Shani G. Radiation Dosimetry: Instrumentation and Methods

Подождите немного. Документ загружается.

380 Radiation Dosimetry: Instrumentation and Methods

the photodiode response in terms of cumulative dose (Si)

to the 20-MeV electron ﬁeld is shown in Figure 8.32. It

can be seen that after a dose of 50 krad, which is the

ionization dose expected within the BELLE SVD per

year in a worst-case scenario, the response is degraded

by ~15%.

In Figure 8.33, the reverse current measured under full

depletion conditions before and after electron irradiation,

with



~ 10

e/cm

, is plotted as a function of reverse

voltage. Following irradiation, the reverse current

increased by a factor of ~13. The reverse current was

monitored as a function of time to determine the effects

of room temperature annealing. Over a period of two

months, the reverse current was seen to decrease by 35%

from the value measured 5 hrs after electron irradiation.

All measurements were made at a constant temperature

of 20°C.

The radiation damage of the detector is measured by

an increase in the reverse bias current per unit volume of

fully depleted detector, according to Equation 8.9:

(8.9)

where



is a current damage parameter which is depen-

dent on the type and energy of radiation

 reverse current before irradiation

I  reverse current after irradiation

Vol  full depletion volume of the detector.

FIGURE 8.31 Comparison of transverse radiation dose proﬁle across the 200-



m-wide microbeam measured by the MOSFET

detector in both ‘‘normal’’ and ‘‘edge-on’’ orientations. (From Reference [18]. With permission.)

FIGURE 8.32 Hamamatsu photodiode response to a 20-MeV electron ﬁeld as a function of absorbed dose. (From Reference

[19]. With permission.)



Vol

--------------







Ch-08.fm Page 380 Friday, November 10, 2000 12:03 PM

Solid-State Dosimeters 381

To obtain calibration curves for the RADFET system

under high-energy electrons in terms of dose in Si, four

such devices were exposed to the 20-MeV electron ﬁeld,

together with the Si photodiode and PIN dosimeter diodes.

The shift of the threshold voltage under a constant current

corresponding to the thermal stable point was measured.

All measurements were done during the 3-hour irradia-

tion. The spread in the change in threshold voltage (V

)

as a function of the dose in silicon (D(Si)) was found to

be less than 3% for different RADFETs from the same

batch. Figure 8.34 shows the calibration curves for both

‘‘R’’ and ‘‘K’’ type RADFETS in the 20-MeV electron

ﬁeld in terms of dose in Si. The analytical expression

derived from the calibration curves for the ‘‘K’’ type

RAFET is D  40.23  (V

), where D is the dose

measured in krad(Si) and V

is the voltage shift in volts.

For the ‘‘R’’-type RADFETs, D  1.81  (V

III. P-TYPE SEMICONDUCTOR DOSIMETER

MOSFETs have a disadvantage of a response which is

dependent upon the incident photon energy, particularly

at low energies. In recent years,

p-FET dosimeters have

been developed with specially-grown thick-gate oxides

which have a large number of oxide traps. Sensitivities of

10 mV/rad(Si) have been achieved. The sensitivity to

radiation can be enhanced by applying a large positive

bias during radiation, which forces more of the positive

FIGURE 8.33 Detector reverse current measured before and after 20-MeV electron irradiation at a ﬂuence. 

 9.2  10

2

(From Reference [19]. With permission.)

FIGURE 8.34 Calibration curves for the RADFETs in terms of dose in Si. (From Reference [19]. With permission.)

Ch-08.fm Page 381 Friday, November 10, 2000 12:03 PM

382 Radiation Dosimetry: Instrumentation and Methods

oxide charge to the interface. The sensitivity to tempera-

ture can be minimized by operating the p-FET with a

current at the temperature-independent point. The main

limitation in using silicon semiconductor detectors is

effects caused by radiation damage. A nonlinear response

with dose rate has proved to be serious for n-type detectors

when pre-irradiated to 10 kGy or more in 20-MeV elec-

trons. A sensitivity drop after pre-irradiation is another

effect which is more pronounced for n-type detectors com-

pared to p-type detectors of the same resistivity. A third

effect of radiation damage is the sensitivity increase with

increasing temperature. [20]

A Pantak HP-320 quasi-monoenergetic x-ray unit was

used by Edwards et al. [21] to determine the response

of two Thomson and Nielson TN-502RD MOSFETs, a

Scanditronix EDP-10 semiconductor (build-up cap 10 mm:

tissue equivalence), a BDD-5 semiconductor (build-up

cap 4.5 mm: tissue equivalence), and a LiF:Mg:Ti TLD

over the energy range 12–208 keV. The sensitivity of

each detector was normalized to the value produced by

exposure to 6-MV x-rays. The maximum relative sensi-

tivities of the two MOSFET detectors were 4.19

 0.25

and 4.44  0.26, respectively, occurring at an incident

x-ray energy of 33 keV. The maximum relative sensitiv-

ity of the Scanditronix EDP-IO of 2.24  0.13 occurred

at 65 keV, and for the EDD-5, it was 7.72  0.45 at 48 keV.

The TLD produced a maximum relative sensitivity of 1.31

 0.09 at 33 keV. Compared with available data based on

heteroenergetic x-ray sources, these measurements have

identiﬁed a more representative response for each detector

to low-energy x-rays.

Scanditronix

p-type semiconductor detectors were

investigated (Scanditronix, Uppsala, Sweden). These

included the EDD-5 and the EDP-IO, both connected to

a DPD-510 direct patient dosimeter. The 0.9-mm sensitive

volume of the EDD-5 detector is encapsulated by Perspex,

which provides an effective tissue-equivalent depth of

4.5 mm. This detector is designed primarily for critical

organ measurements outside the main beam. The 0.9-mm

sensitive volume of the EDP-IO detector is surrounded by

a 0.75-mm-thick stainless steel cap (providing a tissue-

equivalent depth of 10 mm) and is adhered to a Perspex

plate (1-mm thickness). This detector is designed for mea-

surements inside the main beam generated by 4–8 MV

linear accelerators.

A correction factor U

for each detector d was calcu-

lated to allow for the small difference between the SSD

to the center of its sensitive volume (J

) and the SSD to

a rigid surface:

(8.10)

The sensitivity

X,d

of a detector at each x-ray energy

relative to that produced by 6-MV x-rays was calculated

from:

(8.11)

where

x,d

is the detector response, b

is the response to

background radiation, k

is the kerma rate recorded by the

NE2550 Protection Level Secondary Standard with a beam

current of 1 mA and SSD of 1 m, and t

is the exposure time.

The corrected response per unit air-kerma rate was normal-

ized to the detector’s response per unit air kerma rate from a

6-MV linear accelerator by multiplying by two previously

determined calibration factors. The ﬁrst factor L

converted

each detector’s response to the absolute dose rate, measured

at the maximum depth dose in a water phantom irradiated at

an SSD of 1 m with a 10-cm  10-cm ﬁeld from a 6-MV

linear accelerator. The second factor

converted this to the

corresponding air-kerma rate and included any effects due to

attenuation within the detector and from backscatter.

The variation in relative sensitivity of each detector with

x-ray energy is illustrated in Figure 8.35. Relative to its

response to a 6-MV linear accelerator beam, the sensitivity

of each detector increased as the x-ray energy decreased

until it reached a maximum value, and then it decreased

with further decrease in x-ray energy.

On-chip p-FETs were developed by Buehler et al. [22]

to monitor the radiation dose of n-well CMOS ICs by

monitoring threshold voltage shifts due to radiation-

induced oxide and interface charge. The design employs

closed-geometry FETs and a zero-biased n-well to elimi-

nate leakage currents. The FETs are operated using a

constant current chosen to greatly reduce the FETs’ tem-

perature sensitivity. The dose sensitivity of these p-FETs

is about 2.6 mV/krad(Si) and the off-chip instrumenta-

tion resolves about 400 rad(Si)/bit. When operated with a

current at the temperature-independent point, it was found

that the pre-irradiation output voltage is about 1.5 V,

which depends only on design-independent silicon mate-

rial parameters. The temperature sensitivity is less than



V/°C over a 70°C temperature range centered about

the temperature insensitive point.

The schematic cross section of the device shown in

Figure 8.36 indicates that the n-well and source are sep-

arated so that they can operate at slightly different biases

required by the operational ampliﬁer, U

. This allows all

the forced current to ﬂow through the p-FET channel.

Drain-to-well leakage is shunted to ground. Grounding

the n-well is a departure from normal CMOS circuit oper-

ation, where the n-well is normally connected to VDD.

In operation, all terminals of the p-FET are operated

near ground except the drain, which operates near V0 

1.5 V. During irradiation, the device is biased in the off state.

The two RL resistors are used to bleed off any charge

0.4

-------







x,d

()

-----------------------------------------



Ch-08.fm Page 382 Friday, November 10, 2000 12:03 PM

Solid-State Dosimeters 383

remaining on the p-FET. The instrumentation is calibrated

by positioning the switch to CAL, shown in Figure 8.36.

This replaces the p-FET with calibration resistor RC.

The threshold voltage is described by:

(8.12)

where D is the dose, VT

is the threshold voltage at T

and

D  0, VT

 VT T | and .

The temperature and dose dependence of



(transconduc-

tance) is given by:

(8.13)

FIGURE 8.35 The relative sensitivity S

(response per unit air-kerma rate) normalized to that produced by 6-MV x-rays of two

MOSFET

p-type semiconductor diodes (EDP-10 and EDD-5) and two LiF:Mg:Ti TLD (one in PVC, one in Nescoﬁlm™) to low-

energy quasi-monoenergetic x-rays (error bars are

 1 SD). (From Reference [21]. With permission.)

FIGURE 8.36 p-FET total dose circuitry. (From Reference [22]. With permission.)

VT VT

()VT

D



→

 VT



 D

D 0→





T T

()

n



D

Ch-08.fm Page 383 Friday, November 10, 2000 12:03 PM

384 Radiation Dosimetry: Instrumentation and Methods

where



evaluated at T

and D  0,







 ,

n characterizes the mobility temperature dependence, and

T is the absolute temperature. , KP 



. W

 W  DW and L

 L  DL are the effective

channel width and length, respectively,



is the zero ﬁeld

channel mobility, C

is the gate oxide capacitance per unit

area.

The VT values, plotted in Figure 8.37, show a high degree

of linearity during irradiation and a slight recovery with anneal.

The group average slope of the VT vs. dose curve during

irradiation is VT

 1.698  0.038 mV/krad(Si). The shift

in VT with radiation is consistent with the build-up of positive

oxide charge and interface states. The slight recovery of VT

during room-temperature anneal is consistent with the slight

loss of oxide charge. The interface state density is stable, as

seen by the ﬂat response of the mobility during anneal.

The output voltage for the four p-FET samples is

shown in Figure 8.38. This plot was obtained from data

sets. The data shows a nearly linear rise in V

with dose

and a slight annealing effect.

Brucker et al. [23] present the results of studies of

dose enhancement in dual and single dielectric pMOSFET

dosimeters for various package and die designs. Eight

different MOSFET designs and package types were inves-

tigated over a photon energy range from 14 to 1250 keV.

Packages ﬁlled with silicon grease, aluminum oxide, or

FIGURE 8.37 Four 1.2-



m CMOS p-FET threshold voltage dose/anneal responses. (From Reference [22]. With permission.)

FIGURE 8.38 Four 1.2-



m CMOS p-FET output voltage dose/anneal responses determined at the current, ID

. (From Reference

[22]. With permission.)



D 0→



KP W

L



Ch-08.fm Page 384 Friday, November 10, 2000 12:03 PM

Solid-State Dosimeters 385

parafﬁn eliminated the contribution of back scatter to the

enhanced dose. These modiﬁcations allowed measure-

ments of the usual dose enhancement at the aluminum or

polysilicon gate-silicon nitride (dual dielectric devices) or

silicon dioxide interfaces (single dielectric parts) and at the

silicon nitride–silicon dioxide interface. In addition to the

primary peak in the DEF (dose enhancement factor) curve

vs. energy at 45.7 keV, there is a second peak at about 215

keV. This peak might be due to enhancements at the inter-

faces of a MOSFET. These interface effects were small in

the single-insulator parts in standard ceramic packages and

signiﬁcantly larger in the dual-insulator devices. The

effects were reduced by ﬁlling the packages with the mate-

rials as previously described. The geometry of the package,

the size of the air gap between the die’s surfaces, and the

lid of the package impacts the value of the DEF.

The three-dimensional models used by the code were

those in Figure 8.39. The ﬁgure shows a cross-sectional and

top view of the dual-in-line ceramic and TO-18L kovar can

packages. The gold coatings on the bonding pads and in

the well or cavity of the ceramic package are indicated. This

diagram actually represents the REM 501A part type. The

other REM devices did not have the coating in the cavity.

Figure 8.40 shows a comparison of results obtained with

air- and grease-ﬁlled TO-5 kovar cans containing #3 transis-

tors. Figure 8.41 contains plots that compare the DBF values

for Sandia #l parts in ceramic and kovar packages. Results

for air- and grease-ﬁlled ceramic packages are also shown.

A dose-rate-independent p-MOS dosimeter for space

applications was discussed by Schwank, et al. [24] A dual

dielectric p-MOS dosimeter (RADFET) has been designed

at Sandia. The RADFET consists of a thermally grown

oxide and a CVD deposited nitride. With a negatively

applied bias, holes generated in the SiO

transport to and

are trapped at the SiO

/Si

interface, producing a mea-

surable threshold-voltage shift. Because holes are trapped

away from the Si/SiO

interface, hole neutralization by

tunneling and interface-trap build-up are minimized, result-

ing in little fade or annealing of the RADFET output

response. RADFETs were irradiated at dose rates from

0.002 to 50 rad(Si)/s, with biases from 5 to 20 V.

FIGURE 8.39 Cross-sectional and top views of dual-in-line ceramic and TO-18L kovar can packages. (From Reference [23]. With

permission.)

FIGURE 8.40 Dose enhancement vs. energy for Sandia #3 in

TO-5 kovar cans with and without grease in package. (From

Reference [23]. With permission.)

Ch-08.fm Page 385 Friday, November 10, 2000 12:03 PM

386 Radiation Dosimetry: Instrumentation and Methods

RADFETs were also annealed for times up to 10

s at

temperatures up to 100°C. Within experimental uncertainty,

no difference in RADFET output response at a given bias

was observed over the dose rate range examined and for

25°C anneals. At an anneal temperature of 100°C, only a

20% decrease in RADFET output response was observed.

To monitor total dose, RADFETs were characterized

for the threshold-voltage shift, V

, and the change in

source-to-drain voltage, V

, necessary to maintain a

constant source-to-drain current, I

The RADFETs were designed at Sandia and fabricated

at Micrel Semiconductor Corporation. The structure of the

RADFET is shown in Figure 8.42. The RADFETs used by

Schwank et al. consisted of a 60-nm thermal oxide grown

on a (100) n on n



expitaxial substrate, and a low-pressure

CVD 100-nm silicon nitride layer deposited on top of the

oxide. All RADFETs were p-channel transistors with

metal gates and were taken from the same wafer. The

wafer-to-wafer and lot-to-lot variation in RADFET

response is not known, so other wafers would have to be

calibrated independently. Transistors were packaged in

24-pin DIP packages.

p-MOS dosimeters with a single dielectric have been

fabricated using a variety of techniques. Most of these

measure the threshold-voltage shift induced by positive

charge trapping in the oxide. For thermal oxides, most of

the trapped charge occurs in the oxide near either the top

gate-Si/SiO

(negative gate bias during irradiation) or

bottom Si/SiO

(positive gate bias during irradiation)

interface. If holes are trapped near the top interface,

the measured threshold-voltage shift is very small, due

to a small charge moment arm. For this reason, standard

p-MOS dosimeters are normally operated with a positive

applied bias. Unfortunately, for positive applied biases,

interface traps will be generated at the Si/SiO

interface

and trapped holes can be neutralized by thermal emission

and tunneling processes. The time dependence of interface-

trap build-up and the neutralization of holes will result in

fade of the dosimeter’s output characteristics. [24]

The radiation-induced threshold voltage shift, V

of Sandia’s dual dielectric RADFET is given by:

(8.14)

where



and are the dielectric constants for SiO

(



 3.9) and Si

(



 7.5), respectively,



is the

permittivity of free space, L

is the thickness of the oxide

layer, L

is the total dielectric thickness (SiO

 Si

FIGURE 8.41 Illustrates effect of package and ﬁlling in reduc-

ing DBF for Sandia #l in TO-18L and ceramic FLTPAC. (From

Reference [23]. With permission.)

V



------------

-------

x()n



x() xd







---------------------



x() xd







FIGURE 8.42 RADFET transistor structure. (From Reference [24]. With permission.)

Ch-08.fm Page 386 Friday, November 10, 2000 12:03 PM

Solid-State Dosimeters 387

and n

(X) is the distribution of trapped holes in the SiO

layer (or net positive charge if both electrons and holes

are trapped). x is measured relative to the metal gate/Si

interface.

Figure 8.43 is a plot of calculated relative radiation

sensitivity vs. SiO

thickness for several values of insula-

tor thickness (SiO

 Si

At very low temperatures (~100°K), holes generated

in the SiO

will be immobile and the distribution of holes

throughout the SiO

is uniform. Under these conditions,

the radiation sensitivity is proportional to

(8.15)

The radiation response of the RADFETs for dose rates

from 0.002 to 50 rad(Si)/s is shown in Figure 8.44. This

ﬁgure is a plot of V

for a constant source-to-drain

current (~4 mA) for RADFETs irradiated at room tem-

perature to 100 krad(Si) with a constant gate-to-source

bias, V

, of 5 V.

For low total dose levels, the output response of the

RADFETs is linear with total dose. Figure 8.45 is a plot

of V

for RADFETs irradiated to 20 krad(Si) at dose

rates from 0.002 to 50 rad(Si)/s with V

 5 V. Over

this dose range, for all dose rates, the dose is given by

(8.16)

FIGURE 8.43 Relative RADFET sensitivity for the case where holes are trapped near the SiO

/Si

interface for oxide thicknesses

from 0.05 to 100 (arbitrary units) and total insulator thicknesses,

, from 1 to 100 (arbitrary units). (From Reference [24]. With

permission.)

FIGURE 8.44  V

vs total dose for RADFET. (From Reference [24]. With permission.)



----------

()



----------------------------------



Dose[rad(Si)] 8300 V



Ch-08.fm Page 387 Friday, November 10, 2000 12:03 PM

388 Radiation Dosimetry: Instrumentation and Methods

where V

is given in volts. At higher total doses, 20

krad(Si), the RADFET output response becomes sublinear

(see Figure 8.44) as the trapped charge in the insulator

begins to reduce the internal ﬁeld in the oxide. As the oxide

electric ﬁeld is lowered, the amount of electron-hole recom-

bination increases. The increase in electron-hole recombi-

nation will result in fewer trapped holes in the insulator. [24]

A stacked p-FET dosimeter consisting of a RADMON

mother chip with three p-FETs and multiplexer and an

attached RADFET has been developed by Buehler et al.

[25] for the STRV-2/MWIR detector. Calibration of the

dosimeter using an Am-241 source indicates that the

RADFET is about 20 times more sensitive to radiation

than the RADMON. This dosimeter is expected to mea-

sure radiation dose from rads to megarads.

The dosimeter, seen in Figure 8.46, consists of a thin-

oxide RADMON mother chip with three p-FETs, a multi-

plexer, and other devices. A thick-oxide RADFET is

attached to the mother chip. The device operates on the

principle that radiation passing through the gate oxide region

of the p-FET creates hole-electron pairs. Since the oxides

have mainly hole traps, radiation-induced positive charge is

trapped in the oxide. This build-up of positive charge

changes the threshold voltage of the device, which is inter-

preted in terms of radiation dose. A cross-section view is

shown in Figure 8.47.

FIGURE 8.45 V

vs. total dose for RADFETs irradiated to 20 krad(Si) at dose rates from 0.002 to 50 rad(Si)/s with V

 5V.

(From Reference [24]. With permission.)

FIGURE 8.46 RADFET mounted on the RADMON mother chip in a 16-pin ﬂat package. Three radiation sensitivities are achieved:

PK6 has highest sensitivity, PG0 and PG7 have moderate sensitivity, and PG4 has lowest sensitivity, being shielded by the RADFET.

(From Reference [25]. With permission.)

Ch-08.fm Page 388 Friday, November 10, 2000 12:03 PM

Solid-State Dosimeters 389

Room temperature current-voltage measurements are

shown in Figure 8.48 for the ﬁve FETs in this experiment.

These curves were modeled using the saturation drain

current expression:

(8.17)

where

VO is the FET output voltage,



 KP  W/L is

the transconductance, KP 



, W is the channel width,

L is the channel length,



is the channel mobility, and C

is the gate oxide capacitance per unit area. The threshold

voltage, VT, is determined from the extrapolation of the

curves shown in Figure 8.48 to zero drain current using a

least-squares ﬁtting technique. The



values were deter-

mined from the slope of the characteristics shown in

Figure 8.48.

Semiconductor detectors based on p-type silicon but

with different doping levels have been investigated by

Grusell and Rikner. [26] It was shown that a p-type detector

with a low doping level and high resistivity showed a non-

linear dose rate response if it were radiation-damaged in a

high-energy photon beam, which contains neutrons. By

increasing the doping level, it was shown that a detector

with a resistivity of 0.2 cm stayed linear after pre-irradi-

ation in radiation ﬁelds from high-energy electrons, pho-

tons, and protons. Other parameters did not show any

changes of clinical importance at the different doping levels.

FIGURE 8.47 Cross section of the MWIR dosimeter package showing the wire bonds that ground the Au ring and aluminum-coated

mylar lid

. (From Reference [25]. With permission.)

FIGURE 8.48 Typical current-voltage characteristics for the ﬁve FETs located on the MWIR. Note that the FET characteristics for

PG0, PG4, and PG7 overlap. The numbers in the key are

VT in volts and





A/V

. (From Reference [25]. With permission.)



---





VO VT()



Ch-08.fm Page 389 Friday, November 10, 2000 12:03 PM