Online Spectroscopic Investigations (FTIR/Raman) of Industrial Reactions: Synthesis of Tributyltin Azide and Hydrogenation of Chloronitrobenzene

Article abstract of DOI:10.1021/op030009w

The utilization of online spectroscopic methods (FTIR/Raman) for the development and the monitoring of chemical processes was demonstrated with two important chemical reaction types. The synthesis of tri-n-butyltin azide takes place at the solid-liquid interface and is therefore very sensitive to changes of reaction conditions or surface properties. Traditional offline analysis should be avoided due to the high toxicity of the compounds. The concentration variations of all involved compounds were determined. In the same way, the synthesis of 2-chloroaniline was studied to detect the formation of an unstable intermediate (the hydroxylamine) which can lead to uncontrolled decomposition reactions. Online spectroscopy is an opportunity to obtain the information about a reaction system faster and more efficiently than with conventional methods. This allows the rapid optimization of safe chemical processes and provides information for process control.

Full text of DOI:10.1021/op030009w

Organic Process Research & Development 2003, 7, 1059−1066
Online Spectroscopic Investigations (FTIR/Raman) of Industrial Reactions:
Synthesis of Tributyltin Azide and Hydrogenation of Chloronitrobenzene
Jacques Wiss*,† and Arne Zilian‡
NoVartis Pharma AG, Chemical and Analytical DeVelopment, WSJ-94.106a, CH-4002 Basel, Switzerland, and
SolVias AG, Online Technologies, WKL-127.6.06, CH-4002 Basel, Switzerland
Abstract:
reaction calorimetry and simulation software, all necessary
data for the process improvement and optimization can be
obtained. (2) The effect of reaction variables on reaction
performance can be determined. (3) Reduction of the
development time: the majority of the used equipment
(reaction calorimeter, spectrometer) can work fully automated
(even overnight or during the week-end). This time gain can
lead to a significant reduction of the time to market. (4)
Finally, the combination of these methods ensures the
development of a safe, optimized, and well-controlled
process.
Process Monitoring for Manufacturing Efficiency. (1)
Determination of the reaction completion (optimization of
the cycle time, avoid unnecessary stirring periods, no
sampling of toxic reaction mixtures for the traditional
analytics). (2) Monitoring of process variance (e.g. control
of the dosing feed of a reagent to avoid thermal accumula-
tion).
The utilization of online spectroscopic methods (FTIR/Raman)
for the development and the monitoring of chemical processes
was demonstrated with two important chemical reaction types.
The synthesis of tri-n-butyltin azide takes place at the solid-
liquid interface and is therefore very sensitive to changes of
reaction conditions or surface properties. Traditional offline
analysis should be avoided due to the high toxicity of the
compounds. The concentration variations of all involved com-
pounds were determined. In the same way, the synthesis of
2-chloroaniline was studied to detect the formation of an
unstable intermediate (the hydroxylamine) which can lead to
uncontrolled decomposition reactions. Online spectroscopy is
an opportunity to obtain the information about a reaction
system faster and more efficiently than with conventional
methods. This allows the rapid optimization of safe chemical
processes and provides information for process control.
2. Instruments
1. Introduction
The FTIR spectroscopic online measurements were carried
out with a ReactIR 1000 instrument (Applied Systems)
equipped with a diamond probe. This spectrometer operates
according to the Fourier transform method and utilizes the
ATR principle (attenuated total reflection). Therefore, the
penetration depth of the surface waves into the absorbing
medium is very small. The spectral region comprises between
650 and 4000 cm-1.
The Raman spectroscopic online measurements were
carried out with HoloLab Series 5000 and HoloProbe VPT
System (Kaiser Optical Systems, Inc.). These spectrometers
use a fiber-optic cable and immersion probe with sapphire
window to both deliver the excitation laser light to the
reaction mixture and receive the light scattered back. The
spectral region comprises between 50 and 3500 cm-1.
Reaction monitoring is of central interest to chemists
working in all stages of chemical development. It is an
opportunity to obtain, much faster than usual, the information
about a reaction system. This allows us to optimize chemical
processes much faster and to use the information for the
process control. Three main application fields of online
analytics can be distinguished.
Reaction Analysis for Chemical R&D. (1) Determina-
tion of the chemical mechanism and pathway: online
spectroscopy offers a mean to analyze reactions which have
unstable intermediates or are run under conditions which
prevent grab-sample analysis. For example, many reactions
are run at -80 to 0 °C, and it is difficult to have a good
method for tracking the process. Pulling a reaction sample
destroys it, and it cannot be put on a LC column. (2)
Measurement of real-time conversion data: these methods
provide end point of a reaction, conversion, relative con-
centrations, and eventually absolute concentration data for
a wide range of reactions. Therefore, they provide the basis
for developing analytical methodology for process monitoring
in a plant.
3. Examples
Since characteristic differences of the vibrational spectra
of starting material and product were expected, the progress
of the reactions was followed by simultaneous in situ FTIR
and Raman analysis. Both spectroscopic probes were in-
stalled in a RC1 reaction calorimeter. All signals were
simultaneously recorded. The thermal data were used for the
assessment of the thermal safety of the investigated reactions.
Thermal analysis was not the goal of this study, and therefore
these results are not discussed in this paper.
Reaction Engineering for Process R&D. (1) Process
development and scale-up: when used in combination with
* Corresponding author. Telephone: ++41 61 324 44 37. E-mail:
jacques.wiss@pharma.novartis.com.
† Novartis Pharma AG.
‡ Solvias AG.
If no specific literature reference is mentioned for the
evaluation of the spectroscopic data, the correlation between
10.1021/op030009w CCC: $25.00 © 2003 American Chemical Society
Published on Web 06/17/2003
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1059

Scheme 1. Synthesis of tri-n-butyltin azide
The growing awareness of the environmental fate of
organic tin compounds has reflected in the large number of
analytical methods developed for their determination and
specification.10,11 Most methods are too complex to be carried
out routinely in a production plant (goal: determination of
the completion of the reaction).
Finally, detailed knowledge of the reaction kinetics, as
determined from continuous online measurements of reactant
and product concentrations, may help to improve the reaction
conditions and thus increases process safety.
3.1.2. Experimental Section. Tri-n-butyltin chloride was
charged at room temperature in a stirred reactor, and the
reaction system was inerted with nitrogen. Then, the contents
of the reactor was heated to 50 °C. After a first calibration
of the reaction calorimeter, sodium azide was added over 5
min. The reaction mixture was stirred during 21 h at the
same temperature. After a repeat calibration, the reaction
mass was cooled to 25 °C.
3.1.3. Results and Discussion. 3.1.3.1. Normalization of
Spectra. Reaction profiles constructed from normalized
spectra are less noisy. Therefore, trends are more easily
recognizable and interpretable after normalization of the
spectra. Normalizations can help correct fluctuations or
systematic increases or decreases in peak intensity due to
(i) variations in the amount, the composition, or the
morphology of solid, crystalline materials in the reaction
mixture; (ii) the temporary or permanent adherence of
crystals or other particles to the diamond or the sapphire
window of the probes; (iii) the temporary or permanent
formation of thin films on the diamond window of the FTIR
probe; or (iv) changes in the distance between the laser focal
point and the sapphire window of the Raman probe.
Useful peaks for spectra normalizations are peaks resulting
from solvent molecules or from functional groups which are
located far away from the reaction centers of the molecules
involved in the reaction. More general, any peak arising from
species or functional groups which do not undergo any
change in concentration or absolute mass and which can be
regarded as constant is suited for a normalization. Further-
more, the selected peak ideally should be of medium
intensity, i.e., still in the linear range, and located somewhere
in the center of the observed spectral region or close to a
region with significant changes or another region of particular
interest.
the measured wavenumbers and the functional groups was
determined from general spectroscopy books.1-3
3.1. Synthesis of Tri-n-butyltin Azide. 3.1.1. Introduc-
tion. The goal of this study was to evaluate the potential of
online spectroscopic tools for monitoring chemical processes
and detecting deviations from nominal operating conditions.
The literature-known synthesis of tri-n-butyltin azide from
tri-n-butyltin chloride and sodium azide (Scheme 1) was
studied as a model reaction.4
The principal industrial applications of trialkyltin com-
pounds include biocides in marine antifouling paints, fun-
gicides, herbicides, insecticides, miticides, and antifeedants
in agriculture. Tri-n-butyltin azide is of some importance as
a reagent for the preparation of pharmaceutical tetrazole
compounds.5-7 Online monitoring of such a heterogeneous
reaction system by spectroscopic methods may be advanta-
geous for various reasons.
The reaction of crystalline sodium azide with liquid tri-
n-butyltin chloride takes place at the solid-liquid interface
and is therefore very sensitive to changes of reaction
conditions or surface properties. Standard offline methods
may be used to monitor reaction progress; however, they
require repeated sample-taking and -handling during the
analytic process. These operations should be avoided due to
the high toxicity of the reaction mixture and also for reasons
of safety and ecology.
All compounds involved in this reaction have low
maximum allowed concentration (MAK) values. Organotin
compounds have a wide range of toxicity, the most toxic
being trialkyltins as evident from their broad spectrum of
application as biocides. They are capable of blocking the
oxidative phosphorylation in the mitochondria cell.8 Thus,
they influence the basic energetic functions in living cell
systems. Triorganotin compounds can react with active
centers in the cell by forming coordinating links with amino
acids of the cell proteins. Moreover, these compounds can
cause severe skin irritation.9 Further, azides are highly
poisonous and usually sensitive to traces of strong acids and
metallic salts which may catalyze explosive decomposition.
For example, the isolation of explosive silver azide must be
avoided during the offline titration of the unreacted tri-n-
butyltin chloride with silver nitrate.
The normalizations were carried out taking the region
1110.6-1221.1 cm-1 (τCH2 and ωCH2) for the Raman
spectra and the region 815-910 cm-1 (FCH2 an FCH3) for
the FTIR spectra, i.e., the spectra were stretched or com-
pressed so that the peak maxima and the baselines of all the
spectra attained maximum overlap in these regions.
3.1.3.2. Raman Data. The most important changes in the
Raman spectra are summarized hereafter (see Figures 1-3):
(1) Socrates, G. Infrared Characteristic Group Frequencies; Wiley: Chichester,
1994.
(2) Colthup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared and
Raman Spectroscopy; Academic Press: New York, 1975.
(3) Scheinmann, F. An Introduction to Spectroscopic Methods for the
Identification of Organic Compounds; Pergamon Press: Oxford, 1970.
(4) Kricheldorf, H. R.; Leppert, E. Synthesis 1976, 329-330.
(5) Kozima et al. J. Organomet. Chem. 1975, 92, 303, 305, 309.
(6) Sisido et al. J. Organomet. Chem. 1971, 33, 337, 343, 344.
(7) Duncia, J. V.; Pierce, M. E.; Santella, J. B. J. Org. Chem. 1991, 56, 2395-
2400.
(9) Schering, A. G. Organozinn-Verbindungen, Hinweise zur Toxikologie und
sicheren Handhabung 1985, Bergkamen.
(10) Attar, K. M. Appl. Organomet. Chem. 1996, 10, 317-337.
(11) Harrington, C. F.; Eigendorf, G. K.; Cullen, W. R. Appl. Organomet. Chem.
1996, 10, 339-362.
(8) Plum, H. Int. EnViron. Saf. 1982, December, 20-23.
1060
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 1. Raman spectra at intervals of 1 h; normalized data; spectral range: 50-450 cm-1
.
Figure 2. Raman spectra at intervals of 1 h; normalized data; spectral range: 1200-1400 cm-1
.
(i) appearance and decrease of peak at around 125 cm-1
(sole crystal lattice vibration of NaN3 single crystals12):
addition and conversion of NaN3.
3.1.3.3. FTIR Data. The FTIR spectroscopic measure-
ments were performed with an ATR diamond probe. With
this technology, only solvents, liquids, and dissolved com-
pounds can be observed. Thus, for the investigated chemical
reaction, only the two liquid compounds, tri-butyltin chloride
(Bu3SnCl) and tri-butyltin azide (Bu3SnN3), could be de-
tected.
(ii) decrease of peak at around 325 cm-1 (νSn-Cl13):
conversion of Bu3SnCl.
(iii) increase of peak at around 385 cm-1: formation of
Bu3SnN3.
(iv) increases of several peaks between about 1240 and
1390 cm-1 (τCH2, ωCH2, and δsCH3): formation of Bu3-
SnN3; changes in peak intensities probably due to changes
in the dipole moments of the corresponding bonds.
(v) appearance and decrease of peak at around 1360 cm-1
(νsN3 in NaN3): addition and conversion of NaN3.
(vi) appearance and increase of peak around 2065 cm-1
(νasN3 in Bu3SnN3; νsN3 in Bu3SnN3 is said to appear around
1290 cm-1,14 but was not observable as an isolated peak of
significant intensity in our case): formation of Bu3SnN3.
The most important changes in the FTIR spectra are
summarized below (see Figure 4):
(i) Appearance and increase of peak at around 1290 cm-1
(νsN3 in Bu3SnN3): formation of Bu3SnN3.
(ii) Due to the strong absorption of diamond in the spectral
domain between 2000 and 2500 cm-1, the peak caused by
the asymmetric stretching vibrations of N3 in Bu3SnN3 cannot
be observed.
(iii) As for the Raman data, the increase of several peaks
between about 1240 and 1390 cm-1 was observed.
3.1.4. Conclusions. The concentration variations of the
involved compounds were estimated directly from individual
band intensities. Characteristic spectral bands could be
(12) Bryant, J. I. J. Chem. Phys. 1964, 40, 3195/203, 3199/200.
(13) Geissler, H.; Kriegsmann, H. J. Organomet. Chem. 1968, 11, 85-95.
(14) Cheng, H. S.; Herber, R. H. Inorg. Chem. 1970, 9, 1686-1690.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1061

Figure 3. Raman spectra at intervals of 1 h; normalized data; spectral range: 2000-2150 cm-1
.
Figure 4. FTIR spectra at intervals of 1 h; normalized data; spectral range: 650-1800 cm-1
.
observed for all compounds using RAMAN spectroscopy
(see Figure 5). With this method, both solid and liquid
compounds can be observed. The spectra showed sharp, often
isolated peaks (typical for Raman spectroscopy).
in what is probably one of the most important synthetic routes
in aromatic chemistry. Nitro compounds are readily prepared
by direct nitration; when a mixture of o- and p-isomers is
obtained, it can generally be separated to yield the pure
isomers. The primary aromatic amines obtained by the
reduction of these nitro compounds are readily converted to
diazonium salts; the diazonium group, in turn, can be
replaced by a large number of other groups. In many cases
this sequence is the best method of introducing these other
groups into the aromatic ring. In addition, diazonium salts
can be used to prepare an extremely important class of
compounds, the azo dyes.
The catalytic hydrogenation of aromatic nitro compounds
is a class of reaction with a high hazard potential. Neverthe-
less, a large variety of different aromatic nitro compounds
are reduced to amines in the fine chemical industry. The
hazards of such reactions are the following:15
FTIR spectroscopy, however, only allowed the observa-
tion of the formation of tri-n-butyltin azide. Due to the
utilization of an ATR probe, only liquids and dissolved
compounds could be observed. The conversion of Bu3SnCl
cannot be seen (the peak corresponding to νSn-Cl is located
outside of the spectral range of our FTIR instrument). The
spectra show broader peaks, resulting in a more difficult data
analysis. Nevertheless, the corresponding reaction profiles
for the formation of tri-n-butyltin azide agreed well with
those obtained from the RAMAN data (see Figure 6).
The end point of the reaction can be determined quali-
tatively as the point from which the intensity of the product
peaks remains constant within the limits given by the noise
of the methods.
3.2. Synthesis of 2-Chloroaniline. 3.2.1. Introduction.
Reduction of nitro compounds to amines is an essential step
(15) Stoessel, F. J. Loss PreV. Process Ind. 1993, 6, 79-85.
1062
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 5. Normalized reaction profiles (maximum change in peak area set to 100%).
Figure 6. Variation of the relative concentration of Bu3SnN3 measured by FTIR and Raman.
Scheme 2. Synthesis of 2-chloroaniline
(i) hazards related to hydrogen (explosive hydrogen/air
mixtures, very low ignition energy),
(ii) hazards related to the catalyst (handling of pyrophoric
substances, e.g. Raney nickel or platinum on carbon or
palladium on carbon),
(iii) hazards related to the high decomposition energy of
aromatic nitro compounds,
(iv) hazards related to the hydrogenation reaction (very
exothermic reaction; in the case of a cooling failure, the
decomposition of the aromatic nitro compound could be
easily triggered), and
(v) hazards related to unstable intermediates (the ac-
cumulation of intermediate products as N-phenylhydroxyl-
amine can lead to uncontrolled decomposition reactions).
This last point was specially investigated during this work.
Indeed, all hazards concerning the high heats of reaction and
decomposition can easily be determined and quantified using
classical calorimetric methods. Another point was the utiliza-
tion of spectroscopic methods to study heterogeneous
catalytic reactions.16,17
The chemical mechanism18,19 of the investigated reaction
is described in Scheme 2.
(16) Marziano, I.; Sharp, D. C. A.; Dunn, P. J.; Hailey, P. A. Org. Process Res.
DeV. 2000, 4, 357-361.
(17) LeBlond, C.; Wang, J.; Larsen, R.; Orella, C.; Sun, Y.-K. Top. Catal. 1998,
5, 149-158.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1063

Figure 7. Raman spectra of the hydrogenation (385-920 cm-1).
The overall reaction proceeds via nitroso and hydroxyl-
amine intermediates, both of which are reactive and may
undergo undesired condensation or disproportion reactions.20
These reactions are not controlled by the hydrogen mass
transfer and therefore cannot be stopped by reducing the
agitation. Incidents due to the accumulation of N-phenylhy-
droxylamine are reported in the literature. A possible cause
of such accidents is a contamination of the nitro compound
that slows the last reaction step and therefore increases the
proportion of N-phenylhydroxylamine.21 Therefore, the de-
tection and the monitoring of unstable intermediates during
the completion of the reaction are very important for the
assessment of the process safety.
3.2.2. Experimental Section. 1-Chloro-2-nitrobenzene
and methanol were charged at 20 °C in a stirred reactor
(gassing propeller stirrer), and the reaction system was inerted
with nitrogen. Then, Pt/C 5% [catalyst: platinum on carbon]
was added, and a first calibration of the reaction calorimeter
was performed. Then, the reactor was pressurized with
hydrogen (1 bar), and the reaction mixture was stirred at
the same temperature during 20 h. Finally, a repeat calibra-
tion was carried out.
(ii) For the intermediate compound (N-phenylhydroxyl-
amine): the following band is due to deformations in the
-NH-OH group:
(a) appearance of a weak band followed by its decrease
at around 480 cm-1.
(iii) For the final product (2-chloroaniline) the following
bands are due to deformations in the -NH2 group:
(a) appearance and increase of peak at around 470 cm-1,
(b) appearance and increase of peak at around 660 cm-1,
and
(c) appearance and increase of peak at around 820 cm-1.
The concentration variations of the initial compound, of
the hydroxylamine intermediate and of the final product can
be estimated by using band intensities from Raman spec-
troscopy. These concentrations are shown in the Figure 9.
3.2.3.2. FTIR Data. The most important changes in the
FTIR spectra are summarized below (see Figure 10):
(i) for the initial compound (1-chloro-2-nitrobenzene):
(a) decrease of peak at around 780 cm-1: NO2 out-of-
plane wag22,23
(b) decrease of peak at around 853 cm-1: NO2 scissors
deformation;
(ii) for the intermediate compound (hydroxylamine):
(a) the appearance of a weak band followed by its
decrease was observed at around 872 cm-1; this peak
corresponds probably to the N-O stretching;
(iii) for the final product (2-chloroaniline):
(a) appearance and increase of peak at around 680 cm-1:
N-H out-of-plane bending,
3.2.3. Results and Discussion. 3.2.3.1. Raman Data. The
most important changes in the Raman spectra are summarized
below (see Figures 7 and 8):
(i) For the initial compound (1-chloro-2-nitrobenzene):
the following bands are due to deformations of the -NO2
group:
(a) decrease of peak at around 455 cm-1,
(b) appearance and increase of peak at around 749 cm-1:
N-H wagging,
(b) decrease of peak at around 650 cm-1, and
(c) decrease of peak at around 850 cm-1.
(c) appearance and increase of peak at around 930 cm-1:
N-H out-of-plane bending,
(18) Freifeld, M. Practical Catalytic Hydrogenation: Techniques and Applica-
tion; Wiley: Chichester, 1971.
(d) appearance and increase of peak at around 1011 cm-1:
C-N stretching vibration,
(19) Burge, H. D.; Collins, D. J.; Davis, B. H. Ind. Eng. Chem. Prod. Res. DeV.
1980, 19, 389-391.
(20) MacNab, J. I. Runaway Reactions 1981, Paper 3/S, 1-15.
(21) Tong, W. R.; Seagrave, R. L.; Wiederhorn, R. Loss PreV. 1977, 11, 71-
75.
(22) Green, J. H. S.; Harrison, D. J. Spectrochim. Acta 1970, Part 1, 26, 1925.
(23) Green, J. H. S.; Lauwers, H. A. Spectrochim. Acta 1971, Part 1, 27, 817.
1064
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 8. Raman spectra of the hydrogenation (430-500 cm-1).
Figure 9. Normalized reaction profiles (Raman spectroscopy).
(e) appearance and increase of peak at around 2972 cm-1:
N-H stretching vibration.
The concentration variations of the initial compound, of
the hydroxylamine intermediate and of the final product can
be estimated by using FTIR spectroscopy. These concentra-
tions are shown in the Figure 11.
3.2.4. Conclusions. Both methods are utilizable for the
detection of the unstable N-phenylhydroxylamine intermedi-
ate. Under normal reaction conditions, the concentration of
this compound is low (as during our study). Nevertheless,
in case of process deviation, a higher concentration can occur
due to the accumulation of this species. These spectroscopic
methods are appropriate to monitor the reaction to ensure a
safe completion of the chemical process.
the detection of the formation and the accumulation of
undesirable intermediates or side products, so as to determine
the reaction end point, particularly when traditional analysis
is only applicable with difficulty.
A fiber-optic connection between the probe and the
spectrometer is more appropriate for industrial applications
than bypass systems for sampling. With this technology, the
spectrometer can be used in a pilot plant or a production
plant without important modification of the available equip-
ment. This device can be installed in a protected environment
(e.g. in a control room for an explosion-safe installation).
The online evaluation of the data and the process monitoring
can be done with an expert system software.
Online spectroscopy can also be used for other applica-
tions, such as distillation operations, material identification
(through-the-package analysis), catalysts analyses, and so
forth. Undoubtedly, online spectroscopy will become a
standard equipment in the chemical industry in the next years.
4. Final Conclusions
Online spectroscopy is an important tool for the develop-
ment and the monitoring of chemical processes. It allows
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1065

Figure 10. FTIR spectra of the hydrogenation (650-950 cm-1).
Figure 11. Normalized reaction profiles (FTIR spectroscopy).
Acknowledgment
13) for their active cooperation to this work, their helpful
suggestions and keen interest in this problem.
We are grateful to Mr. C. Schwizgebel, Dr. W. Marterer,
Dr. U. Onken (Novartis Pharma AG), Dr. A. Helg (Solvias
AG), Dr. B. Lenain (Kaiser Optical Systems) and Dr. Ph.
Marteau (Laboratoire d’Ingenie´rie des Mate´riaux et des
Hautes Pressions, CNRS (UPR 1311), University of Paris
Received for review March 4, 2003.
OP030009W
1066
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development