Professional Reference Shelf

R3.3 Molecular Dynamics

Abbreviated Lecture Notes

Full Lecture Notes

1. Overview
2. Introduction
1. No Reaction
2. Reaction
3. How the Trajectories are Calculated
1. Equations of Motion
2. Estimates of the Potential Energy Surface(R)
3. Method of Solution to Map Out Trajectories
4. The Monte Carlo Simulation
5. Calculating the Reaction Cross Section
6. Rate of Reaction
7. Problems
8. Closure

I. Overview

II.       Introduction

A. No Reaction

The objective of this Reference Shelf is to use molecule dynamics to provide insight as to how reactions occur. Here we will calculate reaction probabilities, reaction cross sections, and reaction rates. We will observe the effects of vibration, rotation and kinetic energies (velocity) on the colliding molecules. We will find that there is a minimum kinetic energy necessary to react, that the reaction cross section increases with increasing kinetic energy, and that there is a maximum value of the impact parameter, related to the offset of the molecular trajectories, above which no reaction will occur.

            We are going to study the molecular dynamics of the reaction of the hydrogen exchange reaction

written symbolically

                                                      A + BC AB + C

where molecule i has a velocity V_i and is in rotational state J and vibrational state . [E.g. BC (V_BC, J, )].

                           A (V_A) + BC (V_BC, J, ) AB (V_AB, J, ) + C (V_C)

We are going to consider the hydrogen exchange reaction discussed in Karplus article. †

We begin with the A and BC molecules far apart and then calculate the trajectory of the A molecule as it approaches the BC molecule. The A molecule will either replace the C molecule to form AB or it will be deflected and not react.

Let R be the distances of separation for the appropriate species. The distances of separation are shown as a function of time in Figure PRS.3C-1.

Figure PRS.3C-1 Trajectories when no reaction has taken place.

From Karplus, et.al, J.Chem. Phys. 43, p3259 (1965)

B.  Reaction:

1 the A molecule replaces the C molecule and the R_BC distance begins to increase and that the R_AB distance undergoes oscillation and we see that a reaction has taken place. The time it takes the reaction to occur at t₁ is about 10^­14s.

Figure PRS.3C-2  Trajectories when a reaction occurs.
From Karplus, et.al, J. Chem. Phys. 43, p3259 (1965)

III.     How the Trajectories are Calculated

A.  Equations of Motion

We calculate the momentum P of the molecule at a given time and position and then calculate a new position of the molecule using only the definition of force, F, and the potential energy surface (x).

                                                                                  (C-1)

                                                                                                              (C-2)

                                                                                                          (C-3)

                                                                               (C-4)

                                          and

                                                                                                                      (C-5)

Equations (C-4) and (C-5) can be solved simultaneously to obtain x a function of time. Also recall the translational kinetic energy is

                                                                                                  (C-6)

To solve this set of equations, [i.e., (C-4) and (C-5)] we need the potential energy as a function of distance, i.e., . Similarly, for the reaction (A + BC  AB + C) we need the potential energy surface

B.  Estimates of the Potential Energy Surface(R)

Three methods are commonly used to estimate the potential energy surface.

1)   Lennard-Jones 6-12 potential

                                                                                    (C-7)

2)   Morse potential

      where e_LJ is the Lennard-Jones parameter, D is the dissociation energy, r_o equilibrium internuclear distance, is the Morse potential constant.

               For a 3-body system ABC.†

                              _(C-8)

      where D_ij is the dissociation energy for molecules i and j.

3)   The Potential Energy Surface can also be calculated by

      a)   Ab Initio [Cerius²] methods

      b)   Semi-empirical methods such as the London-Eyring-Polanyi-Sato Surface (LEPS surface)

A schematic of the potential energy surface is shown in Figure PRS.3C-3.

Figure PRS.3C-3  Potential energy surface.

C.     Method of Solution to Map Out Trajectories

C.1 Momentum as a function of time and position.

            Consider the motion of molecule A. The x component of momentum for species A is related to the potential energy by

                                                                                                  (C-9)

Integrating we can find the x component of momentum as a function of time and position

                                                                                (C-10)

Similar expressions exist for y and z, e.g.

                                                                            (C-11)

To illustrate the concept we use the Euler method of integration

                                                                                (C-12)

C.2 Location of Molecules (e.g. A) as a function of time.

            The position of A, i.e., x, is related to the momentum P_x by the equation

                                                                                                (C-13)

Integrating we find the location of A as a function of time

                                                                    (C-14)

Because V(x) depends upon x, in practice we need to use a more sophisticated method that the Euler method (C-14) to solve these equations for P_x and x simultaneously to obtain the trajectory of molecule A as a function of time as shown in Table PRS.C-1.

Table PRS.3C-1  3-D Solution Technique to Calculate Trajectory of A

Of course the time interval for each interaction must be small as we carry out each integration.

C.3 Calculating the Trajectories of all the Molecules Using the Hamiltonian

            Because the Hamiltonian is used in classical mechanics to describe the motion of particles, let’s see how it gives the same equation as those given in Table PRS.3C-1. The Hamiltonian is the sum of the kinetic and potential energies

                                                                  (C-15)

Differentiating the Hamiltonian w.r.t. momentum, P_x, we find

                                                                          (C-16)

Similarly

We see that by using the Hamiltonian and coupling these same six equations we can trace out a trajectory for molecule A shown in Table PRS.3C-1.

Change in Coordinate System

            Now lets return to the hydrogen exchange reactor

                                                      A + BC  AB + C

We are going to re-design our coordinate system because we are only interested in the relative positions of the molecules to one another, not where they are in a 3-D space. This re-design is called mass weighted coordinate system or affine transformation.

            Let

                     Q₁, Q₂, Q₃ = Location of C with B as the origin

                     Q₄, Q₅, Q₆ = Location of A wrt center of mass of BC

                     Q₇, Q₈, Q₉ = Location of Center of mass of the 3 particle system

Figure PRS.3C-4  New Coordinate System.

For example, Q₁, Q₂, and Q₃ could represent x₁, y₁, and z₁ components of C with B as the origin or the radial components r, , and with r=0 as the origin. Similarly Q₄, Q₅, and Q₆ could represent the x, y, z coordinates of A with the center mass of BC as the origin or the radial coordinates r, , and . Knowing the location of C and A with regard to their respective origins, the distances between A and B, R_AB, B and C, R_BC and between A and C, R_AC can easily be found. See Figure PRS.3C-6).

IV.    The Monte Carlo Simulation

In classical and statistical mechanics the Hamiltonian is used to express the total energy. The Hamiltonian is the sum of the kinetic energy (KE), , and the potential energy (PE), .

                                    H = KE + PE = KE +

In our new coordinate systems the KE of A is written as

                                                    (C-17)

with

and

For our two-body 3 molecule system, A + BC, we must use the reduced masses

                                                (C-18)

 is the potential energy surface. We only need to specify two distances (e.g. R_AB, R_BC) because the other is then a fixed quantity.

            The distances R_AB and R_BC are shown in the potential energy surface  (R_AB, R_BC) in Figure PRS.3C-5 along with the trajectory of the reaction.

Figure PRS.3C-5  Reactant coordinates and the potential energy surface.

A summary of the above equations used to solve for the molecular trajectories is shown in Table PRS.3C-3.

Table PRS.3C-2  Equations to be solved to predict the trajectories

            For j = 1, 2, 3

(A)                                    

            For j = 4, 5, 6

(B)                                                     

(C)                                         

(D)      

            †

            These equations are analogous to the P_x and x,y,z equations shown on pages 3 and 4.

The above equations can be solved simultaneously using a software package. To predict the location Q_i, from which one can determine the molecular distances, E.g., R_BC, R_AB. However, before we begin to do this we need to specify the parameter values and the initial conditions.

            To map out the molecular trajectories to determine if a reaction has occurred we need three things.

            1.   The Governing Equations. These equations are given in Table PRS.3C-2.

            2.   The specified values of the variables. These values fall in two categories and are shown in Table PRS.3C-3.

                  (a)  Those variables to be studied to learn their effect on the reaction rate. These are the specified variables.

                  (b)  Those variables whose numerical values are specified by the Monte Carlo Simulation.

3.      The initial values to calculate the molecular trajectories. The initial parameter values are given in Table PRS.3C-4 and corresponds to the orientation of the A and BC molecules shown in Figure PRS.3C-6.

Table PRS.3C-3  Categories of Variable Parameter Values

We are going to choose our coordinate system such that molecule A and the center of mass of BC lie the y-z plane and that A approaches BC along the z axis.

Table PRS.3C-4 Initial Conditions to Start the Trajectory

Here  is Plank’s constant divided by 2p and R₊ is the turning point radius shown in Figure PRS.3C-7. The value of r_o is chosen as small as possible to save computing time but not so small as to experience any potential interactions between the A and BC molecules. A schematic of the initial conditions and the orientation is shown below in Figure PRS.3C-6.

Figure PRS.3C-6  Reactive trajectory. Courtesy of J. I. Steinfeld, J. S. Francisco, W. L. Hose,Chemical Kinetics and Dynamics, p.264, Prentice Hall, Upper Saddle River, New Jersey (1989) with v₀ = U = U_R in our notation)

BC, between the B and the C molecule.

Figure PRS.3C-7  Turning points of H­H vibration.

The distance R cannot take on any value, it can only between the maximum and minimum points of the vibration, R_±. That is

                                                           R_­ < R < R₊

We calculate the values R_­ and R₊ by knowing at the turning points where the oscillation changes direction and R₊ and R_­ where all the vibration energy is potential vibrational energy. That is, the molecules are in their most compressed state, R_­, or their most extended state R₊. The potential energy is given by a Morse function D_BC [1 ­ exp [­ _BC(R­R_e)]² where _BC, D_BC and R_e are the appropriate values for H₂. The quantum mechanical energy for the BC molecule in the and J quantum state is Equation 15 of Karplus† article is

                                                  (C-19)

The constants in this equation (e.g. G₁, I₁₁, etc.) are given for H₂ in Table I in Karplus. By equating equations (C-19) and (C-20) we can find the roots of the equation for R to determine the turning points, (R₊ and R_­). There is no angular momentum along the bond direction.

                                                    (C-20)

This calculation is tedious and difficult so we just need to accept that we can find the roots and “move on”. Figures PRS.3C-8, and PRS.3C-9 show the results of the calculations we just outlined.

Figures PRS.3C-8 and PRS.3C-9, Courtesy of ACS, Karplus, M., R.N. Porter, and R. D. Sharma, J of Chemical Physics, 43, p3259

One notes from Figures PRS.3C-8 (b) and (c) that the B-C molecule is rotating as evidenced by the fact that the R_AB and R_AC trajectories cross. On the other hand for the case of no rotation, J=0 in Figure (a), they do not cross. By the two crossings of R_AB and R_AC in Figure 8 (c) , one observes a faster rotation speed, than Figure 8(a) where J=2. In Figure PRS.3C-9 (a) we see that while B-C is not rotating before reaction, the AC molecule is rotating after reaction as evidenced by the crossing of the R_AC and R_BC trajectories. The time of the trajectory calculation is 4­8x10^­14s, the =0 vibration period is 0.5x10^­14s, the rotational period for J=1 is 20x10^­14s and for an any quantum number J is [27x10^­14/J[(J+1)]^1/2]s.

V.    Calculating the Reaction Cross Section

For a specified set of conditions, we now simply run a simulation and see whether or not a reaction occurs. Then we repeat for the same specified conditions but different Monte Carlo chosen values. A number of trajectories were calculated for the specified parameters [V_R, J, , b] using Monte Carlo techniques to calculate many trajectories similar to those shown in Figures PRS.3C-8 and PRS.3C-9. We keep track of the number of trajectories (simulations) that react, N_R, and those that don’t react. We then take the ratio of those trajectories that resulted in reaction to all the trajectories carried, N, out to calculate the reaction probability.

                                                                          (C-21)

Where N is the total number of trajectories calculated and N_r is the number that resulted in reaction. We sum the AB reactions and AC reaction to get N_r.

Reaction

Figure PRS.3C-10  Molecular Trajectories.

Now vary one of the specified parameters by running a number of Monte Carlo simulations for each value of that parameter. First b was varied while holding U, , and J constant. A number of simulations (trajectories) are carried out for each value of b in order to calculate P_r at that value of b. Then b is increased and the simulations repeated to again calculate P_r at another value of b. The results of the calculation are shown in Figure PRS.3C-11. For two different velocities. One notes that even for head-on collision (b=0) the probability is not 1.0, due to orientation effects and that the offset impact parameter, b, of A relative to the center of mass of BC is greater than 1.85 au, then no reaction will occur.

Figure PRS.3C-11 Probability of reaction as a function of the impact parameter.
[Note 1 a.u. = 59.9 pm and 1 hartree (htr) = 627 kcal]

The dashed lines represent the actual calculated values of P_r while the curve represents the smoothed values. We note there is a maximum value of the impact parameter, b_max, above which no reaction will take place.

r

            The reaction cross section, S_r is the probability of reaction, P_r and the cross section pb². In differential form S_r is a function of the relative velocity and the rotational and vibrational quantum numbers and J.

We are going to make an approximation to simplify the calculations. The approximation is that the curve in Figure PRS.3C-11 can be approximated by a cosine function, namely

                                                                                                        (C-23)

Both a and b_max increase with increasing velocity as shown in Table PRS.3C-5.

Table PRS.3C-5

For the curve shown for U_R = 1.17x10⁶ cm/s in Figure PRS.3-11

r as a function of b for each U. Using the cosine approximation, Eqn. (C-23) we can determine a and b_max for the chosen value of U. The reaction smoothed probability is shown as a function b for two different relative velocities in Figure PRS.3C-12

Figure PRS.3C-12 Reaction probability as a function of impact parameter for two different relative velocities

We need to specify the vibration and rotation energies, i.e., quantum number, and J when carrying out these calculations.

From Figure PRS.3C-12 we see as the velocity increases to 1.95 cm/s both a and b_max increase. Substituting for P_r using Equation (C-23) into Equation (C-22) we

                        Let

            Substituting for b and for db

Integrating by parts

                                                                       (C-24)

The reaction cross section S_r(U,J,) can now be found as a function of the relative velocity for which one can determine the corresponding relative transition energy, E_R

Equation (C-24) can be used to calculate the reaction cross section at any relative velocity. For example, when U_R = 1.17x10⁶ cm/s, a = 0.39, and b = 1.85 a.u., then

when,  , then

We continue in this manner to choose U, vary b and find P_r as a function of b to obtain a and b_max to arrive at Figure PRS.3C-12. One notes from this figure that while the cross section at U_R = 1.17x10⁶ cm/s for which S_r = 1.94 (a.u.)² agrees with the simulation. The value at U = 1.95x10⁶ cm/s of S_r = 4.4 (a.u.)² is different than the value of 5.45 (a.u.)² just calculated. The reason for this discrepancy is that we used the cosine function to approximate the P_r very b curve rather than say fitting (P_r vs. b) with a polynomial or plotting the “data” as bP_r as a function b and multiply the area under the curve by 2p to get S_r.

Figure PRS.3C-13

this technique while more tedious and labor intensive will give a more accurate value than the cosine approximation.

Figure PRS.3C-14  Reaction cross section as a function of relative velocity. Here 1 atomic unit º 1 a.u. = 0.59 Å = 59 pm = (1 a.u)²

We note S_r º 0 until we reach .† This energy is threshold kinetic energy necessary for the molecules A and BC to react. If the relative velocity U is such that the threshold energy is not exceeded, no reaction will occur. Now let’s look at the effect of some of the parameters, namely and J, on the reaction cross section.

I = moment of inertia

We now carryout a number of realizations and marking down those runs that result in reaction (e.g., Figure PRS.3C-9) and those that do not result in reaction (e.g., Figure PRS.3C-8) to arrive at Figure PRS.3C-11. Figure PRS.3C-14 shows the results of the calculations that give the reaction cross section as a function of relative velocity (kinetic energy ) for the case J=0 and =0.

            Now lets change J and vary U (i.e., E_R) to calculate the reaction cross section

Figure PRS.3C-15 Effect of rotation quantum number on reaction cross section.

We see that both the threshold kinetic energy, E_T, and the limiting cross section at high kinetic energies increase with increasing rotation quantum number (frequency) For J=2, the rotation period is 11.1x10^­14s compared with the interaction time of 10^­14s. At low kinetic energies the increased rotational energy makes it more difficult to react (steric effects). Also at higher kinetic energies the increased rotational energy increases the reaction cross section. However, the rotational energy is not available for crossing the potential energy barrier, V(R_AB, R_BC, R_AC).

Vibrational Energy

            The vibrational energy is†

                                                                                                          (C-26)

            = vibrational quantum number, v₀ is the frequency of vibration, and h is Plank’s constant.

                                                    Zero point energy =0

            The vibrational energy contributes to the kinetic to supply the energy to pass over the barrier. However, not all the vibrational energy is available for reaction. Now lets change and vary U, (i.e., E_R) and calculate the reaction cross section as shown in Figure PRS.3C-16

Figure PRS.3C-16  Effect of vibrational quantum number on reaction cross section.

At higher vibrational states the threshold energy decreases while the limiting value of S_r increases. However, Karplus notes that too few vibrational states were simulated to reach a definitive conclusion.

            Figure PRS.3C-17 shows the reaction cross as a function of kinetic energy for the hard sphere and line of center models along with the results of molecular dynamics calculations.

Figure PRS.3C-17 Comparison of models.

A discussion of the reaction cross section for the rigid sphere model and for line of centers model is given in Professional Reference Shelf 3A. We see the molecular dynamics (M.D.) trajectory for =0 and J=0 gives a reaction cross section that falls between these two models.

VI.     Rate of Reaction

Knowing the reaction cross section we can now proceed to calculate the overall rate of reaction, ­r_A. The differential rate of reaction d[­r_A(J,)] of species A, which has a velocity V_A and a concentration d_A, with species BC, which has a concentration of d_BC that has velocity V_B and is the vibrational state and the J rotational state, is

                                    (C-27)

                                                                (mole A/s/dm³)

where

­rA(J, )	=	Rate of reaction of A with BC molecules in the J rotational state and vibrational state.  (mol A/dm3/s)
U	=	Relative velocity  (dm/s)
dA	=	Concentration of A molecules with velocities between VA and .
	=	(mol/dm3)
CA	=	The total concentration of A molecules  (mol A/dm3)
f(VA)	=	Distribution of molecular velocities of molecule A. The distribution function is similar for BC molecules BC (c.f. Eqn. (C-27)).
dBC	=	Concentration of BC molecules with rotational state J, vibrational state , and velocities between VBC and  (mol/dm3)
dBC	=	(mol/dm3)
	=	(mol/dm3) = concentration of BC molecules in J rotational state and vibrational state. (mol BC/dm3)
CBC	=	The total concentration of all B­C molecules (mol/dm3), and
FBC(J, )	=	The fraction of B­C molecules in the J rotational state and the vibrational state.
		The distribution functions in the above equations are        for velocity                                         (C-28)        for rotation and vibration                                                           (C-29) where is the rotational-vibrational partition function, fJ is the statistical weight, and                             (C-30) and the values of Gi and Fi are given in Karplus’ article. [Note: also see Ch.12 Laidler pp.449-458.]†

Calculating the Total Rate
First we integrate Equation(C-27) overall all velocities so that velocity is incorporated in the rate and it is now only a function and J.

                                  (C-31)

                                                                      (C-32)

The rate, , Eqn (C-32), is only for those reactions with vibrational and rotation states J. The total rate of reaction is found by summing overall vibrational and rotational states

                                                                                                      (C-33)

                                                          (C-34)

                                                                                                      (C-35)

A comparison of the theoretical and calculate values of the specific reaction rate are shown in Table PRS.3C-6.

   _(C-36)

Equation (C-36) is Eqn. 41 in Karplus’ article.† Where N_Avo is Avogadros number.

            After carrying out the Monte Carlo Simulations for all specified variables b, U (i.e., E_AC), J and at a given temperature we use Equation (C-36) to calculate the specific rate k(T) next. The temperature was changed (e.g., from 300 to 500K) and the specific reaction rate recalculated. The results of some of these of the calculations are shown in Table PRS.3C-6.

Table PRS.3C-6 k(T) Values from M.D. calculations and experiment.^a,h

^a All values of k(T) is units of cubic centimeters per mole•second.

^b Calculated by Eq. (41) with S_r given by Eq. (43) and Table II of Karplus' paper.

^c Calculated by Eq. (C-36) with computed values of S_r.

^d K. Geib and P. Harteck, Z. Physik Chem. Bodenstein Fastband 849(1931).

^e M. van Meersche, Bull. Soc. Chim. Belges 60, 99 (1951).

^f A. Farkas and L. Farkas, Proc. Roy. Soc. (London) A152, 124 (1935).

^g G. Boato, G. Careri, A. Cimino, E. Molinari, and G. G. Volpi, J. Cl. Phys. 24, 783 (1956) have suggested that the values of van Meersche and Farkas and Farkas should be multiplied by 0.5 to correct for the present oxygen in the reaction mixture. Since this point has not been settled unequivocally, we list the range corresponding to 0.5 times the measured value to measured value.

^h Table PRS.3C-6 Courtesy of ACS, Karplus, lit.cit.

Looking at the specific reaction rates k in Table PRS.3C-6 we see

            We can use the theoretical values of k(T) predicted in Table PRS.3C-2 to determine the activation energy. A plot of ln k vs. 1/T is shown in Figure
PRS.3C-17.

Figure PRS.3C-18 Using molecular dynamics to predict activation energy.(From Karplus, et.al, J.Chem. Phys. 43, p3259 (1965))

From the slope we find E_A = 7.4 kcal/mole.

with

which is in excellent agreement with the experimental values.

            A plot of the values in Table PRS.C3-5 in the form of ln k(T) vs. 1/T will yield an activation energy of 7.4 kcal/mol which is very close to the experimental value, of 7.5 kcal/mol. We see there is excellent agreement of both the frequency factor A and the activation energy E_A between the theory and experiment. We also note the differences is the values of the following energies.

            A summary of all the energies obtained from the literature or by calculation are given in Figure PRS.3C-19.

Figure PRS.3C-19  Comparison of Energies

One notes that the activation energy is less than the barrier height which is a result of quantum mechanical tunneling. One also notes that not all the ground state vibrational energy is available to be added to the translational energy to cross the energy barrier.

VII. Problems

Problems for Molecular Dynamics

VIII. Closure

The equation of motion for molecules A and BC were coupled with potential energy surface to calculate trajectories of the A and BC molecules. The reaction probability was calculated by counting up the number of trajectories that resulted in reaction and dividing by all the trajectory trials. Below a threshold value of the translational kinetic energy no reaction occurs. There is also a maximum value of the impact parameter above which no reaction will occur. Owing to steric effects. The reaction probability only reaches a value of 0.6 even for heard on collisions (b-0) and very large translational kinetic energies. The reaction cross can be calculated from the impact parameter and relative velocity for given values of the vibrational and rotational quantum numbers. It was found to have a sigmodal shape, increasing as the kinetic energy (velocity) increased. The threshold kinetic energy decreased with increasing vibration quantum numbers and increased with increasing rotational quantum numbers.

The characteristic times are the rotational vibrational period for J=1 is, 19x10^­4s and for J=5 is 4.9x10^­14s, the vibration period is 0.5x10^­14s, the time of interaction is 10^­14s and the time of the trajectory calculation is between 4 and 8x10^­14s. We note that the rotational period is an order of magnitude greater than the interaction time, while the vibrational period is the same order as the interaction time.

The minimum kinetic energy 5.69 kcal is not sufficient to cross the potential energy barrier of 9.13 kcal and requires some of the vibrational energy from the BC molecules. The sum of the threshold (i.e., minimum energy necessary for reaction) 5.69 kcal and the =0 vibrational state energy of 6.2 kcal gives a total energy of 11.89 kcal which is greater than barrier height of 9.13 kcal. The difference between 11.89 and 9.13 indicates not all the vibrational energy is available for reaction. None of the rotational energy is available for reaction.

The rate constants k(T) were calculated as a function of temperature from first principles with no adjustable parameters. When the ln k was plotted as function of 1/T the activation energy was found to be 7.4 kcal/mol which is in excellent agreement with the experimental value of 7.5 kcal/mol. The fact that the activation energy is smaller than barrier is a consequence of quantum mechanical tunneling.

References

Karplus, M, P.N. Porter, and R. D. Sharma, J. of Chemical Physics, 43, 9, p.3259 (1965).

Laidler, K. J., Chemical Kinetics, 3rd ed., Harper Collins, New York, 1987.

Steinfeld, J. I., J. S. Francisco, and W. L. Hase, Chemical Kinetics and Dynamics, Prentice Hall, Upper Saddle River, NJ, 1989 (see Ch.8, p.246-268).

Figure PRS.3C-18  Energy diagram for the system H + H ® H₂ + H. The values are energy mol^­1 and relate to the calculations of Karplus, Porter, and Sharma.