PROCESS CONSIDERATIONSIN CRUDE OIL BIODESULFURlZATION

Abhijeet P. Borole and Eric N. Kaufman

Bioprocessing Research and Development Center
I Chemical Technology Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee, USA 3783 1-6226
\
"Thesubmitted manuscript has been authored by a contractor of the US. Government
under contract DE-AC05-960R22464. Accordingly, the US. Government retains a
nonexclusive, royalty-pee license to publish or reproduce the publishedform
of this contribution. or allow others to do so,for US. Governmentpurposes. "

ABSTRACT

Biodesulfurization offers an attractive alternative to conventional hydrodesulfurization due to the

mild operating conditions and reaction specificity afforded by the biocatalyst. The enzymatic
pathway existing in Rhodococnrs has been demonstrated to oxidatively desulfurize the organic
sulfur occurring in dibenzothiophene while leaving the hydrocarbon intact. In order for
biodesulfurization to realize commercial success, a variety of process considerations must be
addressed including reaction rate, emulsion formation and breakage, biocatalyst recovery, and
both gas and liquid mass transport. This study compares batch stirred to electro-spray bioreactors
in the biodesulfurization of both model organics and actual cnrdes using a Rhodococcus IGTSS
biocatalyst in terms of their operating costs, ability to make and break emulsions, ability to effect
efficient reaction rates and enhance mass transport. Additionally, sulfur speciation in crude oil is
assessed with respect to sulfur specificity of currently available biocatalysts.

KEY WORDS: crude oil desulhrization, Rhodococcus, electrostatic spraying, dibenzothiophene,

biodesulfurization

INTRODUCTION

Biological processing of fossil fuel feedstocks offers an attractive alternative to conventional

thermochemical treatment due to the mild operating conditions and greater reaction specificity
afforded by the nature of biocatalysis. Efforts in microbial screening and development have
identified microorganisms capable of petroleum desulhrization (see for example, [l-3]),
denitrification [4], demetalization [4], cracking [SI and dewaxing. Biological desulhrization of
petroleum may occur either oxidatively or reductively. In the oxidative approach, organic sulfur is
I
converted to sulfate and may be removed in process water. This route is attractive due to the fact
that it would not require further processing of the sulfur and may be amenable for use at the well
head where process water may then be reinjected. In the reductive desulfurization scheme,
organic sulfur is converted into hydrogen sulfide, which may then be catalytically converted into
elemental sulfur, an approach of utility at the refinery. Regardless of the mode of
biodesulhrization, key factors affecting the economic viability of such processes are biocatalyst
activity and cost, differential in product selling price, sale or disposal of co-products or wastes
from the treatment process, and the capital and operating costs of unit operations in the treatment
scheme.

In all fossil fuel bioprocessing schemes, there is a need to contact a biocatalyst containing aqueous
phase with an immiscible or partially miscible organic substrate. Factors such as liquid / liquid
and gas / liquid mass transport, amenability for continuous operation and high throughput, capital
and operating costs, as well as ability for biocatalyst recovery and emulsion breaking are
significant issues in the selection of a reactor for aqueous / organic contacting. Traditionally,
impeller-based stirred reactors are utilized for such mixing due to their ease of operation and wide
acceptance in the chemical and biological processing industries. Such mechanically stirred
reactors contact the aqueous and organic phases by imparting energy to the entire bulk solution,
i.e. the impeller must move the contents of the reactor.

Recent advances in the area of contactors for solvent extraction have lead to the development of
electrically driven emulsion phase contactors (EPCTM)for efficient contact of immiscible phases
[6]. In this concept, the differing electrical conductivity between the aqueous and organic phases
causes electrical forces t o be focused at the liquid / liquid interface, creating tremendous shear
force. This shear causes the conductive phase to be dispersed ( 5 Om droplet size) into the non-
conductive phase, but does so with decreased energy requirements relative to mechanical
agitators due to the fact that energy is imparted only at the liquid / liquid interface and not the
entire bulk solution. In a configuration of the EPCm developed at the Oak Ridge National
Laboratory, the contactor serves to disperse aqueous phase containing biocatalyst into an organic
23
 r
phase. The EPCW creates droplets of water containing biocatalyst -5 Om in diameter within an
organic phase.

Here, we compare the performance of the EPCW to that of a batch stirred reactor (BSR),
investigate the required level of biocatalyst activity before the surface area afforded by the EPC"
becomes a factor in reactor performance, and characterize the emulsion formed by both reactors
in the presence of bacteria. We have investigated the emulsion quality formed in the EPC,
evaluated the power requirements and analyzed the mass transfer issues in comparison to stirred
reactors. Results on biodesulfurization of actual crude oil by wild type Rhodococcus IGTS8 are
also included. Finally, we assess the sulfur specificity of available biocatalysts with respect to
sulfur compounds present in crude oils.

MATEFUALS AND METHODS

The experimental procedures used for studying biodesulfurization in model systems have been
discussed in detail in previous publications [7-91. A detailed description of oil experiments is
provided here.

Biodesulfurization of Van Texas Crude oil

Biodesulfurization of Van Texas crude oil was studied in batch stirred reactors to evaluate the
substrate specificity of the biocatalyst. The experiment was conducted over a treatment period of
6 days. The crude had an API specific gravity of 31", and a sulfur content of 0.96 wtoh. The
crude oil did not contain volatiles due to production at elevated temperature (-99°C).
Experiments were performed in batch stirred reactors utilizing 50 g of frozen Rhodococcus sp.
wild type strain IGTS8 (ATCC 53968) cell paste which were brought up to 750 mL. with 0.156M
(pH 7.5) phosphate buffer. The cells were suspended in the phosphate buffer prior to addi!icn ?c
the reactor. The reactor VP.QQC! E-:! *:.zz i 1-L -ATis urn-Culture fermentor (model 178657,
Gardiner, NY), utilizing a 6-bladed Rushton-type impeller with 2 baffles. The reactor was kept at
3OoC, agitated at 800 RPM, and aerated with room air at a rate of 0.2 standard liters per minute
(SLPM). A water condenser was used on each reactor to capture volatiles which were expected
to be minimal or non-existent considering the fact that the operating temperature was much less
than that of the oil reservoir. The experiment was conducted with 250 mL of crude oil, treated
with 750 mL of the aqueous phase. Samples (30 mL. from the top of the organic phase) taken
during the course of biological treatment were collected after ceasing the agitation and aeration
for 5 min to allow the aqueous and organic phases to separate. The reactor contents were
emptied at the end of the run and centrifuged at 6000 rpm in a Beckman Model TJ-6 centrifuge to
obtain a sample of treated crude oil. Closed samples were boiled in a closed container for 30 min
to halt biological activity.

Analytical

Model system experiments

In the experiments reported here, DBT and 2-HBP concentrations in the aqueous phase were
below our levels of detection. DBT and 2-hydroxybiphenyl (2-HBP) concentrations in n-
hexadecane were measured by gas chromatography using a Hewlett Packard 5890 gas
chromatograph equipped with a flame ionization detector.

Crnde oil
,
A GC-SCD method was used to determine the sulfur content of the aromatic fraction of the oil.
To allow facilitated observation of sulfur in the treated oil, whole oil samples were fractionated
according to ASTM method D2007. An extended ASTM D2887 procedure was used for
chromatographic separation of the aromatic fraction of the crude oil. Sulfur analysis was
performed by modifylng the ASTM D2887 procedure by adding a Sievers Chemiluminesence
sulfur specific detector after the flame ionization detector.

RESULTS AND DISCUSSION

Rate of biodesulfurization

The specific rate of DBT desulfurization by Rhodococcussp. was typically between 1 and 5 mg 2-
HBP produced per dry g of biocatalyst per hour. Specific rates of 2-HBP production in the batch
stirred reactor and the EPCm reactor systems were within experimental variance and no

24
I
appreciable difference in desulfurization rates were seen between the two reactors. Due to the
hi& surface area reported in the EPCm [7], higher rates were expected in the EPCm, however,
similar performance was observed in both reactors. The reaction rate obtained was without any
supplemental carbon or energy source. Note that the only available carbon and energy source for
\ the biocatalyst other than what may be carried over in the frozen cell paste, was hexadecane and
DBT. However, DBT was not used as the carbon source by the biocatalyst, since the end product
of DBT conversion was 2-HBP (thus preserving the carbon number and fuel value). Other
\ studies (outlined in [7]) have utilized additional external carbon and energy sources and have
reported higher activities with Rhodococcus sp. A commercial scale biodesulfurization process
may require a higher cell density to achieve maximum conversion in a minimum time, provided it
does not affect yield with respect to biocatalyst usage. In order to study effect of cell density,
experiments were conducted in the BSR at different cell loadings. This experiment also lead to
determination of the cell density at which point the BSR becomes mass transfer limited. A
following experiment in the EPCW at this cell density was conducted to evaluate the benefits of
high surface area afforded by the EPCm.

Mass transport issues

The rate of desulfurization, when normalized with respect to cell mass, was found t o decrease
with increasing cell density indicating that mass transfer resistance was the controlling process in
desulfurization. A statistical analysis of the data indicated mass transfer limitation between 5X
and 1OX cell density in the BSR. The mass transfer limitation may be due to gas-liquid or liquid-
liquid mass transport resistance.

The results of experiments conducted in the BSR at IOX cell density indicated no gas-liquid mass
transfer limitation, Increasing the rate of air supply or increasing the oxygen tension in the reactor
through the use of pure oxygen rather than air did not seem to affect HBP production. This
suggests that the system may be limited by liquid-liquid mass transfer Since the EPCTM
reportedly provides larger liquid-liquid interfacial area, the BSR was compared with the EPCW
for desulfurization activity at the high cell density.

Comparison of the EPCW and BSR performance at 1OX cell density showed no difference in the
desulfurization rates between the two reactors. Thus, either the system is not truly mass transport
limited or the EPCm did not provide a larger surface area for reaction under the present
conditions. A detailed characterization of the emulsions formed in the BSR and EPCW in the
presence and absence of biocatalyst was conducted and is reported below.

Emulsion quality in BSR and E P O M

A detailed drop size analysis of the two-phase emulsion formed in the BSR has been reported
previously [7]. Characterization of the emulsion quality in BSR in the absence of biocatalyst has
revealed 100-200 micron droplets under the conditions of experiments conducted here. The
droplets formed in the EPC; however, are in the 1-10 micron range. The ability to form fine
emulsions in the EPCm without increasing energy utilization (see energy utilization section
below) could have tremendous impact upon processing costs assuming that the biocatalyst utilized
is active enough to be mass transport limited.

Emulsion quality in the presence of biocatalyst

Due to the opaqueness rendered by presence of biocatalyst, observations could not be made in
situ during reactor operation. To determine the emulsion quality formed in the E P C m and BSR,
and to determine whether the EPCW offers larger surface area than BSR, samples were collected
from the reactors and observed under a microscope using a lOOx oil emersion objective.
Microscopic examination of samples showed formation of a very fine emulsion in both reactors
with droplet sizes ranging from 1 to 10 Om. Formation of such an emulsion in the BSR may be
presumed due to production of biosurfactants by the biocatalyst IGTSS. Average droplet size for
EPC" and BSR samples were 2.54 _+ 2.40 Om and 3.08 _+ 1.78 Om, respectively (n>300).
Further, a significant amount of the biocatalyst was extracted and existed in the organic phase.
Thus, a very fine emulsion is formed in the EPCm as well as the B S q and it appears that it is for
this reason that an augmentation in desulfirization rate is not seen in the EPCm relative to the
BSR. A couple of process issues warrant consideration here. Firstly, due to the formation of a
fine stable emulsion, downstream separation of the multiphase mixture to obtain clean organic fuel
may require additional separation processes. Secondly, the Rhodococcus biocatalyst used in these
experiments was extracted into the organic phase. If biocatalyst recovery and reuse is desired,
separation ofthe biocatalyst from aqueous as well as organic phases will be required.

25
Energy utilization by E P P

Typically, stirred reactors or impeller based reactors are capable of achieving water or oil droplet
sizes of 100 -300 O m in diameter under the conditions used in this study when surfactants are not
present. Tin order to create such droplet size distribution, the energy required is on the order of
1-6 W/L (based upon empirical correlation’s [IO]). It is estimated that if impeller based systems
were capable of producing 5 Om droplets, it would require -25 kWL [I I] if surfactants are not
present. The EPCW creates droplets of water containing biocatalyst -5 Om in diameter within an
organic phase, and does so with a power requirement of 3 W L [7]. Thus, if a high activity
biocatalyst is available, which is actually limited by mass transport, the EPCm could result in
tremendous savings over the batch stirred reactor. For instance, on a 1 L basis, a BSR using a 3: 1
water to oil ratio and producing oil droplets of 150 mm in diameter creates 1 x IO’ cm2 of
interfacial surface area. On the same volume basis, an EPCm creating 5 Om diameter aqueous
droplets and having a 5% aqueous hold-up creates 6 x 10’ of interfacial area at 1/15” the aqueous
volume to do so. In a mass transport system, the rate of desulfurization would thus be expected
to be six times as large using 93% less biocatalyst. An additional important point which needs to
be noted here is that the fine emulsion formed in the EPCm is an unstable emulsion Le., the
emulsion breaks easily upon removal of the electric fields giving easy separation of the organic
and aqueous streams.

Crude oil biodesulfurization

Oil samples collected from the BSR were analyzed by GC-SCD to obtain the distribution of
organosulfur compounds in the crude oil. Analysis was done on the aromatic fraction of the oil
and not the whole oil, since the baseline did not return to its initial value in case of the whole oil.
This fraction of the Van Texas oil accounted for 22% of the oil’s original volume. A s ~he-:r^ ia
Figure 1, the total sulfur content of this arnma!k !?ski,was reduced from 3.8 to 3.2% in 6 days
nf!re.*--c;.: iGi~8.

The results indicate removal of comparatively low molecular aromatic sulfur compounds;
however, a large portion of the organosulfur fraction does not seem to be affected by the
biodesulfurization process. Additional analysis by GC-MS (not shown here) has revealed up to
90% sulfur removal from DBT and substituted DBT compounds. While it appears that this
biocatalyst is capable of desulfurizing the majority of sulfur species present in diesel @BT and
substituted DBT compounds) and that only improvements in the rate of desulfurization are
needed for the commercialization of this process, a great deal of research is needed for oil
biodesulfurization to be realized. The sulfur specific oxidation of DBT by Rh&occus resulted
!?om over 15 years of research using DBT as the model organic sulfur compound in coal and oil.
Detailed sulfur speciation studies and biocatalyst development is needed to achieve desulfurization
of the broad spectrum of organic sulfur species present in crude oil and to realize the promises of
petroleum biodesulfurization.

CONCLUSIONS
A variety of process considerations in the biodesulhrization of petroleum feedstocks were
addressed in this study including reaction rate, emulsion formation and breakage, biocatalyst
recovery, and both gas and liquid mass transport. Comparison of batch stirred reactor to EPCm
revealed formation of high surface area in the E P P in the absence of surface-active agents.
Presence of biocatalysts capable of producing biosurfactants results in fine emulsions in both
reactors; however, poses a potentially more difficult problem with downstream multiphase
separation. The use of EPCW as a biodesufirization reactor can result in up to several orders of
magnitude energy savings over BSR in the absence of surfactants. Gas-liquid mass transfer was
not a limiting factor in biodesulfurization studies with model systems. Further, biodesulfurization
experiments with actual crude oil showed that presently available biocatalysts such as
Rhodococcus sp. IGTS8 are capable of removing DBT and substituted DBT type compounds but
do not affect the remaining portion of the organosulfur compounds. Thus, there is a need for
hrther development in biocatalysts capable of desulhrization of higher molecular weight non-
DBT type sulfur compounds present in crude oil.

ACKNOWLEDGMENTS

This work was supported by the Office of Oil & Gas Processing, U.S. Department of E n e r a
under contract DE-AC05-960R22464 with Lockheed Martin Energy Research Corp. The
authors wish to thank Dr. Robert Shong of Texaco for GC-SCD analysis of Van Texas crude oil.
The authors acknowledge the material contributions of Energy BioSystems Corp.

26
!

REFERENCES

Lee, M.K., J.D. Senius, and M.J. Grossman, Sulfur-specij7c microbial desulfurizationof
sterically hindered analogs of dibenrothiophene. Applied and Environmental
Microbiology, 1995. 61(12): p. 4362-4366.
2. Monticello, D.J., Biocatalytic desulfurization. Hydrocarbon Processing, 1994. February
1994: p. 39-45.
3. Shennan, J.L., Microbial attack on sulfur-containing hydrocarbons: Implicationsfor the
desulfurization of oils andcoals. J. Chem. Tech. Biotechnol., 1996. 67: p. 109-123.
4. Lin, M.S., et al., Biochemical processing of heavy oils and residuum. Applied
Biochemistry and Biotechnology, 1996.57158: p. 659-664.
5. Premuzic, E.T., et al., Biochemical Alteration of Crude Oils in Microbial Enhanced Oil
Recovery, in Biohydrometallurgical Technologies, A.E. Torma, M.L. Apel, and C.L.
Briadey, Editors. 1993, The Minerals, Metals, and Materials Society p. 401-413.
6. Scott, T.C., D.W. DePaoli, and W.G. Sisson, Further development of the electrically
driven emulsionphase contactor. Ind. Eng. Chem. Res., 1994. 33: p. 1237-1244.
7. Kaufman, E.N., et al., Development of an electro-spry bioreactor for crude oil
processing. Fuel Processing Technology, 1997. 52: p. 127-144.
8. Kaufman, E.N., et al. Biodesulfirization of dibenzothiophene and crude oil using
elecfrosprq reactors. in 3rd International Petroleum Environmenial Conference. 1996.
Albuquerque, NM.
9. Kaufman, E.N., J.B. Harkins, and A.P. Borole. Biodesulfurizationof Dibenzothiophene
and Crude Oils Using Batch Stirred and Electro-sprq Reactors. in International
Petroleum Environmental Conference. 1997. San Antonio, TX.
10. Perry, R.H., D.W. Green, and J.O. Maloney, eds. Perry's Chemical Engineering
Handbook. 6 ed. . 1984, McGraw-Hill: New York. 2336.
11. Scott, T.C. and W.G. Sisson, Droplet size characteristics and energv input requirements
of emulsions formed using high intensity pulsed electric jielh. Separation Science. and
Technology, 1988. 23: p. 1541-1550.

8.OS5

7.005

6.0~5

5.Od

3
4.0d
0
P8 3.0~5

2.oc5

f
0 5 IO 15 20 25

'lirne(min.)

Figure 1. Analysis of the aromatic fraction of Van Texas crude oil by GC-SCD. The
biotreatment results in removal of low molecular weight DBT-type compounds; however,
the higher molecular weight compounds are not affected.

27