Technical Repor t

Automation of Chemical Reaction Kinetics and Product Distribution Studies in

Pharmaceutical Development
Adam M. Fermier,* Alan R. Oyler, Barbara L. Armstrong, Bruce A. Weber,
Ramón L. Rodríguez, James V. Weber, James A. Nalasco
Drug Evaluation, Johnson and Johnson Pharmaceutical Research and Development
Key Words: Laboratory automation; Chemical kinetics; Accelerated degradation studies; Chemical kinetics instrumentation;
Robotics workstation(s)

ABSTRACT BACKGROUND

A
n automated instrument was designed and constructed
to facilitate the performance of pharmaceutical degra-
dation studies. A brief theoretical background on degra-
dation kinetics is given to rationalize the design of the instrument
and representative data are provided to illustrate its successful
application. This system was found to be capable of conducting
multiple simultaneous isothermal and nonisothermal kinetic stud-
ies with user-defined temperature profiles, sampling periods, and
data logging.
INTRODUCTION
The development of a pharmaceutical drug candidate is greatly
assisted by the acquisition of information on the rates and path-
ways of degradation of the active pharmaceutical ingredient and
formulation(s). Rate and product distribution data can be used,
for example, to choose appropriate salt forms or polymorphs, to
optimize formulations for maximum stability, to predict shelf lives
of drug products and outcomes of real-time stability studies, and
to determine the major degradation pathways. Typically, such
studies are conducted under multiple sets of conditions (e.g., tem-
perature, pH, etc.). Often the Arrhenius equation (Equation 1) is
used to correlate reaction rates (k) with temperature:
(1)
where A is the frequency of molecular collisions, E is the activa-
tion energy, R is the gas constant, and T is the absolute tempera-
ture. This equation can be applied to both isothermal and non-
isothermal studies. These kinetics and product distribution studies
involve the sampling of the reaction mixture over time and subse-
quent analysis of these samples. Manual sampling can be tedious
and simply missing a sampling point makes the data less valuable.
Presented here is a description of the design and construction of
an instrument that automates these kinetics and product distribu-
tion studies. The automated instrument is capable of managing
temperature and sampling of up to three simultaneous isothermal
and/or nonisothermal reactions (i.e., 40 vials/reaction).
Traditionally, degradation reactions of pharmaceutical com-
pounds have been studied under isothermal conditions and rate
constants have been measured at multiple temperatures. As an
example, the data for a first-order reaction can be modeled with a
differential rate equation in which the decrease in the concentration
of parent drug (-dC/dt) is directly proportional to the concentration
of the drug at time t (equation 2) where k is the rate constant and C
is the concentration of the parent drug at time t. Prediction of reac-
tion rates at temperatures that are not studied can then be derived
from the Arrhenius equation (equation 1). Typically, experimental
data are acquired at three or more temperatures.
(2)
Alternatively, equivalent rate data (i.e., the ability to calculate
rate constants at any temperature) can be obtained in one experi-
ment using the nonisothermal approach in which a temperature
program is used.1, 2 The resulting data are fitted with a rate equa-
tion that contains terms for temperature and the concentration of
the drug. Thus, for a first-order reaction, equations 1 and 2 can be
combined to give equation 3 where the temperature, T(t), is a
function of time. The utilization of various temperature functions
have been reported and are summarized in Table I. Once the acti-
vation energy (E) and the frequency factor (A) have been deter-
mined by fitting the data to a model such as equation 3, rate con-
stants are then calculated for various temperatures by the use of
*Correspondence
OMP B236
1000 Route 202
Raritan, NJ 08869
Tel: 908-704-4329
Fax: -908-218-0621
Email: afermier@prdus.jnj.com

83
 Technical Repor t Continued
equation 1. Comparison studies of this nonisothermal approach
with isothermal approaches have given nearly identical results.1-7
Thus, the use of nonisothermal techniques for accelerated drug
degradation studies seems appropriate as a means to acquire
more data concerning the behavior of the drug candidate in a
limited time.
(3)
RATIONALE FOR INSTRUMENT DESIGN
In general, conducting isothermal or nonisothermal degradation
reaction studies involves:
1. obtaining sufficient data points (usually 30-40) to obtain
a good fit for the kinetics models,
2. conducting multiple reactions in parallel to complete the
study within a reasonable time period,
3. programming the desired temperature profiles,
4. programming sampling intervals,
5. maintaining solutions of the drug formulation or samples
of the drug substance at programmed temperatures,
6. removing samples of the reaction product mixtures at
programmed intervals,
7. storing the samples at a low enough temperature to
quench the reaction,
8. recording of actual reaction temperatures,
9. recording of actual sampling times,
10. analyzing the samples by techniques such as high perfor-
mance liquid chromatography (HPLC).
Analyses of samples using HPLC (operation 10) can be per-
formed in an automated fashion using commercially available
instrumentation. Therefore, the automated instrument was
designed and constructed to perform operations 1-9 exclusively.
Table 1. Nonisothermal models used for degradation kinetics data.
84
Although HPLC analysis (operation 10) was not incorporated into
the automated instrument, provisions were made for the optional
use of HPLC autosampler vials as reaction vessels. Thus, the
instrument consisted of three “hot blocks” for heating up to 40
reaction vessels in each block (2-mL vials), three “cold blocks” for
storage of these reaction vessels prior to analysis, an “autosampler”
for moving sample vessels from the hot block to the cold block, a
temperature controller to regulate the temperatures of the hot and
cold blocks, and a computer program for a user interface and data
logging. If HPLC autosampler vials are used, then the user has the
option of simply transferring the autosampler vials from the cold
block to an HPLC system for analysis.
An alternative design might have involved a single reaction
vessel rather than multiple independent reaction vessels. The
independent reaction vessel approach was taken for several rea-
sons. First, a single reaction vessel would not be feasible for solid
state reactions since sample aliquots could not readily be set
aside. In addition, we often prefer to pre-weigh solid-state reac-
tion samples that are to be conducted in open vessels in case
there is a loss or gain of water during the reaction period.
Second, with liquid reactions, contamination might occur during
sampling from a single reaction vessel.3, 8 Third, material for each
time point is often required for multiple analyses (HPLC, HPLC-
MS, etc.) and therefore several samples would be required in any
case. And fourth, using smaller, multiple reaction vessels allows for
“sample limited” situations.
For maximum flexibility, each reaction block could be inde-
pendently programmed with either a time/temperature equation
such as any of the approaches given in Table 1 or with a user-gen-
erated time versus temperature table. Thus, the instrument can
support simultaneous isothermal studies and nonisothermal stud-
ies with any conceivable temperature program. Although the
instrument shown here has only three blocks, a larger system

Figure 1. Hot and cold blocks shared a common mechanical design
A. with a phenolic cover,
B. aluminum block,
C. internal casing,
D. and the external housing,
E. to support the blocks.
The heating or cooling source was mounted under the internal casing
(not shown).
could be designed to accommodate additional reactions.
Continuous data logging provided a convenient audit trail to
ensure that the actual temperatures of a defined study followed
user prescribed temperatures.9
DETAILED DESCRIPTION ON INSTRUMENT
Hot and Cold Block Design
The removable hot and cold blocks were made of aluminum
and contained holes to accommodate 40 reaction vessels (Figure
1). The blocks could accommodate standard HPLC autosampler
vials (2-mL; Chromacol - Trumbull, Conneticut, USA, PN: 2-CV)
with silicon Teflon crimp caps (Chromacol - PN: 11-AC-ST15)
or glass sealed ampules (2-mL pre-scored ampules; Wheaton
Science Products - Millville, NJ, USA, PN: 176776). The crimp
cap vials could be used with temperatures up to 100 °C while the
glass sealed vessels could be used with temperatures up to 200 °C.
A silicone rubber heater (Watlow Electric Manufacturing
Company - St. Louis, Missouri, USA, PN: F030050C7-A001B)
with an etched foil element provided uniform heating of the reac-
tion vials in the hot block. A thermal electric cooler (Marlow
Industries - Dallas, TX, USA, PN: ST3353-02) cooled the reac-
tion vials in the cold plate to 5 °C ± 1 °C. The hot and cold
Figure 2. Robotic workcell for autosampling of vials over the course
of the experiment.
blocks were insulated with one-inch thick melamine white foam
(McMaster Carr - New Brunswick, NJ, USA, PN: 86145K54). A
high precision resistive temperature detector (RTD) (Watlow
Electric Manufacturing Company, PN: S80-1000204) monitored
the temperatures of the blocks. The temperatures of the blocks
were controlled by a dual loop temperature controller (Watlow
Electric Manufacturing Company, PN: 999D-22CC-AURG).
The controller provided a voltage signal (24 VDC) to regulate the
power to the heating (120 VAC) and cooling units (12 VDC) via
solid-state relays (Grayhill - La Grange, IL, USA, PN: 70S2-04-
B-06-N; 70S2-01-A-05-N respectively). Power for the thermo-
electric cooler (TEC) was provided by a switching regulated (12
VDC/4.1 amp) power supply (Acopian - Easton, PA, USA, PN:
12WB410). The temperature controller held the blocks to within
0.1 °C of the user-defined temperature. The standard deviation of
temperature between the reaction vials within the hot and cold
blocks was 0.3 oC and this value was comparable to systems
reported in the literature. 10, 11
Autosampler
A three axis robotic workcell (Arrick Robotics - Hurst, TX,
USA, PN: RW-18b-3-Axis) served as the autosampler (Figure 2).
Stepper motors and pulley reducers on each axis provided preci-
sion movements accurate to within 0.002’. Power supplies for the
stepper motors were provided with the unit. A pneumatic gripper
(SMC Corporation of America - Indianapolis, Indiana, USA, PN:
MHQ2-16D) was mounted on the z-axis (Figure 3). Custom
designed fingers allowed the sample vials to be moved from the
hot to the cold blocks. The gripper was operated by toggling a 20
p.s.i. pressure between one of two lines. One line opened and the
second closed the gripper. The two lines were regulated by a sole-
noid valve that, upon activation, pressurized the “closed” line and
vented the “open” line.
85
 Technical Repor t Continued

Figure 3. Gripper Assembly used to pick and place vials from the hot
to the cold blocks. Figure 4. User interface for programming temperature profile for
degradation experiment. Squares identify time intervals when vials will
be moved from hot to cold blocks.
System Integration
A standard desktop computer (IBM 233 MHz, 32 MB RAM)
controlled the instrument. A custom program written in
LabVIEW 5.1® (National Instruments - Austin, Texas, USA)
allowed the user to specify the desired temperature and exposure
time for each vial in the hot block (Figure 4). All temperature
equations listed in Table 1 were capable of being programmed
through this interface. Alternatively, a tab delimited file with time
and temperature columns could be imported. After identifying the
experimental conditions on this user interface, the user loaded the
vials into the preheated hot block and the experiment was subse-
quently started.
The master control panel for the experiment is shown in Figure
5. A master clock tracked the experimental time and the secondary
clock tracked the time until the next vial was to be moved from
the hot to the cold block. A pictorial simulation of the 40-position
hot and cold blocks was presented to the user. The figure demon-
strates an experiment that was 4.5 hours into a 20-hour study. The
status of the vials was shown by respective color changes in the hot Figure 5. Master control panel tracking temperatures and position of
and cold block positions. A thermometer provided the current vials. Master clock is the relative time for the experiment. Simulated
temperature for each block. The graph below the thermometer views of the hot and cold blocks provide a pictorial view of the current
tracked the experimental temperature of the hot block over time. location of all vials. Thermometers reflect the actual temperature of
During the course of an experiment, the temperature controller each plate. The graph tracks the actual temperature of the hot block as
was updated every 0.1 °C change in the user defined temperature a function of time.
profile. A tab delimited text file logged actual temperature changes
greater than 0.1 °C for either the hot or cold blocks. The file also
captured the sampling time of each vial. The text file provided a characterization studies. To provide one example, the system was
convenient audit trail for the user. used to elucidate the degradation chemistry of a new drug candi-
date whose structure suggested a susceptibility to aqueous reactiv-
APPLICATION ity. In solution, the drug molecule (A) could potentially undergo
The system described above has been applied to the study of acyl migration to give a re-arrangement product (B) or hydrolysis
several compounds under development as new pharmaceutical of the acyl moiety to give the hydrolyzed form of the drug (C). In
agents. The system has shown itself to be of great utility in increas- addition, the re-arranged form (B) can also convert to the hydroly-
ing the efficiency and accuracy of performing these necessary drug sis product (C).

86
Figure 6. Degradation profile of a low molecular weight compound
performed on the robotic instrument described here.
The reaction scheme and the data (i.e., drug and drug degra-
dation products molar concentrations versus time plot), from the
above example are shown in Figure 6. The temperature of the
reaction block was 40 °C and the temperature of the cold block
was 5 °C. HPLC analyses were conducted within 24 hours of
sampling and selected traces are shown in Figure 7. Samples were
assayed daily during the first week of the experiment when the
loss of drug was most rapid. This allowed better fitting of the
experimental data over this more curved region of the plot. To
maximize efficiency of HPLC assay time and resources, samples
were assayed only weekly during the more linear later region of
the degradation profile.
The experimental results clearly demonstrated the degradation
profile expected for a molecule of this type.12 The drug substance
(A) most rapidly equilibrated with the re-arrangement product
(B), with a slower conversion of both A and B to the terminal
product C via hydrolysis. Rate constants were obtained by least
squares fitting of the data to the differential rate equations for the
reaction scheme (Figure 6; Scientist Version 2.0, MicroMath
Scientific Software, Salt Lake City, UT). These rate constants
were: k1 = 2.91 ± 0.07 ´ 10-3 h-1, k2 = 5.2 ± 0.2 ´ 10-3 h-1, k3
= 3.5 ± 0.4 ´ 10-4 h-1, and k4 = 8.5 ± 1 ´ 10-4 h-1. The result-
ing fitted curves are shown in Figure 6.
SUMMA RY
The automated instrument described here is capable of per-
forming both isothermal and nonisothermal experiments. We have
demonstrated the utility of the instrument to carry out routine
isothermal kinetic determinations. Currently, the instrument is
capable of performing three simultaneous and independent reac-
tions with 40 vials per reaction (Figure 3). A second instrument
capable of conducting 16 independent reactions is being designed.
Kinetics studies are extremely labor intensive and these instru-
ments remove all of the redundant tasks as outlined above.
Therefore, such instruments are useful in rapidly screening many
potential molecules for their suitability as drug development can-
Figure 7. Selected chromatograms of samples that were analyzed on an
Agilent 1100 HPLC system.
didates based on their potential stability. The performance of such
studies early in development can also play a key role in formula-
tion design and product development. The system outlined here
has been able to save time and labor for the chemists performing
these reactions by enabling unattended sampling and accurate
temperature profiles.
REFERENCES
1. Rogers, A.R. An Accelerated Storage Test with Programmed
Temperature Rise. J. Pharm. and Pharmacol., 1963, 15: p.
101t-105t.
2. Davis, R.E. Temperature as a variable during kinetic
experiments. J. Phys. Chem., 1959, 63: p. 307-309.
3. Kipp, J.E. ; Hlavaty, J.J. Nonisothermal stability assessment of
stable pharmaceuticals: testing of a clindamycin phosphate
formulation. Pharm. Res., 1991, 8(5): p. 570-575.
4. Junnarkar, G.H. ; Stavchansky, S. Isothermal and nonisothermal
decomposition of famotidine in aqueous solution. Pharm.
Res., 1995, 12(4): p. 599-604.
5. Lee, M.-L. ; Stavchansky, S. Isothermal and nonisothermal
decomposition of thymopentin and its analogs in aqueous
solution. Pharm. Res., 1998, 15(11): p. 1702-1707.
6. Zhan, X.; Yin, G.; Wang, L.; Ma, B. Exponential Heating in
Drug Stability Experiment and Statistical Evaluation of
Nonisothermal and Isothermal Prediction. J. Pharm. Sci.,
1997, 86(6): p. 709-715.
7. Zhan, X.; Yin, G.; Ma, B. Determination of Rate Order for
Degradation of Drugs with Nonisothermal Stability
Experiment. J. Pharm. Sci., 1997, 86(10): p. 1099-1104.
8. Kipp, J.E.; Jensen, M.M.; Kronholm, K.; McHalsky, M.
Automated liquid chromatography for non-isothermal kinetic
studies. Int. J. Pharm., 1986, 34(1-2): p. 1-8.
9. Fermier, A.M.; Armstrong, B.L.; Oyler, A.R.; Weber, J.V.;
Nalasco, J., Automated Degradation Instrument. US Patent
Application Serial No. 09/816,787; 2001.
87
 Technical Repor t Continued
10. Zhan, X.; Yin, G.; Ma, B. New heating controller and
computation for linear heating stability experiment. Int. J.
Pharm., 1995, 115(2): p. 161-166.
11. Zhan, X.; Jiang, J.; Liu, S.; Yin, G. Computer-controlled
heating system and new computation for reciprocal heating
stability experiment. Int. J. Pharm., 1995, 115(2): p. 167-
173.
12. Hara, M.; Hayashi, H.; Yoshida, T.; Murayama, H. Studies
on the Kinetics and Mechanism of Drug Degradation. I.
Kinetics and Mechanism of Degradation of Chlorphenesin
Carbamate in Stongly Alkaline Aqueous Solutions. Chem.
Pharm. Bull., 1986, 34(4): p. 1764-1769.
13. Tucker, I.G. An assessment of a logarithmic nonisothermal
storage test. Drug Dev. Ind. Pharm., 1981, 7(2): p. 231-246.
14. Rosenberg, L.S.; Pelland, D.W.; Black, G.D.; Aunet, D.L.;
Hostetler, C.K.; Wagenknecht, D.M. Nonisothermal methods
for stability prediction. J. Parenter. Sci. Technol., 1986, 40(4):
p. 164-168.
15. Cole, B.R.; Leadbeater, L. A critical assessment of an
accelerated storage test. J. Pharm. Pharmacol., 1966, 18: p.
101-111.
16. Anderson, R.A.; Campbell, M. Simple nonisothermal
method for predicting stability of pharmaceuticals. Australas.
J. Pharm., 1971, 52(623): p. S81-S83.
17. Eriksen, S.P. ; Stelmach, H. Single-step stability studies. J.
Pharm. Sci., 1965, 54(7): p. 1029-1034.
18. Zoglio, M.A.; Windheuser, J.J.; Vatti, R.; Maulding, H.V., Jr.;
Kornblum, S.S.; Jacobs, A.L.; Hamot, H. Linear nonisothermal
stability studies. J. Pharm. Sci., 1968, 57(12): p. 2080-
2085.
19. Kay, A.I.; Simon, T.H. Use of an analog computer to simulate
and interpret data obtained from linear nonisothermal
stability studies. J. Pharm. Sci., 1971, 60(2): p. 205-208.
If you are interested in seeing more articles like this in JALA, please email jalanews@virginia.edu, and simply put the code (719) in the subject line.
88
20. Okusa, N. Prediction of the stability of drugs. IV. Prediction
of the stability by a multilevel nonisothermal method. Chem.
Pharm. Bull., 1975, 23(4): p. 803-809.
21. Tucker, I.G.; Owen, W.R. Estimation of all parameters from
nonisothermal kinetic data. J. Pharm. Sci., 1982, 71(9): p.
969-974.
22. Waltersson, J.O.; Lundgren, P. Nonisothermal kinetics
applied to pharmaceuticals. Acta Pharm. Suec., 1982, 19(2): p.
127-136.
23. Ahlneck, C.; Lundgren, P. Methods for the evaluation of solid
state stability and compatibility between drug and excipient.
Acta Pharm. Suec., 1986, 22(6): p. 305-314.
24. Edel, B.; Baltzer, M.O. Nonisothermal kinetics with pro-
grammed temperature steps. J. Pharm. Sci., 1980, 69(3):
p. 287-290.
25. Zoglio, M.A.; Maulding, H.V.; Streng, W.H.; Vincek, W.C.
Nonisothermal kinetic studies. III. Rapid nonisothermal-
isothermal method for stability prediction. J. Pharm. Sci.,
1975, 64(8): p. 1381-1383.
26. Hodgson, S.C.; Ngeh, L.N.; Orbell, J.D.; Bigger, S.W. A
student experiment in non-isothermal chemical kinetics. J.
Chem. Ed., 1998, 75(9): p. 1150-1153.
27. Maulding, H.V., Jr. ; Zoglio, M.A. Flexible nonisothermal
stability studies. J. Pharm. Sci., 1970, 59(3): p. 333-337.
28. Yoshioka, S.; Aso, Y.; Takeda, Y. Isothermal and nonisothermal
kinetics in the stability prediction of vitamin A preparations.
Pharm. Res., 1990, 7(4): p. 388-391.
29. Yoshioka, S.; Aso, Y.; Uchiyama, M. Statistical evaluation of
nonisothermal prediction of drug stability. II. Experimental
design for practical drug products. Int. J. Pharm., 1988, 46(1-
2): p. 121-132.