsorbate, darker parts can be assigned to the aliphatic groups, characterized by a larger energy difference, and bright rods can be recognized as the conjugated backbones, with a smaller energy difference [Laz97].

Figure 6.17: alpha-phenylethynyl-omega-phenyl-ter[1,4-(2,5-dihexylphenylene)ethynylene)] (2).

Unlike most of the organic compounds investigated at the solid-fluid interface with STM [Rab91b], there is no evidence that in the system described here the alkyl chains are aligned along one of the crystallographic axis of the HOPG substrate.


88

Figure 6.18: STM current image of 2 in 1-phenyloctane imaged at the solid-liquid interface on HOPG (Ut=1.2 V , average It= 1.0 nA). 2-D crystal structure with its unit cell, averaged over several images amounts to a=(1.78±0.09) nm, b=(2.90±0.18) nm, alpha=(113±5)°. The distance between neighboring backbones is in this case DeltaL=(1.46±0.11) nn.

The spacing between neighboring parallel backbones, which can be attributed to the width of the molecules, amounts to only DeltaL=(1.46±0.11) nm. It is considerably smaller than the 1.9 nm calculated for the case with the alkyl chains extended. This indicates that the side-chains are disordered between adjacent parallel backbones, since an interdigitation of the hexyl groups can be excluded because of steric hindrance. This can be considered a consequence of their high mobility at room temperature. In this case important for the formation of tightly packed crystals epitaxially grown on the basal plane of the substrate seems the remarkable stiffness of the alternating aryl and ethynyl groups in the backbone.

On the other hand, a related trimer has been studied bearing thiol end-groups in the alpha and omega position which have been protected by carbamoyl functions just in order to increase the stability of the moiety in air environment at room temperature. Its chemical formula and its proper chemical name are given in Fig. 6.19. The crystal structures have been determined both by means of Scanning Tunneling Microscopy (STM) at the solid-liquid interface and by X-Ray Diffraction (XRD) on the single crystal.


89

Figure 6.19: 1,4-Bis-[[2,5-dihexyl-4-[[4-[(N,N-dimethylcarbamoyl)
thio]phenyl]ethynyl]phenyl]ethynyl]-2,5-dihexylbenzene (1).

Figure 6.20: STM constant current image of 1 in 1-phenyloctane imaged at the solid-liquid interface on HOPG. Ut = 1.2 V , It = 1.0 nA. 2-D crystal structure with its unit cell, averaged on several images: a = (1.83 ± 0.11) nm, b = (3.42 ± 0.12) nm, alpha= (108 ± 5)°. The distance between parallel backbones is DeltaL = (1.52 ± 0.08) nm. The angle between the backbones and the lamella main direction is (67 ± 2)°. Superimposed are two molecular models.

STM investigation of 1 at the solid-liquid interface gave rise to a similar result as for the derivative without the protected thiol end-groups (2): the molecules pack in a monocrystalline structure, displayed in Fig. 6.20. A slight interdigitation of the end functionalities in the 2D pattern can be seen. In this case it is likely to be induced by both to hydrogen bonding between adjacent carbamoyl groups and by the steric hindrance of the end-groups that


90

impedes a columnar packing, which was observed for mixtures of alkanes monolayers crystallized on the basal plane of HOPG [Hen92].

On a larger scale the structure is polycrystalline. Single crystallites are characterized by a well-defined molecular orientation with respect to the crystalline substrate. The high resolution imaging made it possible to record defects on the nanometer and the sub-nanometer length scale. The first type of defects are missing molecules within a single molecular crystal; an example of two missing molecules is indicated by an arrow in Fig 6.21. The second type of defects are the domain boundaries that delimitate each crystallite. At these frontiers the molecules are less well packed. This issue will be discussed further in paragraph 6.3.5.

Figure 6.21: STM constant current image of 1 in 1-phenyloctane imaged at the solid-liquid interface on HOPG. Ut = 1.2 V , It = 1.0 nA. Polycrystalline structure made of single crystallites with different molecular orientations. The arrow indicates a defect (two missing molecules) in a crystal lattice.

6.3.3 XRD on single crystals

The crystal structures of phenyleneethynylene trimers 1 and 2 has been determined here for the first time by means of XRD. The structure of the related monomer 3 (Fig. 6.22) has been also investigated for comparison.


91

Figure 6.22: Chemical formula of the monomer: 1,4-Bis[2-[4-[(N,N-dimethylcarbamoyl)thio] phenyl]ethynyl-2,5-dihexylbenzene (3).

ORTEP plots (fORtran Thermal-Ellipsoid Plot program for Crystal Structure Illustrations) [Joh70, Joh72] of the two crystal structures are shown in Fig. 6.23. Their crystal and refinement data are listed in Table 6.4.


92

Table 6.4: Crystallographic data and details of the structure refinements of alpha-[[4-[(N,N-dimethylcarbamoyl)thio]phenyl]ethynyl]-omega-[4-[(N,N-dimethylcarbamoyl)thio]phenyl]-ter[(2,5-dihexylphenylene-1,4)ethynylene)] (1) , alpha-phenylethynyl-omega-phenyl-ter[(2,5-dihexylphenylene-1,4)ethynylene)] (2) and 1,4-Bis[2-[4-[(N,N-dimethylcarbamoyl)thio] phenyl]ethynyl-2,5-dihexylbenzene (3). The structures were solved by direct methods (SIR92) and refined by full-matrix least squares analysis.

Structure

(1)

(2)

(3)

a (Å)

5.9262(4)

5.6655(6)

8.6807(5)

b (Å)

13.897(1)

11.943(2)

9.1165(5)

c (Å)

22.503(2)

23.154(5)

23.696(2)

alpha (°)

94.476(3)

84.848(6)

90

beta (°)

90.998(6)

86.150(9)

98.095(6)

gamma (°)

102.689(6)

80.216(12)

90

V (Å3)

1801.3

1535.5

1856.5

Z

2

2

4

Dx (gcm-3)

1.097

1.064

1.167

µ (cm-1)

1.133

0.552

15.256

space group

P -1

P -1

P 21/n

number of unique
reflections

7427

6061

3863

number of observed
reflections

3039

2406

2111

R a)

0.084

0.0583

0.0477

Rw a)

0.092

0.0775

0.0591

T (K)

210

210

298

lambda b)

Mo Kalpha

Mo Kalpha

Cu Kalpha

diffractometer b)

Nonius KCCD

Nonius KCCD

Nonius CAD4

Technical details:

a) R= refinement; Rw=weighted refinement

b) Diffractometer used either Nonius Kappa-CCD instrument employing graphite-monochromated Mo Kalpha radiation, or Enraf Nonius CAD-4 with graphite-monochromated Cu Kalpha radiation.


93

Figure 6.23: Single molecular structure in single crystals of a) (3), b) (2), c) (1).


94

Figure 6.24: Crystallographic structure of alpha-phenylethynyl-omega-phenyl-
ter[(2,5-dihexylphenylene-1,4)ethynylene)] (2).

In all three cases the phenyl rings are oriented approximately parallel to each other and the hexyl side chains are coplanar. While in the monomer (3) (Fig. 6.23a) the side chains assume a regular all-trans conformation, in the case of the trimers (1 and 2) they are bent (Fig. 6.23b-c). The reason for this behavior are intermolecular interactions as shown in Fig. 6.24 and Fig. 6.25: the hexyl side chains do not interdigitate but are bent towards the main chain direction in order to fill the free volume between them. The tilt angle between the main chain and the side chain amounts to 45° for (2) and 35° for (1); one reason for this difference is likely to be the changed length of the molecule: (1) is longer and therefore possesses more free volume to be filled. Consequently in the case of (1) the side chains can be more tilted than for (2). According to the smaller tilt angles also the distance between the backbones decreases from 9.5 Å to 9.0 Å for (2) and (1), respectively. An important role for this decreased spacing can


95

also be ascribed to the stronger interactions between the end-groups in the case of (1) than for (2).

Figure 6.25: Crystallographic structure of Crystallographic intermolecular structure of alpha-[[4-[(N,N-dimethylcarbamoyl)thio]phenyl]ethynyl]-omega-[4-[(N,N-dimethylcarbamoyl)thio]phenyl]-ter[(2,5-dihexylphenylene-1,4)ethynylene)] (1).

Similarly to the 2D case (STM image in Fig. 6.20) the trimer molecules pack regularly parallel to each other forming lamellar structures. Both 2 and 1 exhibit an arrangement which


96

could suggest a high mobility of the side chains at room temperature. Out of 6 hexyl side chains the molecule 2 posseses 4 that are bent and 2 which are rather straight (Fig. 6.23b), while 1 exhibits in these 2 side chains two approximately statistically occupied orientations of the last propyl function (Fig. 6.23c).

6.3.4 Discussion

The spatial organization of molecules in 2D and 3D assemblies depends on intramolecular, intermolecular as well as interfacial forces. This implies that different environments lead to distinct order on the molecular scale. Hence, it is of prime importance to compare and correlate molecular structures in single crystals and in 2D physisorbed monolayers.

The phenyleneethynylene derivatives consist of three essential parts: the conjugated main chain, the aliphatic side groups and the end functions (in alpha and omega positions). The contribution of each one in the self-assembly can be distinguished. The conjugated backbones exhibit polarizable pi electrons that give rise to strong intermolecular interactions. This is the case in the physisorbed monolayers on HOPG where, due to a rehybridization of the pi-states of the skeletons with the ones of the conductive substrate, the unsaturated main chains tend to adsorb flat on the basal plane of the support, maximizing in this way the overlap of its electronic states with the ones of the substrate, as suggested from ab-initio calculations [Bis00]. Moreover the remarkable stiffness along this backbone is also important for the packing into lamellae both in 2D and 3D.

On the other hand, the lateral chains play a primary role in conjugated oligomers and macromolecules. They are commonly attached to the backbones to enhance the solubility in organic solvents. In addition they play an important role also in the self-organization and self-assembly of the organic system in 2D and 3D architectures on flat solid substrate [Rod89]. In the present case both XRD and STM results give evidence for the remarkable mobility of the side-groups at room temperature. For the first case indeed two different structural conformations of the terminal propyl functions of the hexyl side chains can be recognized in Fig. 6.23c. In the latter one, even with our home made STM apparatus that allows very fast scanning (200 Hz/line), it was not possible to resolve the single hexyl chains. This may be attributed to the high conformational mobility of the relatively short alkyl chains. Indeed, the increasing order of solid state aggregates in other PPE derivatives with the length


97

of the side chains can in fact be interpreted in terms of decreasing dynamics of the lateral substituents [Ofe95].

Finally, the end-functions in alpha and omega position can provide a further stabilization to the molecular arrangement. This is the case of 2 where the carbamoyl end groups interdigitate with the neighboring molecules because of a hydrogen bonding type of interaction.

6.3.5 Dynamics of molecules at the solid-liquid interface

Surfaces grown under non equilibrium conditions are always prone to rearrangement [Bar95,Bisc97]: understanding the mechanisms and kinetics of such a reorganization is necessary to predict the film stability. Subsequent temporal evolution towards the equilibrium state must involve coarsening, i.e., an increase in the characteristic length-scale of the dominant structure. In this context efforts have been mainly devoted to dynamics of metallic or semiconducting thin films [Zin92, Wen96, Mor96, Car97]. On the other hand, physico-chemical phenomena on organic interfaces still need to be understood better at a molecular level [Ulm91], revealing the role of intermolecular interactions in their ensemble. In particular little is known on the processes occurring at the liquid-solid interface [Sta95a].

An investigation on a true molecular scale of the coarsening within an organic monolayer can be carried out by means of STM. It allows to characterize the motion of single molecules at the domain boundary between molecular crystals self-assembled at a surface and to discern the thermodynamic and kinetic contributions to the total energies governing this process. Fig. 6.26 shows the Ostwald ripening process of a phenyleneethynylene trimer monolayer of 1 at the solid-liquid interface. The dynamics of the molecular crystallites on a 22 minutes time scale has been recorded. Small domains tend to shrink and disappear while bigger crystals enlarge.


98

Figure 6.26: STM constant current images. Evolution of the 2-D polycrystal structure of 1 at the solid-liquid interface during 22 minutes: Ostwald ripening phenomena brought about by the reorientation of single molecules in island I, II, III, IV. The white scale bar is 20 nm. Arrow in a) indicates a domain boundary of the type shown in Fig. 6.27a.


99

Observing carefully the first micrograph in Fig 6.26 one can notice that the different domain boundaries in the polycrystalline structure can be divided into two different classes. The first presents equal orientations of the main molecular axes in the two neighboring islands that differ only by a slight translation of the crystal structures relative to each other (scheme in Fig. 6.27a and indication by the arrow in Fig 6.26a). The second type of domain boundaries is made of molecules oriented differently (Fig. 6.27b); several of these types are visible in Fig. 6.26a. The energies associated with the transitions leading to a large single crystal in these two cases can be discussed for the thermodynamics (energetic gain) or the kinetics (activation energy).

Figure 6.27: Scheme of different types of molecular packing in neighbor domains.
Single molecules are represented as a rod. a) translated; b) rotated.

Figure 6.28: Scheme of the molecular packing at the crystallite frontiers.

In the first case (Fig. 6.27a), where the reaction consists of a simple translation, is characterized by a smaller activation energy but the thermodynamic gain is also small. In the latter case (Fig. 6.27b), the system requires to undergo a rotational motion that needs a larger activation energy to occur, albeit the thermodynamic energy gain is bigger since the new state that is achieved is energetic favorable.

Focusing the attention on the second class of domain boundaries because they are more recurring: at room temperature and atmospheric pressure on the minutes time scale individual


100

molecular rods can change their tilt angle with respect to the underlying substrate without a transition to the supernatant solution. This phenomena, known as Ostwald ripening, is a 2D reaction controlled process that takes place at the boundary between different crystals; it is governed by a minimization of the interfacial energy (or line energy in 2D) [Sta95a]. A strong influence of the STM tip in this process has been ruled out [Sta95a] and confirmed in the present case. In fact simply by stopping the scanning for approximately half hour and restarting the imaging in the same area of the sample revealed that the disappearance of small domains took place irrespective of the scanning of the tip. In addition an evaluation of the kinetics of the process made it possible to reject the hypothesis that this coarsening is governed by diffusion [Sta95a]. The tight molecular structure allowed to monitor this reaction with an increased resolution by following the motion of individual rods in real time.

From the tightness of the molecular packing at the domain frontier, the stability of a front or a predominance of one domain with respect to the other may be predicted. A careful observation of the images in Fig. 6.26 allows to classify domain boundaries according to three types (scheme in Fig. 6.28). The most stable interface, characterized by a remarkable tightness, is made up by two domains packed along one head to head direction oriented parallel to the frontier (case A, Fig. 6.28a). A less stable boundary (B) consists of a stable domain that is tightly packed to the front and another one that is more loosely packed (Fig. 6.28b). The area between the two crystals that does not exhibit crystallized molecules is characterized by a remarkable dynamics of the molecules: in these zones the trimers are extremely mobile (not immobilized at surface) and there may be also inclusions of the molecules of the solvent. The more loosely packed crystal, after some time, it will reduce its area. A further extension of the limit (free volume) (C), which reveals higher dynamics of the molecules in this region, gives rise to an even more unstable interface (Fig. 6.28c).

Using this approach it is possible to classify domains in Fig. 6.26. IV is surrounded by three lines of type B, and one of C. These conditions are characteristic of an unstable region and in fact it dissolves very rapidly. II from Fig. 6.26 a to b decreases its area loosing 4 molecules, but acquiring a very stable conformation made by two lateral A type frontiers and two C type. The stability of its contour allowed II to have a relatively long life, albeit the really small size of the crystal. I appears very unstable because limited only by one A front, and three C ones; indeed it disappears in 9 minutes. The more enduring one turned out to be III that is larger than the others and surrounded by four B facets. It expires in 22 minutes.

These results suggest that the stability of the domain boundary is related to two factors:


    101

  1. geometry of the molecular packing at the domain boundary;
  2. size of the crystallite.

Figure 6.29: Evolution of domain areas with time. Filled symbols correspond
to STM images in Fig. 6.26. Open symbols: images not shown.

In addition, an analysis of the kinetics of this process (Fig. 6.29) revealed that first an apparent metastable state exists, where the area of the domains is nearly constant within fluctuations of some nm2; in this apéritiv the domain borders get sharper, along the two head to head directions of the molecular pattern (axes a and b of the unit cell), at the expense of single molecules that were exceeding the frontier, creating a polygonal shape domain. This plateau can last several minutes (for the less stable domains as shown schematically in Fig 6.28b,c) or for hours (for the more stable crystals, packed as shown in Fig. 6.28a). After this, a slight extension of the domain frontier takes place, which is accompanied by a considerable increase of the dynamics of the molecules. This enables the molecular reorientation reaction to occur emerging as an abrupt decrease of the crystallite sizes and a collapse of the domains in up to 2-4 minutes. This coarsening reminds of a phagocytation process carried out by big crystallites at the expense of small ones. The molecular motion terminates in the thermodynamic equilibrium regime where only large crystallites extending over the whole substrate crystallites, i.e. over several thousands of nm2, exists. The velocity of this molecular coarsening, in terms of the disappearance of molecular domains, averaging on several individual crystals, amounts to (160±66) nm2/min. This rate is 1-2 orders of magnitude slower than for 2-hexadecyl-anthraquinone and tetradodecyl-octathiophene


102

[Sta95a]; the difference is attributed to the ability of 1 to form a more closed packed crystal than the other two compounds. Further support of this explanation is the even slower rate detected for didodecylbenezene [Rab91b] that indeed forms a tightly packed crystalline structure on HOPG.

A secondary role in the STM analysis at the solid-liquid interface, that must not be neglected, is played by the concentration of the supernatant solution. A large activation energy barrier, i.e. the nucleation energy, has to be overcome for diluted solutions to form a crystalline monolayer at surfaces. Moreover, a higher number of defects like missing molecules inside a crystal or a broadening of the domain boundaries have been observed in this case. This is therefore likely to affect the lifetime of the metastable state in the Ostwald ripening process. Another important parameter in these kind of investigations is the density of the solution (that is strongly influenced by the density of the solvent) which if large, can slow down the molecular motion both in the solution and at the interface with the solid substrate.

The comprehensive live view of molecular coarsening at surfaces with a sub-molecular resolution have shown that the Ostwald ripening occurring in monolayers at the solid-liquid interface is a phenomenon mainly ruled by the interplay of intermolecular and interfacial forces on the basal plane of the substrate. The experimental data can be explained in terms of thermodynamics and kinetics of the molecular coarsening. After a metastable regime that have a life-time proportional to the tightness of the molecular structure on the surface, the phenomenon is driven by a minimization of the surface free energy.


103

6.4 Visualization of single macromolecules in monolayers

The aim of the present study is visualize for the first time a synthetic polydisperse molecular system self-assembled in monolayer at the solid-liquid interface, to gain insights into the conformation of the adsorbed molecule and to monitor with a sub-molecular resolution the macromolecular fractionation at the interface.

Figure 6.30: Chemical formula of (4): alpha-[[4-[(N,N-dimethylcarbamoyl)thio]phenyl] ethynyl]-omega-[4-[(N,N-dimethylcarbamoyl)thio]phenyl]-poly[(2,5-dihexylphenylene-1,4) ethynylene)].

The STM technique has been employed at the HOPG-solution interface of poly(para-phenyleneethynylene) derivatives with various contour lengths along the main chain.

In Fig. 6.31 the first reported molecularly resolved STM images of a synthetic polydisperse polymer at the solid-liquid interface is shown. The self-assembly occurred from an almost saturated solution of PPE (Fig. 6.30) at the interface with a HOPG substrate [Rab91a]. The monolayer exhibits a two dimensional nematic-like molecular order. Both the conjugated skeletons and the hexyl side groups are lying flat on the (0001) plane of the HOPG. The conjugated backbones appear brighter than the aliphatic chains because of a stronger current as previously noted for the case of the trimers (see paragraph 6.3.2). The spacing between neighboring parallel backbones, which can be attributed to the width of the molecules, amounts to DeltaL=(1.62±0.10) nm. This is considerably smaller than the 1.9 nm calculated for the case with the alkyl chains extended but it fits well with what has been detected for shorter oligomers, where the side-chains have been found to be disordered between adjacent parallel backbones [Sam99a].


104

Figure 6.31: STM constant current image of a PPE 4 at the interface between the basal plane of graphite and an organic solution in 1-phenyloctane. Ut=1.2 V , average It= 1.0 nA. This PPE, according to 1H-NMR analysis on the end-groups, possesses an average contour length 7.9 nm The molecular backbones exhibit a nematic orientation along preferred directions, as evidenced by the hexagonal pattern in the 2D-Fourier Transform. The distance between the backbones averaged over several images amounts to DeltaL=(1.62±0.10) nm.

The 2D molecular arrangement is composed of domains with specific molecular orientations. The conjugated skeletons are aligned along preferred directions, according to the three-fold symmetry of the HOPG lattice, as visualized by the hexagonal spots in the 2D Fourier Transform (Fig. 6.31).

The stiffness of the molecular rods [Mor94, Wau96] and a low polydispersity allowed the molecules to pack in this 2D structure and, therefore, both played a key role for achieving this true molecular resolution imaging of the synthetic macromolecule. Also a finite mobility of the nanorods has been detected, revealing that molecules at the domain boundaries change their tilt angle with respect to the underlying substrate. This molecular dynamics occurs on the time scale of several minutes and can be attributed to a Ostwald ripening process.

Polymeric systems are ensembles of molecules with different sizes, that can be expressed in terms of molecular weight or for linear moieties also as molecular contour lengths. This distribution of molecular lengths (polydispersity) hinders the formation of a monocrystal structure in the solid phase. The formation of 2D and 3D crystals of this kind of systems occurs via self-segregation into areas with molecules of similar dimensions [Wan93]. This molecular domains posses a high degree of order, although their extension is small. The phase segregation process has been observed on a large (micrometer) scale via a different contrast in electron microscopy micrographs [Wan93]. In the present case in fact the


105

difference in the 2D structure of the monodisperse oligomer (see paragraph 6.3.2) and the polydisperse polymer, which on the average is almost three times as long as the trimer, can be is due to the different distribution of molecular lengths, which, for the case of the polydisperse system, does not enable the moieties to assemble into perfect crystals.

The time required for achieving molecular resolution imaging is much longer for the polymer. This is likely to be due to the self-segregation phenomenon and consecutive adsorption that takes place at the molecular level, allowing molecules with different lengths to pack locally well on the substrate.

The true sub-molecular resolution achieved on a synthetic polymer as the one shown in fig. 6.31 opens the possibility to measure for the first time from STM images the true length of the single macromolecules that are self-assembled on the (0001) plane of the HOPG substrate. This allows to study the phase segregation governing the self-assembly of a solid state phase of a polydisperse system. An example of this phenomenon that has been detected on a larger scale with Scanning Force Microscopy will be described later in this chapter in the case of the self-assembly of PPE into dry ribbon-like architectures.

6.4.1 Macromolecular fractionation

Other PPEs with an average contour length of 11.2 nm and 20.3 nm , respectively, (Table 6.5) have been investigated at the solid-liquid interface with STM. Their structures on the nanometer scale are shown in Fig. 6.32 and Fig. 6.33, respectively.

Table 6.5: Samples of PPEs investigated 4 with STM

No. of repeating units

9

14

35

contour length of molecule

7.9 nm

11.2 nm

20.3 nm

U = Mw/Mn

1.92

1.84

10.1

end-groups) and U=Mw/Mn from Gel Permeation Chromatography (PPP calibrated)

Comparing the three cases listed in Table 6.5, upon self-assembly of the macromolecule in 2D


106

monolayers at the solid-liquid interface, it is possible to observe using STM the following features with increasing polymer length (from Fig. 6.31 - 6.33):

  1. the molecules are more mobile. In particular in Fig. 6.32 and Fig. 6.33 the monolayer consists of areas with a stable and tight molecular packing alternated with others possessing a very high molecular dynamics; the structures of these latter ones can not be resolved with a molecular resolution imaging;
  2. the average length of the molecules immobilized on the (0001) plane of HOPG increases;
  3. the achievement of molecular resolution requires a longer waiting time indicating that the activation energy required for the self-segregation to take place increases;
  4. the tunneling parameters that have been used for resolving the molecular structure suggest that with increasing polymer length the molecules tend not to pack flat on the basal plane of the substrate. This feature is likely to be due to the to the self-segregation that hinders the formation of a tightly packed monolayer. In addition the longer molecules possess a higher flexibility, and consequently the system need to loose more entropy for the structural rearrangement, because of the large number of possible states (conformation) which the system can attain.

Figure 6.32: STM current images of a PPE 4 at the interface between the basal plane of graphite and an organic solution in 1-phenyloctane. This PPE, according to 1H-NMR analysis on the end-groups, possesses an average contour length of 11.2 nm. The molecular backbones exhibit a nematic orientation along preferred directions, as evidenced by the hexagonal pattern in the 2D-Fourier Transform. a) Ut=1.7 V , average It= 0.2 nA; b) 1.35 V, average It=0.7 nA.

The macromolecules with an average length of 11.2 nm still organize in a nematic like


107

structure, similar to the one observed for the 7.9 nm moiety with single backbones that are aligned along preferential orientations according to the symmetry of the substrate underneath. This degree of order is lost for the 20.3 nm long polymer.

The histograms in fig. 6.34 describe the distributions of rod lengths obtained from raw data STM current images. Because of the rather high stiffnes of the molecules and their packing that is not characterized by a columnar head to tail alignment, their contour length can be evaluated from STM images. This determination has been executed with an electronic ruler (using the software package Image Tools 1.27 produced by University of Texas - Health Science Center in Saint Antonio). The distributions of the lengths for several hundreds of single rods are plotted together with the Schulz-Zimm distributions, which theoretically describes the molecular weight distribution for a PPE synthesized via a polycondensation route.

Figure 6.33: STM current images of a PPE 4 at the interface between the basal plane of graphite and an organic solution in 1-phenyloctane. This PPE, according to 1H-NMR analysis on the end-groups, possesses an average contour length 20.3 nm long. Ut=1.4 V , average It= 0.55 nA. Besides the molecularly resolved areas there are zones with high molecular dynamics where the molecules are not immobilized at surface.

The curves that refer to the 7.9 nm long PPE (Fig. 6.34a) and to the 11.2 nm long ones (Fig. 6.34b) exhibit three important characteristics:

  1. the peak of the distributions moves to higher rod lengths with the increasing polymer length;
  2. the peak of the rod lengths is shifted with respect to the mole fraction - Schulz-Zimm plots;

    108

  3. the experimental distributions are remarkably narrower than the Schulz-Zimm plots.

In the case of the 20.3 nm long PPE the molecules packed on the HOPG substrate are too few to allow a quantitative investigation and detect a trend on the length of the molecules immobilized at the interface.

Figure 6.34: Histograms of the distribution of lengths of physisorbed rods. The PPEs posses an average contour length according to 1H-NMR analysis on the end groups, a polydispersity determined by GPC using poly(para-phenylene)s for calibration and a number of physisorbed rods as measured: (a) 7.95 nm / 1.92 / 593; (b) 11.2 nm / 1.84 / 253. The mole fraction of the Schulz-Zimm distribution is plotted in solid lines and in (a) Monte Carlo simulation (dashed and dotted lines). Dashed line: ratio of the potential of the interactions molecule-substrate and molecule-molecule = 2; dotted line: interaction molecule-molecule is reduced to zero. The normalization of the curves have been carried out on the tail of the distribution. Simulations have been performed using a repulsive hard core potential and London attractive forces. The model system was designed forming linear rods by overlapping several spheres linearly. The procedure was computed for a distribution of rods according to the Schulz-Zimm distribution. Calculations have been executed varying systematically the ratio between the potential of the interactions molecules-substrate and molecules-molecules in a range that spans from 2 to infinite.

The conformation of the 2,5 dihexyl-para-phenyleneethynylene is well-known to be rather straight. Molecular Dynamics (MD) [Bin95] simulations in vacuum revealed (Fig. 6.35) that upon increasing the contour length of a single PPE molecule the end-to-end distance initially increases linearly up to a length of about 20 r.u. (~ 13 nm), while above this length the bending of the chains increases more for bigger molecular lengths and leads to coiling at more


109

than 35 r.u (~ 22.75 nm) (end-to-end distances for coiled regime not given in Fig. 6.35 because a meaningful statistics would require too long simulation times). This suggests that the polymers investigated here (on average 7.95 nm and 11.2 nm long) are well described by elastic rods.

In order to gain a deeper understanding of the physisorption at the solid-liquid interface Monte Carlo simulations [Bin95, Pal98] have been carried out for the adsorption of rigid rods at a surface. An ensemble of linear rods with a distribution of lengths according to the Schulz-Zimm mole fraction distribution was constructed. Interactions between molecule-molecule and molecule-substrate were simulated making use of a repulsive hard core potential and London attractive forces.

Figure 6.35: Molecular Dynamics simulation of the evolution of the end-to-end distance of a single PPE in vacuum as a function of the increasing polymer length (expressed either in # of repeat units or contour length). The simulations have been computed at room temperature in vacuum applying the pcff force field (Molecular modelling package DISCOVER VERSION 4.0.0, Biosym Technologies Inc., San Diego, CA.) for a time scale of 100 ps for polymer up to 28 r.u., 150 ps for the molecule made of 35 and 42 r.u., and 250 ps for the polymer composed of 49 and 56 r.u.; these turned out to be enough for achieving a constant regime of energy fluctuations. The error bars represents the standard deviation of the end-to-end distance in the final 100 ps calculated.

Using a ratio between the potential of the interactions molecule-substrate and molecule-molecule of 2 (dashed curve in Fig. 6.34a) the distribution is somewhat broader and shifted to even higher rod lengths, approaching the Schulz-Zimm distribution for long chains. Upon increasing the ratio of the potentials the onset of adsorption is shifted to higher molecular lengths, as evidenced by the dotted curve in Fig. 6.34a, for which the potential of interactions molecule-molecule is zero. This fractionation is also accompanied by a notable shift of the peak of the distribution of the adsorbed molecular rod length up to 10 nm (for the 7.9 nm long


110

the polymer).

Figure 6.34a reveals that the Monte Carlo simulation overestimates the adsorption of the very long rods, which needs to be explained. The favored adsorption of long vs. short molecules at the interface can be understood since upon immobilization at the interface they loose translational entropy per particle but they gain potential energy per unit mass. This is in agreement with chromatographic analysis [Fle93] and MD calculations [Xia93]. On the other hand, with the definition of a Kuhn segment (l) [Gro97] one may determine a persistence length (l/2=RL2/2L). If one considers the RL (average end-to-end distance) given by the MD calculations, one obtains l/2=6.85 nm for a PPE chain of 21 r.u., which is close to the experimental case in our experiment. For longer chains the entropic contribution per unit mass to the overall free energy increases, due to the reduced configurational space of the molecule at the interface. This difference between elastic and fully rigid rods can explain the discrepancy between the experimental and the Monte Carlo results (Fig. 6.34a).

The physisorption of molecules at the solid-liquid interface is characterized by a continuous exchange of the adsorbed molecules with those in the solution. The system evolves towards this equilibrium regime with a rate that depends on several factors including the viscosity of the solution, the energy of adsorption of the molecules on the substrate and the concentration of the solution.

The self-assembly into the monolayer is likely to occur via:

  1. stacking of PPE in solution into clusters composed by molecules with similar size;
  2. adsorption of these clusters onto graphite along preferential directions induced by the symmetry of the substrate;
  3. filling of the missing free space with short molecules;
  4. Ostwald ripening at the interface and substitution of short adsorbed molecules with longer ones when sterically permitted.

The segregation phenomenon observed and discussed here on a sub-molecular scale is in good agreement with the one detected on larger scales on dry films of poly(diacetylenes) [Wan93,Hug97]. Since classical methods such as column separation [Rod96], ultracentrifuge [Flo53] and precipitation [Kot67] for pi-conjugated polymers are well-known to suffer from uncontrolled cluster formation in solution [Cot96, Hal98] the use of self-assembly at surfaces might be considered an alternative route for achieving fractionation of a macromolecular solution.


111

In summary, the behavior of rigid (macro)molecules in a physisorbed monolayer has been elucidated on a conductive crystalline substrate. It revealed a macromolecular fractionation at the solid-liquid interface. This phenomenon is governed by the interplay of entropic and enthalpic contributions to the free energy. It indicates that the self-assembly of a polymer on an atomically flat substrate could be a new route for fractionating a polydisperse macromolecular solution. The underlying mechanism may also be operating at solid-liquid interfaces with small inorganic, organic or biological particles exhibiting atomically flat surfaces.

6.5 PPE on insulating substrates

6.5.1 Introduction

In the following paragraph the growth of dry films of PPE from solution will be discussed on different insulating substrates varying systematically several factors:

  1. the concentration of the solution [Sam98a];
  2. the molecular weight of the PPE [Sam99a];
  3. the type of substrate [Sam99c];
  4. the rate of the deposition process [Sam98b];
  5. the side chains [Sch99].

The morphologies of these dry films have been investigated by means of Scanning Force Microscopy in Tapping Mode in air environment at room temperature, exploring a range of scan lengths from 13 µm down to 0.3 µm.

6.5.2 Morphology at different concentrations of the solution

Poly(2,5-dihexyl-1,4-phenyleneethynylene) (PPE), because of its alkyl side chains attached to the aromatic rings, exhibits a good solubility in organic solvents like methanol (MeOH) and tetrahydrofurane (THF). This two solvents have been chosen since they posses a similar boiling point of ca. 65 °C and therefore are likely to evaporate with a similar rate. In this first experiment, therefore, the role of the solvent is not taken into account and will be discussed in paragraph 6.5.5.


112

The self-assembly of alpha-[[4-[(N,N-dimethylcarbamoyl)thio]phenyl]ethynyl]-omega-[4-[(N,N-dimethylcarbamoyl)thio]phenyl]-poly[(2,5-dihexylphenylene-1,4)ethynylene)] (4) (Fig. 6.30) have been investigated on freshly cleaved muscovite mica as a function of the concentration of the cast solution. This PPE possesses an average contour length along the conjugated backbone of 16.4 nm, according to 1H-NMR analysis on the end-groups. SFM images are displayed in Fig. 6.36.

Figure 6.36: TM-SFM images representing the evolution of the morphology of the PPE on mica as a function of the concentration of the polymer. a) 2.0 g/l in MeOH, amplitude image, height range (peak-valley) (h) in the correspondent height image of ~ 70nm; b) 0.14 g/l in THF, height image, h=50 nm; c) 0.07 g/l in THF, height image, h=15 nm.

The structure of a film cast from a concentrated (2.0 g/l) PPE solution in methanol is made up of both areas with grains and areas with more elongated tightly packed architectures that look like ribbons (Fig. 6.36a). Both the diameter of the grains and the width of the ribbons are ca. 25 nm. The mica support is in this case homogeneously coated by a rather thick film. Indeed the height range in the image is ~ 70 nm, which indicates a considerable roughness. The thickness of the film is expected to be some hundreds of nm. At a lower concentration of 0.14 g/l in THF the polymer self-assembles into ribbons with random lengths but with a well defined width and thickness (Fig. 6.36b). Further dilution in THF, down to 0.07 g/l, gives rise to single ribbons lying on the flat mica substrate (Fig. 6.36c).

In both of the last two cases the substrate is only partially covered with the macromolecular moieties. In fact the height ranges are more reduced. In the second case (Fig 6.36b) a medium concentration gives rise to a spider web morphology made of knotty cores where the ribbons are entangled, while the branches are composed of singular ribbons.

It should be pointed out that the same morphologies (grains at high concentrations and


113

ribbons at low concentrations) have been found both in MeOH and in THF. Moreover the ribbons have shown a particularly high stability: upon scanning with SFM in contact mode applying a force of some hundreds of nN on a single assembly it was not possible to deform its shape.

The width of the ribbons is constant for some straight ribbons sections. It is (36±11) nm evaluated on 334 different ribbons from images with 512*512 pixels and a scanlength L le 3µm. The apparent width has to be corrected for the well-known lateral broadening effect in SFM images due to the tip shape [Kel91,But92, Bus97].

Figure 6.37: Scanning Electron Microscope (SEM) micrograph (courtesy of Dr. Rogaschewski) of commercial Si tip (Digital Instruments). Typical shape of the cantilever with a conical tip attached at its end.

The TM-SFM silicon nano-probe is composed of a stiff cantilever that appears as a parallelepiped diving board with a length of 125 µm and a width of 30 µm, as shown in Fig. 6.37. At its edge a conical tip is attached. These probes are commercially available from Digital Instruments, Santa Barbara, CA. The shape of the tip apex may be approximated by a sphere [But92]. The Scanning Electron Microscopy (SEM) micrograph allowed to recognize its shape on the hundred nanometers scale while, however, the tip apex could not be resolved properly.

Figure 6.38: Transmission Electron Microscope (TEM) micrograph (courtesy of Dr. C. Böttcher) of commercial Si tips (Digital Instruments). Tips displayed in a) and b) present a different shape. The average of the terminal tip radius of several tips has been evaluated as R = (13±7) nm.


114

From high resolution TEM imaging on several Si tips, the tip edge and consequently the terminal radius of the spherical end of the probes (Fig. 6.38) could be determined as R=(13±7) nm.

A simple geometrical model (Fig. 6.39) can be used to calculate the broadening induced by the tip size of the ribbon widths from singular profiles on SFM topographical images. Using this model, computed for a spherical tip and a rectangular cross-section of the ribbon, one obtains for the broadening:

2Delta =2 (6.3)

For the present ribbon height, determined as (2.9±0.6) nm from 285 different ribbons profiles, and a mean terminal tip radius of (13±7) nm, 2Delta amounts to (16±6) nm. The effective ribbon width can thus be obtained by subtracting 2Delta from the apparent width. It is comparable to the 16.4 nm average contour length of the molecules as obtained from 1H-NMR analysis on the end-groups. Moreover the mean height of the ribbon of (2.9±0.6) nm is on the order of once or twice the molecular width considering the case where the alkyl side chains are extended, i.e. 1.9 nm.


115

6.5.3 Morphology at different molecular weight

The role of the molecular weight distribution for the self-assembly of PPE in dry films on mica has been also explored, in this case using two different derivatives of PPE, bearing different end-groups in the alpha and omega position. The first one (4) has been already introduced in Fig. 6.30. The latter one, namely alpha-iodo-omega-[(2,5-dihexyl-4-ethynyl)phenyl]-poly[(2,5-dihexylphenylene-1,4)ethynylene)] (5), is shown in Fig. 6.40.

Figure 6.40: Chemical formula of 5, namely alpha-iodo-omega -[(2,5-dihexyl-4-ethynyl)phenyl]-poly[(2,5-dihexylphenylene-1,4)ethynylene)].

First of all the average degrees of polymerization, DP, namely the average number of repeat units, of the different PPEs obtained with a polycondensation synthesis were determined, also in this case, by 1H-NMR end-group analysis and gel permeation chromatography (GPC). All data are reported in Table 6.6. 1H-NMR spectroscopy proved complete end-functionalization of the PPEs within the experimental error. It can therefore be used to evaluate DP, and consequently the number-average molecular weight, Mn, by integrating the relative signals of the end-groups and those of the main chain. The 1H-NMR spectrum of 5 shows proton signals of the terminal repeating unit carrying the iodine atom at delta = 7.69 and 7.30. Polymer 4 exhibits a signal of the methyl groups in the dimethylthiocarbamoyl function at delta = 3.02 and the aromatic protons at delta = 7.48. The main chain-signal of the aromatic protons for both 5 and 4 appears at delta = 7.36. Nominal contour lengths of the molecules have been calculated from the determined DP using literature values for the repeat unit length. GPC was used to estimate the mass-average molecular weight (Mw), Mn , DP and the polydispersity of the polymers 5 and 4. However, this method is very sensitive to the calibration standards. Since poly(styrene) (PS) exhibits a different structure and stiffness from PPE, also a rigid-rod macromolecule, namely a substituted poly(para-phenylene) (PPP) with a known molecular weight [Van96], have been used as a calibration standard.


116

Table 6.6: PPE samples investigated: average number of repeating units and contour lengths according to 1H-NMR results; polydispersity ( ) from GPC measurements with PPP calibration.

Sample

A

B

C

D

E

F

Average no.

of repeating units

9

11

20

22

28

42

Type of PPE

4

4

5

4

4

4

Average contour

Length of molecule

7.9 nm

9.2 nm

13.5 nm

16.4 nm

20.3 nm

29.4 nm

1.92

3.68

3.82

2.65

2.59

8.23

Morphology at low concentration

Ribbons

Ribbons

ribbons

ribbons

grains

grains

Type of PPE: (?) alpha-[[4-[(N,N-dimethylcarbamoyl)thio]phenyl]ethynyl]-omega-[4-[(N,N-dimethylcarbamoyl)thio]phenyl]-poly[(2,5-dihexylphenylene-1,4)ethynylene)]

(?) alpha-iodo-omega -[(2,5-dihexyl-4-ethynyl)phenyl]-poly[(2,5-dihexylphenylene-1,4)ethynylene)]

PPE solutions have been cast onto freshly cleaved mica surfaces. After complete evaporation of the solvent (in this case THF or a mixture of THF/1-phenyloctane have been used), the dry samples revealed a dependence of the morphology on the molecular weight (Fig. 6.41 and Table 6.6). For polymers A through D with an average of 9, 11, 20 or 22 repeating units and a concentration of the applied solution lower than 0.15 g/l, ribbons were observed (Fig. 6.41a). Higher molecular weight (samples E, F) led to a grainy morphology (Fig. 6.41c). Figs. 6.41a,b exhibit ribbon heights on mica of h=(2.9±0.7) nm, as determined from 422 ribbons on different samples.


117

From the comparison of this value with the spacing between neighboring backbones evaluated from molecularly resolved STM images (Fig. 6.31), it is suggested that the ribbons are typically two monolayers thick with their alkyl chains oriented perpendicular to the substrate (Fig. 6.42). Also some sections with single, triple and even higher multi-layers occur. Moreover the widths of the ribbons are constant for some straight ribbon sections, but not for the whole sample.

In Fig. 6.43 the apparent widths (number counting) determined from SFM images of samples A-D are reported. They need to be corrected for the broadening effect due to a finite radius of the tip. Using the broadening model previously described in Fig. 6.39 , for the present ribbon height of h=(2.9±0.7) nm and a terminal tip radius of the commercial Si tips of R=(13±7) nm, 2Delta amounts to (16±7) nm. The true ribbon width, with a 7 nm error bar, is then obtained by subtracting 2Delta from the apparent width.


118

Figure 6.42: Schematic representation of molecular ribbons adsorbed on the mica surface. a) Ribbons are made of several rods packed parallel to each other. b) Each rod is typically made of two PPE molecules packed with the hexyl lateral chains perpendicular to the basal plane of the substrate.

In Fig. 6.43 it is shown that the peaks of the width distributions shift to higher values with increasing polymer length. Since the absolute value of the width is on the order of the length of a single molecule, it is concluded that the extended molecules pack parallel to each other with their long molecular axis perpendicular to the long ribbon axis, as represented in Fig. 6.42. The distribution of the ribbon widths is attributed to the distribution of molecular weights, implying that molecules with similar molecular weights phase segregate into ribbon sections with homogeneous widths. This segregation phenomenon, that governs the ribbons formation, is obviously favored by a small DP (Fig. 6.41) and is likely to be also favored by a low macromolecular polydispersity.


119

Figure 6.43: Histograms of the distribution of ribbon widths for PPE batches A through D with increasing length. Average contour length of the molecule according to 1H-NMR results (Table 6.6) and number of ribbons measured: (A) 7.9 nm / 846; (B) 9.2 nm / 160; (C) 13.5 nm / 264;(D) 16.4 nm / 334. The Schulz-Zimm distribution is plotted in terms of number counting (solid lines) and weight function (dotted line). 2Delta=(16±7) nm is the effective broadening of the tip. Dashed lines are experimental GPC data obtained from PS calibrated measurements.

Attempts to determine the molecular structure directly by electron diffraction were not successful, probably due to the small amount of material in a given ribbon. However, in continuous films of polydiacetylene [Wan93, Hug97] and poly(para-phenyleneethynylene) derivatives [Ofe95] similar molecular architectures have been observed.


120

The polycondensation reaction used for the PPE synthesis gives a molecular weight distribution, which is theoretically described by the Schulz-Zimm distribution [Ofe95]. This function was computed for samples A-D in terms of molecular contour lengths by assuming the number average degree of polymerization DP, according to the number of repeating units estimated by 1H-NMR spectroscopy (Table 6.6).

The mole fraction distribution is expressed as (see 3.7) and the weight fraction distribution as (see 3.8). In the present case , where k is the degree of coupling (in this case k =2), r is the number of monomers (independent parameter) and gamma(k) is the gamma function. These distributions were plotted on the histograms of Fig. 6.43, after being shifted on the x-axis by the tip broadening effect of 16 nm. This made it possible to relate the estimated molecular lengths to the ribbon widths. Noteworthy, a good fitting between the mole fraction distribution and the distribution of ribbon widths can be recognized.

This match also demonstrates that SFM on these nanostructures may provide a reasonable evaluation of the molecular weight distributions for a rigid rod polymer. Nota bene, due to the commonly observed aggregation of the polymer chains it is difficult to determine correct molecular weight distributions for rigid-rod polymers with standard polymer analytical techniques like, e.g., Gel Permeation Chromatography (GPC). Dashed curves in Fig. 6.43 represent the GPC experimental data (PS calibrated) obtained on the respective polymer. A long tailing of the molecular weight distribution curve to high values is observed in the elugrams of the polymers which drastically increases the polydispersity but has only little effect on Mn. A reasonable agreement between the theoretical Schulz-Zimm plot and the experimental GPC curve has been observed only for the shortest PPE (Fig. 6.43a). In fact with increasing chain length the aggregation of the molecules becomes more intense (Fig. 6.43b-d). Moreover, the values are affected by adsorption of the stiff molecules to the column. Additionally, in the present case, the protected thiol end-groups enhance this phenomenon [Fra98].

A comparison between the different methods used to determine the molecular weight distributions (1H-NMR, GPC - PS and PPP calibrated) has shown that Mn evaluated with GPC (PPP calibrated) are overestimated on average by 42% with respect to 1H-NMR data. For the case of GPC (PS calibrated) the overestimation is even bigger. The polydispersity (Mw/Mn)


121

measured with GPC is 50% larger for the case of PS calibration if compared to PPP calibration, while Mw is twice as big. This confirms that for investigating DP and Mn of a stiff polymer 1H-NMR analysis on the end groups is a suitable technique, while for approximating Mw and polydispersity by GPC at least calibration with a rigid-rod polymeric standard like PPP should be used. [Sam99a]

6.5.4 Morphology on different substrates

In order to understand the phenomena governing the growth of PPE into nanoribbons use has been made of different insulating amorphous supports including glass and carbon-coated copper grids; the resulting molecular arrangements on solid surfaces have been compared to the ones obtained on a crystalline mica substrate. SFM investigations of films processed by casting a very dilute polymer solution on a glass support reveal self-assembled solid nanoribbons (Fig. 6.44a) with a length in the micrometer range, lying flat on the substrate.

The hexyl chains, whose length in an extended conformation spans over 1.9 nm, are more likely to pack similarly to the case on mica, i.e. they are standing perpendicular to the substrate in a disordered conformation [Sam99a] spanning only ~ 1.5 nm. The ribbon heights of (3.9±1.0) nm observed here suggest a packing of 2-3 molecular layers, albeit also multiple layers have been detected.


122

Figure 6.44: A) Tapping Mode SFM height image of an on average 11.2 nm long PPE on glass. Singular nanoribbon assembled on the grainy glass surface. B) TEM micrograph showing webs of ribbons on a carbon copper mesh support made by casting an on average 16.4 nm long PPE from a 0.1 g/l solution in THF.

Figure 6.45: Histograms of the distribution of ribbon widths for PPE from profiles on TEM micrographs. Contour length of the molecule according to 1H-NMR results and number of ribbons measured: (A) 7.9 nm / 164; (B) 16.4 nm / 167. Mole fraction (solid line) and weight fraction (dashed line) of the theoretical Schulz-Zimm distribution are in fairly good agreement with the experimental data.


123

Figure 6.46: Schematic representation of the molecular packing in the nanoribbon: a) Top view: a ribbon is composed by parallel, fully extended backbones, stacked perpendicular to the main ribbon axis; b) Side view: the lateral alkyl chains are disordered between adjacent backbones in a bilayer/trilayer aggregate standing on the substrate.

.

The apparent width of these assemblies evaluated from individual profiles on SFM micrographs amounts to (41±15) nm. The effective width can be calculated by deducting the broadening 2Delta due to the finite size of the Si tip apex that for a terminal radius of a clean new tip of R=(13±7) nm amounts to (19±6) nm. The true width of the nanoribbon, with a 6 nm error bar, is also in this case in fair agreement with the macromolecular length of 11.2 nm (U=2.34) considering that the terminal radius of the tip could get slightly broader after several scans

Moreover, a TEM analysis has been executed on PPE films of different molecular weights cast on a carbon-coated copper grids. Nanoribbons have also been found (Fig. 6.44b), and


124

the distribution of their widths was determined from profiles on TEM micrographs. A polymer with an average contour length of 7.9 nm (sample A in Table 6.6) exhibits a ribbon width of (10±3) nm, while a macromolecule with a length of 16.4 nm (sample D in Table 6.6) forms ribbons with an average width of (15±4) nm. The corresponding two distributions have been plotted in histograms in Fig. 6.45. The nanoribbons are remarkably stable arrays under the electron beam of the electron microscope.

Similarly to what has been observed on mica, with increasing PPE length the peak of the experimental data shifts to higher values and the distribution of ribbon widths gets broader. The ribbon widths correlate well with the theoretical Schulz-Zimm distribution which describe the distribution of molecular weights for a polycondensation synthesis [Bra89]. The peak of the ribbons widths distribution, in terms of spatial units, is in both cases similar to the peak of the histograms of the mole fraction distribution and of the weight fraction distribution, indicating a good agreement between the most frequent ribbon width and the average molecular length obtained from 1H-NMR analysis. In addition, a fairly good agreement between the width of the distribution of ribbon widths and the width of the Schulz-Zimm mole fraction distribution can be recognized. This confirms that measuring the width of these nanostructures is an alternative route to evaluate the molecular weight distribution for this kind of polymer as previously demonstrated using mica as a substrate [Sam99a].

In summary, SFM and TEM applied to samples cast on different substrates, give rise to similar results in terms of the sizes of the nanoribbons in the XY plane while their heights have been only detected by SFM [But92]. Similarly to what has been concluded on mica substrates, the nanoribbons appears to consist of macromolecules oriented with their stiff backbones parallel to the substrate and the hexyl side chains perpendicular to the basal plane of the substrate. The backbones are stacked parallel to each other creating a ribbon of stacked pi-conjugated polymers on the substrate (Fig. 6.46).

The demonstrated ability to grow these architectures also on non-crystalline substrates indicates that the main driving force of this self-assembly are intermolecular interactions between the pi-conjugated macromolecules. The substrate plays only a secondary role of inducing a particular orientation of these assemblies. Furthermore, the transparency of the glass support opens the possibility to perform spectroscopical investigations like probing the birefringence and fluorescence of single nanoribbons using Scanning Near-Field Optical Microscopy. This would allow to determine physico-chemical properties of PPE on the nanoscopic level and compare it to single molecular and solid state properties. [Sam99c]


125

6.5.5 Morphology at different rate of the deposition process

In order to gain further insight into the growth process of these PPE nanoribbons, by solution casting, on crystalline mica substrates different solvents have been used. This allowed to vary the rate of evaporation of the solvent and of crystallization of the organic compound.

Figure 6.47: Topographical Tapping Mode Scanning Force Microscopy images of PPE cast on freshly cleaved muscovite mica. a) Solvent : THF, height range (peak-valley) h=30 nm; b) Solvent:mixture THF-phenyloctane (5:1), h=20 nm.

The SFM topographical images in Fig. 6.47 show nanoribbons grown on mica from very dilute solutions of a PPE with an average contour length of 7.9 nm (sample A in Table 6.6). These architectures, obtained by solution casting, exhibit a length of up to several micrometers and a two-dimensional cross-section on the molecular scale, confirming once again what has been described in the previous three paragraphs.

The ribbons in Fig. 6.47a were produced from a PPE solution in tetrahydrofurane (THF). These anisotropic nanostructures, whose width of ~11.1 nm (corrected for the lateral broadening tip effect [Sam98a]) fits relatively well with the size of the molecule, hardly shows any favored orientations.

On the other hand, Fig. 6.47b displays highly oriented nanoribbons obtained from the same PPE polycrystalline powder dissolved in a 5:1 mixture of THF and 1-phenyloctane. These anisotropic assemblies exhibit an effective width of ~7.1 nm and are aligned preferentially along directions according to the three-fold symmetry of the underlying mica, with an average


126

angle of (61±7)° between ribbon segments. It is worth to note that it was not possible to grow well-ordered molecular aggregates with a ribbon like shape using a fast deposition method such as spin-coating. The different molecular organization on the basal plane of mica is attributed to the solvent evaporation process that is followed by the self-assembly of the organic compound into the nanoribbons. The ribbon formation, may be divided in two steps:

  1. self-segregation of the molecules into straight ribbon segments with homogeneous widths, and
  2. self-assembly of long oriented nanoribbons induced by the crystalline substrate.

The first step is governed by intermolecular interactions between the conjugated macromolecules, as previously demonstrated by the ability to grow nanoribbons on amorphous substrates [Sam99c]. The second step takes place on crystalline substrates and is observed only for very slow solvent evaporation rates. In the present case THF has a boiling point (67 °C) which is much smaller than the one of 1-phenyloctane and also of the solvent mixture. Consequently, the time required for the complete evaporation of the solvent was a few hours (2-3) for the first case and a few days (2-3) for the latter. This suggests that in the case of pure THF the solvent evaporation was too fast for step (2) to occur.

The height of the ribbons amounts to (2.9±0.7) nm for the THF deposition, whilst it is (4.7±1.4) nm for the deposition from the solvent mixture. Comparing this value to the width of the molecule of ~1.5 nm [Sam99a] it is suggested that the ribbon is typically made of a molecular bilayer in the first case and of a triple layer in the latter case. The relatively large error bars indicate that the distributions are not sharp but single layers and also higher multiple layers occur. The trend indicates, however, that slowing down the deposition rate leads to thicker nanoribbons. The architecture of these ribbons is as previously described and depicted in Figs. 6.42 and 6.46.

These results demonstrate that it is possible to drive the self-assembly of suitably functionalized PPEs towards nanoribbons. The possibility to align the nanoribbons along the crystallographic axes of the crystalline non conducting substrate by slowing down the evaporation and consequently the crystallization process indicates that the growth of these nanostructures is a kinetically governed phenomenon. [Sam98b]


127

6.5.6 Morphology with different side chains

Further PPE derivatives, synthesized in the group of Prof. Dr. W. Heitz (Department of Chemistry, University of Marburg) has been investigated. They posses acid functions (CPPE in Fig. 6.48) or ester side groups (EHPPE in Fig. 6.48) attached to the main conjugated skeleton. These soluble moieties have been also cast on mica and SFM investigation have been carried out. EHPPE crystalline powder has been solubilized and diluted in THF, while CPPE has been solubilized and diluted in KOH+H20.

Figure 6.48: Chemical formulae of the CPPE and EHPPE.

Thin films of the conjugated polymer have been prepared by casting the solutions (from 0.065 g/l to 0.00325 g/l) on freshly cleaved muscovite mica discs.

The self-assembly of rigid rod polymers is strongly influenced by its distribution of molecular weights, as previously observed, and by the steric hindrance due to the chains [Wed96].

Films of EHPPE cast from 0.013 g/l show anisotropic features composed of spheroids with a preferential orientation which seems to be induced by the hexagonal crystal lattice of the mica substrate (Fig 6.49a). At higher dilution, a 0.0033 g/l solution of EHPPE in THF cast onto mica gives rise to anisotropic flat domains on the substrate (Fig. 6.49b). These elongated features, similar to wetting foots, are made up of one hump in the middle and a surrounding flat corona, which is a self assembled monolayers with a uniform thickness of 2.9 nm. The calculated width of the polymer with the alkyl chains extended is 1.4 nm; therefore a double layer packing of the PPE molecules can be expected, as also obtained for PPE with hexyl side chains. Noteworthy, these ellipsoid features have not been visualized with TEM on carbon-copper coated grids indicating the influence of the crystal mica substrate on the self


128

assembling of this PPE derivative.

Figure 6.49: Tapping Mode SFM height images of: a) EHPPE from 0.013 g/l in THF, h=40 nm; b) EHPPE from 0.0033 g/l in THF, h=40 nm; c) CPPE from 0.017 g/l in H20+KOH, h=7 nm.

Besides, films of CPPE, which possess a lower steric hindrance due to the side chains, exhibit more anisotropic structures (ribbon-like) (Fig. 6.49c), from low concentrated (0.017 g/l) solution, with a thickness of 1.4 nm that is just the double of its calculated polymer width of 0.7 nm, confirming once again the double layer packing.

In both of the two derivatives investigated here the polymer self-assembles in well defined structures in one dimension (Z), while the other dimensions (XY) exhibit a low anisotropy, which can be attributed to both the fairly high polydispersity of the synthetic polymer as evaluated by GPC and in particular to the steric hindrance of the side groups on the aromatic rings.

6.5.7 Morphology of thiol free end functionalized PPE

The reactivity of conjugated systems that exhibit free thiol functions at their edges is very high as previously observed for the case of alkenethiols. Indeed they tend to react into disulphide species. In a THF solution an end-functionalized PPE has been deprotected such that the thiol end groups on the main chains are made to appear. The average contour length of this PPE is 9.2 nm. In order to gain insight into the self-assembly of these deprotected species, thin films have been prepared by solution casting on a SiO2 wafer.

Unlike previous cases, the morphology is not made of spider webs or long nanoribbons,


129

although short ribbon segments have been occasionally observed (indicated with white arrows in Fig. 6.50). The width of these ribbons is ~30 nm, taking into account the tip broadening effect.

This results confirms that the reactivity of the thiol free PPE is rather high; this renders difficult to drive the self-assembly towards well defined and anisotropic architectures.

Figure 6.50: Tapping Mode SFM height image of PPE with free thiol end-groups
assembled on a SiO2 wafer. Height of the gray scale h=50nm.

Arrows indicate short ribbon with a molecular cross-section

6.6 Electronic structure of phenyleneethynylene derivatives

6.6.1 Introduction

Photoelectron spectroscopies (PES) can be used to study conjugated polymers at surfaces and their early stages of metal interface formation. They provide information on both chemical and electronic properties and, on the same time, they are extremely surface sensitive and non-destructive to organic systems [Sal96]. The measurements need to be carried out in vacuum; indeed, to determine the kinetic energy of the photoemitted electrons, without significant collisions with molecules in the background, a pressure in the analysis chamber of p < 10-5 mbar is required. In addition, in order to have a non contaminated surface, in particular when working with reactive metals (donors) doping the conjugated species, it is necessary to work in a ultra-high vacuum (UHV) environment, namely at a pressure < 10-9 mbar.


130

Figure 6.51: Principle of photoelectron spectroscopy of thin organic films on a conductive substrate.

The principle of the technique is represented in Fig. 6.51: a photon (hny) hits and is absorbed by a molecule. This induces an excitation of the electron from its ground state (X in Fig. 6.51) to an excited state (X+ in Fig. 6.51), where (X+ - X) corresponds to the ionization energy (Ii). The electrons is then photoemitted from the molecule with a kinetic energy EK. The binding energy of the electron in the molecule is defined as EB=hny-EK= Ii.

Using soft X-ray photons (namely Xray Photoelectron Spectroscopy, XPS, known also as ESCA), from an excited Mg(K?) radiation (1254.6 eV) it is possible to investigate both the atomic core-electron energy levels (Ci) and the valence electron energy levels (Vi). With ultraviolet radiation (namely Ultraviolet Photoelectron Spectroscopy, UPS) using a helium discharge source - HeI (21.2 eV) or HeII (40.8 eV) - only the valence electronic states may be studied, but with a resolution higher than the one that is usually achieved by XPS. Noteworthy, PES spectra provide a one-to-one correspondence between the peaks in the photoemitted electron energy distribution and the electron energy states in the molecules as depicted in Fig 6.52. The ideal thickness of the organic adsorbate for these kind of measurements, which arises from the short elastic mean free path for low kinetic energy electrons in solids, is between 5 and 10 nm [Sal96].


131

Figure 6.52: Electronic levels in a pi-conjugated molecule. EF=Energy of the Fermi level; EVac=Energy of the vacuum level; EKin=Kinetic energy = hny-Ii; phis=work function. Bottom of the figure: schematic cartoon of the electronic levels. Top: corresponding PES spectra.

The typical electronic structure in a pi-conjugated molecular system is composed of different characteristic features. A core level "band" , a valence "band" whose upper level is the Highest Occupied Molecular Orbital (HOMO), and unoccupied levels that have as the lowest level the Lowest Unoccupied Molecular Orbital (LUMO). In the case of a pristine sample (pure material, non doped) the Fermi level is placed just in the middle of the band gap (energetic gap between HOMO and LUMO). On the other hand the gap between the Fermi level and the vacuum level is known as work function (phis).


132

6.6.2 Work functions of pristine and doped phenylenethynylene trimer and polymer

In this work, the aim was to characterize the electronic structure of oligomeric and polymeric phenyleneethynylene derivatives both of the pristine and of the n-doped moiety.

Four different films were investigated: two spin-coated PPE polymer films, one spin-coated and one UHV-sublimed PPE-trimer film.

XPS has been used to determine the stoichiometric composition and purity of the films. Fig. 6.53 shows XPS survey spectra of the four investigated films together with the spectrum of a clean (aceton/iso-propanol cleaned) Au substrate. The spectra confirmed the overall purity of the adsorbate, with the exception of a small quantity of residual oxygen (C to O ratio was between 0.98 and 0.99). The thicknesses of the four films are lying between 3 and 4 nm, estimated from the reduction of the intensity of the Au 4f 7/2 -line due to the organic adsorbate.

Figure 6.53: XPS survey spectra. a) Au surface cleaned with isopropanol and aceton bath; b) spin coated PPE sample #1 (nominal thickness d=3.1 nm); c) spin coated PPE sample #2 (d= 2.7 nm); d) spin coated PPE trimer (d= 3.0 nm); e) sublimed PPE trimer (d= 4.1 nm).


133

On the other hand UPS allows to explore the electronic structure near the Fermi level: it allows to localize the HOMO and LUMO and to determine the work function phis from the cut-off of the secondary electrons in the HeI spectrum [Sal96]. Fig. 6.54 shows the respective UPS - HeII spectra. The organic films are thick enough to suppress the weakly bound electrons of the gold substrate emitted near the Fermi level. As marked in Fig. 6.54, the UPS spectra of the polymer exhibits in the pi-region three features and the spectrum of the spin-coated trimer four features (here, the feature with the lowest binding energy seems to be split) whereas the pi-features of the sublimed trimer are barely observable. This latter spectrum is dominated by the sigma-bonds of the alkyl chains. In order to increase the pi signal the sublimed trimer film has been annealed for 15 minutes in UHV at 100 °C. At this temperature a good part of the adsorbate desorbed. Hence, only the characterization of the spin-coated trimer and polymer films was pursued further.

Figure 6.54: UPS HeII spectra of a) Au surface cleaned with isopropanol and aceton bath (work function phis=4.1 eV); and pristine samples of: b) spin coated PPE sample #1 (phis =4.2 eV); c) spin coated PPE sample #2 (phis = 4.3 eV); d) spin coated PPE trimer (phis= 4.5 eV); e) sublimed PPE trimer (phis = 4.0 eV).


134

6.6.2.1 Spin coated PPE trimer

The recorded XPS (survey scan, C 1s and Na 1s) and UPS (HeI, HeII spectra at increasing doping are shown in Figs. 6.55 to 6.58. The work functions phis are determined from the HeI spectra (not shown here) and the doping levels (number of Na atoms per monomer) are estimated from the ratio of the intensity of the C 1s and the Na 1s peak of the XPS survey spectra (see Fig. 6.55). The evolution of both values is listed in Table 6.7.

Table 6.7: n-doping of the PPE trimer: evolution of the work function phis with the doping level.

Time of doping (min)

0

1

2

3

5

phis (eV)

4,5

3,2

2,8

2,7

2,6

Na-atoms /monomer

0

0.38

0.84

0.98

1.38

Figure 6.55: XPS survey spectra of spin coated sample of phenyleneethynylene trimer at
increasing level of doping. a) Na/C/Au=0/100/17 ; b) 1.9/100/19.6 (0.38 Na-at/mon.); c)

4.2/100/21.2 (0.84 Na-at/mon.); d) 4.9/100/20.1 (0.98 Na-at/mon.); e) 6.9/100/18.5 (1.38 Na- at/mon.).


135

Figure 6.56: UPS HeII spectra of spin coated film of phenyleneethynylene trimer at increasing level of doping. The spectra are plotted relative to the Fermi level (left) and to the vacuum level (right). a) pristine: phis=4.5 eV ; b) phis= 3.2 eV (0.38 Na-at/mon.); c) phis= 2.8 eV (0.84 Na-at/mon.); d) phis= 2.7 eV (0.98 Na-at/mon.); e) phis= 2.6 eV (1.38 Na-at/mon.).

Figure 6.57: XPS spectra zoomed on the Na 1s peak of spin coated sample of phenyleneethynylene trimer at increasing level of doping. The spectra are plotted relative to the Fermi level (left) and to the vacuum level (right). a) phis= 3.2 eV (0.38 Na-at/mon.); b) phis= 2.8 eV (0.84 Na-at/mon.); c) phis= 2.6 eV (1.38 Na-at/mon.).

Figure 6.58: XPS spectra zoomed on the C 1s peak for spin coated sample of phenyleneethynylene trimer at increasing level of doping. The spectra are plotted relative to the Fermi level (left) and to the vacuum level (right). a) phis= 4.5 eV (pristine); b) phis= 3.2 eV (0.38 Na-at/mon.); c) phis= 2.8 eV (0.84 Na-at/mon.); d) phis= 2.6 eV (1.38 Na-at/mon.).


136

The sequence of HeII spectra of Fig. 6.56a (the spectra are plotted relative to EF) shows that the four pi peaks are smearing out with increasing Na deposition. Due to an increase of the Fermi level and/or the creation of a surface dipole, the whole spectra are shifted to higher binding energies (Dr. M. Keil, IFM Linköping, Sweden, private communications). An appearance of new peaks directly below the Fermi level is not observed.

In order to take into account the shift of the Fermi level due to Na doping, the HeII spectra of Fig. 6.56b are plotted relative to the vacuum level (the position of EF is marked as a dotted line). Since no significant shifts of the whole spectra after the respective doping steps can be observed in Fig. 6.56b the changes of the work function from 4.5 eV to 2.6 eV are necessarily due to an increase of the Fermi level (n-doping).

Therefore, the Fermi level increases by about 1.9 eV after doping of 1.38 Na-atoms per monomer. At higher doping levels (0.84 - 1.38 Na-atoms / monomer) a new structure appears at ca. 5.5 eV. This feature lies around 2.5 eV below the Fermi level whereas the peak of the highest occupied state before doping lies only 1.5 eV below EF. At higher doping levels (0.84 - 1.38 Na-atoms / monomer) unstructured features grow up directly below EF. The shape of these features does not allow to assign them clearly to new levels grown near EF. Figs. 6.57a


137

and b show sequences of Na 1s XPS spectra relative to EF and Evac, respectively. The Na 1s peak of Fig. 6.57a shifts for higher doping levels to higher binding energies (due to a Fermi level shift), whereas no shifts can be observed in Fig. 6.56b. The C 1s spectra of Figs. 6.58a and 6.58b show similar behaviors. Here, the C 1s peak of Fig. 6.59b shifts slightly to a smaller binding energy (relative to Evac) indicating a non-homogeneous distribution of Na in the deeper layers.

6.6.2.2 Spin coated polymer

A polymeric phenyleneethynylene (PPE) solution in THF (1.66 g/l) has been spin-coated on the Au film. Here only the case of the sample # 2 (Fig. 6.53) will be discussed. The recorded XPS and UPS spectra are shown in Figs. 6.59 to 6.60. The following table shows the evolution of the work functions and the doping levels determined from the HeI spectra (not shown here) and the XPS survey spectra (see Fig. 6.59), respectively:

Table 6.8: n-doping of the PPE: evolution of the work function phis with the doping level.

Time of doping (min)

0

1

2

3

phis (eV)

4,3

3,5

3,0

2,6

Na-atoms /monomer

0

0.42

0.94

1.1


138

Figure 6.59: XPS survey spectra of spin coated sample of PPE at increasing level of doping. a) Na/C/Au=0/100/23.4 (pristine); b) 2.1/100/24.8 (0.42 Na-at/mon.); c) 4.7/100/28.1 (0.94 Na-at/mon.); d) 5.5/100/29.3 (1.1 Na-at/mon.).

The sequence of HeII spectra of Fig. 6.60a (relative to EF) shows a smearing of the three pi states together with a shift of the whole spectrum to higher binding energies upon successive doping. After aligning the spectra relative to Evac (Fig. 6.60b) it is possible to observe a slight shift of the spectra of about 0.5 eV to lower binding energies due to dipole effects (non-homogeneous Na-distribution, Dr. M. Keil, IFM Linköping, Sweden, private communications). Therefore the increase of EF after doping of 1.1 Na-atoms / monomer can be estimated as phispristine-phisdoped-shift = 4.3 - 2.6 - 0.5 = 1.2 eV. If compared to the trimer film (see Fig. 6.56b) the behavior of HeII spectra of the polymer film (Fig. 6.60b) seems to be different upon doping. After doping of 0.42 Na-atoms per monomer the peak of the highest occupied orbital (at 6 eV) increases in a first step in intensity and then, in a second step at higher doping levels (0.94 - 1.1 Na-atoms per monomer), the peak either shifts to lower binding energies (to 5.5 - 5 eV) or a second peak at 5.5 - 5 eV additionally appears in the forbidden band gap. This new structure lies ca. 2.5 eV below EF. This value is comparable to the value of the doped trimer film. As seen in the case of the trimer film, the HeII spectra of the polymer film (Fig. 6.60b) also exhibit unstructured features after doping, which occur directly below EF.


139

Figure 6.60: UPS HeII spectra of spin coated sample of PPE at increasing level of doping. The spectra are plotted relative to the Fermi level (left) and to the vacuum level (right). a) pristine: phis=4.3 eV ; b) phis= 3.5 eV (0.42 Na-at/mon.); c) phis= 3.0 eV (0.94 Na-at/mon.); d) phis= 2.6 eV (1.1 Na-at/mon.).

Focusing on the Na 1s and C 1s peaks (here not shown), results similar to the ones of the trimer were obtained, indicating an inhomogeneity of the Na distribution in the polymer. In addition, an analogous study on the polymer sample #1 (Fig. 6.53) confirmed these results.

In summary, the n-doping of the three investigated films with Na was successful. Upon doping, the work functions phis of all samples are drastically decreasing and the Fermi levels are increasing. Additionally, due to doping, a new peak appears in the forbidden band gap directly above the valance band edge and unstructured features grow up below the Fermi level. Because of overlapping structures from the background in this energy region, the line shape of these new structures (especially those appearing directly below EF) can hardly be determined. This complicates the interpretation of the doping process.


140

6.6.3 Optical absorption investigation of PPE

The optical absorption of the polymer in the UV-visible range have been investigated in order to gain further insight into the electronic structure and in particular to determine the band gap. Indeed the peak of the absorption spectra can be assigned to the electronic transition between HOMO and LUMO, i.e. band gap.

Figure 6.61: UV-Visible optical absorption spectra of PPE. a) Degree of polymerization (DP)= 8; b) DP=35. (The spectra have been recorded by Dr. V. Francke in MPI for Polymer Research, Mainz, Germany).

The peak of the absorption spectra shifts to higher wavelengths with increasing polymer length (Fig. 6.61). In fact a polymer with a degree of polymerization (DP) of 8 shows a maximum at 385 nm while for a DP=35 the peak amounts to 402 nm. The band gap amounts to 3.22 eV and 3.09 eV, respectively, i.e. it gets smaller with increasing average length of the macromolecule, as one would expect according to a particle in a box model [Bei69].

6.6.4 Density of valence states calculations

Calculations of the density of valence states (DOVS) were carried out by Dr. Donizetti dos Santos by means of Valence Effective Hamiltonian (VEH) simulations on the basis of an AM1 optimized geometric structure, which usually provides good estimates of the electronic structure of conjugated polymers. The theoretical ionization potential (Ip) and the electron affinity (Ea) of poly(para-phenyleneethynylene) have been determined to Ip = 5.3 eV and Ea


141

= 2.5 eV; as a consequence the band gap amounts to 2.8 eV. This is in fairly good agreement with the experimental results obtained with UV-Visible absorption spectroscopy taking into account that the experiments have been carried out on rather short polymers.

Figure 6.62: Valence Effective Hamiltonian (VEH) of the density of valence states (DOVS) simulated spectra and UPS HeII spectra of a spin coated film of pristine PPE trimer. The spectra are plotted relative to the Fermi level.


142

Figure 6.63: Valence Effective Hamiltonian (VEH) of the density of valence states (DOVS) simulated spectra of a doubly charged system and UPS HeII spectra of spin coated film of a doped PPE trimer (1.38 Na at/mon). The spectra are plotted relative to the Fermi level.

Furthermore a simulation of the UPS HeII spectra for the case of the trimer has been computed. The obtained spectra, plotted relative to the Fermi level, are compared to the experimental UPS HeII plot: for the case of the pristine trimer (Fig. 6.62) the theoretical plot fits well with the experimental one. In the latter case shown in Fig. 6.63 the UPS spectra of the 1.38 Na atoms/molecule are reasonably comparable with the doubly negative charged PPE-trimer, although one extra peak near to the Fermi edge can be observed only in the theoretical spectra.

6.7 Current-Voltage (I-V) Measurements

6.7.1 Visualization of the nanoelectrodes

The characterization of the Au electrodes, developed following the recipe discussed in


143

paragraph 4.3, has been always carried out both performing current-voltage (I-V) type of measurements, in order to proof the existence of a gap, and with Tapping Mode SFM imaging, to visualize the structure and to evaluate the size of the gap. It is fair to note that, due to the finite size of the SFM tip (terminal radius ~ 10 nm), it turns out to be not feasible to visualize of gaps smaller than the diameter of the tip (~ 20 nm). This suggests that SFM is not the ideal technique for this kind of measurements while high resolution TEM seems to be more appropriate.

The Au nanoelectrodes are contacted to macroscopic Au wires using Ag paste. The I-V measurements are performed contacting the macroscopic wires with the voltage source.

Gold nanogaps grown on SiO2 wafers from three different sources have been used:

  1. Dr. S. Rogaschewski (Department of Physics, Humboldt University Berlin);
  2. Dr. K. Kragler (Siemens AG, Erlagen);
  3. Dr. A. C. F. Hoole (Department of Engineering, University of Cambridge, UK).

Figure 6.64: Tapping Mode SFM height image of Au nanoelectrodes produced by Dr. S. Rogaschewski. The height of the gray scale (h) is: a) h= 50 nm ; b) h= 40 nm.

The structure of the first type of electrodes has been studied first. The electrodes exhibit a gap usually of about 50 nm (Fig. 6.64) that is unfortunately still rather big. The value has been determined considering the broadening due to the SFM tip during scanning. The model expressed in formula 6.3 can be used also in this case, with a terminal radius of the tip R=(13±7) nm and a height (h) of the Au layer (electrode) that can be evaluated from singular profiles. For h=10 nm the tip broadening 2Delta ~ 25 nm. Moreover the gap reproducibility is


144

rather poor. Indeed it sometimes has been detected that the lift-off of the gold during the fabrication did not occur completely. In this case the electrodes do not posses any gap (Fig. 6.65).

Figure 6.65: Tapping Mode SFM height image of Au nanoelectrodes produced by Dr. S. Rogaschewski. The lift off process was not perfect; therefore there is no gap between the two electrodes. h= 70 nm.

Figure 6.66: Tapping Mode SFM height image of Au nanoelectrodes
produced by Dr. K. Kragler. h= 40 nm.

The second type of electrodes are shown in Fig. 6.66. The reproducibility of a gap of 10-20 nm is remarkable.


145

Figure 6.67: Tapping Mode SFM height image of Au nanoelectrodes
produced by Dr. A.C.F. Hoole. h= 40 nm.

The third type of electrodes are grown on a 1 µm thick SiO2 insulating layer on a silicon wafer; the leakage current is about 10-13 at a voltage of ± 1V. They are produced by evaporating a 10 nm thick Au layer on the flat substrate. The gap of the nanoelectrodes (Fig. 6.67) determined from topographical profiles on SFM images amounts to ~ 20 - 25 nm, if one considers the SFM tip broadening effect. The reproducibility of this size of the gap is rather high.

The different quality between the three types of electrodes investigated can be due to a variety of factors which can play a role in the electrodes preparation (Fig. 4.3), such as the resolution of the electron beam irradiating the PMMA, the quality of the development, of the metallization and of the lift-off steps.

6.7.2 Bridging metallic nanogaps with molecules

The aim of this experiment is to probe the conductivity of a well defined (macro)molecular architecture by interfacing a PPE nanostructure to the Au nanoelectrodes. The molecules, when bearing the thiol end-groups in the alpha and omega position, are expected to chemisorb on the Au nanoelectrodes.

The deprotection reaction of the end groups needs to be performed in situ on the electrodes. Indeed, it is of prime importance to carry out the deprotection reaction of the carbamoyl


146

groups which cap the thiol functions attached to the main chain of the PPE in a free oxygen environment, in order to avoid the aggregation of PPEs through disulphide bridging.

Figure 6.68: Adsorption of sulphur free PPE. a) Electrochemical adsorption induced by the applied voltage between the two electrodes; b) the self-assembly is likely to be governed by the chemisorption of sulphurs on Au nanoelectrodes.

Initially the chemisorption reaction has been tested as a function of the chemical nature of the end groups. A PPE exhibiting sulphur free end-groups on the main chain (Fig 5.10 formula b) have been self-assembled between the two Au nanoelectrodes, and contemporary a voltage of 1V have been applied between the gap: the moieties adsorb (precipitate) onto the anode as shown in an optical microscope snapshot in Fig. 6.68a. In this case the molecule exhibits two negatively charged end groups playing a paramount role in the self-assembly process. The scale length of the image in Fig. 6.68a is some hundreds of micrometers. On the other hand, a more reduced adsorption of the molecules on the electrodes is obtained by casting the same solution without applying voltage between the two electrodes (Fig. 6.68b). In this case the coating appears to be uniform, although different between the Au and the SiO2 surface; this suggests that in the first example the adsorption was electrochemically driven. In the latter case it is likely that the adsorption of the sulphur end functionalized molecules is governed by the chemisorption of the sulphurs on Au, even though the resolution of the optical microscope does not permit to prove this statement.

The possibility to control the self-assembly of the thiol free PPE on a nanometer scale is rather poor, as discussed in paragraph 6.5.7. Applying a voltage between the electrodes, also in the case of a PPE bearing SH end groups (Fig. 5.10 formula c), the molecular adsorption on the electrodes takes place abruptly (similar to Fig. 6.68a). Because of this reason experiments of the self-assembly of thiol-end PPE have been carried out on the Au nanoelectrodes produced by Dr. K. Kragler, Siemens AG without applying voltage between


147

the gap during the self-assembly.

Figure 6.69: Tapping Mode SFM image of a PPE with an average contour length of 14 r.u. . The height image shows electrodes strongly coated. a) Height image: h=70 nm; b) Phase image.

On this molecular scale, even using low concentrations of the solution (~ 0.04 g/l) in a mix THF-phenyloctane it was not possible to produce any ordered architectures. It is worth to note that this is the same range of concentrations that has been used for producing PPE ribbons, such as the ones in Fig. 6.41. In Fig. 6.69 a disordered adsorbate coats the Au nanoelectrodes which are hardly recognizable. The electronic properties of this molecular adsorbate has been characterized. The I-V plots show that the coating gives rise to an enhancement of the current between the two electrodes, although, due to the lack of molecular ordering in the gap, it is not possible to assign this conduction to a well defined ensemble of molecules. The conductivity of this non-doped material is still in the range of rather insulating materials (Fig. 6.70). The doping of the PPE, due to the existence of the triple bond in the repeat unit, is chemically feasible only by reducing the polymer with an alkali metal (n type doping). This would require to work in a very controlled environment free of oxygen, namely ultra-high vacuum (UHV), which was not done here.


148

Figure 6.70: Current-voltage curve: A) non coated electrodes; b) coated electrodes. The curve
appears to be symmetric and the maximum current is on the order of 10 nA at ± 1V.

6.8 Growth of dry hexakis-dodecyl-hexabenzocoronene films from solutions

By combining chemical sensitive techniques such as photoelectron spectroscopies with the spatial resolution of Scanning Probe Microscopies it is possible to bestow information on both the electronic properties and the molecular arrangements. The aim of the following experiment is to grow and study well defined micrometer size structures of soluble synthetic nanographitic disc like molecules [Mül98c] on conductive substrates, which are atomically flat on the micrometer scale. Physisorbed monolayers at the solid-liquid interface of hexakis-dodecyl-hexabenzocoronene (HBC-C12) [Sta95b] and larger allotropes [Iye98] have been already investigated with STM and a diode-like electrical behaviour of the aromatic part of the molecule in the gap has been probed by means of Scanning Tunneling Spectroscopy (STS) [Sta95b].

HBC-C12 (Fig. 6.71) has been self-assembled from solution as dry thin films on HOPG. It has been characterized by Tapping Mode - Scanning Force Microscopy and UPS. XPS was also used to determine the stoichiometric composition and purity of the films.


149

Figure 6.71: hexakis-dodecyl-hexabenzocoronene (HBC-C12)

As described in paragraph 5.3.3, the films have been prepared following different procedures in order permit the evaporation of the solvent and molecular assembling to occur at decreasing rate.

The diameter of the HBC-C12, in the case where the alkyl side chains are in their fully extended conformation, amounts to 40 Å while the thickness of the planar molecular disc is 3.55 Å. It is worth to note that this thickness is very similar to the one of a monolayer of HOPG (3.35 Å), due to their alike structure. Because of this reason it is hazardous to assign one interface (domain) to HOPG or HBC-C12 just on the base of the thickness detected from SFM topographical profiles. This assignment can be achieved only with a careful observation of the shape of the boundaries and, in cases of uncertainty, by using complementary a chemically sensitive imaging mode such as Phase Imaging SFM [Fin97].

Scanning Force Microscopy images of a HBC-C12 spin coated film on HOPG (method A in paragraph 5.3.3) are displayed in Fig. 6.72. They exhibit a flat morphology made of polygonal planes extended on the micrometer scale. The steps in height between these planes amount to 3.5 Å and multiples of it, in good agreement with HOPG steps [Wie92]. Phase imaging (Fig. 6.72b) reveals the existence of only one phase covering the sample surface. In addition Scanning Tunneling Microscopy investigations on these films have demonstrated that the conductive substrate is covered by a non-conductive adlayer. Hence, the graphite surface appears to be homogeneously coated by the organic semiconducting compound. Albeit the interfacial roughness of this film on the micrometer scale is very little, the degree of molecular order of the organic moiety adsorbed onto the flat substrate may be small on a molecular scale due to the fast rate of physisorption.


150

Figure 6.72: Tapping Mode SFM images of HBC-C12 film prepared by spin coating the quantity to form ~ 1000 layers coating the HOPG (method A described in paragraph 5.3.3). a) Height image: z- scale (h)=10 nm; b) Phase image: z- scale=7°.

Figure 6.73: Tapping Mode SFM micrographs of HBC-C12 grown on HOPG by solution casting (method B in paragraph 5.3.3) a quantity sufficient to coat the HOPG with one layer. a) Height image: h=20 nm; b) Phase image: z- scale=6°; c) Height image: h=20 nm.

On the other hand the morphology of HBC-C12 films obtained by fast casting the solution (procedure B) to produce a single layer is shown in Fig. 6.73. In this case, HBC-C12 molecules self-assemble into inhomogeneous mono-layers (small islands) with a constant height of 3.5 Å, or multiples thereof, which is consistent with a layer by layer growth of the HBC-C12 disk-like molecules oriented flat on the HOPG surface. Thicker domains of higher multiples of 3.5 Å, up to about 25 Å were also observed. The shape of these domains suggests that the stage of the molecular assembly displayed here represents a step towards the formation of more extended clusters occurring via coalescence. The interpretation of the


151

surface composition based on the contrast of the height and phase images, confirmed the existence of different phases on the sample surface that could be due also to a different orientation of the molecules with respect to the basal plane of the substrate.

Figure 6.74: Topographical Tapping Mode SFM image of a HBC-C12 surface produced by slow solution casting of one layer in controlled environment (route C in paragraph 5.3.3). h=15 nm. The white arrows indicate the preferential directions along which the layers grow anisotropic. The angle between them is 60° in perfect agreement with the three-fold symmetry of the HOPG substrate.

In the third case, thin organic layers prepared by a slow deposition procedure (route C, paragraph 5.3.3) are made of well defined monolayers (Fig. 6.74). Noteworthy, the overlayers exhibit a texture with orientation along preferential directions with an angle between them of 60°, according to the three-fold symmetry of the crystalline HOPG substrate, as indicated by the arrows in Fig. 6.74.

Applying relatively high shear forces in the tens of nN range with Scanning Force Microscopy in contact mode it was not easily possible to scratch all the films by moving the upper layers. Nevertheless the first layers in contact with the substrate are relatively strongly bound by van der Waals interactions, and therefore hardly removable [Bis00].

It is likely that in this first layer both the aliphatic part the molecule and the aromatic one are lying flat on the basal plane of the substrate [Bis00] due to hybridization of the electronic states of the organic molecules with the “metallic band“ of the graphite. Then overlayers are grown on the first layer in contact with the substrate. These upper layers exhibit an increasing degree of order with a decreasing rate of the solvent evaporation. The mechanism governing the formation of these layers may be divided in three steps:

  1. homogeneous coating of the substrate with the first layer, which dependent on the rate of formation exhibits more or less order;

    152

  2. creation of overlayers in a layer by layer growth;
  3. ordering of the overlayers along preferential directions induced by the crystalline substrate.

The first step takes place for every type of film preparation discussed above, (A) through (C). The second step, where the molecules are oriented parallel to the (0001) plane of the HOPG, occurs certainly for films prepared by solution casting (B and C). The third step takes place only using a slow casting (C). Note that in both methods (B) and (C), the molecular coverage obtained, due to the dewetting process, is not homogeneous over the macroscopic size (mm2) of the sample. It turned out that these variations over large distances may be controlled to a certain extent by using the slow deposition procedure (C).

The surface compositions of the films have been checked routinely making use of X-ray photoelectron spectroscopy which revealed an overall purity of the adsorbate with the exception of a quantity of residual oxygen limited to a few percents.

UPS allowed to obtain information concerning the order of the adsorbate on a molecular scale. Focusing on the region of the spectra with a binding energy lower then 5 eV, namely the peaks that have been assigned to the pi states of the HBC-C12 [Kei00], it is possible to gain insight into the orientation of the molecules with respect to the substrate.

From the graphs in Fig. 6.75 it is evident that the pi-peaks are growing with decreasing deposition rates (moving from Fig. 6.75a to Fig. 6.75c). The increasing of the thickness of the adsorbate from Fig. 6.75a to Fig. 6.75b is also accompanied by a growth of the pi-peaks due to the higher number of conjugated discs that get irradiated by the photon beam. In addition for the case of Figs. 6.75b and 6.75c thermal annealing carried out for 1 hour at 150 °C led to an increase of the pi peak. Furthermore a preliminary experiment of angle resolved - UPS on slowly coated (C) HBC-C12 have been carried out both with the irradiating light normal to the substrate and at 45° from it (Fig. 6.76). From the intensity of the pi peaks, it is possible to establish that the molecular discs tend to lie flat on HOPG.


153

Figure 6.75: UPS HeII spectra of HBC-C12 plotted relative to the Fermi level of samples prepared at decreasing rates of physisorption: a) casting 1 layer; b) casting ~ 1000 layers; c) slow casting of 1 layer.


154

Figure 6.76: Angle Resolved UPS He(II) spectra of HBC-C12 plotted relative to the Fermi level of sample prepared by slow casting 1 layer. a) sample as prepared; b) thermally annealed sample at 200°C for 2 hours.

Figure 6.77: Scheme of the packing of HBC-C12 into columnar arrays (from [Her96]).

The stacking of several layers is likely to give rise to a columnar structure as the one shown in Fig. 6.77. This order can be further increased by making use of thermal annealing for 2 hours at 200 °C, as shown in Fig. 6.76. This result is in good agreement with the mesophase detected between 60 and 399 °C where the molecules tend to stack in columnar aggregates like in Fig. 6.77 [Her96].


155

In summary, by changing the rate of the self-assembly from solution of the HBC-C12 it was possible to produce well defined and epitaxially oriented layers. Both SFM and Angle resolved UPS spectroscopy suggest that HBC-C12 are lying flat on the HOPG substrates. These results strongly suggest that the growth into layered architectures is a kinetically governed phenomenon which leads to an hetero-epitaxial ordering of the organic interface.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Wed Aug 21 18:53:25 2002