Some initial testing/proof-of-concept was performed in order to determine if potential individual DIY components of a spectrophotometer could indeed perform to a sufficient standard to enable a functional outcome when operated in conjunction. These individual functional components where grouped/tested as follows (and are discussed below in the corresponding sub-sections):
- CD/DVD as a diffraction grating
- DVD will give wider dispersion .: better resolution
- Stepper Motor - to control diffraction grating angle
- Small step size/control required to produce good resolution
- LDR as detector
- In conjunction with light source type/strength, entrance/exit slit width, gives sensitivity
- RGB LED/photoresistor as a "filter photometer"
Diffraction Grating (CD/DVD)
A central component of a spectrophotometer is the diffraction grating used to produce the monochromatic light from the incident beam. As noted (3), (4), (5) a CD or DVD can be used as a diffraction grating. A CD has groove spacing of 1.6 µm whereas a DVD has groove spacing of 0.74 µm.
Table 1 in the Project Background section reports the expected diffraction angle for various wavelengths of light versus incident light angle to the normal of the diffraction grating (i.e. CD). A quick prototype/test of concept was developed to check if "monochromatic" light could indeed be generated that would be suitable as the basis of a spectrophotometer. This involved a 350mA white LED as light source (with a wall-wart power supply to a LM317 in constant current mode), a sliver of CD as the grating, and a piece of white paper as the detector screen (see Album 1 in the Photographs Section).
The incident white light (passing through a slit formed from old razor blades) was indeed split into the expected "rainbow" with the various colours at the expected angles. Therefore, by selectively rotating the grating (e.g. with a stepper motor or similar) a desired wavelength of light can be directed into a sample solution. Visually, the spectrum obtained was "bright" to the unaided eye (the photographs were taken with a 1/4 shutter speed, F=3.5) giving some confidence that the intensity will be sufficient for later transmission/adsorption/detection through a sample solution.
As expected from the literature/how diffraction gratings work, the 1st order spectrum is much "brighter" than the 2nd order diffraction spectrum. Although the 2nd order spectrum has a wider dispersion angle (which would give better resolution). However, the DVD (with closer groove spacing) compared to the CD gives similar dispersion with the 1st order as does the 2nd order with a CD. Therefore, the DVD with a 1st order diffraction spectrum would be the choice.
The 350mA white LED was replaced with a WS2812 RGB LED as the light source. The WS2812 contains three seperate LEDs each producing a narrow band of wavelengths (Red 620-630nm, Green 515-530nm and Blue 465-475nm) - see photograph 7 in Album 1 in the Photographs Section. The idea being that the LED itself already produces "monochromatic" light (of 10nm bandwidth in three regions of the visible spectrum) and the diffraction grating (and associated stepper motor etc) could then be eliminated.
In effect, the WS2812 would provide the equivalent function of the "filters" in a filter photometer. The potential advantage being much simpler construction/fewer components/lower cost/easier setup. However, disadvantages and possible limitations that may prove the idea infeasible include:
- only the three wavelengths are available
- depending upon the colorimetric reactions/analytes of interest this may or may not be significant
- the WS2812 can only provide a maximum of 20mA to an individual LED
- this may not be of sufficient intensity to enable sufficient sensitivity
- the individual LED's in the WS2812 component are physically offset (although much closer than "normal" RGB LED's)
- this means the light paths may need to be aligned, complicating construction somewhat
- if alignment for WS2812 is needed, using a "normal" RGB LED may be advantageous, as higher currents can be used (increasing light intensity, improving sensititivity)
Testing the WS2812 with a CD as a diffraction grating, produced the expected diffraction angle for the various wavelengths as per equation 1 in Project Background Section. This was done to test the CD "diffration grating" with known wavelengths, rather than just the white light source. The following video was produced from still images of operating the WS2812. Note, the WS2812 needed to be located a distance from the slit in order to "align" the light beams from the individual LEDs within the WS2812 (i.e. the WS2812 was sufficiently far away from the slit that it formed a diffuse source).
While a RGB LED shows some promise as a light source in a pseudo "filter" photometer, since the LED's within a WS2812 need to be "aligned" to the entrance slit, this means no particular advantage over a "normal" RGB LED. In fact, a normal RGB LED can be operated at higher current (possibly overcoming the limitation of needing to be situated at a distance from the slit to produce a "diffuse" source so all three individual LEDs within the RGB LED are "aligned") and the necessary control circuitry and software is much simpler.
Information in the Background Section explains the requirements and operation of a stepper motor to alter the incident light angle to the DVD "diffraction" grating in order to produce and direct monochromatic light through a test solution.
A rough working prototype was constructed (see Photographs Section Album 3) to initially test the feasibility of the proposed arrangement/components for a DIY spectrophotometer. Previous testing with a WS2812 RGD LED showed, as expected, that a DVD/CD does indeed diffract light and from the known relationship of wavelength and incident light angle, this diffraction angle can be used to select/determine/calibrate particular wavelengths of light to be used. However, in order for this to be useful in the DIY spectrophotometer, reproducible and fine adjustment to the angle of DVD/CD diffraction grating will be required, and additionally, the monochromatic light produced will need to be of sufficient intensity for the photoresistor to produce a reasonable response. A photoresistor was selected as a potential detector (rather than phototransistor or photodiode which would be more sensitive) because this component was available at the time.
The prototype consisted of:
- A scrounged stepper motor from a disused printer, labelled as manufactured by Shinano Kenshi, model STH-39C013, 0.9o step and 21ohm, with a SN754410 stepper motor driver in "half step drive" mode;
- DVD as a diffraction grating (groove spacing of 0.74µm) with an exit slit of 1.0mm width at a distance of 125mm from the diffraction grating (calculated to give approximately 10nm bandpass with above stepper driver step size);
- GL5516 photoresistor (light resistance 10Lux 5-10Kohm, dark resistance 0.5Mohm) with a 220Kohm resistor to form a voltage divider with 2.5V voltage reference to a PIC 18F248 on board ADC via LM358 opamp as buffer (see Schematics Section);
- White LED operated at 350mA as light source.
- Firmware on PIC18F248 to operate stepper motor, ADC for reading the photoresistor voltage, and RS232 to PC for data transfer/collection
RGB LED Photometer
Previous Sections discussed the use of a WS2812 RGB LED instead of a white LED in conjunction with a diffraction grating. This would produce the equivalent of a "filter" photometer (i.e. a limited bandwidth within a limited portion of the spectrum rather than continuously variable selection of wavelength). Further, instead of using the WS2812 (which has a rather complicated communications protocol) subsitute a "normal" 4-pin RGB LED, which also enables the possibility of having greater light intensity than the WS2812 (which is limited to a maximum of 20mA). The advantage of such an arrangement is simplified construction with no moving parts (i.e. likely more robust). The disadvantage would be the limited wavelengths available. A rough prototype was constructed (see Photographs Section) and some initial testing performed using coloured food dyes.
A common-cathode RGB LED was used with the following specifications:
|RGB LED Specifications|
|Red||Forward Voltage|| ||1.88||2.2||V|
| ||Luminous Intensity||1000||1200|| ||mcd|
| ||Peak Emission Wavelength|| ||620||630||nm|
|Green||Forward Voltage|| ||3.0||3.2||V|
| ||Luminous Intensity||800||1000|| ||mcd|
| ||Peak Emission Wavelength||515||517.5|| ||nm|
|Blue||Forward Voltage|| ||3.0||3.2||V|
| ||Luminous Intensity||600||800|| ||mcd|
| ||Peak Emission Wavelength||460||462.5|| ||nm|
|Peak Current 100 uA, specifications Itest = 20 mA|
A LM317T was used in "constant current mode" (see Schematics Section) as a power supply, with a GL5516 CdS photoresistor as detector. The GL5516 has a light resistance (10Lux) of 5-10 Kohm, dark resistance of 0.5Mohm with a response time of ~30ms. The approximate spectral response curve is given in the following Figure 1.
As test solutions, commerical food dye (Queen Brand) was used, with the following consituents as stated on the product labels.
|Food Colour Dye Constituents|
||INS Food Code
|Queen - Red||Color 124||Ponceau 4R, Cochineal Red A|
| ||Acid 330||Citric Acid|
| ||Preservative 202, 211||Potassium sorbate, sodium benzoate|
|Queen - Green||Color 133||FD&C Blue Dye No1, Triphenylmethane dye|
| ||Acid 330||Citric Acid|
| ||Preservative 211||Sodium benzoate|
|Queen - Blue||Color 102, 133||Tartrazine (102), FD&C Blue Dye No1 (133)|
| ||Acid 330||Citric Acid|
| ||Preservative 211||Sodium benzoate|
The following Figure 1 summarises the visible wavelength absorbance spectra for the dye compounds being tested, overlaid with the RGB LED emission wavelengths, together with the spectral response curve of the GL5516 photoresistor.
An initial proof-of-concept was performed using a cardboard enclosure to mount the photoresistor and the RGB LED (see Album 1, photograph 1 and 2 in the Photographs Section) as a "reading cell". The reading cell provides a means of focussing light from the RGB LED through a test solution (contained in a suitable container transparent to visible wavelengths) onto the photoresistor. The RGB LED was contained within the body of a disused pen to enable ease of direction/focus onto the photoresistor. A DMM was used to read the resistance of the RGB LED versus concentration of various dye solutions.
This initial testing provided the expected increasing resistance with increasing dye concentration (i.e. less light transmitted through the solution onto the photoresistor). The observed resistance versus dye concentration appeared to provide a potentially useful measurement range. The resistance measurement was relatively sensitive to the location of the glass vial containing the test solution (the round vial reflecting/refracting the light on different paths depending upon exact location in relation to the RGB LED and photoresistor).
This initial positive data lead to the manufacture of a more robust reading cell made from perspex scraps (see Album 1, photograph 3 and 4 in the Photographs Section). This perspex reading cell rigidly located the glass vial in relation to the RGB LED and photoresistor. Using various test solutions made from food colouring dyes, the prototype was tested to assess if the expected absorbances versus wavelength were observed and the likely sensitivity of the arrangement. Graphs 1 to 4 below present the results.
Before testing the various dye solutions, photoresistor resistance versus applied LED current was examined (graph 1). This shows increasing LED current (i.e. more "light") corresponds to a decrease in resistance of the photoresistor. Also, the decrease in resistance versus applied current is not linear (greater change in observed resistance at lower currents). This is as per the datasheet for the GL5516 CdS photoresistor, which reports resistance varies logarithmically versus applied light flux. This has implications that a linear relationship between solution concentration and observed photoresistor resistance may not result (i.e. the Beer-Lambert Law, increasing concentration and or path length of a solution will produce increasing absorbance in a linear relationship). A linear relationship is desirable for ease of calculation etc.
Interpreting the observed differences in Graph 1 between the individual red/green/blue LED's within the RGB LED is difficult. This is due to each of the individual LED's having different luminosities at the same applied current, the individual LED's within the RGB LED die are not exactly coincident (i.e. slightly different light paths in relation to the photoresistor), and the photoresistor responds different depending upon incident wavelength.
Test dye solutions were prepared by adding approximately 0.5ml of undiluted dye to 50ml of water. This amount of dye produced visually by eye an opaque solution (i.e. expected maximum absorbance/minimum transmitance). From this maximum concentration solution a series of other solutions were produced by serial dilution with water (see Album 1, photograph 8 and 9 in the Photographs Section).
Graph 2 shows the observed resistance of the photoresistor versus red dye concentration at various applied LED currents for each of red/green/blue LED's within the RGB LED. The red dye (containing cochineal, expected to absorb strongly around 500nm, i.e. corresponding to the emission of both the blue and green LED within the RGB LED) gives the expected absorption of both the blue and green LEDs, while transmiting the emission from the red LED. The absorbance of the blue LED is very sensitive (strongly non-linear), whereas, the green LED absorption is approximately linear (with the range of conditions tested).
Graph 3 shows the observed resistance of the photoresistor versus blue dye concentration at various applied LED currents for each of red/green/blue LED's within the RGB LED. The blue dye has a strong absorbance peak at approximately 630nm, corresponding closely with the emission peak of the red LED. The blue dye absorbance spectrum shows minimal absorbance at wavelengths corresponding to the blue and green LED's within the RGB LED. Graph 3 confirms this expected behaviour (and demonstrates good sensitivity). Due to the sensitivity observed, a further more dilute range of blue dye solutions were prepared. Graph 4 shows that results of measuring photoresistor resistance of these solutions against the red RGB LED at 20mA.
The observed results for the various dye solutions of measuring photoresistor resistance of these solutions against applied light from individual LED's within a RGB LED, shows that a "filter photometer" analogue using a RGB LED as light source with photoresistor as detector shows promise.
Prior initial prototype work (reported above) showed that using a 1W white LED as a light source, with a DVD as a "diffraction grating" (mounted on a stepper motor) and a phototransistor as detector it is possible to obtain 'monochromatic' light through the visible spectrum, that can be directed through a test solution. However, this "minimalist" approach ignores that a parallel polychromatic light beam should be used for illuminating the diffraction grating, in order for the dispersion into monochromatic light to follow the grating equation. Also, since the incident light beam covers the surface of the diffraction grating, the dispersed light needs to be refocused onto the detector. Incorporating "collimation" and "refocusing" is relatively easily done (only requires two concave mirrors), and while complicating construction somewhat, will increase the likelihood of the DIY spectrophotometer having practical utility.
The Czerny-Turner arrangement (6) is used as this only requires a couple of easily sourced concave mirrors. The Wikipedia page (7) explains the elements of a Czerny-Turner monochromator (light source -> entrance slit -> collimating mirror -> diffraction grating -> focussing mirror -> exit slit -> detector). Therefore, the previous DIY spectrophotometer was augmented with two 15cm focal length concave mirrors.
Focal Length of Mirrors
The focal length of the mirrors was checked using both the "distant object method" and the "parallax method" (see Album 2, photograph 1 through 4 in the Photographs Section). The distant object method involved using the light from an illuminated slit (~1mm) directed onto the concave mirror with the reflected image viewed next to the slit (see Album 2, photograph 1). The mirror then being moved back and forth until the image of the slit is as sharp as possible (compare Album 2, photograph 2 and 3). The distance from the slit to mirror when the image is as sharp as possible is the focal length. The parallax method (8) uses two pins which are moved along the focal axis until the real inverted image of the 'object' pin is matched to the 'search' pin and no parallax can been seen between it and the real image. The distance of the 'object' pin and the 'seach' pin to the mirror are then used to calculate the focal length. For the purchased mirrors, both methods resulted in a focal length of 15cm (but since the mirrors were from a "cheap" supplier, best to first check rather than incorporate straight into prototype).
Grating Equation/Checking Dispersion
The grating equation (9) defines the relationship between wavelength and the incident and diffraction angle. This enables calculating from the known incident angle of the polychromatic light what is the expected diffraction angle of the monochromatic light of various wavelenghts. This was tested with a prototype to not only check if the desired dispersed spectrum could be achieved, but that the spectrum was at the expected angles (and thus the geometry of the components of the Czerny-Turner monochromator could be later designed). Using the grating equation and the law of reflection, the following diagram was produced using a 15cm focal length for the collimating mirror and printed to 1:1 scale.
The various components were physically mounted on this template (see Album 2, photograph 5) and the 1st order spectrum were observed at the expected angles (see Album 2, photograph 6 and 7). The white (polychromatic) light from the LED was collimated (parallel light beam) by the concave mirror and directed on the diffraction grating (cut from a DVD). The diffraction grating is assumed to be formed from a series of parallel straight grooves/lines. The DVD does provide the necessary parallel grooves, but obviously these are not 'straight' but curved. This can been seen in the 'curved' appearance of the resultant spectrum (Album 2 photograph 7). Note that the 'outside' edge portion of the DVD has been used as much as possible, as the DVD tracks in this region approximate 'straight' lines closer.
Alignment of optical components
A prototype Czerny-Turner spectrophotometer was constructed (Album 2 photograph 8) using a symmetrical arrangement based on 15cm focal length concave mirrors with a "DVD diffraction grating", using a 1W white LED as light source with a TCS3471 Color Light to Digital Converter as the detector.
In order to align the various optical components, an "optical rail" arrangement was constructed using a tongue-and-groove joint cut into a suitable piece of wood. This enabled "precise" placement of components both in a linear and parallel fashion, while still allowing flexibility in alterating the distances between components etc.
A Red Laser Pointer was mounted in the light source compartment which provided a monochromatic beam of 'known wavelength' (650nm) with which to align the mirrors, diffration grating and detector (Album 2 photograph 9 and 10). Photograph 9 in Album 2 also shows the 1st order diffraction of the red laser pointer light (the red dot on the 'backing screen'). The incident angle to the diffraction grating is 30o when the diffraction grating is perpendicular to the light path from the entrance slit to the collimator mirror. Therefore, with a wavelength of 650nm for the Red Laser Pointer and a groove spacing of 0.74um (DVD diffraction grating), the calculated 1st order diffraction angle is 22.2o. This was indeed the angle at which the the red dot on the 'backing screen' was observed.