Created in COMSOL Multiphysics 6.

Ion Drift Velocity Benchmark

This model is licensed under the COMSOL Software License Agreement 6.3.
All trademarks are the property of their respective owners. See www.comsol.com/trademarks.
Introduction
When ions are subjected to a static electric field in a rarefied buffer gas, the combined
effect of the electric force and collisions between the ions with ambient gas molecules
causes the average ion velocity to approach an equilibrium value known as the ion drift
velocity. The ability to accurately predict the drift velocity is important for the operation
of devices such as ion mobility spectrometers, which are capable of accurately analyzing
gas mixtures containing many different ion species by sorting the different species based
on small variations in their drift velocity.

In this example, the average drift velocity of a group of argon ions is computed and
compared to experimental data. The ions are modeled using the Charged Particle Tracing
interface. Collisions with a neutral background gas are applied using the Collisions node,
which supports various stochastic collision models including Elastic collisions and Resonant
Charge Exchange reactions. The resulting average ion drift velocity and ion temperature
agree closely with values from the literature.

Model Definition
In this example, the model particles are Ar+ ions that are released in a neutral background
gas of Ar atoms of a specified temperature and number density. The objective is to
determine the average steady-state velocity, or drift velocity, of the ions under the
combined influence of the continuous electric force and discrete collisions with the neutral
atoms in the background gas.

The following sections provide an overview of the electric force definition and the theory
of Monte Carlo collision modeling with the Charged Particle Tracing interface. For more
information on the way random collisions are modeled with the Particle Tracing Module,
see the section on Theory for the Charged Particle Tracing interface in the Particle
Tracing Module Users’s Guide.

ELECTRIC FORCE
This model uses a Parametric Sweep over the electric field amplitude to analyze its effect
on the ion drift velocity. Rather than being specified directly, the electric field amplitude is
expressed in terms of the ratio EN in Townsend units, where E (SI unit: V/m) is the
amplitude of the electric field and N (SI unit: 1m3) is the background gas number density.
The Townsend unit is defined as 1 Td  1021 V·m2. In this example, the Parametric
Sweep is run from a minimum of EN = 500 Td to a maximum of EN = 105 Td.

The electric force Fe (SI unit: N) on each particle is simply

2 | ION DRIFT VELOCITY BENCHMARK

 F e = ZeE

where e  1.602176565  1019 C is the elementary charge and Z (dimensionless) is the

charge number. In this example Z  1.

COLLISION TYPES
For the range of energies considered in this benchmark model, the interaction between the
simulated Ar+ ions and the rarefied background gas of neutral Ar atoms is dominated by
the following two reaction types:

1 Elastic collisions, in which only momentum is exchanged between the model particle
and the neutral background argon atom, while the total kinetic energy is conserved; and
2 Resonant Charge Exchange reactions, in which both charge and momentum are
exchanged between the model particle and the background gas, typically resulting in a
fast moving neutral atom and an slow-moving ion.

COLLISION FREQUENCY AND COLLISION PROBABILITY

Collisions between the argon ions and the neutral atoms occur at random times based on
the total collision frequency t (SI unit: 1/s). given by the expression

t =  j  t 
j

where jt is the collision frequency associated with the jth reaction in the model. In this
example, j  1 denotes the Elastic collision while j  2 denotes the Resonant Charge
Exchange reaction.

In general the collision frequency associated with a specific collision type is

 j = n j  v – v g  v – v g (1)

where

• vg (SI unit: m/s) is the velocity of an atom or molecule in the background gas,
• v (SI unit: m/s) is the particle velocity,
• j (SI unit: m2) is the collision cross section associated with the jth reaction, and
• n (SI unit: 1/m3) is the number density of the gas.

The collision cross section for each reaction type can be an arbitrary function of the relative
velocity and is usually determined experimentally. In this example the collision cross

3 | ION DRIFT VELOCITY BENCHMARK

section is approximated as a function of energy using Analytic functions, although it can
also be interpolated from imported tables of data with the Interpolation function type.

The probability Pt (dimensionless) that a particle undergoes a collision in a time interval
0 tis

t
P  t  = 1 – exp  – 0   t'  dt' (2)


For constant collision frequency, Equation 2 simplifies to

P  t  = 1 – exp  – t  (3)

COLLISION DETECTION ALGORITHMS

The Collisions node in the Charged Particle Tracing interface provides two distinct
algorithms for detecting and applying the random collisions between model particles and
the atoms or molecules in a rarefied background gas. The choice of algorithm is controlled
by the Collision detection list, which has options At time steps taken by solver, and Null
collision method, cold gas approximation.

Collision at Time Steps Taken by the Solver

The default option At time steps taken by solver is a brute force approach that checks for
collisions for each model particle at each time step taken by the solver. The following
assumptions are made:

1 The collision frequency for each model particle is assumed constant over the following
time step taken by the solver, even if the particle is accelerating.
2 Each model particle can undergo at most one collision per time step taken by the solver,
which is applied at the beginning of the step.

Both assumptions 1 and 2 restrict the size of the time step taken by the solver for the
results to be meaningful. The step size must be significantly smaller than the average free
time between collisions, which might increase very rapidly if the particles are ions or
electrons subjected to a strong electric field.

For each particle, a uniformly distributed random number U is sampled from the interval
0 1Then, because the collision frequency is assumed constant over the time step taken
by the solver, the logical expression for a collision occurring is obtained by comparing U
to the probability from Equation 3,

U  exp  – t  (4)

4 | ION DRIFT VELOCITY BENCHMARK

where some simplification has been made by noting that the probability distribution
functions of U and 1  U are exactly the same.

The pseudorandom number U is sampled once for every particle at the beginning of every
time step taken by the solver.

Null Collision Method, Cold Gas Approximation

The option Null collision method, cold gas approximation is an attempt to improve on the
option At steps taken by solver by allowing multiple collisions to occur for each particle
within a single time step taken by the solver and allowing these collisions to occur at
distinct times between the discrete time steps.

The “cold gas approximation” is the omission of the gas velocity term vg from Equation 1,

 j = n j  v  v

This collision detection algorithm only applies to highly energetic particles, those that
move at significantly higher speeds than the thermal speed of the background gas.

This algorithm is called the “null collision method” because the general approach is to
overestimate the collision probability over a given time interval, predict collisions based on
this overestimate of the collision frequency, and then discard a fraction of collisions based
on the extent to which the real frequency is less than this artificial constant value, with
these discarded interactions being called “null collisions”. Assume a constant trial
frequency m, such that m  t over the time interval 0 t. If this trial frequency were
the actual collision frequency over the time step, then Equation 3 would apply,

P m  t  = 1 – exp  –  m t 

For each model particle, a trial time tm (SI unit: s) is computed such that Pmtm  U1,
where U1 is a uniformly distributed number in the interval 0 1that is uniquely sampled
for each particle. Solving for the trial time yields

1
t m = – ------- log U 1
m

Now that the trial times for each particle have been computed, a second uniformly
distributed random number U2, not correlated with U1, is also randomly sampled from
the interval 0 1 for each particle that undergoes a collision in the time interval 0 t. If
the inequality

tm tm
U2 0 m dt  0   t  dt (5)

5 | ION DRIFT VELOCITY BENCHMARK

holds true, then the collision is actually a null collision, and must be discarded.

The first integral in Equation 5 can be computed exactly, because the trial frequency is
constant, but that the second integral can only be approximated numerically because the
real collision frequency can be an arbitrary function of time. Therefore the size of the time
step taken by the solver still has some bearing on the accuracy of the solution, even if this
source of error is usually not as large as it is for the option At time steps taken by solver.

CHOOSING BETWEEN COLLISION TYPES

For either collision detection algorithm, it is then necessary to determine whether the
detected collision is of the elastic or resonant charge exchange variety. The probability pj
of a specific collision type occurring is


p j = ----j


where the denominator is, as usual, the total collision frequency over all collision types.

REDUCING SIMULATION TIME WITH A STOP CONDITION

The solver sequence in this example uses a Stop Condition to terminate the study when the
particles have undergone a sufficient number of collisions. The rationale behind specifying
a stop condition based on the maximum number of collisions over all particles, is that the
maximum time needed for the average ion velocity to reach steady state may vary by several
orders of magnitude over the parametric sweep of different electric field amplitudes. Thus,
simply running the simulation to the same maximum solution time for every parameter
value would be either wasteful for those parameter values that statistically converge most
quickly, or inaccurate for those that statistically converge too slowly.

To keep track of the total number of collisions that each particle undergoes, the Count
collisions checkbox is selected in the settings for the Collisions node. The total number of
collisions over all model particles is computed using the expression
comp1.cpt.cptop1(comp1.cpt.col1.cex1.Nc). Here cpt.cptop1() is a built-in
nonlocal coupling that computes the sum of an expression over all model particles.

Results and Discussion

The relationship between the reduced electric field magnitude and the average ion drift
velocity is shown in Figure 1. The average ion temperature is shown in Figure 2.

In both plots, the computed solution shows good agreement with experimental data from
Ref. 1.

6 | ION DRIFT VELOCITY BENCHMARK

Figure 1: Plot of the drift velocity of Ar+ ions in a background gas of neutral argon. The
average drift velocity is compared to experimental data.

Figure 2: Plot of the ion temperature of Ar+ ions in a background gas of neutral argon. The
average ion temperature is compared to experimental data.

7 | ION DRIFT VELOCITY BENCHMARK

Reference
1. A.V. Phelps, “Cross Sections and Swarm Coefficients for Nitrogen Ions and Neutrals in
N2 and Argon Ions and Neutrals in Ar for Energies from 0.1eV to 10keV,” J. Phys. Chem.
Ref. Data, vol. 20, no. 3, pp. 557–573, 1991.

2. A.V. Phelps, “The application of scattering cross sections to ion flux models in discharge
sheaths,” J. Appl. Phys. vol. 76, pp. 747–753, 1994.

Application Library path: Particle_Tracing_Module/

Charged_Particle_Tracing/ion_drift_velocity_benchmark

Modeling Instructions
From the File menu, choose New.

NEW
In the New window, click Model Wizard.

MODEL WIZARD
1 In the Model Wizard window, click 3D.
2 In the Select Physics tree, select AC/DC > Particle Tracing > Charged Particle Tracing (cpt).
3 Click Add.
4 Click Study.
5 In the Select Study tree, select General Studies > Time Dependent.
6 Click Done.

GLOBAL DEFINITIONS

Parameters 1
1 In the Model Builder window, under Global Definitions click Parameters 1.
2 In the Settings window for Parameters, locate the Parameters section.

8 | ION DRIFT VELOCITY BENCHMARK

3 In the table, enter the following settings:

Name Expression Value Description

mAr 0.04[kg/mol]/ 6.6422E-26 kg Ar+ ion mass
N_A_const
ND 3.2956e22[1/m^3] 3.2956E22 1/m³ Background gas
number density
EoverN 500 500 Reduced electric
field in Townsend
Ez 1e-21[V*m^2]* 16478 V/m Electric field
EoverN*ND magnitude
T0 300[K] 300 K Gas temperature
maxCol 1e5 1E5 Maximum number of
collisions for
termination

DEFINITIONS
Enter raw data from Ref. 1 for the ion drift velocity as a function of the reduced electric
field for elastic collisions between Ar+ ions and neutral Ar atoms.

Interpolation 1 (int1)
1 In the Definitions toolbar, click Interpolation.
2 In the Settings window for Interpolation, locate the Definition section.
3 From the Data source list, choose File.
4 Click Browse.
5 Browse to the model’s Application Libraries folder and double-click the file
ion_drift_velocity_benchmark_velocity.txt.

6 Click Import.
7 Locate the Units section. In the Function table, enter the following settings:

Function Unit
Vdrift m/s

8 Locate the Definition section. In the Function name text field, type Vdrift.

Interpolation 2 (int2)
Enter raw data from Ref. 1 for the ion temperature (eV) as a function of the reduced
electric field.

1 In the Definitions toolbar, click Interpolation.

9 | ION DRIFT VELOCITY BENCHMARK

2 In the Settings window for Interpolation, locate the Definition section.
3 From the Data source list, choose File.
4 Click Browse.
5 Browse to the model’s Application Libraries folder and double-click the file
ion_drift_velocity_benchmark_temperature.txt.

6 Click Import.
7 Locate the Units section. In the Function table, enter the following settings:

Function Unit
ionTemp eV

8 Locate the Definition section. In the Function name text field, type ionTemp.

Enter the analytic approximation for momentum cross section for elastic scattering
between Ar+ ions and neutral Ar atoms from Ref. 2, which depends on the kinetic energy
of the particles.

Analytic 1 (an1)
1 In the Definitions toolbar, click Analytic.
2 In the Settings window for Analytic, type Qm in the Function name text field.
3 Locate the Definition section. In the Expression text field, type 1.15e-18*x^(-0.1)*
(1+0.015/x)^0.6.

4 Locate the Units section. In the table, enter the following settings:

Argument Unit
x eV

5 In the Function text field, type m^2.

Analytic 2 (an2)
Enter the analytic approximation for isotropic elastic collision between Ar+ ions and
neutral Ar atoms from Ref. 2, which depends on the kinetic energy of the particles.

1 In the Definitions toolbar, click Analytic.

2 In the Settings window for Analytic, type Qi in the Function name text field.
3 Locate the Definition section. In the Expression text field, type 2e-19/(x^(0.5)*(1+
x))+3e-19*x/(1+x/3)^(2.3).

10 | ION DRIFT VELOCITY BENCHMARK

4 Locate the Units section. In the table, enter the following settings:

Argument Unit
x eV

5 In the Function text field, type m^2.

GEOMETRY 1

Cylinder 1 (cyl1)
1 In the Geometry toolbar, click Cylinder.
2 In the Settings window for Cylinder, locate the Size and Shape section.
3 In the Radius text field, type 2.
4 In the Height text field, type 3.
5 Click Build All Objects.

CHARGED PARTICLE TRACING (CPT)

Particle Properties 1
1 In the Model Builder window, under Component 1 (comp1) >
Charged Particle Tracing (cpt) click Particle Properties 1.
2 In the Settings window for Particle Properties, locate the Particle Mass section.
3 In the mp text field, type mAr.
4 Locate the Charge Number section. In the Z text field, type 1.

Wall 1
1 In the Model Builder window, click Wall 1.
2 In the Settings window for Wall, locate the Wall Condition section.
3 From the Wall condition list, choose Bounce.

Release from Grid 1

1 In the Physics toolbar, click Global and choose Release from Grid.
2 In the Settings window for Release from Grid, locate the Initial Coordinates section.
3 In the qz,0 text field, type 1.
4 Locate the Initial Velocity section. From the Velocity specification list, choose Maxwellian.
5 In the Nv text field, type 30.
6 In the T0 text field, type T0.

11 | ION DRIFT VELOCITY BENCHMARK

Electric Force 1
1 In the Physics toolbar, click Domains and choose Electric Force.
2 In the Settings window for Electric Force, locate the Domain Selection section.
3 From the Selection list, choose All domains.
4 Locate the Electric Force section. Specify the E vector as

0 x
0 y
Ez z

Collisions 1
1 In the Physics toolbar, click Domains and choose Collisions.
2 In the Settings window for Collisions, locate the Domain Selection section.
3 From the Selection list, choose All domains.
4 Locate the Fluid Properties section. In the Nd text field, type ND.
5 In the T text field, type T0.
6 From the Collision detection list, choose Null collision method, cold gas approximation.
7 Locate the Collision Statistics section. Select the Count all collisions checkbox.

Elastic 1
1 In the Physics toolbar, click Attributes and choose Elastic.
2 In the Settings window for Elastic, locate the Collision Frequency section.
3 In the  text field, type Qi(cpt.Ep).
4 Locate the Collision Statistics section. Select the Count collisions checkbox.

Collisions 1
In the Model Builder window, click Collisions 1.

Resonant Charge Exchange 1

1 In the Physics toolbar, click Attributes and choose Resonant Charge Exchange.
2 In the Settings window for Resonant Charge Exchange, locate the Collision Frequency
section.
3 In the  text field, type (Qm(cpt.Ep)-Qi(cpt.Ep))/2.
4 Locate the Collision Statistics section. Select the Count collisions checkbox.

12 | ION DRIFT VELOCITY BENCHMARK

MESH 1
In the Model Builder window, under Component 1 (comp1) right-click Mesh 1 and choose
Build All.

STUDY 1

Parametric Sweep
1 In the Study toolbar, click Parametric Sweep.
Sweep the reduced electric field from 500 Td to 1e5 Td.
2 In the Settings window for Parametric Sweep, locate the Study Settings section.
3 Click Add.
4 In the table, enter the following settings:

Parameter name Parameter value list Parameter unit

EoverN (Reduced electric field 5e2, 1e3, 2e3, 3e3, 5e3,
in Townsend) 1e4, 2e4, 3e4, 5e4, 1e5

Step 1: Time Dependent

To accurately predict the collisions, specify a maximum time and a manual time step size
that are proportional to sqrt(maxCol/EoverN), which in turn is roughly proportional to
the free time between collisions.

1 In the Model Builder window, click Step 1: Time Dependent.

2 In the Settings window for Time Dependent, locate the Study Settings section.
3 In the Output times text field, type 0,5.0e-8*sqrt(maxCol/EoverN).

Solution 1 (sol1)
1 In the Study toolbar, click Show Default Solver.
2 In the Model Builder window, expand the Solution 1 (sol1) node, then click Time-
Dependent Solver 1.
3 In the Settings window for Time-Dependent Solver, click to expand the Time Stepping
section.
4 From the Steps taken by solver list, choose Manual.
5 In the Time step text field, type 2.5e-9*sqrt(maxCol/EoverN).
6 In the Amplification for high frequency text field, type 1.

13 | ION DRIFT VELOCITY BENCHMARK

7 Right-click Study 1 > Solver Configurations > Solution 1 (sol1) > Time-Dependent Solver 1
and choose Stop Condition.
Set the stop condition to stop the time-dependent solver when the maximum number
of collisions is reached. Make sure to store the solution at the time steps before and after
the stop condition is fulfilled.
8 In the Settings window for Stop Condition, locate the Stop Expressions section.
9 Click Add.
10 In the table, enter the following settings:

Stop expression Stop if Active Description

comp1.cpt.sum(comp1.cpt. True (>=1)  Stop expression 1
col1.cex1.Nc)>maxCol

11 Locate the Output at Stop section. From the Add solution list, choose
Steps before and after stop.
12 In the Study toolbar, click Compute.

RESULTS

Drift Velocity
1 In the Results toolbar, click 1D Plot Group.
Compare the drift velocity to the tabulated data. Make this comparison at the last time
step, after the drift velocity has reached an equilibrium value.
2 In the Settings window for 1D Plot Group, type Drift Velocity in the Label text field.
3 Locate the Data section. From the Dataset list, choose Study 1/
Parametric Solutions 1 (sol2).
4 From the Time selection list, choose Last.
5 Locate the Legend section. From the Position list, choose Lower right.

Global 1
1 Right-click Drift Velocity and choose Global.
2 In the Settings window for Global, locate the y-Axis Data section.
3 In the table, enter the following settings:

Expression Unit Description

cpt.ave(cpt.vz) m/s
Vdrift(EoverN) m/s

14 | ION DRIFT VELOCITY BENCHMARK

4 Locate the x-Axis Data section. From the Axis source data list, choose Outer solutions.
5 From the Parameter list, choose Expression.
6 In the Expression text field, type EoverN.
7 Click to expand the Legends section. From the Legends list, choose Manual.
8 In the table, enter the following settings:

Legends
COMSOL
Reference

Drift Velocity
1 In the Model Builder window, click Drift Velocity.
2 In the Settings window for 1D Plot Group, locate the Plot Settings section.
3 Select the y-axis label checkbox. In the associated text field, type Drift velocity (m/
s).
4 Click to expand the Title section. From the Title type list, choose Manual.
5 In the Title text area, type Drift velocity comparison.
6 In the Drift Velocity toolbar, click Plot. The resulting image should look like
Figure 1.

Ion Temperature
Now compare the ion energy to tabulated data. From Ref. 1, the average ion energy
corresponds to a one-dimensional energy distribution, that is, Ep  kBT/(2e).

1 Right-click Drift Velocity and choose Duplicate.

2 In the Settings window for 1D Plot Group, type Ion Temperature in the Label text field.

Global 1
1 In the Model Builder window, expand the Ion Temperature node, then click Global 1.
2 In the Settings window for Global, locate the y-Axis Data section.
3 In the table, enter the following settings:

Expression Unit Description

cpt.ave(2*cpt.Ep) eV
ionTemp(EoverN) eV

Ion Temperature
1 In the Model Builder window, click Ion Temperature.

15 | ION DRIFT VELOCITY BENCHMARK

2 In the Settings window for 1D Plot Group, locate the Plot Settings section.
3 In the y-axis label text field, type Ion temperature (eV).
4 Locate the Title section. In the Title text area, type Ion temperature comparison.
5 In the Ion Temperature toolbar, click Plot. The resulting image should look like
Figure 2.

16 | ION DRIFT VELOCITY BENCHMARK