Study of Petroleum Heat-exchanger Deposits with ATR-FTIR

---

Energy & Fuels 2009, 23, 4059–4067

4059

Study of Petroleum Heat-exchanger Deposits with ATR-FTIR Spectroscopic Imaging Feng H. Tay and Sergei G. Kazarian* Department of Chemical Engineering, Imperial College London, SW7 2AZ, UK ReceiVed April 12, 2009. ReVised Manuscript ReceiVed June 9, 2009

Fouling in heat exchangers is a major economic problem for the crude oil industry. Despite significant research in this area, the fundamentals of the complex fouling process are not fully understood. There are many analytical techniques that are useful in the characterization of such deposits; however, each has its own limitations and drawbacks. Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopic imaging has the advantage of being a nondestructive analytical technique with minimal sample preparation and, most importantly, is able to provide both chemical and spatial information about a sample. This paper introduces novel applications combining macro- and micro-ATR modes in FTIR imaging to characterize real deposits from a heat exchanger of a crude oil refinery. A lab-made aperture was utilized in macro-ATR imaging to correct the distortion of spectral bands that occurs for materials with a high refractive index, such as petroleum deposits. Using different ATR accessories for the macro (Golden Gate with a diamond crystal) and micro (germanium coupled to an infrared microscope) modes, FTIR imaging, with fields of view of 610 × 530 µm2 and 63 × 63 µm2, respectively, yields important information about the spatial distribution of different components in the deposits. Macro-ATR-FTIR imaging revealed clusters of chemically different compounds such as asphaltenes, carbonates, sulfates, sulfoxides, oxalates, and possibly coke. It was also demonstrated that with the enhanced spatial resolution of the micro-ATR approach, a more representative spectrum of the components can be achieved. Different forms of the oxalate functional group, one of which was barely at the detection limit of the macro-ATR approach, were also identified using the micro-ATR method. The sensitivity of the two modes of measurement with different spatial resolutions allows spatial and chemical information to be obtained and analyzed for domains of different sizes in studied samples.

Introduction Crude oil distillation accounts for a large fraction of the energy used in oil refining. Crude oil is normally stored at ambient conditions and heated to 110-150 °C before entering the desalter, where inorganic ionic species such as chlorides, fluorides, sodium, etc. are removed. Particulate matter and sludge (sand, corrosion products, etc.) are also removed in the desalter, although in an operational refinery this is not often the case. Crude oil downstream of the desalter will be heated further to 230-300 °C in the preheat train before it enters the furnace. The integration of the heat exchanger network helps to reduce utility costs of the process, but crude oil is a mixture of substances and some tend to deposit as fouling layers in the heat exchanger units. This fouling layer, as it builds up, decreases the heat flux reaching the process stream; thus, more energy input is required. In a typical (100 000 barrel day-1) unit, a loss of only 1 °C in inlet temperature to the furnace equates to 450 kW and would result in a cost of US $40 000 per annum in additional fuel charges.1 Baudelet and Krueger estimated that an increase of furnace inlet temperature of 1 °C is equivalent to 1 ton of fuel saved per day.2 More fuel (energy) is required for the furnace to bring the process stream to its desired temperature, and the use of more fuels leads increases * To whom correspondence should be addressed. (1) Panchal, C. B.; Huangfu, E. P. Heat Transfer Engineering 2000, 21 (3), 3–9. (2) Baudelet, C. A.; Krueger, A. W. A Practical Approach to Fouling Mitigation in Refinery Units: The Spirelf, 1st ed.; AIChE Spring National Meeting: Houston, TX, 1999.

in the produced amount of CO2 in the process. As the fouling layers accumulate, additional power for the compressors will be needed to overcome the increased pressure drop of the units. Fouling can become serious for a particular batch of crude oil, and a partial shutdown or even a full shutdown of the plant may be needed. In general, fouling in heat exchangers is a major economic problem. Muller-Steinhagen, in 1995, estimated that the total cost of all heat exchanger fouling in the UK is of the order of US $2.5 billion, and the equivalent cost in America is US $15 billion.3 The magnitude and significance of crude oil fouling have led to a number of studies, but the fundamentals of the complex fouling process are not fully understood.4-7 The influences of flow velocity, bulk and surface temperatures, and particulate concentrations on the fouling rate and deposit thickness have been discussed and compared to simple models and correlations, but the detailed chemical characterization of the deposits is still lacking.4,7 Asphaltenes have been closely linked and related to the level of fouling in heat exchangers.4,8 They are defined operationally as a solubility class of crude oil that is insoluble in n-heptane and soluble in light aromatics such as toluene. These materials (3) Muller-Steinhagen, H. Chem. Ind. 1995, (5), 171–175. (4) Watkinson, A. P. Heat Transfer Eng. 2007, 28 (3), 177–184. (5) Takemoto, T.; Crittenden, B. D.; Kolaczkowski, S. T. Chem. Eng. Res. Design 1999, 77 (A8), 769–778. (6) Sheikh, A. K.; Zubair, S. M.; Haq, M. U.; Budair, M. O. J. Energy Res. Technol. 1996, 118 (4), 306–312. (7) Crittenden, B. D.; Kolaczkowski, S. T.; Downey, I. L. Chem. Eng. Res. Design 1992, 70 (A6), 547–557. (8) Rocha, L. C.; Ferreira, M. S.; Ramos, A. C. D. J. Pet. Sci. Eng. 2006, 51 (1-2), 26–36.

10.1021/ef900304v CCC: $40.75  2009 American Chemical Society Published on Web 07/13/2009

4060

Energy & Fuels, Vol. 23, 2009

are typically complex mixtures with high contents of polycyclic aromatic hydrocarbons (PAHs) and with high heteroatom content (N, O, and S). Asphaltenes may be one of the main constituents of the foulant, but in a complex process stream such as crude oil, fouling may be associated with suspended impurities, insoluble gum, inorganic species, and coke.9 There are many analytical techniques that are useful in the characterization of petroleum derived deposits; however, each has its own limitations and drawbacks.10-12 Infrared (IR) spectroscopy has been one of the most versatile techniques in material characterization. One of the earliest applications of IR spectroscopy to study the structure of petroleum asphaltenes was in 1962. Many researchers since then have used different approaches of IR spectroscopy to obtain chemical information to improve the understanding of petroleum fuels,13,14 petroleum asphaltenes, and resins.15-22 The hydrogen bonding capabilities of asphaltenes with phenol (OH group) and pyperidine (NH group) were studied with Fourier transform IR (FTIR) spectroscopy.23 It was shown that asphaltenes with high oxygen and low nitrogen contents have poor interaction with phenols, which indicates that oxygen might be incorporated as acidic hydroxyl groups in asphaltenes. FTIR spectroscopy was also used to monitor chemical changes to the asphaltenes under different treatments. For example, Huang demonstrated that the thermal decomposition of polycyclic asphaltenes only occurs in the temperature range of 450-650 °C.24 In methylation and trimethyl silylation studies on asphaltenes using FTIR spectroscopy, it was observed that the asphaltenes most susceptible to these reaction chemistries were also the most stable asphaltenes in crude oil.25 When asphaltenes are photo-oxidized, an overall increase in carbonyl groups and a decrease in aliphaticity of the molecules were observed from the FTIR spectra.26 It was reported by another researcher that sulfates were produced during the photo-oxidation process.27 The use of a focal plane array (FPA) detector allows the gathering of thousands of IR spectra simultaneously, each from a specific location in the measured area. FTIR spectroscopic (9) Mullins, O. C.; Sheu, E. Y.; Hammami, A.; Marshall, A. G., Asphaltenes, HeaVy Oils and Petroleomics; Springer: New York, 2007. (10) Herod, A. A.; Bartle, K. D.; Kandiyoti, R. Energy Fuels 2008, 22 (6), 4312–4317. (11) Klein, G. C.; Kim, S.; Rodgers, R. P.; Marshall, A. G.; Yen, A. Energy Fuels 2006, 20 (5), 1973–1979. (12) Speight, J. G. Appl. Spectrosc. ReV. 1994, 29 (3-4), 269–307. (13) Castro, L. V.; Vazquez, F. Energy Fuels 2009, 23, 1603–1609. (14) Cooper, J. B.; Wise, K. L.; Welch, W. T.; Sumner, M. B.; Wilt, B. K.; Bledsoe, R. R. Appl. Spectrosc. 1997, 51, 1613–1620. (15) Buenrostro-Gonzalez, E.; Espinosa-Pena, M.; Andersen, S. I.; LiraGaleana, C. Pet. Sci. Technol. 2001, 19 (3-4), 299–316. (16) Calemma, V.; Iwanski, P.; Nali, M.; Scotti, R.; Montanari, L. Energy Fuels 1995, 9 (2), 225–230. (17) Christy, A. A.; Dahl, B.; Kvalheim, O. M. Fuel 1989, 68 (4), 430– 435. (18) Coelho, R. R.; Hovell, I. Pet. Sci. Technol. 2007, 25 (1-2), 41– 54. (19) Coelho, R. R.; Hovell, I.; Monte, M. B. D.; Middea, A.; de Souza, A. L. Fuel Process. Technol. 2006, 87 (4), 325–333. (20) Douda, J.; Llanos, M. E.; Alvarez, R.; Bolanos, J. N. Energy Fuels 2004, 18 (3), 736–742. (21) Elsharkawy, A. M.; Al-Sahhaf, T. A.; Fahim, M. A. Pet. Sci. Technol. 2008, 26 (2), 153–169. (22) Khadim, M. A.; Sarbar, M. A. J. Pet. Sci. Eng. 1999, 23 (3-4), 213–221. (23) Siddiqui, M. N. Pet. Sci. Technol. 2003, 21 (9-10), 1601–1615. (24) Huang, J. Pet. Sci. Technol. 2006, 24 (9), 1089–1095. (25) Juyal, P.; Merino-Garcia, D.; Andersen, S. I. Energy Fuels 2005, 19 (4), 1272–1281. (26) Boukir, A.; Guiliano, M.; Doumenq, P.; El Hallaoui, A.; Mille, G. Comptes Rendus De L Academie Des Sciences Serie Ii Fascicule C-Chimie 1998, 1 (10), 597–602. (27) Huang, J.; Yuro, R.; Romeo, G. A. Fuel Sci. Technol. Int. 1995, 13 (9), 1121–1134.

Tay and Kazarian

imaging has been shown, in recent years, to be a powerful tool in characterizing heterogeneous materials with its enhanced spatial resolution. FTIR imaging is already established in other fields of research such as polymers,28 drug release,29 biomaterials,30 art conservation,31 and forensic science.32 This paper presents the application of FTIR spectroscopic imaging with an FPA detector to characterize petroleum-derived deposits with the aim that this new methodology would be capable of providing new insight into the heat-exchanger deposits with the ultimate goal of contributing to the understanding of the fouling mechanism in crude oil distillation. The attenuated total reflection (ATR) mode, one of the sampling approaches in FTIR imaging, was used in this study. An evanescent wave, which is the result of total internal reflection of the incident wave at the surface of the higher refractive index ATR crystal, penetrates into the sample to a depth of several micrometers. The depth of penetration of this evanescent wave depends on several parameters; refractive index of the ATR crystal and sample, the angle of incidence, and its wavelength. The material in close contact with the reflecting surface selectively absorbs the radiation, and the resulting attenuated radiation is collected and measured. The small penetration depth of the ATR approach makes it a convenient sampling method with little or no sample preparation and can be applied to highly absorbing materials such as carbonaceous hydrocarbons. For solid samples, polishing of the samples is needed for IR measurements in specular reflection mode. In transmission mode, a KBr pellet will have to be prepared. Carboneous materials are often highly absorbing materials in the IR region, thus the particles will have to be ground down to a few micrometers before IR radiation can pass through. For a particular system using the correct setup, ATR-FTIR spectroscopic imaging has the advantage of being a nondestructive approach with spectra measured from a reproducible sample thickness and is able to capture a chemical image of the sample. Chemical maps of petroleum deposits had been reported before using FTIR spectroscopy with a synchrotron source measured in transmission mode.33 The synchrotron source is ∼1000 times brighter compared to an ordinary thermal source, thus in microscopy the IR beam can be masked into a 5 µm spot using an aperture without impairing the quality of the spectrum. However, this facility is expensive and not widely available. Also, the chemical map was generated by raster scanning and it is slower compared to a chemical image obtained using an FPA detector.34 ATR-FTIR imaging with a FPA detector before has been compared to FTIR microscopy in transmission with a synchrotron source, and it was concluded that although the signal-to-noise ratio (SNR) is much better when using the synchrotron, the spatial resolution is higher with micro-ATRFTIR imaging due to the use of the ATR objective with a high refractive index germanium crystal.34 The ATR approach has its limitations as the refractive index of the sample must be low enough such that the angle of incidence of the IR source is (28) Koenig, J. L., Microspectroscopic Imaging of Polymers, 1st ed.; American Chemical Society: 1998; p 411. (29) Kazarian, S. G.; Chan, K. L. A. Macromolecules 2003, 36, 9866– 9872. (30) Kazarian, S. G.; Chan, K. L. A. Biochim. Biophys. Acta, Biomembranes 2006, 1758 (7), 858–867. (31) Spring, M.; Ricci, C.; Peggie, D. A.; Kazarian, S. G. Anal. Bioanal. Chem. 2008, 392 (1-2), 37–45. (32) Ricci, C.; Bleay, S.; Kazarian, S. G. Anal. Chem. 2007, 79 (15), 5771–5776. (33) Chouparova, E.; Lanzirotti, A.; Feng, H.; Jones, K. W.; Marinkovic, N.; Whitson, C.; Philp, P. Energy Fuels 2004, 18 (4), 1199–1212. (34) Chan, K. L. A.; Kazarian, S. G.; Mavraki, A.; Williams, D. R. Appl. Spectrosc. 2005, 59 (2), 149–155.

Study of Petroleum Heat-exchanger Deposits

Figure 1. Schematic diagram showing the IR beam path through (a) the macro-ATR and (b) the micro-ATR crystal. In the macro-ATR accessory, an aperture is inserted to allow beam from the top half of the lens to pass through, which changes the average angle of incidence into the crystal.

well above the critical angle to ensure an artifact-free spectrum. It is shown in this work how a lab-made aperture is incorporated in a commercially available diamond ATR accessory to obtain reliable ATR-FTIR spectra of a high refractive index material, thus addressing the feasibility of using ATR-FTIR imaging to study carboneous materials. The advantages of both the macroand micro-ATR-FTIR imaging approaches to characterize petroleum-derived materials are also discussed. Experimental Section Samples. The heat exchanger deposits were obtained from a preheat train of a distillation column in a refinery with inlet and outlet temperature at around 158 and 166 °C, respectively. The shut down procedure was standard, thus the deposits were only exposed to nitrogen and steam. The deposits were removed from the exchanger and sent to the laboratory. ATR-FTIR Spectroscopic Imaging. Macro-ATR-FTIR images were measured with a continuous scan FTIR spectrometer using a 64 × 64 FPA detector. A diamond ATR accessory (Imaging Golden Gate, Specac, UK) was placed in a large sample compartment. The refractive index of diamond is 2.4. The optics of the accessory utilizes a pair of zinc selenide/germanium lenses that allow IR light to be focused onto the small diamond ATR crystal. The condenser lens was modified with a lab-made aperture to only allow certain angles of incidence of the IR beam to reach the diamond crystal. The aperture used and the alignment of the imaging system have been described before to give an average angle of incidence of the incoming IR source of about 56°, as opposed to 47° without any aperture.35 Spectra were collected from 3800 to 900 cm-1 with a spectral resolution of 8 cm-1. Two hundred coadditions were acquired for spectral averaging to give a reasonable SNR, and the measurement took less than 5 min. The ATR-FTIR imaging spectrometer is patented by Varian Inc.36 Micro-ATR-FTIR images were measured with a Bio-Rad FTS60A coupled to the UMA 600 IR microscope (Varian, Inc.) using a 64 × 64 FPA detector. The ATR crystal used here is germanium with a refractive index of 4. Spectra were collected from 4000 to 875 cm-1 with a spectral resolution of 8 cm-1 and 256 coadditions. Sampling Technique. For the macro-ATR measurement, the sample was placed on the top surface of the ATR diamond crystal and compressed into a tablet using a compaction cell as described before.37 As the samples are hard, compaction allows them to pack closely so that good optical contact can be obtained between the samples and the ATR crystal. The tablet formed after compaction can be measured directly on the macro-ATR accessory. In the micro-ATR setup, the measurement is taken from the top surface of the sample instead of the bottom surface as shown in (35) Chan, K. L. A.; Tay, F. H.; Poulter, G.; Kazarian, S. G. Appl. Spectrosc. 2008, 62 (10), 1102–1107. (36) Burka, E. M.; Curbelo, R. Imaging ATR spectrometer. U.S. Patent No. 6141100, 2000. (37) van der Weerd, J.; Chan, K. L. A.; Kazarian, S. G. Vib. Spectrosc. 2004, 35 (1-2), 9–13.

Energy & Fuels, Vol. 23, 2009 4061 Figure 1b. The sample was pressed ex situ using a piston pump to create a flat surface for the micro-ATR measurement. Note that the compaction pressure used here is different from the pressure used in the compaction cell for the macro-ATR measurement. The measured area of interest is selected using the microscope, and the stage is raised to bring the sample in contact with the germanium crystal. The depth of penetration for both macro- and micro-ATR setup is about 1.5 µm at 1000 cm-1, assuming the refractive index of the sample is 1.7 and using its respective average angle of incidence of 56° (in macro approach with diamond) and 30° (in micro approach with germanium objective).

Results and Discussion FTIR Spectroscopic Measurements with a Single-element Detector. A real heat exchanger deposit was measured on a diamond ATR accessory with a single-element detector and the ATR-FTIR spectrum was shown in Figure 2a. The FTIR spectrum resembles a typical spectrum of extracted asphaltenes, indicating that the bulk of the deposit is asphaltenic.15,16,20 It also shows that when no aperture was used in the diamond ATR accessory (the average angle of incidence of 47°) distortions such as the sharp decrease in absorbance close to the 3000 cm-1 and an increase in baseline after 2800 cm-1 can be observed. The condition for ATR is that the electromagnetic wave must enter a lower refractive index material (sample) at an angle greater than the critical angle of the system. For ATR measurements made near the critical angle, the spectral bands may become distorted. Strong bands can become broadened and redshifted, both effects increasing as the angle of incidence is decreased. This is due to dispersion in the refractive index as a function of light frequency. Harrick had shown that this shift in spectral band will be small when the angle of incidence is greater than a few degrees above the critical angle.38 Thus, the general rule of thumb for the practical application of internal reflection spectroscopy, that is, ATR spectroscopy, to avoid the distortion of the bands, is to keep measurement well above the critical angle of the system. The refractive index of asphaltenes from the literature has been estimated to be around 1.7,39,40 thus the critical angle for the diamond and germanium ATR crystals can be calculated from eq 1 to give 44.6 and 25.1°, respectively. The actual refractive index of the sample is not measured in this study, but the bulk of the deposits have been shown in Figure 2a to be asphaltenic in nature; thus, a typical refractive index of extracted asphaltenes is used here. An angle of incidence of 47° is very close to the estimated critical angle for the diamond and the deposit sample, thus explaining the observed spectral distortion. critical angle ) sin-1

() n2 n1

(1)

On the other hand, introducing the aperture would only allow the greater angles of the incoming radiation, relative to the normal of the top surface of the diamond, to pass through, thus the average angle of incidence is increased to 56°. The resultant spectrum obtained, shown in Figure 2a, has demonstrated that the distortion is removed under this configuration. This develop(38) Harrick, N. J., Internal Reflection Spectroscopy, Third ed.; Harrick Scientific Corporation: New York, 1987; p 327. (39) Buckley, J. S.; Hirasaki, G. J.; Liu, Y.; Von Drasek, S.; Wang, J. X.; Gil, B. S. Pet. Sci. Technol. 1998, 16 (3-4), 251–285. (40) Wattana, P.; Wojciechowski, D. J.; Bolanos, G.; Fogler, H. S. Pet. Sci. Technol. 2003, 21 (3-4), 591–613.

4062

Energy & Fuels, Vol. 23, 2009

Tay and Kazarian

Figure 2. (a) Single-element FTIR spectra of the real deposit collected using a 47 and 56° aperture. Panels b and c show the distortions of the absorption bands at 3000-2800 and 1700-1250 cm-1, respectively.

ment allows high refractive index materials to be measured with ATR spectroscopy without these optical artifacts, which may affect both qualitative and, to a larger extent, quantitative analysis of the spectrum. It is important to ensure that spectra are free from significant distortions for a reliable spectral interpretation and comparison of spectra obtained from different samples and materials. For example, some IR spectra of extracted asphaltenes have a weak aromatic CH stretching band at 3030 cm-1. The artifact near 3000 cm-1 due to dispersion of the refractive index may affect the spectral assignment of this functional group. The band distortions shown in Figure 2 can also cause errors in quantitative analysis of the IR spectra based on the measured absorbance. Panels b and c of Figure 2 show 4 and 10 cm-1 band shifts in the 3000-2800 and 1700-1250 cm-1 ranges, respectively, for the spectrum acquired with no aperture and with a 56° aperture. Structural parameters based on deconvolution of certain spectral bands can be miscalculated due to these band distortions. The ratio nCH2/mCH3 of aliphatic chains in the samples can be correlated to the absorbance ratios of the bands of the asymmetric stretching vibrations of the methylene and methyl group at 2927 and 2957 cm-1, respectively.22,41-43 (41) Lin, R.; Patrick Ritz, G. Org. Geochem. 1993, 20 (6), 695–706. (42) Painter, P. C.; Snyder, R. W.; Starsinic, M.; Coleman, M. M.; Kuehn, D. W.; Davis, A. Appl. Spectrosc. 1981, 35 (5), 475–485. (43) Sharma, B. K.; Sharma, C. D.; Tyagi, O. S.; Bhagat, S. D.; Erhan, S. Z. Pet. Sci. Technol. 2007, 25 (1-2), 121–139.

Calemma et al. had obtained a linear correlation between the nCH2/mCH3 and I2927/I2957 from the spectra of 20 model compounds to give the following relationship, shown in eq 2 with k ) 1.24316 R)

I2927cm-1 n(CH2) ) ×k m(CH3) I2957cm-1

(2)

The absorbances of the bands of the asymmetric stretching vibrations of the methylene and methyl groups have to be obtained from the deconvolution of the spectrum in this region. With the distortion as evidenced in Figure 2, this ratio can be easily miscalculated. In the fingerprint region, where there may be many overlapping bands, a 10 cm-1 shift of the bands may even result in incorrect assignments of functional groups, especially in samples where the components are not known. If the real deposit is assumed to be asphaltenic in nature, the band at 1635 cm-1 can be assigned to highly conjugated carbonyls such as ketones and/or quinone-type structures and amides. The broad band centered at 1600 cm-1 can be assigned to the stretching mode of the CdC aromatic double bond. The band at about 1450 cm-1 can be assigned to the bending modes of CH2 and CH3, and the band at 1370 cm-1 can be attributed to the symmetrical CH bending mode of methyl groups adjacent to alkyl or aryl groups. The band assignment in the 1370-1000 cm-1 region is complicated, and not every band can be assigned to a specific functional group; this again reiterates the importance

Study of Petroleum Heat-exchanger Deposits

of an artifact-free spectrum. The band at 1150 cm-1 may be assigned to a sulfate group, and the band at 1030 cm-1 can be assigned to a sulfoxide group. Good contact between the sample and the crystal is vital for reliable and reproducible measurements using ATR spectroscopy. For hard samples such as the deposits, much force is needed to press the sample onto the ATR crystal to ensure this contact. However, if the sample is harder than the crystal, this will damage the crystal. Not only does diamond have a hardness of 10 on the Mohs scale, it is chemically inert, which makes cleaning of the ATR surface with most of the solvents possible. The issue with the effect of dispersion of the refractive index on the spectrum could have been overcome with a higher refractive index crystal such as germanium or silicon, but diamond still has the advantage of being hard, durable, chemically inert, and relatively nonabsorbing in most of the midIR region. The modification on the optical design of the diamond ATR accessory has made it possible to obtain reliable spectroscopic information on high refractive index materials, such as asphaltenes and petroleum deposits. Macro-ATR-FTIR Imaging. Figure 3 shows the image obtained using a diamond ATR accessory with the FTIR spectroscopic imaging system. The images shown are from the same sample measured using conventional spectroscopy with a single-element detector. The ATR accessory was moved with the sample intact, after using a single-element detector to the macro chamber of spectrometer where the images are acquired using the FPA detector. Each pixel of the FPA detector measured a full mid-IR spectrum. All images are generated from the data of a single measurement by allocating a color to each pixel based on the integrated absorbance of the particular spectral band as indicated below each image in Figure 3. The red in the scale denotes a high value and blue denotes a low value. This color coding represents the concentration of the component, and this can be quantified with a calibration graph of the known component. The univariate analysis of the FTIR images has revealed several chemically different components, and their representative spectra were extracted from the regions where their concentration is the highest as denoted by the symbol X. The image size of this setup is about 610 × 530 µm2, and it has been shown to have a spatial resolution of 12 µm.35 The image showing the spatial distribution of the absorbance of the asymmetrical methylene stretching mode at about 2920 cm-1 in Figure 3a is relatively homogeneous across the measured area, thus showing that good contact had been established in this setup. Note that as lenses are utilized in the design of the ATR accessory, chromatic aberration may exist in this setup. This system was optimized using the fingerprint region of the IR spectrum where most of the spectroscopic information is present, thus the CHx stretching modes in the higher wavenumber region of 3100-2800 cm-1 may not have as-sharp focus as compared to the fingerprint region. The image based on the integration range of 1470-1420 cm-1 is assigned to the bending modes of CH2 and CH3, but as there are other bands such as the characteristic band for carbonates that appears in the same range, it does not correspond to the image of the stretching modes of CH in the 2940-2880 cm-1 region. Multivariate analysis such as principle component analysis and factor analysis can be applied to such data sets with overlapping spectral features in measured FTIR spectroscopic images, but as the focus of the paper is to introduce ATR-FTIR imaging in characterizing petroleum-derived materials, this advanced analytical technique is not elaborated upon here.

Energy & Fuels, Vol. 23, 2009 4063

The images based on the 1675-1620 and 1350-1300 cm-1 ranges show almost the same distribution in the measured area, therefore implying that these two bands belong to the same component. It was found from a spectral library search that these two bands can be assigned to the oxalate functional group. The concentration of this component may be high enough for the conventional FTIR spectroscopy to detect a band at 1635 cm-1, but the band at 1330 cm-1 is not obvious. As mentioned before, the 1635 cm-1 band can be assigned to different conjugated carbonyls, but it is difficult to assign a specific compound for an unknown sample. From the FTIR image, domains of high concentration of this component can be located, thus a more representative spectrum of the component can be extracted. With the “purer” spectrum, chemical species may be identified and spectral assignment made with a higher confidence. For this material, the oxalate compound is not expected normally and thus is not identifiable if based on just the conventional FTIR spectrum measured with a single-element detector. A broad band in the 1500-1350 cm-1 range was also observed from the thousands of spectra acquired in a single measurement with the FPA detector. The image generated in Figure 3b is based on the integrated absorbance of this band, and it shows the distribution of carbonate compounds in the sample. This characteristic carbonate band overlaps with the band assigned to the bending modes of CH2 and CH3 at 1470-1420 cm-1, thus it would not have been obvious if just based on the interpretation of the FTIR spectrum measured with a singleelement detector for the deposit in Figure 2a. The images in Figure 3, panels e and f, are generated based on the distribution of the integrated absorbance in the ranges 1180-1100 and 1080-1000 cm-1, respectively. They can be assigned to the sulfate and sulfoxide functional groups, respectively, which substantiates our tentative assignment of these species based on the conventional FTIR spectrum with a singleelement detector. Although conventional single-element FTIR spectroscopy may have a better SNR and a faster acquisition time, its sensitivity is inferior to the more advanced imaging technique. In single-element measurements, chemical information of the sample is averaged across the whole measured area of the ATR crystal, thus spectral bands of a component in a heterogeneous sample will be diminished by the spectral bands of other components depending on their concentration and absorptivities. On the basis of the images shown, fouling deposits from the heat exchanger are extremely heterogeneous. Clusters of distinct compounds ranging in size from ca. 40 µm for the oxalate to ca. 150 µm for the sulfate can be observed. This emphasizes how the enhanced sensitivity of the imaging approach can provide more accurate chemical information of a sample especially for heterogeneous materials. When components are resolved spatially, spectroscopic information that represents the components can be obtained. It is well accepted that the fouling mechanism of crude oil heat exchangers is complex, which often involves crystallization of inorganics, deposition of particulates, corrosion, and chemical reactions of organics such as oxidative polymerization, asphaltenes precipitation, and coke formation.7,44 The mitigation strategy is therefore plant specific and largely dependent on the feedstock composition of the crude oil. The FTIR images have shown both chemical and spatial information of the organic and inorganic fractions in the deposit analyzed. From the measurements with a single-element detector and the distribution of the methylene asymmetrical stretching mode measured by the (44) Watkinson, A. P.; Wilson, D. I. Exp. Therm. Fluid Sci. 1997, 14 (4), 361–374.

4064

Energy & Fuels, Vol. 23, 2009

Tay and Kazarian

Figure 3. Macro-ATR-FTIR images of the deposits. Images are generated based on the integration of the absorption band as indicated below each images. The imaging area is ca. 610 × 530 µm2. Spectrum is extracted from point X of the chemical image above it. The red box in panel b shows the relative size of a micro-ATR FTIR imaging measurement.

Study of Petroleum Heat-exchanger Deposits

macro-ATR-FTIR imaging approach shown in Figure 3a, the bulk of the deposit was observed to be mainly asphaltenic. Detailed analysis of the organics such as aromaticity, length of the aliphatic side chains, and carbonyl abundances can be carried across the large data set obtained from each imaging measurement. The chemical images also reveal chemical heterogeneities in the deposit, especially for the mineral compounds. This may help in indicating the main mechanisms that contribute to fouling, thus providing information to the heat exchanger specialist in deciding mitigation strategies. For example, the oxalates detected are not commonly found in crude oil but are known to exist in the form of an organic acid salt in sedimentary rocks.45 This particular species is largely present in the deposits for this particular crude oil could be an initiator or a cofactor to the fouling process. Micro-ATR-FTIR Measurement. Figure 3 has shown the overall distribution of the different components in the deposits with the macro-ATR technique using the diamond ATR accessory and how this surpasses conventional single-element FTIR spectroscopy in characterizing such heterogeneous materials. ATR-FTIR spectroscopy with a single-element detector and without the use of a microscope measures the whole sampled area. With a better spatial resolution of ca. 12 µm for macroATR imaging, more information can be obtained about the spatial distribution of different components in the sample. However, particles of a size smaller than 12 µm will not be resolved. A different ATR imaging approach is used here to examine the deposits with a spatial resolution higher than with macro-ATR imaging. The micro-ATR-FTIR imaging technique uses a microscope objective with a germanium crystal. The germanium crystal is hard, 6 on the Mohs scale, but it is brittle. It has a high refractive index of 4 that provides the high spatial resolution when used in immersed objectives, but it is not as robust as the diamond crystal. The image size of the microATR measurements is about 63 × 63 µm2 with the objective used in this study; and the spatial resolution, as it was demonstrated before,46,47 is about 2-4 µm. A red square shown in Figure 3b shows the relative image size of a micro-ATR measurement compared to the size of a macro image obtained with the diamond ATR accessory. From the macro FTIR images, it was shown that the components are dispersed, thus a single micro-ATR-FTIR image, which measures an almost 100 times smaller area of the sample, will not be a good representation of the sample. In this work, multiple micro-ATR imaging measurements were carried out and analyzed, but only four measurements are presented and discussed below. The micro-ATR-FTIR measurement in Figure 4a shows the distribution of a component not identified from initial analysis of the macro FTIR image in the integration range of 1630-1580 cm-1. It shows a cluster of domain size of ca. 7 µm. The spectrum extracted shows 2 strong bands at 1605 and 1313 cm-1. This component was also detected in other separate microATR-FTIR measurements, and it was confirmed that the two bands are from the same component. The oxalates also have two similar bands at 1635 and 1330 cm-1, thus it is speculated from these results that this unknown component may be another complex of the oxalate. The macro-ATR imaging used the same (45) Jehlicka, J.; Edwards, H. G. M. Org. Geochem. 2008, 39 (4), 371– 386. (46) Kazarian, S. G.; Chan, K. L. A.; Tay, F. H., ATR-FT-IR Imaging for Pharmaceutical and Polymeric Materials: From Micro to Macro Approaches. In Infrared and Raman Spectroscopic Imaging; Salzer, R., Siesler, H. W., Eds.; John Wiley & Sons, Ltd: Chichester, 2009; pp 347375. (47) Palombo, F.; Shen, H.; Benguigui, L. E. S.; Kazarian, S. G.; Upmacis, R. K. Analyst 2009, 134, 1107–1118.

Energy & Fuels, Vol. 23, 2009 4065

intergration range of 1630-1580 cm-1 for this compound, and the image generated is shown in Figure 3d. The domain size of the component is just large enough to observe a distribution, but spectral information of this component is still masked by the surrounding contributions. The spectrum from the microATR measurement is able to show stronger bands of the component due to its enhanced spatial resolution. Figure 4b shows the distribution of the carbonate based on the distribution of the integrated absorbance in the range of 1500-1350 cm-1. The spectrum extracted from the red area of the image showed a single band at 1430 cm-1. When this spectrum is compared to the extracted spectrum from the macroATR-FTIR images, it can be seen that this band is narrower and has lesser contribution from other absorbing bands in that region of the spectrum. This shows that a “purer” spectrum of a component can be obtained with the enhanced spatial resolution,48 thus allowing better chemical analysis and spectral interpretation. Figure 4c shows the distribution of the sulfoxide functional group based on the integrated absorbance in the range of 1080-1000 cm-1. When the spectrum extracted from this image is compared to the spectrum from the macro-ATR-FTIR image in Figure 3f, a much stronger sulfoxide band is observed. Figure 4d showed the image generated based on the shift of the baseline of the IR spectrum at 1900 cm-1. Carboneous materials such as graphite and also petroleum deposits can result in an increase of the absorbance baseline in the entire measured IR spectrum. The image in Figure 4d showed clusters of about 30 µm in diameter corresponding to the materials, which result in an increase of the absorbance baseline. Coke formation is one of the mechanisms of fouling, and it is correlated to the thermal degradation of crude oil fractions. The reaction that results in coke formation is still debatable but is closely associated with the asphaltenes.49 The chemical structure of coke is mainly large PAHs with few aliphatic side chains. These aromatic hexagonal sheets have structures similar to graphite, thus they behave similarly in the IR region of the spectrum. The coke material was only observed in the microATR approach and only appeared occasionally. Although the bulk temperature of the line where this deposit came from was between 155 and 166 °C, the temperature of the heat transfer surface will have been higher. Fan and Watkinson compared the diffused reflectance spectra of graphite and industrial samples from a bitumen coker.50 From the spectra, they have deduced that the industrial deposits are mainly graphitic. Although our heat exchanger sample was from a lower temperature stream compared to the 500 °C of a coker, the aging effects of the deposit are not well studied. As the crude oil deposit was in the heat exchanger unit for an undetermined time, it is possible to have materials of a more graphitic nature. These may be formed on the hotspots of the heat exchanger surface. These highly absorbing blackbody materials would not have been observed in techniques used before, and FTIR imaging may provide a way to monitor this aging effect of the deposits. With a higher spatial resolution from the micro-ATR imaging, smaller domains are magnified, thus the sensitivity of the ATRFTIR imaging effectively increases. In-depth analysis of the different components in the deposit can be carried out by using the better-represented spectrum obtained using the micro-ATR approach. The macro-ATR approach, with a larger field of view, (48) Chan, K. L. A.; Kazarian, S. G. Analyst 2006, 131 (1), 126–131. (49) Rahmani, S.; McCaffrey, W.; Gray, M. R. Energy Fuels 2002, 16 (1), 148–154. (50) Fan, Z. M.; Watkinson, A. P. Ind. Eng. Chem. Res. 2006, 45 (18), 6104–6110.

4066

Energy & Fuels, Vol. 23, 2009

Tay and Kazarian

Figure 4. Micro-ATR-FTIR images of the deposits. Each of the images is from different measurement. Images a-c are generated based on the integration of the absorption band as indicated below each images. Image d is generated based on the shift in baseline at 1900 cm-1. The imaging area is ca. 63 × 63 µm2. Spectra are extracted from point X of the chemical image beside it.

can give a better overview picture of the whole sample. Quantitative analysis of the deposit, such as the particle size

analysis of the different components, can be obtained from the FTIR images. Although multiple measurements will still be

Study of Petroleum Heat-exchanger Deposits

needed to obtain a statistically significant representation of the sample, this technique is relatively fast where each measurement only takes less than 2 min. Conclusion This study has addressed the advantages and intrinsic limitations of ATR-FTIR spectroscopic measurements of high refractive index materials, such as petroleum deposits. A labmade aperture is used to correct the distortion of spectral bands due to the dispersion of the refractive index. This allowed reliable spectral information to be obtained from high refractive index materials using ATR-FTIR spectroscopy with a diamond accessory. This development was further extended to acquire ATR-FTIR images of petroleum deposits with the FPA detector. FTIR imaging allows the visualization of different chemical components in a heterogeneous sample. Using the diamond accessory in the macro-ATR setup, clusters of five chemically different compounds, namely, organics which are asphaltene in nature, carbonates, sulfates, sulfoxides, and oxalates were identified in a petroleum deposit. The imaging approach provides

Energy & Fuels, Vol. 23, 2009 4067

both spatial and chemical information of the deposit, was not possible to characterize in this way before. It was also demonstrated that with the enhanced spatial resolution of the micro-ATR approach, a more representative spectrum of the components can be achieved. Different forms of complex of the oxalate, one of which was barely at the detection limit of the macro-ATR approach, were also identified using the microATR-FTIR imaging. Highly IR absorbing materials, which may possibly be coke, were also found and represented in the chemical image from the micro-ATR-FTIR approach. By combining both macro-ATR and micro-ATR-FTIR spectroscopic imaging, the complex petroleum deposit can be better characterized. The ATR-FTIR imaging provides an important tool in the chemical characterization of fouling materials, which will aid in understanding the fundamentals of crude oil fouling. Acknowledgment. We thank EPSRC for support (EP/D503051/ 1) and Dr. Paul Turner for his help and advice. EF900304V