Imaging of Petroleum Asphaltenes Using Scanning Tunneling

---

2358

Ind. Eng. Chem. Res. 1994,33, 2358-2363

Imaging of Petroleum Asphaltenes Using Scanning Tunneling Microscopy Brian A. Watson and Mark A. Barteau' Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716

Scanning tunneling microscopy (STM)was used to image Ratawi (neutral zone) asphaltenes deposited on highly oriented pyrolytic graphite (HOPG).STM images revealed the formation of ordered arrays covering hundreds of angstroms of the surface. Samples deposited from pyridine solutions above and below the critical micelle concentration were quite similar. The individual features imaged in these arrays were attributed to the termini of the pendant aliphatic chains known to be a part of the asphaltene molecular structure. The ordered arrays imaged were extremely flat and uniform, in contrast to the assumed molecular complexity and the absence of unique structure of asphaltenes. The formation of these ordered surface arrays is further evidence of the tendency of asphaltenes to self-associate.

Introduction Asphaltenes are one of the most studied fractions of crude oil (Herzog, 1988; Pfeiffer and Saal, 1940; Ravey, 1988; Ray et al., 1957; Sheu et al., 1991, 1992; Speight, 1991; Storm et ai., 1991; Winniford, 1963). By definition asphaltenes are simply a solubility class: the heptaneinsoluble fraction of the vacuum residue obtained from the crude oil during refining. Asphaltenes are dark brown to black friable solids with no definite melting point. The chemical structure of asphaltenes is extremely complicated, and isolation and identification of individual asphaltene molecules has not proven possible. Asphaltenes do, however, possess several common chemical and geometric characteristics. All asphaltenes are primarily composed of carbon, hydrogen, oxygen, nitrogen, and sulfur (Speight, 1991). In general, asphaltenes are believed to be polynuclear aromatic ring systems bearing alkyl side chains. The condensed ring systems vary in size from 6 to 20 rings and the alkyl chain lengths between 4 and 20 carbon atoms (Hasan et al., 1989; Mojelsky et al., 1992; Payzant et al., 1991; Storm et al., 1993). The apparent high molecular weight of asphaltenes (2000-10 000) is attributed to the linking together of the aromatic sheets with alkyl chains (Speight, 1991). One of the most striking similarities between different asphaltene samples is the near constancy of the ratio of hydrogen to carbon atoms (H/C= 1.15 f 0.5, C = 32 f 3% and H = 8.1 f 0.7% by weight) (Speight, 1991). The constancy of this ratio has led to a general belief that asphaltenes do have a characteristic chemical composition and structure, which may permit a more meaningful description than that of just a solubility class. Another common property of asphaltenes is their tendency to associate in solution. This tendency has been regarded as a key factor hindering the conversion of heavy vacuum residue into more valuable, lighter products. In addition, this self-association has led to uncertainty in the determination of the molecular weight of asphaltenes. Techniques using an ultracentrifuge give asphaltene molecular weights up to 300 000 (Speight, 19911, whereas the osmotic pressure method gives molecular weights up to 80 000 (Speight, 1991). Even lower molecular weights (1000-5000) are obtained using vapor pressure osmometry (Speight, 1991). These discrepancies can likely be at-

* To whom correspondence should be addressed. 0888-5885/94/2633-2358$O4.5O/O

tributed to the varying degrees of asphaltene aggregation that occur during these different weight determinations. Attempts to understand asphaltene self-associationhave been the subject of several studies (Pfeiffer and Saal, 1940; Ray et al., 1957; Sheu et al., 1991,1992; Storm et al., 1991; Winniford, 1963). Using surface tension measurements and rheological data (Sheu et al., 1991,1992; Storm et al., 1991) it has been shown that asphaltenes exhibit many properties seen in traditional colloids, such as the existence of a critical micelle concentration (cmc), the concentration at which aggregates suddenly begin to form. Structurally, however, asphaltene aggregates are quite different from traditional colloid constituents, e.g., cadmium arachidate, which possess hydrophobic tails and hydrophilic head groups that facilitate the formation of colloidal structures such as micelles, bilayers, and vesicles. The size and shape of colloidal asphaltene particles has been investigated using small angle neutron and X-ray scattering. For example Sheu et al. (1991) have shown that Ratawi asphaltenes form spherical aggregates having diameters between 60 and 66 A in toluene and pyridine solutions. Unlike micelles, the Ratawi aggregates did not increase in size with increasing asphaltene concentration. Ravey et al. (1988) studied several asphaltenes derived from Middle East crude oils and found that sheetlike particles formed in tetrahydrofuran solution; these were characterized by diameters between 60 and 200 A with mean thicknesses of 6-8 A. The size of the particles depended on the solvent, but not on the asphaltene sample used. Herzog et al. (1988) examined Pematung and Grenada asphaltene samples in benzene; these formed disks of thickness 3.4 A and diameters of 26-1600 A. We report here an investigation of the molecular and aggregate structure of asphaltenes deposited on graphite. Ratawi (neutral zone-Kuwait/Saudi ArabidIraq border) asphaltenes were deposited on graphite and imaged using scanning tunneling microscopy (STM). Periodic structures on the sub-nanometer scale in these asphaltene images are attributed to intramolecular features rather than to discrete molecules. STM is a powerful surface science tool that permits the real-space imaging of surfaces and adsorbates on surfaces. Thus, STM is an ideal tool for directly determining the configuration of adsorbates on solid surfaces at the molecular level (Chiang et al., 1988; Hallmark et al., 1991; Lippel et al., 1989; Ludwig et al., 1992; McGonigal et al., 1990; Rabe and Buchholz, 1991; Smith et al., 1990; Virtanen, 1991). For example, in 0 1994 American Chemical Society

Ind. Eng. Chem. Res., Vol. 33, No. 10, 1994 2359 Table 1. Compositional Data for Ratawi Asphaltenes (from Storm et al., 1990,1991)

c (%I

82.7 Ni (ppm) 8.4 V (ppm) N (%) 0.9 H/C s (5%) 7.7 molecular weighto VPO measurement in toluene at 50 OC.

H (%I

14.5 308 1.22 2360

ultrahigh vacuum the adsorption geometry of aromatic hydrocarbons such as benzene coadsorbed with carbon monoxide on Rh(ll1) (Chianget al.,1988),perylene-3,4,9,10-tetracarboxylicdianhydride on graphite (Ludwig et al., 1992),and naphthalene on Pt(ll1) (Hallmark et al., 1991) have been determined from molecularly resolved STM images. In the laboratory environment the adsorption geometries on graphite of alkylcyanobiphenyl (Smith et al., 1990), didodecylbenzene (Rabe and Buchholz, 1991), and n-alkanes (McGonigal et al., 1990) have been determined as well using STM. In addition to providing direct observations of the adsorption geometries of molecules, STM can also image discrete moieties within more complicated molecules. For example, STM images of cadmium arachidate deposited on graphite using the Langmuir-Blodgett technique show the individual carbon atoms in the terminal C-C bond of the hydrophobic alkylchain of eachmolecule (Fuchs, 1988; Mizutani et al., 1988). Images of alkylcyanobiphenyl on graphite show resolution of the alkyl chain from the benzene ring in the molecule. STM images of naphthalene on Pt(ll1)resolve each aromatic ring making up the fused molecule (Hallmark et al., 1991).

Experimental Section STM imagingwas carried out in air a t room temperature using the LK lo00 instrument (LK Technologies, Bloomington, IN) described previously (Watson et al., 1992). Mechanically cut Pt tips were used. Unless otherwise noted, images were recorded in the constant current mode at a tunneling current of 1nA and a sample bias of +50 mV with respect to the tip. I-Vspectroscopy was carried out as described previously (Watson et al., 1992). Image 4.0 Software (National Institutes of Health) was used to obtain fast Fourier transforms (FFTs) of the images. FFTs were used to determine the periodicities and angular orientations of the features seen in the images. Ratawi (neutral zone) asphaltenes were used in these studies. Further details of the extraction of the asphaltene fraction from the vacuum residuum can be found elsewhere (Sheu et al., 1991). A summary of the compositional data for the Ratawi asphaltenes from Storm et al. (1990,1991) is given in Table 1. Other characterization details for these samples, including rheological and colloidal properties, are given in Storm et al. (1990,1991,1992,1993)and Sheu et al. (1991,1992). Two different solutions were prepared by dissolving the asphaltene sample in pyridine at concentrations above (5.1 X 1 0 - 2 wt %) and below (1.0 X 10-2 wt %) the critical micelle concentration (cmc). Asphaltene samples were prepared for STM imaging by depositing 10 pL of solution on freshly cleaved graphite and evaporating to dryness.

Results Typical STM images of the asphaltenes are shown in Figure 1. Postimage processing of the images in Figure 1 included low pass filtering to remove high-frequency noise. Histogram equalization was also performed on the image in Figure l a to enhance contrast. The most striking

feature of the images was the observation of ordered arrays of ellipsoidal corrugations which formed a monoclinic unit cell having dimensions 4.2 8, X 4.6 8, (0 = 6 5 O ) for the sample deposited from the low-concentration (LC, below the cmc) solution and 4.7 A X 6.0 A (8 = 5 6 O ) in the case of the high-concentration (HC, above the cmc) samples. Such periodicities were seen over lateral dimensions of more than 500 A as discussed below. The respective unit cells are illustrated in Figure 2. In the HC asphaltene STM image additional fine structure is seen within each ellipsoid. This fine structure is highlighted in the line profile (Figure 3) which was taken between points labeled A and B on the STM image in Figure la. The spacing between the peaks within each doublet in the line profile was determined to have an average distance of 2.1 A. This value was calculated by determining the doublet spacing taken from multiple line profiles of rows in several images. In addition to being well-ordered in the x-y plane, the asphaltenes appear to be rather uniformly packed in the z-direction. In STM images scanned over a 500 A X 500 A area, the z-range varied by at most 5.3 A. It should also be noted that when imaging the LC sample, portions of the surface were adsorbate free, giving images consistent with those of bare HOPG. No images showing discrete adsorbates and bare graphite in the same scan area were seen, and thus it was not possible to determine the sizes of adsorbate-covered domains. The average periodicity and orientation of the features seen in the images of both the LC and HC samples were quite similar. No fine structure, however, was resolved in the LC asphaltene STM images. This may simply be the result of STM tip quality. That the two samples are so similar is not surprising. Both samples were deposited by solvent evaporation. As evaporationoccurs,the asphaltene concentration in the LC sample will exceed the cmc and aggregate formation would be expected to take place. Although the surface population of asphaltene molecules would obviously be greater for the HC sample, the aggregate size and morphology would be expected to be similar in the two samples based on SANS data which showed that, above the cmc, the aggregate size was concentration independent (Sheu et al., 1991). Images of the HC and LC samples revealed the same geometric information in adsorbate-covered regions; however, it was much more common when positioning the tip atop different sites on the LC sample to image adsorbate-free areas. To verify the integrity of the apparent length scales in the micrographs, images of HOPG were routinely recorded. The HOPG images were inspected to ensure that a periodicity of ca. 2.5 8,was observed. In addition, samples prepared with neat pyridine were imaged as well to ascertain what, if any, contribution the solvent made to the images of the adsorbate-coveredsurfaces. STM images of samples prepared by depositing and evaporating the neat solvent on HOPG were consistent with images of bare graphite, indicating that pyridine was not present on the surface after the droplet had evaporated from the HOPG surface. Tunneling spectra characteristic of bare HOPG and of the adsorbate covered surfaces are shown in Figure 4. The I-V spectrum of the adsorbate-covered surface is representative of the spectra taken on both the LC and HC samples. The appearance of a plateau region in the asphaltene I-V spectrum in Figure 4 indicates that the overlayer is less conductive than the graphite substrate, as would be expected. These spectra also indicate that the images of the adsorbate-covered surfaces are the result

2360 Ind. Eng. Chem. Res., Vol. 33, No. 10,1994

Figure 1. (a, top left) STM Image of HC asphaltene sample. Image area is 58 A X 58 A. (b, top right) STM Image of HC asphaltene sample. Image area is 29A X 29 A. (c, bottom left) STM Image of LC asphaltene sample. Image area is 232 A X 232 A. (d, bottom right) STM Image of LC asphaltene sample. Image area is 116 A X 116 A. 4.7

A

2.1

A

A A E Figure 2. Unit cells of (a) HC sample and (b) LC sample as determined from analysis of FFT's of images.

of the molecular overlayers and not a graphite anomaly, e.g., a Moire pattern. In our experience artifacts in images of bare graphite still give rise to tunneling spectra indistinguishable from those of the well-defined graphite surfaces (Johnson, 1992).

Figure 3. Line profile obtained from Figure l a between p i n t a A and B located on the image. The r-dimension has been expanded to highlight the fine structure in the profile.

Discussion These studies illustrate the use of scanning tunneling microscopy to directly image petroleum asphaltenes at the molecular level. The periodicity seen in STM images

Ind. Eng. Chem. Res., Vol. 33, No. 10, 1994 2361 1 2 0 t

t

e d

f

4

4

f 4 +

+

-804 -120

I

’

- 2

F4

CS

-1

0

,

,

v (volts)

,

,

,

1

,

, ,

I

c*

CS

C4 C4

2

Figure 4. I-V spectra of bare and asphaltene-covered graphite surfaces. The solid line is the I-V spectrum of bare graphite. The dotted l i e is the I-V spectrum of the HC sample; however it is representative of spectra obtained for LC samples as well. The I-V spectra clearly differentiate between the bare substrate and the adsorbate-covered surface. The I-V spectrum of the HC sample indicates that it is less conductive than graphite.

a b Figure 6. “Singlemolecule” representations of Ratawi asphaltenes based on (a) liquid state NMR and (b) solid state NMR (Storm et al., 1992, 1993). Aliphatic Aromatic Sheets

I

I-

t

* 6-lSA

Figure 7. Representation of an asphaltene cluster in the solid state (Speight, 1991).

Figure6. Hypothetical representation of asphaltenes extracted from Iraqi crude oil (Speight, 1980).

of Ratawi asphaltenes is quite remarkable given their assumed structural heterogeneity. A typical asphaltene sample is believed to contain over los different molecules. On the other hand, asphaltenes are known to self-assemble into discrete colloidalparticles. As mentioned above, small angle neutron scattering has shown that, in polar solvents, Ratawi 88 haltenes form spherical particles having a radius of 30-32 (Sheu et al., 1991). The periodicity seen in the STM images is consistent with the aggregation phenomenon. The question remains as to what structural features are being imaged. Attempts have been made previously to assign discrete molecular structures that are consistent with nuclear magnetic resonance spectra, mass spectrometry, and elemental analysis of asphaltene samples. For

1

example, Figure 5 (Speight, 1980) shows a hypothetical structure for asphaltenes extracted from Iraqi crude oil that is consistent with such analyses of those samples; Figure 6 (Storm et al., 1992)shows hypothetical structures of Ratawi asphaltenes constructed to explain the results of liquid phase and solid state NMR measurements. It should be stressed that these ”molecules” are simply statistical representations of the chemical analyses. Such structures serve to highlight common constituents that make up an asphaltene, but are not necessarily common molecules that would be found in the crude oil. X-ray scattering, as mentioned above, provides insight into the macromolecular structure of asphaltenes in solution. The structure in Figure 7 (Speight, 1991), illustrating the association between molecules, was constructed to explain results of yet another technique, X-ray diffraction, from asphaltene clusters in a solid sample. From the average structures illustrated in Figures 5 and 6, it can be seen that the two most common building blocks found in chemical analyses of asphaltenes are fused aromatic rings and aliphatic or alkyl chains. Thus the two most likely candidates to explain the features imaged by STM are fused aromatic rings and pendant alkyl chains. In either case these structures could be oriented parallel or perpendicular to the graphite substrate. We discount the possibility that the features in the images of Figure 1 are the aromatic sheets lying parallel to graphite. If indeed individual sheets were being imaged, one might attribute each feature in the image to the ?r-electron cloud associated with each aromatic ring, or to individual carbon atoms within the sheets. However, the periodicities obtained from the FFTs are not consistent with either of these models. Also, sheet boundaries should appear as defects in these images; none were observed. A second possibility is that the edges of the sheets are being imaged. Again this model is discounted since the

2362

Ind. Eng. Chem. Res., Vol. 33, No. 10, 1994

presence of pendant alkyl groups would disrupt the longrange order seen in the STM images. A third possibility is that the pendant alkyl chains lying parallel to the graphite substrate are being imaged. In this case the finite length of the chains should be observable; however, our images show long-range, uninterrupted periodicity over hundreds of angstroms. Ruling out the above possibilities, we assign the features in the STM images of the asphaltenes to the pendant alkyl chains oriented outward from the substrate. Support of this hypothesis comes from dielectric relaxation and conductivity measurements which showed that Ratawi asphaltenes are driven to self-associate via electrontransfer sites such as heteroatoms; however, such sites are not present on the surface of the aggregates (Sheu et al., 1991). Heteroatoms are most likely incorporated into the aromatic sheets of the asphaltenes. Thus a likely mode of aggregation is via the formation of lamellar stacks of aromatic sheets held together by the electron-transfer sites (heteroatoms) and the van der Waals attraction between the x-systems of the aromatic rings. Extended structures formed by stacking of these aromatic sheets would expose at their surfaces the alkyl chains pendant from individual sheets. That the size of colloidal particles formed in solution is concentration independent is assumed to be a consequence of the lack of electron-transfer capability of the alkyl groups (Sheu et al., 1991). Alkyl chains are therefore likely to be present at the surface of the aggregates and therefore accessible to the tip in STM imaging. From the line profile in Figure 3 it is observed that the spacing of the fine structure is ca. 2.0 A, which is on the order of the dimensions of a C-C bond. Thus we attribute these features in the images to the terminal two carbons of pendant alkyl chains. Such an interpretation is not unlike that attributed to STM and atomic force microscopy (AFM) images of Langmuir-Blodgett (LB) films of cadmium arachidate (Fuchs, 1988; Meyer et al., 1991; Mizutani et al., 1988; Schwartz et al., 1992). When cadmium arachidate (CH&H2)&OOCd) mono- and multilayers are deposited on an appropriate substrate, e.g., mica or silicon, using the Langmuir-Blodgett technique the molecules form two-dimensional arrays in which the alkyl chains are aligned perpendicular to the surface. Using AFM or STM the termini of the alkyl chains on the surface of the LB films are imaged. Using AFM, four-layer cadmium arachidate films have been shown to form monoclinic or orthorhombic ordered arrays having periodicities of 5.2 A X 4.7 A and a packing density (molecular area) of 24 A2 (Meyer et al., 1991). Schwartz et al. (1992)prepared LB mono- and multilayers of cadmium arachidate that formed noncentered rectangular two-molecule unit cells having packing densities of 18 A2. The packing densities seen in the images of the asphaltenes described here were 18 A2for the LC sample and 23 A2 for the HC sample. The packing densities for the asphaltenes are therefore consistent with those reported for LB films characterized using AFM. In addition, the asphaltene packing densities are in agreement with those reported for other close-packed aliphatic systems (Bohm et al., 1989; Bonnerot et al., 1985; Tippmann-Krayer et al., 1992). If indeed the pendant alkyl chains of the asphaltenes are being imaged, one must still address the question of the tunneling mechanism through these overlayers. Several mechanisms, such as resonant tunneling, have been proposed to explain electron tunneling through organic overlayers which are expected to have low conductivity

(Chiang, 1992). However, a generally accepted theory is still lacking. In spite of this, there exist many examples of STM images of organic adsorbates deposited on conductive substrates. For example, vanadylnaphthalocyanine aggregates which form ordered arrays of molecular stacks on HOPG have been imaged by STM (Manivannan et al., 1993). Polydiacetylene monolayers have also been imaged by STM (Wilson et al., 1992). In these samples the hydrocarbon chains of the polymer were shown to be oriented in a tilted geometry relative to the substrate. Long chain hydrocarbon monolayers oriented perpendicular to the substrate that have been successfullyimaged by STM include cadmium arachidate (Fuchs, 19881, w-tricosenoic acid (Braun et al., 1988; Fuchs, 1988), 12,8diynoic acid (Braun et al., 1988), and dimyristoylphosphatidic acid (Horber et al., 1988). In all cases molecular resolution images were reported. Several studies, however, have cast doubt on the interpretation of hydrocarbon overlayers whose alkyl chains exceeded lengths of ca. 20 A, such as cadmium arachidate films deposited on gold (Hallmark et al., 1987; Ulman, 1991). In the study by Hallmark et al. (1987),for example, STM images of bare gold surfaces and gold surfaces covered with LB films of cadmium arachidate were compared. The STM imagesof the different samples were indistinguishable, showing the surfaces to be covered with an ordered, close packed array of features that covered hundreds of angstroms of the surface. Thus the features in these images were not attributed to the LB film. Nevertheless, smaller chain lengths (