1.
Introduction

Modern power devices need to possess high voltage, fast recovery, low forward voltage drop and small leakage current characteristics^[1]. The P–i–N diodes are commonly used as freewheeling diodes in the high voltage regime. Because of the poor switching characteristics at high voltages, such as large reverse recovery peak current and long reverse recovery time, the application of P–i–N diodes in high-frequency circuits is limited^[2–4]. In addition, the conduction losses caused by diodes occupy a large portion of the switching losses in high-voltage switching devices (e.g., IGBTs, IEGTs, GTOs)^[5–7]. In order to obtain better soft recovery characteristics, the carrier concentration in the end of reverse recovery process can be adjusted by improving the cathode structure^[8]. To improve the speed of reverse recovery process, the injection efficiency is controlled by changing the anode structure to reduce the number of minority carriers during conduction^{[9, 10]}. Thus, a partial-heavily doped anode structure and cathode extraction structure in the fast recovery diode is proposed. The simulation and experimental results have shown that the reverse recovery characteristic and blocking characteristic have been enhanced. Furthermore, the forward characteristics of the AA–CE (Asymmetric Anode and Cathode Extraction) diode and the P–i–N diodes have been analyzed in detail.

2.
AA–CE diode structure

The cross sections of the proposed AA–CE diode and P–i–N diode are shown in Figs. 1(a) and 1(b), respectively. As is displayed in Fig. 1, the anode region of AA–CE diode consists of a partial-heavily doped P⁺ structure and a lightly doped P structure. N buffer layer is introduced in these two diodes. Other differences are cathode portion consisting of a heavily doped N⁺ structure and a P⁺ structure. In this paper, the physically based 2-D device simulation software ISE-TCAD has been used to analyze the structures^[11].






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-1.jpg'"
class="figure_img" id="Figure1"/>



Download



Larger image


PowerPoint slide






Figure1.
The cross section: (a) the AA–CE diode, (b) the P–i–N diode.




AA–CE diodes and P–i–N diodes are fabricated. Fig. 2 shows the abrasive staining pictures of the AA–CE diode and P–i–N diode. The device sizes and process parameters are as follows: the P⁺ anode depth is 5 μm, and its doping concentration is 1 × 10²⁰ cm^?3. The depth of P region in anode is 2 μm, and the doping concentration is 3 × 10¹⁷ cm^?3. Moreover, the N^? layer thickness is 325 μm and the doping concentration is 2 × 10¹³ cm^?3. The N buffer layer thickness is 50 μm and the doping concentration is 2 × 10¹⁶ cm^?3. Lastly, the cathode depth in both cases is 10 μm, the doping concentration of the cathode N⁺ area is 1 × 10²⁰ cm^?3. The depth of P⁺ region in cathode is 5 μm, and the doping concentration is 3 × 10¹⁷ cm^?3. The fabrication process of AA–CE diode is as follows: N^? wafer is initially used as the substrate. N buffer layer is formed by diffusion on the cathode side. Furthermore, the cathode extraction structure is formed by ion implantation and annealing technique, and then by diffusion on the anode side, the P layer is formed. Additionally, a P⁺ region ohmic contact is formed by a partial ion implantation. Finally, the electrodes are formed by evaporation of aluminum.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-2.jpg'"
class="figure_img" id="Figure2"/>



Download



Larger image


PowerPoint slide






Figure2.
The abrasive staining pictures of (a) AA–CE diode and (b) P–i–N diode.

3.
Results and analyses

3.1
Reverse recovery characteristics

Turning off the power diode from an on state to an off state is called a reverse recovery process. During the on state, a large amount of non-equilibrium carriers are stored in the diode. Those non-equilibrium carriers are released during reverse recovery and result in the reverse current. When a reverse voltage suddenly is added to the diode that works at the forward state, it takes some time for the diode to recover its block ability. In this case, the forward current continues to drop to zero and the internal carriers are rapidly extracted under the reverse voltage. Then current begins to reversely flow to form the reverse recovery current I_rr. When the reverse recovery current is continuously increased to the peak current I_rrm, the diode starts to recover its blocking ability until the reverse current begins to drop to its leakage value. Then a complete reverse recovery process is over. Reverse recovery time (t_rr) is defined as: the time starts when the reverse recovery current is zero to the time when the reverse recovery current is 0.25I_rrm. The diode with a higher reverse current drop rate is called the hard recovery diode. Its softness factor expression is S = t_f/t_s, as shown in Fig. 3. The larger the S value, the smoother the tail current. A hard recovery diode will produce high voltage overshoot and cause electromagnetic oscillation, which should be avoided in design.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-3.jpg'"
class="figure_img" id="Figure3"/>



Download



Larger image


PowerPoint slide






Figure3.
(Color online) Reverse recovery characteristics comparison of the AA–CE diode and P–i–N diode. V_DC = 1800 V, I_F = I_nom/5 = 32 A, L_s = 1.2 μH.




As can be seen in Fig. 3, the reverse recovery time and reverse peak current of AA–CE diode are significantly better than the P–i–N diode. Furthermore, the experimental parameters related to the reverse recovery characteristic of both diodes are presented in Table 1.

Variable	AA–CE diode	P–i–N diode (P–i–N)
Irrm (A)	?27.2	?52
ts (ns)	112	204
tf (ns)	160	132
trr (ns)	272	336
S	1.4	0.6

Table1.
The reverse recovery characteristic parameters.



Table options
-->


Download as CSV

Variable	AA–CE diode	P–i–N diode (P–i–N)
Irrm (A)	?27.2	?52
ts (ns)	112	204
tf (ns)	160	132
trr (ns)	272	336
S	1.4	0.6

Combining the data in Table 1, the reverse recovery current of P–i–N diode is showed by the gray curve in Fig. 3. On the other hand, the reverse recovery current of AA–CE diode is showed by the blue curve. The I_rrm in Table 1 represents the reverse recovery peak current. t_s represents the minority carrier storage time and t_f represents the reverse recovery current fall time. t_rr represents the reverse recovery time which consists of t_s and t_f. It can be seen that there is a long tail current in the reverse recovery curve of the AA–CE diode. Therefore, the soft factor has been greatly improved. Compared to the P–i–N diode, the soft factor of AA–CE diode is doubled. The reverse recovery time is reduced from 336 ns of P–i–N diode to 272 ns of AA–CE diode. The results show that the AA–CE diode peak current is ?27.2 A, while the P–i–N peak current is ?52 A. This asymmetric anode structure can reduce excess carrier concentration at the PN^– junction. The cathode extraction structure can effectively improve the softness factor. Therefore, the reverse recovery peak current and softness factor are well controlled. At the same test conditions, the AA–CE diode reverse recovery peak current and reverse recovery time are reduced for 47% and 20% respectively.



Fig. 4 shows the vertical distribution of the hole density at different time points in the reverse recovery process. A lifetime (τ_p0 and τ_n0) value of 1 μs was used during the numerical simulations. The reverse recovery process starts at t = 1.5 μs. The slope of the carrier profile becomes positive, as shown for the time t = 1.6 μs and t = 1.7 μs. In this time, the P⁺N^? junction is still forward based. The carrier concentration is well above the equilibrium value. This is the rising phase of the reverse recovery current. As shown for the time t = 1.8 μs, holes near the anode portion are pumped away by the electric field. The hole concentration is minimized at the cathode P⁺ region. At t = 2.0 μs, the hole peak density significantly reduced. The P⁺ area in the cathode will inject additional holes to the N buffer area. The reverse recovery current consists of the cathode P⁺ hole current and cathode leakage current. The tail current extends the current fall time (t_f), thus, a soft recovery process is finished^[12]. The N buffer layer can effectively change the impurity distribution on the cathode side and control the carrier extraction speed during reverse recovery process. It also reduces the N^?N junction electric field strength and decreases the carrier extraction rate, which results in a softer recovery characteristic^[13]. Then, the problem of hard recovery characteristics has been greatly improved.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-4.jpg'"
class="figure_img" id="Figure4"/>



Download



Larger image


PowerPoint slide






Figure4.
(Color online) The vertical distribution of hole density in reverse recovery process.

3.2
Blocking characteristics

The blocking characteristics of the AA–CE diode and the P–i–N diode are presented in Fig. 5. From the results presented in Fig. 5, it can be noticed that the P–i–N diode breakdown voltage is about 3600 V, while the AA–CE diode breakdown voltage is about 3900 V.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-5.jpg'"
class="figure_img" id="Figure5"/>



Download



Larger image


PowerPoint slide






Figure5.
(Color online) The reverse I?V characteristics of the AA–CE diode and the P–i–N diode.




The experimental results have shown good agreement with simulation results. In general, the diodes with a buffer layer possesses higher withstand voltage ability than the diodes without a buffer layer^[14–16]. The AA–CE diode has a 300 V higher breakdown voltage than the P–i–N diode. However, the cathode region P⁺ structure weakens the blocking characteristics. Under reverse bias, the breakdown characteristic analysis can learn from the open-base transistor breakdown characteristic. This voltage is given by: ${B_{
m V}} = frac{{q{N_{
m D}}W_{
m N}^2}}{{2{varepsilon _{
m s}}}}$ where N_D is the doping concentration in N^? area. W_N is the thickness of voltage sustainable layer. It is worth pointing out that this calculation is assumed as follows: when the depletion region extends through the entire N^? area, increasing the bias will produce the injection of minority carriers from the forward-biased junction and lead to the onset of high-current flow. The open-base transistor breakdown voltage is always lower than the avalanche breakdown voltage^[17]. Therefore, the design of cathode P⁺ region should be very careful.



The 2D and 3D electric field distributions of both diodes, at the reverse bias voltage of 3000 V, are presented in Fig. 6. As can be seen, the P–i–N diode electric field line near to the anode has a triangular shape. The peak electric field value of AA–CE diodes is higher than P–i–N diodes. But the electric field shows some spikes in the anode area, thus, the increment in breakdown voltage is limited. In addition, the P–i–N diode electric field peak value is 1.7 × 10⁵ V/cm. The AA–CE diode electric field near to the anode has an irregular shape. The AA–CE diode electric field peak appears at cathode P⁺ area with value of 4.6 × 10⁵ V /cm, which is 2.7 times larger than the P–i–N diode electric field peak value. The area of the AA–CE diode surrounded by electric field lines is larger than P–i–N diode. Therefore, the AA–CE diode breakdown voltage is higher than the P–i–N diode breakdown voltage.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-6.jpg'"
class="figure_img" id="Figure6"/>



Download



Larger image


PowerPoint slide






Figure6.
(Color online) The electric field distributions of the AA–CE diode and the P–i–N diode, at the reverse bias of 3000 V.

3.3
On-state characteristic

The on-state characteristics of AA–CE diode and P–i–N diode are shown in Fig. 7. Clearly, the AA–CE diode has a higher on-state voltage drop than the P–i–N diode.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-7.jpg'"
class="figure_img" id="Figure7"/>



Download



Larger image


PowerPoint slide






Figure7.
(Color online) On-state characteristics of the AA–CE diode and P–i–N diode.




The experimental on-state voltage drop at 100 A/cm² of AA–CE diode and P–i–N diode is 2.2 and 1.7 V respectively. In spite of this, the overall trade-off between on-state and reverse recovery losses will be better in the AA–CE diode. At the same forward current densities of 100 A/cm² and the same forward voltage drop of 1.7 V, compared with the P–i–N diode, the simulation results show that the reverse recovery peak current is still reduced by 30%, the reverse recovery time is shortened by 10% and the softness factor is increased by 30%. The on-state characteristics can be improved by optimizing the junction depth and width of the anode P⁺ region. On the one hand, as the width and thickness of the P⁺ region of the anode increase, the forward conduction characteristics become better, but the reverse recovery peak current increases and the softness factor decreases. On the other hand, by optimizing the thickness and doping concentration of the N buffer layer, the electric field intensity at the N^–N junction can be reduced, the carrier extraction rate can be delayed, and the softer recovery characteristic can be obtained. Taking the compromise of these two characteristics into account, we chose the optimal P⁺ region and N buffer structures parameters. In a word, the AA–CE diode possesses a better characteristics curve and a lower recovery loss than the P–i–N diode.



The carrier concentration distributions of the AA–CE diode and the P–i–N diode, when the on-state current density is 100 A/cm², are shown in Fig. 8. The electrons’ and holes’ lifetime in the P–i–N diode N^? and the AA–CE diode N^? area are 10 and 3 μs, respectively. As can be seen in Fig. 7, the high-level injected carrier concentration of anode and cathode areas is greater than the drift region doping concentration. The holes’ concentration (hDensity) and the electrons’ concentration (eDensity) of both diodes are equal in the drift region. Nevertheless, being close to the cathode region, the depletion layer extends to the N buffer layer because of the introduction of the P⁺ region, which causes decrease of the AA–CE diode carrier concentration.






onerror="this.onerror=null;this.src='http://www.jos.ac.cn/fileBDTXB/journal/article/jos/2018/5/PIC/17100013-8.jpg'"
class="figure_img" id="Figure8"/>



Download



Larger image


PowerPoint slide






Figure8.
(Color online) Carrier concentration distribution of the AA–CE diode and P–i–N diode.

4.
Conclusion

A technique of asymmetric anode and cathode extraction has been presented and the diode has been fabricated. A comparative study on the AA–CE diode and the conventional P–i–N diodes has been obtained. Both the simulation and experimental results have shown that the reverse recovery behavior and blocking characteristic of AA–CE diode is superior to the P–i–N diode. Moreover, the AA–CE diode not only has the high blocking voltage but also possesses the soft recovery characteristics. Although the forward conduction characteristics are degraded, this well-designed AA–CE diode is still able to reduce the dynamic power consumption, increase the soft factor and improve the breakdown voltage.