La calorimetria di reazione per lo Sviluppo
e la Sicurezza dei Processi Chimici
Barreca, G.; Bravin, F.M.; Gatti, M.M.; Gigante, L.; Pasturenzi, C.; Poggiali, A.
“Ester reduction: a risky choice”
Milano, 21 Ottobre 2015
Introduction – Aim of the work
Investigation of the main causes of a chemical accident occurred during
the synthesis of a key intermediate of Aliskiren
O
O
O
OH
O
OH
O
H
N
O
O
NH2
O
Aliskiren
The work was commissioned to a Chinese company for tons quantities.
They already produced the intermediate with their own technology on pilot
scale
During the first industrial batch (150 kg batch size) the reactor blow up, and fire
injured five people.
NH2
Introduction – The fact
The accident occurred during the ester reduction.
O
AlCl3,NaBH4
O
O
O
O
O
OH
DME
O
O
Although the most common method to reduce carboxyl groups to alcohols is
using LiAlH4 as reducing agent on an acid substrate, in China is used to use the
complex AlCl3/NaBH4/DME as reducing agent considering it cheaper than the
former one.
O
O
O
O
OH
LiAlH4
O
O
OH
THF
O
They were very confident with this process because they commonly used it for
reductions in this and other processes.
Introduction – The fact
Synthesis ensues dropping of a toluene solution of the ester on to a mixture of
AlCl3 and NaBH4 in dimethoxy ethane.
AlCl3 + 3 NaBH4 -> Al(BH4)3 + 3 NaCl
Al(BH4)3 is the active reducing agent.
Developed reaction conditions:
• Starting temperature: 20-25 ºC
• Max reaction temperature: 50 ºC
• Addition of ester solution: dropwise by controlling the exothermicity
But…
Reactor was too big to allow the temperature probe to be dipped into the mass.
Anyway, plant director decided to proceed cooling the jacket at the minimum
temperature reachable with the industrial brine (-22ºC).
Introduction – The fact
The production method provided the addition of the substrate in about one
hour with temperature control. After 2/3 of addition:
Some mists began to surface the reaction mass.
Pressure began to rise.
An operator opened the vacuum valve trying to reduce the rising internal
pressure.
Reactor blew up and fire invaded the department.
Five persons was injured by fire.
What were wrong?
Investigation – Literature review
The owner of the company engaged Chemo to understand the causes of the
accident.
We began our investigation starting from an analysis of the available literature.
As usual, we accessed Bretherick’s and Sax’s as the very first two sources of
information.
Starting from this two sources we retrieved some interesting articles focussing
the fact that this kind of reducing system is not safe as it seems.
Literature Review – Sax’s
SFF500 CAS: 16940-66-2 HR: 3 SODIUM BOROHYDRIDE
Ignites in air above 288 ºC when exposed to spark. Potentially explosive reaction with aluminum
chloride + bis(2-methoxyethyl) ether. Reacts with ruthenium salts to form a solid product which
explodes when touched or on contact with water. Reacts to form dangerously explosive hydrogen gas
on contact with alkali, water and other protic solvents (e.g., methanol, ethanol, ethylene glycol, phenol),
aluminum chloride + bis(2-methoxyethyl)ether. Reacts violently with anhydrous acids (e.g., sulfuric,
phosphoric, fluorophosphoric) to form diborane. Violent exothermic reaction with dimethyl formamide
has caused industrial explosions. Mixtures with sulfuric acid may ignite. Incompatible with palladium,
diborane + bis(2-methoxyethyl) ether, polyglycols, dimethylacetamide, oxidizers, metal salts, finely
divided metallic precipitates of cobalt, nickel, copper, iron, and possibly other metals. Emits flammable
vapors on contact with acid fumes. Materials sensitive to polymerization under alkaline conditions, such
as acrylonitrile, may polymerize upon contact with sodium borohydride. Avoid storage in glass
containers. When heated to decomposition it emits toxic fumes of Na2O. See also HYDRIDES, BORON
COMPOUNDS, and SODIUM COMPOUNDS.
Literature Review – Bretherick’s
Sodium tetrahydroborate (Sodium borohydride) [16940-66-2] NaBH4
Aluminium chloride, Bis(2-methoxyethyl) ether
Addition of a 4% solution of sodium tetrahydroborate in diglyme containing 0.09% of
water to a 27% solution of aluminium chloride in the same solvent led to a violent
explosion, attributed to formation and ignition of hydrogen. The ignition source arose
from contact of the hydroborate solution with the solid chloride, as demonstrated
experimentally. Nitrogen purging is essential for all hydride reductions, and also for
hydroboration, organoborane, Grignard and organometallic reactions generally. Previous
work had shown that clear solutions of the sodium tetrahydroborate—aluminium chloride
reagent did not ignite in dry air, but the solid-containing reagent could lead to ignition.
Literature Review – de Jongh, H. A. P.,
Chem. Eng. News, 1977, 55(31), 31
The ignition most probably
resulted from reaction of
NaBH4 solution with
some undissolved AlCl3.
Preparation of this reducing agent had been carried out in exactly
the same way 12 times on a 10-times smaller scale.
Literature Review – Brown, H.C., Chem. Eng.
News,1977, 55(35), 5
I agree with de Jongh that the presence
of water, solid AICI3, and possibly sodium
borohydride, indicated as present in the
reaction mixture that exploded, could
produce a hydrogen-air mixture in the
explosive range, ignited by spontaneously
inflammable materials produced in the
interaction of solid AICI3 with NaBH4
Literature Review – Riv. Combustibili,
Vol. 54, fasc. 2, marzo-aprile 2000.
Experimental stability data (C80)
on NaBH4 in different solvents:
DMF
DMAC
Alcohols
N-methyl pyrrolidone
Acetonitrile
Pyridine
TEA
Diglyme
Methyl tert buthyl ether
also in presence of water and
contaminants.
NaBH4: thermally stable in diglyme and Methyl tert buthyl ether
Literature Review – JACS, 1949, 71, 2950
Al(BH4)3. This volatile
compound (b.p. 44.5 ºC)
was first prepared by
Schlesinger and associates,
and was found by them to
ignite spontaneously when
exposed to laboratory air.
Since Schlesinger has
reported that aluminum
borohydride is subject to
rapid hydrolysis, it would
appear that this reaction is a
prerequisite to explosion at
room temperature. In the
absence of water vapor,
decomposition at a higher
temperature may serve the
same purpose.
Literature Review – Chemistry of the boron
hydride Q. Rev. Chem. Soc., 1955, 9, 174-201
Literature Review – J. Haz. Mat., 142 (2007)
639-646
Clearly, the presence of a metal
chloride boosts sodium borohydride
decomposition to the extent that such
contact raises a serious safety
concern. Furthermore, this
enhancement of sodium borohydride
decomposition by a metal chloride
was new information for some of our
staff.
Literature Review – J. Haz. Mat., 142 (2007)
639-646
Temperature history of sodium borohydride solutions in the APTAC.
Literature Review – J. Haz. Mat., 142 (2007)
639-646
Pressure history of sodium borohydride solutions in the APTAC
Literature Review - French Ministry of
Environment – DPPR/SEI/BARPI – ARIA no.4708
Hydrogen explosion - September 1, 1993
Gennevilliers – [Hauts de Seine] France
On September 1st, 1993 at 6.45 am, an explosion and fire occurred in one of
the plant's workshops during a chemical reaction. In this reaction, an imide is
converted into an amine in anhydrous conditions and in the presence of
aluminum chloride-activated sodium borohydride. The reaction, inerted
with nitrogen, is performed in a triethylene glycol methyl ether (triglyme) and
chloroform mixture. The synthesis was already performed at an industrial scale
21 times in 2 years.
Four of the eight people present in the workshop were hospitalized for burns
and respiratory difficulties : 2 were discharged from the hospital the same day.
The cost of property damage was evaluated at 14 MF.
The €15 of the "economic consequences" rating is 3: the amount of property
damage is estimated to be 14 MF, or roughly 2.13 M€ (€15 between 2 M€ and
10 M€).
http://www.aria.developpement-durable.gouv.fr
Experimental part – Easymax
Easymax trial was performed to verify heat accumulation
Reaction temperature was set up at 25 ºC and -20 ºC
Methylester was dropped into the reducing mixture by syringe pump
Experimental part – Easymax - results
While at 25 ºC the reaction proceeds immediately (green line) and shows
minimal heat accumulation when we executed the experiment at -20 ºC we
noticed an induction time (red line) before the reaction starts and a strong heat
accumulation.
Reaction at -20 ºC shows an heat accumulation of 58%.
Experimental part - Phitec
We began our investigation trying to simulate the occurred runaway.
To do this we set up an adiabatic low Phi factor experiment first (PhiTecII).
After loading the reducing agent into the calorimetric bomb we dosed 2/3 of the
reacting mixture at 30 ºC
Experimental part – Phitec - results
The figure below shows the temperature and pressure trends vs. time.
The red arrows shows the injection time.
T0 = 30 ºC
Tfin = 88 ºC
Pfin = 9,3 bar
∆Tad = 66 K
Max SHR = 188 ºC/min
Max Prate = 85 bar/min
Gas Analysis at the
end of the test
Toluene, DME,
Methane, Hydrogen,
Oxygen, Nitrogen
All data presented are Φ corrected.
Experimental part – RC1
To evaluate the heat of the reaction and the adiabatic temperature rise we set up
an RC1 experiment.
The substrate solution was added into the reactive system in about 1 hour through
a dosing funnel.
Due to the lack of an automatic doser we couldn’t evaluate the heat accumulation in
standard operative conditions.
Experimental part – RC1 - results
40.0
100%
35.0
Qr
Me-ester
30.0
80%
Heat (W)
25.0
60%
20.0
15.0
40%
10.0
20%
5.0
0.0
0
15
30
45
60
75
90
105
120
135
0%
150
time (min)
∆Hreaz: 219 kJ/mol
Accumulation: n.a. (probably very low at 30 º C - from EasyMax data) ~22%
∆Tad: 86 K
MTSR : Teb and 29% solvent loss
∆H reaz is calculated on limiting reagent.
∆H>0 correspond to an exothermic effect
∆T ad is calculated on final reaction mass and average cp.
Data refers to performed experiment
Experimental part – Another choice
O
O
O
O
OH
LiAlH4
O
O
OH
THF
O
On request of our supplier we provided to study the reduction with another reducing
system.
We selected classical LiAlH4 as reactant and changed substrate from ester to the
corresponding acid.
DME was replaced with THF.
A solution of the substrate was added dropwise into the suspension of LiAlH4 at 4550 ºC.
Substrate addition must be modulate on hydrogen evolution.
We set up an RC1 experiment.
Experimental part – RC1
10,0
Qr
Acid
100%
8,0
80%
Heat (W)
6,0
60%
4,0
40%
2,0
20%
0,0
0
15
30
45
60
75
90
105
120
135
0%
150
time (min)
∆Hreaz: 157 kJ/mol
Accumulation: < 5%
∆Tad: 29 K
MTSR : Teb and 4% solvent loss
∆H reaz is calculated on limiting reagent.
∆H>0 correspond to an exothermic effect
∆T ad is calculated on final reaction mass and average cp.
Data refers to performed experiment
Literature Review – Bretherick’s
Lithium tetrahydroaluminate (Lithium aluminium hydride) [16853-85-3]
LiAlH4
Care is necessary in handling this powerful reductant, which may ignite if lumps are
pulverized with a pestle and mortar, even in a dry box. (…)The explosive thermal
decomposition of the aluminate at 150-170 C is due to the interaction with partially
hydrolysed decomposition products. (…).
Experimental part – RC1 – Data Summary
comparison
∆H reaz
accumulation
(kJ/mol)
∆Τ ad
Teb (ºC)
MTSR
(K)
Tp
(ºC)
NaBH4/AlCl3
218
n.a. (~22%)
86,5
30
85
Teb with 29%
solvent loss
LiAlH4
157
< 5%
28,6
40
65
Teb with 4%
solvent loss
Although RC1 experiment was done adding all the substrate in semi-batch
conditions, we advised an in process control every 20% of substrate added to
verify no reactant accumulation.
∆H reaz is calculated on limiting reagent.
∆H>0 correspond to an exothermic effect
∆T ad is calculated on final reaction mass and average cp.
Data refers to performed experiment
Results
From the data we obtained we can confirm that they started the reaction at a too low
temperature.
Easymax experiment demonstrated that at lower temperature the reaction has an
induction time.
This led to an accumulation of reagents (loaded bomb).
Easymax experiment demonstrated that at -20 ºC the reaction shows an heat
accumulation of 58%.
Once the reaction has started, temperature and pressure arose in few seconds
bringing on instant reactor opening.
Adiabatic experiment demonstrated that once the reaction takes place with the quantities
of the added reagent leads to an immediate temperature and pressure increase.
Leaking gases and solvents (methane, hydrogen, DME, toluene) have caused
the fire.
FTIR, GC-MS and Micro GC analysis of Phitec’s end of experiment gases demonstrated
that hydrogen and methane arises from the decomposition mixture.
Results
RC1 experiment showed the heat of reaction and permits to scale up the dosing in
safety conditions for both systems if boundary conditions were respected.
But Al(BH4)3 is the hazardous one.
the contact of the hydroborate solution with the solid chloride is an ignition
source
aluminum borohydride ignites spontaneously when exposed to laboratory air. It
is subject to rapid hydrolysis, it would appear that this reaction is a prerequisite
to explosion at room temperature.
Conclusions
Never trust false friends
the lower the temperature the more safety
the lower the accumulation the more safety
if I can drink it is not dangerous
Bhopal
India, 1984
we are used to do this way and nothing ever
happened
Oppau
Germany, 1921
we are confident this way because we have
studied
Literature Review – Bretherick’s
CAN OF BEANS
Foote, C. S., private comm., 1965
An unopened can of beans, placed in a laboratory oven originally at 110 ºC but
later reset to 150 ºC, exploded causing extensive damage. Comments were
judged to be superfluous.