Sequential determination of free and total cyanide by flow injection

This study presents a flow injection system for the sequential determination of free (CN-) and total (CN-+ HCN) cyanide using a potentiometric method which employs two different processes for the determination of these two chemical species. The first process is based on direct detection of CN- using an ion-selective electrode for cyanide. In the second process, the sample is mixed with acid, and the released HCN is transferred across a PTFE membrane. The flow system employs three solenoid valves, a gas diffusion chamber, an ion-selective electrode, a potentiometer, and a computer with an AID conversion card. A Turbo Pascal® computer program automatically performs all the steps involved in data acquisition and processing. The standard deviation for the results obtained with the proposed method was 0.5%.

This study presents a .flow injection system for the sequential determination offree (CN-) and total (CN-+ HCN) cyanide using a polentiometric method which employs two different processes for the determination of these two chemical species. The first process is based on direct detection of CNusing an ion-selective electrode for cyanide. In the second process, the sample is mixed with acid, and the released HCN is transferred across a P TFE membrane. The flow system employs three solenoid valves, a gas &.sion chamber, an ion-selective electrode, a potentiometer, and a computer with an AID conversion card. A Turbo Pascal computer program automatically performs all the steps involved in data acquisition and processing. The standard deviation for the results obtained with the proposed method was 0.5%. different technique tbr determination of total cyanide [9]. In this paper, we present an automated flow system which is capable of sequentially determining both free and total cyanide in aqueous samples. For the determination of free cyanide, we employ a technique of direct flow injection with potentiometric detection through an ion-selective electrode (potentiometric flow-through detection). For determination of total cyanide, we employ an acid-flow injection technique, followed by permeation across a gas-diffusion membrane and posterior potentiometric detection (gas-diffusion flow injection potentiometry).
In the system being proposed, these two techniques are implemented concurrently in a single flow system which employs three three-way solenoid valves, a gas diffusion chamber, an ion electrode for cyanide, and a reference electrode. This system performs the sequential determination of free and total cyanide using two distinct analytic curves.

Introduction
Numerous works based on flow injection analysis (FIA) have been published lately [1]. Many of them [2][3][4] describe the use of loop-based injectors to introduce the sample into the carrier stream. A common way of changing the volume of the sample being introduced into the flow system is to change the length of the sampling loop. In other cases, three-way solenoid valves are used lbr sample introduction, and the system is triggered either by a built-in electronic circuit [5] or microcomputer [6]. Usually, in methods that employ manual injection, signals coming from the detection system are sent directly to a printer, generating a real-time diagram. In automated methods, a microcomputer recieves, processes and records the signals.
Many projects have been developed to monitor cyanide in waste waters [7]. One of the most commonly employed techniques is direct potentiometry with an ion-selective electrode (ISE) [8]. The classic method for determination of total cyanide in effluents is distillation, to eliminate interference, followed by colorimetric detection. This is an onerous method, since it requires expensive reagents, as well as several hours of laboratory work. Besides, it is used only tbr determination of low levels of cyanide, and it has the additional disadvantage of not being easily automated. Finally, this method does not allow quantification of free cyanide in the original sample.

Solutions
All solutions were prepared with distilled or deionized water. For the cyanide stock solution (1.0 x 10-1moldm-), we weighed 0.65g KCN. This was diluted in 50 ml water; ml NaOH tool dmwas added, and the volume was completed to 100ml. From this solution, we prepared analytic solutions of 1.0 x 10-5, Description of the process Figure shows the flow chart for the determination of free and total cyanide. In the process of free cyanide determination, valves V2 and V3 are triggered simultaneously, so that the sample is directly injected into the carrier stream. Triggering of valve V3 causes interruption of the carrier stream. Triggering of valve V2 causes the sample to be introduced through junction C1. Therefore, when valves V3 and V2 are triggered, a sample zone is formed, whose volume depends on the triggering time interval of valves V2 and V3.
For determination of total cyanide, only valve V1 is triggered. The sample flows together with a sulphuric acid solution through junction C2. This causes liberation of HCN, which crosses a PTFE membrane in the gas diffusion chamber (CD). After having crossed the membrane, at the other side of the chamber, HCN is converted into CNby the carrier solution (NaOH solution). The newly formed CNis determined by the same electrode in both processes. To quantify the two species of cyanide in a sample, it is necessary to produce two distinct analytic curves, one for the determination of free cyanide, and the other for determination of total cyanide.

Automation process
The electric circuit is described in figure 2 drive triggers the solenoid valves (Vs). The signals coming from the electrode system are read by the potentiometer and sent to the analogic/digital conversion card, and then acquired by the software. At the same time, a data file is created and the diagram appears in the computer screen in real time.
The program to control triggering of the valves, data acquisition and mathematical treatment of the data was developed in Turbo Pascal (R) 7.0 [6], as shown in the flow chart presented in figure 4. The basic reading and conversion routine performed by the AD card was developed in Assembler, and allows up to 20 000 readings per second. The time interval for valve opening can vary from msec to min. Throughout the experiment, after each reading, the converted value appears in the computer screen, and the final recording of all values appears as a diagram. Two data files (ASCII), one containing the diagram, and another containing the maximum values for each injection, are stored in the computer disk for later mathematical treatment. System performance assessment To assess system performance, we determined free cyanide, and HCN in synthetic samples, with different pH values and with the same total cyanide concentration, which was fixed as 10 -4 mol dm-3. The different pH values were used to produce a variation in the ratio between CNand HCN concentrations. For that, we considered the HCN pKa in an aqueous solution (pKa =9.14). The pH values necessary to obtain [CN-]/[HCN] rates equal to 1/100, 1/10 and 10/1 were calculated using equation (1): (1) The calculations revealed that pH(/00) 7.14; pH(/0) 8.14 and pH(10/l) 10.14.
These same samples were submitted to determination using our proposed method and the traditional potentiometric method, so that results could be compared. HCN concentration was indirectly determined in both methods, since this value was obtained from the difference between total cyanide concentration and CNconcentration.
To compare the precision of the two methods, all determinations were replicated six times. To perform the determination of the two species, HCN and CN-, using the method proposed herein, we used three consecutive injections. The potential value taken into consideration was obtained from the average among the three resulting peaks. The injections performed in triplicate were also used to plot the two analytic curves.

Results and discussion
Baseline stability and injection reproducibility are shown in figure 5, which represents the diagrams for potential versus time for the two processes carried out in the proposed method. After statistical evaluation, we obtained the following functions in relation to the analytic curve for processes and  Time (see) Figure 5. Diagrams of the two processes, obtained with five standard solutions and injections performed in triplicate.
their respective correlattion coefficients rl =0.998 and r2 0.996. The lower detection limit was approximately 1.0 x 10 -5 moldm -3 for both processes. The injection system described in figure presents a high level of stability and repeatability of the measurements, and can be applied for cyanide determination of industrial waste waters. Tables  1-3 show the results obtained in the experiments performed for assessment of equipment performance.
The comparison between CNand HCN concentrations as determined by the two methods reveals good agreement in terms of the results for the three rates under study (tables 1-3). The percentile difference between these two methods can be seen in the last column of tables 1-3. For the three rates there was positive deviation in relation to CNconcentration, and negative deviation in relation to HCN concentration.
However, this deviation is not very significant, since, with the exception of HCN determination using the 10[ rate, the deviation values are smaller than the standard deviation of the results. This suggests that the proposed method has a high degree of precision in comparison to   Table 3. CNand HCN concentrations at an approximate rate oflO/1 (pH= 10), determined with the traditional potentiometric method and with the proposed method (replicated six times for each method the stationary method. Data analysis using the t-test resulted in a 95% confidence level.

Conclusions
The main characteristic of the analytical procedure being proposed here is operational simplicity, since the sensor used for detection consists merely of a cyanide-indicator electrode.
The system responds efficiently and quickly, has a good level of repeatability (r.s.d. <0.5%, n--6), allows performance of 60 measurements per hour, and has good baseline stability (+1 mV). Also, our method is not significantly influenced by temperature; there is no need to temperature control if it does not vary significantly during measurements. The response linear range is wide (frown 1.0 x 10 .5 to 1.0 x 10moldm-3), with a lower detection limit of 8.0 x 10 .6 mol dm-3. The computer program necessary for data aquisition and data treatment can be obtained fi'om the authors. These include source code write in Turbo Pascal 7.0 and executable codes for use with IBM-compatible computers.