Tel:886-3-5131379, fax:886-3-5736952, email:ccmeng@mail.nctu.edu.tw
*Department of Electrical Engineering, National Chung-Hsing University, Taichung, Taiwn, R. O. C.
Abstract — A method to monitor the GaInP/GaAs HBT device structure including ledge thickness is demonstrated in this paper. The base thickness and base doping density are obtained through base transit time and base sheet resistance measurements while the base transit time is measured through the cut-off frequency measurements at various bias points. A large size HBT device with two emitters is used to measure the ledge thickness. Emitter doping profile and collector doping profile can be obtained by the large size HBT device through C-V measurement. An FATFET device formed by two emitters as drain and source terminals and the interconnect metal as the Schotty gate on the ledge between two emitters is used to measure the ledge thickness.
I. INTRODUCTION
The HBT (Heterjunction Bipolar Transistor) structure strongly affects HBT’s high frequency characteristics and reliability. It is desirable to confirm the layer structure and ledge thickness after the device has been fabricated. It has become especially important in today’s foundry business model because normally the device structure is proprietary to the foundry company and will not be released to the circuit designer. On the other hand, it is advantageous for a circuit design company to make sure the device quality from lot-to-lot and wafer-to-wafer. Furthermore, it is necessary to fine tune the device structure and choose the right process if a better circuit performance is needed. In this paper, a method is established to find the device structure including ledge thickness for GaInP/GaAs HBT devices. Ledge thickness is an important parameter for reliability concern. The ledge thinning process in an HBT is not precisely controlled. Normally, a thicker-than-needed emitter is designed and a selective etching is used to remove the emitter cap for the emitter mesa. A wet etching dipping is used to thin down the emitter ledge to a totally depleted condition. The ledge thinning by chemically wet etching is somewhat uncontrollable. In some case, a thin emitter is designed such that, after the emitter cap is removed, the remaining ledge is already depleted. A quick wet etching dip is used to expose the fresh InGaP surface. In any case, a way to monitor the final ledge thickness is important to guarantee the device reliability. The method needs a special two-emitter large size HBT for ledge thickness measurement. The base
thickness and doping density are obtained through base transit time and base sheet resistance measurements. The base transit time is obtained through S parameter measurement at various bias points. The cut-off frequencies at various collector current and collector voltage can be used to remove the effect of base-collector transit time and emitter charge time to obtain base transit time. A standard transmission line measurement is used to obtain the base sheet resistance. Thus, base thickness and base doping density can be obtained from base transit time and base sheet resistance. A large size HBT device with two emitters is developed in this paper to measure the ledge thickness, emitter and collector doping profiles.
Emitter doping profile and collector doping profile can be obtained by the large size HBT device through C-V measurement. An FATFET device formed by two emitters as drain and source terminals and the interconnect metal as Schotty gate on the ledge between two emitters is used to measure the ledge thickness. The resulting measurement fits well with the material data.
II. DEVICE STRUCTURE CHARACTERIZATION METHOD
A large size two-emitter HBT device is designed as shown in figure 1. A GaInP/GaAs HBT has a heavily doped base, thus a conventional C-V measurement can be used to obtain the emitter and collector doping profiles.
The doping profiles for emitter and collector are illustrated in figure 2 and figure 3, respectively. Emitter doping is 3x1017/cm3 and 0.067 µm thickness. Collector is 2x1016/cm3 and 0.6 µm thickness. The ledge is strongly related to the device reliability. A fully depleted ledge prevents injected electron minority carriers from recombining at the exposed surface by inverting the band diagram at the surface. An FATFET device formed by two emitters as drain and source terminals and the interconnect metal as Schotty gate on the ledge between two emitters as illustrated in figure 1 is used to measure the ledge thickness. The C-V measurements in figure 4 reveal an unusually high doping starting at 0.051µm depth from the Schottky barrier surface. The measured unusually high doping density at this depth is caused by the heavily doped p type base. Thus, the C-V measurement can be
used to probe the ledge thickness even through the exact doping of the ledge can not be obtained.
A 2.4X3X2 HBT device is used to perform the high frequency S parameter measurement. The corresponding I-V curve is illustrated in figure 5. Forward Gummel plot and reversed Gummel plot are also illustrated in figure 5.
The absence of the conduction band discontinuity in GaInP/GaAs interface can be evident by the fact that the Cummel plot for collector current overlays the reverse Gummel plot for reverse collector current. The almost zero conduction band discontinuity means that the GaInP material is an ordered structure instead of the disorder structure. An order GaInP emiiter has demonstrated a good reliability and thus the method here can also used to check the material quality.
S parameters are measured at various bias points and the cut-off frequency as a function of collector current and collector voltage is illustrated in figure 7. A Kirk effect is observed at high current. The emitter-collector charging time is the reciprocal of the cut-off radian frequency and is decomposed into emitter charging time, base transit time, base-collector transit time and collector charging time as follows.
( )
It will be evident that the collector charging time is negligible. Thus, a plot of emitter-collector charging time as a function of the reciprocal of collector current as shown in figure 8 can be used to extrapolate the sum of base-transit time and base-collector transit time time. The collector transit time is a function of the base-collector voltage through the formula below.
bi
The base-emitter voltage is 1.4 Volt for the range of collector currents interested. Thus, a mapping between depletion width and base-collector voltage is obtain through C-V measurement. The sum of base transit and base-collector transit time can be ploted as a function of the depletion width to extrapolate the base-transit time.
The base transit time is related to the base thickness by the formula below. The base layer sheet resistance can be obtained through the transmission line measurement. The sheet resistance is from a four point probe measurement.
The sheet resistance is related to the base doping density and base thickness by the formula below.
( )
The minority mobility and the majority mobility are related to the base doping density by the formula below.
( )
majority 38010 641
Thus, by solving both equations, the base doping and thickness are.
( )
The whole HBT structure together with ledge thickness is thus obtained.
III. CONCLUSION
The HBT (Heterjunction Bipolar Transistor ) structure strongly affects HBT’s high frequency characteristics and reliability. It is advantageous for a circuit design company to make sure the device quality from lot-to-lot and wafer-to-wafer. A method to determine the HBT device structure and material quality by DC and RF measurements have been developed in this paper. The whole HBT structure together with ledge thickness is very useful to fine tune the device structure and choose the right process if a better circuit performance is needed.
ACKNOWLEDGEMENT
This work was supported by the National Science Council of Republic of China under contract NSC 92-2219-E-009-023 and by the Ministry of Education under contract 89-E-FA06-2-4.
REFERENCE
[1] W. Liu, “Handbook of III-V heterojunction bipolar transistors”, John Wily & Sons.
[2] W. Liu, T. Henderson, E. Beam III and S. K. Fan, “Electron saturation velocity in Ga0.5In0.5P measured in a GaInP/GaAs/GaInP double-heterojunction bipolar transistor”, Electronics Letters, Vol. 29, No. 21, p1885-1887, October 1993.
Figure 1. The two-emitter HBT device for measuring ledge thickness, emitter doping profile and collector doping profile.
The two emitters are used as drain and source while the interconnect metal is used as the gate metal for the ledge between two emitters.
0.050 0.055 0.060 0.065 0.070
1E17 1E18 1E19
Emiiter doping profile Doping(cm-3)
Doping density(cm-3)
Depth(um)
Figure 2. The emitter doping profile obtained through C-V measurement.
0.3 0.4 0.5 0.6
1E15 1E16 1E17 1E18
Collector Doping profile Doping(cm-3)
Doping density(cm-3)
Depth(um)
Figure 3. The collector doping profile obtained through C-V measurement.
0.035 0.040 0.045 0.050 0.055 0.060
1E17 1E18 1E19 1E20
Fat-FET doping profile doping(cm^-3)
Doping density(cm-3)
Depth(um)
Figure 4. The ledge thickness measurement.
0 1 2 3 4 5 6 7
1112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101
ABC
lmnopq r s t uvwxyzaaabacadaeafagahaiajakalamanaoapaqarasatauavawaxayazbabbbcbdbebfbgbhbibjbkblbmbnbobpbqbrbsbtbubvbwbxbybzcacbcccdcecfcgchcicjckclcmcncocpcqcrcsctcucvcw
Ib=0uA Ib=10uA Ib=20uA Ib=30uA Ib=40uA Ib=50uA Ib=60uA Ib=70uA Ib=80uA Ib=90uA Ib=100uA Ib=110uA Ib=120uA Ib=130uA Ib=140uA Ib=150uA
1 Ib=160uA
A Ib=170uA
a Ib=180uA Ib=190uA Ib=200uA
Ic(A)
Vce(volt)
Figure 5. I-V curve of 2.4X3X2 HBT device.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
1 ForwardIB
ForwardIC ReverseIB ReverseIC
Current
Vbe(volt)
Figure 6. GUmmel plot of 2.4X3X2 HBT device.
Gate
Metal Source
Drain
75u
100u
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 5
10 15 20 25 30 35 40
Vc=0.5V Vc=1.0V Vc=1.5V Vc=2.0V Vc=2.5V Vc=3.0V
Ft(GHz)
Ic(A)
Figure 7. Cutoff frequency as a funvtion of collector current with collector-emitter voltage as a parameter
0 200 400 600 800 1000 1200 1400
2 3 4 5 6 7 8 9 10 11 12
Vc=0.5V Vc=1.0V Vc=1.5V Vc=2.0V Vc=2.5V Vc=3.0V tec(psec)
1/Ic(1/Amp)
Figure 8. A plot of emitter-collector charging time as a function of the reciprocal of collector current. The plot can be used to extrapolate the sum of base-transit time and base-collector transit time.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
2.0 2.5 3.0 3.5 4.0 4.5 5.0
tB+Xdep/2Vs(psec)
Xdep(um)
Figure 9. A plot of the sum of base-transit time and base-collector transit time as a function of the depeltion width. The plot can be used to extrapolate the base transit time.
GaAs n+
n G aA s+ n GaAs p GaAs+ n InGaAs+
n InGaP 1 1 7 5 A
D
6.755E19(cm−3)
6000 A
D
510A 670A D
D
2E16(cm−3) 3E17(cm−3)
Figure 10. The HBT layer structure including ledge determined by the DC and RF measurements.