Identification and Analysis of Hazardous Materials Using Optical Spectroscopy

This investigation is aimed towards using optical spectroscopy for remote identification and quantitative analysis of hazardous substances for safety and security applications. We introduce a new model employing portable photosensor devices that are based on the double-barrier and vertically placed silicon structure, for such applications. The different absorption depths of individual waves allow us to carry out their spectral selection using an algorithm developed for this specific objective. We tested the proposed model on experimental Ag-p-Si-n-Si structures. The algorithm is developed for the spectral analysis without the preliminary calibration. The low dark currents (several dozens of pA) permit us to carry out the spectral analysis of the integral flux of the electromagnetic radiation of low intensity. The quantitative data from light current-voltage characteristics allow us to obtain an intensity distribution spectrum characteristic of the material by using red LED and the green laser. The results of this investigation divulge new possibilities for the creation of a new type of the portable semiconductor spectrophotometer and due to its stand-off detection capability, offer potential pathways to evaluate hazardous substances.


Introduction
In recent years, increasing demand for the stand-off detection/sensing for hazardous chemical has significantly increased due to their use in safety, security, environmental protection and detect suspicious packages, wire, fragmentation materials, or other signs of improvised explosive devices (IEDs) or crisis situations. Standoff techniques enable detection of threats without contact. The possibility of contamination (for operators and/or equipment) is thus avoided, and the need for subsequent decontamination is eliminated. In addition, standoff detection of explosives allows identification of threats from safe distances, i.e., outside the blast zone. There are now ongoing efforts to extend these methodologies to even biological threats. It is, hence of paramount importance to increase the efficiency of such methods and solution pathways which will respond quickly to such crisis situations. It is particularly important to create user-friendly sensors platform which will likely enable the remote analysis of the different materials in different media.
Remote spectrophotometric sensors are of special interest these days, as they provide us with necessary information on the composition of the mediupm under investigation and hence solve the omnipresent problem of identification from the security stand-point [1] [2] [3] [4] [5]. Nevertheless, the information received from studies of natural objects by existing means is not satisfactory from the point of their quantitative and qualitative analysis. There is a significant gap in terms of dispersion of the spectral data of the identical objects even in the advanced methods of investigation of the optical spectral analysis, which often excludes correlation between the measurement results. The processing of the spectral data is obtained by quantitative analysis and the reliability of such results depends directly on the parameters of the device in use.
It is critical that several sensors are required for monitoring large areas. To optimize cost and to strategically place sensors in regions which are likely to provide a fair assessment of the contamination scenario, it is urgent to develop; a): inexpensive and small-size sensors with high spectral sensitivity, suitable for remote field identification, and, b): appropriate algorithms for accurate registration of data measured by these sensors. In methods that are currently available for spectral analysis, the spectral distribution of the electromagnetic radiation is obtained by light filters, a prism, a diffraction grating and high accuracy mechanical devices [6] [7] [8]. The spectrophotometric systems like this are not multi-purpose and require additional devices and external software support for the fulfillment of every new function. Thus, they are rather expensive and are not suitable for field deployment.
The effective solution to the above-stated problems is by development of a semiconductor structure in which the electronic processes will provide high accuracy spectral analysis of the electromagnetic radiation. Several researchers have conducted study using multi-colored photoreceivers [9]- [17] in which multilayer structures or a cascade chain of active layers with different base thicknesses are used. In such photoreceivers, the different penetration depths of individual waves of the radiation provide different degrees of photoconductivity.
The mathematical modeling of the measurement results provides information on the spectral distribution of the intensity. The registration accuracy of structures mentioned above is essential due to the necessity to having identical ab-sorption conditions with nanometric accuracy for multiple photodiodes and multilayer structures. The complexity involved during fabrication of photodiodes and multilayer structures during their development and the difficulty in controlling the spectral sensitivity by the external voltage impedes their development and usage. The present investigation offers a viable means of high level of accuracy during spectral analysis with the help of the electronic processes in n + -p-n + structures.

Device Structure
The structure under investigation is a silicon structure, as prepared using standard integrated circuit processing methods. The structure consists of an n + -p-n + structure ( Figure 1). The p-base is occupied by the depleted layers of two barriers ( Figure 1). The current through the structure is the difference of the cur-   Since a qN ρ = , we obtain, where a N is the concentration of acceptor impurities, ε is the relative dielectric permeability of the substance, 0 ε is the dielectric permeability of the vacuum, and q is the electron charge.
The boundary conditions for the given equation are If both barriers are identical, then 2 1 0 ϕ ϕ − = . Consequently: With the help of Equation (5) For that, we first determine the total photocurrent flowing through the structure. In the investigated photodetectors, the photocurrent has diffusion and drift components. To determine the diffusion photocurrent, it is necessary to find p n density of minority charge carriers-holes in the n-semiconductor. For that, the following one dimensional diffusion equation is solved [18], Khudaverdyan et al. [19] obtained an expression for the drift currents gener- Taking into consideration Equations (8), (9) and (7), the expression for the total current flowing through the structure is: At normal functioning, the term that contains n p0 is much smaller than the first term. Thus, it may be neglected, and Equation (10) is: When irradiated by the integral flux (e.g. of the Sun), Equation (11) for the photocurrent is: where (i = 1, 2, 3, …) changes in the integral flux with the change of the emission wavelength, and (j = 1, 2, 3, …) changes with the change of bias voltage, F 0 (λ i ) is the total flux of incident photons with the wavelength λ i . Since the width of the n-region in Equation (11) was less than L p , the value of L p in Equation (12) was replaced by W ( Figure 1). Using algorithm described below, it is possible to register a separate wavelength λ i and its intensity by changing the voltage ∆V, thus obtaining the spectral dependence for wave intensities. In the present investigation, we used the AM0 spectral distribution for Si using radiation of LED (L-813SRC-J4) with a dominant wavelength 660 nm and a laser with the radiation wavelength of 530 nm. Since, with increase of the bias voltage, the registration region of the rear barrier widens towards the surface [19] [20], it allows us to carry out the sequential registration of the waves from depth to the surface.

Result of the Experiment
To determine the spectral composition of an integral flux of the electromagnetic radiation, the wave intensities of the radiation are equally divided so that the absorption depth corresponds to the width of the registering environment (~4 -6 μm). The photocurrent corresponding to the largest possible value of x is normalized by the most deeply penetrated wave. Under these conditions, the process of the selection of separate waves and their intensities from the integral flux of radiation can be described as in the following.
Assume that the leading indicator signal is the photocurrent. We also assume that the largest values of x m1 and x m2 of x m correspond to the change of bias vol-tage 1 mV, and the relevant photocurrents I 1 and I 2 . They are obtained only from the absorption of the most deeply penetrated wave λ i . With the help of these photocurrents, and using Lambert's law on radiation absorption in the homogeneous environment, we will have the coefficient of wave absorption,  (13) with the corresponding program, we can determine the length of the wave for the initial material of the photodetector, e.g. silicon. By means of the formula for the photocurrent (11), we can obtain the intensities of separate waves in the absorbed radiation as, With the help of Equations (5), (13) and (14), we determine the absorption coefficient of the most deeply penetrated wave, the wavelength and the intensity of the wave. Then, with the help of the expression (12), considering the absorption surface, we form the dependence I 1 = f(V) and subtract it from the experimental dependence I = f(V) of the summed current. As a result, we obtain a new dependence of the summed photocurrent on V, without the dependence I 1 = f(V).
By means of relevant software, this method helps us determine the lengths and the intensities of all the waves in the radiation. The experiments were carried out on the n + -p-n + structure. The cross-section of the structure with the parameters of the layers is presented in Figure 3. The spectral region of the monochromator is from 400 nm up to 1150 nm. Based on these parameters, the design height of the rear barrier was φ к = 0.89 eV, and the distance of Fermi level from the valence band top of the p-epitaxial layer was E F − E V = 0.29 eV ( Figure 4). The height of the silicide Schottky barrier (p-silicon with Ag) was 0.54 eV [18], and, consequently, the height of the surface barrier was ϕ c = 0.25 eV.
The photocurrents of the two oppositely directed barriers are oppositely directed, and they compensate each other depending on the applied voltage. It is noticeable in the relative spectral characteristics ( Figure 5). The figure shows the mirror image of the long wave photocurrent. In the absence of the bias voltage ( Figure 5, curve. 1), the spectral photocurrent is mainly governed by the surface potential barrier, which creates a short-wave peak.
The long-wave peak has small height and is governed by the rear barrier, which is located at the depth of 6 µm from the illuminated surface (Figure 3).
When the bias voltage is "+" on the Schottky contact, the surface barrier is shifted in the reverse direction, and the barrier of the p-n junction is shifted in the forward direction, which increases the photocurrent of the first barrier ( Figure 5, Curve 2). When the bias voltage sign is "−", the rear barrier is shifted in the reverse direction and the spectral photocurrent is governed by its increase    (13) and (14), the current-voltage characteristic data in the dark and under illumination, and the developed algorithm [21], we modelled the process of receiving of the spectral distribution of the radiation intensity. As a result, the received spectral curve (Figure 8) is close to the stated curve of the red LED and the laser with the radiation wavelength of 530 nm (Figure 9).