1.
Introduction

In recent years, high electron mobility transistors (HEMT) biosensors have been applied to detect protein and DNA concentration^[1–5]. High electron mobility transistors attract a great deal of attention due to their high work cut-off frequency, low power consumption and low noise^[6]. HEMT utilizes modulation doped heterojunction structure to separate the doped layer and conduct layer; this kind of structure possesses higher two dimension electron gas concentration and electron mobility^{[7, 8]}. Attributed to the wurtzite lattice structure, GaAs and GaN structure lack a symmetric center, so there is spontaneous polarization in their structure^{[9, 10]}. This phenomenon further enhances the two dimension electron gas concentration. The two dimension electron gas channel is close to the surface so it could be affected by external factors, such as stress^{[11, 12]}.



Based on this performance, there is a great deal of research pertaining to HEMT sensors^{[1, 13, 14]}. For example, piezoelectric polarization induced by mechanical stress influences two dimension electron gas in the interface, so this phenomenon is utilized as a pressure sensor^[15]. There is other research focusing on the application of HEMT as gas sensors and biosensors^{[16, 17]}. HEMT sensing devices that are modified with biomolecules are reported to realize label-free detection of the C-reactive protein^{[18, 19]}. HEMT biosensors have the potential to realize real-time detection. A GaN/AlGaN HEMT cardiac troponin I real-time biosensor has been reported recently^[20]. This suggests the possibility of HEMT biosensor application as a real-time detection method in clinic applications. As high-sensitivity and real-time biosensors, HEMT biosensors have great market prospects. For the application of HEMT biosensors, the electric properties consistency of the inter-chip performance have an important influence on the stability and repeatability of the detection. Compared to GaN/AlGaN, GaAs/AlGaAs HEMT has a more mature fabrication technology. GaAs/AlGaAs HEMTs have the potential to be used as biosensors. In this research, we fabricated GaAs/AlGaAs HEMT biosensors of different epitaxial structures and device structures to study the electric properties consistency.

2.
Experimental material and details

The familiar HEMT device structure is that the source and drain electrodes are located on each side of the gate separately. When the HEMT is applied as a biosensor, the gate will be modified with biomolecules as the sensing district. During the modification and detection process, the gate needs to be dropped with a variety of solutions. In order to avoid short circuiting of the device, the source and drain electrodes are arranged far away from where the gate electrode is on one side. We investigated the effect of GaAs epitaxial structures and different device structures on the device electrical performance.



We firstly grow three kinds of epitaxial structures on 2-inch semi-insulating GaAs substrate. The first kind of HEMT epitaxial structure is labeled as A: 500 nm buffer layer/12 nm InGaAs channel layer/4 nm AlGaAs spacer layer/Si δ-doping layer (100 s)/28 nm AlGaAs barrier layer/30 nm cap layer; the second kind of HEMT epitaxial structure is labeled as B: 500 nm buffer layer/12 nm InGaAs channel layer/4 nm AlGaAs spacer layer/Si δ-doping layer (80 s)/28 nm AlGaAs barrier layer/30 nm cap layer; the third kind of HEMT epitaxial structure is labeled as C: 500 nm buffer layer/50 nm AlGaAs barrier layer/ Si δ-doping layer (40 s)/ 4 nm AlGaAs spacer layer/12 nm InGaAs channel layer/6 nm AlGaAs spacer layer/Si δ-doping layer (80 s)/28 nm AlGaAs barrier layer/30 nm cap layer. The difference between A and B is the Si δ-doping time. The special epitaxial structure of C is double-layer doped.



For each kind of epitaxial structure, we designed two types of device structures. The familiar HEMT structure where the source and drain electrodes are located on separate sides is named as the H-type structure. The special structure that source and drain electrodes are arranged far away from the gate electrode on one side is named as the U-type structure. Four different kinds of channel size are designed for each kind of device structure. Fig. 1 shows the microscope image of the device structure and details of the channel size. The electrical characterization of the HEMT biosensor was measured at 25 °C by the CHI 660D electrochemical workstation.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/12/PIC/18040001-1.jpg'"
class="figure_img" id="Figure1"/>



Download



Larger image


PowerPoint slide






Figure1.
(Color?online) The microscope image of device structure and details of channel size.

3.
Result and discussion

On a 2-inch epitaxial wafer of A/B and C, there are about 70 devices of the same device structure and channel size. We measure the source–drain current I_SD (mA) of each device under the same conditions. The source–drain voltage is 1 V, and source–gate voltage is 0 V (V_SD = 1 V; V_SG = 0 V). Based on the date, we calculate of the average and standard deviation of the current distribution. The statistical results are shown in Table 1.

	ISD (mA)	A (100 s)	B (80 s)	C (double-δ)
U-type	L = 40 μm; W = 50 μm	0.82 ± 0.05	0.27 ± 0.03	0.09 ± 0.01
	L = 20 μm; W = 50 μm	1.80 ± 0.12	0.60 ± 0.08	0.21 ± 0.02
	L = 40 μm; W = 200 μm	3.26 ± 0.31	1.16 ± 0.14	0.41 ± 0.02
	L = 20 μm; W = 200 μm	6.85 ± 0.72	2.45 ± 0.25	0.91 ± 0.09
H-type	L = 40 μm; W = 50 μm	0.55 ± 0.07	0.09 ± 0.01	0.08 ± 0.01
	L = 20 μm; W = 50 μm	1.04 ± 0.10	0.19 ± 0.03	0.17 ± 0.03
	L = 40 μm; W = 200 μm	2.81 ± 0.41	0.56 ± 0.14	0.39 ± 0.03
	L = 20 μm; W = 200 μm	4.55 ± 0.71	0.86 ± 0.14	0.80 ± 0.07

Table1.
The statistical results of the source–drain current I_SD (mA) of each device when the source–drain voltage is 1 V, and source–gate voltage is 0 V (V_SD = 1 V; V_SG = 0 V).



Table options
-->


Download as CSV

	ISD (mA)	A (100 s)	B (80 s)	C (double-δ)
U-type	L = 40 μm; W = 50 μm	0.82 ± 0.05	0.27 ± 0.03	0.09 ± 0.01
	L = 20 μm; W = 50 μm	1.80 ± 0.12	0.60 ± 0.08	0.21 ± 0.02
	L = 40 μm; W = 200 μm	3.26 ± 0.31	1.16 ± 0.14	0.41 ± 0.02
	L = 20 μm; W = 200 μm	6.85 ± 0.72	2.45 ± 0.25	0.91 ± 0.09
H-type	L = 40 μm; W = 50 μm	0.55 ± 0.07	0.09 ± 0.01	0.08 ± 0.01
	L = 20 μm; W = 50 μm	1.04 ± 0.10	0.19 ± 0.03	0.17 ± 0.03
	L = 40 μm; W = 200 μm	2.81 ± 0.41	0.56 ± 0.14	0.39 ± 0.03
	L = 20 μm; W = 200 μm	4.55 ± 0.71	0.86 ± 0.14	0.80 ± 0.07

For the HEMT devices used as biosensors, the device should possess larger current and better inter-chip consistency. The relationship between device current and channel size is in accordance with I ∝ W/L. Under the same device structure and channel size, the current of A epitaxial structure is greater than B and C in the statistical results. Taking U-type structure L = 40 μm W = 50 μm as an example, the current of A epitaxial structure is 0.82 mA, the current of B epitaxial structure is 0.27 mA, and the current of C epitaxial structure is 0.09 mA. The A epitaxial structure modulation doping time is larger than B, so the carrier concentration is higher. The structure of C has double-layer modulation doping and the upper layer is the same as B, but the current of B is larger than C. This result indicates that double-layer modulation doping structure is not conducive to enhance the current.



For A epitaxial structure with the largest current, we used the ratio of standard deviation (δ) to average value (μ) to assess the inter-chip current consistency. The smaller ratio means the better current consistency. There are three variables: device structure (H-type U-type), channel length, and channel width. We used a control variables method to study the effect of each variable on device current performance consistency. The result is shown in Fig. 2.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/12/PIC/18040001-2.jpg'"
class="figure_img" id="Figure2"/>



Download



Larger image


PowerPoint slide






Figure2.
(Color?online) (a) The comparison of ratio (δ/μ) of U-type and H-type device structure when the channel size is the same. (b) The comparison of ratio (δ/μ) of channel length when the device structure and channel width are the same. (c). The comparison of ratio (δ/μ) of channel width when the device structure and channel length are the same.




The comparison of the ratio (δ/μ) of standard deviation to average value under a different situation is shown in the above figure. When the channel size is the same, the ratio (δ/μ) of standard deviation to average value of the U-type device structure is less than that of H-type except for the conditions of L = 20 μm W = 200 μm. When the channel width and device structure are the same, the ratio (δ/μ) of standard deviation to average value of the L = 20 μm device is less than that of L = 40 μm. When the channel length and device structure are the same, the ratio (δ/μ) of standard deviation to average value of W = 200 μm device is better than that of W = 50 μm except for L = 20, U-type. The smaller ratio means the better current consistency. Based on the result and analysis, the current consistency of the U-type device structure is better than H-type, the channel length 40 μm is better than 20 μm, and the channel width 200 μm is better than 50 μm. Based on the above studies, the optimal epitaxial structure of the GaAs HEMT biosensor is the A epitaxial structure, the optimal device structure is U-type, and the optimal channel size is L = 40 μm and W = 200 μm.



For the A epitaxial structure, U-type device structure, L = 40 μm and W = 200 μm device, we investigated the distribution of device current with location on 2 inch GaAs wafer. The normalized current distribution with location is shown in Fig. 3.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/12/PIC/18040001-3.jpg'"
class="figure_img" id="Figure3"/>



Download



Larger image


PowerPoint slide






Figure3.
(Color?online) The normalized current of A epitaxial structure, U-type device structure, L = 40 μm, W = 200 μm HEMT device distribution with location on 2-inch GaAs substrate.




The color in the figure represents the magnitude of current. The white part of the circle shows the failure of the device. The above figure shows that the current increases from the inside to the outside on a 2-inch GaAs wafer. The closer to the edge of the wafer, the current is greater and the current consistency is poorer. In order to obtain HEMT biosensors with better current consistency for biosensing detection, the device should be located in the 1 inch diameter circle in the center.

4.
Conclusion

The influences of epitaxial structure, device structure, and device size on device current and consistency are studied. The results show that the performance of doping 100 s epitaxial structure is better the other two kinds of epitaxial structure. The comparison between B epitaxial structure and C epitaxial structure reveals that the double-layer doped structure has no significant effect for improving the device current. The double-layer doped epitaxial structure is not conducive to HEMT biosensor performance. The U-type structure is better than the H-type structure. When the channel size and epitaxial structure are the same, the U-type device current is greater than the H-type device, and the current consistency of the U-type is also better than the H-type. The relationship between device current and channel size is in accordance with I ∝ W/L. The current consistency of channel length L = 40 μm is better than L = 20 μm. The current consistency of channel width 200 μm device is better than 50 μm. The closer to the edge of the wafer, the current is greater and the current consistency is poorer. Based on the above studies, the optimal device of GaAs HEMT biosensors is the A epitaxial structure, U-type device structure, L = 40 μm, W = 200 μm, located in a 1 inch diameter circle in the center.