Professional Documents
Culture Documents
net/publication/313165615
CITATIONS READS
7 231
7 authors, including:
P. Lorenzini
University of Nice Sophia Antipolis
100 PUBLICATIONS 1,795 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Vincent Mosser on 18 April 2017.
Abstract
Hall sensors mth a high senahvlty, a low thermal dnft and a low offset voltage based on AMXAs/InGrAs/GaAs
heterostructures have been developed The physnzal phenomena responsible for the thermal dnft of the Hall
senntwlty are rmewed and mvestlgated usmg a set of test devices Hnth well-controlled structure parameters
These results have been used to optmuze the design of sensors v&h a Hall factor m the 1000 V/AfT range m
order to reduce the temperature sensltmty of the channel electron denstty down to a few 100 ppmiT The
dependence of the Hall factor and of Its thermal drift on the bias current has been mvestlgated We show that
the bias-current level can be tuned to achieve a very low thermal dnft together mth a high absolute sennhwty
m the range 0 4-O 5 V/T
the devuzes have the same general structure (Fig l), nesses A model was developed 111order to calculate
the only varymg parameters being the spacer and cap- and predict the electron density, and thus the Hall
layer thicknesses and the integrated sheet concentration factor, of a gven device as a function of the fabncatton
of the dopmg atoms The Al and In contents are equal parameters of the heterostructure This model, which
to 0 3 and 0 15, respectwely, and the quantum-well does not involve any fittmg parameters, performs the
thickness amounts to 130 A A band diagram 1s shown self-consistent resolution of the Schriidmger equation,
m Fig 2 Poisson equation and Ferm-Duac statlstlcs m the
Dependmg on the fabncatlon parameters, the channel Hartree approxunation, using the Fermi-level pmmng
electron density n, as measured by the Hall effect, is at the surface as the boundary condition Modelhng
between 5 X 101’ and 15 X 10” cmm2 The good control results are also shown 111Fig 3
of the fabncatlon parameters IS clear m Fig 3, which
shows n,, as a function of the mtegrated dopm density
for gwen spacer (40 A) and cap-layer (600 1 ) thlck- 3. Electrical characteristics
In a biased device the channel potenttal drops wn- As the channel resistance &=(ekn,)-‘(LIP’) de-
tmuously between the source and dram contacts How- pends on the temperature, mamly through the mob&y
453
800
a
Baa0
B
e 400
4
1 200
~~,,,,'
8 \
0
-500 250 0 250 500
Blascurrent Iti @A) Btas current @A)
Fig 5 Hall factor vs btas current for an AlGaAsiInGaMGaAs Fig 7 Thermal sensltlvlty of the Hall factor as a function of
Hall device with a Hall electron dens@ n,= 15 X 1Ol’ cmW2 the bias current for the same device as III Fig 6
16. . . . . . . ..I.‘. I.... The value of the slope, 0 75 K-l A-l, 1s to be compared
with the expected value of the term m brackets in eqn
.
14- (3) Takmg into account the expenmental value of
3
the Hall factor under low biasing conditions, I&,=
+- 12
810 V/m, a device aspect ratio LIW=6, the chan-
“5
1 nel-substrate capacitance C, = lo-’ F/cmm2, and the
zi expenmental thermal denvatlve of the inverse mobility
OS from Fig 6, the calculated slope amounts ot 09 K-’
_ I,,_:,:
A-’ In spite of the crudeness of the model, this value
06
150 200 250 300 350 1s m good agreement with the expenmental one, thus
Temperature (K) demonstrating the vahdlty of this approach
Fig 6 Inverse mob&y vs temperature for an AlGaAs/InGaAs/ Interestingly, eqn (3) predicts the existence of a
GaAs Hall devxe with a Hall factor Km=-810 V/&T, showmg constant bias current for w&h the thermal drift of
the hnear relatlonshlp /L.-‘-T
the experunental Hall factor will remam very small
Tins allows a large operatmg temperature range without
external compensation This fact was also noticed by
p,,, the magnitude of the backgatmg effect will be Kyburz ei al [3] m 3D InAlAs/InGaAs/InP Hall sensors
temperature dependent If we now consider a Hall For the devxes shown m Fig 7, the value of this low-
device where the thermal dnft of the free electron thermal-d& bias current 1s about 05 mA, and the
density m the channel has already been reduced to a resultmg absolute sensmvlty amounts to 0 4-O 5 VfT
few hundred ppm/?Z, eqn (1) can be differentiated to The corresponding voltage drop m the device IS about
yield 3 V, and the power dissipation remains low, m the
1 dnH 1 dn* c, --- 1 L d(~L,-l)
mW range Usmg the same heterostructure, the value
--=----
(2) of the absolute magnetic sensltlvlty could be m pnnclple
nH dT n, dT eras0 enso 2W dT
increased by 50% by reducmg the aspect ratio of the
The thermal behavlour of the moblhty appeanng m Hall bar down to L/W=4
eqn (2) can be determined expenmentally As shown
m Fig 6, m such 2D devxes the inverse moblhty has
a hnear temperature dependence CL,,-’=aT 111the
temperature range of interest This 1s not surpnsing
since the moblhty 1s hmlted mainly by acoustic phonon 7. conclusiolls
scattermg
Thus, the relatnre thermal drift of the Hall factor Magnetic sensors Hnth supenor performance can be
reads designed by makmg use of the band-gap engmeenng
of III-V heterostructures The pseudomorphic AlGaAs/
s,=s,- c&,.Z$p-’ 1 (3) InGaAs/GaAs system, whose technology is available
[ 1 from mdustnal foundnes, 1s of particular interest [4,
As expected from eqn (3), a plot of the experunental 51 Hall-effect devices with a Hall factor amounting to
thermal dnft of the Hall factor S, shows a hnear at least 1000 V/A/T can be designed The mtnnslc
dependence on the bias current, as shown in Fig 7 temperature coefficient of the active layer electron
454