Applied Catalysis A: General 318 (2007) 190–198

www.elsevier.com/locate/apcata

Deactivation of nitrile hydrogenation catalysts: New mechanistic

insight from a nylon recycle process
Michael W. Duch a, Alan M. Allgeier b,*
a
DuPont Corporate Center for Analytical Science, P.O. Box 80302, Wilmington, DE 19880-0302, United States
b
DuPont Textiles and Interiors (now Invista) Experimental Station, P.O. Box 80302, Wilmington, DE 19880-0302, United States
Received 12 September 2006; received in revised form 29 October 2006; accepted 5 November 2006
Available online 14 December 2006

Abstract
In the recycling of nylon-6 and nylon-6,6, high temperature ammonolysis leads to reversion of the polymers to monomer units and dehydration of
amide monomers to nitrile molecules. The resulting product is a mixture of 6-aminocapronitrile, caprolactam, adiponitrile, hexamethylenediamine and
other species. To complete the recycle loop the nitrile molecule can be hydrogenated to hexamethylenediamine. This liquid phase hydrogenation has
been studied over Raney1 Ni and Raney1 Co catalysts at less than 100 8C and 3.5 MPa in semi-batch mode. Raney1 Ni exhibited rapid deactivation in
the absence of sodium hydroxide, while Raney1 Co provided long catalyst life. In comparison studies, Raney1 Ni also deactivated during
hydrogenation of pure adiponitrile, in the absence of sodium hydroxide. While sodium hydroxide inhibits Raney1 Ni deactivation, it presents an
economic and environmental challenge for waste handling. Analysis of recovered catalyst samples by ESCA showed an increase in nitrogen and
carbon on the surface of deactivated Raney1 Ni samples compared to recovered Raney1 Co and control samples. The observed overlayer was
consistent with the formation of oligomeric secondary amines, which covered the catalyst surface and prevented further reaction. Amine coupling
reactions were favored for Raney1 Ni compared to Raney1 Co.
# 2006 Elsevier B.V. All rights reserved.

Keywords: Deactivation; Raney; Nitrile hydrogenation; Nylon; Recycle; Polyamine; Hexamethylenediamine; Adiponitrile; Caprolactam; Ammonolysis

1. Introduction and backing materials such as rubber, polyurethane and even

Recycling of waste nylon carpets has become an increas-
ingly important problem in recent years [1] since landfill
capacity is nearing maximum utilization and nylon fibers can
last 30–40 years in conventional landfills [2]. In the United
States, a minimum of 25% recycle content is required in
government carpet installations, and major fiber producers have
developed and/or commercialized recycling processes. Carpet
waste generally includes both industrial carpet waste and post-
consumer carpet waste. Challenges to carpet recycling include
not only the logistic of collection and distribution but also the
complexity of materials used in carpet construction. Carpet
may include the carpet facing polymer such as nylon or
polypropylene, as well as the dyes, adhesives, compounding
* Corresponding author at: Amgen Inc., United States. Tel.: +1 805 313 5237;
fax: +1 805 480 1346.
E-mail address: Allgeier@amgen.com (A.M. Allgeier).
0926-860X/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2006.11.003
calcium carbonate. Furthermore, nylon polymers present a
critical problem arising from their high melting point and low
solubility. These characteristics, which are valued for produ-
cing high quality carpets, lead to challenges to recycling
processes. While recycling of some polymers such as
polystyrene and polyethylene may rely on dissolution or
melting, many nylon recycling processes rely on conversion of
the polymer back to monomer for effective recycling.
The nylon recycle process for recovering monomers from
post-industrial and post-consumer nylon-6 and nylon-6,6 carpet
fibers is a multistep operation. A plethora of depolymerization
processes have been explored in past laboratory studies,
including steam stripping, water hydrolysis, nitric acid
hydrolysis, and acetic acid acetolysis [3], but depolymerization
via ammonolysis has been shown to be a technically feasible
route for the recovery of high purity nylon intermediates [4].
In the ammonolysis process (Fig. 1), first the recovered
carpets are mechanically separated to remove the carpet backing
and latex from the nylon fiber. The nylon (-6 and/or -6,6) fiber is
 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198 191

Fig. 1. Block diagram of nylon-6 and -6,6 recycling process. ADN, adiponitrile; ACN, 6-aminocapronitrile; HMD, hexamethylenediamine; CPL, caprolactam.

then depolymerized at elevated temperature and pressure in the
presence of excess ammonia to form a mixture of monomers. In
this chemistry, the secondary amides of the nylon polymers
react with ammonia to break the nylon chain and form a
primary amide and an amine. The primary amide can
subsequently be dehydrated to form a nitrile group. The
resulting ammonolysis product is predominantly a mixture of
four major components. Hexamethylenediamine (HMD) and
adiponitrile (ADN) are formed from nylon-6,6, while 6-
aminocapronitrile (ACN) and caprolactam (CPL) are formed
from nylon-6 (Scheme 1). In addition to these major
components, several degradation products and products from
incomplete reaction remain, including 5-cyanovaleramide, 6-
aminocaproamide and low molecular weight oligomers. The
ammonolysis product may subsequently be separated by
distillation into components, followed by hydrogenation of the
components to make HMD or the ammonolysis product may be
directly hydrogenated to form HMD, CPL and other products.
CPL remains intact during the hydrogenation reaction. The
crude recycled HMD and CPL can be separated by distillation
and subsequently refined to produce monomer grade HMD and
CPL, respectively. The CPL can be directly polymerized to
nylon-6 or it can be recycled back to the ammonolysis reactor,
where it can be converted into ACN.
A significant factor impacting the feasibility of the nylon
recycle process is the productivity or lifetime of the
hydrogenation catalyst. Nitrile hydrogenation catalysts are
known to deactivate. For acetonitrile hydrogenation in the gas
phase over nickel catalysts, the deactivation has been attributed
to a nitrile ‘‘cracking’’ mechanism, which leads to an unreactive
nickel carbide surface [5]. We, recently, investigated the
deactivation of Raney Ni catalysts during liquid phase nitrile
hydrogenation and suggested an alternate deactivation mechan-
ism [6]. Rather than a ‘‘cracking’’ mechanism, we suggested
that amine condensation reactions lead to insoluble oligomers,
which adsorb to active catalyst sites and prevent further
reaction. The mechanism was consistent with three key
observations:
 Deactivation was observed for a,v-dinitriles, while mono-
nitriles such as hexanenitrile showed no deactivation.
 The only mononitriles that did exhibit deactivation were
aminonitriles, which are capable of participating in amine
condensation oligomerization.

Scheme 1. The idealized ammonolysis and hydrogenation chemistries of nylon-6,6 and nylon-6 recycle.
192 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198
 Deactivation rate was dependent upon dinitrile chain length
with succinonitrile deactivating slowly but dodecanedinitrile
deactivating very rapidly. A corresponding correlation of
oligomer solubility was inferred.
In this paper we present a study of cobalt and nickel
hydrogenation catalysts in the context of a nylon recycle
process and compare the selectivity, productivity and,
especially lifetime for two catalysts. Nitrile hydrogenation
catalysts commonly comprise cobalt and nickel in their active
forms but other transition metals including iron, ruthenium and
palladium may be useful in the hydrogenation of nitriles and
thereby, the nylon recycle process.
2. Experimental
The depolymerized nylon used in the hydrogenation process
was obtained by the ammonolysis of a mixture of nylon-6 and
nylon-6,6, described elsewhere [4a]. Major products from the
ammonolysis reaction include hexamethylenediamine (HMD),
5-cyanovaleramide (CVAM), and adiponitrile (ADN) from
nylon-6,6; and caprolactam (CPL), 6-aminocaproamide
(ACAM) and 6-aminocapronitrile (ACN) from nylon-6. The
crude products also contain over twenty separate partial
reaction or degradation products in small quantities, including
amide oligomers, amine coupling products, carbon dioxide,
ammonia, and water.
The carbon dioxide in the ammonolysis products can react
with the HMD present in the product to form carbamate
derivatives, which form solids at room temperature. The
carbamate derivatives, when held below their decomposition
temperature of about 120 8C, can cause plugging in commercial
equipment. Therefore, carbon dioxide as well as ammonia is
usually removed from the depolymerized products by conven-
tional distillation techniques before the hydrogenation process.
Semi-batch reactions were conducted in either a 100 mL or a
300 mL pressure vessel constructed of Hastelloy-C1 (nickel
alloy) and stirred via an impeller to maintain efficient gas/
liquid/solid contact. The reactors were heated with external
band heaters and the 300 mL reactor was equipped with an
internal cooling coil for efficient temperature control. All
reactants were charged to the reactor, which was then
assembled, tested for pressure leaks, and heated to reaction
temperature while stirring at very low rates under a low
hydrogen pressure. On reaching reaction temperature, the
pressure in the reactor was increased to the operating pressure
and the stirring rate was increased to 700–1000 rpm to initiate
the reaction [7]. Hydrogen was fed on demand from a 1 l steel
reservoir to maintain constant pressure in the reactor during
operation. The pressure of the hydrogen reservoir was
monitored to follow the course of the reaction. All pressure
vessels were protected with a rupture disk or emergency
pressure relief valve.
All products were analyzed by gas chromatography using 1-
methylpyrollidinone as internal standard, a 30 m, 1 mm RTX-5
Amine column from Restek, a method with maximum heating
temperature of 275 8C and a flame ionization detector.
The hydrogenation process was performed at a relatively
low temperature (<100 8C) and pressure (<70 bar), which
advantageously offers reduced investment cost for an industrial
process. Iron and chromium promoted Raney1 Ni 2400, and
chromium and nickel promoted Raney1 Co 2724 catalysts,
obtained from W.R. Grace & Co, were used in this study [8].
The Raney catalysts were prepared by leaching Al from a metal
aluminum alloy. The BET surface area of Raney1 Ni 2400 and
Raney1 Co 2724 were 140 m2/g and 76 m2/g, respectively, and
the active surface area of the Ni and Co catalysts were 52 m2/g
and 18 m2/g, respectively, based on CO chemisorption of
30 cm3 CO/g and 8 cm3 CO/g for Ni and Co, respectively [9].
These values correspond to 1.34 mmol/g and 0.36 mmol/g at
1:1 stoichiometry.
Samples of Raney catalysts are pyrophoric and were
generally handled as aqueous slurries. Samples for ESCA
analysis were passivated by first washing the samples with three
portions of water (approximately 15 mL each), followed by
three portions of methanol (approximately 15 mL each). The
methanol slurries were then bubbled with laboratory grade
nitrogen for 10 min and finally they were purged with air to
dryness.
ESCA spectra were obtained using a Physical Electronics
PHI 5600ci spectrometer operated with an Al monochromatic
X-ray source at 350 W power (15 kV and 23.3 mA). High-
resolution spectra were recorded at an analyzer pass energy of
23.5 eV and a take-off angle of 458. An analyzer slit of
dimensions 0.8 mm  2 mm was used to image the sample
analysis area. Spectra were scanned at 0.1 eV step interval and
50 ms step time. A PHI Model 06-350 ion gun and a Model NU-
04 neutralizer were used to compensate for charging effects.
The pressure in the analysis chamber was typically below
5  108 Torr during analysis. The number of scans used to
accumulate spectra for each element varied, depending on
signal intensity. All spectral peak positions were referenced
relative to the carbon 1s (C 1s) line at 284.8 eV. The linearity of
the electron binding energy scale was checked by recording the
Au 4f7/2 line at 84.0 eV and the Cu 2p3/2 line at 932.7 eV. The
catalyst samples were mounted for analysis on a PHI Model
190 flat sample holder using double-coated tape (3M Scotch1 #
665). PHI MultiPak@ software version 6.0A was used for data
analysis.
3. Results and discussion
Two main process options exist for the recycle of nylon
polymers via ammonolysis (Fig. 1) and the options may have
very different implications for the lifetime of the hydrogenation
catalyst, as well as the affordability of the process. In option A
the crude ammonolysis product is distilled to form one or more
pure fractions containing adiponitrile (ADN), 6-aminocaproni-
trile (ACN) and hexamethylenediamine (HMD) and possibly
caprolactam (CPL), in addition to a high boiling impurities
stream. Oligomers, other unreacted amides and high boiling
products are thus separated before hydrogenation. Option B
offers an economically attractive process simplification by
eliminating the intermediate distillation step and treating the
 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198 193

Ni 2400 is a common catalyst for the production of HMD, it

Fig. 2. Comparison of Raney-type nickel and cobalt catalysts used under
identical conditions for the hydrogenation of crude nylon-6 ammonolysis
product, i.e. containing residual by-products and oligomers. (&) Raney1
Co 2724 and (~) Raney1 Ni 2400.
entire crude ammonolysis product in the hydrogenation step but
demands a robust catalyst system, which will not deactivate in
the presence of ammonolysis by-products. Indeed, the
ammonolysis product is a complex mixture, derived from a
reaction at 320 8C. In one example, a feedstock of only nylon-6
yielded a crude ammonolysis product, which after removal of
dissolved ammonia, water and carbon dioxide (CO2 from
decomposition reactions) by standard distillation techniques,
comprised 43% ACN, 46% CPL and 2% 6-aminocaproamide
and about 9% higher molecular weight components. To evaluate
the feasibility of Option B (Fig. 1), we subjected this crude
product to hydrogenation without further purification.
3.1. Demonstration of reactivity and process
Drawing heavily from prior experience in hydrogenation of
ADN to ACN and/or HMD, we decided to restrict the scope of
this investigation to Raney-type nickel and cobalt catalysts,
which can be used at low pressures (<6.89 MPa, i.e.
<1000 psig) and temperatures (<100 8C) [10]. While Raney1
showed strikingly different behavior than Raney1 Co 2724 for
the hydrogenation of the above nylon recycle crude product.
Thus in a 100 mL reactor, 0.7 g catalyst, 1.7 g water, 25 g
methanol, and 25 g of the above crude nylon-6 ammonolysis
product were reacted with hydrogen at 3.5 MPa (500 psig)
pressure and 75 8C. The Raney-type Ni catalyst showed rapid
deactivation in comparison to the Co catalyst, and reached
merely 40% conversion before the reaction essentially ceased.
The Co catalyst reaction reached essentially full conversion
after 340 min (0.18 mol H2 consumed versus 0.19 mol
theoretical) (Fig. 2). Concentration profiles for the major
components of each reaction are shown in Table 1.
The product from the Co catalyst comprised 2% ACN, 32%
HMD, 46% CPL and other by-products, suggesting that some
ACN had been converted to undesirable by-products including
amine condensation products (bis(hexamethylene)triamine = -
bis(6-aminohexyl)amine, BHMT) and perhaps amine amide
condensation products.
To develop further understanding of Raney1 Ni deactivation
we evaluated the kinetics of the reactions described in Table 1.
The nylon recycle feed presents many challenges to a thorough
kinetic analysis, most notably the number of different
components in the feed, which may have significant adsorption
strength to the catalyst. Prior studies of the kinetics of ADN and
ACN hydrogenation have been published and describe the
complexity of this system [11]. Additionally, a study of
butyronitrile hydrogenation chemistry over Raney Ni catalysts,
showed that the primary amine product of nitrile hydrogenation
competes with hydrogen for adsorption sites [12].
Acknowledging these limitations, the data for Raney1 Co
(Table 1), was modeled as first order and made a good fit to that
reaction rate equation (d[ACN]/dt = 0.0089[ACN] + 0.117;
R2 = 0.985). Hydrogen concentration was constant throughout
[7]. Raney1 Ni deviated significantly from first order behavior
(Fig. 3).

Table 1
Profile of reaction components for Raney Co and Raney Ni, provided as weight percent
Time (min) ACN (%) HMD (%) CPL (%) Sum of coupling productsa (%) Other componentsb (%)
1
Raney Co 2724
0 39.0 0.0 47.0 0.2 11.0
60 19.2 17.2 46.6 3.3 13.7
120 12.7 22.8 46.3 4.4 13.8
180 6.0 28.4 46.2 5.6 13.8
240 3.5 30.8 46.2 6.1 13.5
340 2.0 32.0 46.2 6.3 13.6
Raney1 Ni 2400
0 39.0 0.0 47.0 0.2 11.0
30 34.5 4.4 46.6 0.7 14.0
60 30.9 7.3 46.6 1.7 13.5
120 25.0 11.6 46.7 3.0 13.7
180 22.0 13.5 46.6 2.7 14.2
240 20.5 14.7 46.8 4.1 13.8
300 20.1 14.9 46.8 4.0 14.3
340 19.6 15.0 46.5 4.2 14.7
a
The sum of ACN hydrogenation coupling products: tetrahydroazepine (THA), hexamethyleneimine (HMI) and bis(hexamethylene)triamine (BHMT).
b
Includes adiponitrile, 1-amino-2-cyanocyclopent-1-ene, 5-cyanovaleramide, 6-aminocaproamide, and several other components, which were not fully identified.
194 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198

Fig. 3. First order plots for Raney-type nickel and cobalt catalysts for the
hydrogenation of crude nylon-6 ammonolysis product, i.e. containing residual
by-products and oligomers. (&) Raney1 Co 2724 and (~) Raney1 Ni 2400.
Fit lines are shown for the Raney1 Co data set and for the first 20% conversion
of the Raney1 Ni data.
Fig. 4. First order plots for estimated [sites] of Raney1 Nickel 2400 during the
hydrogenation of crude nylon-6 ammonolysis product, suggesting an induction
period and late onset of first order deactivation behavior.

By analogy to Raney1 Co, the reactivity inherent to Raney1 With the intrinsic rate constant k0 in hand, the equation
Ni in the absence of deactivation should also fit a first order ln([ACN]0/[ACN]) = kt could be applied to each integral region
reaction rate equation. One can utilize the method of initial first (defined by two consecutive data points) of the data of Fig. 2
order kinetics to determine an idealized reaction rate profile for and could be solved for [sites], where k = k0 [sites]. In doing so
Raney Ni in the absence of deactivation. Thus a rate constant we made the simplifying assumption that [sites] = constant
(k = 0.0039 min1) was determined from the first 20% within an integral region. The data are shown in Table 2. A first
conversion of the Raney Ni reaction, where the influence of order plot derived from the determined [sites] is shown in
deactivation is likely negligible. Fig. 4. There appears to be an induction period in the
It remains to make some assessment of the rate of deactivation, followed by a first order decay in sites. This may
deactivation or loss of active sites over Raney1 Ni. The above suggest a period of coupling reactions, which produce soluble
discussion employed apparent first order behavior: rate = - amine dimers/oligomers and only after building a critical
k[ACN]. The Raney1 Ni reaction must also be dependent upon molecular weight do the oligomers adsorb to the active sites.
the concentration of active sites, which are decreasing with Molecular weight growth in condensation polymerizations
conversion. Thus rate = k0 [ACN][sites]x. In the absence of follows such a profile, qualitatively. Further supporting the
deactivation, k0 [sites]x would equal the first order rate constant, significance of coupling reactions, we note on Table 1 the
k. For the Raney1 Ni reaction, deactivation is taken as increasing concentration of soluble coupling products as a
negligible in the first 20% of conversion. Knowing the function of reaction time. Our analytical method did not allow
concentration of active sites at time zero (CO chemisorption) for analysis of higher order oligomers in the liquid phase. The
and assuming x = 1, we solve for k0 . data of Fig. 4 certainly show deviations from first order
behavior or scatter in the data late in the reaction. These may be
k ¼ k0 ½sites; 0:0039 min1 ¼ k0 ð0:7 g  1:34 mmol=gÞ; attributable to the reaction being essentially dead after 180 min.
k0 ¼ 0:0042 min1 mmol1 Changes in reactant concentration are small after this point and
imprecision in the analytical methods is accordingly more
significant. Further investigation of kinetic experiments on a
Table 2 more controlled system (e.g. neat ACN hydrogenation), along
Determination of the decrease in active sites by fit to experimental [ACN] data with quantitation of all coupling products, will provide further
and a first order reaction rate law for Raney1 Ni 2400
insight into this complex catalytic system.
Time mol k (min1) = ln[(ACNt  1)/ [Sites] ln[sites(0)/ While the above discussion points to amine condensation
(min) ACN (ACNt)]/Dt (mmol) = k/k0 sites] reactions generating a strongly adsorbed deactivating layer on
0 0.0871 0.937a 0 the catalyst, we recognized that the feed used in those
30 0.0771 4.1  E-3 0.987 0.052 experiments, prepared by process option B, contained a
60 0.0690 3.7  E-3 0.886 0.055 significant concentration of amide oligomers. To rule out the
120 0.0558 3.5  E-3 0.853 0.094
180 0.0490 2.1  E-3 0.517 0.595
dependence of Raney Ni deactivation exclusively upon the
240 0.0457 1.2  E-3 0.286 1.19 presence of high molecular weight oligomers and by-products
300 0.0448 0.3  E-3 0.079 2.47 derived from process option B, a further experiment was
340 0.0438 0.5  E-3 0.127 2.00 performed. In this case, process option A was applied to a
a
Sites at time = 0 were taken from measured CO chemisorption; mixture of nylon-6 and nylon-6,6 but the crude ammonolysis
k0 = 0.0042 min1 mol1. product was purified by removal of high boiling oligomers via
 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198
Fig. 5. Comparison of Raney-type nickel and cobalt catalysts for the hydro-
genation of purified nylon ammonolysis product, i.e. a mixture of HMD and
ACN with only small amounts of by-products and oligomers. (&) Raney1 Co
2724, (~) Raney1 Ni 2400, and (*) Raney1 Ni 2400 with NaOH added to the
reaction.
distillation. One major fraction comprised predominantly HMD
and ACN (24.4% HMD, 66.8% ACN, 0.03% ADN, 0.02% CPL
and small amounts of other impurities). Hydrogenation of this
fraction using Raney1 Ni 2400 also led to rapid deactivation,
while a corresponding reaction with Raney1 Co 2724
proceeded to full conversion (Fig. 5). Conditions for these
two batch reactions were identical to the reactions of Fig. 2,
with the exception of the feed composition. Addition of 0.15 g
50% NaOH(aq.) to the reaction, inhibited Raney Ni deactiva-
tion (Fig. 5). It has been reported that addition of sodium
hydroxide improves the selectivity of nitrile hydrogenation
reactions [11] and we have found that it, likewise, hinders the
deactivation of Raney Ni in the nylon recycle process.
Clearly the Raney-type Co catalyst was very robust in the
hydrogenation step of nylon recycle via ammonolysis. While
the use of sodium hydroxide with Raney Ni did mitigate
deactivation problems, it may lead to further complications in
downstream processing. The high temperatures of distillation
may encourage side reactions catalyzed by the base.
Additionally, sodium hydroxide presents an environmental
and economic challenge. After the hydrogenation process,
NaOH must be removed via neutralization or via a ‘‘tails
purge’’ in a distillation. In either case the NaOH (or resulting
salt) cannot be disposed via incineration and may have to be
disposed via the undesirable method of deep-welling. Use of
the robust cobalt catalyst may be the key to utilizing the
economical process option B and we have shown through
continuous reaction studies that the cobalt catalyst is effective
for over 1 month of continuous operation (>1750 kg product/
kg catalyst) [13].
3.2. ESCA study of deactivated catalysts
The contrast in behavior between the Ni and Co catalysts
remains of great fundamental interest, beyond the practical
considerations. Deactivation of Raney-type Ni catalysts was
observed in the hydrogenation of both crude ammonolysis
product and the fractionated product, from which high and low
boiling components were removed by distillation. That fact
suggests that caprolactam and oligomeric amide impurities are
195
not required to effect the deactivation but that the process may
be inherent to the hydrogenation reaction. Indeed, our earlier
work on dinitrile hydrogenation with Raney Ni demonstrated
rapid deactivation for a series of dinitriles molecules, whereas
alkyl mononitriles did not deactivate at an appreciable rate [6].
Cobalt catalysts were not evaluated in that study.
To further understand the difference in catalyst deactivation
observed for Raney Ni and Raney Co, the surfaces of the used
catalysts were analyzed by electron spectroscopy for chemical
analysis (ESCA). ESCA is an invaluable tool for characteriza-
tion of the surface of catalysts and for the observation of
metallic states and deposition layers. For Raney catalysts a
complication exists in the analysis as a result of the pyrophoric
nature of the catalysts. To facilitate sample handling and since
our studies were confined to evaluation of materials deposited
during reaction, samples were washed and then passivated
before analysis. Samples of unused catalysts were similarly
washed and passivated for use as control samples. An additional
complication in the analyses is the presence of interfering
Auger lines in key parts of the spectra. Using the Al
monochromatic source allowed collection of good spectra in
the Ni 2p and Co 2p region of the spectra with no interference;
however data for Fe 2p were less reliable and not considered in
this study.
Evaluation of the surfaces of Raney1 Co 2724 catalysts
revealed only a very small N 1s signal in the ESCA spectra at
399 eV binding energy (Fig. 6) and the level of nitrogen did not
differ significantly between the control and the catalyst
recovered from the hydrogenation of the crude nylon-6
ammonolysis product prepared by process option B
(Table 3). The source of nitrogen in these samples is not
known at this time but we note that the spectra of Fig. 6 are not
consistent with reported values for cobalt nitride (396.8 eV)
[14]. For Raney1 Ni 2400, the spectra for control versus used
catalyst differed greatly (Fig. 6 and Table 3). A large N 1s signal
(399.5 eV) in the spectrum of the catalyst recovered from the
hydrogenation of crude nylon-6 ammonolysis product was
consistent with deposition of an oligomeric overlayer on the
catalyst surface. The N 1s signal is inconsistent with the
Fig. 6. The N 1s region of ESCA spectra of Raney-type nitrile hydrogenation
catalysts. (A) Raney1 Ni 2400 recovered from the hydrogenation of crude
nylon-6 ammonolysis product, (B) Raney1 Ni 2400 control, (C) Raney1 Co
2724 recovered from the hydrogenation of crude nylon-6 ammonolysis product,
and (D) Raney1 Co 2724 control—unused catalyst.
196 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198

Table 3
Percent atomic concentration determined by ESCA
Sample C 1s N 1s O 1s Al 2s a Cr 2p Ni 2p Co 2p N/Ni N/Co N/C
Raney Ni control 16.9 0.4 35.5 21.7 1.1 17.4 0.02 0.02
Raney Ni from reaction 28.5 1.6 39.7 15.2 0.4 10.6 0.15 0.06
Raney Ni from reaction with NaOH additive 23.0 nd 39.4 20.4 1.8 10.4 nd nd
Raney Ni with a monolayer of CPL 30.5 3.1 43.4 4.4 0.9 14.5 0.21 0.10
Raney Ni from ADN hydrogenation 39.3 5.2 35.0 5.5 1.2 13.0 0.40 0.13
Raney Co control 22.4 0.4 50.7 3.2 0.2 21.9 0.02 0.02
Raney Co from reaction 22.2 0.3 47.1 4.4 0.4 25.0 0.01 0.01
Fe spectra not analyzed due to Ni/Co LMM Auger line interferences.
a
Ni/Co 3s line interference limits accuracy.

reported value for nickel nitride (398.1 eV) [14] but similar to
nitrogen of organic molecules [15]. The C 1s region of the
spectra are shown in Fig. 7. We are cautious to make strong
conclusions based on C spectra alone as C is a ubiquitious
surface contaminant for any catalyst handled in open laboratory
conditions. Table 3 shows a significant amount of carbon in the
control samples due to handling and the N/C ratio of the
deactivated catalyst indicates that there is more carbon present
than would be attributable to a simple oligomeric amine
overlayer, further reinforcing the challenge of analyzing real
world catalyst samples. Qualitatively, however, the C 1s signals
were consistent with a hydrocarbon oxidation state. Notably,
there was no detectable signal 281–283 eV, the binding energy
region for C 1s of metal carbides [15], suggesting the
deactivation is not a result of Ni3C formation as reported
for gas phase hydrogenation of acetonitrile over Ni/SiO2 [5].
Rather, the signals are consistent with an aliphatic amine
layer, possibly incorporating carboxylate groups from the
passivation. This overlayer, which may prevent further
reaction, was reminiscent of previously observed behavior
of Raney Ni in ADN hydrogenations [6]. In fact, spectra of
Raney1 Ni 2400 catalyst recovered from the hydrogenation
of crude nylon-6 ammonolysis product and catalyst recovered
from hydrogenation of ADN, each in the absence of sodium
hydroxide showed very large N 1s signals, with the ADN
sample showing the greatest intensity (Fig. 8). The C 1s
spectra of these two samples are less informative owing to C
contamination on all samples due to handling and/or
preparation (Fig. 9).
It was shown above that addition of sodium hydroxide
inhibited deactivation of Raney Ni in the present study (Fig. 5).
Correspondingly, the ESCA spectrum of Raney Ni catalyst
recovered from the hydrogenation of crude nylon-6 ammono-
lysis product in the presence of sodium hydroxide showed a
negligible N 1s peak (Fig. 8), suggesting minimal deposition of
an amine overlayer. Since it has been shown that secondary
amine formation occurs on the catalyst surface [16] and sodium
hydroxide competes for primary amine adsorption sites [11],
one may speculate that amine coupling reactions require an
ensemble of surface sites and that strong adsorption of sodium
hydroxide breaks up this ensemble. With this speculation in
mind we have identified other organic based additives, which

Fig. 8. N 1s ESCA spectra of Raney1 Ni 2400 catalysts. (A) Recovered from

Fig. 7. The C 1s region of ESCA spectra of Raney-type nitrile hydrogenation
catalysts. (A) Raney1 Ni 2400 recovered from the hydrogenation of crude
nylon-6 ammonolysis product, (B) Raney1 Ni 2400 control, (C) Raney1 Co
2724 recovered from the hydrogenation of crude nylon-6 ammonolysis product,
and (D) Raney1 Co 2724 control—unused catalyst.
hydrogenation of nylon-6 crude ammonolysis product in the absence of sodium
hydroxide, (B) recovered from the hydrogenation of ADN in the absence of
sodium hydroxide, (C) comparison sample prepared by the physisorption of
caprolactam onto Raney Ni surface, (D) recovered from hydrogenation of
nylon-6 crude ammonolysis product in the presence of sodium hydroxide, and
(E) control—unused catalyst.
 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198
Fig. 9. C 1s ESCA spectra of Raney1 Ni 2400 catalysts. (A) Recovered from
hydrogenation of nylon-6 crude ammonolysis product in the absence of sodium
hydroxide, (B) recovered from the hydrogenation of ADN in the absence of
sodium hydroxide, (C) comparison sample prepared by the physisorption of
caprolactam onto Raney Ni surface, (D) recovered from hydrogenation of
nylon-6 crude ammonolysis product in the presence of sodium hydroxide, and
(E) control—unused catalyst.
alter the reactivity of the catalyst surface and reduce secondary
amine formation [17]. As organic additives these species, such
as tetramethylammonium hydroxide or tetrabutylammonium
cyanide, may be more environmentally benign.
All of our observations are consistent with the concept that
amine coupling reactions occur on the surface of Raney Ni and
for disubstituted molecules, they eventually lead to insoluble
oligomeric species which deposit on the catalyst and block
active sites (Scheme 2). The formation of secondary amines is
consistent with current mechanistic understanding of nitrile
hydrogenation [16]. Additionally, the formation of an overlayer
on the surface of a catalyst has recently been implicated to play
an important role in the mechanism of nitrile hydrogenation
[18]. Isotopic labeling studies have shown that the overlayer
serves as a reservoir for activated hydrogen and these hydrogen
atoms may transfer to an activated nitrile or partially
hydrogenated intermediate during acetonitrile hydrogenation
[18]. It is, perhaps, ironic that this overlayer, which early in a
reaction may serve a symbiotic role to enable nitrile
hydrogenation, eventually serves a parasitic role of over-
powering the catalyst surface and preventing further reaction.
Scheme 2. Amine condensation occurs on the surface of Raney1 Ni 2400 in competition with direct hydrogenation.
197
It is unlikely that caprolactam or oligomeric amides
contribute to the deactivation. Amides are very stable at these
reaction temperatures (75–90 8C) and unlikely to form
polymeric species. A comparison sample for ESCA was
prepared by deposition of approximately one monolayer of
caprolactam by evaporation of a methanol solution of
caprolactam onto Raney1 Ni 2400. Examination of the ESCA
spectrum shows the anticipated peak in the N 1s region (Fig. 8).
Comparison of this spectrum to the above samples does not
allow one to unequivocally rule out the presence of amide
species in the Raney Ni catalyst deactivated during the
hydrogenation of the nylon recycle feed. Nonetheless,
circumstantial evidence suggests amides do not contribute to
the deactivation. Caprolactam, a non-participating component
of the feed, did not reduce in concentration during the course of
either the Raney Co or Raney Ni reactions in this study, as
measured by GC analysis. If oligomeric amides, which were
pre-existing in the crude ammonolysis product, deposited on
the surface, one would expect very rapid deactivation. On the
contrary, deactivation of Raney Ni in these reactions was not
immediate and, in fact, the catalyst deactivated more
aggressively during ADN hydrogenation as inferred from a
larger ESCA N 1s signal. The presence of amides such as
caprolactam in the feed may actually retard the deactivation
relative to pure ADN feed by serving as a simple inert diluent or
by exerting a greater solubility for the proposed oligomeric
amines. Lactams are frequently used as solvents for a range of
organic molecules.
To further support the proposed deactivation scheme, the
selectivities of the Raney Ni and Raney Co catalyzed reactions
were compared. For the reactions of Fig. 2, Raney Ni exhibited
a greater yield of amine coupling by-products than Raney Co
did, at comparable conversions. Thus at 43% conversion, the
Raney Ni reaction comprised 3.2% hexamethyleneimine (HMI)
and 0.8% bis(hexamethylene)triamine (BHMT), while at 46%
conversion the Raney Co reaction comprised 1.8% HMI and
0.2% BHMT. HMI is the intramolecular amine condensation
product and BHMT is the intermolecular condensation product
(Scheme 2, n = 1) and the first step in formation of higher
oligomers.
The observations presented here are additive to our earlier
study of Raney Ni deactivation during ADN hydrogenation and
are in contrast to a study of the deactivation of nickel catalysts
during acetonitrile hydrogenation, which concluded that amine
‘‘cracking’’ led to carbide formation on the surface of the nickel
catalyst [5]. That mechanism would not distinguish between
dinitriles and mononitriles. If a cracking mechanism was
operative in the current study, it was certainly disfavored
198 M.W. Duch, A.M. Allgeier / Applied Catalysis A: General 318 (2007) 190–198
kinetically, compared to the amine condensation mechanism
proposed here.
4. Conclusions
Nylon-6 and nylon-6,6 fibers and polymers may be recycled
via the combination of ammonolysis, accompanying dehydra-
tion, and nitrile hydrogenation to produce predominantly
hexamethylenediamine. While Raney1 Co 2724 showed little
or no sign of deactivation during the semi-batch hydrogenation
of the ammonolysis products, Raney1 Ni 2400 showed clear
signs of deactivation, which could be abated in the presence of
sodium hydroxide and other basic additives. The deactivation
of Raney1 Ni 2400 is ascribed to the formation of oligomeric
compounds arising from amine condensation reactions, which
can be suppressed with ammonia or sodium hydroxide. In
support of this conclusion, ESCA spectra of used Raney Ni
catalysts reveal an increase in nitrogen and carbon containing
species at the surface, relative to Raney Ni control samples and
used Raney Co samples.
Acknowledgements
DuPont and DuPont Textiles and Interiors (now Invista) are
acknowledged for support of this work. Xiaowei Hu, Angel
Vargas and Jeff Crawford are acknowledged for valuable
technical assistance.
References
[1] A.H. Tullo, Chem. Eng. News 78 (4) (2000) 23.
[2] http://www.laep.org/uclasp/issues/landfills/litter.html, cited 5 June 2003.
[3] E.F. Moran Jr., U.S. Patent 5,310,905 (1994).
[4] (a) R.J. McKinney, U.S. Patent 5,302,756 (1994);
(b) A. Lambert, G.H. Lang, GB Patent 1,172,997 (1969)
(c) G. Kalfas, Polym. React. Eng. 6 (1) (1998) 41.
[5] M.J.F.M. Verhaak, A.J. van Dillen, J.W. Geus, J. Catal. 143 (1993) 187.
[6] A.M. Allgeier, M.W. Duch, Chem. Ind. (Marcel Dekker) 82 (2001) 229
(Catal. Org. Reactions).
[7] Stirring rates were chosen based on prior lab experience for avoiding
diffusion limitations in neat adiponitrile (ADN) hydrogenation reactions.
While an evaluation of the influence of stirring rate on kinetics was not
conducted in this study, rates of gas uptake were slower here than in the
ADN hydrogenations.
[8] Raney1 is a registered trademark of W.R. Grace.
[9] Analysis by W.R. Grace. Samples were dried in pure flowing hydrogen at
130 8C for 60–70 min prior to adsorption. CO adsorption was measured
via a pulsed adsorption procedure at 0 8C with 3 cc pulses of CO in He
carrier gas (20 sccm). Further details may be found in S.R. Schmidt,
Chem. Ind. (Marcel Dekker) 62 (1995) 45 (Catal. Org. Reactions).
[10] (a) S.K. Sengupta, T.A. Koch, K.R. Krause, U.S. Patent 5,900,511 (1999);
(b) S.B. Ziemecki, U.S. Patent 5,151,543 (1992)
(c) D.J. Ostgard, M. Berweiler, S. Roder, P. Panster, Chem. Ind. (Marcel
Dekker) 89 (2003) 273 (Catal. Org. Reactions).
[11] (a) C. Mathieu, E. Dietrich, H. Delmas, J. Jenck, Chem. Eng. Sci. 47
(1992) 2289;
(b) C. Joly-Vuillemin, D. Gavroy, G. Cordier, C. DeBellefon, H. Delmas,
Chem. Eng. Sci. 49 (1994) 4839;
(c) D. Gavroy, C. Joly-Vuillemin, G. Cordier, P. Fouilloux, H. Delmas,
Catal. Today 24 (1995) 103.
[12] S.N. Thomas-Pryor, T.A. Manz, Z. Liu, T.A. Koch, S.K. Sengupta, W.N.
Delgass, Chem. Ind. (Dekker) 75 (1998) 195 (Catal. Org. Reactions).
[13] A.M. Allgeier, T.A. Koch, S.K. Sengupta, Chem. Ind. (Marcel Dekker)
104 (2005) 37 (Catal. Org. Reactions).
[14] G. Soto, W. de la Cruz, M.H. Farias, J. Electron Spectrosc. Relat. Phenom.
135 (2004) 27.
[15] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray
Photoelectron Spectroscopy, Physical Electronics Inc., Eden Prairie, MN,
USA, 1995.
[16] Y. Huang, W.M.H. Sachtler, Appl. Catal. A 182 (1999) 365.
[17] A.M. Allgeier, T.A. Koch, S.K. Sengupta, U.S. Patent 6,376,714 (2002).
[18] Y. Huang, W.M.H. Sachtler, Appl. Catal. A 191 (2000) 35.