The composition of transformation products of 2,4,6-trinitrobenzoic acid in the aqueous-phase hydrogenation over Pd/C catalysts

Article abstract of DOI:10.1002/mrc.4931

Due to a detailed analysis of NMR spectra of the reaction solutions with different composition obtained by the aqueous-phase catalytic (Pd/C) hydrogenation of 2,4,6-trinitrobenzoic acid, the intermediate compounds were identified and a more substantiated mechanism was proposed for the formation of the main reaction products—1,3,5-triaminobenzene and cyclohexane-1,3,5-trione trioxime. The condensation of the 1,3,5-triaminobenzene molecules produced by a complete hydrogenation of 2,4,6-trinitrobenzoic acid was shown to result in the formation of a paramagnetic heterocyclic compound.

Full text of DOI:10.1002/mrc.4931

Talsi Valentin (Orcid ID: 0000-0003-4008-6497)
Yurpalov Vyacheslav (Orcid ID: 0000-0002-7243-5498)
The Composition of Transformation Products of 2,4,6-
trinitrobenzoic Acid in the Aqueous-Phase Hydrogenation
over Pd/C Catalysts
Valentin P. Talsi1, Olga B. Belskaya1,2 and Vyacheslav L. Yurpalov*1
Address: 1 Center of New Chemical Technologies of the Federal Research Center Boreskov
Institute of Catalysis of Siberian Branch of the Russian Academy of Sciences, 54
Neftezavodskaya St., 644040 Omsk, Russia and 2Omsk State Technical University, 11 Mira
Av., 644050 Omsk, Russia
Email: Vyacheslav L. Yurpalov - yurpalovv@mail.ru
* Corresponding author
Abstract
Due to a detailed analysis of NMR spectra of the reaction solutions with different
composition obtained by the aqueous-phase catalytic (Pd/C) hydrogenation of 2,4,6-
trinitrobenzoic acid, the intermediate compounds were identified and a more substantiated
mechanism was proposed for the formation of main reaction products – 1,3,5-
triaminobenzene and cyclohexane-1,3,5-trione trioxime. Condensation of the 1,3,5-
triaminobenzene molecules produced by complete hydrogenation of 2,4,6-trinitrobenzoic acid
was shown to result in the formation of a paramagnetic heterocyclic compound.
Keywords
1H and 13C NMR spectroscopy; palladium catalyst; trinitrobenzoic acid; catalytic
hydrogenation; aqueous-phase reaction
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/mrc.4931
This article is protected by copyright. All rights reserved.

Introduction
Investigation of catalytic transformations of trinitroaromatic compounds toward the complete
hydrogenation yielding triamine derivatives or toward the synthesis of intermediate reduction
products is of practical interest for utilization of explosives and production of useful chemical
substances [1]. One of the most available and processable trinitrocompounds is the oxidation
product of trinitrotoluene, 2,4,6-trinitrobenzoic acid (TNBA), the main technological
advantage of which is good water solubility in the composition of alkaline metal salts [2-4].
This allows the reaction to be performed in the medium of this nontoxic and inexpensive
solvent. Selective synthesis of the specified products via the multistep hydrogenation of
TNBA makes it necessary to develop selective catalysts and obtain a detailed knowledge
about individual transformation steps and composition of the intermediates. In this context, a
difficult problem is the identification of the reduction products, especially in the intermediate
steps of the reaction.
In our earlier studies [5,6], the NMR analysis of the composition of hydrogenation products
formed a basis for proposing the scheme of TNBA transformation into 1,3,5-triaminobenzene
during its aqua-phase catalytic (Pd/С) hydrogenation. The main intermediate (two isomers of
cyclohexane-1,3,5-trione trioxime) and a series of minor components of reaction solutions
were identified. Analysis of the additional spectral information obtained by increasing the
substrate concentration and varying the catalyst activity made it possible to refine the
identification of the intermediates as well as the condensation products of the final 1,3,5-
triaminobenzene. As a result, a more reasonable mechanism was proposed for the final stages
of TNBA hydrogenation, which is presented in this paper.
Materials and Method
Catalyst Synthesis
The study was carried out using the 5(1)%Pd/С catalysts that were synthesized similar to [5-
7], by hydrolytic precipitation of palladium polyhydroxo complexes (PHC) on the surface of
carbon material Sibunit (synthesized at the Department of Experimental Technologies,
Institute of Hydrocarbons Processing SB RAS, SBET 422 m2g-1). To this end, an H2PdCl4
aqueous solution was supplemented with alkaline precipitant to obtain the required pH. The
pH was controlled on a SevenMulti (Mettler Toledo) system using a combined electrode. The
obtained PHC solution was mixed with an aqueous suspension of Sibunit. This was
This article is protected by copyright. All rights reserved.

accompanied by a complete absorption of the complexes with decoloration of the solution.
The further liquid-phase reduction of palladium on the support surface was carried out with
sodium formate in an alkaline medium [8, 9]. After palladium reduction, the samples were
washed with water to pH 7, filtered and stored moist (the moisture content of ca. 50%).
Before further experiments, no any pretreatments of the synthesized catalyst were performed.
To obtain highly dispersed particles of supported palladium, the synthesis of PHC was
performed using KHCO3 alkaline precipitant at pH 4 (catalysts 5(1)%Pd/C(K)). In order to
increase the particle size of palladium, NaHCO3 was used as an alkaline precipitant during
the synthesis of PHC, and the synthesis of PHC was carried out at pH 5 (catalysts
5%Pd/C(Na)). The TEM data on the particle sizes of the supported palladium for these
samples are presented in supplementary materials (Figure 4S).
In addition, to reduce the active surface in samples we employed the partially deactivated
catalysts that were obtained by repeated use of the initial catalyst 5%Pd/C(K) after two or
three hydrogenation cycles of the substrate (5%Pd/C(K)-D2 and 5%Pd/C(K)-D3,
respectively). After removal of the reaction products, a fresh 10% solution of TNBA was
added to the used and partially deactivated catalyst; and then the hydrogenation was repeated.
Some details of the multi-cycle experiment are presented in supplementary materials (Figure
5S).
Catalytic Tests
The sodium salt of TNBA was obtained by the addition of sodium bicarbonate (2.95 g) to the
aqueous suspension of TNBA (10.00 g) until it completely dissolved and CO2 release ceased.
The catalytic hydrogenation of the obtained salt (100 mL of 10 % aqueous solution) was
performed at a temperature of 323 K and pressure of 0.5 MPa in the presence of 500 mg (dry)
catalyst (the loading was about 400 g of TNBA per 1 g of Pd). The reaction was conducted in
a 180 mL steel autoclave equipped with a valve for hydrogen input and an external
thermostatted jacket. The reaction mixture was stirred by a magnetic stirrer at 1400 rpm to
prevent external diffusion limitations. The progress of the reaction was monitored by
measuring the volume of consumed hydrogen with a mass flow meter. The reaction was
stopped when the required amount of hydrogen was consumed. After cooling, the sampling
was carried out with a syringe.
NMR Spectroscopy
13
1
The С and Н NMR spectra of hydrogenation products were recorded on an Avance 400
(Bruker) NMR spectrometer using DOCD3 for deuterium stabilization and as the internal
This article is protected by copyright. All rights reserved.

standard. The pulse programs zg, jmod (APT) with a relaxation delay of 10 s were employed.
The lengths of 90° pulses were 3 μs (1H) and 7 μs (13C). The composition of the
hydrogenation products was estimated from the integrated intensity of signals from
components in the 1H NMR spectra.
EPR Spectroscopy
The EPR spectra were recorded on a BrukerEMXplus (Bruker) spectrometer operating in the
X range (9.7 GHz) with an ER 4105 DR resonator at a microwave power 0.2–2.0 mW, at a
modulation frequency 100 kHz, amplitude 0.3 G, and temperature 25°C. The spectra were
analyzed with Bruker WinEPR Processing software. An accurate determination of g-values
was made using a Bruker ER 4119HS-2100 marker (g = 1.9800±0.0006). Spectra of aqueous
solutions were taken in glass capillaries with the inner diameter of 1 mm. Simulation of the
spectra was performed using Bruker WinEPR SimFonia (ver. 1.2) for the rapidly tumbling
radicals in solution.
Results and Discussion
The occurrence of consecutive and parallel reactions in the TNBA hydrogenation process
makes it difficult to identify the intermediate transformation products and elucidate the
reaction mechanism. Earlier we have proposed the reaction scheme based on the NMR
analysis of products in the individual steps of TNBA transformation [5]. Based on previously
obtained data [5,6] the following mechanism of the final stages of TNBA reduction (Scheme
1) can be represented, according to which 2,4,6-substituted nitrobenzoic acids 1 - 3 are
hydrogenated in two directions. The left path in Scheme 1 includes an intramolecular redox
reaction followed by decarboxylation, and the right path goes through the reduction of the
hydroxyamino group of 1 – 3 with a low probability of decarboxylation. The first of the
indicated reaction routes determines its most essential feature and leads to the formation of
oximes 14 and 15, which, depending on the catalyst properties, can either be the main
components of the hydrogenation product or occur in minor amounts, as was shown before
[6,7]. The following quantitative assessment of the relative content of hydrogenates (wt. %,
excluding solvent) was made on the basis of 1H NMR spectra.
It was found [6,7] that at a high dispersion of supported palladium, oximes 14 and 15
virtually completely transform into triaminobenzene 9 due to the reversible step of their
formation and further consecutive reactions, according to Scheme 1. The 1Н NMR spectrum
of transformation products displayed in Figure 1 shows the corresponding signals
This article is protected by copyright. All rights reserved.

characterizing compound 9 (~85%) with insignificant admixtures of compounds 14, 15 and 8
(1-2%). The other minor components noted in Figure 1 (12 - 15%) are not included in
Scheme 1. They are formed by the condensation and hydrolysis of compound 9 and will be
considered below. In the case of catalysts containing supported palladium particles of a
greater size, the strength of the interaction of oximes 14 and 15 with the metal surface is
insufficient for their further transformation into compounds 11, 12, 13 and 9; the oximes are
desorbed from the surface into solution [6,7]. The main components of the hydrogenation
products are oximes 14 and 15 (~ 44%), the content of triamine 9 is less than 23% (Figure 2).
The most detailed composition of the intermediate compounds was described in our earlier
work [5] using NMR spectra of the reaction solution sampled at the intermediate step of the
reaction (Figure 3). Although the hydrogenate giving the spectrum in Figure 3 contains up to
5% of acid 5 formed via the right reaction path, the left path is the main one since the total
concentration of intermediate oximes 14 and 15 (24%) is much higher than the concentration
of compound 5 and is comparable to the concentration of the final product 9 (38%).
However, due to a low concentration of the solution, some compounds were not detected
although their formation was implied by the transformation scheme, and identification of
several previously described compounds was not quite reliable. To solve these problems, in
the present work the concentration of TNBA solution used in the reaction was increased 5-
fold (a 10% aqueous solution of TNBA sodium salt). Therewith, the amount of Pd/C catalyst
was also increased at a constant ratio of 400 g TNBA per 1 g Pd. To solve the same problem,
the reaction was decelerated with the use of partially deactivated catalysts (Experimental
Part).
Figure 4 shows the 1H NMR spectrum of the hydrogenate obtained with deactivated catalyst
5%Pd/C(K)-D2. It contains ~ 37% of the final product 9, up to 20% of each of the acids 5
and 6 but not more than 2% of oximes 14 and 15. Therefore, in this case, the hydrogenation
proceeds mainly along the right path of Scheme 1. Among the intermediates formed via this
reaction route, only acid 5 was unambiguously identified in ref. [5]. The identification of four
more compounds (4, 6, 8, 16) is described below.
2,6-diamino-4-nitrobenzoic acid 4
In the 1Н NMR spectrum in Figure 4, the singlet at 6.40 ppm, whose intensity was high for
minor products, was attributed to acid 4. This characteristic signal was not noted in ref. [5]
This article is protected by copyright. All rights reserved.

because interpretation of the 13C NMR data, which could confirm its attribution to compound
4, appeared less reliable than in the case of other benzoic acids.
The region of 90–120 ppm, where absorption of С1, С3 and С5 atoms of benzoic acids under
consideration is observed, in case of compound 4 is the sole spectral fragment that contains
signals suitable for its identification (Figure 5; Table 1). In the indicated spectral region, there
are two signals at 102.7 and 111.3 ppm that can be assigned to compound 4 taking into
account similar parameters of the spectra for benzoic acids 3, 5, 6, 8 (Figure 5, Table 1).
However, the signal at 102.7 ppm from CH carbons of compound 4 is strongly broadened, in
contrast to other acids. It is known that broadening and even disappearance of signals in 13С
NMR spectra may be caused by chemical exchange [10]. We have not found a convincing
mechanism for the chemical exchange; nevertheless, identification of compound 4 only by
1Н NMR spectrum appears to be reasonable because the key intermediate 6 discussed below
can be obtained just as the result of reduction of compound 4.
2,6-diamino-4-hydroxyaminobenzoic acid 6 and 2,4,6-triaminobenzoic acid
8
In our earlier work, the signals in NMR spectra of the main minor component of
solutions displayed in Figures 4 and 5 were erroneously assigned to compound 8, which was
not distinguished from a similar symmetric compound 6 because they had close singlet
1
signals at 5.84 ppm (8) and 5.85 ppm (6) in Н NMR spectra. The error was corrected by
13
refining the attribution of signals to compounds 6 and 8 in С NMR spectra (Table 1). The
asignment of the indicated signals exactly to compound 6 is based primarily on their presence
in the spectra of solutions of the reaction products obtained with the use of less active
catalysts, along with other intermediate hydrogenation products (acids 4 and 5). At the same
time, compound 6 is absent in the products of exhaustive hydrogenation of TNBA obtained
with the use of active fresh catalysts, whereas 2,4,6-triaminobenzoic acid identified now as
compound 8 (Table 1) is a constant and noticeable minor component despite the mentioned
above preference of the left reaction path in the Scheme 1 with early decarboxylation. We
could explain the latter circumstance only by some probability of compound 9 carboxylation
with the formation of compound 8. Although we have not found any examples of the
appropriate Kolbe–Schmitt reaction for compound 9, it is known that the similar substance
phloroglucinol is carboxylated by alkali carbonates in water solution at the same pressure and
temperature as in our case [11].
This article is protected by copyright. All rights reserved.

A high-field shift of a part of signals in 13С NMR spectra of compound 6, in comparison
with the similar signals of compound 8, agrees with the same shift observed upon substitution
of amino group by hydroxylamino one in the case of compounds 9 and 13 (Table 1).
2-hydroxyamino-4,6-diaminobenzoic acid 7,2,4-diamino-6-[(3,5-
diaminophenyl)-NNO-azoxy]benzoic acid 16
We detected only hardly perceptible signals (not visible in the 1H NMR spectra) that can be
attributed to intermediate 7. Most likely, acid 5 transforms, missing compound 7, via
azoxybenzene 16 as suggested in Scheme 3.
The signals corresponding to compound 16 are observed in the spectra of solutions obtained
with the use of low active catalysts (Figures 5 and 6) and have high intensities in the case of
some deactivated catalyst (Figure 6). The compound under consideration has two substituted
benzene rings. One of them gives a doublet (6.98 ppm, 2H) and a triplet (6.92 ppm, 1H) with
the J constant of ~2 Hz in the 1Н NMR spectrum (Figure 6). These signals correspond to 1,3-
diamino-5-X-substituted benzene, where substituent X has the pronounced electron acceptor
properties. For example, such effect of the electron acceptor substituent is observed for 1,3-
diamino-5-nitrobenzene [12]: the experimental spectrum has doublet and triplet with the
chemical shifts of 6.60 and 6.15 ppm, respectively. Two other signals corresponding to
compound 16 are represented by the doublet of doublets (6.49 ppm, 1H), which is
undistinguishable from the triplet, and the doublet (5.84 ppm, 1H) with the J constant ~ 2 Hz.
Thus, the second benzene ring in compound 16 with two protons in meta-position has four
substituents, one of them causes an additional diastereotopic splitting of the signal from the
adjacent proton [13]. This substituent was identified as azoxy group which is characteristic of
the typical reduction products of aromatic nitrocompounds – azoxybenzenes [14].
The formation of azoxybenzenes upon reduction of, for example, nitrobenzene is explained
by the addition of the aniline amino group to nitroso group of the intermediate
nitrosobenzene, which is accompanied by dehydrogenation. Azoxybenzene 16 is formed by a
similar mechanism upon interaction of triaminobenzene 9 with acid 5. The signals in the 1Н
NMR spectrum (Figure 6) indicated in Scheme 3 were assigned to compound 16. The
assignment agrees with the polarization of the molecule 16 depicted in Scheme 4 which
explains unexpectedly wide range (5.84 – 6.98 ppm) of proton chemical shifts and
diastereotopic splitting of the signal 6.49 ppm as the result of hindered inversion of
stereogenic pyramidal nitrogen –NO-- in azoxybenzene 16 [13].
This article is protected by copyright. All rights reserved.

As shown by the analysis of the 13С NMR spectrum of a solution of hydrogenation products
obtained with the use of deactivated catalyst (Figure 1S), some characteristic signals
assignable to compound 16 are detected in the spectrum (Table 1).
The final reduction products of azoxybenzenes are mainly the corresponding
hydrazobenzenes. In accordance with this statement, all 1H NMR spectra of the solutions
obtained as the result of TNBA exhaustive hydrogenation contain weak but characteristic
triplet signals with the chemical shift 5.53 ppm (J ~ 2 Hz). The only reason to make protons
of the aminobenzenes involved to absorb in a higher field than 1,3,5 protons in compound 9
is condensation with the formation of dimer 18 discussed below. Protons in position 4 of
compound 18 (Scheme 2) absorb at 5.64 ppm (Figure 1). In order to shift this signal to a still
higher field, it is necessary to replace –NH- fragment in compound 18 by hydrazine group –
NH-NH-. So the signal 5.53 ppm belongs to hydrazobenzene 24 formed by the reduction of
compound 16. Most of the remaining signals of this compound are overlapped partially or
completely by the signals of other components, but the second triplet signal 5.93 ppm (J ~ 2
Hz) belongs to it definitely. Taking into account this scarce information and simulated
spectra, the following structure of the reduction product of compound 16 was nevertheless
suggested (Scheme 5).
As follows from Figure 6, in addition to the product of exhaustive hydrogenation 9 (63%), an
intermediate product with a high (~20%) content in solution is compound 16. The content of
oximes 14 and 15 in this case does not exceed 2%. It has not been established under which
conditions the main intermediate on the right reaction route (Scheme 1) is compound 6
(Figure 4), and under which conditions it is compound 16 (Figure 6); however, it is obvious
that in case of the deactivated catalysts hydrogenation selects predominantly the right path.
This experimental information is consistent with the fact that there are two ways of the
adsorption of aromatic compounds: either due to the aromatic ring with the formation of π-
complex, or due to atoms of functional groups with the orientation of the plane of the
aromatic ring normal to the surface [15]. Substrates 1 - 3 are probably adsorbed by the first
way on a fresh active catalyst, which induces the intramolecular redox process and
decarboxylation (left path); on the deactivated sediment-covered catalyst, they are anchored
by oxygen atoms of hydroxylamine or nitro groups so that the plane of the benzene ring is
oriented normal with respect to the surface, which reduces the probability of decarboxylation
(right path). In ref. [15], such a qualitative change in the adsorption behavior was observed
with a change in the surface acidity.
This article is protected by copyright. All rights reserved.

Identification of condensation products of 1,3,5-triaminobenzene 2,8,14-
triazatetracyclo[13.3.1.13,7.19,13]henicosa-1(19),3(21),4,6,9(20),10,12,15,17-
nonaene-5,11,17-triamine 17, N1-(3,5-diaminophenyl)benzene-1,3,5-
triamine 18
It is known [16] that, in contrast to aniline, triamine 9 is characterized by a relatively easy
polycondensation. Its possible mechanism is displayed in Scheme 6. In our previous paper [5]
we failed to identify the condensation products of compound 9 and suggested uncertain
polyaniline structures. In fact all noticeable signals of condensation products in the 1H and
13C NMR spectra of hydrogenates are originate from compounds 17 and 18. Signals that can
be attributed to the analogous linear trimer or to the dendrimer type tetramer were detected
only at the limit of 1H NMR sensitivity.
Signals from compounds 17 and 18 are most noticeable in the spectra of reaction solutions
obtained using the catalyst that is highly active and selective toward the final product 9; the
most pronounced signals marked in Figures 1 and 7 correspond to cyclic trimer 17. Along
with the signals from compounds 17 and 18, signals assigned to compounds 19 – 22, which
will be discussed below, are marked in the spectra in Figures 1 and 7.
Identification of compounds 17 and 18 is reasonable due to a good agreement of their
experimental and simulated 13С NMR spectra displayed in Table 1 and Figures S2 and S3.
Despite some differences in chemical shifts for experimental and simulated spectra, it is
difficult to find other suitable structures for compounds 17 and 18.
However, it should be noted that the simulated 1Н NMR spectrum of compound 17 (doublet
at 6.34 ppm, 6Н, and triplet at 6.90 ppm, 3Н, J ~ 2 Hz) even qualitatively differs from the
experimental spectrum by relative positions of the doublet (6.27 ppm) and triplet (5.90 ppm).
The indicated contradiction can be explained and even used as an additional argument in
favor of correctness of the identification.
Compound 17 resembles porphyrins and can actually be presented as a macrocyclic aromatic
compound using a special variant of prototropic tautomerism, which is typical of porphyrins
(Scheme 7). A cyclic translocation of protons and π-electrons results in the formation of the
aromatic system including 18 electrons [17]. A pronounced strong-field shift of signals from
protons in the porphyrin or other aromatic macrocycles caused by the ring electron current is
well known [18]. The ring electron current is the reason why the internal CH protons of
This article is protected by copyright. All rights reserved.

compound 17 have much smaller chemical shifts than the external CH protons, in contrast to
the simulated spectra.
The doublet and triplet corresponding to compound 17 in the 1Н NMR spectra are always
accompanied by a singlet signal with the chemical shift of 7.1 ppm (Figure 1), which is
attributed to NH protons involved in the tautomeric transformations displayed in Scheme 7.
The 1H NMR spectrum of compound 18 consists only of a 5.64 ppm triplet, attributed to a
proton in position 4 (Scheme 2). The doublet signal of protons in positions 2 and 6 of dimer
18 overlaps with the absorption of aminobenzene 9.
Paramagnetism of cyclic trimer 17
The EPR study of the hydrogenation product whose NMR spectra are shown in Figures 1 and
7 revealed the presence of a stable intense signal with parameters g = 2.0040, ΔH1/2 = 0.20 G,
aN = 2.10 G (2NH), aN = 0.65 G (NH2), aH = 2.20 G (3HAr), aH = 0.85 G (2HNH2), and aH =
0.75 G (HNH), caused by splitting of a paramagnetic center on magnetic nuclei of hydrogen
and nitrogen (Figure 8). This signal appears when the reaction mixture gets in contact with
air; it may be related to the oxidation of the reduction product(s) by analogy with derivatives
of aniline oligomers [16,18,20]. The EPR spectrum analysis and its simulation indicate the
delocalization of the unpaired electron on benzene core and its hyperfine interaction with 1H
and 14N nuclei of cyclic trimer 17 fragments. The values of the hyperfine splitting constants
(HFSC) are close to similar structural fragments of paramagnetic N-cryptand [21,22].
The mechanism of radical particle formation (Scheme 8) can be proposed via the oxidation
[20] of compound 17 to the corresponding radical 17• or via the formation of a radical cation
NH+• (17+•) as a result of single-electron transfer (SET) [23]. The formation of the cation
radical 17+• is more likely for such compounds [19], however, it is difficult to determine the
charge of a paramagnetic particle using its EPR spectrum [24].
A long-term storage in air of the hydrogenation product whose composition is characterized
by the NMR spectra displayed in Figures 1 and 7 led to the formation of a solid product
consisting mostly of water-insoluble polycondensation products of aromatic amines [5]. The
1Н NMR spectrum of an aqueous extract of this product contains only three signals (Figure
9). The chemical shift of 5.75 ppm singlet corresponds to compound 8. The second rather
intense singlet at 8.45 ppm was earlier assigned to formic acid [5]. The third very weak
This article is protected by copyright. All rights reserved.

doublet at 6.27 ppm corresponds to compound 17. Since the EPR spectrum of this aqueous
extract shows a similar signal as for the initial hydrogenation product, this reliably confirms
that the observed EPR signal is due to the presence of compound 17, taking into account that
1
the samples of analogous extracts whose H NMR spectra contain only signals of 5.75 and
8.45 ppm are EPR silent.
It should be noted that the NMR spectra of hydrogenation products (Figures 1, 4, 5 and 7)
contain also the signals that are assigned to 2,4,6-triaminobenzamide 20 and weak signals
from the products of successive hydrolysis of 1,3,5-triaminobenzene: 3,5-diaminophenol 19
and 5-aminobenzene-1,3-diol 22, which were considered in our earlier work [5] (Scheme 9).
Two not noticed earlier 13C NMR signals of 160.2 and 148.3 ppm (Figure 7) are attributed to
carbonate of 1,3,5-triaminobenzene with tentative structure 21, taking into account that its
ammonium and carbonate ions do not participate in fast chemical exchange with free amine 9
and other carbonate ions which are present in the solution.
The identity of compound 19 was completely confirmed when analyzing new 13C NMR
spectra. Phenol 22 still remains as the tentative structure suggested only on a basis of the
solitary 1HNMR signal. But the identification of phloroglucinol 23 in [5] was wrong. In
reality amide 20 yields the relatively intense signal in 1H NMR spectra at 5.89 ppm assigned
earlier to compound 23. Compound 20 was reliably identified on a basis of 13C NMR spectra
(Table 1). Its formation by the amidation of acid 8 with ammonia released upon condensation
of amine (leading to compounds 17 and 18) is quite probable.
Conclusion
Reaction mixtures were obtained by catalytic (Pd/C) aqua-phase hydrogenation of 2,4,6-
trinitrobenzoic acid using concentrated solutions and catalysts with different activity. A
detailed analysis of NMR spectra of these mixtures made it possible to assign all the signals
observed in the spectra. Refined identification was used to propose a more substantiated
reaction mechanism describing a multicomponent composition of the hydrogenation product.
In the case of exhaustive hydrogenation, the composition of the hydrogenation product is
determined in the final step of the reaction upon reduction of 2-amino-6-hydroxyamino-
nitrobenzoic acid and its two minor regioisomers. The first route, which is related to
intramolecular redox reaction and subsequent decarboxylation, leads to 1,3,5-
This article is protected by copyright. All rights reserved.

triaminobenzene via the formation of intermediate cyclohexane-1,3,5-trione trioximes. The
second route, which is caused by reduction of the hydroxyamino group of 2-amino-6-
hydroxyamino-nitrobenzoic acid and its isomers in the first step, also leads mainly to 1,3,5-
triaminobenzene through a series of intermediate benzoic acids, among which 2,6-diamino-4-
hydroxyaminobenzoic acid and 2,4-diamino-6-[(3,5-diaminophenyl)-NNO-azoxy]benzoic
acid are likely to be the key ones. The second way is characteristic of the catalysts whose
activity is reduced as a result of repeated hydrogenation cycles. The reason for the existence
of different routes can be a different orientation of the adsorbed aromatic molecule relative to
the catalytic surface.
It was demonstrated for the first time that hydrogenation of 2,4,6-trinitrobenzoic acid is
accompanied by condensation of three molecules of the main product of the reaction, 1,3,5-
triaminobenzene, with the formation of a paramagnetic aromatic macrocyclic compound.
Acknowledgements
The authors thank S. V. Vysotskii for participation in catalytic experiments. Characterization
of catalysts and identification of reaction products were performed using facilities of the
Omsk Regional Collaborate Center, Siberian Branch of the Russian Academy of Sciences.
The study was financially supported by the Russian Foundation for Basic Research (Project
No. 16-03-00601-a).
This article is protected by copyright. All rights reserved.

References
1. A. L. Rusanov, L. G. Komarova, D. Yu. Likhatchev, S. A. Shevelev, V. A. Tartakovsky,
Russ. Chem. Rev. 2003, 72, 899.
2. G. S. Samuelsen, V. L. Garik, G. B. L. Smith, J. Am. Chem. Soc. 1950, 72, 3872.
3. V. A. Kuz’mina, Yu. T. Lapina, S. V. Sysolyatin, Polzunovsky Vestnik. 2008, 3, 129.
4. O. B. Belskaya, R. M. Mironenko, V. P. Talsi, V. A. Rodionov, S. V. Sysolyatin, V. A.
Likholobov, Procedia Engineering. 2016, 152, 110.
5. O. B. Belskaya, V. P. Talsi, R. M. Mironenko, V. A. Rodionov, S. V. Sysolyatin, V. A.
Likholobov, J. Mol. Catal. A: Chem. 2016, 420, 190.
6. R. M. Mironenko, O. B. Belskaya, V. P. Talzi, V. A. Rodionov, S. V. Sysolyatin, V. A.
Likholobov, Russ. Chem. Bull. Int. Ed. 2016, 65, 1535.
7. O. B. Belskaya, R. M. Mironenko, V. P. Talsi, V. A. Rodionov, T. I. Gulyaeva, S. V.
Sysolyatin, V. A. Likholobov, Catal. Today. 2018, 301, 258.
8. P. A. Simonov, S. Yu. Troitskii, V. A. Likholobov, Kinet. Catal. 2000, 41, 255.
9. O. B. Belskaya, R. M. Mironenko, T. I. Gulyaeva, M. V. Trenikhin, V. A. Likholobov,
Russ. Chem. Bull. 2018, 1, 71.
10. V. P. Talsi, S. N. Evdokimov, Journal of Chemical Engineering and Bioanalytical
Chemistry. 2017, 1, 1.
11. V. Hessel, C. Hofmann, P. Löb, H. Löwe, M. Parals, Chem. Eng. Technol. 2007, 30, 355.
12. A. Preiss, M. Elend, S. Gerling, S. Tränckner, Magn. Reson. Chem. 2005, 43, 736.
13. V. P. Talsi, S. N. Evdokimov, Russ. J. Org. Chem. 2017, 53, 1630.
14. A. Mahata, R. K. Rai, I. Choudhuri, S. K. Singh, B. Pathak, Phys. Chem. Chem. Phys.
2014, 16, 26365.
15. K. Tanabe, Stud. Surf. Sci. Catal. 1985, 20, 1.
16. M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 1997, 119, 4492.
17. A. S. Ivanov, A. I. Boldyrev, Org. Biomol. Chem. 2014, 12, 6145.
18. F. Sondheimer, R. Wolovsky, Y. Amiel, J. Am. Chem. Soc. 1962, 84, 274.
19. M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 1996, 118, 10626.
20. H. Garcıa, H. D. Roth, Chem. Rev. 2002, 102, 3947.
21. A. S. Ichimura, Q. Xie, L. P. Szajek, J. Lema, A. Burns, R. H. Huang, J. E. Jackson, J. L.
Dye, F. Demol, F. X. Sauvage, M. G. DeBacker, J. Phys. Chem. A. 2000, 104, 3038.
22. A. S. Ichimura, L. P. Szajek, Q. Xie, R. H. Huang, S. Z. Huang, M. J. Wagner, J. L. Dye,
J. E. Jackson, J. Phys. Chem. B. 1998, 102, 11029.
This article is protected by copyright. All rights reserved.

23. T. Wozniewski, E. Fedorynska, S. Malinowski, J. Colloid Interface Sci. 1982, 87, 1.
24. J. R. Boltont, G. K. Fraenkel, J. Chem. Phys. 1964, 40, 3307.
This article is protected by copyright. All rights reserved.

Table 1: Assignment of signals in the 13C NMR spectra of TNBA hydrogenation productsа,с.
Compound
C-1
C-2
C-3
C-4
C-5
C-6
C-7
C-2’
C-4’
C-6’
148.8
104.0
102.7
101.6
95.36
95.62
148.4
148.2
100.2
142.8
152.0
102.0
102.7
106.7
95.36
95.62
148.4
148.2
105.2
147.8
150.0
172.0
3
4
114.9
111.3
109.8
106.0
107.7
148.4
152.1
106.3
147.8
172.8
160.6
160.0
5
146.3
148.6
96.07
94.23
147.0
149.6
96.07
94.23
146.3
148.6
96.07
98.53
6
8
9
13
16
101.4
108.8
101.4
98.76
98.76
98.38
17b
18b
(147.7) (103.8) (144.8) (103.8) (147.7) (102.3)
146.4 92.21 146.3 92.21 146.3 94.68
(148.3) (86.4) (148.0) (86.4) (148.0) (90.95)
157.8
98.9
94.87
148.1
148.8
95.4
96.71
147.3
148.8
95.4
94.87
148.1
19
20
161.4
aChemical shifts are given in ppm. bValues in parentheses were obtained by simulation of the
spectra using ACD/Labs 6.00 (Advanced Chemistry Development Inc.) software. cThe
numbering of carbon atoms in the compounds listed in Table 1 is shown in Scheme 2.
This article is protected by copyright. All rights reserved.

Figure 1: The 1Н NMR spectrum of the hydrogenation products of TNBA sodium salt (10%
aqueous solution) that were obtained using the highly dispersed 5%Pd/C(K) catalyst.
This article is protected by copyright. All rights reserved.

Figure 2: The 1H NMR spectrum of the reaction mixture obtained by hydrogenation of
TNBA in aqueous solution (5%Pd/C(Na), concentration of TNBA is 10 wt.%) at a
temperature of 343 K and pressure of 0.5 MPa [5].
This article is protected by copyright. All rights reserved.

Figure 3: The 1Н NMR spectrum of the products formed in the intermediate step of TNBA
sodium salt hydrogenation (2% aqueous solution) over 1%Pd/C(K) catalyst [5].
This article is protected by copyright. All rights reserved.

Figure 4: The 1Н NMR spectrum of the hydrogenation products of TNBA sodium salt (10%
aqueous solution) that were obtained with the use of 5%Pd/C(K)-D2 catalyst (deactivated,
second cycle).
This article is protected by copyright. All rights reserved.

Figure 5: The 13С NMR spectrum of the hydrogenation products of TNBA sodium salt (10%
aqueous solution) that were obtained with the use of 5%Pd/C(K)-D2 catalyst (deactivated,
second cycle).
This article is protected by copyright. All rights reserved.

Figure 6: The 1Н NMR spectrum of the hydrogenation products of TNBA sodium salt (10%
aqueous solution) that were obtained with the use of 5%Pd/C(K)-D3 catalyst (deactivated,
third cycle).
This article is protected by copyright. All rights reserved.

Figure 7: The 13С NMR spectrum of the hydrogenation products of TNBA sodium salt (10%
aqueous solution) that were obtained with the use of a highly dispersed 5%Pd/C(K) catalyst.
This article is protected by copyright. All rights reserved.

Figure 8: Experimental (1) and simulated (2) EPR spectra of an aqueous solution of TNBA
sodium salt hydrogenation products (NMR spectra of this solution are displayed in Figures 1
and 6) stored on air for 1 h.
This article is protected by copyright. All rights reserved.

Figure 9: The 1H NMR spectrum of the aqueous solution obtained by water extraction from
solid condensation products of the TNBA hydrogenates. An arrow indicates the signal of
compound 17. Two more intense signals may belong to 1,3,5-triaminobenzene carbonate
(5.75 ppm) and formate ion (8.45 ppm).
This article is protected by copyright. All rights reserved.

Scheme 1: The mechanisms of TNBA hydrogenation over Pd/Sibunit catalysts according to
[5].
This article is protected by copyright. All rights reserved.

Scheme 2: The numbering of carbon atoms in the compounds listed in Table 1.
This article is protected by copyright. All rights reserved.

Scheme 3: The formation of azoxybenzene 16 from compounds 5, 9 via nitrosoderivative of
5.
This article is protected by copyright. All rights reserved.

Scheme 4: The displacement of the electrons in the molecule 16.
This article is protected by copyright. All rights reserved.

Scheme 5: The structure of the final product of compound 16 reduction.
This article is protected by copyright. All rights reserved.

Scheme 6: The possible polycondensation pathway of the compound 9.
This article is protected by copyright. All rights reserved.