The initial flux isn't zero in TDS simulation

User Support
1

Hi,I’m doing 1D TDS simulation.The gas is deuterium,and the material is tungsten.
I found when I set E_p near to E_D(such as E_p = 0.6 and E_D = 0
.39),there will be a situation that the initial value of flux is not zero when the temperature is very low.

guess10
guess103840×2880 200 KB

guess8
guess83840×2880 205 KB

guess9
guess93840×2880 182 KB

And here’s my code:

import festim as F
import os
my_model = F.Simulation()

import numpy as np

vertices = np.linspace(0, 1e-4,num = 1000)

my_model.mesh = F.MeshFromVertices(vertices)

atom_density =6.28e28 
W = F.Material(
    id=1,
    D_0=2.9e-7,  # m2/s
    E_D=0.39,  # eV
)

my_model.materials = W


import sympy as sp
trap_conc = 7e-5*atom_density
my_model.initial_conditions = [
		F.InitialCondition(field="1", value=trap_conc)
	]
trap_1 = F.Trap(
        k_0=2.9e12/atom_density,
        E_k=0.39,
        p_0=2e13,
        E_p=1,
        density=trap_conc,
        materials= W
    )

my_model.traps = [trap_1]


my_model.boundary_conditions = [
    F.DirichletBC(surfaces=[1, 2], value=0, field=0)
]

ramp = 1  # K/s
my_model.T = F.Temperature(value=300.1 + ramp * (F.t))



#Stepsize
my_model.dt = F.Stepsize(
    initial_value=0.1,
    stepsize_change_ratio=1.2,
    dt_min=1e-6,
    max_stepsize=1,
)

#Settings
my_model.settings = F.Settings(
    absolute_tolerance=1e8,
    relative_tolerance=1e-12,
    final_time = 1173
)


list_of_derived_quantities = [
        F.TotalVolume("solute", volume=1),
        F.TotalVolume("retention", volume=1),
        F.TotalVolume("1", volume=1),
        F.HydrogenFlux(surface=1),
        F.HydrogenFlux(surface=2),
	F.AverageVolume(field="T", volume=1),
    ]

derived_quantities = F.DerivedQuantities(
    list_of_derived_quantities,
    # filename="tds/derived_quantities.csv"  # optional set a filename to export the data to csv
)


my_model.exports = [derived_quantities]

my_model.initialise()
my_model.run()


t = derived_quantities.t
T = derived_quantities.filter(fields="T").data
flux_left = derived_quantities.filter(fields="solute", surfaces=1).data
flux_right = derived_quantities.filter(fields="solute", surfaces=2).data

flux_total = (-np.array(flux_left) - np.array(flux_right))/2
kk = flux_total.astype(str)
with open('10.txt', 'w') as file:
    for row in kk:
        file.write('\t'.join(map(str, row)) + '\n')
ref = np.genfromtxt('480k-0dpa.csv', delimiter=',')
tt = ref[:,0]
ttt = tt - 300.1
flux = ref[:,1]

import matplotlib.pyplot as plt
plt.plot(t, flux_total, linewidth=0.5)
#plt.plot(ttt,flux)

plt.ylabel(r"Desorption flux (m$^{-2}$ s$^{-1}$)")
plt.xlabel(r"t(s)")
plt.title("E=1,n=7e-5,0.1mm")
plt.savefig("guess10.png",dpi=600)

I wonder why that happens. Thank you!

2

Hi @Jellycarrot!

I suppose this behaviour, particularly the artifact at low temperatures for weak traps, is related to your initial conditions. Since hydrogen can escape from weak traps even at room temperatures, its initial “realistic” distribution at 300 K (i.e. one that could be obtained by simulating the exposure phase prior to TDS) might not correlate with the used one.

At lower initial temperatures, there are no artificats on TDS spectra:

image
image606×454 38.7 KB

For higher detrapping energies, the spectra shift to higher temperatures, so higher initial temperatures can be considered without any artifacts.

Please, let me know if it helps. Possibly, @remidm and @jhdark could provide more advices.

1 Like
3

@Jellycarrot I believe that increasing the waiting time before the tds phase should also help removing this initial peak as the weak traps will dissociate.

Consider something like:

ramp = 1  # K/s
start_tds_time = 200  # s
T_value = sp.Piecewise(
    (300.1 + ramp * (F.t - start_tds_time), F.t >= start_tds_time),  # increase temp after sometime
    (300.1, True)  # otherwise T=300.1
)
my_model.T = F.Temperature(value=T_value)
4

Thank you!
That’s a good revolution to my problem.

5

Yes,it really help.I have tried,and the curve truly started from 0 at 200K.
But I’m trying to fit the experimental data, which is beginning at 300K.So,maybe I need to use Professor remidm’s method.
Anyway,thank you for you good advice and promt reply!

2 Likes