Photocatalytic redox-combined synthesis of L-pipecolinic acid from L-lysine by suspended titania particles: Effect of noble metal loading on the selectivity and optical purity of the product

Article abstract of DOI:10.1016/S0021-9517(03)00049-6

Photocatalytic (> 300 nm) conversion of L-(S)-lysine (L-Lys), in its neutralized aqueous solution, into L-pipecolinic acid (L-PCA) under deaerated conditions at 298 K was investigated in detail using suspended TiO₂ powders (Degussa P-25, Ishihara ST-01, and HyCOM TiO₂) loaded with platinum (Pt), rhodium (Rh), or palladium (Pd). A common feature of the results of experiments using a wide variety of metal-loaded TiO₂ photocatalysts is that the rate of PCA formation (r_PCA) was greatly reduced when higher optical purity of PCA (OP_PCA), i.e., enantio excess of the L-isomer of PCA, was obtained; higher r_PCA was achieved by the use of Pt-loaded TiO₂ powders, while these powders gave relatively low OP_PCA. Selectivity of PCA yield (S _PCA), i.e., amount of PCA production based on L-Lys consumption, also tended to increase with decrease in OP_PCA, giving a master curve in the plots of OP_PCA versus S_PCA. Among the TiO₂ powders used in this study, HyCOM TiO₂ showed relatively high OP_PCA and S_PCA but not optimum S _PCA and OP_PCA simultaneously. In order to interpret such relations, the mechanism of stereoselective synthesis of the L-isomer of PCA (L-PCA) was investigated using isotope-labeled α-¹⁵N-L-lysine with quantitative analysis of incorporation of ¹⁵N in PCA and ammonia (NH₃), a by-product. It was observed for several photocatalysts that the ¹⁵N proportion (P₁₅) in PCA was almost equal to OP_PCA, suggesting that oxidative cleavage by photogenerated positive holes of the ε-amino moiety of L-Lys gave optically pure L-PCA through retention of chirality at the α-carbon in the presumed intermediate, a cyclic Schiff base (α-CSB), which undergoes reduction by photoexcited electrons into PCA. From P₁₅ in NH ₃ and PCA, the selectivity of oxidation between α and ε-amino groups in L-Lys by photoexcited positive holes (h⁺) and the efficiency of reduction of α-CSB (produced via ε-amino group oxidation to give optically pure PCA) and ε-CSB (produced via α-amino group oxidation to give racemic PCA) by photoexcited electrons (e^-) were calculated. The former was found to be independent of the kind of photocatalyst, especially the loaded metal, while the latter was influenced markedly only by the loaded metal. It was clarified that OP _PCA and S_PCA obtained for various TiO₂ powders used in the present study were strongly governed by the reduction stage, i.e., the efficiency of reduction of two types of CSB. When S_PCA was relatively low, photocatalysts, favoring the reduction of α-CSB rather than ε-CSB, gave higher OP_PCA but lower S_PCA, since some ε-CSB remained unreduced to give racemic PCA. In contrast, at higher S_PCA, both CSBs were reduced nonselectively and OP_PCA was found to be determined mainly by the selectivity in the oxidation stage. The relatively low yield of molecular hydrogen (H₂) when higher S _PCA was achieved is consistent with the mechanism in which H ₂ liberation occurs instead of the reduction of CSBs by e ^-. Thus, the general tendency of plots between OP_PCA and S_PCA could be explained by the above-described redox-combined mechanism of photocatalysis.

Full text of DOI:10.1016/S0021-9517(03)00049-6

Journal of Catalysis 217 (2003) 152–159
www.elsevier.com/locate/jcat
Photocatalytic redox-combined synthesis of L-pipecolinic acid from
L-lysine by suspended titania particles: effect of noble metal loading
on the selectivity and optical purity of the product
Bonamali Pal,a Shigeru Ikeda,a Hiroshi Kominami,b Yoshiya Kera,b and Bunsho Ohtani a,∗
a
Catalysis Research Center, Hokkaido University, Sapporo 060-0811, Japan
b
Faculty of Science and Engineering, Kinki University, Kowakae 3-4-1, Higashiosaka, Osaka 577, Japan
Received 8 October 2002; revised 23 January 2003; accepted 29 January 2003
Abstract
Photocatalytic (> 300 nm) conversion of L-(S)-lysine (L-Lys), in its neutralized aqueous solution, into L-pipecolinic acid (L-PCA) under
deaerated conditions at 298 K was investigated in detail using suspended TiO powders (Degussa P-25, Ishihara ST-01, and HyCOM TiO )
2
2
loaded with platinum (Pt), rhodium (Rh), or palladium (Pd). A common feature of the results of experiments using a wide variety of metal-
loaded TiO photocatalysts is that the rate of PCA formation (r ) was greatly reduced when higher optical purity of PCA (OP ),
2
PCA
PCA
i.e., enantio excess of the L-isomer of PCA, was obtained; higher r
was achieved by the use of Pt-loaded TiO powders, while these
PCA
2
powders gave relatively low OP
. Selectivity of PCA yield (S
), i.e., amount of PCA production based on L-Lys consumption, also
PCA
PCA
tended to increase with decrease in OP
, giving a master curve in the plots of OP
versus S
. Among the TiO powders used in this
PCA 2
PCA
PCA
study, HyCOM TiO showed relatively high OP
and S
but not optimum S
and OP
simultaneously. In order to interpret such
2
PCA
PCA
PCA
PCA
15
relations, the mechanism of stereoselective synthesis of the L-isomer of PCA (L-PCA) was investigated using isotope-labeled α- N-L-lysine
15
with quantitative analysis of incorporation of N in PCA and ammonia (NH ), a by-product. It was observed for several photocatalysts that
3
15
the N proportion (P ) in PCA was almost equal to OP
, suggesting that oxidative cleavage by photogenerated positive holes of the
15
PCA
ε-amino moiety of L-Lys gave optically pure L-PCA through retention of chirality at the α-carbon in the presumed intermediate, a cyclic
Schiff base (α-CSB), which undergoes reduction by photoexcited electrons into PCA. From P in NH and PCA, the selectivity of oxidation
15
3
+
between α and ε-amino groups in L-Lys by photoexcited positive holes (h ) and the efficiency of reduction of α-CSB (produced via ε-amino
group oxidation to give optically pure PCA) and ε-CSB (produced via α-amino group oxidation to give racemic PCA) by photoexcited
electrons (e ) were calculated. The former was found to be independent of the kind of photocatalyst, especially the loaded metal, while the
−
latter was influenced markedly only by the loaded metal. It was clarified that OP
and S
obtained for various TiO powders used
PCA 2
PCA
in the present study were strongly governed by the reduction stage, i.e., the efficiency of reduction of two types of CSB. When S
was
PCA
relatively low, photocatalysts, favoring the reduction of α-CSB rather than ε-CSB, gave higher OP
but lower S
, since some ε-CSB
PCA
PCA
remained unreduced to give racemic PCA. In contrast, at higher S
, both CSBs were reduced nonselectively and OP
was found to
PCA
PCA
be determined mainly by the selectivity in the oxidation stage. The relatively low yield of molecular hydrogen (H ) when higher S
was
2
PCA
−
achieved is consistent with the mechanism in which H liberation occurs instead of the reduction of CSBs by e . Thus, the general tendency
2
of plots between OP
and S
could be explained by the above-described redox-combined mechanism of photocatalysis.
PCA
PCA
2003 Elsevier Science (USA). All rights reserved.
Keywords: Photocatalysis; Redox-combined mechanism; Metal-loaded titania particles; L-Lysine; L-Pipecolinic acid; Schiff-base intermediate
1. Introduction
of medicines and/or biologically active chemical com-
pounds [1]; for example, a commercially available local
anesthetic, L-N-propylpipecolinic acid 2,6-xylidide, is syn-
thesized from L-(S)-PCA (L-PCA) [2]. In order to obtain
optically active L-PCA, deaminocyclization of L-(S)-lysine
(L-Lys) is a preferable green chemistry process, because the
optically pure, but cheap, starting material is readily avail-
able and only ammonia is liberated as a by-product in the
Pipecolinic acid (PCA) is one of the most important in-
termediate compounds for the syntheses of a wide range
*
Corresponding author.
E-mail address: ohtani@cat.hokudai.ac.jp (B. Ohtani).
0021-9517/03/$ – see front matter
doi:10.1016/S0021-9517(03)00049-6
2003 Elsevier Science (USA). All rights reserved.

B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
153
reaction stoichiometry. Several routes of stoichiometric pro-
duction of L-PCA from L-Lys have been reported [3–5],
but they are noncatalytic and multistep routes. In 1990, we
reported a one-step photocatalytic synthesis of L-PCA by
photoirradiation of a deaerated aqueous solution of L-Lys
with suspended particulate semiconductor photocatalysts,
such as titanium(IV) oxide (TiO2) or cadmium(II) sulfide
(CdS) [6,7]. This photocatalytic synthesis proceeds under at-
mospheric pressure at room temperature and needs no pro-
tection of the functional groups in L-Lys. To the best of our
knowledge, this is a unique example of a catalytic one-step
synthesis of L-PCA from L-Lys and a greener process com-
pared with the conventional synthetic methods.
Photocatalysis −is a redox reaction driven by photoex-
cited electrons (e ) in the conduction band and simulta-
neously generated positive holes (h+) in the valence band
of semiconducting materials when photoirradiated with the
light of energy greater than the band gap between the va-
lence and the conduction bands [8,9]. Generally speaking,
in the presence of molecular oxygen (O2) under ordinary
atmospheric conditions, the target of reduction by e− is
limited to O2 and the proposed resulting products, super-
oxide anion [10], hydroperoxy [11], and/or hydroxyl [12]
radicals, are all oxidants. Therefore, the photocatalytic re-
action under aerated conditions is oxidation of substrates
by both h+ and the oxidants derived from e−, and this
is one of the main reasons for the potential environmen-
tal applications of photocatalysis, i.e., complete oxidation
and mineralization of organic pollutants in air and/or wa-
ter [13,14]. In order to apply this photocatalytic oxidation
to organic synthesis, it is necessary to avoid overoxidation
(mineralization) of substrates; several examples of photocat-
alytic oxidative conversion of organic substrates have been
reported [15–18]. On the other hand, when the photocat-
alytic reactions are carried out under deaerated conditions,
both reduction and oxidation by e− and h+, respectively, can
be utilized for conversion of organic substrates; for example,
transfer hydrogenation of nitrogen-containing compounds
along with the oxidation of alcohols [19–21] has been re-
ported. Similar to the reported photocatalytic cyclization of
α, ω-diamines [22], the proposed mechanism of the above-
noted photocatalytic deaminocyclization of L-Lys involves
successive oxidation of one of the amino groups in L-Lys
into imines by h+, hydrolysis of the imines into aldehyde
or ketone, condensation of them with the remaining amino
group into a cy−clic Schiff base (CSB), and reduction of CSB
into PCA by e [6], as shown in Scheme 1. Such combina-
tion of reduction and oxidation is one of the characteristic
features of photocatalytic PCA synthesis, since equimolar
oxidation and reduction by positive holes and electrons, re-
Scheme 1. Proposed mechanism of the photocatalytic N-cyclization of L-Lys on platinized TiO powders.
2

154
B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
spectively, proceed separately in the ordinary photocatalytic
reaction systems,+giving different reduction and oxidation
products. Since h and e− produced in a particle must be
consumed on the surface of each particle in the photocat-
alytic reaction by semiconductor particles, the sites for ox-
idation and reduction might be located nearby, enabling in-
duction of a redox-combined reaction. Several examples of
such redox-combined photocatalytic reactions have been re-
ported [23–29].
and rhodium chloride (RhCl3) (Wako Pure Chemicals) were
used as received for the metal loading. L-Lys hydrochlo-
ride (Wako) and α-15N–L-Lys dihydrochloride (99%, Cam-
bridge Isotope Laboratories) were used in the photocatalytic
reaction after neutralization by sodium hydroxide (NaOH).
Benzaldehyde, acetonitrile, and an amino acid-derivatizing
agent, BSTFA (N,O-bis(trimethylsilyl)trifluoroacetamide,
GL Sciences), were used without further purification.
Optical purity of the product PCA (OPPCA) was found
to depend on the kind of photocatalyst; TiO2 and CdS-
based particulate photocatalysts gave L excess and almost
racemic PCA, respectively [6]. This dependence of OPPCA
has been attributed to the difference in the position, i.e.,
which amino group in L-Lys undergoes preferential oxida-
tive attack by h+; as shown in Scheme 1, oxidation of an
α-amino group into α-keto acid leads to the loss of chiral-
ity, whereas oxidation of an ε-amino group has no effect on
the optical activity. Infrared spectroscopic analyses of the
behavior of L-Lys adsorbed on thin film photocatalysts sug-
gested that h+ in CdS can oxidize only the α-amino group,
while TiO2 oxidizes the ε-amino group predominantly [30].
Therefore, in order to achieve high OPPCA in the photocat-
alytic synthesis of PCA from L-Lys, the use of a TiO2-based
photocatalyst is preferable, though we have not been able to
obtain optically pure PCA by TiO2 photocatalysis due to the
partial contribution of the α-amino group oxidation route,
and further regulation is therefore needed for application of
this photocatalytic process to practical organic synthesis.
The above results lead to a strategy for producing
optically active PCA through the preferential oxidation
of an ε-amino group in L-Lys by changing the surface
physicochemical properties of the catalysts with the loading
of cocatalysts. In this paper, we describe the results of a
study on the effects of various TiO2 powders and cocatalysts
on photocatalytic PCA synthesis. We have attempted to
provide an overview for the correlation and improvement of
OPPCA, selectivity for PCA production, and PCA yield of
the above-noted reaction systems with focus on the reaction
mechanism, which was investigated by measurement of
isotopic distributions of nitrogen (14N and 15N) in NH3
and PCA using α-15N–L-lysine (α-15N–L-Lys) as a starting
material.
2.2. Metal loading
Loading of noble metals onto the TiO2 powders was
achieved by impregnation and subsequent H2 reduction
as follows. To an aqueous suspension (50–100 cm3) a
calculated amount of TiO2, a metal chloride solution (in
most cases ca. 1 mmol dm−3), was added dropwise for 1 h
with magnetic stirring at room temperature. The mixture was
gently stirred at ca. 353 K, evaporated to dryness, and then
further dried at 383 K overnight in an oven. The powder thus
obtained was ground in a mortar to produce fine powder
and placed in a quartz cylindrical vessel. The vessel was
heated in an electric furnace up to 573 K with an argon (Ar)
flow through a quartz tube inserted in the vessel and kept
at this temperature for 15 min. Then a flow of hydrogen
(H2) (40 cm3 min−1) was introduced into the vessel with
increase in temperature up to 773 K, and the vessel was
kept at this temperature for 3 h. The vessel was inclined and
rotated at ca. 70 rpm throughout the reduction process to
avoid heterogeneity. In all catalyst preparation procedures,
the content of metal was 2%, calculated as the weight ratio
of loaded metal to TiO2.
2.3. Photocatalytic reaction and product analysis
A metal-loaded catalyst (50 mg) was suspended in an
aqueous solution (5.0 cm3) containing L-Lys (100 µmol) and
photoirradiated by a high-pressure mercury arc (Eiko-sha,
400 W) under Ar. The photoirradiation was performed
through a cylindrical Pyrex glass filter and a glass reaction
tube (18 mm in diameter and 180 mm in length) so that light
of wavelength > 300 nm reached the suspension. The tem-
perature of the suspension during photoirradiationwas main-
tained at 298 0.5 K by the use of a thermostated water bath.
For a rough estimation of photonic efficiency of the photo-
catalytic reaction, incident photon flux (1500 µmolh−1 at a
wavelength between 300 and 400 nm for the irradiated area
of the reaction tube, 5.1 cm2) was estimated by total emis-
sion energy of the mercury arc at the sample position and
a spectrum measured using a Molectron Power Max 5200
laser power meter and a Hamamatsu Photonics C7473-36
photonic multichannel analyzer (resolution: 0.8 nm), respec-
tively.
2. Experimental and methods
2.1. Materials
Commercial TiO2 samples, Degussa P-25 (ca. 50 m2 g−1
,
a mixture of anatase and rutile crystallites), Ishihara ST-01
(320 m2 g−1, mainly anatase), and synthesized HyCOM
(Hydrothermal Crystallization in Organic Media) [31,32],
After irradiation for 1 h, a portion (0.2 cm3) of the gas
phase of the sample was withdrawn with a syringe and
subjected to gas chromatographic analysis (GC, Shimadzu
GC-8A with an MS-5A column (GL Sciences) and a TCD
which was calcined at 823 K for 1 h under air (67 m2 g−1
,
anatase), were used in this study. Metal salts such as chloro-
platinic acid (H2PtCl6 · 6H2O), palladium chloride (PdCl2),

B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
155
detector) of molecular hydrogen (H2). The yield of enan-
tiomers of PCA, as well as the amount of unreacted L-Lys,
was measured by high-performance liquid chromatography
(HPLC, Shimadzu LC-6A equipped with a Daicel Chiral-
Pak MA(+) column and an ultraviolet absorption detector).
These product analyses have been reported in detail else-
where [33,34].
Derivatization of NH3, a by-product, into benzylamine
(BA) was performed as follows. To 5 cm−3 of a reaction
mixture containing products (NH3 and PCA) and unreacted
L-Lys (or α-15N–L-Lys) as well as a metal-loaded catalyst,
neat benzaldehyde (1 mmol) was injected through a rubber
septum. The suspension was vigorously stirred under Ar
for 1 h and subsequently stirred under H2 for 1 h. After
acidification of the suspension with hydrochloric acid to
convert benzylamine into its water-soluble hydrochloride
salt, unreacted benzaldehyde was removed by extraction
with diethyl ether. The aqueous layer was evaporated to
dryness to yield a yellowish residue. A portion of the
residue was again dissolved in water and was analyzed by
HPLC with a Daicel Crownpak CR(+) column (for BA)
or a Daicel ChiralPak MA(+) column (for PCA). GC-mass
spectroscopy (MS) measurements were performed with a
Shimadzu QP5050A equipped with a fused silica capillary
(J&W DB-1). A small amount of the yellowish residue was
treated with BSTFA in acetonitrile in a sealed glass tube
under Ar, heated at 423 K for 2.5 h, and injected. MS
patterns of fractions were recorded in chemical ionization
(CI with isobutene) mode.
Fig. 1. Rate (r
), selectivity (S
), and optical purity (OP
) of
PCA
PCA
PCA
photocatalytic N-cyclization of L-Lys into PCA on metal-loaded P-25
(circles), ST-01 (triangles), and HyCOM TiO (squares). Open, half-filled,
and filled symbols denote Pt-, Rh-, and Pd-loaded samples, respectively.
2
SPCA (ca. 70%) was achieved by the use of Pt-loaded
HyCOM-TiO2. This highest rPCA corresponded to photonic
efficiency of 11% on the assumption of a two-electron
process as discussed later. It is notable that almost complete
consumption of L-Lys was achieved by 0.5-h irradiation
of this photocatalyst, while other TiO2 powders required
an hour or more of irradiation. However, with respect to
OPPCA, the photocatalytic performance of these Pt-loaded
TiO2 powders was poor in comparison with that of other
metal-loaded samples.
3. Results and discussion
3.1. Photocatalytic activity of metal-loaded TiO2 powders
Fig. 1 shows rate of PCA formation (rPCA), optical
As described before, the present photocatalytic reaction
has been proved to proceed via a+combination of oxidation
of L-Lys with positive holes (h ) and reduction of CSB
intermediates with electrons (e−), as shown in Scheme 1. It
h−as been proposed that Pt loaded on photocatalysts captures
or other reaction intermediates [33]. In the case of TiO2,
loaded Pt should work as a sole reduction site, because
bare TiO2, without modification with Pt, has poor ability
for reduction of adsorbed substrates except for suitable
electron acceptors such as molecular oxygen. Actually, the
photocatalytic activity of bare TiO2 powders in the present
system was negligibly small (data not shown). Therefore,
it is thought that the functions of other loaded metals such
as Rh and Pd used in the present study are similar to those
of Pt, a catalytic site not for the oxidation of L-Lys but for
the reduction of CSBs by e−, and the observed dependence
of photocatalytic performance on the kind of loaded metal
is thought to be due to the difference in catalytic abilities of
these metals for CSB reduction as discussed later.
purity of PCA (OPPCA), i.e., enantio excess percentage of
the L-isomer of PCA, and selectivity of PCA yield (SPCA
)
based on L-Lys consumption for photocatalytic reaction
of L-Lys in aqueous suspensions of various metal-loaded
TiO2 powders. A preliminary experiment showed that TiO2
loaded with nickel, copper, or silver gave appreciable but
very low photocatalytic activity for PCA production (data
not shown), and noble metals, Pt, Pd and Rh, were used as
representative cocatalysts in this study. All of the powders
used in this study gave PCA from L-Lys, and a wide
variety of photocatalytic activities was observed depending
on the kind of TiO2 and the loaded metal. A common
feature was that rPCA was greatly reduced when higher
OPPCA and lower SPCA were obtained, e.g., Pd-loaded P-25
exhibited highest OPPCA but lowest rPCA and SPCA. Similar
photocatalytic activity, relatively high OPPCA but low rPCA
and SPCA, was observed for the other Pd-loaded samples.
On the other hand, when any of the TiO2 powders were
e
and acts as a cathode for the reduction of protons
used, loading of Pt tended to induce high rPCA and SPCA
.
The highest rPCA (ca. 80 µmol h−1) and relatively high

156
B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
significant influence on SPCA and OPPCA. This relation
of SPCA and OPPCA is discussed later on the basis of
mechanistic findings.
3.2. Mechanistic insights into photocatalytic reaction of
L-Lys
As shown in Scheme 1, the overall process of photo-
catalytic production of PCA from L-Lys consists of three
st+eps: oxidation of one of the amino groups of L-Lys with
h
to yield imines, hydrolysis of imines into α-keto acid
and δ-aldehyde along with the release of ammonia (NH3)
followed by intramolecular condensation between residual
amino groups and the carbonyl group into two types of cyclic
Schi−ff base intermediates, and reduction of CSBs formally
by e along with addition of protons to yield the final prod-
uct, PCA. Thus, observed OPPCA and SPCA must be gov-
erned by the selectivities of both oxidation (which amino
group in L-Lys is oxidized) and reduction (efficiency in re-
duction of CSBs). In order to interpret the observed differ-
ence in photocatalytic performances of several metal-loaded
TiO2 powders, the selectivity in each stage should be evalu-
ated separately, and we have therefore tried to determine the
isotopic distributions of nitrogen (14N and 15N) in NH3 (as
a form of benzylamine) and PCA using α-15N–L-lysine as a
starting material.
Fig. 3 shows parts of GC-MS patterns of PCA synthe-
sized photocatalytically from L-Lys and α-15N–L-Lys and
those of BA prepared by postirradiation condensation be-
tween NH3 and benzaldehyde followed by H2 reduction
(Scheme 2). The data obtained using Pt-loaded P-25 are
shown in this figure. Peaks at m/z 274 and m/z 252 are
assigned to parent peaks ([M + 1]+) of bis(trimethylsilyl
(TMS)) derivative of PCA and BA, respectively. It is clear
by comparison of MS patterns of PCA and BA synthesized
from α-15N–L-Lys with those from L-Lys that both products
gave fragments of one atomic unit larger molecular weight
together with standard fragments, indicating the incorpora-
tion of 15N atoms in PCA and BA molecules. Based on the
assumption that the isotopic fractions (m/z 274:275:276 =
100:24:10 for PCA and m/z 252:253:254 = 100:29:10 for
BA) in samples obtained from L-Lys are also applicable to
the 15N-incorporated molecules, the isotopic distributions
of photocatalytic products, defined as the proportion of the
amount of 15N-incorporated products in the total amount of
products (P15, %), of PCA and NH3 on this photocatalyst
were estimated to be 46 and 45 for PCA and NH3, respec-
tively.
Fig. 2. Correlation between optical purity and selectivity of various
metal-loaded catalysts. The symbols are the same as those in Fig. 1.
There was appreciable dependence of photocatalytic ac-
tivity on the kind of TiO2 powder; the HyCOM TiO2 pow-
ders showed high rPCA when any kind of metal was loaded.
For example, even when HyCOM TiO2 loaded with Rh,
which reduced each rPCA of Pt-loaded HyCOM, P-25, and
ST-01 powders, was used, the rPCA was still higher than
those of other TiO2 powders loaded with Pt. High photo-
catalytic activity of HyCOM TiO2 powders, similar to that
found in the present study, has been observed in various
liquid-phase photocatalytic systems and might be due to
well-balanced structural features of HyCOM TiO2 powders;
i.e., high crystallinity and large surface area, leading to lesser
recombination between e− and h+ and greater adsorption
of substrates, respectively [7,31,35]. Detailed discussions on
the correlation between photocatalytic activity and physico-
chemical properties of HyCOM TiO2 powders will be pub-
lished elsewhere.
Fig. 2 shows plots of OPPCA versus SPCA extracted
from the three-dimensional plot of Fig. 1. There is a clear
relation between them, i.e., SPCA increased with decrease in
OPPCA, which gave a master curve, though HyCOM TiO2
powders tended to show slightly higher SPCA and OPPCA
among the TiO2 samples used. These results suggest that
achievement of optimum SPCA and OPPCA simultaneously
by each catalyst is difficult.
It can also be seen in Fig. 2 that both SPCA and OPPCA
depend mainly on the kind of loaded metal. Among them,
Table 1 summarizes P15 in PCA and NH3, OPPCA, SPCA
and the amount of molecular hydrogen (H2) production
,
Pt loading tended to give higher SPCA but lower OPPCA
.
(YH , µmol). Based on the fact that SPCA, OPPCA, and YH
2
2
This trend was reversed for Pd-loaded catalysts, while
Rh loading gave moderate photocatalytic activity. Since
loaded metals are thought to act as reduction sites, as
noted above, these results also indicate that the difference
in catalytic performances in the reduction stage has a
were not greatly affected by the source material, L-Lys and
α-15N–L-Lys, by a few representative photocatalysts, it was
assumed that L-Lys and α-15N–L-Lys behave similarly in
the photocatalytic reaction. Mainly due to the relatively low
yield of NH3 by some Pd- and Rh-loaded photocatalysts,

B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
157
Scheme 2. Schematic procedure for conversion of ammonia with benzaldehyde into benzylamine (BA) using metal-loaded TiO particles as hydrogenation
2
catalysts.
Table 1
15
N proportion in PCA and NH , optical purity, selectivity, amount of H liberation, and the efficiency of reduction of α-CSB or ε-CSB in the photocatalytic
3
2
reaction of L-Lys on several metal-loaded TiO powders
2
a
b
c
d
e
f
Catalyst
Loaded
metal
P
(%)
OP
(%)
S (%)
PCA
Y
(µmol)
f
(%)
f
(%)
ε-CSB
H2
15
PCA
α-CSB
PCA
NH
3
P-25
P-25
HyCOM
HyCOM
Pt
Rh
Pt
46
63
56
68
45
44
47
47
55
61
62
69
68
52
66
48
7
39
29
35
57
49
71
62
78
39
65
32
Pd
a 15
15
N proportion of products obtained from α- N-L-Lys.
b
c
d
e
f
Optical purity of PCA.
Selectivity of PCA production based on amount of consumed L-Lys.
Amount of liberated molecular hydrogen.
Efficiency of α-CSB reduction.
Efficiency of ε-CSB reduction.
was observed+on Pt-loaded P-25, indicating that oxidative
cleavage by h of an ε-amino moiety of L-Lys gave optically
pure L-PCA through retention of chirality of the α-carbon
in the presumed intermediate of α-CSB, which undergoes
reduction by e− into L-PCA. On the other hand, P15 in NH3
reflects the proportion of oxidation at the α-amino moiety
in consumed L-Lys, since NH3 is produced via oxidation
of L-Lys followed by hydrolysis of the imines. It is clear
that P15 in NH3 was almost constant, i.e., independent of
the kind of photocatalyst used in the present experiments. In
our previous study [30], we showed that L-Lys is adsorbed
on the TiO2 surface through protonation of an α-amino
group by acidic hydroxyl groups on the surface. Because a
basic amino group undergoes oxidation more easily than a
protonated amino group such as an α-amino group of Lys in
its neutral aqueous solution, protonation on the TiO2 surface
enhances the relative probability of oxidation of the ε-amino
group to give optically active PCA. The constant P15 in NH3
observed in this study indicates that the surface properties of
TiO2 samples are not so different and that the selectivity in
the first oxidation step is less controllable by metal loading.
On the other hand, significant differences have been ob-
served in P15 in PCA depending on the kind of photocat-
alyst, especially a loaded metal. These results suggest that
the observed difference in OPPCA is due to the difference in
efficiencies of reduction of two types of CSB, α-CSB, and ε-
CSB, the former of which is produced via ε-amino group ox-
idation to give the L-isomer of PCA and the latter of which is
produced via α-amino group oxidation to give racemic PCA.
For more precise interpretation of the observed trend of
plots between OPPCA and SPCA as shown in Fig. 2, we
tried to quantify the efficiency of reduction of two types of
CSB by analyzing photocatalytic activity and isotopic dis-
Fig. 3. GC-MS patterns of TMS derivatives of PCA (top) obtained from
15
L-Lys (upper) or from α- N-PCA (lower) and of BA (bottom) obtained
15
from L-Lys (upper) or from α- N-PCA (lower).
leading to an insufficient amount of BA yield for GC-MS
analysis, we could not detect any peak of BA to obtain
P15. It was observed for several photocatalysts that P15
in PCA was comparable to OPPCA, though a slight difference

158
B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
Scheme 3. An example of determination of the efficiency of reduction of two types of CSB on various metal-loaded TiO powders.
2
tribution. On the basis of the reaction mechanism shown in
Scheme 1, we could estimate the amounts of δ-aldehyde and
α-keto acid intermediates from the total amounts of con-
sumption of L-Lys and P15 in NH3. Assuming a quantita-
tive intramolecular condensation into CSB, reduction effi-
ciency of the two kinds of CSB (fα and fε), defined as
yields of 15N- and 14N-incorporated PCAs based on the
amounts of produced α-CSB (fα) and ε-CSB (fε), respec-
tively, was determined. The results for Pt-loaded P-25 are
shown in Scheme 3. The reaction started with 100 µmol of
L-Lys, and PCA was produced through the intermediacy of
28 and 23 µmol of α-CSB and ε-CSB, respectively, as calcu-
lated from the consumption of L-Lys (51%) and P15 in NH3
(45%). From the amounts of α-CSB and ε-CSB, P15 in PCA
(46%), and total yield of PCA (34%), fα and fε were cal-
culated to be 57 and 78%, respectively. Similarly calculated
fα and fε for specific metal-loaded TiO2 powders are given
in Table 1.
As expected from the above results and discussion, fα
and fε varied depending on the kind of photocatalyst. It
seems that fε showed a more marked dependence than fα,
and the dependence was closely related to the photocatalytic
performance. As for the results of photocatalytic reaction in
which SPCA is relatively low, e.g., by Pd-loaded HyCOM
TiO2, favoring the reduction of α-CSB gave higher OPPCA
as well as a higher yield of H2. This trend is consistent with
the proposed mechanism of a relatively large part of ε-CSB
remaining unreduced and production of a by-product of H2
preferably proceeding instead of the reduction of CSBs,
especially ε-CSB, by photoexcited electrons. On the other
hand, at higher SPCA, e.g., by Rh-loaded P-25, both CSBs
were reduced nonselectively, and OPPCA thereby became
lower than that of Pd-loaded HyCOM TiO2. Thus, OPPCA
is governed mainly by the selectivity in the oxidation stage.
Relatively high SPCA was achieved by Pt-loaded P-25 and
HyCOM TiO2. For these metal-loaded photocatalysts, the
reduction of both ε-CSB and α-CSB proceeds efficiently,
leading to much lower OPPCA. Accordingly, the general
relation between SPCA and OPPCA on several metal-loaded
TiO2 powders, as shown in Fig. 2, could be explained on the
basis of the redox-combinedmechanism of PCA production.
Thus, the stereoselectivity of the final product, PCA, is
determined by the kind of metal that works as a catalyst
for the reduction of CSB intermediates. It is reasonable
to expect that reduction of α-CSB is easier than that of
ε-CSB due to the steric hindrance by the carboxyl group.
The catalytic ability of Pd, which gave a relatively low fε,
is thought to be lower that of Pt or Rh; it is well known
that Pt acts as an efficient catalyst for reduction of protons
or some organic compounds in deaerated aqueous TiO2
photocatalytic systems, but there have been few reports
on Pd-loaded samples. Actually, we have observed much
lower photocatalytic activity on Pd-loaded TiO2 powders in
H2 production from deaerated aqueous methanol solution
(data not shown). Thus, high selectivity of α-CSB reduction
by Pd-loaded samples might be due to both the structural
differences between two types of CSB and the insufficient
catalytic ability of Pd to induce ε-CSB reduction. Pt and Rh
have sufficient ability for CSB reduction, and this ability
was enhanced by loading onto P-25 and HyCOM TiO2
powders of relatively large surface area to induce high
dispersion. Studies on structural characterization of loaded
metals correlating with photocatalytic activity, especially for
effective control of stereoselectivity, are now in progress.

B. Pal et al. / Journal of Catalysis 217 (2003) 152–159
159
4. Conclusions
[7] B. Ohtani, K. Iwai, H. Kominami, T. Matsuura, Y. Kera, S.-I. Nishi-
moto, Chem. Phys. Lett. 242 (1995) 315.
[8] P.V. Kamat, Chem. Rev. 93 (1993) 267.
A detailed study on the effects of loading of various
metals on the photocatalytic activity of TiO2 powders for
PCA synthesis from L-Lys was conducted. The selectivity
for oxidation of two kinds of amino groups in L-Lys and
reduction of two kinds of cyclic Schiff base intermediates
were found to be important factors for stereoselective
production of optically active PCA. From the analyses of
distribution of 15N-incorporated products using α-15N–L-
Lys as a substrate, it was found for the first time that
the selectivity in the reduction stage can be controlled by
appropriate metal loading, but the effect on selectivity in
the oxidation stage was negligible; i.e., just loading of
noble metals as reduction sites is not sufficient to obtain
a TiO2 photocatalyst having both high OPPCA and SPCA
simultaneously. Consequently, in order to achieve optimum
photocatalytic activity, i.e., selective production of optical
pure L-PCA with high selectivity, other strategies for further
control in the oxidation stage, e.g., modification of the
manner of adsorption of L-Lys on TiO2 to change the
physicochemical properties of TiO2 surface, are needed.
[9] M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341.
[10] T. Hirakawa, H. Kominami, B. Ohtani, Y. Nosaka, J. Phys. Chem.
B 105 (2001) 6993.
[11] P. Pichat, C. Guillard, C. Maillard, L. Amalric, J.C. D’Oliveira, Trace
Met. Environ. 3 (1993) 207.
[12] R.I. Bickley, R.K.M. Jayanty, V. Vishwanathan, J.A. Navio, NATO
ASI Ser., Ser. C 174 (1986) 555.
[13] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem.
Rev. 95 (1995) 69.
[14] A. Fujishima, K. Hashimoto, T. Watanabe, in: TiO Photocatalysis—
2
Fundamentals and Applications, BKC, Tokyo, 1999, p. 176.
[15] M.A. Fox, M.J. Chen, J. Am. Chem. Soc. 105 (1983) 4497.
[16] T. Ohno, K. Nakabeya, M. Matsumura, J. Catal. 76 (1998) 176.
[17] T. Ohno, T. Kigoshi, K. Nakabeya, M. Matsumura, Chem. Lett. (1998)
877.
[18] B. Ohtani, T. Ohno, in: M. Kaneko, I. Okura (Eds.), Photocatalysis—
Science and Technology, Kodansha–Springer, Tokyo, 2002, p. 185.
[19] F. Mahdavi, T.C. Bruton, Y. Li, J. Org. Chem. 58 (1993) 744.
[20] K.H. Park, H.S. Joo, K.I. Ahn, K. Jun, Tetrahedron Lett. 36 (1995)
5943.
[21] B. Ohtani, Y. Goto, S.-I. Nishimoto, T. Inui, J. Chem. Soc., Faraday
Trans. 92 (1996) 4291.
[22] S.-I. Nishimoto, B. Ohtani, T. Yoshikawa, T. Kagiya, J. Am. Chem.
Soc. 105 (1983) 7180.
[23] B. Ohtani, H. Osaki, S.-I. Nishimoto, T. Kagiya, Tetrahedron Lett. 27
(1986) 2019.
[24] K. Shibata, T. Mimura, M. Matsui, T. Sugiura, H. Minoura, J. Chem.
Soc., Chem. Commun. (1988) 1318.
Acknowledgments
Mr. Tetsuzo Habu and Mr. Kazuhito Matsudaira (Catal-
ysis Research Center, Hokkaido University) are acknowl-
edged for their assistance in the construction of the pho-
toirradiation apparatuses. This work was partly supported
by Grants-in-Aid for Scientific Research on Priority Areas
(412 and 417, Nos. 12874083, 14044003, and 14050007)
from the Ministry of Education, Culture, Sports, Science,
and Technology (MEXT) of the Japanese Government.
[25] H. Minoura, Y. Katoh, T. Sugiura, Y. Ueno, M. Matsui, K. Shibata,
Chem. Phys. Lett. 173 (1990) 220.
[26] K. Shibata, A. Hida, M. Matsui, H. Muramatsu, T. Sugiura, H. Mi-
noura, in: Conference on Recent Development of Photocatalytic Reac-
tion, Tokyo, 1994, p. 24.
[27] H. Kisch, W. Lindner, Chem. Unserer Zeit. 35 (2001) 250.
[28] A. Reinheimer, R. van Eldik, H. Kisch, J. Phys. Chem. B 104 (2000)
1014.
[29] H. Keck, W. Schindler, F. Knoch, H. Kisch, Chem. Eur. J. 3 (1997)
1638.
[30] B. Ohtani, T. Yako, Y. Samukawa, S.-I. Nishimoto, K. Kanamura,
Chem. Lett. (1997) 91.
References
[31] H. Kominami, T. Matsuura, K. Iwai, B. Ohtani, S.-I. Nishimoto, Y.
Kera, Chem. Lett. (1995) 693.
[32] H. Kominami, S. Murakami, Y. Kera, B. Ohtani, Catal. Lett. 56 (1998)
125.
[1] P.D. Bailey, P.A. Millwood, P.D. Smith, Chem. Commun. (1998) 633.
[2] B.T.A. Ekenstam, C. Bovin, 4695576 A 870922, US Patent, 1987.
[3] T. Fujii, M. Miyoshi, Bull. Chem. Soc. Jpn. 48 (1975) 1341.
[4] K. Miyazaki, A. Yasutake, H. Aoyagi, N. Izumiya, Mem. Fac. Sci.,
Kyushu Univ. 12 (1980) 165.
[5] M.T. Beck, A. Katho, L. Dozsa, Inorg. Chim. Acta 55 (1981) L55.
[6] B. Ohtani, S. Tsuru, S.-I. Nishimoto, T. Kagiya, K. Izawa, J. Org.
Chem. 55 (1990) 5551.
[33] B. Ohtani, J.-I. Kawaguchi, M. Kozawa, Y. Nakaoka, Y. Nosaka, S.-I.
Nishimoto, J. Photochem. Photobiol. A: Chem. 90 (1995) 75.
[34] B. Ohtani, J.-I. Kawaguchi, M. Kozawa, S.-I. Nishimoto, T. Inui, K.
Izawa, J. Chem. Soc., Faraday Trans. 91 (1995) 1103.
[35] H. Kominami, J.-I. Kato, M. Kohno, Y. Kera, B. Ohtani, Chem. Lett.
(1996) 1051.