Equipment, Setup and Contact Materials

Editor’s Note—The paper on which this article is based was originally presented at the 2018 IEEE Product Safety Engineering Society Symposium, where it received recognition as the Best Symposium Paper. It is reprinted here, with permission, from the proceedings of the 2018 IEEE Product Safety Engineering Society International Symposium on Product Compliance Engineering. Copyright 2018 IEEE.

The physiological effects of electrical shock are mainly induced by current [1], thus current limits are often specified in safety standards to protect the human body from electrical shock hazards [2]. However for certain standards or applications, voltage limits are often preferred. In such cases, human body impedance can be used to estimate the voltage limit based on safe current limits. Furthermore, human body impedance can be used to build electrical circuit models representing conduction paths through the human body for estimating touch currents. For example, as given by UL 101 [2], the human body impedance is modeled with a 1500 Ω resistance in parallel with a 0.22 µF capacitor, placed in series with another 500 Ω resistor. Such a measurement circuit is used to evaluate the touch current of perception-level shock effects for 60 Hz sinusoidal AC.

DC applications, particularly those at hazardous voltages, have become increasingly prevalent due to increased use of renewable energy sources (such as photovoltaics), energy storage systems, etc. Therefore, it is beneficial to review the human body resistance at DC to better understand its effect on electrical shock physiological effects for DC applications. IEC 60479-1 [3] specified the human body resistance for DC human body resistance between 25 V up to 1000 V. However, the DC resistance values included in [3] were based on experimental data conducted at 25 V only, with the rest of the values mathematically extrapolated based on AC human body resistance. This makes the assumption that body impedance scales identically with DC as it does with AC, which may or may not be the case. In addition, the DC body resistances given in [3] are only for dry conditions. To the knowledge of the authors, currently there are no data available on DC human body resistance for wet conditions based directly on experimental observations. For wet conditions, the DC body resistance values provided in the IEC standard [3] is assumed to be identical to the AC body resistance under wet conditions at each voltage. Again, this makes assumptions that are not firmly substantiated by experimental data.

As DC at hazardous voltages and the potential for human exposure to such hazards becomes more common, it is imperative to have comprehensive data on human body impedance at DC. Such a dataset will require measurements on many individuals to have any statistical significance (i.e., 50 or more). It is the ultimate goal of the authors to develop typical DC body impedance values based on experimental data, similar to what is currently available for AC.

However, it is yet unclear to what level of influence various measurement parameters will have, and it is also to date unclear how repeatable any given measurement would be on any particular individual. The authors concluded that it was therefore premature to move forward with a large-scale experimental program involving a significant number of human volunteers before a better understanding of the consistency of measurement on any particular individual was established. For example, it was unclear whether repeatable values were attainable when measuring the same individual at different times under the same test conditions. To the knowledge of the authors, no previous work has evaluated this. Furthermore, no published data were found on the effect of contact material on human body impedance testing, which is presumed to have an effect on measured body impedance and may better explain the relationship of measured body impedance and applied voltage. (It was reported in [3] the body resistance decreases linearly as the applied voltage increases.) Upon consideration of these unknown effects on DC body impedance measurement, the authors have taken the first step to further understand these factors; the results of which are reported here. The larger-scale test program was therefore postponed to be conducted as a second phase of our work.

The initial work that is reported here explored the effect of three test variables (contact material, wet or dry conditions, and time of day) on three test subjects. Copper and aluminum contact materials were used to better understand the potential influence on measured body impedance, and to test a hypothesis that the nonlinear behavior observed in body impedance behavior is analogous to what is observed in metal-semiconductor contacts. [6] Wet and dry conditions were used to evaluate the effect on measured body impedance, as well as to determine the relative repeatability of measurements under each condition. For safety, this work limited the scope of the investigation to voltages at and below 60 V.

Review of DC Voltage Shock Threshold

There are four distinct electrical shock physiological effects for either DC or AC: perception, inability of let go, ventricular fibrillation, and burn. According to the experiments conducted by Dalziel in the 1940s [1], the electrical shock threshold of direct current is higher than that for alternating current. In other words, the human body is less vulnerable to electrical shock at DC compared to 50/60 Hz AC signals at similar magnitudes. In regards to the voltage limit, the DC limit is 60 V under dry conditions, and 30 V under wet conditions as specified in UL 1310, [4] intending to protect against the inability of let-go shock effects. This limit was selected with the intent to protect 95% of the population including children. Note that this limit is defined based on a hand-to-both feet pathway: for other current pathways, the allowable voltage limit could be different. In order to collect a wider range of experimental data, the voltage limit was set at 60 V instead of 30 V for both dry and wet conditions in this work, though 60 V is the limit for the inability of let-go under only dry conditions. It is noted however that these limits were set with the consideration of children, and in this work the subjects were adults only (and therefore are able to endure higher voltages). For safety and the comfort of the test subjects, each subject was able to break the circuit anytime during the test by lifting his hand from the electrode (Figure 2) if the perception sensation became too uncomfortable. The current was limited to below 20 mA, both through a current limit setting on the power supply, as well as through the inclusion of a fast-blow 20 mA fuse placed in series with the current supply to the test subject.

Equipment and Experimental Setup

The BK Precision Model 9183B was used for supplying DC for the tests. During the test, the power supply voltage output was controlled by laptop computer. The Dewetron Model DEWE-50-USB2-8 was used for data acquisition, including the voltage and current output from the DC power supply. Current and voltage connections to the metal plates were physically separated so that contact resistance effects were minimized. Body resistance was calculated from the voltage and current readings using Ohm’s Law.

Figure 2 shows the experimental setup, shown with a subject in position during testing. The subject stood on a metal plate which was alternated between copper and aluminum. Subjects were instructed to stand on the plate in bare feet. Each subject placed their right hand on a metal plate of the same material as that of the plate at the feet. Prior to each testing session, the hands were cleaned using an alcohol wipe to remove surface oils and dirt, as well as to dry the hands. No effort was made to clean or dry the feet. The size of the metal plate for the hand contact was 100 mm by 100 mm. This size is defined as a “large contact area” by the IEC Standard 60479-1 [2]. According to this standard, the larger contact area is expected to result in the lowest resistance to the body, which is considered as a “worst case” comparing to similarly defined “medium” and “small” contact areas. The large contact area or the “worst case” is assumed in this work as worst-case conditions are of most interest for safety-related applications.

The human body resistance is known to be affected by the moisture level of the skin surface [3]. In this work, the surface of the hands were tested under two conditions: dry and simulated sweat water wet. The concentration of sodium in the sweat is in the range between 6 to 85 mEq. per L [5], equivalent to 13.8 mg/dL to 195.5 mg/dL. A concentration of sodium on the higher end of this range is selected for this work, as this will lead to the lowest resistance, and therefore a “worst case” electrical safety hazard. The test was run with a salinity of 80 mEq per L of water which approximates the 95th percentile of the maximum sodium concentration of 85 mEq. This equates to 1.85 g NaCl per liter of water. Human sweat also contains potassium and other salts, but these concentrations are much lower compared to the sodium concentration [5]; therefore, the effects due to potassium were ignored in this research.

As shown in Figure 2, two bags each filled with 0.5 kg of metal shot were applied on the back of the hand. The test subjects were instructed to relax their hand, with the intent that the pressure was only applied by the weight of the bags. This was an attempt to control the variable of pressure onto the metal plates, which may have an effect on body contact resistance.

Results

The supply voltage was applied linearly from 0 V to 60 V, ramped linearly at a rate of 1 V/s. The current was limited to 20 mA, and when either current or voltage reached the limit, the power supply switched to a constant-current source at 20 mA. Each test is continued until 60 V was reached or the test subject removed his hand from the plate due to discomfort. Figure 3 shows an example of the measured voltage and current under dry conditions for one subject. In this case, the applied DC voltage reached 60 V then ramped back to 0 V, with the current observed to be around 12 mA at the applied voltage of 60 V. In Figure 3 the x-axis is the time step, which is a sample count of the measurements is taken during the 60-second test.

Table 1: Test conditions

Four combinations of test conditions were measured: Table 1 shows the conditions which were used for each test. For each test condition and volunteer, the test was repeated 20 times over a timeframe of several weeks. Figure 4 shows the boxplot results for measured current for each of the four tests and three volunteers at an applied voltage of 25 V. The upper edge of the outer box with light blue color represents the first quartile (Q1), while the lower edge represents the third quartile (Q3). Thus this outer box portion of the plot represents the interquartile range (IQR), or the middle 50% of the observations. The inner box represents the 95% confidence interval. The vertical lines represent the upper and lower whiskers which extend outward to indicate the lowest and highest values in the data set (excluding outliers). The horizontal line in the box represents the mean value; the circle with the cross mark represents the median value.

The data in Figure 4 show that the variability of body resistance under wet conditions (tests 3 and 4) were significantly smaller than that observed for dry conditions (tests 1 and 2). Moreover, the resistance of the wet condition was lower than the corresponding dry condition, which implies that the wet condition is a worse case (i.e., more hazardous) from a safety perspective. As conservatism is generally preferred in safety analysis, and the fact that the wet condition data exhibited lower variability, the results suggest that future body impedance testing will be conducted under wet conditions only.

Figure 5 shows the boxplot of measured current for both tests 3 (aluminum) and 4 (copper, both conducted using wet skin) at 5 V. It was observed that the metal electrode had an influence on the measured current. For all test subjects, copper exhibited a higher measured current at 5 V and 10 V, with this effect found to diminish as voltage increased. It is noted that this is not likely attributable to the higher electrical conductivity of copper relative to aluminum, as the four-probe configuration used for measuring resistance includes only the metal-skin contact into the measurement and does not include the bulk resistance of the metal contact. Additionally, any bulk resistance contribution would be observed over all voltages and would not diminish as voltage increased.

The analysis of variance (ANOVA) method can facilitate in determining the significance of a factor for a particular output parameter. In this study, it was used to determine the statistical significance of the metal material of the electrode on the body resistance measurement as a function of voltage. Figure 6 shows the R values calculated for the influence of the electrode material on the measured current: a higher R value suggests higher influence on the output parameter. It was found that the R value is high at low voltages, then drops quickly as the voltage increases from 10 V to 20 V. This suggests that the metal electrode is found to have a statistically significant influence on measured impedance at voltages below 20V, which is consistent with presence of a Schottky barrier at the metal-skin interface. [6]

Figure 7 shows the median value of resistance of voltage for test 4 (wet condition, copper electrode). It was observed that body resistance generally decreased as the touch voltage increased. The body resistance also was found to possess a nonlinear relationship with voltage, observations consistent with the IEC 60479-1 standard. IEC 60479-1 notes this nonlinear behavior, and also mentions further increases as electrical breakdown of the skin occurs [3]. The nature of this nonlinear behavior is not described in IEC 60479-1. The authors propose that this nonlinearity can be explained in the context of a Schottky barrier, where the skin-to-metal plate contact forms a metal-semiconductor junction that leads to a nonohmic current-voltage behavior. The difference in measured resistance between aluminum and copper would then be a function of work function (which for the two metals are approximately 4.3 and 4.7 eV, respectively) [6]. Measurements using additional metal surfaces would help to substantiate or refute this hypothesis, for example conducting measurements using materials of a lower work function (i.e., magnesium, 3.7 eV) and higher work function (i.e., nickel and platinum, 5.2 and 5.7 eV, respectively), both potential subjects of future work.

The coefficient of variance (CV) represents the ratio of the standard deviation to the mean, which is useful for comparing the degree of variation of the measured resistance for each individual volunteer. As establishing the repeatability of body impedance measurement for the same individual was a key goal of this work, the CV helps to quantify this variability. Figure 8 shows the CV for all three volunteers using copper electrodes under wet conditions (test 4), comparing the relative variance of the data among volunteers. It was observed that volunteer No. 3 exhibited much larger variation in body resistance compared to the other two volunteers (this can also be observed in Figure 4). For volunteer No. 2, the measured body resistance was less consistent at lower voltages, but as the voltage increased over 20 V, the CV of the body resistance drops to around 10%, consistent with the data from volunteer No. 1.

To further investigate the larger variation observed from the results of volunteer No 3, the data were separated by time of day (morning and afternoon). The morning is defined as measurements conducted during the normal workday before 12 PM local time, and afternoon is defined as measurements completed after 12 PM. For the duration of this work, typically two measurements were completed each day on each test subject: one during the morning and one during the afternoon. The time reference was defined as the time when the test results were completed and saved in the computer. Figure 9 shows the coefficient of variance for morning (a) and afternoon (b). As was the case shown in Figure 8, Figure 9 also shows data with copper electrodes and skin under wet conditions. It is interesting to note that the CV varies significantly for volunteer No. 3 between morning and afternoon, a trend that was consistent for this subject over all the applied voltages used in this study. The difference for the other two volunteers between morning and afternoon was found to be less significant, especially for volunteer No. 1. It is noted that for tests conducted in the morning, the variation of the data for volunteer No. 3 were actually lower than for volunteer No. 2 at voltages less than 25 V. For both morning and afternoon, the CV decreases as the voltage increases. The exact nature of the statistically significant differences in morning and afternoon data for volunteer No. 3 are unknown at this time. Since this difference in behavior was observed across 20 measurements over several weeks, it is less likely that the issues was due to measurement error and more likely attributable to some metabolic or other body condition affected over the noon hour (i.e., lunch). Another, though less likely, possibility is some sort of unconscious change in behavior of volunteer No. 3 between morning and afternoon, though it would be difficult to affect such a change consistently over 20 test sessions. Regardless of the cause, of importance here is less the specific cause and more of the overall effect on body resistance. These observations do suggest that future measurements may need to be conducted during both the morning and afternoon and the corresponding measurement time noted for each test subject for future studies.

Conclusion

The data showed an influence of contact material on the measured body resistance and may be due to formation of a Schottky barrier, similar to what is observed with metal- semiconductor devices. This would explain the nature of the nonohmic behavior that has been long known for human body impedance, though further investigation is needed to confirm this hypothesis. The results show that it is imperative to report the composition of contact materials used for body impedance measurements with the results, and that only one type of contact material should be used for both contacts.

Wet conditions exhibited more consistent test results of the body resistance than dry conditions. Considering this and the fact that wet conditions exhibit lower body resistance than the corresponding dry condition, future work will focus on utilizing wet conditions only. This study also demonstrated that the measured resistance can vary significantly between different times of the day, namely morning and afternoon as was investigated here. It was also observed that this variation with respect to time of day was not observed with all test subjects, and was of unknown origin. Regardless of cause, the results suggest time of day is a potential variable for body impedance and needs to be continued to be included in future investigations, preferably obtaining data at different times of day for the same volunteer. When this time-of-day effect is separated from the data, it is observed that the coefficient of variance tends to be around 10%, with higher values observed at lower voltages.

The experimental work described here shows that human body impedance measured for a particular test subject, conducted under the same test conditions, is expected to be repeatable over time. Data are expected to be normally distributed with standard deviations roughly 10% of the mean value for most test subjects and conditions, though larger variability is possible for some test subjects (particularly due to changes due to the time of day, with as-yet unknown origins).

These findings confirm that data from a larger sample set of test volunteers would likely be representative of each individual’s DC body resistance within a predictable level of uncertainty, even if only one measurement session was conducted on a volunteer. However, measurements conducted several times on additional volunteers would be beneficial for building a better understanding of the effects of test variables on individuals. This suggests that in future work a subset of volunteers will be asked to return to repeat measurements over several days, while the larger population may only be asked to participate in one or two test sessions (preferably two, with one conducted during the morning and a second that same afternoon). To further investigate the nature of the nonohmic contact behavior, the three original test subjects will be asked to repeat testing using additional contact materials. Additional test subjects may also be requested to conduct tests using several contact materials as well. For all testing, it is anticipated that only wet conditions will be used, as the measured currents were higher and data variability were lower. Tests are anticipated to continue with the same body current path (right hand to both feet), though it would be beneficial to conduct additional investigations with other body current pathways.

References

1. 1. C.F. Dalziel, E. Ogden and C. Abbott, “Effect of Frequency on Let-go Currents,” Transactions of the American Institute of Electrical Engineers,
      vol. 62 1943.
   2. UL 101, “Leakage Current for Appliance,” Underwriters Laboratories Inc.
   3. IEC Technical Committee 64 Working Group 4, “Effects of current on human beings and livestock –
      Part 1: General aspects,” IEC 60479-1.
   4. UL 1310, “Class II Power Unit,” Underwriters Laboratories LLC.
   5. I. Schwartz, et al., “Excretion of Sodium and Potassium in Human Sweat,” Fall Meeting of the American Physiological Society, pp. 114-119, Madison, Wisconsin.
   6. R. Steim, F. René Kogler, and Christoph J. Brabec, “Interface materials for organic solar cells,”
      J. Mater. Chem., 2010, 20, 2499-2512.
   7. R. T. Tung, (2014). “The physics and chemistry of the Schottky barrier height,” Applied Physics Reviews, 1 (1).

Hai Jiang received his Ph.D. and Masters in Electrical Engineering from University of Dayton (Ohio). He is currently a Senior Research Engineer and global expert for electrical shock and leakage current at Underwriters Laboratories (UL). Jiang is a senior member of the IEEE Society and a Professional Engineer in the U.S. He is also the primary designated engineer (UL Standard Engineer) for UL101 Leakage Current For Appliances. Jiang can be reached at hai.jiang@ul.com.

Paul W. Brazis, Jr. is a Research Manager and Distinguished Member of the Technical Staff with UL Corporate Research at UL LLC (Northbrook, IL, USA). He has a background in electrical and thermal characterization, electronic materials, and device physics, receiving his BS, MS, and PhD in Electrical Engineering in 1995, 1997, and 2000 respectively, all from Northwestern University (Evanston, IL, USA). Brazis joined UL in 2008 and leads the Electrical & Mechanical Research team. He can be reached at paul.brazis@ul.com.