Reduction of Cr(VI) polymerization catalysts by non-olefinic hydrocarbons

Article abstract of DOI:10.1016/j.apcata.2012.02.023

The Phillips Cr(VI)/silica catalyst, which is widely used for commercial ethylene polymerization, is usually considered to be reduced to a lower-valent active Cr species in the reactor upon contact with ethylene or other α-olefin monomers. In this paper, however, the case is presented that Cr(VI) is actually quite reactive with other hydrocarbons to which it is also often exposed, including alkanes and aromatics. Redox products from these reactions are identified, and the effect on catalyst polymerization activity and polymer character is described.

Full text of DOI:10.1016/j.apcata.2012.02.023

Applied Catalysis A: General 423–424 (2012) 91–99
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Reduction of Cr(VI) polymerization catalysts by non-olefinic hydrocarbons
∗
Eric Schwerdtfeger, Richard Buck, Max McDaniel
Chevron-Phillips Chemical Co, Bartlesville, OK 74004, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The Phillips Cr(VI)/silica catalyst, which is widely used for commercial ethylene polymerization, is usually
considered to be reduced to a lower-valent active Cr species in the reactor upon contact with ethylene
or other ␣-olefin monomers. In this paper, however, the case is presented that Cr(VI) is actually quite
reactive with other hydrocarbons to which it is also often exposed, including alkanes and aromatics.
Redox products from these reactions are identified, and the effect on catalyst polymerization activity and
polymer character is described.
Received 17 December 2011
Received in revised form 14 February 2012
Accepted 16 February 2012
Available online 25 February 2012
Keywords:
©
2012 Elsevier B.V. All rights reserved.
Phillips catalyst
Ethylene polymerization
HDPE
Polyethylene
Reduction
Chromium
Alkane oxidation
1
. Introduction
lower-valent species, the potential coordination sphere is
expanded, and in the absence of other ligands, coordinative unsat-
uration is created that initiates polymerization.
Polyethylene is the world’s most commonly used polymer, and
a significant portion of it (HDPE in particular) is made with the
Phillips Cr/silica catalyst. Discovered in 1951 at Phillips Petroleum
Company [1,2], this catalyst is made by impregnating a chromium
In commercial practice, the hexavalent catalyst precursor usu-
ally also comes into contact with other hydrocarbons, such as the
reaction solvent or diluent. These commonly include paraffins,
such as isobutane, n-butane, cyclohexane, hexanes, isopentane,
and even mineral oil. Ethylene comprises only a small part of the
typical reaction mixture, e.g. only 3–5 wt% in isobutane (slurry at
compound onto a high-porosity silica carrier to a loading of about
◦
1
wt% Cr, followed by calcination at >500 C. During this high tem-
perature “activation” step the chromium is oxidized and reacts with
surface hydroxyl groups to become anchored to the support as
hexavalent chromate or dichromate species such as illustrated in
Scheme 1A.
◦
◦
80–110 C) or in cyclohexane (solution at 125–150 C). Typically,
the hexavalent catalyst is charged into a storage vessel, some-
◦
times while still quite hot (100–250 C) after calcination, into which
This is the catalyst precursor. The active sites responsible for
ethylene polymerization are formed when this surface-anchored
Cr(VI) is then reduced to a lower-valent isolated species [2,3]. This
is usually thought to occur in the polymerization reactor with ethy-
lene as the reducing agent, such as shown in Scheme 1B. Surface
Cr(II) and formaldehyde have most often been suggested as the
by-products of this redox reaction [4–7], although other possi-
ble products have also been proposed [8,9], and some recent data
from this laboratory (to be presented elsewhere) likewise argues
for a more complicated reduction pathway. As the chromium
is reduced, from tetrahedral Cr(VI) to a probably octahedral
one of these non-olefinic hydrocarbons is immediately introduced.
Afterward the catalyst is then stored under one of these hydro-
carbons at ambient temperature for hours, days, and sometimes
weeks, prior to introduction into the reactor where it can contact
ethylene.
Although the reactivity of the Phillips Cr/silica catalyst with
alkanes has not been investigated to our knowledge, some hexava-
lent chromium compounds have been reported to oxidize many
organics, including olefins, alcohols, carbonyls, and even alkanes
[10]. Thus, there is reason to suspect that in some commercial
usage the Cr(VI) could be reduced even before contact with ethy-
lene. In this paper we explore this possibility. The reactivity of the
hexavalent catalyst with various paraffins and aromatics has been
investigated, and the resultant implications for the formation of the
active species and the polymerization mechanism is discussed.
∗ Corresponding author. Tel.: +1 918 661 9974; fax: +1 918 661 1900.
E-mail address: Max.McDaniel@sbcglobal.net (M. McDaniel).
0
926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2012.02.023

9
2
E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
ethanol and 5–10% water, as indicated in each experiment, was
then added to again just submerge the catalyst. The solution also
often contained 0.3% HCl to encourage hydrolysis of ligands on the
chromium. After the ethanol solution was allowed to contact the
catalyst for the specified time, usually about 1 day, it was sampled
and analyzed by GC–MS.
In the second method of reduction the catalyst was left in the
fluidized bed where it was activated, as described. Dry nitrogen
was used to fluidize the catalyst, and the temperature was set as
◦
described (usually at 100 C). About 5 mL of non-olefinic hydrocar-
Scheme 1. (A) Formation of Cr(VI)/silica catalyst precursor; (B) reduction of Cr(VI)
by ethylene to lower valent active species, as suggested by Ref. [4] and others.
bon was then injected into the nitrogen stream, where it evaporated
over a period of about half an hour. The vapor was carried up
through the catalyst bed where it reduced Cr(VI). After passing
through the catalyst bed, the nitrogen was directed through a cold
trap where the oxidized hydrocarbon was collected and later ana-
lyzed by GC–MS.
2
. Experimental
2.1. Catalyst preparation
As a second step in this second method of contact, the catalyst
bed was then exposed to water vapor at 100 C in order to hydrolyze
◦
An attempt was made to select a range of catalysts (Cr/silica
◦
◦
and Cr/silica-titania), activation temperatures (600 C to 850 C), Cr
loadings (1% to 5%), and Ti loadings (0%, 2.5%, 4.5%) that are still rep-
resentative of commercial operations. Detection of products was
enhanced by higher Cr loadings and lower activation temperatures,
which produce higher conversion to Cr(VI).
any remaining ligands left on the chromium. About 5 mL of water
was injected into the nitrogen stream where it evaporated over
about half an hour and passed up through the catalyst bed. After
exiting the catalyst bed, the nitrogen was then directed through
another cold trap, where the water and hydrocarbon oxidation
products were collected and later analyzed by GC–MS.
Cr/silica catalysts were prepared from grade 951 silica, obtained
from W.R. Grace Co., which was impregnated to incipient wetness
with an aqueous solution of chromic trioxide, followed by drying
◦
overnight in a vacuum oven at 110 C. The final catalyst contained
2.3. Analysis by GC–MS
either 1 or 3 wt% chromium, and had a surface area of approx-
2
imately 600 m /g and a pore volume of approximately 1.0 mL/g.
Gas chromatography was performed using a Varian 3800 GC
analyzer equipped with two separate all-purpose capillary columns
(30 m × 0.25 mm × 0.25 ␮m) and flame ionization (FI) detector. In
a typical injection, a 5 ␮L aliquot was removed by syringe from the
Cr/silica-titania catalysts were made by ter-gelation of silica, tita-
nia, and chromium (III) hydroxide (from a solution of Cr nitrate)
according to Dietz [11]. This yields a catalyst containing either 2.5
2
◦
or 4.5 wt% Ti, 1.0 or 3.0 wt% Cr, a surface area of 480 m /g and a pore
sample vial and injected into a GC port held at 250 C using a split
volume of 2.5 mL/g. Like the impregnation technique, this method
leaves the Cr on the surface.
The next step, activation or calcination, anchors the chromium
to the surface and converts trivalent chromium on the catalyst
into the hexavalent form [3]. All of the catalysts used in this
study, whether Cr/silica or Cr/silica-titania, were found to contain
ratio of 20:1. The carrier gas used was ultra-high purity helium and
was electronically controlled throughout the run to a constant flow
rate of 1.0 mL/min. In a typical run, the initial column temperature
◦
◦
◦
was held at 70 C for 2 min, then ramped at 20 C/min to 250 C and
◦
then held at 250 C for 9 min. In cases where FI detection was used,
◦
the FI detector on the GC was maintained at 300 C.
9
0–100% of the chromium as Cr(VI). To activate these catalysts,
Mass spectral analysis was performed in conjunction with a Var-
ian 320 MS instrument connected to the GC unit using electron
ionization at 70 eV. The nominal mass range scanned was 50–900
m/z using a scan time of 0.5 s. Nominal detector voltage used was
1200 V.
about 10 g was placed in a 4.75 cm quartz tube fitted with a sin-
tered quartz disk at the bottom. While the catalyst was supported
on the disk, dry air was blown up through the disk at the linear rate
of about 1.6–1.8 standard cubic feet per hour. An electric furnace
around the quartz tube was then turned on and the temperature
◦
was raised at the rate of 400 C/h to the indicated temperature, such
◦
◦
as 600 C or 800 C. At that temperature the catalyst was allowed to
fluidize for 3 h in the dry air. Afterward the catalyst was collected
and stored under dry nitrogen, where it was protected from the
atmosphere.
2.4. Polymerization
Polymerization runs were made in a 2.5 L steel reactor equipped
with a marine stirrer rotating at 400–500 rpm. The reactor was sur-
rounded by a steel jacket through which a coolant was circulated.
The temperature of the coolant was precisely controlled through
instrumentation so that the temperature in the reactor could be
2.2. Reduction of the catalyst
◦
Activated catalyst containing Cr(VI) was then exposed to various
controlled to within 0.5 C.
non-olefinic hydrocarbons in one of two ways. In the first method,
the catalyst was contained in a glass vessel sealed with Teflon-
seated valves. Enough of the hydrocarbon liquid was added to just
submerge the catalyst. Then the sealed vessel was allowed to stand
for hours or perhaps days as desired. Sometimes it was shielded
from light in a foil wrap. Other times it was exposed to ambient
fluorescent light. After the specified time had passed the liquid
was sampled and injected into a GC–MS instrument for analysis
of products.
Unless otherwise stated, a small amount (0.01–0.10 g normally)
of the solid catalyst was first charged under nitrogen to the dry
reactor. Next 1.2 L of isobutane liquid was charged and the reactor
◦
heated up to 105 C. Finally ethylene was added to the reactor to
equal 3.8 MPa (550 psig), which was maintained during the exper-
iment. The stirring was allowed to continue for the specified time,
usually about 1 h, and the activity was noted by recording the flow
of ethylene into the reactor to maintain the set pressure.
After the allotted time, the ethylene flow was stopped and the
reactor depressurized and opened to recover a granular polymer
powder. In all cases the reactor was clean with no indication of any
wall scale, coating or other forms of fouling. The polymer powder
As a second step the vessel was then drained of as much of
the hydrocarbon as possible, and the remainder was removed by
◦
evaporation in a nitrogen flow at 25 C. A solution of methanol or

E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
93
Table 1
by some researchers [4]. However, it is not entirely clear in these
experiments whether it is the reduction of Cr(VI) that is the slow-
est step in the production of activity, or a later desorption of
by-products, or even a rearrangement of ligands that leads to alky-
lation.
◦
Activity change at −30 C resulting from brief exposure to hydrocarbons at elevated
temperatures.
a
Reducing treatment
Reductant/Cr
ratio
Activity
−
4
−1 −1
g )
(×10 min
None
Ethylene, 140
Propylene, 100
NA
7
3
3
1
0
27
25
19
28
26
13
14
23
22
25
0.1
5.2
In high pressure polymerization runs conducted in laboratory
autoclaves, it is common for the catalyst to contact the reaction
solvent or diluent, even at elevated temperature, before ethylene is
introduced. This exposure can last for several minutes. In the com-
mercial reactors the catalyst is first stored in the hydrocarbon at
ambient temperature for hours, days, and sometimes even weeks,
before being introduced into the reaction zone, where the solvent
concentration is still 20–30 times higher than that of ethylene. Thus,
under these circumstances, it is reasonable to ask whether alkanes
could also reduce the Cr(VI) and whether the resulting site would
still be active.
◦
C
C
◦
◦
1
2
-Butene, 100
-Butenes, 100
C
◦
C
◦
Isobutene, 100
C
3
◦
◦
◦
◦
◦
Cyclohexane, 150
Cyclohexane, 150
Cyclohexane, 150
Cyclohexane, 150
Cyclohexane, 150
Acetone, 150
Benzene, 150
C
C
C
C
C
0.13
0.25
1
4
15
3
◦
◦
C
C
2
a
First order rate constant measured at −30 ◦C.
Therefore, the above experiment was repeated, but the cata-
lyst was exposed to small amounts of cyclohexane (used as solvent
in the solution process) instead of ethylene. A color change from
orange to green indicated Cr(VI) was reduced by cyclohexane. Then
was then removed and weighed. Activity was specified as grams of
polymer produced per gram of solid catalyst charged per hour.
◦
the reduced catalyst was tested at −30 C as before. The results are
again listed in Table 1. It is clear that cyclohexane was indeed an
◦
3
. Results and discussion
effective reducing agent at 150 C and that it produced a catalyst
◦
that was active for ethylene polymerization at −30 C. Acetylene,
◦
3
.1. Polymerization at −30 C
benzene, and acetone were effective as reducing agents, although
they produced considerably weaker polymerization activity at -
◦
Cr(VI)/silica catalysts are not active for ethylene polymeriza-
30 C.
◦
tion at low temperatures. When contacted with ethylene at 25 C
or below, Cr(VI)/silica catalysts display little or no activity. This is
because reduction of Cr(VI) by ethylene is slow at low tempera-
tures. As the temperature is raised the catalyst gradually develops
activity after an initial dormant period. This interval is called the
induction time”, during which reduction of Cr(VI) occurs to form
a lower-valent active species. At 25 C the induction time is so long
that activity is usually regarded as being near zero. At higher tem-
peratures the induction time decreases, and by about 100 C, where
commercial slurry and gas phase polymerization occurs, only a
short induction time is observed, from approximately 5 to 15 min.
At temperatures higher than 125 C, where the solution process
operates, there is no induction time. Polymerization begins imme-
diately when the catalyst contacts ethylene.
◦
3.2. Storage in alkanes or aromatics at 25 C
n-Heptane: Since commercial catalysts are typically stored
under non-olefinic hydrocarbons for prolonged periods before use,
it was of interest to determine whether reduction by alkane can
“
◦
◦
occur over time at 25 C. Therefore, a Cr/silica-titania catalyst was
◦
calcined at 800 C to produce one of the most active Phillips cat-
◦
alysts used commercially. It had a bright orange color, indicating
near quantitative conversion to Cr(VI), as would be expected. Pro-
tected under a nitrogen blanket, the catalyst was immersed in
n-heptane (industrial grade, dried twice over 13× molecular sieve).
The bright orange powder immediately started to darken to a black-
ish orange color, which indicates a reduction of the hexavalent
chromium. Within an hour the color turned green-black. The cat-
alyst was allowed to stand under n-heptane for six days with no
further color change.
Therefore, far from being the inert diluent that is usually imag-
ined, this industrial grade of n-heptane was found to be surprisingly
reactive with the catalyst, reducing Cr(VI) almost immediately. This
means that the mechanism of reduction shown in Scheme 1B may
not always represent catalyst formation in a commercial process,
and the oxidation products could be different.
◦
As an example of this behavior, the ethylene uptake of a
◦
Cr(VI)/silica catalyst was measured at −30 C. At this low temper-
ature no activity was observed. However, in a second experiment
the Cr(VI)/silica was first exposed to a small amount of ethylene (7
◦
per Cr) at 140 C, then it was tested for polymerization activity at
◦
−
30 C. This time it exhibited strong activity. Thus, it is clear that
ethylene performs more than one function. First it reduces Cr(VI)
to a lower valent active form [4], and second it acts as a monomer.
As might be expected, the optimum temperature for reduction by
◦
ethylene was found to be about 130–140 C. At lower temperatures
A similar result was obtained using a Cr/silica catalyst, although
the color change occurred a little more slowly.
the reduction is perhaps incomplete, and at higher temperatures
some instability in the reduced form must lower the polymerization
activity.
In similar experiments, other olefins were also effective at
reducing the Cr(VI)/silica to create a catalyst that would polymerize
Although this industrial grade of n-heptane is sometimes used
commercially, we wondered if the observed reduction could have
been due to traces of olefinic impurities rather than to the alkane
itself, since olefins are more reactive with Cr(VI)/silica-titania.
Therefore, the experiment was repeated with research grade n-
heptane, again after it was dried over molecular sieve two times.
This grade, which assayed at 98.6% n-heptane, contained no trace of
olefins. The remaining 1.4% was a mixture of branched heptane iso-
mers, dimethylcyclopentane, methylcyclohexane, and isooctane.
When the catalyst was immersed in this grade it exhibited very sim-
ilar behavior, although the color changes, which are shown in Fig. 1,
occurred a little more slowly. Within an hour the bright orange
color darkened to orange-black. After three days the catalyst was
dark green. The catalyst was allowed to stand under the n-heptane
◦
ethylene at −30 C. This includes propylene, 1-butene, 2-butene,
and isobutene. Such activity developed at lower temperatures
when the catalyst was first exposed to propylene compared to
ethylene. This suggests that propylene was more easily oxidized
than ethylene. 1-Butene was also tested as the reducing agent, and
it developed activity at perhaps even lower temperatures. These
results, and others, are summarized in Table 1.
Thus, it is clear that olefins reduce Cr(VI) easily and that
olefins with alkyl substituents are even more effective. As noted
in Scheme 1, aldehydes have been proposed as the by-product

9
4
E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
Table 2
for six days without further color change. It was then dried under
nitrogen in order to be tested for polymerization activity (experi-
ment A2).
No redox by-products were found in the n-heptane by GC–MS
analysis. This suggests that these products were still attached to
the chromium as ligands. However, wet methanol was added to the
catalyst to hydrolyze or displace whatever ligands remained on the
catalyst. The methanol solvent was then analyzed by GC–MS, and
a C7 ketone was then found.
Oxidation products detected from reaction of Cr(VI) catalysts with alkanes or aro-
matics followed by hydrolysis.
Hydrocarbon
n-Pentane
Main products
Minor products
Pentanedione
2-Pentanone
3-Pentanone
n-Hexane
2-Hexanone
3-Hexanone
2-Hexanol
Dimer,
hexanedione
3-Hexanol
In a variation of the experiment, n-heptane vapor was applied
◦
to the Cr(VI)/silica-titania catalyst as a vapor at 100 C. At this tem-
n-Heptane
Isobutane
C7 Ketone
t-Butanol
perature it took only a small amount of n-heptane to obtain the
green-black color previously observed. Again nothing was observed
in the n-heptane off-gas, but as before, a C7 ketone was observed
when the catalyst was treated with wet methanol. Table 2 summa-
rizes the results of this and similar experiments. This catalyst was
then tested for polymerization activity (experiment A3).
Cyclohexane
Cyclopentane
Cyclohexanone
Dimers
Cyclohexanedione
n-Hexane: The experiment was repeated using n-hexane instead
2
of n-heptane. A silica having a surface area of 600 m /g was impreg-
O
O
◦
nated with 3 wt% Cr and then calcined in dry air at 650 C, which
converts most of the chromium into the hexavalent state. One sam-
ple of this catalyst was soaked in n-hexane liquid for 18 h, which
caused the same darkening of the orange Cr(VI) catalyst to a black-
green color. No oxygenate redox by-products were found in the
n-hexane liquid when it was subjected to GC–MS analysis. After the
n-hexane had been decanted off, the catalyst was dried by passing
Toluene
Benzene
Benzaldehyde
Not detected
◦
dry nitrogen over it for several hours. It was then slurried at 80 C in
several mL of a solution of ethanol containing 5% H O and 0.3% HCl
2
in an attempt to hydrolyze or displace the redox products from the
chromium. After several hours of soaking, this liquid was then ana-
lyzed by GC–MS. The major redox products found were 2-hexanone
and 3-hexanone, and also 2-hexanol and 3-hexanol.
GC–MS, but no redox by-products were observed. Then the catalyst
was exposed at 100 C to a few mL of vaporized aqueous solution
containing 0.3% concentrated HCl. This off-gas was condensed and
analyzed. It was found to contain 2-hexanone and 3-hexanone, and
also 2-hexanol and 3-hexanol.
◦
In a second experiment, this same Cr(VI)/silica catalyst was
◦
exposed to n-hexane vapor in nitrogen at 100 C for a few min-
n-Pentane: The experiment was repeated, but using another
common solvent, n-pentane (purity 99.0%, dried over 13× molec-
utes. Again a darkening of the color, from orange to green, was
observed. The n-hexane off-gas was condensed and analyzed by
◦
ular sieve). The 800 C Cr(VI)/silica-titania catalyst was sealed in a
◦
Fig. 1. Cr(VI)/silica-titania catalysts reduced at 25 C in the presence of (A) n-heptane, (B) n-pentane, and (C) isobutane.

E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
95
glass vessel, then wetted and covered with the n-pentane. Once
again the orange color of the catalyst darkened to brown after
chromium. Therefore the catalyst was wet with a dilute solution
of dilute solution of water in methanol. After stirring for an hour
the methanol solution was also sampled for GC–MS analysis. Cyclo-
hexanone was found, and a small amount of cyclohexanedione as
well.
1
–2 h, indicating reduction. It darkened further over the next few
days, quickly reaching a green-black color. The color transition is
shown in Fig. 1. After 4.5 days the n-pentane was sampled for anal-
◦
◦
ysis, and the remainder was evaporated off under nitrogen at 25 C.
Cyclopentane: The experiment above was repeated, using 650 C
The resulting dry catalyst had a sky-blue color, as shown in Fig. 1.
This catalyst was then tested for ethylene polymerization activity
Cr(VI)/silica and cyclopentane. While the catalyst was fluidizing
at 100 C in dry nitrogen, a large amount (32 mL) of cyclopentane
◦
(
experiment A4).
was evaporated into the nitrogen stream and up through the cata-
lyst bed. A color change marked the reduction of the catalyst, but
no redox by-products could be found in the condensed cyclopen-
In another experiment, a catalyst containing 5% Cr on silica was
◦
calcined at 600 C and immersed in n-pentane. After about 3 h it too
turned from orange to an orange-black color, indicating the start of
reduction. It was then allowed to stand for 18 days, after which it
was entirely black. The n-pentane was again sampled and analyzed.
GC–MS analysis of the n-pentane solutions from both experi-
ments revealed no oxidation products. This again suggests that the
products were still held on the catalyst by the chromium, either
through bonding or simple coordination. Therefore, these catalysts
were then washed in a solution of 10% water in methanol, or in
the second experiment of 0.3% HCl in methanol, to liberate the oxi-
dation products by hydrolysis or displacement. GC–MS analysis of
the decanted methanol detected the presence of C5 carbonyls. Two
peaks were observed which were identified from the fragmentation
pattern as 2-pentanone (∼65%) and 3-pentanone (∼35%). Hydro-
gen on primary carbons (methyl) is thought to be less reactive
with chromate than those on secondary carbons (methylene), and
those in turn much less reactive than tertiary carbon hydrogens
◦
tane off-gas. The temperature of the bed was then lowered to 80 C
and an aqueous solution of 0.7% HCl was then evaporated into the
nitrogen stream that moved up through the bed. Analysis of the
condensed vapor detected only dimers of cyclopentane as the redox
by-products.
Isobutane: In the Phillips loop-slurry polymerization process,
which accounts for most of the world’s PE produced using
chromium catalysts, the calcined catalyst is usually stored in isobu-
tane for days or sometimes even weeks before being introduced
◦
into the reactor. Therefore a sample of Cr(VI)/silica-titania (800 C),
◦
◦
calcined at 800 C, was immersed in liquid isobutane at 25 C. After
10 min the color began to darken, turning brown, and in 1–4 h it
had turned green-black, which is an indication of a rapid reduction.
◦
Cr/silica catalyst (800 C) also went through the same color progres-
sion when exposed to isobutane, although a bit more slowly.
In one experiment the catalyst was allowed to soak in isobu-
tane for 10 days. It appeared to be green-black. The flask was then
vented, allowing the isobutane liquid to vaporize, leaving a dry
blue-gray powder. In another experiment the flask was pressur-
ized with about 3 atm of isobutane gas. The catalyst changed colors
quickly in the same way. After a week the pressure was vented,
leaving a gray-green powder.
In one case the dry catalyst was then slurried in a dilute solution
of water in methanol. After being stirred for about 1 h the methanol
solution was analyzed by GC–MS. The oxidation/hydrolysis product
was found to be t-butanol.
(
methine) [10]. This would explain why pentanal was not found.
The preference for formation of 2-pentanone over 3-pentanone
probably reflects the relative abundance of the 2 position in n-
pentane (two possible carbons) over the 3 position (only one
carbon). A small amount of what may have been pentanedione was
also found, possibly from the double oxidation of n-pentane.
The influence of titania: The presence of titania on the cata-
lyst is known to create new Cr(VI) species that are more easily
reduced [3,12,13], thus improving the activity. This is thought to
be due to the creation of strong Bronsted acid sites that, when
bonded to Cr(VI), pull electron density from the chromium. There-
fore, for comparison, the experiments above used two catalysts,
one Cr(VI)/silica-titania and the other Cr(VI)/silica. Both were cal-
In another experiment the calcined Cr/silica-titania catalyst was
◦
treated with isobutane vapor at 250 C for about 5 min to simulate
the charging of a still hot catalyst immediately after calcination.
This catalyst was a dark green, indicating that Cr(VI) had been
mostly reduced. It was then tested for ethylene polymerization
activity (experiment A9).
◦
cined at 800 C to yield an orange powder, and then submerged
in research-grade n-pentane. Both catalysts turned a dark green,
however, the change seemed to occur a little more slowly when
the catalyst was Cr(VI)/silica, indicating that titania does make the
catalyst more reactive with these hydrocarbons.
Toluene: Occasionally in commercial operations catalysts can
also be exposed to small amounts of aromatics. Therefore two sam-
ples of Cr(VI)/silica-titania (3.2% Cr, 2.5% Ti, and 1% Cr, 4.5% Ti) both
A similar result was observed with n-heptane when the cat-
◦
◦
◦
alysts (700 C Cr(VI)/silica and 700 C Cr(VI)/silica-titania) were
submerged in the liquid for up to two weeks. They also turned
dark green, indicating reduction of Cr(VI). However, the change
occurred more slowly on Cr(VI)/silica, indicating that Cr(VI)/silica-
titania was more reactive.
calcined at 850 C, were immersed in dry toluene. Both immedi-
ately turned green-black indicating a rapid reduction. They were
allowed to stand for two weeks, and then the toluene was sam-
pled and analyzed by GC–MS. In both experiments, the toluene
was found to contain a dimer of toluene, which had been produced
during the reduction. The toluene was then evaporated off under
Cylclohexane: The preferred solvent in the Phillips solution HDPE
process, used around the world with chromium catalysts since the
◦
flowing nitrogen at 25 C, leaving a dry gray-brown powder which
1
950s by various polyethylene producers, was cyclohexane, which
was later tested for polymerization activity (experiment C2).
A dilute solution of water in methanol was then added and
the mixture was stirred for 1 h. GC–MS analysis of the methanol
solution indicated that the primary oxidation product was ben-
zaldehyde in both experiments. The toluene dimer was also found
in smaller amounts, as well as small amounts of two isomers of
a dimer composed of toluene and benzaldehyde. These structures
are summarized in Table 2.
contains only methylene C H bonds. A catalyst containing 5% Cr
on silica was calcined at 650 C to produce an orange powder, indi-
cating Cr(VI). It was then immersed in dry cyclohexane and within
minutes it turned to a darker orange-brown, indicating the begin-
ning of reduction. After about 3 h it was a full brown, and within
◦
5
h it became green-brown. It was allowed to set at room tempera-
ture for 29 days, during which time it turned completely black. The
solvent was then sampled for analysis, and the remainder evapo-
rated off under flowing nitrogen at 25 C, leaving a dry green-brown
Benzene: In still another test, Cr(VI)/silica-titania catalyst, cal-
◦
◦
cined at 850 C, was submerged in benzene. The catalyst darkened
catalyst.
immediately, indicating a rapid reduction. It was allowed to set
for two weeks, during which time it turned green-black. After
three weeks the benzene was sampled and then evaporated off
GC–MS analysis of the cyclohexane solvent detected no oxida-
tion products, which again indicates that they were attached to the

9
6
E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
Table 3
Summary of polymerization results using Cr(VI) catalysts reduced by alkanes or aromatics.
Experiment/condition
Hydrocarbon reductant
Reduction treatment
Non-reduced control
6 days in C7 liquid at 25 C in light
10 min in C7 vapor at 100 C in dark
5 days in C5 liquid at 25 C in light
3 days in iC4 liquid at 25 C in dark
10 days in iC4 liquid at 25 C in light
10 days in iC4 liquid at 25 C in dark
10 min in iC4 vapor at 90 C in dark
10 min in iC4 vapor at 250 C in dark
1 h in iC4 liquid at 105 C in dark
Non-reduced control
1h in iC4 liquid at 105 C in dark
3h in iC4 liquid at 120 C in dark
Non-reduced control
13 days in toluene liquid at 25 C in light
Activity (gPE/g/h)
Melt index (g/10 min)
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
B1
B2
B3
C1
C2
None
4526
50
3.49
NA
◦
n-Heptane
n-Heptane
n-Pentane
Isobutane
Isobutane
Isobutane
Isobutane
Isobutane
Isobutane
None
◦
3019
2398
3558
1810
1749
3128
1911
3880
6173
5232
5627
4416
855
2.93
2.57
2.84
1.76
1.17
2.67
0.53
NA
1.42
1.25
0.51
5.32
2.75
◦
◦
◦
◦
◦
◦
◦
◦
◦
Isobutane
Isobutane
None
◦
Toluene
◦
Catalyst: 1% Cr(VI) on silica-titania, reaction at 105 C, 1.45 molar ethylene.
◦
◦
◦
A: 2.5% Ti, calcination at 800 C, yield to ∼2500 g/g.
B: 2.5% Ti, calcination at 800 C, yield to ∼3500 g/g.
C: 4.5% Ti, calcination at 850 C, yield to ∼3600 g/g.
◦
◦
in a nitrogen stream at 25 C leaving a green powder. Although
and at 350 C it does not remain coordinated by the chromium but
instead it is flushed away by the CO stream. A final flush in nitrogen
the color change indicates a reduction of Cr(VI), we were unable
to determine what the oxidation products were, either in the
benzene solution or in subsequent treatment with wet methanol.
◦
at 350 C removes even residual CO, which can be strongly adsorbed
◦
◦
at 25 C if the catalyst is cooled in CO. But the nitrogen flush at 350 C
results in a clean divalent catalyst containing no adsorbed oxi-
dation products. The chromium tends to be highly coordinatively
unsaturated. Such Cr(II)/silica catalysts are blue (green if cooled
in a vacuum because the Cr weakly coordinates nitrogen to yield
the blue color), and quite active for ethylene polymerization. And,
because the chromium is already reduced and vacant upon contact
with ethylene, it develops activity more quickly than Cr(VI)/silica.
When exposed to oxygen such CO-reduced catalysts display a
brilliant yellow chemiluminescence that resembles a flash of fire.
Cr(II) is cleanly converted back to Cr(VI) and the energy is released
as light. This is probably an indication that the oxidation occurs
easily with little or no molecular rearrangement. However, if the
chromium coordinates a ligand, such as CO or ethylene, there is no
chemiluminescence, even though the oxidation does proceed.
The catalysts of this study, that were reduced with alkanes or
aromatics, displayed no such chemiluminescence upon exposure to
oxygen. For example, one Cr(VI)/silica-titania catalyst was allowed
to soak in isobutane for a total of 10 days before the isobutane was
vented. The dry catalyst had a blue-gray color. It turned a yellow-
ish green when exposed to dry air, signifying partial restoration
of Cr(VI), but it did not chemiluminesce like a CO reduced cata-
lyst. In another experiment the same Cr(VI)/silica-titania catalyst
was reduced with n-pentane to yield a beautiful blue catalyst after
drying, identical in color to the CO-reduced catalysts. Upon expo-
sure to air it also turned a yellow-green color, but again it did not
chemiluminesce.
3
.3. Acceleration by light
While these tests were being conducted, a suspicion devel-
oped that the reduction of Cr(VI) was being accelerated by light.
Therefore two tests were made in which the reaction mixture
was stored in the dark, protected from ambient lighting. In one
◦
test, Cr(VI)/silica calcined at 700 C was submerged in n-heptane
and stored in the dark. It was only uncovered on occasion for
less than a minute for observation. The catalyst darkened and
after a week was found to be black, indicating that the Cr(VI) was
reduced even in the dark. However, the reduction occurred some-
what more slowly than had occurred in the presence of ambient
lighting.
In the second test, a portion of Cr(VI)/silica-titania catalyst,
◦
calcined at 800 C, was sealed in a glass tube, then wetted and
covered with isobutane liquid. The tube was then shielded with
aluminum foil to prevent exposure to light. The sample was exam-
ined 21.5 h later, and compared to a similar experiment conducted
in the presence of ambient lighting. Although the covered cata-
lyst had darkened slightly, it was clear that the reduction occurred
more slowly in the absence of light. Even after 21.5 h the catalyst
still had a dark orange color, in contrast to the un-covered sam-
ple which was considerably darker even after an hour, and turned
green-black after 21 h.
This sample was then monitored for the next 10 days, always
protected from light. It continued to darken, reaching an orange-
black color after five days. At the end of the experiment the
isobutane was vented and the dry catalyst had a blue-gray color,
which was just like the catalyst exposed to light.
Thus, light did accelerate the reduction process, although the
final reduced state appeared to be similar. The presence of titania
did not seem to be necessary for light to accelerate the reduc-
tion. However, it is not clear whether titania may have been more
efficient at harnessing the power of light than simple Cr(VI)/silica
catalysts.
The lack of chemiluminescence is probably an indication that
these catalysts still retain ligands coordinated from the oxidation
of the hydrocarbon. As noted above, these oxidation products were
usually not detected in the hydrocarbon solvent after reduction,
which also indicates that they remained coordinated. Only the addi-
tion of water/methanol solution liberated them.
3.5. Polymerization activity of alkane- or aromatic-reduced
catalysts
Some of the catalysts described above (i.e. reduced by expo-
sure to non-olefinic hydrocarbon in various ways) were then dried
and tested for polymerization activity in a laboratory reactor at
105 C. Most of these alkane- or aromatic-reduced catalysts were
active when placed in a reactor containing ethylene and isobutane
3
.4. Chemiluminescence
◦
Cr(VI)/silica can be cleanly reduced to Cr(II)/silica by carbon
◦
monoxide at 350 C [3]. The oxidation product is carbon dioxide,

E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
97
Table 4
◦
Polymerization results from Cr(VI)/silica-titania (850 C) before and after reduction
by alkane. (Each entry is an average of 5–7 runs.).
Treatment
Activity
gPE/g/h)
Melt index
(g/10 min)
(
Untreated
Isobutane
Isobutane, Air
n-Hexane
5203
4535
1349
3054
3.12
2.48
1.95
1.63
come on stream more rapidly (producing lower MW polymer)
than others [3]. In Fig. 2 the melt index obtained from various
control runs using non-reduced catalyst is plotted against polymer
yield, and one can see the gentle decline in melt index. The melt
indices from various alkane- or aromatic-reduced catalysts are
also plotted in Fig. 2, and it is immediately obvious that these are
generally below the non-reduced control curve. This means that
the melt index potential of the catalyst has been diminished. The
most severe example of this is the catalyst exposed to isobutane
Fig. 2. Catalysts reduced in isobutane tended to produce PE of higher MW (lower
melt index) for a given polymer yield.
◦
vapor at 250 C, which would be equivalent to the immediate
charging of a hot catalyst into the storage vessel, which is then
quickly filled with isobutane.
◦
diluent at 105 C. However, they often tended to exhibit diminished
Fig.
3 shows the effect of this reduction treatment on
activity. The results of these tests are listed in Table 3, where they
are compared to non-reduced catalyst control runs.
Some of the experiments in Table 3 show the loss in activity that
resulted from prolonged or intense exposure to various alkanes or
the molecular weight distribution. In another experiment, a
◦
Cr(VI)/silica-titania catalyst (800 C) was allowed to polymerize
ethylene and then stored under isobutane in the light for about
1
0 days. Then it was tested again. As expected, it had lost activity
◦
aromatics. For example, in experiments A2 and A4, storage at 25 C
during this time, and the melt index of the resultant polymer had
also dropped. The size exclusion chromatograms of the polymers
are shown in Fig. 3. Also shown is the polymer from experiment
A9, in which the catalyst was briefly treated with isobutane vapor
in simple alkanes (n-heptane or n-pentane) caused a significant loss
of activity. Similarly, in experiment C2 one can see that storage in
toluene also lowered the activity significantly.
Other catalysts were treated with isobutane, which is of spe-
cial interest because it is widely used for catalyst storage in many
commercial operations. Catalysts were treated with isobutane in
various ways in Table 3, and the results were somewhat variable.
Nevertheless, it is clear that the catalyst activity is lowered when
Cr(VI) has the opportunity to be reduced by isobutane rather than
by the more reactive ethylene. For example, in experiments A5, A6
◦
at 250 C. Each curve has been normalized according to the activity
of the catalyst, so that one may see which part of the distribution
was most affected. The change observed was significant. Isobutane
treatment caused a narrowing of the MW distribution from a loss
on the low-MW side. This would correspond to a loss of some of
the most reactive Cr(VI) sites, which suggests that they were pref-
erentially inhibited by the reduction treatment. On this particular
catalyst, it is the titania-associated sites that are known to be more
reactive, and to contribute to this low-MW region [3,12,13].
◦
and A7, Cr(VI) catalyst was stored at 25 C under liquid isobutane
for varying lengths of time. This would be similar to its treatment
in a manufacturing plant. Then each catalyst was charged to the
reactor where it simultaneously contacted isobutane and ethylene
3.6. Oxygen sensitivity after reduction
◦
at 105 C. Reduction in isobutane again seems to cause a loss in
activity.
◦
Cr(VI) catalysts are not reactive with oxygen at 25 C. If exposed
This caused some concern that the observed loss in activity
might be caused by contamination from the storage vessel. There-
fore, to avoid using a storage vessel, some catalysts were treated
to dry air the catalyst remains active, provided that the air is first
purged with nitrogen before the catalyst is introduced into the
reactor. However, alkane-reduced catalysts seemed to be quite
◦
with isobutane vapor at the end of the calcination step, at 90 C
◦
(
A8, which produced a dull yellow color) and then at 250 C (A9,
which resulted in a green color). These catalysts also lost activity,
especially the latter.
In other experiments in Table 3 (A10, B2, B3), the catalyst was
treated with isobutane in the reactor itself. That is, the catalyst was
charged to the reactor with isobutane, and allowed to stir for 1–3 h
at reaction temperature or higher in the absence of ethylene. Then
ethylene was added and the polymerization activity was recorded.
Even this treatment produced a loss in activity.
Also shown in Table 3 is the melt index of the polymer obtained,
which is an indication of the ease of flow of the molten polymer.
Melt index usually varies inversely with the molecular weight, and
is often used to monitor the character of the polymer. In the exper-
iments of Table 3, melt index tends to parallel the activity. That is,
reduction by alkanes tended to lower the melt index produced.
Melt index is also known to decrease as the amount of polymer
produced rises. That is, melt index is a function of reaction time,
more commonly expressed as the yield of polymer (g of PE pro-
duced per g of catalyst). This happens because some active sites
Fig. 3. MW distribution of polymer made with Cr/silica-titania, before and after two
different treatments with isobutane.

9
8
E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
Scheme 2. Possible oxidation of alkanes by Cr(VI)/silica catalysts.
sensitive to oxygen. During the course of this study, it was noted
that such exposure to oxygen resulted in a strong loss of activity –
beyond that caused solely by the reduction in alkane. This is shown
in Table 4, where alkane-reduced catalysts were tested and then
exposed to dry air briefly before being tested again. The melt index
potential of the catalyst seemed to parallel the activity. Exposure of
the alkane-reduced catalyst to air greatly lowered the melt index
produced, even below that of the alkane-reduced catalyst itself.
4. Conclusions
Cr(VI)-containing catalysts were found to be reduced, not only
by 1-alkenes during the course of their polymerization, but also
by alkanes, which are frequently used as solvents and diluents for
reaction and for catalyst storage. The reaction with alkanes, which
have long been considered inert, is indeed quite slow in comparison
to reaction with 1-alkenes.
◦
The −30 C data indicates that reduction by alkanes can some-
times help activate the catalyst (i.e. generate active sites) in
◦
comparison with the original Cr(VI) catalyst at −30 C. However,
3.7. Adsorption/desorption of ketones
◦
it is also clear from the 105 C polymerization data that reduction
by alkanes can diminish the activity and melt index potential, in
comparison to the the olefin-reduced catalysts. That is, catalysts
Finally, several experiments were done to better understand
the nature of ketones generated by reduction of Cr(VI) with alka-
◦
◦
reduced by olefins at 105 C tended to produce higher overall activ-
nes. A Cr(VI)/silica-titania catalyst was calcined at 800 C, then it
◦
◦
ity (at 105 C) than catalyst reduced by alkanes. Changes in the
was treated with carbon monoxide at 350 C for half an hour to
MW distribution suggest that it is the more reactive sites, those
that normally produce low-MW polymer, that are most affected by
reduction with alkanes. Possibly the by-products from alkane oxi-
dation inhibit the development of activity on those sites, compared
to reduction by olefins. Thus, the effect of prolonged contact with
alkanes can sometimes be seen in diminished activity and higher
PE molecular weight.
convert the chromium into the divalent state [3]. A final 30 min
◦
flushing in dry nitrogen at 350 C removed traces of residual CO,
◦
and then the catalyst was cooled in nitrogen to 25 C. This Cr(II) cat-
alyst is well known as a very active catalyst exhibiting a shortened
induction time, in comparison to its Cr(VI) counterpart. A sample of
◦
this catalyst was then exposed to 2-hexanone vapor at 25 C in the
amount of 2.5/Cr and then another sample was exposed to 1.0 2-
hexanone/Cr. The addition of the ketone changed the color of both
catalysts from sky-blue to a bluish grey. Both catalysts were tested
for polymerization activity, and both were found to be inactive,
indicating that the ketone had adsorbed onto the Cr and blocked
polymerization. This is not surprising.
The methine hydrogen (R CH) on isobutane, and the methylene
3
hydrogens ( CH2 ) on other alkanes, were found to be more reac-
tive than the methyl ( CH ) hydrogens, resulting after hydrolysis in
3
the corresponding secondary ketone or alcohol, or the tertiary alco-
hol. Scheme 2 shows one possible pathway in which a methylene
C
H is attacked by Cr(VI) to form a Cr(IV) species. Adding water at
Then, in order to drive off the adsorbed ketone and regener-
ate the active site, both of these catalysts were heated in flowing
this point could result in hydrolysis to the corresponding alcohol.
Alternatively, a second hydrogen shift could result in the ketone
coordinated to a Cr(II) species. This species could further react with
water to desorb the ketone, or even to oxidize the Cr(II) to Cr(IV) or
Cr(III) hydroxide and hydrogen gas.
◦
nitrogen to 300 C for 1 h. Both were then cooled in nitrogen and
retested for polymerization activity. They were still inactive. And,
unlike the original Cr(II) catalyst, they did not chemiluminescence
upon exposure to oxygen.
The exception to this view is isobutane, which has only the one
hydrogen on the reactive carbon. Thus, adding water to a catalyst
reduced by isobutane released only t-butanol.
Although the activation temperatures and the Cr and Ti load-
ings were varied over a wide range in this study, similar results
were usually obtained. Only the rate of reduction seemed to vary,
especially with titania content.
This result can only mean that the ketone did not reversibly
coordinate to the coodinatively unsaturated Cr(II) and then des-
orb upon further heating. This indicates that the ketone behaved
differently from CO, which adsorbs and desorbs reversibly with
heating and cooling. It suggests that the ketone oxidized the Cr(II)
sites to form a different species, probably containing an oxy-hexane
group, that permanently destroyed the active site. It will become
evident later that this observation is important in the interpretation
of reduction to form active sites.
Finally it is interesting to note that species B in Scheme 2 is com-
parable to the catalyst prepared above by exposing coordinatively

E. Schwerdtfeger et al. / Applied Catalysis A: General 423–424 (2012) 91–99
99
unsaturated Cr(II) to 2-hexanone vapor (last section above). The
[2] J.P. Hogan, J. Polym. Sci.: Part A-1 8 (1970) 2637–2652.
[
3] M.P. McDaniel, Adv. Catal. 33 (1985) 47–98;
ketone could not be removed by heat to regenerate the original
Cr(II) site. This indicates that the ketone probably oxidizes the ini-
tial Cr(II) site, probably back to Cr(IV) as in species A of Scheme 2.
M.P. McDaniel, Adv. Catal. 53 (2010) 123–606 (Chapter 3).
4] L.M. Baker, W.L. Carrick, J. Org. Chem. 33 (2) (1968) 616–618.
[
[5] L.M. Baker, W.L. Carrick, J. Org. Chem. 35 (3) (1969) 774–776.
[
[
6] B. Liu, Y. Fang, M. Terano, Mol. Simul. 30 (15) (2004) 963–971.
7] B. Lui, H. Nakatani, M. Terano, J. Mol. Catal. A: Chem 184 (2002) 387–
98.
Acknowledgments
3
[
[
8] H.L. Krauss, H. Schmidt, Z. Anorg. Allg. Chem. 392 (1972) 258–270.
9] R.E. Cunningham, B.H., Ashe, A. Clark, G.C. Bailey, Phillips Petroleum Co.
Research Report 2613 from 1960.
The authors wish to thank Alfred Clark. R.E. Cunningham, B.H.
Ashe, Phil M. Stricklen, Kathy S. Collins and Alan Solenberger who
performed some of the experiments described in this report.
[
10] G. Cainelli, G. Cardillo, in: K. Hafner, C.W. Rees, B.M. Trost, J.M. Lehn, P. von
Rague Schleyer, R. Zahradnik (Eds.), Reactor and Structure Concepts in Organic
Chemistry, vol. 19, Springer-Verlag, Berlin, 1984.
[
[
11] R.E. Dietz, U.S. Patent 3,887,494, issued June 3, 1975 to Phillips Petroleum Co.
12] C.E. Marsden, in: V.E.G. Poncelet, P.A. Jacobs, P. Grange, D. Delmon (Eds.), Studies
in Surface Science and Catalysis (Preparation of Catalysts V), vol. 63, Elsevier
Science Publishers B.V., Amsterdam, 1991, pp. 215–227.
References
[1] J.P. Hogan, R.L. Banks, U.S. Patent 2,825,721 to Phillips Petroleum Company,
filed August 1954, Issued March 1958, and Belgian Patent 530,617, filed January,
[13] M.P. McDaniel, M.B. Welch, M.J. Dreiling, J. Catal. 82 (1983) 118–126.
1
955, filed in the U.S. Jan 1953, S.N. 333,576, also Austrian Patent App. 864/54,
filed August 6, 1954.