Tetramethylbenzenium radical cations as major active intermediates of methanol-to-olefin conversions over phosphorous-modified HZSM-5 zeolites

Article abstract of DOI:10.1016/j.jcat.2012.12.014

The reaction intermediates of methanol-to-olefin (MTO) conversion over phosphorous-modified HZSM-5 (P-MFI) catalysts were investigated by electron spin resonance (ESR) and gas chromatography-mass spectroscopy (GC-MS). Phosphorous modification partially neutralized the strong acid sites in HZSM-5. The phosphorous compounds suppress the migration of polymethylbenzenium radical cations, lowering spin-spin interactions and thus preserved their hyperfine splitting. The ESR spectra of the organic species formed in P-MFI zeolites during MTO conversion were similar to those of the cation species generated from 1,2,4,5-tetramethylbenzene in mordenite. GC-MS of organic extracts revealed the predominant formation of polymethylbenzenes with 4-6 methyl groups. The correlation between conversion and the number of spins of tetramethylbenzenium radical cations exhibited that these radical cations might be major active intermediates of MTO conversion over P-MFI catalysts.

Full text of DOI:10.1016/j.jcat.2012.12.014

Journal of Catalysis 299 (2013) 240–248
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Tetramethylbenzenium radical cations as major active intermediates of
methanol-to-olefin conversions over phosphorous-modified HZSM-5 zeolites
Hoi-Gu Jang a, Hyung-Ki Min b, Suk Bong Hong b, Gon Seo a,
⇑
a School of Applied Chemical Engineering and Research Institute for Catalysis, Chonnam National University, Gwangju 500-757, Republic of Korea
b Department of Chemical Engineering and School of Environmental Science and Engineering, POSTECH, Pohang 790-784, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
The reaction intermediates of methanol-to-olefin (MTO) conversion over phosphorous-modified HZSM-5
(P-MFI) catalysts were investigated by electron spin resonance (ESR) and gas chromatography–mass
spectroscopy (GC–MS). Phosphorous modification partially neutralized the strong acid sites in HZSM-5.
The phosphorous compounds suppress the migration of polymethylbenzenium radical cations, lowering
spin–spin interactions and thus preserved their hyperfine splitting. The ESR spectra of the organic species
formed in P-MFI zeolites during MTO conversion were similar to those of the cation species generated
from 1,2,4,5-tetramethylbenzene in mordenite. GC–MS of organic extracts revealed the predominant for-
mation of polymethylbenzenes with 4–6 methyl groups. The correlation between conversion and the
number of spins of tetramethylbenzenium radical cations exhibited that these radical cations might be
major active intermediates of MTO conversion over P-MFI catalysts.
Received 11 October 2012
Revised 11 December 2012
Accepted 12 December 2012
Keywords:
Methanol-to-olefin conversion
HZSM-5
Phosphorous modification
Electron spin resonance
Tetramethylbenzenium radical cations
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
and propene through side-chain alkylation and paring mechanisms
[10].
Lower olefins such as ethene, propene, and butenes, industrially
important petrochemicals, are mainly produced through the ther-
mal cracking of naphtha, which is energy intensive and leads to
the emission of large volumes of carbon dioxide. Therefore, alter-
native production methods have been sought to generate lower
olefins from non-petroleum sources with reduced energy con-
sumption [1]. Methanol-to-olefin (MTO) conversion over zeolitic
acid catalysts such as HSAPO-34 (framework type CHA) and mod-
ified HZSM-5 (framework type MFI) is a promising synthetic route
[2,3], and the first commercial MTO unit was built in China in 2010
[4].
SAPO-34 has a cylindrical cage structure with molecular dimen-
sions of 6.7 ꢀ 6.7 ꢀ 10.0 Å and small 8 membered-ring (MR)
(3.8 ꢀ 3.8 Å) windows [5]. Its large cages can be used to produce
branched, aliphatic, and aromatic compounds, but the small en-
trances may allow the transfer of only small linear molecules, lead-
ing to high selectivity to lower olefins [6,7]. The formation of
hexamethylbenzene (hexaMB) in HSAPO-34 has been repeatedly
observed in situ by 13C MAS NMR or ex situ by gas chromatogra-
phy–mass spectroscopy (GC–MS) [8,9]. Electron spin resonance
(ESR) spectra with the highly resolved hyperfine splitting of hex-
aMB’s radical cations formed in HSAPO-34 cages suggest that they
are major reaction intermediates in the MTO production of ethene
ZSM-5 has a different pore system from SAPO-34: There are two
intersecting straight (5.6 ꢀ 5.3 Å) and sinusoidal (5.5 ꢀ 5.1 Å)
10MR channels that give this medium-pore zeolite notable coke
resistance when used as a solid acid catalyst [7,11]. HZSM-5 was
the first commercial catalyst used for methanol-to-gasoline pro-
duction, which has light olefins as reaction intermediates. Modifi-
cation with phosphorous neutralizes strong acid sites and thus
limits further reactions of the lower olefins produced, resulting
in a high selectivity to lower olefins [12–14]. The methylation
activity of lower olefins over HZSM-5 is also reduced by phospho-
rous modification, leading to a very high selectivity to propene
[12,15].
Unlike hexaMB in HSAPO-34, trimethylbenzene (triMB) and
tetramethylbenzene (tetraMB) formed in HZSM-5 pores have been
suggested as active MTO conversion intermediates because the rel-
atively small 10MR pore intersections of this medium-pore zeolite
may not allow the further side-alkylation of hexaMB [9]. The neg-
ligible detection of hexaMB in the product stream also indicates its
limited mobility in the HZSM-5 pores.
Polymethylbenzenes (polyMBs) generated over the acid sites of
zeolite catalysts should be radical cations by accepting protons and
serve as active reaction intermediates in the catalytic cycle of
acid-catalyzed MTO conversions as radical cations. Heptamethyl-
benzenium radical cations therefore are suggested as active inter-
mediates on Hbeta (BEA), which has relatively large pores
(6.6 ꢀ 6.7 and 5.6 ꢀ 5.6 Å) [16]. Theoretical work has also
⇑
Corresponding author. Fax: +82 62 530 1899.
E-mail address: gseo@chonnam.ac.kr (G. Seo).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.12.014

H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
241
confirmed the stabilization of polyMB’s radical cations by the
frameworks of MFI and BEA zeolites, but the small energy differ-
ences among them were not sufficient to verify the formation of
prominent species within their pores [17]. However, there is no
experimental evidence to confirm the formation of radical cations
from triMB or tetraMB during MTO conversions over HZSM-5,
although various polyMBs are simultaneously formed within the
zeolite’s medium pores.
JSM-7500F scanning electron microscope (SEM). N2 sorption
experiments were carried out on a Mirae SI nanoPorosity-XG ana-
lyzer. Temperature-programmed desorption (TPD) of ammonia
was recorded using a laboratory-made apparatus, following estab-
lished procedures [20]. The samples saturated with ammonia and
purged with helium (Shinil, 99.999%) at 150 °C before being heated
to 700 °C at 10 °C minꢁ1. The desorbed ammonia was detected by a
Balzers QMS200 mass spectrometer.
ESR is a powerful technique for detecting radical cations formed
during heterogeneously catalyzed reactions [10,18]. On the other
hand, ESR signals with no hyperfine splittings do not provide any
useful structural information. Radical cations in HZSM-5 pores re-
main close to each other, unlike in the cage-based, small-pore
material HSAPO-34, in which polyMB’s radical cation produced in
one cage cannot effectively communicate with those in another
cage. Cations’ proximity in HZSM-5 allows strong spin–spin inter-
actions between them, lowering the feasibility of observing their
hyperfine splitting.
Uptakes of o-xylene (Yakuri, 99%) and methanol (Aldrich, 99.8%)
on the P-MFI catalysts were measured using a gravimetric adsorp-
tion system equipped with a quartz spring. The samples were first
evacuated at 300 °C for 1 h. The mass gain of o-xylene was mea-
sured under 2 Torr at 90 °C for 180 min; that of methanol was mea-
sured under 32 Torr at 30 °C for 60 min. IR spectra of pyridine
adsorbed on the zeolite catalysts were recorded on a BIO-RAD
175C FT-IR spectrophotometer with a Graseby Specac in situ cell.
A self-supported catalyst wafer (10 mg) was evacuated at 500 °C
for 1 h. 1 ll pyridine (Aldrich, 99%) was injected at 50 °C and main-
This work reports the modification of HZSM-5 zeolites by phos-
phorous impregnation in an attempt to more effectively separate
from one another the aromatic reaction intermediates formed dur-
ing MTO conversions and also to partially neutralize intrazeolitic
strong acid sites to allow observation of hyperfine splitting by a
reduction of the spin–spin interactions. Powder X-ray diffraction,
scanning electron microscopy, NH3 temperature-programmed
desorption, IR spectrophotometer of adsorbed pyridine, methanol
and o-xylene uptake measurements, and ESR were performed to
characterize the phosphorous-modified HZSM-5 catalysts. The
in situ IR measurement of the organic species formed and occluded
in these catalysts during MTO conversions allowed the examina-
tion of the effect of phosphorus modification on the conversions.
The hyperfine splittings of polyMB’s radical cations formed on
the HZSM-5 catalysts were interpreted by comparing the ESR sig-
nals generated by the adsorption of the neutral forms of corre-
sponding molecules on a large-pore acidic zeolite, Hmordenite
(MOR).
tained for 30 min. With increasing desorption temperature, differ-
ential IR spectra were recorded across the range 4000–700 cmꢁ1
with a resolution of 4 cmꢁ1
.
2.3. MTO conversion
MTO conversions were carried out at an atmospheric pressure
in a continuous-flow microreactor as described elsewhere [20].
At usual runs, 0.1 g catalyst charged in the center of 1/200 quartz
tube was activated at 550 °C for 1 h and used at 350 °C and
3.0 hꢁ1 WHSV. The amounts of P(0.0)-MFI(25) and P(2.0)-MFI(25)
varied from 0.07 g to 0.25 g, in the examination of the correlation
between acidity and activity. This range was set to prevent chan-
neling due to decreasing catalyst loading and pressure build-up
by increasing. The products were analyzed by an on-line Donam
DS 6200 gas chromatograph with a CP-Volamine capillary column
(60 m ꢀ 0.32 mm) and a flame ionization detector. Conversion was
defined as the percentage of methanol consumed during MTO con-
versions; dimethyl ether was not considered as a product. The
yield of each product was calculated with respect to the molar
amount of methanol converted to a given hydrocarbon.
2. Experimental
2.1. Catalyst preparation
2.4. Analysis of occluded materials
The proton forms of two HZSM-5 zeolites with Si/Al molar ra-
tios of 25 and 75 were from Zeolyst and used as parent catalysts.
Ammonium phosphate monobasic (Wako, 98%) was impregnated
on the HZSM-5 zeolites by an incipient wetness method [19]. Sam-
ples were then stood at room temperature for 5 h to stabilize and
dried at 100 °C for 2 h. The phosphorous-modified HZSM-5 (P-MFI)
catalysts were obtained by calcination at 550 °C for 4 h. Since the
aluminum contents of the zeolites with Si/Al ratios of 25 and 75
were 0.61 mmol gꢁ1 and 0.21 mmol gꢁ1, respectively, the maxi-
mum amounts of phosphorous impregnated were adjusted to
2.0 wt% (0.65 mmol gꢁ1) or 1.0 wt% (0.32 mmol gꢁ1), sufficient to
neutralize all the acid sites present. The phosphorous impregnated
was still remained after calcination [19]. The zeolite catalysts pre-
pared here are labeled as follows: P(x)-MFI(y), where x (wt% of P
loaded) is varied between 0.0 6 x 6 2.0 and 0.0 6 x 6 1.0, and y
(Si/Al ratio) is 25 or 75, respectively.
The zeolite catalyst samples were charged in an ESR tube cou-
pled with an on–off top valve to prevent contact with the atmo-
sphere and to allow their activation at 550 °C in a 50 ml minꢁ1
argon (Shinil, 99.999%) flow for 1 h. After cooling to 350 °C, meth-
anol diluted in the argon flow was fed for a given time at similar
WHSV as the MTO conversions. The ESR tube containing the cata-
lyst powder was transferred to a JEOL JES-FA200 ESR spectrometer
within 5 min at room temperature, and ESR spectra were recorded
at X-band (9.17 GHz) with 100 kHz field modulation.
For comparison, ESR signals of polyMB’s radical cations formed
on larger pore MOR zeolite were also observed. 0.1 g MOR(15)
charged in the ESR tube was activated in an argon (Shinil,
99.999%) flow of 100 ml minꢁ1 at 550 °C for 1 h. After cooling to
ambient temperature, a small amount of polyMBs, ca. 0.1 ml, was
added and isolated from the flow by locking the on–off valve. After
heating at 200 °C, polyMB’s radical cations were formed on the
MOR(15) and their ESR spectra were recorded. All the polyMBs:
o-, m-, and p-xylenes, 1,2,3-, 1,2,4-, and 1,3,5-triMBs, 1,2,3,5-, and
1,2,4,5-tetraMBs, pentaMB, and hexaMB had purities higher than
98% and were from Aldrich. They were used as received.
MOR(15) with a Si/Al molar ratio of 15 was obtained from Tosoh
and used as a reference zeolite in ESR study.
2.2. Characterization
The hydrocarbon deposits formed on the zeolite catalysts after
the MTO conversions at 350 °C were extracted as reported else-
where [21] and further characterized by GC–MS total ion chroma-
tography (Varian CP 3800 gas chromatograph with a Varian
Powder X-ray diffraction (XRD) patterns were recorded on a
Rigaku Ultima III X-ray diffractometer with Cu K
a radiation.
Crystals’ shapes and average sizes were determined using a JEOL

242
H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
320-MSD mass selective detector) with electron impact ionization
at 70 eV. The extracted organic compounds were identified
through comparison with the NIST database [22]. The adsorption
and occlusion of hydrocarbon species on the zeolite catalysts dur-
ing the MTO conversions were also monitored using the IR spectro-
photometer described above. Self-supporting wafers of 10 mg were
activated under flowing nitrogen (100 ml minꢁ1, 99.9%, Shinil) at
500 °C for 1 h in the in situ cell. Methanol vapor was fed under a
nitrogen flow of 50 ml minꢁ1 at 350 °C using a Sinmyung KdSeries
100 syringe pump. Methanol (CH3OH, Aldrich, 99.9%) and tetra-
on P(2.0)-MFI(25). The amount of acid sites estimated from the
aluminum content was 0.61 mmol gꢁ1, though the amount of
strong acid sites determined from the TPD of ammonia was, about
a fifth, indicating that not all the aluminum atoms contributed to
the formation of strong acid sites. Phosphorous modification at
2.0 wt%, corresponding to 0.65 mmol gꢁ1, was sufficient to mask
all the acid sites of P(0.0)-MFI(25). However, about a fourth of
the strong acid sites remained on P(2.0)-MFI(25). Some of the
strong acid sites of P(0.0)-MFI(75) also remained after phospho-
rous modification. These results show that a part of the phospho-
rous compound was used to neutralize the strong acid sites, but
some formed small aggregates since the amount of phosphorous
impregnated still remained after calcination [19]. The absence of
phosphorous diffraction peaks and no phosphorous aggregates ob-
served on the external surface suggest that very small aggregates
formed in the pores of MFI(25) and MFI(75).
IR spectra of the P-MFI catalysts activated in nitrogen flow at
500 °C clearly exhibited the effect of phosphorous modification
(Fig. S3). A band at 3743 cmꢁ1 attributed to external hydroxyl
groups and that around 3600 cmꢁ1 to acidic hydroxyl groups were
definitely observed on both P(0.0)-MFI(25) and P(0.0)-MFI(75)
[23]. However, the increase of phosphorous modification gradually
decreased these bands, indicating phosphorous impregnated re-
acted with surface hydroxyl groups and considerably diminished
their intensity on P(2.0)-MFI(25) and P(1.0)-MFI(75). The partial
neutralization of strong acid sites of MFI zeolites by phosphorous
modification might be due to the reaction of phosphorous with
external hydroxyl groups, not only with acidic hydroxyl groups.
IR spectra of pyridine adsorbed on the MFI catalysts followed by
purging at 50 and 300 °C (Figs. S4 and S5) consistently show
absorption bands at 1546 and 1446 cmꢁ1, attributed to pyridinium
ions formed on Brønsted acid sites and pyridine coordinated to Le-
wis acid sites, respectively [24]. The bands at 1546 cmꢁ1 on
MFI(25) and MFI(75) after purging 50 °C definitely decreased with
increasing phosphorous modification, while those at 1446 cmꢁ1
did not show a significant decrease on both catalysts (Fig. S4).
Shoulder bands at 1446 cmꢁ1 attributed to physically adsorbed
pyridine appeared on the catalysts with high phosphorous modifi-
cation. These spectra meant that the phosphorous impregnated on
the P-MFI catalysts mainly neutralize Brønsted acid sites, while the
effect of phosphorous modification on Lewis acid sites was rela-
tively weak. The IR spectra of pyridine adsorbed on the P-MFI cat-
alysts after purging at 300 °C mainly reflected the amount of strong
acid sites. The areas of the bands at 1546 cmꢁ1 on the catalysts
deutromethanol (CD3OD, Aldrich, 99%) were fed at 0.63 l .
l minꢁ1
3. Results and discussion
Powder XRD patterns of the P-MFI catalysts prepared here
(Fig. S1) show that each sample maintained the MFI structure dur-
ing phosphorus impregnation and no reflections other than those
from the parent materials were observed, even after thermal treat-
ment at 550 °C. No noticeable change of their crystallite or mor-
phology was detected by SEM (Fig. S2).
TPD profiles of ammonia from the P-MFI catalysts (Fig. 1) show
that the temperatures at peak maxima and the areas of the ammo-
nia desorption peaks gradually decreased with increasing phos-
phorous modification [19]. The temperature at peak maximum of
P(0.0)-MFI(25) was 382 °C, which decreased to 329 °C in P(2.0)-
MFI(25), the most modified sample. A similar trend was shown
by MFI(75): The temperature at peak maximum of P(0.0)-MFI(75)
was 371 °C, which decreased to 348 °C in P(1.0)-MFI(75).
In order to discuss the catalytic role of acid sites, the TPD pro-
files were deconvoluted with three peaks, as shown in Fig. 1: The
first peak with the temperature at peak maxima (Tm) was around
250 °C, the second peak around 350 °C, and the third peak around
500 °C. Although the acid sites of zeolites were generally divided
by weak and strong acid sites, the third peak was essentially re-
quired to accomplish a high accordance between experimental
and deconvoluted profiles of the P-MFI catalysts. Table S1 lists
the deconvoluted peak areas and their temperatures at peak max-
ima. The areas of peak II and III gradually decreased with increas-
ing phosphorous modification, while those of peak I did not follow
such trends. Therefore, the formers were considered as strong acid
sites, which could react with phosphorous compounds impreg-
nated and generate polyMB’s radical cations in MTO conversions.
Ammonia desorption associated with strong acid sites de-
creased from 0.125 mmol gꢁ1 on P(0.0)-MFI(25) to 0.034 mmol gꢁ1
Fig. 1. Deconvoluted NH3-TPD profiles of (A) P(x)-MFI(25) and (B) P(x)-MFI(75) catalysts.

H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
243
of methanol adsorption on MFI(25) did not decrease with phospho-
rous modification, and the adsorbed amount was reduced by 13%
on P(2.0)-MFI(25). However, phosphorous modification consider-
ably lowered both the rate and amount of o-xylene adsorption;
adsorption was reduced 37% on P(2.0)-MFI(25), indicated that
phosphorous compounds located in the pores restricted mass
transfer. The greater decrease of o-xylene adsorption than metha-
nol adsorption suggests that the phosphorous compounds in the
pores act to obstruct the adsorption of o-xylene.
Samples’ phosphorous impregnation, pore volume, surface area,
strong acid sites, and methanol and o-xylene adsorptions are listed
in Table 1. Phosphorous impregnation reduced the pore volume:
The micropore volume of P(0.0)-MFI(25) was 0.16 cm3 gꢁ1, but that
of P(2.0)-MFI(25) decreased to 0.12 cm3 gꢁ1. Mesopore volume also
decreased from 0.25 cm3 gꢁ1 to 0.16 cm3 gꢁ1 upon phosphorous
modification. Modification with phosphorous caused similar de-
creases in P(0.0)-MFI(75). However, the decreases of mesopore
and micropore volumes in both samples were less than 25%, indi-
cating no severe blockage of the pores.
Fig. 4 exhibits the MTO conversions over the P-MFI catalysts.
Phosphorous modification less than 1.0 wt% on MFI(25) did not re-
duce the conversion, though greater phosphorous modification
considerably reduced the conversion. P(1.5)-MFI(25) showed a
conversion of 75% and P(2.0)-MFI(25), 25%. P(1.0)-MFI(75) exhib-
ited low conversion (42%) because P(0.0)-MFI(75) had fewer acid
sites than P(0.0)-MFI(25). However, none of the catalysts showed
deactivation. These results show that the decrease of strong acid
sites by phosphorous modification is responsible for the reduction
of the conversion. The negligible deactivation is likely due to the
catalysts’ unique channel system that may have inhibited the for-
mation of large molecules, the precursors of coke.
Compositions of the product streams in MTO conversions over
the catalysts (Fig. 5) did not alter with time on stream (TOS).
P(0.0)-MFI(25) allowed considerable amounts of aromatic com-
pounds and C5AC9 olefins to be produced, and their contents
slightly increased with increasing catalyst loading from 0.07 g to
0.10 g. The further reaction of lower olefins to aromatics occurred,
resulting in an increase of aromatics accompanying a decrease of
lower olefins. Ethene and propene were the main products over
P(2.0)-MFI(25), but the increase of its loading caused considerable
increases of higher olefins and aromatics. The product composi-
tions over P(0.0)-MFI(25) and P(2.0)-MFI(25) were definitely dif-
ferent when their loadings were 0.1 g, while those on both
Fig. 2. The correlation between the area of the IR band attributed to Brønsted acid
sites and the amount of strong acid sites measured from NH3-TPD.
after purging at 300 °C gradually decreased with increasing phos-
phorous modification, while those at 1446 cmꢁ1 were too weak
to discuss quantitatively. These results strongly support that phos-
phorous modification reduces the amount of strong Brønsted acid
sites, but a part of strong acid sites is retained even after excessive
phosphorous modification.
Fig. 2 shows the correlation between the area of the band at
1546 cmꢁ1 and the amount of strong acid sites: The former associ-
ated with the amount of Brønsted acid sites was determined from
the IR spectra after purging at 50 °C and the latter from the sum of
the desorbed amount of ammonia associated with peak II and III in
TPD profiles. A linear correlation between them indicated the ma-
jor type of strong acid sites on the P-MFI catalysts was Brønsted
acid which could donate protons to others.
Fig. 3 shows the uptake curves of methanol and o-xylene on the
P-MFI(25) catalysts. The large kinetic diameter of o-xylene
(0.68 nm) compared with the size of the ZSM-5 zeolite’s pores
led to high temperature, that is, 90 °C, being required to overcome
the resistance to adsorption; methanol molecules (kinetic diame-
ter, 0.37 nm) could rapidly enter the pores of the zeolite. The rate
Fig. 3. Uptake curves of (A) methanol at 30 °C and (B) o-xylene at 90 °C over P(x)-MFI(25) catalysts. The pressures of methanol and o-xylene were adjusted to 32 and 2 Torr
throughout the uptake measurements, respectively.

244
H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
Table 1
Characterization data for P-MFI catalysts.
Pore volume (cm3 gꢁ1
)
SBET (m2 gꢁ1
)
The amount of strong
Uptake amount (mg gꢁ1
)
c
Catalyst
Loading amount
of P (wt%)
acid sitesd (mmol gꢁ1
)
a
b
Vmicro
Vmeso
Methanol
o-Xylene
P(0.0)-MFI(25)
P(0.5)-MFI(25)
P(1.0)-MFI(25)
P(1.5)-MFI(25)
P(2.0)-MFI(25)
P(0.0)-MFI(75)
P(0.2)-MFI(75)
P(0.6)-MFI(75)
P(1.0)-MFI(75)
0.0
0.5
1.0
1.5
2.0
0.0
0.2
0.6
1.0
0.16
0.15
0.14
0.14
0.12
0.15
0.14
0.13
0.12
0.25
0.21
0.16
0.16
0.16
0.21
0.19
0.16
0.16
385
365
340
335
290
370
350
320
310
0.125
0.087
0.088
0.077
0.034
0.065
0.046
0.040
0.035
100.2
95.8
95.3
92.6
87.2
–
–
–
–
63.7
60.5
56.1
50.6
40.2
–
–
–
–
a
b
c
Calculated using the Dubinin–Radushkevich (DR) algorithm.
Calculated using the Barrett–Joyner–Halenda (BJH) algorithm.
Calculated using the BET equation.
d
Sum of acid sites related to peak II and III of NH3-TPD.
Fig. 4. Methanol conversion with respect to TOS in MTO conversion over (A) P(x)-MFI(25) and (B) P(x)-MFI(75) catalysts with different levels of P loading at 350 °C and
3.0 hꢁ1 WHSV.
catalysts at similar conversions were very similar. The low conver-
sion over P(2.0)-MFI(25) with a low catalyst loading was due to a
small amount of strong acid sites, and the increase of its loading in-
creased the conversion and the contents of higher olefins and aro-
matics. These results mean that the phosphorous modification only
lowers the amount of strong acid sites, not to change the reaction
mechanism of MTO conversion.
The use of ’turnover frequency’ in representing the activity of
catalysts has been recommended, but the vagueness of active sites
and intermediates in MTO conversions retards the use of turnover
frequency. Although the polyMB’s radical cations are generally
considered as active intermediates, the in situ measurement of
their numbers had not been reported yet. However, they are
formed on strong acid sites, and the number of acid sites may cor-
respond to that of active sites. Fig. S6 shows a correlation between
the conversion and the amount of strong acid sites on P(2.0)-
MFI(25), indicating the significance of strong acid sites, but the
correlation coefficient was low, 0.87. The turnover frequency of
methanol over a strong acid site could be calculated from the slope
was 0.028 molecule siteꢁ1 sꢁ1. More knowledge on the formation
of polyMB’s radical cations on strong acid sites and their catalytic
activities required to use turnover frequency as a unit denoting
the catalytic activity in MTO conversion.
Fig. 5. Product composition after MTO conversion for 10 min over P(0.0)-MFI(25)
and P(2.0)-MFI(25) catalysts at 350 °C with different amounts of catalyst loadings.

H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
245
Fig. 6. Differential IR spectra of organic species formed and occluded in (A) P(0.0)-MFI(25) and (B) P(2.0)-MFI(25) catalysts during MTO conversion using deuterated CD3OD
as reactant at 350 °C and 3.0 hꢁ1 WHSV.
Differential IR spectra of organic materials adsorbed and oc-
cluded on the P(0.0)-MFI(25) and P(2.0)-MFI(25) catalysts during
MTO conversion were recorded using deuterated CD3OD as a reac-
tant (Fig. 6). The negative bands at 3750–3300 cmꢁ1 and the
appearance of bands at 2722 and 2654 cmꢁ1 were attributed to
the conversion of OAH bonds to OAD. The bands at 2230 and
2080 cmꢁ1 indicate the accumulation of hydrocarbons containing
CAD bonds. The intense bands at 1588 and 1467 cmꢁ1 shown by
P(0.0)-MFI(25) indicate the presence of aromatic C@CAC bonds,
while these bands were very weak on P(2.0)-MFI(25) [25]. The IR
spectra show that alkyl aromatics formed during MTO conversion
over MFI(25), but phosphorous modification significantly sup-
pressed aromatic formations.
IR spectra of the organic materials accumulated during the MTO
conversion of normal methanol (Fig. S7) show similar general fea-
tures to the IR spectra recorded using CD3OD (Fig. 6). However, the
Fig. 7. GC–MS total ion chromatograms of CH2Cl2 extracts (A) from P(0.0)-MFI(25) used for 30 min and 120 min, and (B) from P(2.0)-MFI(25) catalyst used for 30 min and
120 min after MTO conversions at 350 °C and 3.0 hꢁ1 WHSV. Asterisk, mass signal of the C2Cl6 internal standard.

246
H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
Fig. 8. ESR spectra of (A) P(x)-MFI(25) and (B) P(x)-MFI(75) catalysts used in MTO conversions at 350 °C and 3.0 hꢁ1 WHSV for 5 min.
CAH stretching bands at 1600–1400 cmꢁ1 hindered differentiation
of the CAH and C@CAC bands. The HZSM-5 catalysts with fewer
strong acid sites, the result of either phosphorous modification or
reduced aluminum in the framework, lowered the formation of
polyMBs during MTO conversion.
The suppressed formation of polyMBs over the modified
MFI(25) was confirmed by GC–MS analysis of the organic extracts
from the used catalysts (Fig. 7). The organic extracts from used
P(0.0)-MFI(25) included polyMBs with 3–6 methyl groups, while
those from used P(2.0)-MFI(25) very small amount of polyMBs
with 4–6 methyl groups. The increase of reaction time from
30 min to 120 min on P(0.0)-MFI(25) considerably increased the
contents of polyMBs, while only the content of hexaMB slightly in-
creased on P(2.0)-MFI(25) with increasing reaction time. The phos-
phorous modification significantly reduced the formation of
polyMBs in MTO conversions because strong acid sites, which
acted as active sites for the formation of polyMB, were partially
neutralized. PentaMB and hexaMB were present in the extracts
at higher levels than triMBs and tetraMBs. GC–MS signals of pen-
taMB and hexaMB on HZSM-5 (Si/Al = 140) were barely decreased
after N2 flushing at 370 °C and concluded that these polyMBs did
not serve as reaction intermediates [9]. 13C labeling experiments
over the same catalyst also suggested that pentaMB and hexaMB
were less important than triMB or tetraMBs in MTO conversions
[26]. The contents of polyMBs in used P(0.0)-MFI(25) and P(2.0)-
MFI(25) after flushing with N2 flow were listed in Table S2. The
changes of the contents of polyMBs in zeolites by flushing were
definite on P(0.0)-MFI(25), indicating the high accumulation of
polyMBs on its external surface. On the other hand, the effect of
flushing on the contents of polyMBs was negligible on P(2.0)-
MFI(25). Since phosphorous compounds mainly reacted with the
hydroxyl groups on the external surface and neutralized strong
acid sites on it, polyMBs formed in P(2.0)-MFI(25) were mainly lo-
cated in the pores. The very low contents of polyMBs enhanced the
feasibility of observation for their hyperfine splittings of ESR sig-
nals because of weak spin–spin interaction.
Fig. 9. ESR spectra of (A) P(2.0)-MFI(25) and (B) P(1.0)-MFI(75) catalysts used in MTO conversions of various durations at 350 °C and 3.0 hꢁ1 WHSV.

H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
247
Fig. 10. Simulated ESR spectra of 1,2,4,5-tetraMB cation radicals using the Isotropic Simulation software. Different hfsc values were used for (A) methyl and (B) proton of the
benzene ring. (C) Comparison of the simulated ESR spectrum (dotted line) with methyl hfsc = 0.8 mT and proton hfsc = 0.1 mT with the experimental spectrum of P(2.0)-
MFI(25) (solid line) used in MTO conversions at 350 °C and 3.0 hꢁ1 WHSV for 5 min.
ESR spectra of the P-MFI catalysts were recorded after their use
in MTO conversions at 350 °C for 5 min (Fig. 8). Since all the fresh
P-MFI catalysts did not show any ESR signals with g = 2.0012, the
signals observed on the used P-MFI catalysts were attributed to or-
ganic radicals formed during the MTO conversion. The used P(0.0)-
MFI(25) and P(0.0)-MFI(75) catalysts showed symmetric ESR sig-
nals without any hyperfine splitting. The used P-MFI catalysts—
P(2.0)-MFI(25) and P(1.0)-MFI(75)—exhibited ESR signals with
hyperfine splittings, with greater splitting shown by the sample
with greater phosphorous modification.
xylenes with two methyl substituents were observed in the prod-
uct streams, they were sufficiently stable not to behave as active
intermediates under the tested reaction conditions. The remaining
polyMB’s radical cation active intermediates likely resulted from
tetraMBs, and their hyperfine splittings were very similar to that
of 1,2,4,5-tetraMB’s radical cations.
The chemical environment of the pores, the interaction between
radical cations and acid sites, and the spin–spin interaction be-
tween the radical cations excised complex influences over the
ESR signals, leading complex hyperfine splitting of the polyMB’s
radical cations in the pores of the P-MFI zeolite. ESR signals of
1,2,4,5-tetraMB’s radical cations were simulated (Fig. 10) using Iso-
tropic Simulation Software (JEOL) by varying the hyperfine split-
ting constant (hfsc) of the methyl groups and protons combined
to benzene ring with reference to reported values [27]. Both the
methyl group and the proton’s hfsc altered the hyperfine splitting
(Fig. 10A and 10B), with the proton showing a greater effect. The
simulated hyperfine splitting of 1,2,4,5-tetraMB’s radical cations
(Fig. 10C) obtained by adjusting the hfsc of the methyl group and
proton to 0.8 and 0.1 mT, respectively, highly coincided with that
The ESR signals became more intense with as MTO conversions
over the catalysts progressed (Fig. 9). After 5 min, P(2.0)-MFI(25)
exhibited a small ESR signal with hyperfine splitting; after a fur-
ther 5 min, the ESR signal increased by about four times but its
hyperfine splitting became weak. After a total of 15 min, the ESR
signal was smooth with no hyperfine splitting. The ESR signals
shown by P(1.0)-MFI(75) showed similar changes of signal inten-
sity and hyperfine splitting with TOS to those of P(2.0)-MFI(25).
At the commencement of MTO conversions, low concentrations
of active intermediates, polyMB’s radical cations were sparsely
spread in the pores with weak spin–spin interactions between
them, allowing them to preserve their hyperfine splitting. How-
ever, as the concentrations of radical cations in the pores of the
P-MFI catalysts increased with increasing TOS, spin–spin interac-
tion strengthened and so weakened the hyperfine splitting.
Since the polyMBs with over three methyl substituents could
not enter the pores of the ZSM-5 zeolite, the generation of
polyMB’s radical cations was impossible. However, the 12MR pores
of MOR(15) were sufficiently large for the adsorption of polyMBs,
even including hexaMB. Fig. S8 compares the ESR spectra of vari-
ous polyMB’s radical cations generated on MOR(15) by the adsorp-
tion of polyMBs with subsequent heating. The different polyMB’s
radical cations showed different hyperfine splitting according to
the number and position of methyl groups substituted. Radical cat-
ions of xylenes and tetraMBs exhibited similar hyperfine splitting
with a steep line at the center position. The outer lines showed
weak intensities compared with the inner lines on the hyperfine
splittings of triMB’s, pentaMB’s, and hexaMB’s radical cations.
However, the signals of polyMB’s radical cations formed on the
P-MFI catalysts in the MTO conversions clearly differed from those
of triMB’s, pentaMB’s, and hexaMB’s radical cations. Therefore,
these polyMBs could be excluded from the pool of active interme-
diates of the MTO conversion over the P-MFI catalysts. Since
Fig. 11. The variations of the conversion of methanol and the number of spins
generated over P(2.0)-MFI(25) in MTO conversion at 350 °C and 3.0 hꢁ1 WHSV.

248
H.-G. Jang et al. / Journal of Catalysis 299 (2013) 240–248
of the polyMB’s radical cations generated on P(2.0)-MFI(25) during
the MTO conversions.
Acknowledgments
Since the ESR signal with a hyperfine splitting of tetraMBs was
observed even on P(2.0)-MFI(25) in a very short TOS, it is difficult
to assign the chemical structure of active intermediates in MTO
conversions. Fig. 11 exhibits the variations of the conversion of
methanol and the number of spins generated with TOS. The con-
version was zero at an early stage of MTO conversions due to so-
called induction period, while it increased to 13% at 5 min TOS.
The conversion increased to 23% at 10 min TOS, but did not in-
crease further in these experimental conditions. The number of
spins generated also increased similar to the conversion to
10 min TOS, and the hyperfine splitting of tetraMBs was retained.
However, the number of spin drastically increased after 15 min
TOS with continuing MTO conversions, but the hyperfine splitting
disappeared after 15 min TOS. The further reaction of polyMBs
with methanol produced many kinds of polycyclic aromatic hydro-
carbons (PAHs) and cokes in zeolite pores, and the ESR signals lost
their hyperfine splitting because of high spin–spin interaction.
Many PAHs including various polyMBs involve MTO conversion
as active intermediates, but the simultaneous increases of the
conversion and the number of spins with the hyperfine splitting
of tetraMB suggest that tetraMB’s radical cations generated on
P(2.0)-MFI(25) may act as major active intermediates in MTO
conversion at these experimental conditions.
This work was supported by the Priority Research Centers pro-
gram (2009-0094055) through the National Research Foundation
of Korea funded by the Ministry of Education, Science, and
Technology.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2012.12.014.
References
[1] M. Stöcker, Micropor. Mesopor. Mater. 29 (1999) 3.
[2] J.F. Haw, D.M. Marcus, Top. Catal. 34 (2005) 41.
[3] J.Q. Chen, A. Bozzano, B. Glover, T. Fuglerud, S. Kvisle, Catal. Today 106 (2005)
103.
[4] U. Olsbye, S. Svelle, M. Bjørgen, P. Beato, T.V.W. Janssens, F. Joensen, S. Bordiga,
K.P. Lillerud, Angew. Chem. Int. Ed. 51 (2012) 2.
[5] M. Salmasi, S. Fatemi, A.T. Najafabadi, J. Ind. Eng. Chem. 17 (2011) 755.
[6] B.P.C. Hereijgers, F. Bleken, M.H. Nilsen, S. Svelle, K.-P. Lillerud, M. Bjørgen, B.M.
Weckhuysen, U. Olsbye, J. Catal. 264 (2009) 77.
[7] H.-G. Jang, H.-K. Min, J.K. Lee, S.B. Hong, G. Seo, Appl. Catal. A: Gen. 437–438
(2012) 120.
[8] J.F. Haw, W. Song, D.M. Marcus, J.B. Nicholas, Acc. Chem. Res. 36 (2003)
317.
[9] M. Bjørgen, S. Svelle, F. Joensen, J. Nerlov, S. Kolboe, F. Bonino, L. Palumbo, S.
Bordiga, U. Olsbye, J. Catal. 249 (2007) 195.
4. Conclusions
[10] S.J. Kim, H.-G. Jang, J.K. Lee, H.-K. Min, S.B. Hong, G. Seo, Chem. Commun. 47
(2011) 9498.
[11] P. Sazama, B. Wichterlova, J. Dedecek, Z. Tvaruzkova, Z. Musilova, L. Palumbo,
S. Sklenak, O. Gonsiorova, Micropor. Mesopor. Mater. 143 (2011) 87.
[12] N. Rahimi, R. Karimzadeh, Appl. Catal. A: Gen. 398 (2011) 1.
[13] K.W. Song, G. Seo, C.-H. Shin, Korean Chem. Eng. Res. 49 (2011) 521.
[14] W.W. Kaeding, S.A. Butter, J. Catal. 61 (1980) 155.
[15] J. Liu, C. Zhang, Z. Shen, W. Hua, Y. Tang, W. Shen, Y. Yue, H. Xu, Catal.
Commun. 10 (2009) 1506.
[16] M. Bjørgen, U. Olsbye, S. Svelle, S. Kolboe, Catal. Lett. 93 (2004) 37.
[17] H. Fang, A. Zheng, J. Xu, S. Li, Y. Chu, L. Chen, F. Deng, J. Phys. Chem. C 115
(2011) 7429.
[18] A. Lund, M. Shiotani, S. Shimada, Principles and Applications of ESR
Spectroscopy, Springer, New York, 2011. p. 273.
[19] G. Seo, R. Ryoo, J. Catal. 124 (1990) 224.
[20] K.Y. Lee, H.-J. Chae, S.-Y. Jeong, G. Seo, Appl. Catal. A: Gen. 369 (2009) 60.
[21] H.-K. Min, M.B. Park, S.B. Hong, J. Catal. 271 (2010) 186.
[22] NIST Chemistry WebBook, <http://webbook.nist.gov/chemistry/.
[23] T. Blasco, A. Corma, J. Martínez-Triquero, J. Catal. 237 (2006) 267.
[24] X. Gao, Z. Tang, H. Zhang, C. Liu, Z. Zhang, G. Lu, D. Ji, Korean J. Chem. Eng. 27
(2010) 812.
The modification of HZSM-5 zeolite with phosphorous partially
neutralized its strong acid sites. Aggregates of phosphorous com-
pounds located in the pores impeded the mass transfer of aromatic
molecules. The reduction of the strong acid sites enhanced selec-
tivity for lower olefins by suppressing their further reaction. Or-
ganic extracts from the used P-MFI catalysts contained various
polyMBs: a small amount of triMBs, more tetraMBs, and high levels
of pentaMB and hexaMB. Phosphorous-modified P-MFI catalysts
when used in MTO conversions for a short time exhibited ESR sig-
nals of polyMB’s radical cations with hyperfine splittings, because
the cations’ low concentration in the pores weakened their spin–
spin interactions. Hyperfine splitting differed from those of pen-
taMB’s and hexaMB’s radical cations, but was similar to that of
1,2,4,5-tetraMB’s radical cations generated on MOR(15). Consider-
ing the simultaneous increase of the conversion and the number of
spins with the hyperfine splittings of tetraMB’s radical cations sug-
gested that these radical cations formed during MTO conversions
acted as active intermediates over the P-MFI catalysts rather than
hexaMB in the cages of SAPO-34.
[25] J.W. Park, S.J. Kim, M. Seo, S.Y. Kim, Y. Sugi, G. Seo, Appl. Catal. A: Gen. 349
(2008) 76.
[26] S. Svelle, F. Joensen, J. Nerlov, U. Olsbye, K.-P. Lillerud, S. Kolboe, M. Bjørgen, J.
Am. Chem. Soc. 128 (2006) 14770.
[27] R.M. Dessau, S. Shih, E. Heiba, J. Am. Chem. Soc. 92 (1970) 412.