1. Introduction

The many exceptional properties (electronic, optical, mechanical, etc.) of synthetic diamond film grown by CVD techniques give it excellent features in terms of high radiation resistance, low leakage current, high temperature operation, high chemical inertness, and tissue-equivalence [1–4]. Thus, it has been expected that CVD diamond radiation detectors can be used for many applications in several fields, among which there is the realization of nuclear detectors which are being actively investigated (for example, the RD42 activity in CERN) [5]. The high resistivity of diamond permits a simple structure, and the very high band gap results in an extremely low number of free carriers leading to very low noise and power dissipation, while the high carrier mobilities and saturation field allow fast charge collection. Even more important is the extreme high radiation hardness and high temperature operation capability, since particle fluxes in the next generation of particle accelerators will be so high as to be beyond the operational limit of the conventional silicon based detectors.

In the past years, tremendous attempts have been made to develop CVD diamond α-particle detectors. However, more attention has been paid in the improvement of the detector performance. The polycrystalline nature of films makes the detector performance strongly depend on the microstructure of CVD diamond films [6]. In general, diamond films grown by CVD method had different micro-structural features between the final growth side and the nucleation side. Clarifying the microstructure of CVD diamond is helpful to understand both the basic device physics and the device application. In this paper, two types of detectors, one with a coplanar and the other with a sandwich geometry using an identical CVD diamond film, are fabricated in order to investigate the effects of the microstructure on the performance of diamond film α-particle detectors.

2. Experiment

CVD diamond films are grown on single crystal (100) n-type silicon substrates using a hydrogen-acetone precursor mixture in a hot-filament chemical vapor deposition (HFCVD) system. The resistivity of the silicon substrate is about 4–7Ω·cm. The usual scratching procedure is adopted to promote nucleation on the silicon surface. The deposition parameters in detail can be found in Ref [7]. A typical growth rate of approximately 1μm/h is obtained. The structure and surface morphology of the films are characterized by Raman spectroscopy and scanning electron microscopy (SEM). The diamond films, used in our experiments, has a typical microcrystalline structure with a grain size about several microns and a thickness about 50μm. Prior to detector fabrication, diamond films are treated in strongly oxidizing solutions (H₂SO₄+50%H₂O₂) for 30min to reduce greatly the surface conductivity of the material, and then annealed at 500°C in an argon atmosphere for 1h to get rid of surface hydrogen. This constitutes a critical step in the device fabrication, resulting in a magnitude reduction of device leakage current [8].

Figure 1 shows a_schematic diagram of the α-particle detector fabricated in our lab. The detector works with a coplanar geometry when a bias voltage is applied between A and B electrodes and with a sandwich geometry when a bias voltage is applied between A and C electrodes. The frontside interdigitated electrodes (shown in Fig. 2), with a width and inter-electrode gap of 150μm, are fabricated using conventional photolithography and dry etching techniques on the unpolished “growth” side of the film. Metal contact electrodes are realized using thermally evaporated Cr (50nm) followed by gold (100nm). The silicon substrate is used as the backside contact. To obtain ohmic contacts, the sample is annealed at 450°C in the Ar atmosphere for 45min.

The pulse height distribution (PHD) measurements are carried out on the diamond film detector under different electrode mode while irradiated with uncollimated ²⁴¹Am α-particle source. The α-particle irradiation is carried out through a pinhole, in the normal direction to the sample surface. The detector output is connected, through a charge pre-amplifier (Ortec 142IH) and a shaping amplifier (Ortec 575A) with a 1.5μs shaping time, to a multichannel analyzer (Ortec Trump-PCI-2K). Keithley 4200-SCS is used to measure the bias voltage-dependent photocurrent characteristics of these detectors under α-particle irradiation.

 

Fig. 1. The schematic diagram of the α-particle detector.

Download Full Size | PDF

 

Fig. 2. The optical microscopic image of detector electrode structure patterned on the unpolished “growth” side of the diamond film.

Download Full Size | PDF

3. Results and discussion

The surface morphology of the diamond film observed on the entire 2×2 cm² surface indicates a typical microcrystalline structure (as shown in Fig. 3) with a gain size of about several microns. It also exhibits (111) facets crystal. The cross-section of the film indicates a distinct columnar structure that is always observed in CVD diamond growth and is an important parameter to identify the morphology of the film. The film thickness is about 50 μm, confirmed by SEM.

 

Fig. 3. SEM image of CVD diamond film used for α-particle detectors.

Download Full Size | PDF

 

Fig. 4. Raman spectra for the final growth side and the nucleation side of the CVD diamond film.

Download Full Size | PDF

Raman spectroscopy, a powerful technique, is widely used not only to detect graphitic carbon contained in CVD diamond, but also to estimate the quality of diamond. Figure 4 shows the typical micro-Raman spectra of the diamond film in the final growth side and in the nucleation side. One can find that the full width at half maximum (FWHM) of the diamond line in the nucleation side is larger than that in the final growth side. The non-diamond carbon phase including amorphous carbon and defects such as dangling bonds at the boundaries between the diamond grains can broaden the diamond peak. By considering the sensitivity of Raman signal for non-diamond carbon phase is about 75 times of that for diamond, Raman spectroscopy can be used to estimate the non-diamond carbon content in the film by the expression, C_nd=1/[1+75(I_d/I_nd)], where I_d is Raman peak intensity for diamond crystals, and I_nd is the Raman peak intensity for a non-diamond carbon phase. The Raman results show that the non-diamond carbon content C_nd.in the nucleation side is about 0.89% and is higher than that in the final growth side, 0.64%.

The pulse height distribution (PHD) of the two types of CVD diamond detectors is measured at room temperature. The detector with coplanar geometry is found to have a similar PHD to that with sandwich geometry (shown in Fig. 5). However, the more counts at the 5.5MeV²⁴¹ Am full-energy peak can be obtained from the detector with coplanar geometry (shown in Fig. 6).

 

Fig. 5. Pulse height distributions obtained by the CVD diamond detector with different device geometry exposed to α particles.

Download Full Size | PDF

 

Fig. 6. Counts and average charge collection efficiency plot of diamond-based alpha-particle detector with coplanar and sandwich electrode geometry respectively.

Download Full Size | PDF

According to Hecht’s theory, the charge collection efficiency η, which is defined as the ratio of the electrical charge (Q _ind) induced in the external circuit to the charge (Q ₀) generated by a particle in detector, is given by η=Q _ind/Q ₀=μτE/L, where μ and τ are the carrier mobility and lifetime respectively, E is the mean internal electric field, and L the interelectrode spacing. The average collection efficiencies as a function of the applied electric field from both types of the 50μm thick devices are also shown in Fig. 6. For both types of diamond detectors, the characteristic increasing curves with a tendency to saturate at high electric field values are found. At higher field strengths the increase of η flattens due to increased phonon creation and the resulting decrease of the carrier mobility.

We also find that the charge collection efficiency is different for the two types of the detectors fabricated using an identical diamond film. The detector with coplanar geometry has higher charge collection efficiency η than that with sandwich geometry. Under an applied electric field of 66.7kV/cm for the coplanar geometry and 200kV/cm for the sandwich geometry, the average charge collection efficiency is 42.9% and 37.4%, respectively. The photocurrent responses of two CVD diamond detectors to 5.5 MeV ²⁴¹Am α particles are given in Fig. 7, where the photocurrent denotes the net photocurrent which is subtracted the dark-current from the total current. The photocurrent almost proportionally increases with voltage, due to the proportionally linear relationship between the collected carriers and external electrical field, conformed to a simple photo-generation and collection model where the photocurrent can be written as I _ph=qF ₀ η _abs η, where q is the electron charge, F ₀ the number of incident photons per unit time and η _abs the optical absorption efficiency. It also can be seen that the photocurrent of the coplanar structure detector is higher than that of the sandwich structure detector. At a bias voltage of 30 V, the photocurrents obtained by above two types of detectors are about 1.08 nA and 0.80 nA, respectively.

 

Fig. 7. The I-V curves of the detector with different electrode geometry under various applied bias voltage.

Download Full Size | PDF

 

Fig. 8. The ratio of the full width at half maximum (FWHM, ∆E) to the 5.5MeV ²⁴¹Am full-energy peak (E) of the pulse height distribution of the detector with different electrode geometry under various applied electric field.

Download Full Size | PDF

This difference of the charge collection efficiency and the photocurrent is believed to result mainly from the different structural imperfection distributions between the final growth side and the nucleation side of the diamond film. The SEM observation exhibits that the grain boundary density is lower at the final diamond growth side than that at the nucleation side. The Raman analyses (see Fig. 4) indicate that the non-diamond phase is fewer in the final diamond growth side than that in the nucleation side. In other words, more structural imperfection presented in the nucleation side. However, the grain boundaries, the sp2-bonded carbon impurities and other defect centers could act as trapping centers or recombination centers in the process of the carrier drift. When α particles irradiating CVD diamond films, free carriers (electron-hole pairs) are generated, move towards and then are collected by the relevant electrodes at external electrical field. For the coplanar geometry detector, the electrodes are only fabricated on the final diamond growth side. Under α-particle irradiation, the created free carriers of the electrons or holes could more easily travel to the electrodes under bias since less structural imperfection presented in the final diamond growth region, leading to fewer trapping and recombination centers there, resulting in longer lifetimes τ of the carriers and therefore higher charge collection efficiency η. Whereas the electrical contacts are fabricated on both the final growth and the nucleation sides for the sandwich structure. When positively biased the electrode at the nucleation side, the created electrons have to pass through the highly defective region near the nucleation side before they are collected by the electrode, in the process of electron drift, the recombination and trapping of the electrons must happen more frequently, decreasing the lifetimes of the electrons and their drift distance, resulting in a lower charge collection efficiency η.

The energy resolution (ϕ), expressed by the ratio of the full width at half maximum (FWHM, ΔE) to the 5.5MeV ²⁴¹Am full-energy peak (E) of the pulse height distribution, of the CVD diamond detector operating under different electrode modes is also investigated. The dependence of the ΔE/E as a function of the applied electric field is shown in Fig. 8. The characteristic decreasing curve with a tendency to saturate in high electric field values is found. The energy resolution of the two types of detectors is significantly improved_with the applied electric field from 6.67kV/cm to 200kV/cm, which corresponds to carriers accelerated and collected by electrodes quickly. To make carriers collected entirely, higher applied electric field should be applied. Under α particle irradiation the detector with sandwich geometry has a similar trend on energy resolution with coplanar geometry. It is obvious that the energy resolution of the coplanar detector is better than that of the sandwich detector, which may result from the different micro-structure feature between the final growth side and the nucleation side of the diamond film. A good energy resolution of 1.1% is achieved at 66.7kV/cm for the coplanar geometry and 200kV/cm for the sandwich geometry.

4. Conclusions

Two types of detectors, one with a coplanar and the other with a sandwich geometry using an identical CVD diamond film, are fabricated in order to investigate the effects of the film microstructure on the performance of diamond film α-particle detectors. The pulse height distribution (PHD) and photocurrent measurements indicate that the diamond film detector with coplanar geometry irradiated with ²⁴¹Am α-particle has higher values on counting rate, charge collection efficiency and photocurrent than that with sandwich geometry. Under an applied electric field of 66.7kV/cm for the coplanar geometry and 200kV/cm for the sandwich geometry, an average charge collection efficiency of 42.9% and 37.4% is obtained, respectively. Scanning electron microscopy and Raman scattering studies directly demonstrate that the different counts, efficiencies and photocurrents of the two types of detectors mainly resulte from the different micro-structural features between the final growth and the nucleation sides of the diamond film. Under α particle irradiation the detector with sandwich geometry has a similar trend on energy resolution with coplanar geometry. When the applied electric field is 66.7kV/cm for the coplanar geometry and 200kV/cm for the sandwich geometry, a good energy resolution of 1.1% is obtained for two types of the diamond detector. Based on the above results, the coplanar geometry is considered as a good electrode mode for good detector performance.