Light Emitting Diodes (LED) for Aqueous Light Bleaching of Paper

1 Introduction

Light bleaching, historically practiced by sunlight exposure, was rediscovered in conservation in the 1970s (Branchick et al. 1982; Keyes 1982; Smith 2009) and has since become a recognised bleaching method (Brückle 2009b; Hofmann et al. 1991; Schaeffer et al. 1997). The intensity of light exposure determines the rate of bleaching action (Feller et al. 1982; Palmer Eldrige 1982; van der Reyden et al. 1988). With sunlight and fluorescent tubes, the noted drawback concerns the potentially long duration of aqueous exposure (Baker 1982, 1986; Saur-Aull 1996; Brückle 2009a; Henniges and Potthast 2009), due to their low light intensity (Gowland and McAusland 2003). The first light source with significantly higher luminous flux was the high-intensity-discharge lamp (HID) introduced by Perkinson (2001). The illuminance was adjusted by the distance between the wetted or immersed object and the lamp. The 300 W HID lamps in this study provide a maximal illuminance of about 50,000 lx when used at the nearest recommended distance to the object of 60 cm. However, other published data suggests that HID laps with greater power are capable of reaching levels as high as 100.000 lx (118.000 lm, 120°, 60 cm) (Schopfer 2012). Due to the high thermal radiation emitted by all HID lamps, they must not be placed closer to the object to avoid heat accumulation and corresponding negative effects on the object, in particular protein sizing and media (Schaeffer 1991).

Light emitting diodes (LED) offer several advantages over the established light sources. Invented over 30 years ago, they have gone through a tremendous development toward a high level in efficacy and luminous flux. Today, they are present in countless lighting systems for indoor and outdoor applications. The currently available LED chips (≤1 cm²) are significantly smaller than any traditional light source. High-performance LEDs by far exceed all other sources in their illumination potential (Figure 1). The efficacy of LEDs also tops most other lighting technologies. Available in a wide range of specifications and pre-assembled modular systems, they have replaced numerous traditional lighting technologies also in museum lighting (Garside et al. 2017; Herzberg 2015; Michalski et al. 2012; Miller and Druzik 2012). Considering the past development, it seems likely that luminous flux and efficacy will see further increases in the future. Due to their monochromatic light emission, LEDs usually do not emit ultraviolet (UV) radiation, which is found in the emission of all other light sources used in bleaching. UV radiation has an accelerating effect on paper bleaching but also on cellulose degradation independent of the visible (VIS) light intensity (Feller et al. 1982; van der Reyden 1988).

Figure 1:

Illuminance of LED in comparison to fluorescent tubes (FT) and high-intensity discharge lamps (HID) according to published studies (see references). Bars indicate achievable illuminance in lux (lx); dots indicate the closest recommended distance (cm). Error bars indicate fluctuations in illuminance due to setup and ageing of the lamps.

For these reasons, LEDs are an interesting alternative radiation source for light bleaching. This study focuses on the characteristics of modular LED systems, their possible application in bleaching, their effectiveness and the cellulose stability. They are also compared to HID lamps available in the lab.

2 Light emitting diodes (LED)

2.1 Technology

LEDs are made of semiconductor crystals that can allow or block the flow of electrons depending on direction of the applied current (Appendix A). When an aligning current of sufficient voltage is applied to the crystals, they emit light. The light emission increases with increasing voltage until a maximum is reached. The emitted light is of a specific wavelength which depends on the elemental composition of the crystal. To produce white light, a LED with a peak in the shorter wavelengths (blue) is combined with a phosphor-based filter layer that transforms a part of the blue emission into light with greater wavelengths. The mixed light appears white (Dohlus 2015). The thickness and composition of the filter determines the colour temperature and spectral composition of the white light. Unlike other light sources, white LED do not emit high levels of infrared radiation. The energy that is not transformed into light remains in the crystal in form of heat. The diodes are usually attached to an aluminium board as a thermal transmitter (Figure 2) that absorbs the process heat and avoids overheating and disintegration of the crystal. Due to their power, high-performance LED usually require additional cooling units such as heat sinks and fans.

Figure 2:

Model of a white light LED semiconductor: (a) the phosphorous filter covering a crystal containing (b) the p-doped layer, (c) the active area that emits radiation under aligning electrical current and (d) the n-doped layer. The semiconductor is attached to (e) the substrate.

3 Tested light sources

Two LED were selected, Aventrix and MiniMatrix, supplied by the international supplier LUMITRONIX^® LED Technik GmbH (Hechingen, Germany). The modules can be assembled to create larger lighting both units.

3.1 Aventrix (AVX)

Originally designed for street lighting, the Aventrix 4×4 module (Figure 3) features 16 high-performance LEDs of the Nichia E21 series (NVSLE21AT; R70) (Lumitronix (1) 2017; Nichia (1) 2017) (summary of properties, see Table 1). The chips are mounted on a square aluminium core circuit board (12 × 12 cm). With a maximum luminous flux of 7,785 lm per module, they provide a high level of luminous flux. The modules are dimmable by adjusting the voltage between 44 V and 48 V (DC). Each module consumes a total radiant flux of 68.8 W (Lumitronix (2) 2017). With a medium luminous efficacy of 136 lm/W, they are about 60 % more efficient than the HID lamps currently used in the lab (Lumitronix (1) 2017; Osram 2017). A square lighting unit of 1 m² requires 64 modules (8×8; 96×96 cm) and emits a total luminous flux of around 498,240 lm while consuming a radiant flux of 4,400 W. The modules can be attached either with screws though pre-drilled holes in the circuit board or with heat transmitting epoxy resin. The lifespan is estimated with above 60,000 h. With daily 8-hour use, the AVX will last 20 + years. Since most labs use light bleaching not continuously, the actual life span will be longer. Due to their high radiant flux, AVX need a passive cooling unit such as a heat sink to remove process heat and avoid heat accumulation as well as diminished performance.

Figure 3:

Aventrix 4×4 (AVX) module with 16 LED chips (yellow) mounted on an aluminium core circuit board (drawing courtesy of LUMITRONIX^® LED Technik GmbH, 2017).

Table 1:

Characteristics of the MMX and AVX LED and the Osram HID.

	MiniMatrix (MMX) (Nichia NT2W757DRT; R8000)	Aventrix (AVX) (Nichia NVSLE21AT)	HID (Osram Power Star HQI TS 400 W/D; Norka Pollux 1 400W)
dimensions	30 mm×30 mm	120 mm×120 mm	579 mm×318 mm
colour temperature	6,500 K	4,000 K	5,500 K
angle of radiation	120°	120°	–
segments/m2	1,024	64	2
power consumption/segment	0.48 W	68.8 W	400 W
power consumption/m2	492 W	4,400 W	800 W
luminous flux/segment	78 lm	7,785 lm	35,000 lm
luminous flux/m2	79,870 lm	498,240 lm	70,000 lm
Voltage	24 V	48.8 V	118 V
Amperage/Segment	0.002 A	1.4 A	4 A
Amperage/m2	20 A	90 A	8 A
luminous efficacy	163 lm/W	136 lm/W	87 lm/W
colour rendering index (Ra)	>80	>70	>90
lifespan	>200,000 h	>60,000 h	12,000 h
external cooling	no	yes	yes
dimmable	yes	yes	no

3.2 Minimatrix (MMX)

The MiniMatrix system of the Nichia 757 series (NT2W757DRT; R8000) (Nichia (2) 2017) was designed for backlighting in architectural and advertising elements (Figure 4) (summary of properties, see Table 1). In the factory-made state, 126 segments, each 3×3 cm and featuring four LEDs, are connected in a 27×42 cm module. By breaking the perforated joints between the segments, the module can be divided in any preferred number of segments. Each segment features electronic contacts on each edge. By attaching single segments or segment clusters with special board connectors, a module of a preferred shape can be built (Figure 5), for example to backlight single letters in an LED billboard (Lumitronix (3) 2017). Their spectral luminous efficacy of 163 lm/W is 96 % higher compared to the HID used in our study. Each segment provides a luminous flux of 78 lm. In a square lighting unit of 1 m² (32×32; 96×96 cm) they would emit a total luminous flux of 79,870 lm and consume a radiant flux of 492 W. The segments are dimmable between 20 V and 24 V. With daily 8 hour use, the MMX will last ca. 68 years. Since most labs use light bleaching not continuously, the actual life span will be longer, and exceed that of HID (Osram Powerstar HQI TS 400 W/D) 16 times. The segments are built of a lightweight honeycomb aluminium board. Because of their relatively high colour temperature, high efficacy and lower luminous flux, the MMX does not require external cooling. By attaching the modules with plastic spacers at a short distance to the supporting board, the airflow around the module ensures sufficient cooling.

Figure 4:

MiniMatrix (MMX) segment. Four LED chips (yellow) are mounted on honeycomb aluminium circuit board with perforated contacts around the outer edges. Additional gaps are added at each side for insertion of connecting bridges (drawing courtesy of LUMITRONIX^® LED Technik GmbH, 2017).

Figure 5:

MMX segments held together by connectors that ensure power supply (drawing courtesy of LUMITRONIX^® LED Technik GmbH, 2017).

3.4 Differences between MMX, AVX, and HID

Both LED modules fulfil the desired properties (see 2.2) while they also show certain differences (Table 1).

By far the most obvious difference between the two modules is their luminous flux. Per module, the luminous flux of the AVX is 99 times higher than the MMX which is only in part due to dimensional difference. When calculated per 1 m², the AVX still reaches a luminous flux about 6 times higher than the MMX. Compared to two HID lamps illuminating 1 m², the MMX are slightly (14 %) and the AVX more than 7 times (712 %) more intense. Also, LED can be placed at a shorter distance to a substrate than HID, which allows an exponential increase in illuminance (lx), so it was expected that especially the powerful AVX would show a greater bleaching efficiency.

MMX segments are 1/16 the size of AVX, requiring 16 more segments of MMX than of AVX for a light bleaching unit of given dimension. A large number of segments may prove disadvantageous because the relatively frail connections between the segments may break, causing power disconnection. Segments can then only be reattached by inserting a separate connector, which requires extra space and disrupts the uniformity of light distribution. However, multi-segment modules allow more flexible uses.

MMX segments mounted on corrugated board are significantly more lightweight than the AVX mounted on solid board. Also, MMX shows a higher colour temperature and spectral distribution than AVX. While the AVX shows a sharp peak around 450 nm and a broad second peak around 590 nm, the MMX shows a more prominent sharp peak around 460 nm and a low curve peaking at 580 nm (Figure 6). The spectral luminous efficiency of MMX is slightly higher than that of AVX and is significantly higher than for the featured HID lamps.

Figure 6:

Light spectrum of the 4,000 K Aventrix 4×4 (AVX) with NVSLE21AT chip (R70) and 6,500 K MiniMatrix (MMX) with NT2W757DRT chip (R8000) compared to the Osram HID lamps (Power Star HQI TS 400 W/D); (after data by LUMITRONIX^® LED Technik GmbH (2017) and Osram (2017)).

4 Experimental

A test light bleaching unit was designed that features an aluminium profile frame, metal brackets for suspending a removable polypropylene tray at a desired distance from the light source, a set of four cooling fans and the LED source. The MMX and AVX modules and thermal transmission supplies each were mounted on a 1.5 mm aluminium plate (Figure 7). Each LED panel was connected via a lustre terminal and cable to an adjustable 300 W laboratory DC power supply. Since the DC power supply heats up during operation and with increasing power, the cable was sufficiently long (1.5 m) to avoid heat transfer to the LED modules. The LED modules were easy to mount and exchange in the frame. The MMX module featured 64 segments with a total luminous flux of 4,992 lm, the AVX module featured four segments with a total luminous flux of 31,140 lm and included an aluminium heat sink on the back. Although the LED spectrum is UV-free, at maximum intensity it is hazardous, requiring tinted goggles with 60 % light transmission rate during operation.

Figure 7:

Light bleaching test device with (a) cooling fan, inserted in (b) corrugated cardboard, (c) heatsink, attached to (d) aluminium plate supporting the (e) LED module (here: AVX). The distance between the LED and the tray (f) is adjustable between 3–30 cm. The DC power supply (g) serves as an on-off control and determines the illuminance by setting the desired voltage.

4.2 UVA content and illuminance of the tested light sources

The energy of UVA radiation emitted by the LED and HID (mW/m²) as well as its correlation with the luminous flux (µW/lm) was determined with an ELSEC 764 photometer (Table 3). The HID emits a significant level of UV radiation as illustrated in Figure 6. At a distance of 60 cm the exposure is 18,700.80 mW/m² with a ratio to the VIS of 423.00 µW/lm. At 120 cm distance, the exposure dropped by 60 % to 7101.00 mW/m² but the ratio of 405.60 µW/lm remained. For both LEDs, the relative amount was too low to be displayed by the photometer, indicating that there is no UV radiation in the emitted light. The small amount of W/m² is likely due to daylight UV transmitted through the room windows during measurement. The HID was used unfiltered for UV in this study.

Table 3:

UV radiation determined for both settings of the tested light sources (five measurements each).

type	mW/m2	standard deviation	µW/lm	standard deviation
HID 60 cm	18,700.80	±159.89	423.00	±0.50
HID 120 cm	7,101.00	±14.20	405.60	±4.19
MMX 10 cm	6.88	±0.29	0.00	±0.00
MMX 20 cm	3.16	±0.17	0.00	±0.00
AVX 10 cm	137.00	±1.71	0.00	±0.00
AVX 20 cm	53.60	±0.15	0.00	±0.00

The homogeneity of the LED illuminance was checked by exposing undeveloped cyanotype paper to the LED sources at the closest chosen treatment distance (10 cm) (Figure 8). One quadrant of the paper was covered with Marvelseal^® 360 before the start of exposition. The remaining quadrants were covered, respectively, after 2, 5 and 10 minutes. The exposed paper sheets were developed in demineralised water and air-dried. Any variances in the developed cyanotype’s blue colour would represent uneven exposure caused by lower or higher illuminance. Both cyanotypes were evenly developed (disregarding a small contamination-induced error in the bottom left segment of MMX), indicating a homogeneous LED light distribution.

Figure 8:

Cyanotype papers (each 7×10 cm) developed after different exposure durations under AVX (left) and MMX (right), to verify homogeneous illuminance at a short distance (10 cm).

5 Results

All treated samples show a brightness increase (measured as colour differences Δ_E) (Figure 9). Immersion washing alone increased the brightness of both papers about Δ_E= 3.37 (±0.21), which is a slight, though perceptible change. The brightness increase effected by bleaching shows more diverse results. Both papers’ brightness increased with intensified exposure. Brightening was more intense for bleaching at a shorter distance or over a longer time. Furthermore, paper 1 showed a greater brightness increase than paper 2 (Figure 10).

Figure 9:

Colour distance (Δ_E, CIE1976) of paper 1 and 2 after four different bleaching treatments (see Table 4) compared to washed papers (dots and dotted line). See also Figure 10.

Figure 10:

Paper 1 (top) and paper 2 (bottom) after four different bleaching treatments (groups, see Figure 9, treatment ID, see Table 4). Colour measurement spots are marked with pencil.

For samples in group 1 that were bleached at the far distance and for the shortest time, the average bleaching effectiveness can be ranked HID above AVX above MMX, though the difference between HID and LEDs is statistically significant only for paper 1. Group 2 samples which were bleached at the same distance for 4 h, results were quite similar, effectiveness was similar for HID and AVX and better than for MMX. Paper 1 was bleached with HID and AVX intensely, while paper 2 showed no significant brightness increase. The MMX showed the lowest brightening effect on paper 1.

In group 3, which was bleached at the minimum exposure distance and for 2 h, the AVX modules effected a brightness increase that was about 28 % (paper 1) and 40 % (paper 2) more intense than the HID. This is the most significant absolute brightness increase achieved within 2 h of exposure. MMX and HID achieved no significant brightness increases for paper 1, and only a slight one for paper 2. Within group 4, which was bleached at the same distance for 4 h, HID and MMX effected a significantly greater brightness increase than in group 3. For paper 2, a similar ranking like in group 3, with an overall increase of brightness, can be observed with HID being 19 % lower than AVX. The difference between the brightening achieved by AVX when compared to HID and MMX indicates that for those two light sources the exposure time was not long enough to reach the full potential brightening.

When comparing the four groups, the most significant overall colour change (equalling brightening) of paper 1 was Δ_E ≥ 10. It was achieved by HID in both distances after 4 h exposure, by MMX in 10 cm distance after 4 h and by AVX in 10 cm distance after 2 h and 4 h. All papers underwent a colour shift in the same direction: They became brighter (ΔL), less red (Δa) and less yellow (Δb). Washing caused a greater ΔL increase for paper 2 than for paper 1.

The weight average molar mass showed a decrease for HID by 15 % (±1.29 %) and for MMX-bleached papers by 17 % (±1.32 %) while for AVX there was no notable change (Figure 11). The carbonyl group content, an indicator for oxidation of the cellulose, was increased after HID bleaching by 11 % (±1.02 %) and after MMX bleaching by 14 % (±3.17 %), but was slightly decreased for the AVX sample.

Figure 11:

Results of cellulose analysis. Left: Molar mass (kg/mol) of the untreated reference (REF), and the three samples with the highest exposure for each light source (e. g. smallest distance and longest exposure: HID-1–2, MMX-1–2, AVX-1–2). Right: Carbonyl group content (µmol/g) of the untreated reference (REF), and the three samples.

6 Discussion

Although a significant bleaching effect could be observed for both tested LED sources, they showed notably different performances. Especially Aventrix 4×4 (AVX) was highly efficient in reducing discolouration. Due to its luminous flux of 31,140 lm and the resulting high illuminance, the bleaching efficiency was considerably higher than that achieved with the MiniMatrix (MMX). HID in 60 cm distance showed a similar brightening effect as AVX in 20 cm distance. However, this is most likely due to its UVA content, which makes up, in bleaching, for its low illuminance (see Figure 1). Therefore, the paper brightening achieved by HID may not be related only to the VIS but also to UV which was not filtered in this study. The significant drop in molar mass of the HID exposed samples when compared to the reference confirms the known fact that UV radiation causes photocatalytic breakdown of the cellulose, as noted by Verborg (2012). A corresponding slight increase in carbonyl group content furthermore illustrates the UV radiation-related cellulose oxidation. For the AVX bleached samples, no significant negative effect on the cellulose was evident despite the intense illumination. With MMX, the molar mass dropped significantly, and carbonyl group content increased. Since this cannot be linked to UV radiation, the high amount of blue light present in the spectrum might be the reason for the cellulose breakdown and oxidation. It can be assumed that the different colour temperature of the two LED sources had a more significant effect of the cellulose than the illuminance. The blue light content, known as a key factor for light damage, must also be considered a risk factor concerning cellulose damage during light bleaching. However, this connection is not entirely conclusive due to the relatively small sample size.

The most intense brightening occurred at 10 cm distance during the first 2 h exposure, indicating that the high illuminance increased the rate of bleaching. Compared to AVX, the MMX showed a lower performance in brightening the paper, explainable by its lower illuminance. The applied external cooling with heat sinks and fans was effective in keeping the operating temperature of the AVX within the acceptable range.

7 Conclusion

The present study demonstrates that high-performance LEDs have a great potential of raising the efficacy and sustainability of artificial light bleaching of paper. The high luminous flux and outstanding long lifespan make them an interesting light source for bleaching. The absence of UV radiation in the emitted spectrum and the low level of emitted thermal radiation allow LED sources to be placed in a much closer distance to the object than any other light source. The wide angle of radiation of 120° ensures a homogeneous illumination even in close distance. Especially when LED with moderate colour temperature, i. e. lower short-wave VIS content, and high luminous flux, like the AVX, are applied, excellent brightening results with no notable negative effect on the cellulose substrate can be expected. Since increased oxidation of the cellulose likely accelerates the extent of colour reversion (Burgess 1982), the results confirm the importance of minimizing short-wave VIS (blue light). Thus, the total intensity of the applied radiation may not be as important for the preservation of the cellulose as the spectral composition of the applied radiation. The AVX in this regard proved the better choice than the MMX. Nevertheless, the studied effects need to be complemented by subsequent research. Certainly, the influence of the colour temperature on the degradation of the cellulose and the relation between illuminance and bleaching need further attention. As LEDs are likely to replace most common light sources in the future, they hold significant potential for paper conservation practise.

The LED modules can be constructed to units of any size, weight and power. Effective cooling of the LED is always an important construction feature. LED are generally safer than HID since they operate at lower voltage, do not emit UV and can be dimmed to any desired luminous flux. No pre-heating or cooling breaks are needed, allowing the conservator to pause the exposure any time during the treatment. The illuminance can be adjusted to meet the treatment requirements through voltage adjustment without changing the setup. Furthermore, by masking parts of the object during treatment and adjusting the illuminance, light bleaching at different rates is possible at shorter exposure times than those that are common with other light sources. LED exposure has some health hazards, as the intense illumination can cause stress and dizziness. Protective eyewear, such as tinted goggles or light-blocking shields must be considered as a safety precaution, especially when working with illuminances over 25,000 lx.

A series of ready-to-use countertop and overhead LED bleaching units is in the process of construction and will be marketed by Belo Restaurierungsgeräte GmbH.

Figure A1:

Semi-conductor crystal with an n-doped layer (left; in red) and a p-doped layer (right; in green). (a): p-n-junction under normal condition. (b): increased p-n-junction under electric current opposing flow direction. (c): irradiation reaction under electric current matching flow direction (after Heinz 2009).