def rampvoltage(device, Vsource, begin_bias, end_bias, init_step_size,
min_step, max_iter, rel_error, abs_error, callback):
'''
Ramps bias with assignable callback function
'''
start_bias = begin_bias
if (start_bias < end_bias):
step_sign = 1
else:
step_sign = -1
num_successes = 0
last_bias = start_bias
step_size = init_step_size
while (abs(last_bias - end_bias) > min_step):
print(("last end %e %e") % (last_bias, end_bias))
next_bias = last_bias + step_sign * step_size
if next_bias < end_bias:
next_step_sign = 1
else:
next_step_sign = -1
if next_step_sign != step_sign:
next_bias = end_bias
print("setting to last bias %e" % (end_bias))
print("setting next bias %e" % (next_bias))
ds.circuit_alter(name=Vsource, value=next_bias)
try:
ds.solve(type="dc",
absolute_error=abs_error,
relative_error=rel_error,
maximum_iterations=max_iter)
except ds.error as msg:
if msg[0].find("Convergence failure") != 0:
raise
ds.circuit_alter(name=Vsource, value=last_bias)
step_size *= 0.5
print("setting new step size %e" % (step_size))
if step_size < min_step:
raise NameError("Min step size too small")
num_successes = 0
continue
num_successes += 1
if (num_successes > 5) and (step_size < init_step_size):
step_size *= 2
if step_size > init_step_size:
step_size = init_step_size
print("setting new step size %e" % (step_size))
num_successes = 0
print("Succeeded")
last_bias = next_bias
callback()
def solve(self, *args, **kwargs):
for region in self.device.mesh.regions:
log.info('Computing photogeneration for region: %s' % region)
# Assume 1D for now
nodes = ds.get_node_model_values(device=self.device.name,
region=region,
name='x')
pg = np.zeros((len(nodes), len(self.light_source)), dtype=float)
rfidx = RefractiveIndex(region.material)
for idλ, λ in enumerate(self.light_source):
# I know, using unicode names here is asking for trouble
# ... but looks nice when using greek letters
# TODO: compute the real reflection
R = 0
# λ is nm, but irradiance's units are W⋅m–2⋅nm–1
P = self.light_source.irradiance(λ)
Φ_0 = (P * λ * 1e-9 / PhysicalConstants.hc) * (1 - R)
# α(λ)'s units are cm-1. But we use m
α = rfidx.alpha(λ) * 1e+2
# TODO: do not assume conversion efficiency is 100% (1 photon = 1EHP by now)
η_g = 1.0
# Debug
print('\n' + '-' * 70)
print('Photogeneration dump. Parameters:')
print(
'λ={:.2f} nm; P(λ)={:.2f}; Φ_0(λ)={:8.2e} m-2 s-1; α(λ)={:8.2e} m-1'
.format(λ, P, Φ_0, α))
print('-' * 24)
print(' x(μm) | G_op(x, λ)')
print('-' * 24)
for idx, x in enumerate(nodes):
pg[idx, idλ] = η_g * Φ_0 * exp(-α * x)
print('{:8.2f} | {:8.2e}'.format(x * 1e6, pg[idx, idλ]))
# Total photogeneration for each node
# Conditions:
# 100% quantum efficiency
# Ignoring reflection
pgen_by_node = np.add.reduce(pg, 1)
for i, v in enumerate(pgen_by_node):
ds.set_node_value(device=self.device.name,
region=region,
name='G_op',
index=i,
value=float(v))
if 'type' in kwargs:
ds.solve(**kwargs)
else:
ds.solve(type='dc', **kwargs)
def rampvoltage(device, Vsource, begin_bias, end_bias, init_step_size, min_step, max_iter, rel_error, abs_error, callback):
'''
Ramps bias with assignable callback function
'''
start_bias=begin_bias
if (start_bias < end_bias):
step_sign=1
else:
step_sign=-1
num_successes = 0
last_bias=start_bias
step_size=init_step_size
while(abs(last_bias - end_bias) > min_step):
print("last end %e %e") % (last_bias, end_bias)
next_bias=last_bias + step_sign * step_size
if next_bias < end_bias:
next_step_sign=1
else:
next_step_sign=-1
if next_step_sign != step_sign:
next_bias=end_bias
print "setting to last bias %e" % (end_bias)
print "setting next bias %e" % (next_bias)
ds.circuit_alter(name=Vsource, value=next_bias)
try:
ds.solve(type="dc", absolute_error=abs_error, relative_error=rel_error, maximum_iterations=max_iter)
except ds.error as msg:
if msg[0].find("Convergence failure") != 0:
raise
ds.circuit_alter(name=Vsource, value=last_bias)
step_size *= 0.5
print "setting new step size %e" % (step_size)
if step_size < min_step:
raise NameError("Min step size too small")
num_successes = 0
continue
num_successes += 1
if (num_successes > 5) and (step_size < init_step_size):
step_size *= 2
if step_size > init_step_size:
step_size = init_step_size
print "setting new step size %e" % (step_size)
num_successes = 0
print "Succeeded"
last_bias=next_bias
callback()
def rampbias(device, contact, end_bias, step_size, min_step, max_iter,
rel_error, abs_error, callback):
'''
Ramps bias with assignable callback function
'''
start_bias = ds.get_parameter(device=device,
name=GetContactBiasName(contact))
if (start_bias < end_bias):
step_sign = 1
else:
step_sign = -1
last_bias = start_bias
while (abs(last_bias - end_bias) > min_step):
print(("last end %e %e") % (last_bias, end_bias))
next_bias = last_bias + step_sign * step_size
if next_bias < end_bias:
next_step_sign = 1
else:
next_step_sign = -1
if next_step_sign != step_sign:
next_bias = end_bias
print("setting to last bias %e" % (end_bias))
print("setting next bias %e" % (next_bias))
ds.set_parameter(device=device,
name=GetContactBiasName(contact),
value=next_bias)
try:
ds.solve(type="dc",
absolute_error=abs_error,
relative_error=rel_error,
maximum_iterations=max_iter)
except ds.error as msg:
if msg[0].find("Convergence failure") != 0:
raise
ds.set_parameter(device=device,
name=GetContactBiasName(contact),
value=last_bias)
step_size *= 0.5
print("setting new step size %e" % (step_size))
if step_size < min_step:
raise "Min step size too small"
continue
print("Succeeded")
last_bias = next_bias
callback()
def solve(self, *args, **kwargs):
if not args and not kwargs:
self.initial_solution()
elif kwargs.get('type', '') == 'ramp':
kwargs['type'] = 'dc'
start = kwargs.pop('start')
stop = kwargs.pop('stop')
step = kwargs.pop('step')
contact = kwargs.pop('contact')
for v in range(start, stop, step):
set_parameter(device=self.name,
name=self._contact_bias_name(contact),
value=v)
solve(**kwargs)
self.print_currents()
else:
solve(*args, **kwargs)
for mdl in self._models:
mdl.solve(*args, **kwargs)
def rampbias(device, contact, end_bias, step_size, min_step, max_iter, rel_error, abs_error, callback):
'''
Ramps bias with assignable callback function
'''
start_bias=ds.get_parameter(device=device, name=GetContactBiasName(contact))
if (start_bias < end_bias):
step_sign=1
else:
step_sign=-1
last_bias=start_bias
while(abs(last_bias - end_bias) > min_step):
print("last end %e %e") % (last_bias, end_bias)
next_bias=last_bias + step_sign * step_size
if next_bias < end_bias:
next_step_sign=1
else:
next_step_sign=-1
if next_step_sign != step_sign:
next_bias=end_bias
print "setting to last bias %e" % (end_bias)
print "setting next bias %e" % (next_bias)
ds.set_parameter(device=device, name=GetContactBiasName(contact), value=next_bias)
try:
ds.solve(type="dc", absolute_error=abs_error, relative_error=rel_error, maximum_iterations=max_iter)
except ds.error as msg:
if msg[0].find("Convergence failure") != 0:
raise
ds.set_parameter(device=device, name=GetContactBiasName(contact), value=last_bias)
step_size *= 0.5
print "setting new step size %e" % (step_size)
if step_size < min_step:
raise "Min step size too small"
continue
print "Succeeded"
last_bias=next_bias
callback()
CreateMesh(device=device, region=region)
SetParameters(device=device, region=region)
set_parameter(device=device, region=region, name="taun", value=1e-8)
set_parameter(device=device, region=region, name="taup", value=1e-8)
SetNetDoping(device=device, region=region)
print_node_values(device=device, region=region, name="NetDoping")
InitialSolution(device, region)
# Initial DC solution
solve(type="dc",
absolute_error=1.0,
relative_error=1e-10,
maximum_iterations=30)
DriftDiffusionInitialSolution(device, region)
###
### Drift diffusion simulation at equilibrium
###
solve(type="dc",
absolute_error=1e10,
relative_error=1e-10,
maximum_iterations=30)
####
#### Ramp the bias to 0.5 Volts
####
# bias_volt = 0.0
tag=bottom_tag,
name=bottom_tag)
ds.finalize_mesh(mesh=meshname)
ds.create_device(mesh=meshname, device=device)
# Add some physics
for region, params in zip(regions, [CdS_params, CdTe_params]):
for name, value in params.items():
ds.set_parameter(device=device, region=region, name=name, value=value)
set_dd_parameters(device=device, region=region)
# Initial DC solution
ds.solve(type="dc",
absolute_error=1.0e10,
relative_error=1e10,
maximum_iterations=150)
# for region in regions:
# DriftDiffusionInitialSolution(device, region)
###
### Drift diffusion simulation at equilibrium
###
currs = []
volts = []
bias_contact = "top_n1"
for volt in range(-20, 20, 1):
volt = volt / 10
volts.append(volt)
ds.set_parameter(device=device,
name=f"{bias_contact}_bias",
def setup_drift_diffusion(self,
dielectric_const=11.9,
intrinsic_carriers=1E10,
work_function=4.05,
band_gap=1.124,
Ncond=3E19,
Nval=3E19,
mobility_n=1107,
mobility_p=424.6,
Nacceptors=0,
Ndonors=0):
"""
Sets up equations for a drift-diffusion style of dc current transport.
kwargs here are device-level parameters, imbuing all regions with these properties
If a specific region or material is to have a different set of parameters, they can be set through the
Region constructor.
:param dielectric_const:
:param intrinsic_carriers:
:param work_function:
:param band_gap:
:param Ncond:
:param Nval:
:param mobility_n:
:param mobility_p:
:param Nacceptors:
:param Ndonors:
:return:
"""
# Begin legacy copy-paste job
# Set silicon parameters
device = self.name
region = self.regions[0].name
eps_si = dielectric_const
n_i = 1E10
k = kb
mu_n = mobility_n
mu_p = mobility_p
set_parameter(device=device,
region=region,
name="Permittivity",
value=eps_si * eps_0)
set_parameter(device=device,
region=region,
name="ElectronCharge",
value=q)
set_parameter(device=device, region=region, name="n_i", value=n_i)
set_parameter(device=device, region=region, name="T", value=T)
set_parameter(device=device, region=region, name="kT", value=k * T)
set_parameter(device=device,
region=region,
name="V_t",
value=k * T / q)
set_parameter(device=device, region=region, name="mu_n", value=mu_n)
set_parameter(device=device, region=region, name="mu_p", value=mu_p)
# default SRH parameters
set_parameter(device=device, region=region, name="n1", value=n_i)
set_parameter(device=device, region=region, name="p1", value=n_i)
set_parameter(device=device, region=region, name="taun", value=1e-5)
set_parameter(device=device, region=region, name="taup", value=1e-5)
# CreateNodeModel 3 times
for name, value in [('Acceptors', "1.0e18*step(0.5e-5-x)"),
('Donors', "1.0e18*step(x-0.5e-5)"),
('NetDoping', "Donors-Acceptors")]:
result = node_model(device=device,
region=region,
name=name,
equation=value)
logger.debug(f"NODEMODEL {device} {region} {name} '{result}'")
print_node_values(device=device, region=region, name="NetDoping")
model_name = "Potential"
node_solution(name=model_name, device=device, region=region)
edge_from_node_model(node_model=model_name,
device=device,
region=region)
# Create silicon potentialOnly
if model_name not in get_node_model_list(device=device, region=region):
logger.debug("Creating Node Solution Potential")
node_solution(device=device, region=region, name=model_name)
edge_from_node_model(node_model=model_name,
device=device,
region=region)
# require NetDoping
for name, eq in (
("IntrinsicElectrons", "n_i*exp(Potential/V_t)"),
("IntrinsicHoles", "n_i^2/IntrinsicElectrons"),
("IntrinsicCharge",
"kahan3(IntrinsicHoles, -IntrinsicElectrons, NetDoping)"),
("PotentialIntrinsicCharge", "-ElectronCharge * IntrinsicCharge")):
node_model(device=device, region=region, name=name, equation=eq)
node_model(device=device,
region=region,
name=f"{name}:{model_name}",
equation=f"simplify(diff({eq},{model_name}))")
# CreateNodeModelDerivative(device, region, name, eq, model_name)
### TODO: Edge Average Model
for name, eq in (("ElectricField",
"(Potential@n0-Potential@n1)*EdgeInverseLength"),
("PotentialEdgeFlux",
"Permittivity * ElectricField")):
edge_model(device=device, region=region, name=name, equation=eq)
edge_model(device=device,
region=region,
name=f"{name}:{model_name}@n0",
equation=f"simplify(diff({eq}, {model_name}@n0))")
edge_model(device=device,
region=region,
name=f"{name}:{model_name}@n1",
equation=f"simplify(diff({eq}, {model_name}@n1))")
equation(device=device,
region=region,
name="PotentialEquation",
variable_name=model_name,
node_model="PotentialIntrinsicCharge",
edge_model="PotentialEdgeFlux",
variable_update="log_damp")
# Set up the contacts applying a bias
is_circuit = False
for contact_name in get_contact_list(device=device):
set_parameter(device=device,
name=f"{contact_name}_bias",
value=0.0)
# CreateSiliconPotentialOnlyContact(device, region, contact_name)
# Start
# Means of determining contact charge
# Same for all contacts
if not InNodeModelList(device, region, "contactcharge_node"):
create_node_model(device, region, "contactcharge_node",
"ElectronCharge*IntrinsicCharge")
#### TODO: This is the same as D-Field
if not InEdgeModelList(device, region, "contactcharge_edge"):
CreateEdgeModel(device, region, "contactcharge_edge",
"Permittivity*ElectricField")
create_edge_model_derivatives(device, region,
"contactcharge_edge",
"Permittivity*ElectricField",
"Potential")
# set_parameter(device=device, region=region, name=GetContactBiasName(contact), value=0.0)
contact_bias_name = f"{contact_name}_bias"
contact_model_name = f"{contact_name}nodemodel"
contact_model = f"Potential -{contact_bias_name} + ifelse(NetDoping > 0, -V_t*log({CELEC_MODEL!s}/n_i), V_t*log({CHOLE_MODEL!s}/n_i))"\
CreateContactNodeModel(device, contact_name, contact_model_name,
contact_model)
# Simplify it too complicated
CreateContactNodeModel(
device, contact_name, "{0}:{1}".format(contact_model_name,
"Potential"), "1")
if is_circuit:
CreateContactNodeModel(
device, contact_name,
"{0}:{1}".format(contact_model_name,
contact_bias_name), "-1")
if is_circuit:
contact_equation(device=device,
contact=contact_name,
name="PotentialEquation",
variable_name="Potential",
node_model=contact_model_name,
edge_model="",
node_charge_model="contactcharge_node",
edge_charge_model="contactcharge_edge",
node_current_model="",
edge_current_model="",
circuit_node=contact_bias_name)
else:
contact_equation(device=device,
contact=contact_name,
name="PotentialEquation",
variable_name="Potential",
node_model=contact_model_name,
edge_model="",
node_charge_model="contactcharge_node",
edge_charge_model="contactcharge_edge",
node_current_model="",
edge_current_model="")
# Biggie
# Initial DC solution
solve(type="dc",
absolute_error=1.0,
relative_error=1e-10,
maximum_iterations=30)
drift_diffusion_initial_solution(device, region)
solve(type="dc",
absolute_error=1e10,
relative_error=1e-10,
maximum_iterations=30)
def solve(self, *args, **kwargs):
if not args and not kwargs:
self.initial_solution()
else:
solve(*args, **kwargs)